Final Report Summary - VMERGE (Emerging viral vector borne diseases)
Executive Summary:
In the field, intense sampling campaigns (WP1-WP3) have been undertaken by the Vmerge consortium in 2014 and in 2015, allowing the collection of mosquitoes, sera and environmental data. In laboratory, molecular and serological diagnostic tools have been developed and validated (D1.1 and D1.2). The use of these assays on the field-collected samples has provided a wealth of data on arbovirus prevalence, especially on RVFV prevalence, in the targeted regions. Coupled to a metagenomics approach, well-known pathogenic arboviruses and new viruses with a potential interest for human and livestock health have been identified (D1.3 and D1.4). Beyond known viruses, a huge amount of new viruses have been identified, significantly increasing current knowledge on virus diversity. Finally, the microbiomes of several mosquito vectors have been analysed, leading to the first detailed characterisation of the microbiome composition of an important vector (D1.5). Despite the long process for having authorizations to work at high levels of biosecurity and containment, Dutch and Spanish populations of Culex pipiens were found to be capable of transmitting RVFV (D2.1). Generally, the proportion infected increased with dose (D2.3) and this proportion continued to increase between days 14 and 28 post-infection, but no substantial effect of the different temperature regimes tested was observed (D2.2). Coinfection with Brugia nematodes did not increase infection rate (D2.5). Results also suggested that coinfection of Culicoides midges with multiple strains of bluetongue virus (BTV) resulted in reassortment confirming in vitro results (D2.4) and that the MP-12 strain of RVFV may be a promising vaccine candidate for use in camelid livestock in Europe (D2.6).
In silico, Vmerge successfully developed two approaches to assess the transmission risk of RVFV in endemic areas of Western Africa: a mechanistic model using field data as inputs and a climate-based modelling approach (D3.3 and D4.2). RVFV risk maps for up-to-now free areas (Northern Africa and Europe) were produced using results on mosquito abundance, but the figure is not substantially different from the risk maps created before (D3.1). In-field studies on the existing pathways of animal trade of importance for RVF introduction and spread in Egypt have been designed carried out (D4.3). A genetic module for RVF between EMPRES-i and Virus Pathogen Resource (ViPR) was developed (D4.7). Furthermore, a preliminary spatial analysis to track the climatic events able to favour RVF occurrence over North Africa was conducted (D4.4). For Culicoides-borne viruses, Vmerge developed a complete framework to: 1) confirm that opposition between Mediterranean and non-Mediterranean climates are an important change driver in Culicoides communities, 2) describe the structure of Culicoides imicola at different spatial scales and develop new genetic tools (D3.5) 3) characterize spatial and temporal overlap of wild and domestic hosts and various Culicoides species (D3.4) and 4) produce European risk maps for BTV and SBV (D3.2). To facilitate an integrative approach between all partners, different services were provided including access to quality data, optimization of sampling design for collection campaigns (D5.1 D5.4 and D5.2) training in spatial modelling, modelling software, and assistance to upscale regional vector distribution models (D5.3 and D5.6). Research outputs have included upgrading modelling software (D5.9 and D5.8) spatial distribution models for Rift Valley fever vectors, livestock movement models (D5.7) and hotspot framework overlapping hosts and Culicoides (D5.5). Automated risk map scripts were produced to ensure regular updated maps (D5.10). Dissemination included the maintenance of a website to promulgate information to sister projects and public.
One of the major concerns of the Vmerge project was to strengthen interactions between research and surveillance actors, to ensure translation of the scientific productions into policy, guidelines and manuals. The analysis of EU legislation in force for RVF and CBD and its consistency with the epidemiology of the diseases have been performed as well the national contingency plans in place in some southern EU countries have been assessed in collaboration with the Mediterranean Animal Health Network (D4.1 and D4.5). The path of transferring knowledge generated through academia into practical awareness via policy change is not a straight-forward or short-term venture. To disseminate the achievements of Vmerge at the country level, a variety of trainings, manuals, guidelines, and other tools have been adapted into realistic strategies and epidemiological scenarios in the different countries via 1) mapping of projects (D6.1) 2) translating results into policy (D6.2 and D6.3) 3) disseminating results to stakeholders across various institutional levels (D6.4 and D6.5).
Project Context and Objectives:
In the last decades, world was facing the emergence and the recrudescence of vector-borne diseases due to current and predicted environmental and socio-economic changes. In a retrospective analysis, Jones et al (2008) have shown that, while emerging-infectious disease events were most often detected in Europe or North America, the major risks are in or originate from southern countries. Indeed, many demographic, environmental, social and economic factors favouring the emergence of VBD are faced by these countries which only benefit from limited health facilities to prevent, monitor or control their spread. Therefore, understanding and controlling VBD is a global public good. European research into VBD should be seen as a tool for protecting and improving animal health and economics, public health and the welfare of the European economies and their citizens, as part of a coordinated, international effort at the global scale.
The majority of transboundary animal diseases are of viral origin. Nonetheless, the World Organization for Animal Health (OIE) has listed 82 notifiable diseases of major sanitary and economic importance for farm animals of which 43 are being caused by a virus and 20 are vector borne (11 arboviroses). Recent epizootics of bluetongue and Schmallenberg viruses in Europe and of Rift Valley fever virus outside Africa showed that viral vector-borne diseases have the potential to emerge and spread rapidly at remote distances, and are difficult to control. In this context, we need to develop a better understanding of the complex relations between vectors, viruses and their hosts at different scales (individual, vector population, landscape, and ecosystem). Moreover, the availability of predictive models and relevant scenarios (environmental changes, control strategies...) are essential for the development of adequate prevention and contingency programs. We need then to anticipate their establishment and spread by developing specific research programs and innovative surveillance and control strategies that can be translated into tools that can be quickly brought into action for a better cost-efficient control of vector-borne diseases, including those possibly still unknown.
These diseases have considerable impacts on animal farming as an important economic activity in Europe, but also as an important subsistence resource for more than 70% of poor people living in developing countries. Almost 2/3 of world farm animal populations are reared in developing countries (source FAO stat). They represent an essential source of proteins and contribute to the economic development of these countries and overall to food security. The demand in animal products in developed countries is decreasing while in developing countries it is growing by almost 3% per year (source World Bank).
Production losses resulting from animal diseases are estimated to be 20% in the world on average with a major impact in developing countries. For instance, the ban in Ethiopian livestock exports to Yemen and Saudi Arabia after the Rift Valley fever epizootics and epidemics in the Horn of Africa and the Middle East resulted in an estimated GDP loss of 25% in the Somali province of Ethiopia.
In a very different context, the net costs of the bluetongue virus-serotype 8 epizootic of 2007 was valued at 164–175 million Euros for the single case of the Netherlands, based on the output of a deterministic economic model. The cost to the EU as a whole was obviously substantially higher, but would have been even greater if there had not been an available body of expertise derived from EU and national research programs.
The Vmerge proposal relies on the ‘One Health’ concept, first suggested in 2004 by the Wildlife Conservation Society to strengthen links between human health, wild and domestic animal health and management of the environment, particularly biodiversity and ecosystem services. In 2008, a reference framework based on this concept was drawn up by six leading international organisations – World Health Organisation (WHO), Food and Agriculture Organisation (FAO), World animal health organisation (OIE), United Nations Children's Fund (UNICEF), the United Nations System Influenza Coordination (UNSIC), and the World Bank. In April 2010, FAO, OIE and WHO reiterated the importance and usefulness of this approach in a tripartite concept note on ‘Sharing responsibilities and coordinating global activities to address health risks at the animal-human-ecosystems interfaces’. The vision developed is: “A world capable of preventing, detecting, containing, eliminating, and responding to animal and public health risks attributable to zoonosis and animal diseases with an impact on food security through multi-sectorial cooperation and strong partnerships”. This approach was confirmed and developed at policy level by the G20 meeting of Ministers of Agriculture held in Paris (22-23 June 2011): “As far as public health, animal health are concerned, we stress the importance of strengthening international and regional networks, international standard setting taking into account national and regional differences, information, surveillance and traceability systems, good governance and official services, since they ensure an early detection and a rapid response to biological threats, facilitate threats, facilitate trade flows and contribute to global food security”.
The overall objective is to address the risk of introduction, emergence and spread of vector-borne viruses associated with mosquitoes (Diptera, Culicidae) and Culicoides biting midges (Diptera, Ceratopognidae), i.e. mainly Rift Valley fever, bluetongue and Schmallenberg viruses but much broader information is expected from field sampling of insect and host populations. We want to improve our currently limited understanding of these emerging vector-borne diseases and their potential for spread throughout northern Africa and Europe, and enhance epidemiological surveillance strategies and tools for better disease detection.
The specific objectives are:
• to provide innovations in diagnostic procedures for more reliable cost effective and quicker identification of emerging vector-borne diseases for Europe, northern Africa and the Sahelian region;
• to improve our understanding of vector competence of insect populations in ecosystems at risk of emerging vector-borne diseases;
• to model vector population dynamics and virus transmission at a fine spatial and temporal resolution, and upscale the models for better assessment of emergence and spread capacities of selected vector-borne diseases throughout the range of ecosystems;
• to assess the existing surveillance networks for emerging vector-borne diseases and propose new surveillance strategies in Europe, northern Africa and the Sahel.
Vmerge was designed to be integrative and comprehensive and embraces the entire knowledge acquisition chain from high quality research and experimentation to identification of control and surveillance methods and tools, and provision of sustainability and capacity building. The objectives were, following the identification of viruses with emerging potential, to develop experimental models on animals, and then, based on transmission models, to develop new tools for control and surveillance such as on line predictive and preventive risk mapping. Ensuring this continuum requires a strong partnership between research teams and labs from both North and South of the Mediterranean. Southern research teams and labs are from low income countries (Mauritania, Senegal and Tunisia) where diseases represent a major threat to Europe They carried out experimentation and contribute to setting up the surveillance methods and tools. Moreover, the participation of FAO to the project brings an institutional dimension which added international authority to effectively adopt the surveillance tools that Vmerge were identified.
The coordination and management structure was organised as represented in the chart in attachment.
Vmerge consortium was made of 16 beneficiaries from 12 countries (see figure below), including 5 international cooperation partner countries: Senegal, Mauritania, Morocco, Tunisia, and Egypt. It encompassed most of European and Mediterranean ecosystems, plus Mauritania and Senegal making the eco-epidemiological link with Sahelian Africa. Vmerge intended to establish a continuum between field and experimental research, and vector/vector-borne disease surveillance through a logical framework linking innovative diagnostics, prospective surveys in vector populations and their hosts sampled in selected ecosystems. This aimed to the identification of emerging risks of VBD for Europe and the Mediterranean Basin, and the implementation of improved surveillance strategies in this region, as well as in the Sahelian region of Africa, through a reinforced North-South partnership.
The Vmerge beneficiaries were:
Partner 1. Centre de coopération internationale en recherche agronomique pour le développement (CIRAD), Montpellier, France
Partner 2. Centre de Recerca en Sanitat Animal (CReSA), Spain
Partner 3. The Pirbright Institute (PIR), formerly Institute of Animal Health, UK
Partner 4. National Environment and Research Council, Centre for Ecology and Hydrology (NERC-CEH), UK
Partner 5. Agence nationale de sécurité sanitaire, de l’alimentation, de l’environnement et du travail (ANSES), France
Partner 6. Friedrich-Loeffler Institut (FLI), Germany
Partner 7. Euro-AEGIS, Belgium
Partner 8. Central Veterinary Institute of Wageningen University and Research Centre (CVI-WUR), The Netherlands
Partner 9. Food and Agriculture Organisation of the United Nations (FAO), Roma (Italy) FAO sub-regional office in Tunis (Tunisia)
Partner 10. Pathoquest, Paris, France
Partner 11. Istituto Zooprofilattico Sperimentale dell’Abruzzo e Molise “G. Caporale” (ICT), Teramo, Italy
Partner 12. Institut Agronomique et Vétérinaire Hassan II (IAV), Rabat, Morocco,
Partner 13. Institut sénégalais de recherché agricoles (ISRA), Dakar, Senegal,
Partner 14. Faculty of Veterinary Medicine-Alexandria University (FVM-AU), Alexandia, Egypt
Partner 15. Institut de recherche vétérinaire de Tunis (IRVT), Tunis,Tunisia
Partner 16. Centre national d’élevage et de recherches vétérinaires (CNERV), Nouakchott, Mauritania
Project Results:
During the Vmerge project, many efforts have been done to ensure a collaborative and integrative approach between all European and African partners. In WP1, partners have defined several projects structuring sampling designs beyond classical surveillance and now follow a common sampling strategy. The WP2 was mostly organized as an integrated international study to measure the competence of European Culex pipiens, the common house mosquito, for Rift Valley fever (RVF) virus. Joint protocols for mosquito infections have been discussed and agreed between partners. In WP3 and WP4, technical meetings were organized on ‘Spatio-temporal methods in RVF risk modelling and their integration’ to compare risk modelling approaches as well as facilitate data sharing among the project partners, and on ‘Animal movement and spread of RVF’ to discuss new methods to simulate and predict livestock movements. WP5 members helped to define sampling strategy, and WP6 leader has produced a dissemination plan, which includes guidelines to help partners understand their responsibilities in terms of dissemination and the main communication channels. It ensures that all Vmerge products are disseminated in a timely and ordered fashion.
In the field, intense sampling campaigns (WP1-WP3) have been undertaken in 2014 and in 2015, allowing the collection of mosquitoes, sera and environmental data.
Vector sampling. Campaigns for mosquito sampling have been carried out by all partners involved in sampling all along the mosquito season in 2015 (ISRA, CNERV, IAV, IRVT, FMV-AU, IZS, Cirad and Cresa). In addition, some partners (e.g. ISRA and IAV) have continued mosquito sampling during the 2016 season. Mosquito sampling has been structured mainly following three approaches. The first approach focused on Culex pipiens and involved the capture of Cx. pipiens females in several sites in five countries (IRVT, FMV-AU, IZS, Cirad and Cresa). The captured mosquitoes have been analysed using both specific and no-a-priori approaches (i.e. microarrays and shotgun sequencing). Beyond Cx. pipiens individuals, other species have been captured and analysed. Due to field constraints during the 2015 sampling campaign, IAV has carried out a complementary module within the Cx. pipiens approach. The goal of this module is to compare the virome of Cx. pipiens between larval and adult stages. Consequently, IAV has performed sampling during the 2016 season and the analysis of these samples is on-going. The second and third approaches involved the sampling of three mosquito vectors in different habitats in Senegal. ISRA has performed sampling campaigns in 2014, 2015 and 2016 in these habitats. The mosquitoes from 2014 and 2015 have been already processed and analysed. Finally, CNERV has carried out a sampling campaign in different Mauritanian sites where RVFV outbreaks have taken place.
Serum sampling. In parallel to mosquito sampling, several partners have carried out serum sampling campaigns. For example, ISRA has collected and analysed sera from sentinel herds (small ruminants) in different Senegalese localities (3 campaigns in 2015 in 3 different sites, 5 sentinel herds of 30 animals each) and from transhumant livestock (6 field trips, 120 samples/trip). CNERV has launched an extensive sampling campaign, collecting over 1000 sera (ruminants and camels) from sites involved in the Mauritanian RVFV outbreak in 2015. IRVT has also collected 850 sera from sheep.
Molecular and serological diagnostic tools have been developed and/or validated. Field-collected samples were analyzed with different methods in parallel.
Validation of FTA cards as arbovirus surveillance tools. Honey-soaked FTA cards were analysed for their suitability as arbovirus surveillance tools (Cresa). First, standardisation assays were performed to determine the honey concentration required to avoid PCR inhibition and mosquito mortality while allowing feeding. Moreover, validation with experimentally-infected mosquitoes was carried out. Finally, the cards were deployed during field tests (8 sites) and cards showing feeding were analysed with a Flavivirus-specific PCR and shotgun metagenomics. No Flavivirus was detected with the Flavivirus-specific RT-PCR. Shotgun metageomics, on the other hand, allowed the detection of different viruses, supporting the suitability of the FTA cards coupled to shotgun metagenomics as a tool in arbovirus surveillance programs.
Development of a Pan-Flavivirus PCR and a microarray for Flavivirus detection. In order to develop a new pan-Flavivirus microarray (Alere®), FLI has first developed and validated a new pan-Flavivirus SYBRGreen-based PCR. The goal of this development was to produce a robust method to get the amplicons to be used as template with the microarray. The new PCR design showed an improved sensibility when compared with an established design (Chao, 2007) using 26 flaviviruses.
The design of the microarray involved the selection of 84 probes, including probes specific for the whore Flavivirus genus, several Flavivirus groups (JEVG, YFVG, TBEV, Non-Vector), certain species (e.g. WNV, JEV, USUV, MVEV) and sub-species (e.g. WNV lineages). The microarray was then tested with 27 reference strains, some mixed together, showing excellent discrimination power.
Analysis of field-captured mosquitoes with pan-family PCRs and the pan-Flavivirus microarray. FLI analysed the presence of arboviruses in pools of mosquitoes collected. To this end, the Flavivirus microarray plus four PCRs were used: a pan-Flavivirus PCR, a pan-Bunyavirus PCR, a pan-Phlebovirus PCR and a pan-Alphavirus PCR. 1788 pools from 12 different mosquito species and were analysed in total, some with all the PCRs. Positive pools were sequenced. Isolation in cell culture was attempted from positive pools but without success. Several arboviruses were detected, including West Nile virus in Senegal (4 pools) and Italy (3 pools) or Usutu virus in France (13 pools). Beyond arboviruses, 544 pools were positive for mosquito-only flaviviruses.
Development of a SYBRGreen-based RT-PCR to study the prevalence of Alphamesonivirus 1. Alphamesonivirus 1 was largely predominant in several of the samples analysed with shotgun metagenomics (up to 60% of the viral reads in certain samples). Given its abundance, this virus may have a strong influence on the mosquito biology. A detailed knowledge of its ecology is required if this virus is to be used in mosquito or virus control. Cirad launched a study on its prevalence in Cx. pipiens populations in the French Carmargue. A real-time PCR was developed to detect the virus. This PCR was used to analyse 147 pools of females (mean number of mosquitoes per pool = 30; max/min= 34/6) collected in 6 sites along 2016. As shown in the figure below, mean infection rate was 8.1% (95% CI = 6.7-10.1). This infection rate is several orders of magnitude larger than that observed for most arboviruses infecting mosquitoes. No influence of site, habitat type or sampling month on the infection rate was detected.
Development of innovative Microsphere Immunoassay for specific Detection of orbiviruses or flaviviruses Antibodies. ANSES has developed two new microsphere-based immunoassays. The goal of these assays was to simplify the determination of serotypes of the orbivirus Bluetongue virus (BTV) or of Flavivirus species which, when determined by serological analyses, can only be performed by neutralization tests (VNT) which are time-consuming. To improve the serological characterization of BTV infections, the recombinant VP7 and VP2 viral proteins were synthesized using insect cells expression system. The purified antigens were covalently bound to fluorescent beads and then assayed with 822 characterized ruminant sera from BTV vaccinations or infections in a duplex microsphere immunoassay (MIA). The results demonstrated that MIA detected the anti-VP7 antibodies with a high specificity as well as a competitive ELISA approved for BTV diagnosis, with a better efficiency for the early detection of the anti-VP7 antibodies. The VP2 MIA results showed that this technology is also an alternative to VNT for BTV diagnosis. Comparisons between the VP2 MIA and VNT results showed that VNT detects the anti-VP2 antibodies in an early stage and that the VP2 MIA is as specific as VNT. This novel immunoassay provides a platform for developing multiplex assays, in which the presence of antibodies against multiple BTV serotypes can be detected simultaneously. In order to improve the serological differentiation of flavivirus infections, the recombinant soluble ectodomain of WNV E (WNV.sE) and EDIIIs (rEDIIIs) of WNV, JEV, and TBEV were synthesised using the Drosophila S2 expression system. Purified antigens were covalently bonded to fluorescent beads. The microspheres coupled to WNV.sE or rEDIIIs were assayed with about 300 equine immune sera from natural and experimental flavivirus infections and 172 nonimmune equine sera as negative controls. rEDIII-coupled microspheres captured specific antibodies against WNV, TBEV, or JEV in positive horse sera. This innovative multiplex immunoassay is a powerful alternative to ELISAs and VNTs for veterinary diagnosis of flavivirus-related diseases.
Serological surveys for RVFV infections in Senegal, Mauritania and Tunisia. Serum sampling and analysis for RVFV antibodies has been carried out by ISRA along 2015 and 2016. Sampling focused on sentinel herds (small ruminants) in different Senegalese localities (3 campaigns in 2015 in 3 different sites, 5 sentinel herds of 30 animals each) and on transhumant livestock (6 field trips, 120 samples/trip). The results show that RVFV infections of small ruminants have taken place in Senegal although at a low frequency. Positive sera, including a IgM positive, found in transhumant livestock strongly suggest RVFV introduction in Senegal from animal trade. Large serological surveys on RVFV prevalence have also been carried out in Mauritania (CNERV) and in Tunisia (IRVT) in 2015. CNERV found 75 positive sera among 1082 sera collected in several sites affected by an epizooty. On the contrary, IRVT did not found any positive serum among 850 sera collected from sheep all over the country.
Characterisation of differences in microbiota depending on vector and geographic distribution. The study of vector microbiomes was structured in three approaches, each with a different goal. The first approach, named the Culex pipiens project, targeted the characterisation of the virome of Culex pipiens, a RVFV vector, and the potential influence on virome composition of geographical location and human activity on the environment. Culex pipiens populations in sites with two different levels of human activity (peri-urban and rural sites) have been sampled in parallel in five countries (IRVT, FMV-AU, IZS, Cirad and CreSA) along the 2015 season. These sampling campaigns and the subsequent processing of mosquitoes have produced 10 samples (one sample per habitat and country).
The second approach studied the complementarity of classical serological surveys in livestock with virus surveys in mosquito vectors. More specifically, this project analyses the virome of Aedes vexans, a major RVFV vector, in three ecologically-similar sites along the rainy seasons in 2014 and 2015 (ISRA, 16 samples in total, some samplings did not lead to any capture). The results will be compared to those of a serological survey for the prevalence of different arboviruses carried out in the same sites and periods (ANSES).
The last approach explored the influence of habitat and taxonomy on the virome through the comparison of the viromes of two related mosquito vectors, Culex tritaeniorhynchus and Culex poicilipes, in three different habitats with specific surface-water conditions. The three habitats were sampled by ISRA two consecutive years (2014 and 2015), leading to the production of 12 samples.
For all approaches, mosquito sampling finished in late 2015, after the mosquito season, and was followed by a huge work on mosquito identification. For example, over 355,000 mosquitoes were identified to the species level by ISRA. Mosquitoes of the targeted species were pooled in groups of 30 individuals and their RNA extracted. Then, RNA from selected pools was subjected to shotgun sequencing by Pathoquest. Sequencing results were bioinformatically processed to obtain a list of viruses potentially present in each sample (Pathoquest). Then, statistical analyses were carried out to compare the virus communities from the different samples (Cirad). Despite the fact that the whole pipeline is highly demanding in terms of time, the results of the bioinformatics analysis are already available for all the samples. Nevertheless, statistical analysis has only been achieved for the Culex pipiens project. The results provided by this project represent a breakthrough in our knowledge of the mosquito virome. For the first time and thanks to, among others, the large geographical scale of the study, the virome of an important mosquito vector has been dissected in detail showing a virus community highly diverse comprising a large fraction of new virus species. The composition of the virome resembles that of the bacteriome in insects with a few dominant families representing most of the sequences. Variations depending on geographical location were observed, with African samples generally clustering apart from European samples. Nevertheless, a relatively large group of virus families was conserved among samples strongly suggesting that this group forms the core virome of Cx. pipiens and probably influences mosquito biology. An isolate of a member species of the core virome was obtained to be used in future studies on its influence of vector fitness and competence.
Key activities on vector competence and transmission were to determine the ability of European insects and livestock breeds to become infected with emerging arboviruses, primarily Rift Valley fever virus (RVFV), and how this is affected by external variation in population origin, environment, and ingested dose, by conducting experimental infections.
Rift Valley fever virus (RVFV) is transmitted between vertebrate hosts primarily via the bites of mosquitoes, particularly those in the genera Aedes and Culex. The commonest mosquito species known to be capable of transmitting RVFV is Culex pipiens, and preliminary studies of Culex pipiens obtained from France (Moutailler et al. 2008) suggest that some European populations of this species are relatively susceptible to RVFV infection. However, different populations of a species may have different levels of competence (e.g. Moutailler et al. 2008). Extrapolation from studies using a single vector population may poorly estimate the potential for a virus to spread and the extent to which that potential could vary geographically. Therefore, in order to accurately estimate possible regional variation in the potential for RVFV to spread within Europe, which is a key objective of the Vmerge project, it is essential to conduct competence studies using several geographically distinct populations of the species.
For their studies, DLO-CVI used Laboratory-reared Culex pipiens mosquitoes originating from an above-ground collected pool of egg rafts were generously provided by the laboratory of Entomology of Wageningen University. The colony originates from an above-ground collected pool of egg rafts collected in Brummen, The Netherlands and was established in 2010. The mosquitoes were maintained at 23ºC in BugDorm cages with a 16:8 light:dark cycle and 60% relative humidity. The mosquitoes were provided with 6% sucrose and bovine or chicken blood using a Hemotek PS5 (Discovery Workshops) for egg production. Egg rafts were hatched in tap water supplemented with Liquifry No. 1 (Interpet Ltd.). Larvae were fed with a mixture (1:1:1) of bovine liver powder, ground koi food and ground rabbit food.
At CReSA, two different populations of Cx. pipiens were used: Cx. pipiens form molestus from Empuriabrava and a hybrid between pipiens form and molestus form, from Gavà. The population of Ae. albopictus came from Sant Cugat de Vallés, a town belonging to the metropolitan area of Barcelona and the place where the Asian tiger was first identified in Spain in 2004. All mosquito populations have been reared in the laboratory to obtain stable colonies and the number of filial generations was more than 30. Mosquitoes were reared under the following environmental conditions: photoperiod was 14 h:10 h (light:dark) with two crepuscular cycles of 30 min between to simulate dawn and dusk; the mean temperature during day was 26˚C and 22˚C during night. Relative humidity (HR) was maintained constant at 80%. The mosquito colonies were tested for the presence of Flavivirus, Alphavirus and Phlebovirus by reverse transcription nested polymerase chain reaction (RT-nPCR) to confirm the absence of other viral infections. The colonies were also tested for the presence of Wolbachia spp. by PCR analysis of a fragment of wsp gene as previously described. All colonies resulted Wolbachia spp. positives (data not shown).
The Pirbright Institute experiments used Cx. pipiens form pipiens and Cx. pipiens form molestus hybrid mosquitoes originating from mosquito larvae and pupae collected from a container habitat in an allotment area in Surrey, UK (Manley et al., 2015). Mosquitoes have been reared in the laboratory to obtain stable colonies and the number of filial generations was more than 45. Mosquitoes were reared under the following environmental conditions: photoperiod was 16 h:8 h (light:dark) at 25 ±1°C. Relative humidity (HR) was maintained constant at 60 ±10%.
The Clone 13 strain of RVFV was used by TPI and CVI. This particular strain was used for two reasons: to increase comparability with published results from French mosquitoes and because biosecurity regulations in NL permit the use of this field-attenuated strain under lower containment requirements. The virus was propagated in mosquito cells C6/36. As this strain is the property of the Institut Pasteur, the KE&C department at IP was contacted and agreed to the use of the strain for the purposes of the project. CReSA used RVFV virulent strain 56/74 due to repeated difficulties in shipping Clone 13 within the timeframe required for the infection studies. The virus was propagated in mammalian BHK-21 cells and blood virus mixtures were prepared by resuspending cattle erythrocytes in freshly harvested virus-containing culture medium (DLO-CVI), mixing bovine blood with virus, heparin and ATP (CResA) or by mixing commercially available defibrinated sheep blood with freshly harvested virus containing culture medium from C6/36 cells.
Feeding was performed using a Hemotek feeding system. Mosquitoes were maintained at 26 °C (TPI), 28°C (CVI) or 26°C / 22°C day/night (CReSA) for either 14 (CReSA) or 21 (CVI, TPI) days and given 6% (CVI) or 10% (TPI, CReSA) sucrose in water ad libitum. At the end of the period, saliva was collected by forced salivation (capillary tube method). Saliva samples and bodies were tested by virus isolation using Vero cells (TPI, CVI and CReSA) and/or BHK cells (TPI). Virus isolation was detected by cytopathic effect and confirmed by quantitative real-time PCR (TPI, CVI and CReSA) or immunoperoxidase monolayer assay (IPMA; CVI).
Low or negligible levels of transmissible infections were observed in most populations tested, with exception of the experiments performed with Dutch Cx. pipiens mosquitoes using the Clone 13 at high viral dose. Analysis of the results obtained from Dutch populations indicates that approximately 80% of Dutch Culex pipiens develop a transmissible infection as detectable by virus isolation from saliva, when a dose of 1010.5 TCID50/ml was used and an incubation of 21 days at 28ºC. The results of infections at other doses are described under D2.3. No transmissible infections were seen in UK mosquitoes.
One factor which is well-established to affect the proportion of individuals developing a disseminated infection is the environmental temperature. The epidemiological importance of a mosquito species in the field ultimately depends on its ability to transmit virus to a new host. After ingesting a potentially infectious blood meal, transmission requires that the replicating virus overcomes or bypasses various barriers to dissemination to ultimately infect the salivary glands. Several factors affect mosquito competence by changing the potency or effectiveness of these barriers or by providing the virus with a way to evade them. One such factor is the simultaneous oral ingestion of microfilarial blood or tissue nematodes, such as Onchocerca or Brugia, which mechanically disrupt midgut integrity during infection. In one previous study, Aedes taeniorhynchus became more competent to RVFV infection and transmission (31% compared to 5%) when simultaneously fed Brugia malayi (Turell et al. 1984). Insect-transmitted blood and tissue nematodes such as Brugia are endemic in many parts of the world and their distribution overlaps with that of several important and high-impact veterinary, zoonotic and human arboviruses. The net effect of such coinfection is unclear, however, since it has also been noted in separate studies that this damage affects other parameters of vectorial capacity, namely the duration of the extrinsic incubation period (EIP) and the post-feeding mortality of the mosquitoes.
We investigated the consequences of ambient temperature and ‘shock’ temperature treatment on the proportion of Culex pipiens mosquitoes from European populations that develop transmissible infections with RVFV, and consider its possible implications for changes to the emergence potential of RVFV if introduced into Europe during extreme weather conditions. We also investigate the consequences of coinfection of mosquitoes with microfilarial nematodes (Brugia spp.).
To study the influence of temperature during the extrinsic incubation period (EIP), mosquitoes were allowed to feed on a blood meal containing a high dose of virus (1010.5 TCID50). Feeding was performed using a Hemotek feeding system. After feeding, the mosquitoes were maintained at different temperatures and a humidity of 70% and given 6% sucrose in water ad libitum. After an EIP of 14-, 21- or 28 days, the infection rates (IRs) and transmission rates (TRs) were determined. Saliva was collected by forced salivation (capillary tube method). Saliva samples and bodies were tested by virus isolation using Vero cells. Virus isolation was detected by cytopathic effect and confirmed by immunoperoxidase monolayer assay (IPMA). To study the effects of Brugia malayi coinfection, microfilariae were obtained from the Liverpool School of Tropical Medicine. Microfilariae were sourced from infected gerbils, counted, and re-suspended in RPMI medium at 10,000-12,000 microfilariae/ml, and kept at 37°C. Microfilariae were transported from the Liverpool School of Tropical Medicine to the Pirbright Institute on the day of collection, at approximately 25°C (~5-7 hours). After the arrival at the Pirbright Institute, the microfilaria were kept at 37°C until infection (<24 hours). Wild type (wt) strain of Bunyamwera virus (BUNV) was used. This virus is member of the Bunyaviridae virus family, is closely related to Rift Valley fever virus (RVFV), and is often used as a model bunyavirus. In the UK, BUNV is classified as a hazard group 2 biological agent by Advisory Committee on Dangerous Pathogens (ACDP) and it can be handled safely at containment level (CL) 2. BUNV has been shown to disseminate efficiently in mosquitoes of genus Aedes and Culex.
Seven to nine day old non-blood-fed female mosquitoes starved for 20-24 hours (no sucrose, no water) were used for infection experiment. Feeding was performed using a Hemotek feeding system. Mosquitoes were maintained at 26 °C for up to 14 days and given 10% w/v sucrose in water ad libitum. To investigate the competence of mosquitoes for B. malayi, individual mosquitoes were anesthetised and dissected (head, thorax and midguts) live into phosphate buffered saline with 0.75% bovine albumin (PBSA) or RPMI medium with 10% foetal bovine serum (FBS), incubated at 37 °C for 1 hour and then observed microscopically for the presence of B. malayi. To investigate the effect of B. malayi co-infection on the competence of mosquitoes for BUNV, individual mosquitoes were anesthetised, snap-frozen on dry ice and then dissected (head, thorax and midguts) into PBSA. Dissected body parts were homogenized using Bio-Rad TissueLyser at 20 Hz for 80 seconds and assayed for BUNV infection by plaque assay on BHK-21 cells.
The temperature study results show that the IRs found after EIPs at 20˚C are not significantly different from those found after incubation at 28˚C (compare Figs 3A and B). Virus titres in the bodies as well as the TRs were however lower after an EIP at 20ºC. Interestingly, when mosquitoes were fed and maintained for 7 days at 20˚C and subsequently maintained at 28˚C for 7, 14 or 21 days (corresponding to 14, 21 and 28 DPI), infection and transmission rates were similar as those obtained after corresponding EIPs at 28˚C.
Our findings suggest that the proportion of exposed mosquitoes with saliva from which infectious virus can be isolated increases between 14 and 28 days post-infection. However, the total proportion developing transmissible infections was not substantially different over the temperature maintenance regimes tested. We therefore conclude that for the conditions tested, there is not strong evidence to conclude that the efficiency of the process of infection itself changes with temperature, only the rate at which infection progresses.
For the Brugia studies, the competency of Aedes aegypti and Culex pipiens for Brugia malayi was investigated at 3, 7, 11 and 14 days post-infection (mosquitoes fed with 20,000 mf/ml). As anticipated based on literature research, Cx. pipiens mosquitoes were found to be non-competent for B. malayi while Ae. aegypti mosquitoes were found to be competent for B. malayi. Based on this, Aedes aegypti mosquitoes were chosen to investigate the effect of B. malayi co-infection on the competence of mosquitoes for BUNV. Analysis of mosquitoes 7 days post-infection revealed no increased infection and/or dissemination rate in mosquitoes co-infected with BUNV and B. malayi as compared to the mosquitoes infected with BUNV only. The survival rates were also comparable.
Analysis of mosquitoes 14 days post-infection suggested increased infection rate in mosquitoes co-infected with BUNV and B. malayi as compared to the mosquitoes infected with BUNV only. However, low numbers of available samples due to high mosquito mortality do not permit to analyse these data with high confidence.
We conducted additional experimental infection studies using the protocols developed above but with different infectious doses to allow the impact of feeding dose on competence to be estimated, improving our understanding of the uncertainties that can arise when extrapolating from laboratory infection studies to field conditions. At DLO-CVI, mosquitoes were maintained for 21 days at 28ºC with a humidity of 70% and a 16:8 light:dark photoperiod. After the extrinsic incubation period (EIP), body and saliva samples were evaluated for the presence of virus by virus isolation. All results were confirmed either by quantitative reverse transcriptase real-time PCR (qPCR) or immunoperoxidase monolayer assay (IPMA). Post EIP analysis of mosquitoes fed with the low-dose blood meal revealed an infection rate (IR) of 20% and a transmission rate (TR) of 8%. Analysis of mosquitoes fed a medium-dose revealed an IR of 64% and a TR of 14%. Mosquitoes fed a high-dose revealed a IR of 91% and a TR of 24%. In sporadic cases, virus isolations from saliva samples were positive, whereas isolations from corresponding bodies were negative (Fig. 4).
At CReSA, in the Cx. pipiens hybrid colony line, the rates of infection and dissemination tended to increase with higher doses of blood-virus mixture. During infections at the Pirbright Institute, the mosquito feeding rate ranged from 0.0% to 82.4%. The initial experiments were characterised by low feeding rates (0.0-6.1%). Following extensive protocol optimisation the feeding rates increased to 59.2-82.4% and were highly reproducible. For initial experiments, the mosquitoes were maintained for extrinsic incubation period (EIP) of 21 days at 26ºC with a humidity of 70% and a 16:8 light:dark photoperiod. The above conditions were altered to 14 days at 26ºC with a humidity of 70% and a 16:8 light:dark photoperiod for later experiments. Mosquitoes reared at two different larval densities (500 and 2000 larvae per 1 L of water) were tested. Two virus doses were used, low (~5.0 Log10 TCID50 per ml of blood meal) and high (~7.0 Log10 TCID50 per ml of blood meal). After the extrinsic incubation period (EIP), body and saliva samples were evaluated for the presence of virus by virus isolation. Post EIP analysis of mosquitoes fed with the low-dose or high-dose blood meal and incubated for 21 days revealed no evident infection. Survival rates for mosquitoes fed with low dose of the virus (26.8-34.2%) were higher than survival rates for mosquitoes fed with the high dose of the virus (14.7-17.8%).
In summary, the proportion of mosquitoes becoming infected, and developing transmissible infections, was observed to increase with higher oral infection doses. Additional experiments to investigate the infectiousness of blood obtained from infected animals were also conducted at CVI and suggested that the period of infectiousness to vectors may be shorter than the period for which viral RNA can be detected.
The susceptibility of European livestock breeds to RVFV, and their ability to infect mosquitoes that feed on them, are significant unknown components in estimating the threat RVFV poses to Europe. The infection is most significant in productive livestock such as small ruminants, cattle and camels. Characteristic ‘abortion storms’ with up to 100% new-borne case fatality rates, can be seen during acute infections. However, only moderate symptoms are observed in infected adult cattle and camels. The significant role of camelids in transmission and spread of the virus were already demonstrated by their involvement during the introduction of RVFV to Egypt 1977. During an epidemic of RVF in Mauritania 2010, severe clinical manifestations, such as ocular discharge, foot lesions, hemorrhages, neurological disorders as well as fatal hyperacute progressions were observed for the first time in camels. An association between infected camels and subsequent human infections highlights the previously underestimated role of camelids in terms of RVFV epidemiology. In particular, although camelids have been implicated as playing a significant role in the epidemiology of previous outbreaks of RVFV and the farming of camelids such as alpaca, vicuna and llamas in Europe has increased substantially in recent years, little is known about the outcome of infection in these species. The purpose of this deliverable is to determine the susceptibility of livestock breeds used in Europe, including camelids and sheep, to infection with RVFV, as well as the potential for European mosquitoes biting infectious animals to become infectious themselves and thereby sustain local transmission.
Three alpacas (used as model camelids) were challenged with the live attenuated MP-12 strain of RVFV, and immunogenicity and pathogenicity were measured. The MP-12 strain was chosen both for practical reasons, since in Germany it can be used for animal experiments under BSL-2 conditions, and because its attenuation means it has been proposed as a potential vaccine in the event of an outbreak. In addition, several comparable studies using the same strain were also carried out recently in cattle and sheep, allowing our results to be evaluated in a broader context. The MP-12 strain of RVFV was kindly provided by Professor Richard Elliot of the MRC Centre for Virus Research (CVR) at the University of Glasgow in the United Kingdom. MP-12 was propagated in Vero 76 cells (Collection of Cell Lines in Veterinary Medicine, Friedrich-Loeffler-Institut, Germany). Three male alpacas (Vicugna pacos), at the age of 6 month (#A1, #A2) and 4 years (#A3) respectively, were acquired from a local breeder. All three alpacas were immunized subcutaneously with 2 ml of a 106 TCID50/ml virus dilution. After inoculation, serum, whole blood, nasal- and conjunctival swabs and faeces were taken at 3,5,7,10,13,17,19,21,24,28 and 31 days post infection (dpi). Body temperatures were taken on the same days and additionally in the first two days post infection for recognition of any acute adverse effects of immunization. Serum neutralization test (SNT) was performed as described in the OIE Terrestrial Manual (2014). The presence of RVFV-derived RNA was verified using a quantitative real-time RT-PCR (RT-qPCR) by Bird et.al.
To compare both acute and subacute effects in this study, animals were euthanized at 3 (#A1) and 31 days (#A2, #A3) post infection (dpi). Clinical monitoring, analysis of liver enzymes and haematological parameters demonstrated the tolerability of the MP-12 strain, as no significant adverse effects were observed. Comprehensive analysis of serological parameters illustrated the immunogenicity of the vaccine, eliciting high neutralizing antibody titres and antibodies targeting different viral antigens. Significant serological responses were detectable in all assays for animal #A2 and #A3, whereas no antibody generation was found in animal #A1. For both #A2 and #A3 the generation of neutralizing antibodies started at 7 dpi. In the course of the experiment high neutralizing titres of up to 1:1280 were reached. RVFV was detected in serum and liver of the alpaca #A1 euthanized 3 dpi, whereas no virus was detectable at 31 dpi (#A2 and A3#). Viral replication has been determined by detection of various RVFV-antigens in hepatocytes by immunohistochemistry and the presence of mild multifocal necrotizing hepatitis.
In conclusion, the MP-12 RVFV strain was found to be a promising candidate vaccine for European camelid livestock.
To determine the period in which Culex pipiens mosquitoes become infected after feeding on viremic lambs, two 12-week-old lambs of the Dutch Texel-Swifter breed were inoculated with 105 TCID50 of the highly virulent RVFV strain 35/74 via intravenous route. EDTA blood samples, to be used for virus isolation, were collected every day after challenge. On days, 1, 2, 3 and 4, cardboard boxes containing 50 mosquitoes were placed on the inner thigh and fixed with elastic bandage (Fig. 5). After 45 min. the boxes were removed and brought to the biosafety level-3 (BSL-3) laboratory. Engorged mosquitoes were collected and transferred to new containers. The mosquitoes were held at 20˚C for 7 days and were subsequently transferred to 28˚C. After 5-6 days, the mosquitoes were assayed for the presence of virus in bodies and saliva. Both lambs developed viremia one day post infection (DPI) followed by a peak in body temperature on DPI 2 (Fig. 5A, B). The first lamb succumbed to the infection on DPI 3, whereas the second lamb had to be euthanized on DPI 4 after reaching a humane end-point. Mosquitoes were maintained at 20˚C for 7 days and 28˚C for 7 (mosquitoes that fed on DPI 1), 6 (mosquitoes that fed on DPI 2 or 3) or 5 (mosquitoes that fed on DPI 4) days. Analysis of mosquito samples after the extrinsic incubation time (EIP) revealed very high infection rates (IRs, 86-91%) in the groups of mosquitoes fed on DPI 2, but very low (11-18%) IRs in the groups of mosquitoes fed on DPI 3. This correlated with transmission rates (TRs), being 29-30% in mosquitoes that fed on DPI 2 and 0-11% in mosquitoes fed on DPI 3. The IR and TR of mosquitoes that fed on DPI 4 on lamb 1 were also very low, 8% and 0% respectively (Fig. 5C, D).
Interestingly, these results suggest that efficient transmission from viraemic lambs to mosquitoes takes place in a short time period of little more than 24 hrs.
In silico, extending models with host/vector interaction has been explored for RVF and Culicoides-borne diseases (CBD) to capture the interaction between hosts and vectors and the importance of environmental factors in both endemic and non-endemic situations. We successfully developed two approaches to assess the transmission risk of RVFV in endemic areas of Western Africa: a mechanistic model using field data as inputs and a climate-based modelling approach. RVFV risk maps for up-to-now free areas (Northern Africa and Europe) were produced using results on mosquito abundance, but the figure is not substantially different from the risk maps created before.
Forecasting RVF outbreaks in an endemic area. This was carried out using two different approaches. The first used a mechanistic approach using field data (serological analysis, entomological collections, analysis of nomadic herds movements) to predict the transmission of Rift Valley Fever virus (RVFV) in Senegal (led by Cirad, with strong interactions with ISRA, CNERV and ANSES). The second (led by FAO) aimed to test, calibrate and validate an existing climate-based model to identify and predict RVF at-risk areas in Senegal and Mauritania.
The mechanistic modelling of RVF transmission in Senegal was divided into the following steps:
1. Modelling the dynamics of Rift Valley fever mosquito vectors in different ecosystems of Northern Senegal: a rainfall-driven model of vector population dynamics developed by Soti et al. (2012) was applied to different study areas and its accuracy assessed using entomological field data (2015-2016). Upscaling the model allowed the development of a regional model, based on satellite-derived temperature, rainfall and humidity estimates;
2. Analysis the nomadic herd movements around Younoféré, and their potential role in the virus introduction into the pilot area: a field survey has been performed involving arriving nomadic herds during the 2015 rainy season, to measure key parameters, such as the main origin areas, the proportion of resident/nomadic herds according to time, and the proportion of seropositive nomadic animals upon arrival.
3. Modelling the dynamics of Rift Valley fever in different ecosystems of Northern Senegal: a deterministic compartmental model of RVF transmission integrating nomadic movements and using entomological model outputs as inputs was developed in Younofere pilot site, and calibrated with 2015 and 2016 serological data.
1. Modelling the dynamics of Rift Valley fever mosquito vectors in different ecosystems of Northern Senegal. In 2015 and 2016, mosquito populations were sampled by ISRA in the same 10 sampling sites as in 2014 from three different geographic areas (Diama, Keur Momar Sarr, Younouféré) using CDC light traps with CO2. Five trapping sessions were performed (jul-aug-sept-oct-nov). These captures allowed to collect 115,075 mosquitoes over 100 trap/nights, belonging to 31 species in 6 genera (Aedes, Aedeomyia, Anopheles, Culex, Mansonia and Uranotaenia). In Diama and Keur Momar Sarr, the preponderant species was Culex tritaeniorhynchus with respectively 53% and 33.6% and Mansonia uniformis with 36.7% and 46.4% respectively; whereas, Ae. vexans arabiensis (79.6%) was the most common mosquito in Younoufere. The model was initially developed for Ae. vexans and Cx. poicilipes in the Ferlo pastoral area (Northern Senegal). The dynamics of these two vector species is modelled by combining a hydrological model of the dynamics of the water bodies, with mosquito population models describing different stages of the mosquito life cycle. For Diama and Keur Momar Sarr, the dynamics of Cx. poicilipes and Culex tritaeniorhynchus was modeled using the same mosquito population model. For the three species, parameters and functions were defined according to a bibliographic review and expertise of ISRA partners. The parameters are constant values, whereas the functions (e.g. mortality, transition rates from one stage to the next one) are temperature, humidity, rain or flood-dependent. The model was implemented in the three study sites. Daily rainfall, humidity and temperature data were used as inputs. Entomological data collected during the 2014 and 2015 rainy seasons (July-November) were used to validate the models for the three species. The results showed a good fit of the predicted mosquito abundances with field data. Satellite-derived rainfall, temperature and humidity estimates were used as input of the models to simulate RVF vector population dynamics at a regional scale. The data were freely downloaded from internet databases. The assimilation of environmental data into the mosquito population model was realized with Ocelet, a domain specific language for modelling spatial dynamics and manipulating geographic information (www.ocelet.fr).
2. Analysis the nomadic herd movements around Younofere site. The objectives were: (a) To collect data on nomadic herders’ movements (spatial, temporal, reasons for movements), nomadic livestock characteristics (composition, size, RVF status) and the main features of the locations; (b) Graph an “observed” nomadic herder’s movement network and compute general network metrics, centrality measures and village and herder descriptive statistics; (c) Fit a model of the probability of occurrence of nomadic herd movements using an Exponential Random Graph Model (ERGM) and potential factors such as nodal attributes (village characteristics), edge attributes (number of animals moved, time of movement, seroprevalence in the herd) and structural and relational attributes (factors specific to nomad’s network) of the observed network; (d) Graph a simulated a nomads’ movement network from the fitted ERGM. During the 2015 rainy season, a 10km radius buffer zone centered on Younoféré was defined, as well as a census of the main temporary ponds in this zone. Among these ponds, 10 ponds were selected, based on their surface and the high/low frequentation by nomadic herds. From early August to mid-September 2015, 3 field sessions were conducted with (i) census of all nomadic herds settled around the ten ponds and (ii) movement questionnaire (ii) for each session, 4 herds newly arrived, in which 30 animals were sampled. This survey has been be replicated from early November to mid- December (3 sessions of census, questionnaire and sampling). Data has been gathered in a dedicated database. The total number of animals seen during the study period was 33,600 with 7,335 cattle heads, 19,466 sheep heads, 5,240 goats and 1,559 donkeys. The proposed exponential random graph model showed a good goodness of fit, and the simulated movement network showed several similarities with observed movement network when assessed on several parameters. Overall, nomads tend to own large herds and have movements geographically spread over all Senegal which highlights the importance of this work. Several nodes (towns or villages) such as Payar, Younoufere and others seem to be important hubs for nomads. Thus, planning interventions in these locations is highly recommended. Herders that have more goats tend to do less stops for water and food and thus do longer distances without stops. The presence of good quality grass in a stopping point makes nomads stay for longer periods of time and thus slow their movement habits.
3. Modelling the dynamics of Rift Valley fever in different ecosystems of Northern Senegal. The complete model, combining entomological, resident and transhumant herd components was first built and calibrated for Younoféré site. The epidemiological system was considered at the pond scale, considering that mosquitoes come from ponds and that the area of influence of that pond is of 2.5 km radius. Four dynamical processes were modelled: Host population renewal (arrival, departure of transhumants); Host population dynamics (gestation, birth, death, sales); Infection dynamics (virus transmission); Vector population dynamics (output of the entomological model). Host population dynamics was established from expert opinion and field observations. For nomadic populations, we recorded nomadic data as previously presented, within an area of 75km2. Regarding the vector population, we used the outputs of the entomological model previously described. To calibrate the model we used the results of a serological survey performed in 2016 around Younoféré. Results were in favor of an endemic circulation, at least since 2013. Last steps will be (i) to perform a sensitivity analysis on the model to identifiy the parameters that strongly influence the outputs, (ii) to test similar scenarios with Barkedji data (2003-2004), (iii) to combine this network with the 3 calibrated local models and see in what extent these movements influence the transmission/persistence of the virus in Northern Senegal, and spatialize the model through northern Senegal (iv) use this model to identify the most cost-efficient control measures and surveillance designs.
Time-series analysis on NDVI-based model. Within the framework of the Vmerge Work Package 3, FAO focused on testing, calibrating and validating an existing climatic model (hereafter, climate-based model) in Senegal and Mauritania, based on the work by Anyamba and collaborators (e.g. Anyamba et al. 2009), who developed a monitoring and risk mapping system that utilizes the Normalized Difference Vegetation Index (NDVI) as a proxy for ecological dynamics to map areas at potential risk of Rift Valley fever in East Africa. This is based on the findings that NDVI is a complex proxy representing the interaction of several variables, including rainfall, temperature, and soil moisture that are important for the emergence, survival, and propagation of Rift Valley fever mosquito vectors. Using this approach Anyamba and collaborators successfully retrospectively predicted areas where Rift Valley fever outbreaks occurred (Anyamba et al. 2002; Anyamba et al. 2009, 2010). Based on literature surveys and thresholding of climate variables (rainfall and long-term NDVI), the climate-based model creates a static mapping of the potential epizootic area mask, hereafter RVF vector suitability area (Anyamba et al. 2009). Within the suitability area, identification of risk areas is determined from NDVI anomaly calculations, taking into consideration the persistence of positive anomalies above a threshold of 0.025 NDVI units (based upon a desert calibration) over a 3-month period. The model thus captures these temporal dynamics in ecology using time-series measurements of NDVI. Pixels with persistently average positive anomalies over a three-month period exceeding a threshold of 0.1 NDVI are considered hotspots for vector amplification. An exhaustive description and explanation of the original model is available elsewhere (Anyamba et al. 2009).
A preliminary analysis and application of the uncalibrated climate-based model for the year 2003 and 2005 showed that: a) the RVF outbreaks occurred outside the vector suitability map, particularly in Mauritania and southern Senegal, indicating a low accuracy of the model for these regions; 2) the RVF at-risk areas were not predicted in semi-arid areas of Senegal (e.g. temporary ponds in the Ferlo region) and Mauritania; 3) however, the RVF outbreaks were accurately predicted along the Senegal river and in south-western Mauritania. The analysis of satellite-derived rainfall and NDVI data (source: FEWSNET) available at subnational level for areas of permanent water and temporary ponds, showed that: 1) the rainfall patterns and NDVI anomalies in areas of permanent water such as along the Senegal river were similar to those observed in the flooding areas of Kenya; 2) the rainfall patterns observed in the Ferlo region highlighted the presence of dry spell events even at subnational level. As described below, following these preliminary results, we updated the RVF vector suitability map and we developed two models to predict RVF at-risk areas for vector amplification under different ecological conditions: 1) the climate-based model for irrigated areas and flooded areas (e.g. along the Senegal river), and 2) the dry spell model for semi-arid areas and temporary ponds (e.g. the Ferlo region).
FAO compiled historical information available on outbreaks, serological, and entomological data collected from previous studies (e.g. ACI, CORUS, EDEN and EDENEXT projects; FAO EMPRES-i outbreaks database; FAO/Oxford RVF database) in Senegal and Mauritania, the WorldClim dataset (i.e. 19 temperature and rainfall bioclimatic variables, Hijmans et al. 2005) as well as time-series of satellite-derived climate (e.g. CHIRPS, TRMM, TAMSAT, RFE) and vegetation (e.g. NDVI from: eMODIS NDVI, MODIS AQUA NDVI, SPOT-VGT NDVI, METOP, PROBA-V) data spanning from 1981 to 2016. A database of 556 RVF records from 1988 to 2015 was used to train, calibrate, and validate the models. Then ecological niche modelling (i.e. MaxEnt (Phillips et al. 2006); kernel density) and time-series analysis were applied to identify: a) areas historically suitable for RVF occurrence; b) main RVF climatic predictors and c) climate anomalies that may trigger vector population dynamics in different ecosystems, such the Senegal River (e.g. St. Luis; Diama, Ross Bethio) and semi-arid areas (e.g. temporary ponds of the Ferlo area). A kernel density function was applied to the georeferenced RVF outbreaks (spanning from 1988 to 2015) to estimate a historical RVF distribution (90% contour) for model interpretation. Kernel density function 50% contours were used to identify yearly RVF hotspots for model validation. The RVF vector suitability area was updated using Principal Component Analysis and Iso-clustering of bioclimatic variables (FAO 2015) to define regions of similar ecological conditions for RVF vectors. The NDVI anomaly thresholds for Senegal and Mauritania were defining based on the analysis and comparison of climatic data (e.g. NDVI and rainfall) extracted at outbreak locations and RVF outbreak hotspots level (e.g. kernel contour) as well as by performing sensitivity analysis of the anomaly thresholds. The PCA and Iso-clustering method identified two main clusters including all RVF outbreaks of Senegal and Mauritania respectively (Fig. 6). The MaxEnt model, which was run for the Mauritania end Senegal clusters separately, was very accurate (AUC 0.98 for Senegal and 0.96 for Mauritania) and highlighted five main climatic predictors: Precipitation between August-October; Precipitation seasonality; Temperature seasonality; Mean Temperature of wettest quarter (between August-October); Mean Temperature of warmest quarter (between March-May). An important predictor for Mauritania was also Total annual precipitation. The two different models were then averaged to generate a final suitability map for RVF occurrence (Fig. 7).
Results confirmed that the original climate-based model used by FAO for East Africa performed well for years of well above-normal rains (e.g. year 2005, 2010, 2012, 2013), particularly along the Senegal river and in areas located in south-western Mauritania. However, it was not able to identify hotspots for potential RVF amplification in semiarid areas of the Ferlo region, particularly for years with irregular rainfall patterns, such as 2003 and 2008. The new 3-monthly NDVI average anomaly was set at 0.08 NDVI value for Senegal and Mauritania. In order to capture rainfall anomalies such as dry spell events that trigger Aedes sp. vector population dynamic in semi-arid areas (such as in the Ferlo region and Mauritania), a second model was developed. Based on literature review (Caminade et al. 2014), a dry spell was defined as a rainfall pause of about 10 days that is preceded and followed by a significant rainfall event (> 20 mm per day). The dry spell events were identified using CHIRPS pentadal data and TAMSAT decadal data (min and max thresholds of 5 and 20 mm respectively). The output maps of the ‘dry spell’ model were verified using the water pond monitoring system developed by USGS/USAID/FEWSNET (available at http://earlywarning.usgs.gov/fews/software-tools/21). The final RVF maps consisted in overlaying the output of the calibrated climate-based model (i.e. average cumulative anomalies) with the dry spell model to capture environmental and climatic drivers of vector dynamic in different ecosystems of the region (Fig. 8).
In-field studies on the existing pathways of animal trade of importance for RVF introduction and spread in Egypt have been designed carried out. CReSA and IZSAM decided to focus the efforts on Egypt because of its relevance in relation to the risk of RVF introduction into North Africa and to complement the studies CIRAD was already carrying out in other Western African countries. After the meeting held in Egypt in May 2015 between CreSA team and the Official Veterinary Authorities, they agreed for Egypt’s participation within VMERGE project and the questionnaires designed for Northern African countries were adapted into a new revised English version. Data on legal movements of cattle and camels was collected through the questionnaires above mentioned. The main objective of the studies carried out in Egypt was to gain an understanding of the legal trade of animals into Egypt and evaluate its impact in relation to the risk of RVF introduction. To do that, data on all legal importations of animals into Egypt between 2012 and 2015 (i.e. cattle and camels imported from Ethiopia and Sudan) was provided by the Quarantine & Inspection Department of the Ministry of Agriculture and Land Reclamation of Egypt (QID-MALRE). All cattle and camels legally imported into Egypt need to be quarantined. The QID-MALRE also provided data on the specific quarantine facility to which each of the lots of animals imported during this 4 years-period were taken. Once released from the quarantine facilities, animals may be directly sent to slaughterhouses or to animal markets. Data on the location and characteristics of slaughterhouses and animal markets were provided by the Department of Epidemiology of the Ministry of Agriculture and Land Reclamation of Egypt (DE-MALRE). The second objective of the studies carried out in Egypt was to evaluate the risk of RVFV transmission within Egypt. It was evaluated using a GIS-based multi-criteria evaluation (MCE) approach, which allows to combine the data on the distribution of the different factors may influencing the risk of a given disease. In the case of RVF it is essential to account for the distribution of both susceptible hosts of RVF (cattle, buffalo, sheep, goats and camels), and the distribution of the vectors of RVFV. For the estimation of the distribution of susceptible host, data on the number of vaccine doses administered by animal species and by governorates were provided by the ED-MALRE. Finally, animal trade data and risk of RVFV transmission data were combined to identify those areas and periods in which the introduction of RVFV is more likely, so that adequate control measures can be implemented.
A genetic module for RVF between EMPRES-i and Virus Pathogen Resource (ViPR) was developed. The primary objectives were to:
1. To explore which countries, individuals and research organizations have genomic/genetic RVFV sequences that could be annotated and submitted to the repositories (e.g. GenBank/DDBJ/EMBL) of the International Nucleotide Sequence Database Collaboration (INSDC) or to another centralized database;
2. To explore if these databases include sequences that need further annotations or quality checks;
3. Which steps are needed to provide open source bioinformatic tools to analyze genetic data and link this information to epidemiological data;
4. To develop a RVF sequence centralized database and an EMPRES-i genetic module for RVF according to the work-plan developed under (3).
SIB conducted a review of the existing virus sequence platforms that would be appropriate for the development of a genetic module and they identified the Virus Pathogen Resource (ViPR2) as an excellent option for this linkage for several reasons, including the uniqueness and completeness of the database for RVF, the fact that developers were already known and appreciated (previously developed the Influenza Research Database, IRD), the presence of analytical and visualization tools, and the multiple pathogens addressed in this database (allowing for genetic modules to be developed for other pathogens in the future). FAO consulted the RVF specialist members of Vmerge to explore whether such a genetic module would be useful to their scientific needs. The interest for such an initiative was fully confirmed but with the caveats that a larger proportion of long sequences and even whole genome sequencing should be generated to enable more accurate phylogenetic studies, and that the annotation of the RVF virus would also be needed (this step should be performed by SIB in the near future and will be published in ViralZone).
FAO first conducted extensive desktop work to identify linkages between recorded RVF events in EMPRES-i (all species, including humans) and deposited RVF sequences in ViPR. After analysing the roughly 2000 RVF events in EMPRES-i against the 898 RVF genetic sequences in ViPR, it was determined that many of the listings could be linked through similarities of species, date, and location. Importantly however, discrepancies were discovered and suggestions made to improve data on both sides in order to facilitate linkages in the future. The sharing of unpublished RVF genetic sequences from external will be encouraged in the future, and once publicly available, the linkage between these sequences and matching events will be released online through the EMPRES-i Genetic Module (currently only available through ViPR).
For Culicoides-borne viruses, we developed a complete framework to: 1) confirm that opposition between Mediterranean and non-Mediterranean climates are an important change driver in Culicoides communities, 2) describe the structure of Culicoides imicola at different spatial scales and develop new genetic tools, 3) characterize spatial and temporal overlap of wild and domestic hosts and various Culicoides species, and 4) produce European risk maps for BTV and SBV.
Culicoides species communities and population structure. The strategy was first to describe the evolution of Culicoides community species along a transect from Mediterranean areas to non-Mediterranean areas, then to describe the structure of Culicoides imicola populations as the main vector in Mediterranean areas, and to develop genetic tools which will allow the further study of Culicoides obsoletus and Culicoides chiopterus populations, as important vector species for non-Mediterranean Europe.
We first confirmed that BTV transmission areas could be separated into two distinct epidemiological systems where vector communities are distinct: the Mediterranean areas where Culicoides imicola is dominant and non-Mediterranean areas where communities are dominated by Culicoides obsoletus/Culicoides scoticus associated to Culicoides dewulfi and Culicoides chiopterus in north-western Europe.
Then, we characterized, using a multi-locus approach including maternally and bi-parentally inherited markers, the colonization history of C. imicola, exploring the demographic, historic and evolutionary events shaping the current distribution of C. imicola at different scales: within the current distribution areas of this species, within the Mediterranean basin, and more locally between France and Spain. More specifically, we depicted the geographical pattern of genetic variation of C. imicola in the Afrotropical region and in the Mediterranean basin by a phylogeographical approach. We then characterized the genetic structure using Bayesian clustering and traditional population genetics tools. We finally performed approximate Bayesian computation analyses (ABC) in order to retrace and date the colonization events. We also combined entomological surveys, movement simulations of air-borne particles, and population genetics to reconstruct the chain of events that led to a newly colonized French area nestled at the northern foot of the Pyrenees.
This allowed us to propose a new hypothesis explaining the current distribution of C. imicola within the Mediterranean basin. This hypothesis supports an ancient presence of C. imicola in the Mediterranean basin along its African and European coasts. This invalidates the consensual hypothesis of a very recent a northward expansion of C. imicola. The reinforced entomological surveillance, following BTV outbreaks, would have only revealed the presence of long-settled populations. Other factors than the presence/absence of the vector have driven the spread of BTV in the Mediterranean basin such an increase of the population sizes observed in the western Mediterranean in connection with an increase of the temperature.
At a more local scale, entomological surveys conducted from 2008 to 2012 indicate the establishment of C. imicola populations in the in the French Pyrenees-Orientales department, suggesting a close genetic relationship to the nearby (~80 km south) populations settled in Catalonia. Instead, the newly settled C. imicola population was shown by both nuclear and mitochondrial genetic loci to be closely related to far more distant populations (360 to 1,000 km east or south-east) in the Var department, Corsica, Sardinia and Algeria. Corsica was further supported as the most likely source of introduction by the ABC analyses, suggesting that establishment of C. imicola in Pyrénées-Orientales could have occurred through long-distance dispersal from abundant populations in the island (> 500 km from the mainland sampling site). Simulating the movement of air-borne particles evidenced frequent wind-transport events allowing, within at most 36 hours, the immigration of midges from north-eastern Spain and Balearic Islands, and, as rare events, their immigration from Corsica at dates corresponding to the period of abundance of this species.
Finally, we developed microsatellite markers for C. obsoletus and C. chiopterus, as important BTV vectors in Europe, to address their genetic diversity at different scales. This will also quantify gene flows and dispersion abilities of both species. Culicoides obsoletus is an ubiquitous species, able to breed in a range of habitats, in both farm holdings and other landscape types. In contrast, C. chiopterus is a specialist species that has only been isolated as larvae from cattle dung. This provides a contrasting ecology that we hypothesize will have a significant influence on dispersal behavior. The markers were tested for C. obsoletus on three populations collected in England, in France and in Balearic Islands. Individuals were identified morphologically and by PCR species-specific diagnostic assay. In the population from France, we observed many non-amplified samples and some allele sizes outside of the normal range at some loci. All samples were hence sequenced for a portion of the mitochondrial genes cytochrome oxidase subunit I. This population was actually composed of C. obsoletus sensu stricto but also by the recently described "dark obsoletus" and by C. obsoletus of the O2 group which are certainly distinct species. Culicoides montanus specimens were also detected in the Spanish and English populations. The following analysis were carried out on the C. obsoletus s.s. only. For 8 loci, a significant heterozygote deficit was observed which for four loci could be explained by null alleles. A genetic structure was highlighted differentiating the populations from France and England and from the population of the Balearic Islands. A Wahlund effect may thus explain some of the observed heterozygous deficits. Here again we will continue the technical development of these markers and test them on a larger number of populations. The markers were tested also for C. chiopterus on three populations collected in France. For 9 loci there is a significant heterozygote deficiency which for four loci can be explained by null alleles, i.e. mutations in the nucleotide sequences of flanking regions that may prevent the primer annealing to template DNA during amplification of the microsatellite locus by PCR. No genetic structure could be evidenced. It will therefore be necessary to continue the technical development of these markers and to test them on a larger number of populations.
Interaction between domestic and wildlife hosts and their vectors. A risk framework was developed to explore how the spatial and temporal overlap of wild and domestic hosts and various Culicoides biting midge vector taxa may affect vector-borne disease risk across three European landscapes. We particularly focus on the risk posed by the bluetongue (BTV) and Schmallenberg (SBV) viruses which recently emerged in Europe and caused huge impact on the livestock sector.BTV has long been considered as endemic in wild ruminants in large parts of Africa and North America and wild ruminants can act as maintenance hosts. The role of European wild ruminants in BTV and SBV transmission and maintenance began to be investigated only following the arrival of exotic strains of these viruses in Northern Europe and is still under debate. Given the diversity of wild and domestic ruminants and midge species involved in transmission of both BTV and SBV in Europe, it is critical to consider how, where and when these actors in the transmission cycle will coincide and how this may be modulated by environmental variability, such as drought and flooding regimes.
In this research we develop predictive environmentally-driven models for the seasonal abundance, group size and group frequency of wild deer species in three regions of Europe, and overlay these predictions with those from a statistical model predicting the seasonal abundance of key vector species in the same locations. These overlays allow for a landscape level examination of spatial variability in potential disease risk, accounting for host and vector affinities for different habitats, meteorological variation and other environmental factors.
We demonstrate considerable spatial and seasonal variations in the risk posed by BTV and SBV in three regions of Europe. Specifically:
- Simulated drought and flood conditions had significant impacts on the densities and number of groups of wild ruminant hosts and also on the length and intensity of adult Culicoides biting activity, with the latter varying between landscapes, between Palearctic versus Afro-tropical vector taxa and between habitats.
- In particular, drought was predicted to reduce the density and frequency of red deer groups per 9km2 landscape, probably due to the reduced forage availability and quality. Flood reduced deer densities probably because saturated soils reduced the growth of dry-adapted grasses but could sometimes increase the number of groups because it can increase vegetation productivity in more elevated areas favoured by red deer.
- For Culicoides vectors, wetter conditions reflecting flood tended to produce adult seasons that were longer with larger peaks in abundance, whilst dry conditions reflecting drought had the opposite effect, with both effects being habitat dependent. Adult seasons were generally longer with larger peak abundances in less open habitats i.e. in grassland versus open shrub or savanna for C. imicola in Spain and in forest or crop-mosaic habitat versus grassland in northern Europe. A potential biological explanation for this finding may be that the semi-aquatic breeding sites used by immature Culicoides may dry out more rapidly in more open habitats.
Considering the Scotland landscape, the risk framework identified:
- Late June to early September as a key risk period where high adult midge vector biting density coincides with high density and group sizes of the major wild ruminant reservoir for BTV, red deer, particularly in north and central areas, and when red deer, cattle and sheep all tend to move to higher elevation areas to access higher quality forage or avoid heat.
- That the risk of SBV reproductive losses is low in an average year because the risk period for in-lamb ewes in mid-September occurs just prior to the red deer rut when adult vector abundance is low.
- However, vector abundance can be high in some areas and years during this period, making delays in the timing of breeding of sheep in Scotland potentially advantageous.
For the Netherlands landscape, the risk framework identified:
- A higher potential for SBV to circulate between sheep, deer and Culicoides during the second seasonal peak of adult Culicoides activity (July-October) because adult vector season extends for longer after the start of the risk period for SBV reproductive losses in in-lamb ewes.
- Under conditions of increased wetness, the vector season is extended much later in the year, which is predicted to increase SBV impacts by increasing the exposure of red deer during the rutting season and exposure of livestock prior to be moved indoors for the winter. This suggests that livestock should be moved indoors earlier in a wet year if avoidance of SBV impacts is of high priority relative to other health and nutritional constraints on management.
- Conversely, drier conditions reduce the magnitude of the second seasonal peak in adults and the end adult vector activity earlier, vastly reducing the risk of reproductive losses from SBV.
- Geographically, the greatest risk of SBV and BTV is predicted to occur in the central region of the country, in which large predicted group sizes for red deer and seasonal adult vector abundance coincide with high densities of sheep and cattle.
For the Spanish landscape, the risk framework identified:
- A higher potential for SBV to circulate between sheep, deer and Culicoides during the second seasonal peak of adult Culicoides activity (July-October) than in Scotland and under wetter conditions because adult vector season extends for longer after the start of the risk period for SBV reproductive losses in in-lamb ewes. Drier conditions again reduced late adult vector season activity of C. obsoletus complex midges, again lowering the risk of SBV impacts
- A significant risk period for transmission of BTV or SBV to cycle between livestock and roe deer in April and early May when roe deer graze on pastures to exploit greening up vegetation after the overwinter dry season though vector abundance is relatively low at this time.
- The spatial pattern of predicted relative risk of transmission differed considerably between the two vector species, with SBV risk for the C. obsoletus complex concentrated in the southwest and northern, more mountainous region of Andalucia, and risk for C. imicola tending to be greatest in central Andalusia.
From snapshot sampling of the midge vector communities in sheep, red deer and roe deer habitats in Scotland, the C. obsoletus/scoticus taxa was identified as having the largest potential for disease transmission amongst wild and domestic hosts, being present in all three host habitats in large numbers. The abundance of C. pulicaris was also high over all three habitats, representing the potential for this species to also act as a bridge vector among wild and domestic hosts. These results highlight the ongoing need for more ecologically-driven estimates of disease risk in European landscapes that capture changes to critical eco-epidemiological processes under different environmental conditions. Such work will allow for more effective and pro-active disease control measures to be developed, tailored for specific landscapes, habitat types and climatic scenarios.
Riskmapping with virus transmission models of BTV and SBV. We have used the models as published in Fisher et al, 2013 and Hartemink et al. 2011, to prepare a method for the creation of risk maps. We assume that SBV and BTV behave very similarly, since they are transmitted by the same vectors and affect the same host species. Other partners in the VMERGE project created methods to extrapolate from vector data collection to predictions on vector abundance beyond the sampled locations, using geographically detailed information on aspects like soil type, temperature, humidity, land use, etc. This map of predicted vector abundance is input for the risk map for transmission of vector-borne infections, when combined with the transmission models. Using the available information, we derive risk maps for BTV-8 and SBV in Europe. The risk maps for other BTV serotypes are different, since other Culicoides species are driving their transmission. In figure 11 a risk map for an average temperature of 16 degrees Celsius is shown. Exact scaling of the risk is not possible due to lack of data. The results are therefore all given in terms of relative risk.
To facilitate an integrative approach between all partners, different services were provided including access to quality data, optimization of sampling design for collection campaigns, training in spatial modelling, modelling software, and assistance to upscale regional vector distribution models. Research outputs have included upgrading modelling software, spatial distribution models for Rift Valley fever vectors, livestock movement models, and hotspot framework overlapping hosts and Culicoides. Automated risk map scripts were produced to ensure regular updated maps. Dissemination included the maintenance of a website to promulgate information to sister projects and public.
Spatial Services and Data Management. Provision of spatial data and customization of the data to VMERGE needs were provided to project partners through the VMERGE data website (http://www.VMERGEdata.com/). This website ensured access to large number of spatial datasets amounting several terabytes supporting the needs of other work packages. As examples, was produced a very extensive updated archive of remotely sensed climate indices derived by Temporal Fourier Analysis from MODIS imagery for 2001 – 2015 for day and night time land surface temperature; and for normalised and enhanced vegetation indices. These covariates represent the foundation of much of the spatial modelling undertaken in VMERGE and in other sister projects. New spatial distribution presence absence models of potential hosts, including horses and roe deer, as well as updates of presence absence and abundance indices for red deer for use in hotspot mapping (see below). These were produced as ensembles of Boosted Regression Tree and zoned Random Forest models, implemented by the VECMAP software suite. All data outputs are deposited on the VMERGE data website either for direct access or access on request. A mosaicked and processed dataset of MODIS derived of vegetation phenology indicators for 2005-12. Access to all archived was controlled by a password mediated registration process so that VMERGE partners were able to download public domain datasets customised and standardised for partners, as well as many datasets initially restricted to VMERGE partners only, including those mentioned above. Crucially, these data will continue to be accessible to all VMERGE project partners (and indeed to partners from other completed FP7 projects such as EDENext, IDAMS and EDEN) for a number of years as funding has been allocated from ongoing projects to maintain the data site, and to expand it into new topics and areas under the names www.palebludata.com and for generically www.spatialdatasite.com.
Upscaled RVF models. Important gaps of information remain about the distribution of potential vectors of RVF in Africa. The objective of this task was to upscale RVF models through the generation of habitat suitability maps for Aedes and Culex in Morocco, Tunisia, Mauritania, and Senegal. Maps generated can be used to target surveillance areas (WP4) and to inform transmission modelling (WP3). In total 445 sites distributed in Morocco (partners #12), Tunisia (partners #15), Mauritania (partners #16), and Senegal (partners #13), were visited to detect mosquitoes. Mosquitoes of the genus Aedes and Culex were detected in 115 sites and 272 sites respectively. The data were merged and used by partners #7 to produce species distribution models. Additional Aedes and Culex presence data compiled during a systematic literature review on RVF vectors (Versteirt et al., 2013) were combined to the data collected during the project of VMERGE. Using a Mahalanobis distance mask, data collected in areas with distinct environmental conditions than those observed within the four African countries involved in VMERGE project were withdrawn. Based on the combined number of occurrence data, four species were retained for species distribution modelling: Ae. vexans sI (n=122), Cx. pipiens (n=338), Cx. theileri (n=142), and Cx. tritaeniorhynchus (n=139). Random forest modelling technique was used to estimate the distribution of the four species. For this approach, balanced datasets for both presence and absence training data were required to avoid bias towards predicting presence or absence. The absence data was constituted by a random subset of visited sites in the VMERGE countries and pseudo-absences locations generated by the method of surface range envelope, as suggested by Barbet-Massin et al. (2012), within the boundaries of the Mahalanobis distance mask.
To predict the occurrence of the four mosquitoes, predictor variable datasets from the VMERGE data repository (http://www.vmergedata.com) were used; they include environmental eco-climatic variables and anthropogenic variables such as urbanization. Using the VECMAP® software, a random forest of 600 trees was trained for each species dataset (Breiman, 2001). The random forest models produced probability maps of occurrence based on environmental conditions. When conditions are suitable, the probability values are high, indicating where the species would be able to establish after introduction. The training data were only taken from areas that are similar of the VMERGE countries; therefore, environmental conditions homogeneous within the VMERGE area are not designated as important variables, while in fact those conditions are important for the species to occur. For Cx. pipiens, the probability of occurrence is presented in Figure 12. The spatial distribution was determined by temperature and urbanization (AUC = 0.95).
Mapping hotspots of interaction between Culicoides and domestic and wild ruminants. The objective was to develop a novel, continental-scale framework for identifying and mapping key geographical hotspots of overlap between wild deer hosts, domestic ruminant hosts and Culicoides biting midge vectors of Schmallenberg (SBV) and bluetongue viruses (BTV) across Europe, assuming that these hotspots would be most favourable for establishment and spread of transmission. As well as domestic ruminants, transmission of these Culicoides-borne viruses in Europe involves both Afro-tropical and Palaeartic vector species and diverse wild ruminant species. The hotspot framework was developed by overlaying suitable habitat for deer species and abundance patterns of Culicoides vectors and susceptible and reservoir livestock species derived from environmentally-driven, Europe-wide spatial models.
Livestock movement analysis. The objective was to generate a model, which quantifies the factors influencing animal movement patterns at the national and international levels. This model might be used to predict the probability of a trading connection between any spatial units and, consequently, could be used to inform disease transmission models. At the international scale, the dataset of animal movement was of too low quantity and quality to allow its used to train a predictive spatio-temporal animal movement model for the region of interest of Vmerge. At a regional level, however, a large and detailed spatio-temporal data set of animal movement records was assembled by partner #1 in collaboration with partners #16 in Mauritania and #13 in Senegal. For Mauritania, the trade network included 2203 trade connections between 71 national nodes and 42225 possible paths linking these nodes. Population data on human and livestock distribution was estimated by extracting populations from the Worldpop database (human population) and from the Gridded Livestock of the World v.2. The primary analytical framework consisted of the use of gravity models, where the levels of movement are determined by the size of the populations and origin and destination. This type of model is frequently used in migration studies to predict movement patterns between spatial units (Garcia et al., 2014). A great advantage of gravity models is that they can easily be fitted to observation data sets using linear models, and that they can be used for predictive purposes to establish levels of flow between locations. However, they do not allow characterizing the structure of the trade network, so they are frequently used in complement to social-network analysis metrics. A strong seasonal pattern was observed in the animal movement data, corresponding to an important shift between the means of transport between movements on foot and movements by truck, those latter mainly during the Tabaski period. In both cases, models predicting the probability of a trading connexion between two spatial units were produced with high accuracy.
For Mauritania, there was no noticeable difference between the use of Euclidian and least-cost distance as predictor of the probability of a movement connection. This pattern is typical of a trade induce movement – animals are moved more to areas with many livestock because that is where the markets are, and the distance is a deterrent as the potential losses are greater for longer movements. Once validated against the trade movement data collected in Senegal, this model could be used to predict the probability of a trading connection between any spatial units and inform disease models by providing a direct quantification of the transmission risk over long distances through trade at different points in time within a year.
Software Development. VECMAP®, a system developed by Avia-GIS (http://www.vecmap.com Figure 27), was used by the partners from Morocco (partners #12), Tunisia (partners #15), Mauritania (partners #16), and Senegal (partners #13). Based on their experience in using the system, new relevant features have been integrated into VECMAP®. Those new features have been tested successfully and validated during this project. The following sections present the new features included in VECMAP®. VECMAP® can create sampling strategy adapted to budgets constraints and allocate sample sites to the closest team of field staff. To support surveys done in Mauritania (#16) and in Senegal (#13), VECMAP® was upgraded by adding features enabling longitudinal sampling. To facilitate further the collection of field data, VECMAP® offers an automated approach for smartphones and tablets. Using the built-in GPS, every sample location can be recorded ensuring accurate data location. Furthermore, downloaded maps can be accessed offline. It also contains precise instructions for the field operators, telling them exactly what to look for, where to look for it and what to record. To standardise data as much as possible, various field types (textarea, radio buttons, barcodes, comboboxes, lists, etc.) were constructed to facilitate fieldwork and data processing. The collected data can be sent to the server when online or extracted when no Internet is available. In the context of the VMERGE project, an update mosquitoes list with African species was incorporated in VECMAP® ensuring correct classification of the vectors identified. Identifications are logged and stored centrally in the VECMAP database. Another new feature was a tool allowing to export species data stored as text file.
After spatial analysis, the training dataset is used as input for species distribution models that are developed via a wide range of statistical methods. These models are predictive; filling in distribution gaps and predicting species occurrence probability and abundance in locations on the map where no sampling has taken place. During the VMERGE project, one of the modelling techniques, the Generalised Linear Model, has been updated to handle spatial autocorrelation. It is now possible to assess the degree of coefficient variability using a generalised geographically weighted regression (GGWR). Two remaining options handle spatial autocorrelation in different ways: the autoregressive term (RAC) and the GLSM (mixed model). In addition, a zoning option was developed and the user can choose whether to make the model using pre-defined eco-zones. To improve the user friendly graphical interface, it is now possible to apply a default legend to model output.
Automatic procedure to update risk maps. The objective was to develop an automatic procedure to produce updated risk maps on a monthly-basis for Senegal and Mauritania (see above for modelling description). Results showed that the original climate-based model used by FAO for East Africa performed well for years of well above-normal rains, particularly along the Senegal river and in areas located in south-western Mauritania. However, it was not able to identify hotspots for potential RVF amplification in semiarid areas of the Ferlo region, particularly for years with irregular rainfall patterns. The scripts developed for identifying RVF potential areas based on the two models, i.e. a) the calibrated climate-based model (e.g. average cumulative NDVI anomalies) and b) the dry spell model. The scripts were written in Python and in JavaScript for their application in GEE environment and use of near-real time images. Five python scripts have been written to produce hotspots for vector amplification and potential RVF epizootic, and the scripts were imported into ArcToolBox. Based on the definition of dry spell given in the previous sections, one python script has been written and imported into ArcToolbox, to identify dry spell events during the wet season. Recently, both the climate-based model and the dry spell model have been implemented in GEE to generate updated, near-real time RVF risk maps. FAO is focusing on this activity beyond the Vmerge project. The application demo is shown in Figure 30. The demo is currently available for the GEE community.
One of the major concerns of the Vmerge project was to strengthen interactions between research and surveillance actors, to ensure translation of the scientific productions into policy, guidelines and manuals to disseminate and update Vmerge-generated knowledge.
The analysis of EU legislation in force for RVF and CBD and its consistency with the epidemiology of the diseases have been performed as well the national contingency plans in place in some southern EU countries have been assessed in collaboration with the Mediterranean Animal Health Network – REseau MEditerranéen de Santé Animale as REMESA. To assess the existing surveillance systems for RVF, a specific questionnaire has been designed and drafted aiming at collecting information on the surveillance systems in place in some Northern African countries. An important preparatory work has been done with through the release of a systematic review on published compartmental models on RVF transmission dynamics.
Furthermore, a preliminary spatial analysis to track the climatic events able to favour RVF occurrence over North Africa was conducted. A methodology to track the climatic events that can favour RVF risk over North Africa has been identified (Cyril Caminade et al., 2014): a rainless period lasting at least for a week, followed by large precipitation (at least above 10mm) at the end of rainy season might favour the risk in triggering RVF outbreaks. High numbers of these events, called embryogenesis risk events, occurred in Mauritania in the years of RVF epidemics (Cyril Caminade et al., 2014). This approach has been applied to the whole Africa (the Maghreb included) for the years 2008 – 2015 (these maps were based on data collection from 2008 to 2015 and further analysis of the climatic events that might have favoured the risk to trigger RVF outbreaks in Mauritania in 1998, 2003, 2010 and 2012 from previous studies). The result was the production of yearly maps reporting the sum of embryogenesis events and maps with the embryogenesis events constrained to animal presence (population density greater than 1/km2). These findings allowed the identification of geographic areas and periods of the year more at risk for RVF persistence in case of introduction, thus representing a support for planning early warning activities. These maps should be coupled with pragmatic suggestions for early detection of RFV and other vector-borne diseases mainly in unaffected North African countries by implementing risk-based surveillance approaches to have reliable data at the least cost possible (simple random selection vs convenience sampling). These results can be used for the planning of risk-based surveillance. Linked to this activity, with respect to the assessment of the risk of RVF and Culicoides-borne infections spread from sub-Saharan countries, three targeted areas (in correspondence with oases) were defined in Morocco, serum samples were taken from 3 sentinel sheep flocks (one each oasis) containing 15 sentinel animals each. Vector abundance dynamics for Culicoides and for mosquitoes were assessed by trapping activities. Traps used were of the blacklight type for Culicoides and Carbon dioxide traps for mosquitoes. Tests were carried out by Biopharma in collaboration with IZSAM. At the end of the project, in November 2016, laboratory findings were still in progress and the activities on vectors were still ongoing because the peak of the vector season was not finished yet.
All these activities are functional to the development of proposals for risk-based surveillance programs to be activated in at risk countries. Several workshop and training sessions were carried out.
Many capacity building activities were carried out within Vmerge, to develop capacities in Vmerge beneficiary countries and veterinary services.
Training and capacity building to Vmerge beneficiaries. The first training course consisted of two parts: an entomological part that focused on how to capture and identify mosquitoes, and a second part that focused on how to use the VECMAP mobile app to record data in the field. This face-to-face course was organized at three different locations: Rabat (period: 3-7 June 2014; language: French), Rosso (period: 9-13 June 2014; language: French) and Alexandria (period: 11-15 May 2015; language: English). Afterwards, continuing support was provided to the partners to select sampling sites and finalize fieldwork logistics (mobile app, materials, etc.).
The second VECMAP course provided an initiation to species distribution modelling using VECMAP Lite software (http://www.vecmap.com/vecmap-lite-version). Preliminary to this course from March to April 2016, a distance learning training on geographic information system was provided to the project partners, which could not attend the face-to-face GIS workshop. From the 9th to the 13th of May 2016, the 2nd VECMAP course was delivered face-to-face in Antwerp (Belgium). The partners from Morocco (partners #12), Tunisia (partners #15), Mauritania (partners #16), and Senegal (partners #13) participated to this training (Figure 26). This workshop was very practical focusing on mapping and modelling using three techniques implemented in VECMAP Lite, i.e. Non-Linear Discriminant Analysis, Generalised Linear Model and Random Forest. The first day was launched by a general presentation of VECMAP Lite and the datasets provided for the different exercises. Those datasets included two types of response presence/absence datasets: a subset of a real dataset on Anopheles labranchiae and different virtual vector datasets (benchmark spatial test dataset) geographically localised in the four countries of the participants and created specifically for this workshop. Those datasets were created with the goal to provide a better understanding on the impact of sample size and prevalence on the vectors distribution models. In addition, the participants received 110 environmental raster layers and were introduced to satellite imagery and spatial resolution. The second and third days were dedicated to theoretical and practical presentations of the three spatial modelling techniques mentioned above. Using the datasets provided, the participants gain experience in the modelling techniques, and they reach a better understanding of the challenges related to modelling process and the advantages and disadvantages of each technique. For the two last days of the workshop, the participants could handle the modelling techniques on the virtual vector datasets and on their own datasets and generate risk map showing the probability of vector presence.
Training and capacity building to veterinary services. Following training needs assessment in the region (largely through REMESA) and considering the already-planned trainings and workshops for the region (by FAO or USDA) it was agreed that capacity building focused on veterinary services in order to improve preparedness and disease reporting (or syndrome reporting) from the field. Within the capacity buildings framework, the following activities have been conducted: a training course on epidemiology was carried out in Tunis from 8 to 10 October 2015, back to back with a REMESA meeting. The training focused on the preparedness of at-risk countries in Northern Africa, namely the development of contingency plans and surveillance for early detection (see next section). FAO together with other partners (IZSAM and CReSA) conducted the training in French. The targeted group was mostly REMESA network, including all REMESA focal points (Mauritania, Tunisia, Morocco and Algeria). Participants from Egypt and Senegal attended the training. Although not planned in the principle, a course on “Introduction to Risk Analysis” was carried out in Egypt, Cairo from 23 to 25 February 2016.
By accessing stakeholders across various institutional levels, the greater understanding of vector ecology and vector-borne diseases attained through Vmerge will allow for the development of better prevention and control strategies, as well as tools to fight the disease in affected and at risk countries. Furthermore, the dissemination of data within the project consortium will lead to more robust and structured research work. The path of transferring knowledge generated through academia into practical awareness via policy change is not a straight-forward or short-term venture. To disseminate the achievements of Vmerge at the country level, a variety of trainings, manuals, guidelines, and other tools have been adapted into realistic strategies and epidemiological scenarios in the different countries via 1) mapping of projects, 2) translating results into policy, 3) disseminating results.
FAO has compiled and categorized past, ongoing and future projects on Rift Valley fever (RVF) and the involved vectors into one database. The database was developed as an excel file, where projects were classified by category in separate spreadsheets. The focus of attention lies on projects from around the year 2000 until the planned future ones. This RVF database was completed by sharing it with project partners and external experts/institutions in June 2014. The final file was published online on both the FAO and Vmerge websites (http://www.fao.org/ag/againfo/programmes/en/empres/news_050614.html). FAO released the third version of this dynamic database in July 2016, which was updated after sharing with the RVF community created from the stakeholder list (~ 300 members).
For details on dissemination actions see the ‘potential impact’ section below.
Potential Impact:
Consolidation of collaborations among European and African partners improves the international integration of research on important aspects of the epidemiology of pathogens burdening human and herd health in Africa and in Europe.
We produced a better understanding of the disease emergence process for selected viruses at the level of vector populations and their ecosystem, including the development of innovative methods and tools for pathogen detection, and disease surveillance. Final results on vector competence would have impact in both pure and applied fields. Collectively, the findings of the “cofactors of competence” programme has substantially increased our understanding of how common variations in the environmental conditions experienced by insect vectors might result in differences between estimates of competence achieved under normal laboratory conditions and those exhibited by vector populations in the field. This suggested in turn ways to improve the design of laboratory vector competence studies as well as the extent to which the estimates obtained during previous studies might require adjustment. More accurate measures of the competence of field populations of vectors have increased our ability to predict the potential for establishment and spread of exotic arboviruses in Europe. We expect from transmission modelling risk maps for Rift Valley fever (RVF) and Culicoides-borne diseases (CBD) in Europe and northern Africa and a forecast model for RVF set up and formalized to provide near real-time predictions for regions at risk of RVF virus occurrence, thus forming the basis for an early-warning system for RVF and CBD in Sahelian, northern Africa and Europe.
Up to now, 20 papers have been published or are under the revision process. More than 30 talks were given by Vmerge partners. Many other papers are in preparation. So, these research and experimentation results will continued to be disseminated in the scientific community through both Vmerge websites in the aim to disseminate data and information to sister projects and the wider public, thereby drawing attention to the project itself.
Moreover, one of the major Vmerge aim was to translate research results to policy to contribute to setting up the surveillance methods and tools. We thus expected to provide national veterinary services with practical recommendations and standardized methods and tools for surveillance of vector-borne diseases, and to provide international organizations with science-based proposals to improve control of vector-borne diseases, resulting in safer livestock trade and better food security, as well as better protection of public health.
In this context, capacity building activities have been planned and performed for veterinary services: a first training on epidemiology was held in Tunisia from 8 to 10 October 2015 and a risk analysis course was held in Egypt from 23 to 25 February 2016. The two-day training in Tunis was organized by the Food and Agriculture Organization of the United Nations (FAO) and focused on preparedness for early detection and contingency planning of RVF. The training was attended by 30 participants, including all REMESA (REseau MEditerranéen de Santé Animale) network focal points from Algeria, Mauritania, Morocco, and Tunisia. REMESA was created in 2009 to generate a common framework for work and cooperation, having the necessary capabilities to assist and coordinate the development and implementation of animal health regional projects and programs. The specific objective of REMESA is the improvement of prevention and control against the major transboundary animal diseases and zoonoses through the strengthening of the national and regional resources and capacities, and the harmonization and coordination of surveillance and control activities. Also present was one representative from both the veterinary services of Senegal and the OIE in Tunis. The trainers came from Vmerge partner institutions, namely the FAO in collaboration with the Centre for Agricultural Research for Development (CIRAD) in France, the Centre de Recerca en Sanitat Animal (CReSA) in Spain, and the Istituto Zooprofilattico Sperimentale dell'Abruzzo e del Molise "Giuseppe Caporale" (IZSAM) in Italy.
This training, highly participative, covered the topics of RVF surveillance, disease recognition, establishing an early warning system, and contingency plan development. Participants had the opportunity to learn from the experience of both Mauritania and Senegal in their responses to RVF outbreaks and ongoing surveillance, prevention, and control programs. A desktop simulation exercise on RVF was also organized.
At the end of training, participants expressed need for:
(i) the development of contingency plans for RVF;
(ii) clear guidance on the vaccination strategy;
(iii) the availability of vaccines;
(iv) capacity building for conducting risk modelling to determine the main risk areas (Tunisia, Morocco, and Algeria).
FAO also reminded its commitment in assisting countries in disease surveillance. In this regard, the development of a RVF surveillance manual is underway.
A manual was produced to provide professional and para-professionals with information needed to design and implement effective animal health surveillance for RVF in a One Health context. The manual emphasizes a risk-based approach to surveillance and utilizes the principles outlined in the OIE Guide to Terrestrial Animal Health Surveillance for designing and implementing surveillance programs. The statements in the manual are intended to provide guidance and should not be treated as prescriptions. Appropriate surveillance procedures at the national level depend on the local epidemiological conditions, production systems, culture and institutional arrangements.
Currently FAO has a manual on Preparation for RVF Contingency Plans published in 2002 that provides information on the nature of RVF and the principles and strategic options for its prevention, control and elimination. However, with the increased risk of vector spread into new regions, the importance of surveillance cannot be over emphasised. Furthermore, the lessons learnt from RVF outbreaks since the manual’s publication in 2002 and the emergence of One Health both signify the need for an updated guide. The manual has the goal of disseminating prevention, surveillance, and control knowledge to all regions susceptible to RVFV (East, West, and North Africa plus the Middle East). By increasing the resilience of livelihoods to threats and crises, its goals were based upon the successful fulfilment of the following priorities:
- Priority 1 - Improving the capacity to formulate and promote risk reduction and crisis management policies, strategies and plans;
- Priority 2 – Setting up and/or improving the mechanisms which identify and monitor threats and assess risks, and to deliver integrated and timely early warning; and
- Priority 3 - Capacity development to support institutions at national and regional levels.
The purpose of the manual is to provide professional and para-professionals with information needed to design and implement effective animal health surveillance for RVF in a One Health context. As a contextual foundation for surveillance, the manual provides general information on the disease and the causative agent, RVF virus. The background data stresses epidemiology and determinants of risk.
The manual suggests appropriate surveillance objectives and activities for a range of countries according to their epidemiological status. The goal is to aid countries to develop risk-based surveillance systems that are fit-to-purpose and cost effective. In line with lessons learnt during recent outbreaks, the manual highly recommends the use of participatory epidemiology tools to engage all stakeholders to achieve improved disease surveillance. Participatory surveillance is an evidenced-based approach to surveillance that involves diverse stakeholder groups to identify problems and formulate solutions. The approach taps in to traditional knowledge systems and networks. The public often have local language names for common disease problems and are aware of when and where diseases are occurring. This knowledge frequently includes environmental, clinical and epidemiological characteristics of the disease. Participatory surveillance is action-oriented approach for the collection of epidemiological intelligence to inform decision-making. These approaches rely on cooperation among farmers, animal and human health care providers, and governments.
Given the zoonotic and vector-borne nature of the disease, RVF surveillance is ideally approached from a One Health perspective. Surveillance systems need to cover a range of human, animal and environmental issues and the risk of an RVF outbreak can only be assessed in a trans-disciplinary manner. Although the manual focuses on animal health surveillance, it attempts to do so in the context of an integrated surveillance and response system.
To be accessible to a large target audience (ex: both national level veterinarians and para-vets), it was decided that the wording and length should not be a barrier, and the scope broad (while remaining within the focus of surveillance). The manual is approximately 40 pages and contains the following sections (see attachment).
Proper dissemination of the manual is crucial to its relevance in combatting RVF. Initially, the manual will be available in only an electronic version, and will be translated from English into French, Swahili, Portuguese, and Arabic. The primary channel will be through the FAO website in the FAO Document Repository where all FAO public documents are available. A database to directly contact RVF stakeholders has also been created in order to distribute the RVF Newsletter. The Stakeholder database contains approximately 300 members, including all VMERGE partners, as well as individuals from various types of institutions, including research, international organisations, NGOs, governments, the private sector, and private individuals working directly on RVF or in RVF affected regions. After subsequent distribution of the newsletters, requests from colleagues of listserv members receiving the emails has kept the database updated and constantly expanding. Lastly, the manual will also be available on the Vmerge website. At this time, FAO is still exploring the need for a printed version of the manual.
A workshop on RVF concerning its transfer, modelling, forecasting, as well as animal mobility and vector surveillance was conducted as part of the project activities of the REMESA on the 6th October 2016 Tunis, Tunisia. A policy brief document “Enhancing Surveillance and Early Detection of RVF” was written based on the presentations and followed discussions. This policy brief recognized that differences in RVF risk of infection occur between (and within) infected and non-infected countries in the region. Furthermore, those countries are of different levels in managing the risk of RVR infection or incursion. For all of these issues, the work of Vmerge applied an approach in which each national member of REMESA is encouraged to develop a national RVF surveillance strategy that is supportive of the greater regional effort. It is not a top –down prescribed approach but a though mapping exercise on how to build a RVF taskforce in the REMESA region to adapt and mitigate the risk of RVF in a sustainable manner. The real work should begin in depth to start understanding all the outlines of the vector-borne animal diseases in the region, which has been greatly influenced under Vmerge.
The connectedness and interdependence of animal and human health demand full embrace of the One Health concept as a strategy to prevent the emergence and global spread of vector-borne pathogens. Any sustainable approach must include building relevant capacity, both human and technical, and implementing intervention programs at the source of potential infection and epidemics. Any decisions for dealing with these emerging diseases should take into account the major causes (rather than just the consequences) and also the types of re-emergence in question. There can be no single response. The purpose of this policy brief was to highlight recent advances in the fields of epidemiology, detection, and the prevention of diseases transmitted by arthropod vectors.
The design of control strategies for vector–borne diseases should be guided by research into emergence mechanisms in order to understand how a wild cycle can produce a pathogen that goes on to cause devastating urban epidemics. In addition to studying the emergence mechanisms, combining field work with basic research, we must also: improve surveillance techniques by developing tools for rapid pathogen detection; find alternative means of prevention by testing vaccines against arthropod bites; and propose new control strategies, without neglecting to improve existing anti-vector control techniques. Answers to such types of question are essential for gauging the present risks of disease, predicting future threats, and formulating methods for monitoring and control. Compared with many other diseases, the ever increasing threat of vector-borne diseases represents a great challenge to public and animal health managers. Complex life cycles, changing distribution ranges, a variety of potential vectors and hosts, and the possible role of reservoirs make surveillance for vector-borne diseases a grave concern in a changing environment with increasing technical constraints. Surveillance activities may have various specific objectives and may focus on clinical disease, pathogens, vectors, hosts and reservoirs, but ultimately such activities should improve our ability in a timely manner to predict, prevent and control the diseases of concern.
This surveillance therefore becomes a serious issue for local and national organizations and is being conducted more and more at the regional and international level through multidisciplinary networks. With financial and logistical constraints, tools for optimizing and evaluating the performance of surveillance systems are essential. The continuous development of mapping, analytical and modelling tools provides with an enhanced ability to interpret, visualize and communicate surveillance results. This also demonstrates the importance of the link between surveillance and research, with interactions and benefits in both directions.
Passive surveillance. The Purpose of RVF surveillance is intended for use in outbreak management, to inform disease control measures, to inform trade policy, and to prioritize further surveillance activities. The specific purpose of the surveillance system as a whole, and of each activity, should be carefully considered and expressly stated. The aims of surveillance for VBD may include one or more of the following:
- Early detection of outbreaks;
- Detection of cases to enable the timely implementation of control measures;
- Detection of vector-borne pathogens in hosts and /or vectors;
- Describing the distribution and level of disease or pathogen circulation;
- Monitoring space-time trends in levels of disease or pathogen circulation;
- Determining the geographic distribution of known or potential vector populations; and
- Mapping transmission risk.
In addition, surveillance activities may also contribute to improving our knowledge of the disease and to help inform and refine further surveillance by:
- identifying risk factors and determinants of pathogen occurrence and transmission;
- generating hypotheses regarding potential pathogen reservoirs or vectors; and
- identifying risk factors for the emergence in non –endemic areas.
Depending on the objectives of a surveillance program, data may be collected about one or more of the following parameters:
- Seroprevalence in host populations, including seasonal and annual changes;
- infection prevalence in hosts (and if warranted in vectors as well);
- Species composition, abundance and temporal/spatial variation in vector populations;
- Density of hosts and spatial and temporal variation in those populations;
- Incidence of clinical disease, mortality or other health –related outcomes in host populations;
- Presence and level of potential risk factors for disease, related to the agent, host, vector or environment.
Early warning surveillance. This is an important component of surveillance for RVF and it is widely used, since it potentially allows for the detection of multiple known and unknown emerging diseases. Early warning surveillance often forms part of comprehensive early warning contingency plans that also include many other components.
Sentinel surveillance. Sentinel surveillance employs repeated sampling subjects at selected sites, which act as proxies for the entire population. The sites may be randomly selected or located in areas where the probability of disease detection is increased (risk-based sampling). To detect recent seroconversion, young animals, free of maternal antibodies, are sampled at weekly to monthly intervals. Sentinel surveillance can also be deployed at a regional scale as demonstrated in West Africa for RVF, where a network of sentinel small ruminant herds was located in potential high-risk areas for RVF in Mauritania and Senegal.
Risk-based surveillance. Risk-based surveillance aims to increase the efficiency and cost-benefit ratio of surveillance by employing risk assessment methods to prioritize surveillance requirements in either the sampling or analysis of the surveillance data. Risk-based sampling methods are built into most surveillance systems for RVF, particularly those aimed at the early detection of disease, since preferentially sampling those strata of the population, or certain geographic areas, considered being at the greatest risk of disease occurrence increases the likelihood of early detection, and improves the cost-effectiveness of the system.
The risk assessment may be based on expert opinion; the presence of known risk factors and the outputs of qualitative or quantitative risk analysis that could be used to inform risk-based surveillance is an assessment of the risk of transporting pathogen. It could include for example some approaches to improve a risk-based surveillance system by selecting relevant animal movements for risk-based sampling. This significantly increased the sensitivity of the surveillance, when compared to completely random sampling.
Spatial analysis and modelling. Advances in mapping and geographic information system have made it easier to communicate surveillance data in various visual formats to aid understanding and facilitate decision-making and have also allowed more complex spatial and spatio-temporal analysis to identify risk factors for RVF occurrence (or prediction). Identifying high-risk areas is one of the important outputs of a surveillance programs that can then be fed into the planning and implementation of prevention and control measures, as well as used to refine further risk-based surveillance activities.
Information on vector distribution can be compiled into risk maps, indicating potential high-risk areas for disease. These risk maps can be further refined by incorporating data on other known risk factors, as well as previous disease outbreak data. A spatial risk model is used to estimate vector presence or abundance, or RVF presence or incidence, in a particular geographic area, with the outputs generally displayed in map format, as a risk surface.
Evaluation of surveillance systems. The development of new methods using up-to-date tools and scientific knowledge is crucial for RVF surveillance. We have to take into consideration that specific local conditions, rapid economic development and a plethora of management options in a financially restricted environment result in animal health managers facing situations where the wrong decisions could lead to hazardous consequences.
Surveillance and research. Surveillance is by nature interdisciplinary, hence collaborative networks and systems must be developed to facilitate the planning and co-ordination of surveillance activities, the integration of surveillance and research, efficient data management and analysis, and the communication of outputs to decision-makers and other stakeholders. It is also crucial to increase surveillance activities in regions where new vector-borne diseases are more likely to emerge.
Finally, dissemination actions were organized by FAO. These actions communicated project outcomes that are judged to be of interest to stakeholders outside the consortium, including disease updates, early warning messages, reports, scientific publications including those concerning Mauritania or Senegalese studies, training materials, manuals, guidelines, announcements of Vmerge project events and their outcomes, outbreaks (or their updates) including the 2015 Mauritania outbreak, and climate maps.
Audience. Stakeholder database contains approximately 300 members, including all VMERGE partners. They were emailed the newsletters using the FAO administered account rvf@fao.org. The database contains individuals from various types of institutions, including research, international organizations, NGOs, governments, the private sector, and private individuals working directly on RVF or in RVF affected regions. After subsequent distribution of the newsletters, requests from colleagues of listserv members receiving the emails kept the database updated and constantly expanding. The database may be used in the future to quickly access RVF stakeholders (see attachment).
Review process. Newsletters are timed according to the available information that needs to be shared. They are often timed around alerts or new RVF outbreaks (as reported by ProMED-mail or OIE) or when Vmerge partners requested specific news to be disseminated. Accompanying but less time sensitive RVF related news were then sourced, such as recent publications, before all topics of the newsletter were summarized and distributed to the Vmerge consortium for further input. After Vmerge partners provided input and or further news to share the newsletter was disseminated.
Route of dissemination. Stakeholders were sent the final newsletter in HTML format, which was subsequently advertised via EMPRES, FAO’s VPH newsgroup, and finally posted on the Vmerge website’s ‘publications’ section (http://www.vmerge.eu/publications/rvf-newsletters).
List of past newsletter items:
- FAO compiles past, ongoing, and future projects on Rift Valley fever and its vectors
- RVF training with REMESA network
- RVF human cases in Mauritania
- Euro-Aegis has produced a virtual species distribution map (SDM)
- Summary of climate update (19 Oct 2015)
- African climate forecast (5 Nov 2015)
- FAO and its partners are putting an effort together to improve Rift Valley fever preparedness in the Maghreb
- First RVF outbreaks in Mauritania reported to the OIE
- Update on FAO activities on RVF in East Africa
- Analysis of RVF early warning systems in place in the EU
- Africa climate forecast (15 Nov 2015)
- Rift Valley fever newsletter provides outbreak, forecasting, and project updates on RVF
- FAO helps countries prevent and control RVF
- RVF risk factors in central Sudan
- EMRPES Watch: Africa – El Nino and increased risk of Rift Valley fever – Warning to countries
- USDA Emerging Health Risk Notification, 20 Dec 2015. El Nino and Rift Valley fever, East Africa
- Climate update for RVF in East Africa (8 January 2016)
- RVF update for East Africa prepared by NASA, including rainfall hotspots, vegetation anomalies, and the RVF risk map (5 Feb 2016)
- Serological evidence of exposure to Rift Valley fever virus detected in Tunisia
- RVF update for East Africa prepared by NASA, including rainfall hotspots, veg anomalies, and the RVF risk map
- Chimpanzee Adenovirus vaccine provides multispecies protection against Rift Valley fever
- Climate update for RVF in East Africa (4 March 2016)
- Media reported RVF disease event in Uganda
- RVF paper: Current challenges and future prospects
- FAO compiles past, ongoing, and future projects on Rift Valley fever and its vectors
- OIE confirmed RVF event in Ugandan goat
- Rift Valley fever in kidney transplant recipient returning from Mali
- Complete genome sequence of Rift Valley fever virus strain Lunyo
- Geographically weighted discriminant analysis of environmental conditions associated with Rift Valley fever outbreaks in South Africa
- Culex pipiens and Stegomyia albopicta (=Aedes albopictus) populations as vectors for lineage 1 and 2 West Nile virus in Europe
- Possible Human RVF Case in Uganda
- Risk or Loss Function and Quality Measures for Rift Valley Fever Vaccine Production
- China confirms human case of RVF imported from Angola
- FAO updates RVF mapping tool
- Summary of RVF publications released as of June 2016
- RVF event in Niger
- Summary of RVF related publications released since August 2016
- Bulletin Epidémiologique sur la Fièvre de la Vallée du Rift : WP 6.5 Vmerge Project
- Two highlighted papers focused on RVF
- Likely spread of RVF outbreak from Niger to Mali
- Final Vmerge meeting
- 13th REMESA Joint Permanent Committee Meeting
- RVF risk maps
- Summary of presentations given at the final Vmerge meeting held
- Update on RVF Outbreak in Niger and Mali
List of Websites:
www.vmerge.eu
Coordinator: Thomas Balenghien, thomas.balenghien@cirad.fr
In the field, intense sampling campaigns (WP1-WP3) have been undertaken by the Vmerge consortium in 2014 and in 2015, allowing the collection of mosquitoes, sera and environmental data. In laboratory, molecular and serological diagnostic tools have been developed and validated (D1.1 and D1.2). The use of these assays on the field-collected samples has provided a wealth of data on arbovirus prevalence, especially on RVFV prevalence, in the targeted regions. Coupled to a metagenomics approach, well-known pathogenic arboviruses and new viruses with a potential interest for human and livestock health have been identified (D1.3 and D1.4). Beyond known viruses, a huge amount of new viruses have been identified, significantly increasing current knowledge on virus diversity. Finally, the microbiomes of several mosquito vectors have been analysed, leading to the first detailed characterisation of the microbiome composition of an important vector (D1.5). Despite the long process for having authorizations to work at high levels of biosecurity and containment, Dutch and Spanish populations of Culex pipiens were found to be capable of transmitting RVFV (D2.1). Generally, the proportion infected increased with dose (D2.3) and this proportion continued to increase between days 14 and 28 post-infection, but no substantial effect of the different temperature regimes tested was observed (D2.2). Coinfection with Brugia nematodes did not increase infection rate (D2.5). Results also suggested that coinfection of Culicoides midges with multiple strains of bluetongue virus (BTV) resulted in reassortment confirming in vitro results (D2.4) and that the MP-12 strain of RVFV may be a promising vaccine candidate for use in camelid livestock in Europe (D2.6).
In silico, Vmerge successfully developed two approaches to assess the transmission risk of RVFV in endemic areas of Western Africa: a mechanistic model using field data as inputs and a climate-based modelling approach (D3.3 and D4.2). RVFV risk maps for up-to-now free areas (Northern Africa and Europe) were produced using results on mosquito abundance, but the figure is not substantially different from the risk maps created before (D3.1). In-field studies on the existing pathways of animal trade of importance for RVF introduction and spread in Egypt have been designed carried out (D4.3). A genetic module for RVF between EMPRES-i and Virus Pathogen Resource (ViPR) was developed (D4.7). Furthermore, a preliminary spatial analysis to track the climatic events able to favour RVF occurrence over North Africa was conducted (D4.4). For Culicoides-borne viruses, Vmerge developed a complete framework to: 1) confirm that opposition between Mediterranean and non-Mediterranean climates are an important change driver in Culicoides communities, 2) describe the structure of Culicoides imicola at different spatial scales and develop new genetic tools (D3.5) 3) characterize spatial and temporal overlap of wild and domestic hosts and various Culicoides species (D3.4) and 4) produce European risk maps for BTV and SBV (D3.2). To facilitate an integrative approach between all partners, different services were provided including access to quality data, optimization of sampling design for collection campaigns (D5.1 D5.4 and D5.2) training in spatial modelling, modelling software, and assistance to upscale regional vector distribution models (D5.3 and D5.6). Research outputs have included upgrading modelling software (D5.9 and D5.8) spatial distribution models for Rift Valley fever vectors, livestock movement models (D5.7) and hotspot framework overlapping hosts and Culicoides (D5.5). Automated risk map scripts were produced to ensure regular updated maps (D5.10). Dissemination included the maintenance of a website to promulgate information to sister projects and public.
One of the major concerns of the Vmerge project was to strengthen interactions between research and surveillance actors, to ensure translation of the scientific productions into policy, guidelines and manuals. The analysis of EU legislation in force for RVF and CBD and its consistency with the epidemiology of the diseases have been performed as well the national contingency plans in place in some southern EU countries have been assessed in collaboration with the Mediterranean Animal Health Network (D4.1 and D4.5). The path of transferring knowledge generated through academia into practical awareness via policy change is not a straight-forward or short-term venture. To disseminate the achievements of Vmerge at the country level, a variety of trainings, manuals, guidelines, and other tools have been adapted into realistic strategies and epidemiological scenarios in the different countries via 1) mapping of projects (D6.1) 2) translating results into policy (D6.2 and D6.3) 3) disseminating results to stakeholders across various institutional levels (D6.4 and D6.5).
Project Context and Objectives:
In the last decades, world was facing the emergence and the recrudescence of vector-borne diseases due to current and predicted environmental and socio-economic changes. In a retrospective analysis, Jones et al (2008) have shown that, while emerging-infectious disease events were most often detected in Europe or North America, the major risks are in or originate from southern countries. Indeed, many demographic, environmental, social and economic factors favouring the emergence of VBD are faced by these countries which only benefit from limited health facilities to prevent, monitor or control their spread. Therefore, understanding and controlling VBD is a global public good. European research into VBD should be seen as a tool for protecting and improving animal health and economics, public health and the welfare of the European economies and their citizens, as part of a coordinated, international effort at the global scale.
The majority of transboundary animal diseases are of viral origin. Nonetheless, the World Organization for Animal Health (OIE) has listed 82 notifiable diseases of major sanitary and economic importance for farm animals of which 43 are being caused by a virus and 20 are vector borne (11 arboviroses). Recent epizootics of bluetongue and Schmallenberg viruses in Europe and of Rift Valley fever virus outside Africa showed that viral vector-borne diseases have the potential to emerge and spread rapidly at remote distances, and are difficult to control. In this context, we need to develop a better understanding of the complex relations between vectors, viruses and their hosts at different scales (individual, vector population, landscape, and ecosystem). Moreover, the availability of predictive models and relevant scenarios (environmental changes, control strategies...) are essential for the development of adequate prevention and contingency programs. We need then to anticipate their establishment and spread by developing specific research programs and innovative surveillance and control strategies that can be translated into tools that can be quickly brought into action for a better cost-efficient control of vector-borne diseases, including those possibly still unknown.
These diseases have considerable impacts on animal farming as an important economic activity in Europe, but also as an important subsistence resource for more than 70% of poor people living in developing countries. Almost 2/3 of world farm animal populations are reared in developing countries (source FAO stat). They represent an essential source of proteins and contribute to the economic development of these countries and overall to food security. The demand in animal products in developed countries is decreasing while in developing countries it is growing by almost 3% per year (source World Bank).
Production losses resulting from animal diseases are estimated to be 20% in the world on average with a major impact in developing countries. For instance, the ban in Ethiopian livestock exports to Yemen and Saudi Arabia after the Rift Valley fever epizootics and epidemics in the Horn of Africa and the Middle East resulted in an estimated GDP loss of 25% in the Somali province of Ethiopia.
In a very different context, the net costs of the bluetongue virus-serotype 8 epizootic of 2007 was valued at 164–175 million Euros for the single case of the Netherlands, based on the output of a deterministic economic model. The cost to the EU as a whole was obviously substantially higher, but would have been even greater if there had not been an available body of expertise derived from EU and national research programs.
The Vmerge proposal relies on the ‘One Health’ concept, first suggested in 2004 by the Wildlife Conservation Society to strengthen links between human health, wild and domestic animal health and management of the environment, particularly biodiversity and ecosystem services. In 2008, a reference framework based on this concept was drawn up by six leading international organisations – World Health Organisation (WHO), Food and Agriculture Organisation (FAO), World animal health organisation (OIE), United Nations Children's Fund (UNICEF), the United Nations System Influenza Coordination (UNSIC), and the World Bank. In April 2010, FAO, OIE and WHO reiterated the importance and usefulness of this approach in a tripartite concept note on ‘Sharing responsibilities and coordinating global activities to address health risks at the animal-human-ecosystems interfaces’. The vision developed is: “A world capable of preventing, detecting, containing, eliminating, and responding to animal and public health risks attributable to zoonosis and animal diseases with an impact on food security through multi-sectorial cooperation and strong partnerships”. This approach was confirmed and developed at policy level by the G20 meeting of Ministers of Agriculture held in Paris (22-23 June 2011): “As far as public health, animal health are concerned, we stress the importance of strengthening international and regional networks, international standard setting taking into account national and regional differences, information, surveillance and traceability systems, good governance and official services, since they ensure an early detection and a rapid response to biological threats, facilitate threats, facilitate trade flows and contribute to global food security”.
The overall objective is to address the risk of introduction, emergence and spread of vector-borne viruses associated with mosquitoes (Diptera, Culicidae) and Culicoides biting midges (Diptera, Ceratopognidae), i.e. mainly Rift Valley fever, bluetongue and Schmallenberg viruses but much broader information is expected from field sampling of insect and host populations. We want to improve our currently limited understanding of these emerging vector-borne diseases and their potential for spread throughout northern Africa and Europe, and enhance epidemiological surveillance strategies and tools for better disease detection.
The specific objectives are:
• to provide innovations in diagnostic procedures for more reliable cost effective and quicker identification of emerging vector-borne diseases for Europe, northern Africa and the Sahelian region;
• to improve our understanding of vector competence of insect populations in ecosystems at risk of emerging vector-borne diseases;
• to model vector population dynamics and virus transmission at a fine spatial and temporal resolution, and upscale the models for better assessment of emergence and spread capacities of selected vector-borne diseases throughout the range of ecosystems;
• to assess the existing surveillance networks for emerging vector-borne diseases and propose new surveillance strategies in Europe, northern Africa and the Sahel.
Vmerge was designed to be integrative and comprehensive and embraces the entire knowledge acquisition chain from high quality research and experimentation to identification of control and surveillance methods and tools, and provision of sustainability and capacity building. The objectives were, following the identification of viruses with emerging potential, to develop experimental models on animals, and then, based on transmission models, to develop new tools for control and surveillance such as on line predictive and preventive risk mapping. Ensuring this continuum requires a strong partnership between research teams and labs from both North and South of the Mediterranean. Southern research teams and labs are from low income countries (Mauritania, Senegal and Tunisia) where diseases represent a major threat to Europe They carried out experimentation and contribute to setting up the surveillance methods and tools. Moreover, the participation of FAO to the project brings an institutional dimension which added international authority to effectively adopt the surveillance tools that Vmerge were identified.
The coordination and management structure was organised as represented in the chart in attachment.
Vmerge consortium was made of 16 beneficiaries from 12 countries (see figure below), including 5 international cooperation partner countries: Senegal, Mauritania, Morocco, Tunisia, and Egypt. It encompassed most of European and Mediterranean ecosystems, plus Mauritania and Senegal making the eco-epidemiological link with Sahelian Africa. Vmerge intended to establish a continuum between field and experimental research, and vector/vector-borne disease surveillance through a logical framework linking innovative diagnostics, prospective surveys in vector populations and their hosts sampled in selected ecosystems. This aimed to the identification of emerging risks of VBD for Europe and the Mediterranean Basin, and the implementation of improved surveillance strategies in this region, as well as in the Sahelian region of Africa, through a reinforced North-South partnership.
The Vmerge beneficiaries were:
Partner 1. Centre de coopération internationale en recherche agronomique pour le développement (CIRAD), Montpellier, France
Partner 2. Centre de Recerca en Sanitat Animal (CReSA), Spain
Partner 3. The Pirbright Institute (PIR), formerly Institute of Animal Health, UK
Partner 4. National Environment and Research Council, Centre for Ecology and Hydrology (NERC-CEH), UK
Partner 5. Agence nationale de sécurité sanitaire, de l’alimentation, de l’environnement et du travail (ANSES), France
Partner 6. Friedrich-Loeffler Institut (FLI), Germany
Partner 7. Euro-AEGIS, Belgium
Partner 8. Central Veterinary Institute of Wageningen University and Research Centre (CVI-WUR), The Netherlands
Partner 9. Food and Agriculture Organisation of the United Nations (FAO), Roma (Italy) FAO sub-regional office in Tunis (Tunisia)
Partner 10. Pathoquest, Paris, France
Partner 11. Istituto Zooprofilattico Sperimentale dell’Abruzzo e Molise “G. Caporale” (ICT), Teramo, Italy
Partner 12. Institut Agronomique et Vétérinaire Hassan II (IAV), Rabat, Morocco,
Partner 13. Institut sénégalais de recherché agricoles (ISRA), Dakar, Senegal,
Partner 14. Faculty of Veterinary Medicine-Alexandria University (FVM-AU), Alexandia, Egypt
Partner 15. Institut de recherche vétérinaire de Tunis (IRVT), Tunis,Tunisia
Partner 16. Centre national d’élevage et de recherches vétérinaires (CNERV), Nouakchott, Mauritania
Project Results:
During the Vmerge project, many efforts have been done to ensure a collaborative and integrative approach between all European and African partners. In WP1, partners have defined several projects structuring sampling designs beyond classical surveillance and now follow a common sampling strategy. The WP2 was mostly organized as an integrated international study to measure the competence of European Culex pipiens, the common house mosquito, for Rift Valley fever (RVF) virus. Joint protocols for mosquito infections have been discussed and agreed between partners. In WP3 and WP4, technical meetings were organized on ‘Spatio-temporal methods in RVF risk modelling and their integration’ to compare risk modelling approaches as well as facilitate data sharing among the project partners, and on ‘Animal movement and spread of RVF’ to discuss new methods to simulate and predict livestock movements. WP5 members helped to define sampling strategy, and WP6 leader has produced a dissemination plan, which includes guidelines to help partners understand their responsibilities in terms of dissemination and the main communication channels. It ensures that all Vmerge products are disseminated in a timely and ordered fashion.
In the field, intense sampling campaigns (WP1-WP3) have been undertaken in 2014 and in 2015, allowing the collection of mosquitoes, sera and environmental data.
Vector sampling. Campaigns for mosquito sampling have been carried out by all partners involved in sampling all along the mosquito season in 2015 (ISRA, CNERV, IAV, IRVT, FMV-AU, IZS, Cirad and Cresa). In addition, some partners (e.g. ISRA and IAV) have continued mosquito sampling during the 2016 season. Mosquito sampling has been structured mainly following three approaches. The first approach focused on Culex pipiens and involved the capture of Cx. pipiens females in several sites in five countries (IRVT, FMV-AU, IZS, Cirad and Cresa). The captured mosquitoes have been analysed using both specific and no-a-priori approaches (i.e. microarrays and shotgun sequencing). Beyond Cx. pipiens individuals, other species have been captured and analysed. Due to field constraints during the 2015 sampling campaign, IAV has carried out a complementary module within the Cx. pipiens approach. The goal of this module is to compare the virome of Cx. pipiens between larval and adult stages. Consequently, IAV has performed sampling during the 2016 season and the analysis of these samples is on-going. The second and third approaches involved the sampling of three mosquito vectors in different habitats in Senegal. ISRA has performed sampling campaigns in 2014, 2015 and 2016 in these habitats. The mosquitoes from 2014 and 2015 have been already processed and analysed. Finally, CNERV has carried out a sampling campaign in different Mauritanian sites where RVFV outbreaks have taken place.
Serum sampling. In parallel to mosquito sampling, several partners have carried out serum sampling campaigns. For example, ISRA has collected and analysed sera from sentinel herds (small ruminants) in different Senegalese localities (3 campaigns in 2015 in 3 different sites, 5 sentinel herds of 30 animals each) and from transhumant livestock (6 field trips, 120 samples/trip). CNERV has launched an extensive sampling campaign, collecting over 1000 sera (ruminants and camels) from sites involved in the Mauritanian RVFV outbreak in 2015. IRVT has also collected 850 sera from sheep.
Molecular and serological diagnostic tools have been developed and/or validated. Field-collected samples were analyzed with different methods in parallel.
Validation of FTA cards as arbovirus surveillance tools. Honey-soaked FTA cards were analysed for their suitability as arbovirus surveillance tools (Cresa). First, standardisation assays were performed to determine the honey concentration required to avoid PCR inhibition and mosquito mortality while allowing feeding. Moreover, validation with experimentally-infected mosquitoes was carried out. Finally, the cards were deployed during field tests (8 sites) and cards showing feeding were analysed with a Flavivirus-specific PCR and shotgun metagenomics. No Flavivirus was detected with the Flavivirus-specific RT-PCR. Shotgun metageomics, on the other hand, allowed the detection of different viruses, supporting the suitability of the FTA cards coupled to shotgun metagenomics as a tool in arbovirus surveillance programs.
Development of a Pan-Flavivirus PCR and a microarray for Flavivirus detection. In order to develop a new pan-Flavivirus microarray (Alere®), FLI has first developed and validated a new pan-Flavivirus SYBRGreen-based PCR. The goal of this development was to produce a robust method to get the amplicons to be used as template with the microarray. The new PCR design showed an improved sensibility when compared with an established design (Chao, 2007) using 26 flaviviruses.
The design of the microarray involved the selection of 84 probes, including probes specific for the whore Flavivirus genus, several Flavivirus groups (JEVG, YFVG, TBEV, Non-Vector), certain species (e.g. WNV, JEV, USUV, MVEV) and sub-species (e.g. WNV lineages). The microarray was then tested with 27 reference strains, some mixed together, showing excellent discrimination power.
Analysis of field-captured mosquitoes with pan-family PCRs and the pan-Flavivirus microarray. FLI analysed the presence of arboviruses in pools of mosquitoes collected. To this end, the Flavivirus microarray plus four PCRs were used: a pan-Flavivirus PCR, a pan-Bunyavirus PCR, a pan-Phlebovirus PCR and a pan-Alphavirus PCR. 1788 pools from 12 different mosquito species and were analysed in total, some with all the PCRs. Positive pools were sequenced. Isolation in cell culture was attempted from positive pools but without success. Several arboviruses were detected, including West Nile virus in Senegal (4 pools) and Italy (3 pools) or Usutu virus in France (13 pools). Beyond arboviruses, 544 pools were positive for mosquito-only flaviviruses.
Development of a SYBRGreen-based RT-PCR to study the prevalence of Alphamesonivirus 1. Alphamesonivirus 1 was largely predominant in several of the samples analysed with shotgun metagenomics (up to 60% of the viral reads in certain samples). Given its abundance, this virus may have a strong influence on the mosquito biology. A detailed knowledge of its ecology is required if this virus is to be used in mosquito or virus control. Cirad launched a study on its prevalence in Cx. pipiens populations in the French Carmargue. A real-time PCR was developed to detect the virus. This PCR was used to analyse 147 pools of females (mean number of mosquitoes per pool = 30; max/min= 34/6) collected in 6 sites along 2016. As shown in the figure below, mean infection rate was 8.1% (95% CI = 6.7-10.1). This infection rate is several orders of magnitude larger than that observed for most arboviruses infecting mosquitoes. No influence of site, habitat type or sampling month on the infection rate was detected.
Development of innovative Microsphere Immunoassay for specific Detection of orbiviruses or flaviviruses Antibodies. ANSES has developed two new microsphere-based immunoassays. The goal of these assays was to simplify the determination of serotypes of the orbivirus Bluetongue virus (BTV) or of Flavivirus species which, when determined by serological analyses, can only be performed by neutralization tests (VNT) which are time-consuming. To improve the serological characterization of BTV infections, the recombinant VP7 and VP2 viral proteins were synthesized using insect cells expression system. The purified antigens were covalently bound to fluorescent beads and then assayed with 822 characterized ruminant sera from BTV vaccinations or infections in a duplex microsphere immunoassay (MIA). The results demonstrated that MIA detected the anti-VP7 antibodies with a high specificity as well as a competitive ELISA approved for BTV diagnosis, with a better efficiency for the early detection of the anti-VP7 antibodies. The VP2 MIA results showed that this technology is also an alternative to VNT for BTV diagnosis. Comparisons between the VP2 MIA and VNT results showed that VNT detects the anti-VP2 antibodies in an early stage and that the VP2 MIA is as specific as VNT. This novel immunoassay provides a platform for developing multiplex assays, in which the presence of antibodies against multiple BTV serotypes can be detected simultaneously. In order to improve the serological differentiation of flavivirus infections, the recombinant soluble ectodomain of WNV E (WNV.sE) and EDIIIs (rEDIIIs) of WNV, JEV, and TBEV were synthesised using the Drosophila S2 expression system. Purified antigens were covalently bonded to fluorescent beads. The microspheres coupled to WNV.sE or rEDIIIs were assayed with about 300 equine immune sera from natural and experimental flavivirus infections and 172 nonimmune equine sera as negative controls. rEDIII-coupled microspheres captured specific antibodies against WNV, TBEV, or JEV in positive horse sera. This innovative multiplex immunoassay is a powerful alternative to ELISAs and VNTs for veterinary diagnosis of flavivirus-related diseases.
Serological surveys for RVFV infections in Senegal, Mauritania and Tunisia. Serum sampling and analysis for RVFV antibodies has been carried out by ISRA along 2015 and 2016. Sampling focused on sentinel herds (small ruminants) in different Senegalese localities (3 campaigns in 2015 in 3 different sites, 5 sentinel herds of 30 animals each) and on transhumant livestock (6 field trips, 120 samples/trip). The results show that RVFV infections of small ruminants have taken place in Senegal although at a low frequency. Positive sera, including a IgM positive, found in transhumant livestock strongly suggest RVFV introduction in Senegal from animal trade. Large serological surveys on RVFV prevalence have also been carried out in Mauritania (CNERV) and in Tunisia (IRVT) in 2015. CNERV found 75 positive sera among 1082 sera collected in several sites affected by an epizooty. On the contrary, IRVT did not found any positive serum among 850 sera collected from sheep all over the country.
Characterisation of differences in microbiota depending on vector and geographic distribution. The study of vector microbiomes was structured in three approaches, each with a different goal. The first approach, named the Culex pipiens project, targeted the characterisation of the virome of Culex pipiens, a RVFV vector, and the potential influence on virome composition of geographical location and human activity on the environment. Culex pipiens populations in sites with two different levels of human activity (peri-urban and rural sites) have been sampled in parallel in five countries (IRVT, FMV-AU, IZS, Cirad and CreSA) along the 2015 season. These sampling campaigns and the subsequent processing of mosquitoes have produced 10 samples (one sample per habitat and country).
The second approach studied the complementarity of classical serological surveys in livestock with virus surveys in mosquito vectors. More specifically, this project analyses the virome of Aedes vexans, a major RVFV vector, in three ecologically-similar sites along the rainy seasons in 2014 and 2015 (ISRA, 16 samples in total, some samplings did not lead to any capture). The results will be compared to those of a serological survey for the prevalence of different arboviruses carried out in the same sites and periods (ANSES).
The last approach explored the influence of habitat and taxonomy on the virome through the comparison of the viromes of two related mosquito vectors, Culex tritaeniorhynchus and Culex poicilipes, in three different habitats with specific surface-water conditions. The three habitats were sampled by ISRA two consecutive years (2014 and 2015), leading to the production of 12 samples.
For all approaches, mosquito sampling finished in late 2015, after the mosquito season, and was followed by a huge work on mosquito identification. For example, over 355,000 mosquitoes were identified to the species level by ISRA. Mosquitoes of the targeted species were pooled in groups of 30 individuals and their RNA extracted. Then, RNA from selected pools was subjected to shotgun sequencing by Pathoquest. Sequencing results were bioinformatically processed to obtain a list of viruses potentially present in each sample (Pathoquest). Then, statistical analyses were carried out to compare the virus communities from the different samples (Cirad). Despite the fact that the whole pipeline is highly demanding in terms of time, the results of the bioinformatics analysis are already available for all the samples. Nevertheless, statistical analysis has only been achieved for the Culex pipiens project. The results provided by this project represent a breakthrough in our knowledge of the mosquito virome. For the first time and thanks to, among others, the large geographical scale of the study, the virome of an important mosquito vector has been dissected in detail showing a virus community highly diverse comprising a large fraction of new virus species. The composition of the virome resembles that of the bacteriome in insects with a few dominant families representing most of the sequences. Variations depending on geographical location were observed, with African samples generally clustering apart from European samples. Nevertheless, a relatively large group of virus families was conserved among samples strongly suggesting that this group forms the core virome of Cx. pipiens and probably influences mosquito biology. An isolate of a member species of the core virome was obtained to be used in future studies on its influence of vector fitness and competence.
Key activities on vector competence and transmission were to determine the ability of European insects and livestock breeds to become infected with emerging arboviruses, primarily Rift Valley fever virus (RVFV), and how this is affected by external variation in population origin, environment, and ingested dose, by conducting experimental infections.
Rift Valley fever virus (RVFV) is transmitted between vertebrate hosts primarily via the bites of mosquitoes, particularly those in the genera Aedes and Culex. The commonest mosquito species known to be capable of transmitting RVFV is Culex pipiens, and preliminary studies of Culex pipiens obtained from France (Moutailler et al. 2008) suggest that some European populations of this species are relatively susceptible to RVFV infection. However, different populations of a species may have different levels of competence (e.g. Moutailler et al. 2008). Extrapolation from studies using a single vector population may poorly estimate the potential for a virus to spread and the extent to which that potential could vary geographically. Therefore, in order to accurately estimate possible regional variation in the potential for RVFV to spread within Europe, which is a key objective of the Vmerge project, it is essential to conduct competence studies using several geographically distinct populations of the species.
For their studies, DLO-CVI used Laboratory-reared Culex pipiens mosquitoes originating from an above-ground collected pool of egg rafts were generously provided by the laboratory of Entomology of Wageningen University. The colony originates from an above-ground collected pool of egg rafts collected in Brummen, The Netherlands and was established in 2010. The mosquitoes were maintained at 23ºC in BugDorm cages with a 16:8 light:dark cycle and 60% relative humidity. The mosquitoes were provided with 6% sucrose and bovine or chicken blood using a Hemotek PS5 (Discovery Workshops) for egg production. Egg rafts were hatched in tap water supplemented with Liquifry No. 1 (Interpet Ltd.). Larvae were fed with a mixture (1:1:1) of bovine liver powder, ground koi food and ground rabbit food.
At CReSA, two different populations of Cx. pipiens were used: Cx. pipiens form molestus from Empuriabrava and a hybrid between pipiens form and molestus form, from Gavà. The population of Ae. albopictus came from Sant Cugat de Vallés, a town belonging to the metropolitan area of Barcelona and the place where the Asian tiger was first identified in Spain in 2004. All mosquito populations have been reared in the laboratory to obtain stable colonies and the number of filial generations was more than 30. Mosquitoes were reared under the following environmental conditions: photoperiod was 14 h:10 h (light:dark) with two crepuscular cycles of 30 min between to simulate dawn and dusk; the mean temperature during day was 26˚C and 22˚C during night. Relative humidity (HR) was maintained constant at 80%. The mosquito colonies were tested for the presence of Flavivirus, Alphavirus and Phlebovirus by reverse transcription nested polymerase chain reaction (RT-nPCR) to confirm the absence of other viral infections. The colonies were also tested for the presence of Wolbachia spp. by PCR analysis of a fragment of wsp gene as previously described. All colonies resulted Wolbachia spp. positives (data not shown).
The Pirbright Institute experiments used Cx. pipiens form pipiens and Cx. pipiens form molestus hybrid mosquitoes originating from mosquito larvae and pupae collected from a container habitat in an allotment area in Surrey, UK (Manley et al., 2015). Mosquitoes have been reared in the laboratory to obtain stable colonies and the number of filial generations was more than 45. Mosquitoes were reared under the following environmental conditions: photoperiod was 16 h:8 h (light:dark) at 25 ±1°C. Relative humidity (HR) was maintained constant at 60 ±10%.
The Clone 13 strain of RVFV was used by TPI and CVI. This particular strain was used for two reasons: to increase comparability with published results from French mosquitoes and because biosecurity regulations in NL permit the use of this field-attenuated strain under lower containment requirements. The virus was propagated in mosquito cells C6/36. As this strain is the property of the Institut Pasteur, the KE&C department at IP was contacted and agreed to the use of the strain for the purposes of the project. CReSA used RVFV virulent strain 56/74 due to repeated difficulties in shipping Clone 13 within the timeframe required for the infection studies. The virus was propagated in mammalian BHK-21 cells and blood virus mixtures were prepared by resuspending cattle erythrocytes in freshly harvested virus-containing culture medium (DLO-CVI), mixing bovine blood with virus, heparin and ATP (CResA) or by mixing commercially available defibrinated sheep blood with freshly harvested virus containing culture medium from C6/36 cells.
Feeding was performed using a Hemotek feeding system. Mosquitoes were maintained at 26 °C (TPI), 28°C (CVI) or 26°C / 22°C day/night (CReSA) for either 14 (CReSA) or 21 (CVI, TPI) days and given 6% (CVI) or 10% (TPI, CReSA) sucrose in water ad libitum. At the end of the period, saliva was collected by forced salivation (capillary tube method). Saliva samples and bodies were tested by virus isolation using Vero cells (TPI, CVI and CReSA) and/or BHK cells (TPI). Virus isolation was detected by cytopathic effect and confirmed by quantitative real-time PCR (TPI, CVI and CReSA) or immunoperoxidase monolayer assay (IPMA; CVI).
Low or negligible levels of transmissible infections were observed in most populations tested, with exception of the experiments performed with Dutch Cx. pipiens mosquitoes using the Clone 13 at high viral dose. Analysis of the results obtained from Dutch populations indicates that approximately 80% of Dutch Culex pipiens develop a transmissible infection as detectable by virus isolation from saliva, when a dose of 1010.5 TCID50/ml was used and an incubation of 21 days at 28ºC. The results of infections at other doses are described under D2.3. No transmissible infections were seen in UK mosquitoes.
One factor which is well-established to affect the proportion of individuals developing a disseminated infection is the environmental temperature. The epidemiological importance of a mosquito species in the field ultimately depends on its ability to transmit virus to a new host. After ingesting a potentially infectious blood meal, transmission requires that the replicating virus overcomes or bypasses various barriers to dissemination to ultimately infect the salivary glands. Several factors affect mosquito competence by changing the potency or effectiveness of these barriers or by providing the virus with a way to evade them. One such factor is the simultaneous oral ingestion of microfilarial blood or tissue nematodes, such as Onchocerca or Brugia, which mechanically disrupt midgut integrity during infection. In one previous study, Aedes taeniorhynchus became more competent to RVFV infection and transmission (31% compared to 5%) when simultaneously fed Brugia malayi (Turell et al. 1984). Insect-transmitted blood and tissue nematodes such as Brugia are endemic in many parts of the world and their distribution overlaps with that of several important and high-impact veterinary, zoonotic and human arboviruses. The net effect of such coinfection is unclear, however, since it has also been noted in separate studies that this damage affects other parameters of vectorial capacity, namely the duration of the extrinsic incubation period (EIP) and the post-feeding mortality of the mosquitoes.
We investigated the consequences of ambient temperature and ‘shock’ temperature treatment on the proportion of Culex pipiens mosquitoes from European populations that develop transmissible infections with RVFV, and consider its possible implications for changes to the emergence potential of RVFV if introduced into Europe during extreme weather conditions. We also investigate the consequences of coinfection of mosquitoes with microfilarial nematodes (Brugia spp.).
To study the influence of temperature during the extrinsic incubation period (EIP), mosquitoes were allowed to feed on a blood meal containing a high dose of virus (1010.5 TCID50). Feeding was performed using a Hemotek feeding system. After feeding, the mosquitoes were maintained at different temperatures and a humidity of 70% and given 6% sucrose in water ad libitum. After an EIP of 14-, 21- or 28 days, the infection rates (IRs) and transmission rates (TRs) were determined. Saliva was collected by forced salivation (capillary tube method). Saliva samples and bodies were tested by virus isolation using Vero cells. Virus isolation was detected by cytopathic effect and confirmed by immunoperoxidase monolayer assay (IPMA). To study the effects of Brugia malayi coinfection, microfilariae were obtained from the Liverpool School of Tropical Medicine. Microfilariae were sourced from infected gerbils, counted, and re-suspended in RPMI medium at 10,000-12,000 microfilariae/ml, and kept at 37°C. Microfilariae were transported from the Liverpool School of Tropical Medicine to the Pirbright Institute on the day of collection, at approximately 25°C (~5-7 hours). After the arrival at the Pirbright Institute, the microfilaria were kept at 37°C until infection (<24 hours). Wild type (wt) strain of Bunyamwera virus (BUNV) was used. This virus is member of the Bunyaviridae virus family, is closely related to Rift Valley fever virus (RVFV), and is often used as a model bunyavirus. In the UK, BUNV is classified as a hazard group 2 biological agent by Advisory Committee on Dangerous Pathogens (ACDP) and it can be handled safely at containment level (CL) 2. BUNV has been shown to disseminate efficiently in mosquitoes of genus Aedes and Culex.
Seven to nine day old non-blood-fed female mosquitoes starved for 20-24 hours (no sucrose, no water) were used for infection experiment. Feeding was performed using a Hemotek feeding system. Mosquitoes were maintained at 26 °C for up to 14 days and given 10% w/v sucrose in water ad libitum. To investigate the competence of mosquitoes for B. malayi, individual mosquitoes were anesthetised and dissected (head, thorax and midguts) live into phosphate buffered saline with 0.75% bovine albumin (PBSA) or RPMI medium with 10% foetal bovine serum (FBS), incubated at 37 °C for 1 hour and then observed microscopically for the presence of B. malayi. To investigate the effect of B. malayi co-infection on the competence of mosquitoes for BUNV, individual mosquitoes were anesthetised, snap-frozen on dry ice and then dissected (head, thorax and midguts) into PBSA. Dissected body parts were homogenized using Bio-Rad TissueLyser at 20 Hz for 80 seconds and assayed for BUNV infection by plaque assay on BHK-21 cells.
The temperature study results show that the IRs found after EIPs at 20˚C are not significantly different from those found after incubation at 28˚C (compare Figs 3A and B). Virus titres in the bodies as well as the TRs were however lower after an EIP at 20ºC. Interestingly, when mosquitoes were fed and maintained for 7 days at 20˚C and subsequently maintained at 28˚C for 7, 14 or 21 days (corresponding to 14, 21 and 28 DPI), infection and transmission rates were similar as those obtained after corresponding EIPs at 28˚C.
Our findings suggest that the proportion of exposed mosquitoes with saliva from which infectious virus can be isolated increases between 14 and 28 days post-infection. However, the total proportion developing transmissible infections was not substantially different over the temperature maintenance regimes tested. We therefore conclude that for the conditions tested, there is not strong evidence to conclude that the efficiency of the process of infection itself changes with temperature, only the rate at which infection progresses.
For the Brugia studies, the competency of Aedes aegypti and Culex pipiens for Brugia malayi was investigated at 3, 7, 11 and 14 days post-infection (mosquitoes fed with 20,000 mf/ml). As anticipated based on literature research, Cx. pipiens mosquitoes were found to be non-competent for B. malayi while Ae. aegypti mosquitoes were found to be competent for B. malayi. Based on this, Aedes aegypti mosquitoes were chosen to investigate the effect of B. malayi co-infection on the competence of mosquitoes for BUNV. Analysis of mosquitoes 7 days post-infection revealed no increased infection and/or dissemination rate in mosquitoes co-infected with BUNV and B. malayi as compared to the mosquitoes infected with BUNV only. The survival rates were also comparable.
Analysis of mosquitoes 14 days post-infection suggested increased infection rate in mosquitoes co-infected with BUNV and B. malayi as compared to the mosquitoes infected with BUNV only. However, low numbers of available samples due to high mosquito mortality do not permit to analyse these data with high confidence.
We conducted additional experimental infection studies using the protocols developed above but with different infectious doses to allow the impact of feeding dose on competence to be estimated, improving our understanding of the uncertainties that can arise when extrapolating from laboratory infection studies to field conditions. At DLO-CVI, mosquitoes were maintained for 21 days at 28ºC with a humidity of 70% and a 16:8 light:dark photoperiod. After the extrinsic incubation period (EIP), body and saliva samples were evaluated for the presence of virus by virus isolation. All results were confirmed either by quantitative reverse transcriptase real-time PCR (qPCR) or immunoperoxidase monolayer assay (IPMA). Post EIP analysis of mosquitoes fed with the low-dose blood meal revealed an infection rate (IR) of 20% and a transmission rate (TR) of 8%. Analysis of mosquitoes fed a medium-dose revealed an IR of 64% and a TR of 14%. Mosquitoes fed a high-dose revealed a IR of 91% and a TR of 24%. In sporadic cases, virus isolations from saliva samples were positive, whereas isolations from corresponding bodies were negative (Fig. 4).
At CReSA, in the Cx. pipiens hybrid colony line, the rates of infection and dissemination tended to increase with higher doses of blood-virus mixture. During infections at the Pirbright Institute, the mosquito feeding rate ranged from 0.0% to 82.4%. The initial experiments were characterised by low feeding rates (0.0-6.1%). Following extensive protocol optimisation the feeding rates increased to 59.2-82.4% and were highly reproducible. For initial experiments, the mosquitoes were maintained for extrinsic incubation period (EIP) of 21 days at 26ºC with a humidity of 70% and a 16:8 light:dark photoperiod. The above conditions were altered to 14 days at 26ºC with a humidity of 70% and a 16:8 light:dark photoperiod for later experiments. Mosquitoes reared at two different larval densities (500 and 2000 larvae per 1 L of water) were tested. Two virus doses were used, low (~5.0 Log10 TCID50 per ml of blood meal) and high (~7.0 Log10 TCID50 per ml of blood meal). After the extrinsic incubation period (EIP), body and saliva samples were evaluated for the presence of virus by virus isolation. Post EIP analysis of mosquitoes fed with the low-dose or high-dose blood meal and incubated for 21 days revealed no evident infection. Survival rates for mosquitoes fed with low dose of the virus (26.8-34.2%) were higher than survival rates for mosquitoes fed with the high dose of the virus (14.7-17.8%).
In summary, the proportion of mosquitoes becoming infected, and developing transmissible infections, was observed to increase with higher oral infection doses. Additional experiments to investigate the infectiousness of blood obtained from infected animals were also conducted at CVI and suggested that the period of infectiousness to vectors may be shorter than the period for which viral RNA can be detected.
The susceptibility of European livestock breeds to RVFV, and their ability to infect mosquitoes that feed on them, are significant unknown components in estimating the threat RVFV poses to Europe. The infection is most significant in productive livestock such as small ruminants, cattle and camels. Characteristic ‘abortion storms’ with up to 100% new-borne case fatality rates, can be seen during acute infections. However, only moderate symptoms are observed in infected adult cattle and camels. The significant role of camelids in transmission and spread of the virus were already demonstrated by their involvement during the introduction of RVFV to Egypt 1977. During an epidemic of RVF in Mauritania 2010, severe clinical manifestations, such as ocular discharge, foot lesions, hemorrhages, neurological disorders as well as fatal hyperacute progressions were observed for the first time in camels. An association between infected camels and subsequent human infections highlights the previously underestimated role of camelids in terms of RVFV epidemiology. In particular, although camelids have been implicated as playing a significant role in the epidemiology of previous outbreaks of RVFV and the farming of camelids such as alpaca, vicuna and llamas in Europe has increased substantially in recent years, little is known about the outcome of infection in these species. The purpose of this deliverable is to determine the susceptibility of livestock breeds used in Europe, including camelids and sheep, to infection with RVFV, as well as the potential for European mosquitoes biting infectious animals to become infectious themselves and thereby sustain local transmission.
Three alpacas (used as model camelids) were challenged with the live attenuated MP-12 strain of RVFV, and immunogenicity and pathogenicity were measured. The MP-12 strain was chosen both for practical reasons, since in Germany it can be used for animal experiments under BSL-2 conditions, and because its attenuation means it has been proposed as a potential vaccine in the event of an outbreak. In addition, several comparable studies using the same strain were also carried out recently in cattle and sheep, allowing our results to be evaluated in a broader context. The MP-12 strain of RVFV was kindly provided by Professor Richard Elliot of the MRC Centre for Virus Research (CVR) at the University of Glasgow in the United Kingdom. MP-12 was propagated in Vero 76 cells (Collection of Cell Lines in Veterinary Medicine, Friedrich-Loeffler-Institut, Germany). Three male alpacas (Vicugna pacos), at the age of 6 month (#A1, #A2) and 4 years (#A3) respectively, were acquired from a local breeder. All three alpacas were immunized subcutaneously with 2 ml of a 106 TCID50/ml virus dilution. After inoculation, serum, whole blood, nasal- and conjunctival swabs and faeces were taken at 3,5,7,10,13,17,19,21,24,28 and 31 days post infection (dpi). Body temperatures were taken on the same days and additionally in the first two days post infection for recognition of any acute adverse effects of immunization. Serum neutralization test (SNT) was performed as described in the OIE Terrestrial Manual (2014). The presence of RVFV-derived RNA was verified using a quantitative real-time RT-PCR (RT-qPCR) by Bird et.al.
To compare both acute and subacute effects in this study, animals were euthanized at 3 (#A1) and 31 days (#A2, #A3) post infection (dpi). Clinical monitoring, analysis of liver enzymes and haematological parameters demonstrated the tolerability of the MP-12 strain, as no significant adverse effects were observed. Comprehensive analysis of serological parameters illustrated the immunogenicity of the vaccine, eliciting high neutralizing antibody titres and antibodies targeting different viral antigens. Significant serological responses were detectable in all assays for animal #A2 and #A3, whereas no antibody generation was found in animal #A1. For both #A2 and #A3 the generation of neutralizing antibodies started at 7 dpi. In the course of the experiment high neutralizing titres of up to 1:1280 were reached. RVFV was detected in serum and liver of the alpaca #A1 euthanized 3 dpi, whereas no virus was detectable at 31 dpi (#A2 and A3#). Viral replication has been determined by detection of various RVFV-antigens in hepatocytes by immunohistochemistry and the presence of mild multifocal necrotizing hepatitis.
In conclusion, the MP-12 RVFV strain was found to be a promising candidate vaccine for European camelid livestock.
To determine the period in which Culex pipiens mosquitoes become infected after feeding on viremic lambs, two 12-week-old lambs of the Dutch Texel-Swifter breed were inoculated with 105 TCID50 of the highly virulent RVFV strain 35/74 via intravenous route. EDTA blood samples, to be used for virus isolation, were collected every day after challenge. On days, 1, 2, 3 and 4, cardboard boxes containing 50 mosquitoes were placed on the inner thigh and fixed with elastic bandage (Fig. 5). After 45 min. the boxes were removed and brought to the biosafety level-3 (BSL-3) laboratory. Engorged mosquitoes were collected and transferred to new containers. The mosquitoes were held at 20˚C for 7 days and were subsequently transferred to 28˚C. After 5-6 days, the mosquitoes were assayed for the presence of virus in bodies and saliva. Both lambs developed viremia one day post infection (DPI) followed by a peak in body temperature on DPI 2 (Fig. 5A, B). The first lamb succumbed to the infection on DPI 3, whereas the second lamb had to be euthanized on DPI 4 after reaching a humane end-point. Mosquitoes were maintained at 20˚C for 7 days and 28˚C for 7 (mosquitoes that fed on DPI 1), 6 (mosquitoes that fed on DPI 2 or 3) or 5 (mosquitoes that fed on DPI 4) days. Analysis of mosquito samples after the extrinsic incubation time (EIP) revealed very high infection rates (IRs, 86-91%) in the groups of mosquitoes fed on DPI 2, but very low (11-18%) IRs in the groups of mosquitoes fed on DPI 3. This correlated with transmission rates (TRs), being 29-30% in mosquitoes that fed on DPI 2 and 0-11% in mosquitoes fed on DPI 3. The IR and TR of mosquitoes that fed on DPI 4 on lamb 1 were also very low, 8% and 0% respectively (Fig. 5C, D).
Interestingly, these results suggest that efficient transmission from viraemic lambs to mosquitoes takes place in a short time period of little more than 24 hrs.
In silico, extending models with host/vector interaction has been explored for RVF and Culicoides-borne diseases (CBD) to capture the interaction between hosts and vectors and the importance of environmental factors in both endemic and non-endemic situations. We successfully developed two approaches to assess the transmission risk of RVFV in endemic areas of Western Africa: a mechanistic model using field data as inputs and a climate-based modelling approach. RVFV risk maps for up-to-now free areas (Northern Africa and Europe) were produced using results on mosquito abundance, but the figure is not substantially different from the risk maps created before.
Forecasting RVF outbreaks in an endemic area. This was carried out using two different approaches. The first used a mechanistic approach using field data (serological analysis, entomological collections, analysis of nomadic herds movements) to predict the transmission of Rift Valley Fever virus (RVFV) in Senegal (led by Cirad, with strong interactions with ISRA, CNERV and ANSES). The second (led by FAO) aimed to test, calibrate and validate an existing climate-based model to identify and predict RVF at-risk areas in Senegal and Mauritania.
The mechanistic modelling of RVF transmission in Senegal was divided into the following steps:
1. Modelling the dynamics of Rift Valley fever mosquito vectors in different ecosystems of Northern Senegal: a rainfall-driven model of vector population dynamics developed by Soti et al. (2012) was applied to different study areas and its accuracy assessed using entomological field data (2015-2016). Upscaling the model allowed the development of a regional model, based on satellite-derived temperature, rainfall and humidity estimates;
2. Analysis the nomadic herd movements around Younoféré, and their potential role in the virus introduction into the pilot area: a field survey has been performed involving arriving nomadic herds during the 2015 rainy season, to measure key parameters, such as the main origin areas, the proportion of resident/nomadic herds according to time, and the proportion of seropositive nomadic animals upon arrival.
3. Modelling the dynamics of Rift Valley fever in different ecosystems of Northern Senegal: a deterministic compartmental model of RVF transmission integrating nomadic movements and using entomological model outputs as inputs was developed in Younofere pilot site, and calibrated with 2015 and 2016 serological data.
1. Modelling the dynamics of Rift Valley fever mosquito vectors in different ecosystems of Northern Senegal. In 2015 and 2016, mosquito populations were sampled by ISRA in the same 10 sampling sites as in 2014 from three different geographic areas (Diama, Keur Momar Sarr, Younouféré) using CDC light traps with CO2. Five trapping sessions were performed (jul-aug-sept-oct-nov). These captures allowed to collect 115,075 mosquitoes over 100 trap/nights, belonging to 31 species in 6 genera (Aedes, Aedeomyia, Anopheles, Culex, Mansonia and Uranotaenia). In Diama and Keur Momar Sarr, the preponderant species was Culex tritaeniorhynchus with respectively 53% and 33.6% and Mansonia uniformis with 36.7% and 46.4% respectively; whereas, Ae. vexans arabiensis (79.6%) was the most common mosquito in Younoufere. The model was initially developed for Ae. vexans and Cx. poicilipes in the Ferlo pastoral area (Northern Senegal). The dynamics of these two vector species is modelled by combining a hydrological model of the dynamics of the water bodies, with mosquito population models describing different stages of the mosquito life cycle. For Diama and Keur Momar Sarr, the dynamics of Cx. poicilipes and Culex tritaeniorhynchus was modeled using the same mosquito population model. For the three species, parameters and functions were defined according to a bibliographic review and expertise of ISRA partners. The parameters are constant values, whereas the functions (e.g. mortality, transition rates from one stage to the next one) are temperature, humidity, rain or flood-dependent. The model was implemented in the three study sites. Daily rainfall, humidity and temperature data were used as inputs. Entomological data collected during the 2014 and 2015 rainy seasons (July-November) were used to validate the models for the three species. The results showed a good fit of the predicted mosquito abundances with field data. Satellite-derived rainfall, temperature and humidity estimates were used as input of the models to simulate RVF vector population dynamics at a regional scale. The data were freely downloaded from internet databases. The assimilation of environmental data into the mosquito population model was realized with Ocelet, a domain specific language for modelling spatial dynamics and manipulating geographic information (www.ocelet.fr).
2. Analysis the nomadic herd movements around Younofere site. The objectives were: (a) To collect data on nomadic herders’ movements (spatial, temporal, reasons for movements), nomadic livestock characteristics (composition, size, RVF status) and the main features of the locations; (b) Graph an “observed” nomadic herder’s movement network and compute general network metrics, centrality measures and village and herder descriptive statistics; (c) Fit a model of the probability of occurrence of nomadic herd movements using an Exponential Random Graph Model (ERGM) and potential factors such as nodal attributes (village characteristics), edge attributes (number of animals moved, time of movement, seroprevalence in the herd) and structural and relational attributes (factors specific to nomad’s network) of the observed network; (d) Graph a simulated a nomads’ movement network from the fitted ERGM. During the 2015 rainy season, a 10km radius buffer zone centered on Younoféré was defined, as well as a census of the main temporary ponds in this zone. Among these ponds, 10 ponds were selected, based on their surface and the high/low frequentation by nomadic herds. From early August to mid-September 2015, 3 field sessions were conducted with (i) census of all nomadic herds settled around the ten ponds and (ii) movement questionnaire (ii) for each session, 4 herds newly arrived, in which 30 animals were sampled. This survey has been be replicated from early November to mid- December (3 sessions of census, questionnaire and sampling). Data has been gathered in a dedicated database. The total number of animals seen during the study period was 33,600 with 7,335 cattle heads, 19,466 sheep heads, 5,240 goats and 1,559 donkeys. The proposed exponential random graph model showed a good goodness of fit, and the simulated movement network showed several similarities with observed movement network when assessed on several parameters. Overall, nomads tend to own large herds and have movements geographically spread over all Senegal which highlights the importance of this work. Several nodes (towns or villages) such as Payar, Younoufere and others seem to be important hubs for nomads. Thus, planning interventions in these locations is highly recommended. Herders that have more goats tend to do less stops for water and food and thus do longer distances without stops. The presence of good quality grass in a stopping point makes nomads stay for longer periods of time and thus slow their movement habits.
3. Modelling the dynamics of Rift Valley fever in different ecosystems of Northern Senegal. The complete model, combining entomological, resident and transhumant herd components was first built and calibrated for Younoféré site. The epidemiological system was considered at the pond scale, considering that mosquitoes come from ponds and that the area of influence of that pond is of 2.5 km radius. Four dynamical processes were modelled: Host population renewal (arrival, departure of transhumants); Host population dynamics (gestation, birth, death, sales); Infection dynamics (virus transmission); Vector population dynamics (output of the entomological model). Host population dynamics was established from expert opinion and field observations. For nomadic populations, we recorded nomadic data as previously presented, within an area of 75km2. Regarding the vector population, we used the outputs of the entomological model previously described. To calibrate the model we used the results of a serological survey performed in 2016 around Younoféré. Results were in favor of an endemic circulation, at least since 2013. Last steps will be (i) to perform a sensitivity analysis on the model to identifiy the parameters that strongly influence the outputs, (ii) to test similar scenarios with Barkedji data (2003-2004), (iii) to combine this network with the 3 calibrated local models and see in what extent these movements influence the transmission/persistence of the virus in Northern Senegal, and spatialize the model through northern Senegal (iv) use this model to identify the most cost-efficient control measures and surveillance designs.
Time-series analysis on NDVI-based model. Within the framework of the Vmerge Work Package 3, FAO focused on testing, calibrating and validating an existing climatic model (hereafter, climate-based model) in Senegal and Mauritania, based on the work by Anyamba and collaborators (e.g. Anyamba et al. 2009), who developed a monitoring and risk mapping system that utilizes the Normalized Difference Vegetation Index (NDVI) as a proxy for ecological dynamics to map areas at potential risk of Rift Valley fever in East Africa. This is based on the findings that NDVI is a complex proxy representing the interaction of several variables, including rainfall, temperature, and soil moisture that are important for the emergence, survival, and propagation of Rift Valley fever mosquito vectors. Using this approach Anyamba and collaborators successfully retrospectively predicted areas where Rift Valley fever outbreaks occurred (Anyamba et al. 2002; Anyamba et al. 2009, 2010). Based on literature surveys and thresholding of climate variables (rainfall and long-term NDVI), the climate-based model creates a static mapping of the potential epizootic area mask, hereafter RVF vector suitability area (Anyamba et al. 2009). Within the suitability area, identification of risk areas is determined from NDVI anomaly calculations, taking into consideration the persistence of positive anomalies above a threshold of 0.025 NDVI units (based upon a desert calibration) over a 3-month period. The model thus captures these temporal dynamics in ecology using time-series measurements of NDVI. Pixels with persistently average positive anomalies over a three-month period exceeding a threshold of 0.1 NDVI are considered hotspots for vector amplification. An exhaustive description and explanation of the original model is available elsewhere (Anyamba et al. 2009).
A preliminary analysis and application of the uncalibrated climate-based model for the year 2003 and 2005 showed that: a) the RVF outbreaks occurred outside the vector suitability map, particularly in Mauritania and southern Senegal, indicating a low accuracy of the model for these regions; 2) the RVF at-risk areas were not predicted in semi-arid areas of Senegal (e.g. temporary ponds in the Ferlo region) and Mauritania; 3) however, the RVF outbreaks were accurately predicted along the Senegal river and in south-western Mauritania. The analysis of satellite-derived rainfall and NDVI data (source: FEWSNET) available at subnational level for areas of permanent water and temporary ponds, showed that: 1) the rainfall patterns and NDVI anomalies in areas of permanent water such as along the Senegal river were similar to those observed in the flooding areas of Kenya; 2) the rainfall patterns observed in the Ferlo region highlighted the presence of dry spell events even at subnational level. As described below, following these preliminary results, we updated the RVF vector suitability map and we developed two models to predict RVF at-risk areas for vector amplification under different ecological conditions: 1) the climate-based model for irrigated areas and flooded areas (e.g. along the Senegal river), and 2) the dry spell model for semi-arid areas and temporary ponds (e.g. the Ferlo region).
FAO compiled historical information available on outbreaks, serological, and entomological data collected from previous studies (e.g. ACI, CORUS, EDEN and EDENEXT projects; FAO EMPRES-i outbreaks database; FAO/Oxford RVF database) in Senegal and Mauritania, the WorldClim dataset (i.e. 19 temperature and rainfall bioclimatic variables, Hijmans et al. 2005) as well as time-series of satellite-derived climate (e.g. CHIRPS, TRMM, TAMSAT, RFE) and vegetation (e.g. NDVI from: eMODIS NDVI, MODIS AQUA NDVI, SPOT-VGT NDVI, METOP, PROBA-V) data spanning from 1981 to 2016. A database of 556 RVF records from 1988 to 2015 was used to train, calibrate, and validate the models. Then ecological niche modelling (i.e. MaxEnt (Phillips et al. 2006); kernel density) and time-series analysis were applied to identify: a) areas historically suitable for RVF occurrence; b) main RVF climatic predictors and c) climate anomalies that may trigger vector population dynamics in different ecosystems, such the Senegal River (e.g. St. Luis; Diama, Ross Bethio) and semi-arid areas (e.g. temporary ponds of the Ferlo area). A kernel density function was applied to the georeferenced RVF outbreaks (spanning from 1988 to 2015) to estimate a historical RVF distribution (90% contour) for model interpretation. Kernel density function 50% contours were used to identify yearly RVF hotspots for model validation. The RVF vector suitability area was updated using Principal Component Analysis and Iso-clustering of bioclimatic variables (FAO 2015) to define regions of similar ecological conditions for RVF vectors. The NDVI anomaly thresholds for Senegal and Mauritania were defining based on the analysis and comparison of climatic data (e.g. NDVI and rainfall) extracted at outbreak locations and RVF outbreak hotspots level (e.g. kernel contour) as well as by performing sensitivity analysis of the anomaly thresholds. The PCA and Iso-clustering method identified two main clusters including all RVF outbreaks of Senegal and Mauritania respectively (Fig. 6). The MaxEnt model, which was run for the Mauritania end Senegal clusters separately, was very accurate (AUC 0.98 for Senegal and 0.96 for Mauritania) and highlighted five main climatic predictors: Precipitation between August-October; Precipitation seasonality; Temperature seasonality; Mean Temperature of wettest quarter (between August-October); Mean Temperature of warmest quarter (between March-May). An important predictor for Mauritania was also Total annual precipitation. The two different models were then averaged to generate a final suitability map for RVF occurrence (Fig. 7).
Results confirmed that the original climate-based model used by FAO for East Africa performed well for years of well above-normal rains (e.g. year 2005, 2010, 2012, 2013), particularly along the Senegal river and in areas located in south-western Mauritania. However, it was not able to identify hotspots for potential RVF amplification in semiarid areas of the Ferlo region, particularly for years with irregular rainfall patterns, such as 2003 and 2008. The new 3-monthly NDVI average anomaly was set at 0.08 NDVI value for Senegal and Mauritania. In order to capture rainfall anomalies such as dry spell events that trigger Aedes sp. vector population dynamic in semi-arid areas (such as in the Ferlo region and Mauritania), a second model was developed. Based on literature review (Caminade et al. 2014), a dry spell was defined as a rainfall pause of about 10 days that is preceded and followed by a significant rainfall event (> 20 mm per day). The dry spell events were identified using CHIRPS pentadal data and TAMSAT decadal data (min and max thresholds of 5 and 20 mm respectively). The output maps of the ‘dry spell’ model were verified using the water pond monitoring system developed by USGS/USAID/FEWSNET (available at http://earlywarning.usgs.gov/fews/software-tools/21). The final RVF maps consisted in overlaying the output of the calibrated climate-based model (i.e. average cumulative anomalies) with the dry spell model to capture environmental and climatic drivers of vector dynamic in different ecosystems of the region (Fig. 8).
In-field studies on the existing pathways of animal trade of importance for RVF introduction and spread in Egypt have been designed carried out. CReSA and IZSAM decided to focus the efforts on Egypt because of its relevance in relation to the risk of RVF introduction into North Africa and to complement the studies CIRAD was already carrying out in other Western African countries. After the meeting held in Egypt in May 2015 between CreSA team and the Official Veterinary Authorities, they agreed for Egypt’s participation within VMERGE project and the questionnaires designed for Northern African countries were adapted into a new revised English version. Data on legal movements of cattle and camels was collected through the questionnaires above mentioned. The main objective of the studies carried out in Egypt was to gain an understanding of the legal trade of animals into Egypt and evaluate its impact in relation to the risk of RVF introduction. To do that, data on all legal importations of animals into Egypt between 2012 and 2015 (i.e. cattle and camels imported from Ethiopia and Sudan) was provided by the Quarantine & Inspection Department of the Ministry of Agriculture and Land Reclamation of Egypt (QID-MALRE). All cattle and camels legally imported into Egypt need to be quarantined. The QID-MALRE also provided data on the specific quarantine facility to which each of the lots of animals imported during this 4 years-period were taken. Once released from the quarantine facilities, animals may be directly sent to slaughterhouses or to animal markets. Data on the location and characteristics of slaughterhouses and animal markets were provided by the Department of Epidemiology of the Ministry of Agriculture and Land Reclamation of Egypt (DE-MALRE). The second objective of the studies carried out in Egypt was to evaluate the risk of RVFV transmission within Egypt. It was evaluated using a GIS-based multi-criteria evaluation (MCE) approach, which allows to combine the data on the distribution of the different factors may influencing the risk of a given disease. In the case of RVF it is essential to account for the distribution of both susceptible hosts of RVF (cattle, buffalo, sheep, goats and camels), and the distribution of the vectors of RVFV. For the estimation of the distribution of susceptible host, data on the number of vaccine doses administered by animal species and by governorates were provided by the ED-MALRE. Finally, animal trade data and risk of RVFV transmission data were combined to identify those areas and periods in which the introduction of RVFV is more likely, so that adequate control measures can be implemented.
A genetic module for RVF between EMPRES-i and Virus Pathogen Resource (ViPR) was developed. The primary objectives were to:
1. To explore which countries, individuals and research organizations have genomic/genetic RVFV sequences that could be annotated and submitted to the repositories (e.g. GenBank/DDBJ/EMBL) of the International Nucleotide Sequence Database Collaboration (INSDC) or to another centralized database;
2. To explore if these databases include sequences that need further annotations or quality checks;
3. Which steps are needed to provide open source bioinformatic tools to analyze genetic data and link this information to epidemiological data;
4. To develop a RVF sequence centralized database and an EMPRES-i genetic module for RVF according to the work-plan developed under (3).
SIB conducted a review of the existing virus sequence platforms that would be appropriate for the development of a genetic module and they identified the Virus Pathogen Resource (ViPR2) as an excellent option for this linkage for several reasons, including the uniqueness and completeness of the database for RVF, the fact that developers were already known and appreciated (previously developed the Influenza Research Database, IRD), the presence of analytical and visualization tools, and the multiple pathogens addressed in this database (allowing for genetic modules to be developed for other pathogens in the future). FAO consulted the RVF specialist members of Vmerge to explore whether such a genetic module would be useful to their scientific needs. The interest for such an initiative was fully confirmed but with the caveats that a larger proportion of long sequences and even whole genome sequencing should be generated to enable more accurate phylogenetic studies, and that the annotation of the RVF virus would also be needed (this step should be performed by SIB in the near future and will be published in ViralZone).
FAO first conducted extensive desktop work to identify linkages between recorded RVF events in EMPRES-i (all species, including humans) and deposited RVF sequences in ViPR. After analysing the roughly 2000 RVF events in EMPRES-i against the 898 RVF genetic sequences in ViPR, it was determined that many of the listings could be linked through similarities of species, date, and location. Importantly however, discrepancies were discovered and suggestions made to improve data on both sides in order to facilitate linkages in the future. The sharing of unpublished RVF genetic sequences from external will be encouraged in the future, and once publicly available, the linkage between these sequences and matching events will be released online through the EMPRES-i Genetic Module (currently only available through ViPR).
For Culicoides-borne viruses, we developed a complete framework to: 1) confirm that opposition between Mediterranean and non-Mediterranean climates are an important change driver in Culicoides communities, 2) describe the structure of Culicoides imicola at different spatial scales and develop new genetic tools, 3) characterize spatial and temporal overlap of wild and domestic hosts and various Culicoides species, and 4) produce European risk maps for BTV and SBV.
Culicoides species communities and population structure. The strategy was first to describe the evolution of Culicoides community species along a transect from Mediterranean areas to non-Mediterranean areas, then to describe the structure of Culicoides imicola populations as the main vector in Mediterranean areas, and to develop genetic tools which will allow the further study of Culicoides obsoletus and Culicoides chiopterus populations, as important vector species for non-Mediterranean Europe.
We first confirmed that BTV transmission areas could be separated into two distinct epidemiological systems where vector communities are distinct: the Mediterranean areas where Culicoides imicola is dominant and non-Mediterranean areas where communities are dominated by Culicoides obsoletus/Culicoides scoticus associated to Culicoides dewulfi and Culicoides chiopterus in north-western Europe.
Then, we characterized, using a multi-locus approach including maternally and bi-parentally inherited markers, the colonization history of C. imicola, exploring the demographic, historic and evolutionary events shaping the current distribution of C. imicola at different scales: within the current distribution areas of this species, within the Mediterranean basin, and more locally between France and Spain. More specifically, we depicted the geographical pattern of genetic variation of C. imicola in the Afrotropical region and in the Mediterranean basin by a phylogeographical approach. We then characterized the genetic structure using Bayesian clustering and traditional population genetics tools. We finally performed approximate Bayesian computation analyses (ABC) in order to retrace and date the colonization events. We also combined entomological surveys, movement simulations of air-borne particles, and population genetics to reconstruct the chain of events that led to a newly colonized French area nestled at the northern foot of the Pyrenees.
This allowed us to propose a new hypothesis explaining the current distribution of C. imicola within the Mediterranean basin. This hypothesis supports an ancient presence of C. imicola in the Mediterranean basin along its African and European coasts. This invalidates the consensual hypothesis of a very recent a northward expansion of C. imicola. The reinforced entomological surveillance, following BTV outbreaks, would have only revealed the presence of long-settled populations. Other factors than the presence/absence of the vector have driven the spread of BTV in the Mediterranean basin such an increase of the population sizes observed in the western Mediterranean in connection with an increase of the temperature.
At a more local scale, entomological surveys conducted from 2008 to 2012 indicate the establishment of C. imicola populations in the in the French Pyrenees-Orientales department, suggesting a close genetic relationship to the nearby (~80 km south) populations settled in Catalonia. Instead, the newly settled C. imicola population was shown by both nuclear and mitochondrial genetic loci to be closely related to far more distant populations (360 to 1,000 km east or south-east) in the Var department, Corsica, Sardinia and Algeria. Corsica was further supported as the most likely source of introduction by the ABC analyses, suggesting that establishment of C. imicola in Pyrénées-Orientales could have occurred through long-distance dispersal from abundant populations in the island (> 500 km from the mainland sampling site). Simulating the movement of air-borne particles evidenced frequent wind-transport events allowing, within at most 36 hours, the immigration of midges from north-eastern Spain and Balearic Islands, and, as rare events, their immigration from Corsica at dates corresponding to the period of abundance of this species.
Finally, we developed microsatellite markers for C. obsoletus and C. chiopterus, as important BTV vectors in Europe, to address their genetic diversity at different scales. This will also quantify gene flows and dispersion abilities of both species. Culicoides obsoletus is an ubiquitous species, able to breed in a range of habitats, in both farm holdings and other landscape types. In contrast, C. chiopterus is a specialist species that has only been isolated as larvae from cattle dung. This provides a contrasting ecology that we hypothesize will have a significant influence on dispersal behavior. The markers were tested for C. obsoletus on three populations collected in England, in France and in Balearic Islands. Individuals were identified morphologically and by PCR species-specific diagnostic assay. In the population from France, we observed many non-amplified samples and some allele sizes outside of the normal range at some loci. All samples were hence sequenced for a portion of the mitochondrial genes cytochrome oxidase subunit I. This population was actually composed of C. obsoletus sensu stricto but also by the recently described "dark obsoletus" and by C. obsoletus of the O2 group which are certainly distinct species. Culicoides montanus specimens were also detected in the Spanish and English populations. The following analysis were carried out on the C. obsoletus s.s. only. For 8 loci, a significant heterozygote deficit was observed which for four loci could be explained by null alleles. A genetic structure was highlighted differentiating the populations from France and England and from the population of the Balearic Islands. A Wahlund effect may thus explain some of the observed heterozygous deficits. Here again we will continue the technical development of these markers and test them on a larger number of populations. The markers were tested also for C. chiopterus on three populations collected in France. For 9 loci there is a significant heterozygote deficiency which for four loci can be explained by null alleles, i.e. mutations in the nucleotide sequences of flanking regions that may prevent the primer annealing to template DNA during amplification of the microsatellite locus by PCR. No genetic structure could be evidenced. It will therefore be necessary to continue the technical development of these markers and to test them on a larger number of populations.
Interaction between domestic and wildlife hosts and their vectors. A risk framework was developed to explore how the spatial and temporal overlap of wild and domestic hosts and various Culicoides biting midge vector taxa may affect vector-borne disease risk across three European landscapes. We particularly focus on the risk posed by the bluetongue (BTV) and Schmallenberg (SBV) viruses which recently emerged in Europe and caused huge impact on the livestock sector.BTV has long been considered as endemic in wild ruminants in large parts of Africa and North America and wild ruminants can act as maintenance hosts. The role of European wild ruminants in BTV and SBV transmission and maintenance began to be investigated only following the arrival of exotic strains of these viruses in Northern Europe and is still under debate. Given the diversity of wild and domestic ruminants and midge species involved in transmission of both BTV and SBV in Europe, it is critical to consider how, where and when these actors in the transmission cycle will coincide and how this may be modulated by environmental variability, such as drought and flooding regimes.
In this research we develop predictive environmentally-driven models for the seasonal abundance, group size and group frequency of wild deer species in three regions of Europe, and overlay these predictions with those from a statistical model predicting the seasonal abundance of key vector species in the same locations. These overlays allow for a landscape level examination of spatial variability in potential disease risk, accounting for host and vector affinities for different habitats, meteorological variation and other environmental factors.
We demonstrate considerable spatial and seasonal variations in the risk posed by BTV and SBV in three regions of Europe. Specifically:
- Simulated drought and flood conditions had significant impacts on the densities and number of groups of wild ruminant hosts and also on the length and intensity of adult Culicoides biting activity, with the latter varying between landscapes, between Palearctic versus Afro-tropical vector taxa and between habitats.
- In particular, drought was predicted to reduce the density and frequency of red deer groups per 9km2 landscape, probably due to the reduced forage availability and quality. Flood reduced deer densities probably because saturated soils reduced the growth of dry-adapted grasses but could sometimes increase the number of groups because it can increase vegetation productivity in more elevated areas favoured by red deer.
- For Culicoides vectors, wetter conditions reflecting flood tended to produce adult seasons that were longer with larger peaks in abundance, whilst dry conditions reflecting drought had the opposite effect, with both effects being habitat dependent. Adult seasons were generally longer with larger peak abundances in less open habitats i.e. in grassland versus open shrub or savanna for C. imicola in Spain and in forest or crop-mosaic habitat versus grassland in northern Europe. A potential biological explanation for this finding may be that the semi-aquatic breeding sites used by immature Culicoides may dry out more rapidly in more open habitats.
Considering the Scotland landscape, the risk framework identified:
- Late June to early September as a key risk period where high adult midge vector biting density coincides with high density and group sizes of the major wild ruminant reservoir for BTV, red deer, particularly in north and central areas, and when red deer, cattle and sheep all tend to move to higher elevation areas to access higher quality forage or avoid heat.
- That the risk of SBV reproductive losses is low in an average year because the risk period for in-lamb ewes in mid-September occurs just prior to the red deer rut when adult vector abundance is low.
- However, vector abundance can be high in some areas and years during this period, making delays in the timing of breeding of sheep in Scotland potentially advantageous.
For the Netherlands landscape, the risk framework identified:
- A higher potential for SBV to circulate between sheep, deer and Culicoides during the second seasonal peak of adult Culicoides activity (July-October) because adult vector season extends for longer after the start of the risk period for SBV reproductive losses in in-lamb ewes.
- Under conditions of increased wetness, the vector season is extended much later in the year, which is predicted to increase SBV impacts by increasing the exposure of red deer during the rutting season and exposure of livestock prior to be moved indoors for the winter. This suggests that livestock should be moved indoors earlier in a wet year if avoidance of SBV impacts is of high priority relative to other health and nutritional constraints on management.
- Conversely, drier conditions reduce the magnitude of the second seasonal peak in adults and the end adult vector activity earlier, vastly reducing the risk of reproductive losses from SBV.
- Geographically, the greatest risk of SBV and BTV is predicted to occur in the central region of the country, in which large predicted group sizes for red deer and seasonal adult vector abundance coincide with high densities of sheep and cattle.
For the Spanish landscape, the risk framework identified:
- A higher potential for SBV to circulate between sheep, deer and Culicoides during the second seasonal peak of adult Culicoides activity (July-October) than in Scotland and under wetter conditions because adult vector season extends for longer after the start of the risk period for SBV reproductive losses in in-lamb ewes. Drier conditions again reduced late adult vector season activity of C. obsoletus complex midges, again lowering the risk of SBV impacts
- A significant risk period for transmission of BTV or SBV to cycle between livestock and roe deer in April and early May when roe deer graze on pastures to exploit greening up vegetation after the overwinter dry season though vector abundance is relatively low at this time.
- The spatial pattern of predicted relative risk of transmission differed considerably between the two vector species, with SBV risk for the C. obsoletus complex concentrated in the southwest and northern, more mountainous region of Andalucia, and risk for C. imicola tending to be greatest in central Andalusia.
From snapshot sampling of the midge vector communities in sheep, red deer and roe deer habitats in Scotland, the C. obsoletus/scoticus taxa was identified as having the largest potential for disease transmission amongst wild and domestic hosts, being present in all three host habitats in large numbers. The abundance of C. pulicaris was also high over all three habitats, representing the potential for this species to also act as a bridge vector among wild and domestic hosts. These results highlight the ongoing need for more ecologically-driven estimates of disease risk in European landscapes that capture changes to critical eco-epidemiological processes under different environmental conditions. Such work will allow for more effective and pro-active disease control measures to be developed, tailored for specific landscapes, habitat types and climatic scenarios.
Riskmapping with virus transmission models of BTV and SBV. We have used the models as published in Fisher et al, 2013 and Hartemink et al. 2011, to prepare a method for the creation of risk maps. We assume that SBV and BTV behave very similarly, since they are transmitted by the same vectors and affect the same host species. Other partners in the VMERGE project created methods to extrapolate from vector data collection to predictions on vector abundance beyond the sampled locations, using geographically detailed information on aspects like soil type, temperature, humidity, land use, etc. This map of predicted vector abundance is input for the risk map for transmission of vector-borne infections, when combined with the transmission models. Using the available information, we derive risk maps for BTV-8 and SBV in Europe. The risk maps for other BTV serotypes are different, since other Culicoides species are driving their transmission. In figure 11 a risk map for an average temperature of 16 degrees Celsius is shown. Exact scaling of the risk is not possible due to lack of data. The results are therefore all given in terms of relative risk.
To facilitate an integrative approach between all partners, different services were provided including access to quality data, optimization of sampling design for collection campaigns, training in spatial modelling, modelling software, and assistance to upscale regional vector distribution models. Research outputs have included upgrading modelling software, spatial distribution models for Rift Valley fever vectors, livestock movement models, and hotspot framework overlapping hosts and Culicoides. Automated risk map scripts were produced to ensure regular updated maps. Dissemination included the maintenance of a website to promulgate information to sister projects and public.
Spatial Services and Data Management. Provision of spatial data and customization of the data to VMERGE needs were provided to project partners through the VMERGE data website (http://www.VMERGEdata.com/). This website ensured access to large number of spatial datasets amounting several terabytes supporting the needs of other work packages. As examples, was produced a very extensive updated archive of remotely sensed climate indices derived by Temporal Fourier Analysis from MODIS imagery for 2001 – 2015 for day and night time land surface temperature; and for normalised and enhanced vegetation indices. These covariates represent the foundation of much of the spatial modelling undertaken in VMERGE and in other sister projects. New spatial distribution presence absence models of potential hosts, including horses and roe deer, as well as updates of presence absence and abundance indices for red deer for use in hotspot mapping (see below). These were produced as ensembles of Boosted Regression Tree and zoned Random Forest models, implemented by the VECMAP software suite. All data outputs are deposited on the VMERGE data website either for direct access or access on request. A mosaicked and processed dataset of MODIS derived of vegetation phenology indicators for 2005-12. Access to all archived was controlled by a password mediated registration process so that VMERGE partners were able to download public domain datasets customised and standardised for partners, as well as many datasets initially restricted to VMERGE partners only, including those mentioned above. Crucially, these data will continue to be accessible to all VMERGE project partners (and indeed to partners from other completed FP7 projects such as EDENext, IDAMS and EDEN) for a number of years as funding has been allocated from ongoing projects to maintain the data site, and to expand it into new topics and areas under the names www.palebludata.com and for generically www.spatialdatasite.com.
Upscaled RVF models. Important gaps of information remain about the distribution of potential vectors of RVF in Africa. The objective of this task was to upscale RVF models through the generation of habitat suitability maps for Aedes and Culex in Morocco, Tunisia, Mauritania, and Senegal. Maps generated can be used to target surveillance areas (WP4) and to inform transmission modelling (WP3). In total 445 sites distributed in Morocco (partners #12), Tunisia (partners #15), Mauritania (partners #16), and Senegal (partners #13), were visited to detect mosquitoes. Mosquitoes of the genus Aedes and Culex were detected in 115 sites and 272 sites respectively. The data were merged and used by partners #7 to produce species distribution models. Additional Aedes and Culex presence data compiled during a systematic literature review on RVF vectors (Versteirt et al., 2013) were combined to the data collected during the project of VMERGE. Using a Mahalanobis distance mask, data collected in areas with distinct environmental conditions than those observed within the four African countries involved in VMERGE project were withdrawn. Based on the combined number of occurrence data, four species were retained for species distribution modelling: Ae. vexans sI (n=122), Cx. pipiens (n=338), Cx. theileri (n=142), and Cx. tritaeniorhynchus (n=139). Random forest modelling technique was used to estimate the distribution of the four species. For this approach, balanced datasets for both presence and absence training data were required to avoid bias towards predicting presence or absence. The absence data was constituted by a random subset of visited sites in the VMERGE countries and pseudo-absences locations generated by the method of surface range envelope, as suggested by Barbet-Massin et al. (2012), within the boundaries of the Mahalanobis distance mask.
To predict the occurrence of the four mosquitoes, predictor variable datasets from the VMERGE data repository (http://www.vmergedata.com) were used; they include environmental eco-climatic variables and anthropogenic variables such as urbanization. Using the VECMAP® software, a random forest of 600 trees was trained for each species dataset (Breiman, 2001). The random forest models produced probability maps of occurrence based on environmental conditions. When conditions are suitable, the probability values are high, indicating where the species would be able to establish after introduction. The training data were only taken from areas that are similar of the VMERGE countries; therefore, environmental conditions homogeneous within the VMERGE area are not designated as important variables, while in fact those conditions are important for the species to occur. For Cx. pipiens, the probability of occurrence is presented in Figure 12. The spatial distribution was determined by temperature and urbanization (AUC = 0.95).
Mapping hotspots of interaction between Culicoides and domestic and wild ruminants. The objective was to develop a novel, continental-scale framework for identifying and mapping key geographical hotspots of overlap between wild deer hosts, domestic ruminant hosts and Culicoides biting midge vectors of Schmallenberg (SBV) and bluetongue viruses (BTV) across Europe, assuming that these hotspots would be most favourable for establishment and spread of transmission. As well as domestic ruminants, transmission of these Culicoides-borne viruses in Europe involves both Afro-tropical and Palaeartic vector species and diverse wild ruminant species. The hotspot framework was developed by overlaying suitable habitat for deer species and abundance patterns of Culicoides vectors and susceptible and reservoir livestock species derived from environmentally-driven, Europe-wide spatial models.
Livestock movement analysis. The objective was to generate a model, which quantifies the factors influencing animal movement patterns at the national and international levels. This model might be used to predict the probability of a trading connection between any spatial units and, consequently, could be used to inform disease transmission models. At the international scale, the dataset of animal movement was of too low quantity and quality to allow its used to train a predictive spatio-temporal animal movement model for the region of interest of Vmerge. At a regional level, however, a large and detailed spatio-temporal data set of animal movement records was assembled by partner #1 in collaboration with partners #16 in Mauritania and #13 in Senegal. For Mauritania, the trade network included 2203 trade connections between 71 national nodes and 42225 possible paths linking these nodes. Population data on human and livestock distribution was estimated by extracting populations from the Worldpop database (human population) and from the Gridded Livestock of the World v.2. The primary analytical framework consisted of the use of gravity models, where the levels of movement are determined by the size of the populations and origin and destination. This type of model is frequently used in migration studies to predict movement patterns between spatial units (Garcia et al., 2014). A great advantage of gravity models is that they can easily be fitted to observation data sets using linear models, and that they can be used for predictive purposes to establish levels of flow between locations. However, they do not allow characterizing the structure of the trade network, so they are frequently used in complement to social-network analysis metrics. A strong seasonal pattern was observed in the animal movement data, corresponding to an important shift between the means of transport between movements on foot and movements by truck, those latter mainly during the Tabaski period. In both cases, models predicting the probability of a trading connexion between two spatial units were produced with high accuracy.
For Mauritania, there was no noticeable difference between the use of Euclidian and least-cost distance as predictor of the probability of a movement connection. This pattern is typical of a trade induce movement – animals are moved more to areas with many livestock because that is where the markets are, and the distance is a deterrent as the potential losses are greater for longer movements. Once validated against the trade movement data collected in Senegal, this model could be used to predict the probability of a trading connection between any spatial units and inform disease models by providing a direct quantification of the transmission risk over long distances through trade at different points in time within a year.
Software Development. VECMAP®, a system developed by Avia-GIS (http://www.vecmap.com Figure 27), was used by the partners from Morocco (partners #12), Tunisia (partners #15), Mauritania (partners #16), and Senegal (partners #13). Based on their experience in using the system, new relevant features have been integrated into VECMAP®. Those new features have been tested successfully and validated during this project. The following sections present the new features included in VECMAP®. VECMAP® can create sampling strategy adapted to budgets constraints and allocate sample sites to the closest team of field staff. To support surveys done in Mauritania (#16) and in Senegal (#13), VECMAP® was upgraded by adding features enabling longitudinal sampling. To facilitate further the collection of field data, VECMAP® offers an automated approach for smartphones and tablets. Using the built-in GPS, every sample location can be recorded ensuring accurate data location. Furthermore, downloaded maps can be accessed offline. It also contains precise instructions for the field operators, telling them exactly what to look for, where to look for it and what to record. To standardise data as much as possible, various field types (textarea, radio buttons, barcodes, comboboxes, lists, etc.) were constructed to facilitate fieldwork and data processing. The collected data can be sent to the server when online or extracted when no Internet is available. In the context of the VMERGE project, an update mosquitoes list with African species was incorporated in VECMAP® ensuring correct classification of the vectors identified. Identifications are logged and stored centrally in the VECMAP database. Another new feature was a tool allowing to export species data stored as text file.
After spatial analysis, the training dataset is used as input for species distribution models that are developed via a wide range of statistical methods. These models are predictive; filling in distribution gaps and predicting species occurrence probability and abundance in locations on the map where no sampling has taken place. During the VMERGE project, one of the modelling techniques, the Generalised Linear Model, has been updated to handle spatial autocorrelation. It is now possible to assess the degree of coefficient variability using a generalised geographically weighted regression (GGWR). Two remaining options handle spatial autocorrelation in different ways: the autoregressive term (RAC) and the GLSM (mixed model). In addition, a zoning option was developed and the user can choose whether to make the model using pre-defined eco-zones. To improve the user friendly graphical interface, it is now possible to apply a default legend to model output.
Automatic procedure to update risk maps. The objective was to develop an automatic procedure to produce updated risk maps on a monthly-basis for Senegal and Mauritania (see above for modelling description). Results showed that the original climate-based model used by FAO for East Africa performed well for years of well above-normal rains, particularly along the Senegal river and in areas located in south-western Mauritania. However, it was not able to identify hotspots for potential RVF amplification in semiarid areas of the Ferlo region, particularly for years with irregular rainfall patterns. The scripts developed for identifying RVF potential areas based on the two models, i.e. a) the calibrated climate-based model (e.g. average cumulative NDVI anomalies) and b) the dry spell model. The scripts were written in Python and in JavaScript for their application in GEE environment and use of near-real time images. Five python scripts have been written to produce hotspots for vector amplification and potential RVF epizootic, and the scripts were imported into ArcToolBox. Based on the definition of dry spell given in the previous sections, one python script has been written and imported into ArcToolbox, to identify dry spell events during the wet season. Recently, both the climate-based model and the dry spell model have been implemented in GEE to generate updated, near-real time RVF risk maps. FAO is focusing on this activity beyond the Vmerge project. The application demo is shown in Figure 30. The demo is currently available for the GEE community.
One of the major concerns of the Vmerge project was to strengthen interactions between research and surveillance actors, to ensure translation of the scientific productions into policy, guidelines and manuals to disseminate and update Vmerge-generated knowledge.
The analysis of EU legislation in force for RVF and CBD and its consistency with the epidemiology of the diseases have been performed as well the national contingency plans in place in some southern EU countries have been assessed in collaboration with the Mediterranean Animal Health Network – REseau MEditerranéen de Santé Animale as REMESA. To assess the existing surveillance systems for RVF, a specific questionnaire has been designed and drafted aiming at collecting information on the surveillance systems in place in some Northern African countries. An important preparatory work has been done with through the release of a systematic review on published compartmental models on RVF transmission dynamics.
Furthermore, a preliminary spatial analysis to track the climatic events able to favour RVF occurrence over North Africa was conducted. A methodology to track the climatic events that can favour RVF risk over North Africa has been identified (Cyril Caminade et al., 2014): a rainless period lasting at least for a week, followed by large precipitation (at least above 10mm) at the end of rainy season might favour the risk in triggering RVF outbreaks. High numbers of these events, called embryogenesis risk events, occurred in Mauritania in the years of RVF epidemics (Cyril Caminade et al., 2014). This approach has been applied to the whole Africa (the Maghreb included) for the years 2008 – 2015 (these maps were based on data collection from 2008 to 2015 and further analysis of the climatic events that might have favoured the risk to trigger RVF outbreaks in Mauritania in 1998, 2003, 2010 and 2012 from previous studies). The result was the production of yearly maps reporting the sum of embryogenesis events and maps with the embryogenesis events constrained to animal presence (population density greater than 1/km2). These findings allowed the identification of geographic areas and periods of the year more at risk for RVF persistence in case of introduction, thus representing a support for planning early warning activities. These maps should be coupled with pragmatic suggestions for early detection of RFV and other vector-borne diseases mainly in unaffected North African countries by implementing risk-based surveillance approaches to have reliable data at the least cost possible (simple random selection vs convenience sampling). These results can be used for the planning of risk-based surveillance. Linked to this activity, with respect to the assessment of the risk of RVF and Culicoides-borne infections spread from sub-Saharan countries, three targeted areas (in correspondence with oases) were defined in Morocco, serum samples were taken from 3 sentinel sheep flocks (one each oasis) containing 15 sentinel animals each. Vector abundance dynamics for Culicoides and for mosquitoes were assessed by trapping activities. Traps used were of the blacklight type for Culicoides and Carbon dioxide traps for mosquitoes. Tests were carried out by Biopharma in collaboration with IZSAM. At the end of the project, in November 2016, laboratory findings were still in progress and the activities on vectors were still ongoing because the peak of the vector season was not finished yet.
All these activities are functional to the development of proposals for risk-based surveillance programs to be activated in at risk countries. Several workshop and training sessions were carried out.
Many capacity building activities were carried out within Vmerge, to develop capacities in Vmerge beneficiary countries and veterinary services.
Training and capacity building to Vmerge beneficiaries. The first training course consisted of two parts: an entomological part that focused on how to capture and identify mosquitoes, and a second part that focused on how to use the VECMAP mobile app to record data in the field. This face-to-face course was organized at three different locations: Rabat (period: 3-7 June 2014; language: French), Rosso (period: 9-13 June 2014; language: French) and Alexandria (period: 11-15 May 2015; language: English). Afterwards, continuing support was provided to the partners to select sampling sites and finalize fieldwork logistics (mobile app, materials, etc.).
The second VECMAP course provided an initiation to species distribution modelling using VECMAP Lite software (http://www.vecmap.com/vecmap-lite-version). Preliminary to this course from March to April 2016, a distance learning training on geographic information system was provided to the project partners, which could not attend the face-to-face GIS workshop. From the 9th to the 13th of May 2016, the 2nd VECMAP course was delivered face-to-face in Antwerp (Belgium). The partners from Morocco (partners #12), Tunisia (partners #15), Mauritania (partners #16), and Senegal (partners #13) participated to this training (Figure 26). This workshop was very practical focusing on mapping and modelling using three techniques implemented in VECMAP Lite, i.e. Non-Linear Discriminant Analysis, Generalised Linear Model and Random Forest. The first day was launched by a general presentation of VECMAP Lite and the datasets provided for the different exercises. Those datasets included two types of response presence/absence datasets: a subset of a real dataset on Anopheles labranchiae and different virtual vector datasets (benchmark spatial test dataset) geographically localised in the four countries of the participants and created specifically for this workshop. Those datasets were created with the goal to provide a better understanding on the impact of sample size and prevalence on the vectors distribution models. In addition, the participants received 110 environmental raster layers and were introduced to satellite imagery and spatial resolution. The second and third days were dedicated to theoretical and practical presentations of the three spatial modelling techniques mentioned above. Using the datasets provided, the participants gain experience in the modelling techniques, and they reach a better understanding of the challenges related to modelling process and the advantages and disadvantages of each technique. For the two last days of the workshop, the participants could handle the modelling techniques on the virtual vector datasets and on their own datasets and generate risk map showing the probability of vector presence.
Training and capacity building to veterinary services. Following training needs assessment in the region (largely through REMESA) and considering the already-planned trainings and workshops for the region (by FAO or USDA) it was agreed that capacity building focused on veterinary services in order to improve preparedness and disease reporting (or syndrome reporting) from the field. Within the capacity buildings framework, the following activities have been conducted: a training course on epidemiology was carried out in Tunis from 8 to 10 October 2015, back to back with a REMESA meeting. The training focused on the preparedness of at-risk countries in Northern Africa, namely the development of contingency plans and surveillance for early detection (see next section). FAO together with other partners (IZSAM and CReSA) conducted the training in French. The targeted group was mostly REMESA network, including all REMESA focal points (Mauritania, Tunisia, Morocco and Algeria). Participants from Egypt and Senegal attended the training. Although not planned in the principle, a course on “Introduction to Risk Analysis” was carried out in Egypt, Cairo from 23 to 25 February 2016.
By accessing stakeholders across various institutional levels, the greater understanding of vector ecology and vector-borne diseases attained through Vmerge will allow for the development of better prevention and control strategies, as well as tools to fight the disease in affected and at risk countries. Furthermore, the dissemination of data within the project consortium will lead to more robust and structured research work. The path of transferring knowledge generated through academia into practical awareness via policy change is not a straight-forward or short-term venture. To disseminate the achievements of Vmerge at the country level, a variety of trainings, manuals, guidelines, and other tools have been adapted into realistic strategies and epidemiological scenarios in the different countries via 1) mapping of projects, 2) translating results into policy, 3) disseminating results.
FAO has compiled and categorized past, ongoing and future projects on Rift Valley fever (RVF) and the involved vectors into one database. The database was developed as an excel file, where projects were classified by category in separate spreadsheets. The focus of attention lies on projects from around the year 2000 until the planned future ones. This RVF database was completed by sharing it with project partners and external experts/institutions in June 2014. The final file was published online on both the FAO and Vmerge websites (http://www.fao.org/ag/againfo/programmes/en/empres/news_050614.html). FAO released the third version of this dynamic database in July 2016, which was updated after sharing with the RVF community created from the stakeholder list (~ 300 members).
For details on dissemination actions see the ‘potential impact’ section below.
Potential Impact:
Consolidation of collaborations among European and African partners improves the international integration of research on important aspects of the epidemiology of pathogens burdening human and herd health in Africa and in Europe.
We produced a better understanding of the disease emergence process for selected viruses at the level of vector populations and their ecosystem, including the development of innovative methods and tools for pathogen detection, and disease surveillance. Final results on vector competence would have impact in both pure and applied fields. Collectively, the findings of the “cofactors of competence” programme has substantially increased our understanding of how common variations in the environmental conditions experienced by insect vectors might result in differences between estimates of competence achieved under normal laboratory conditions and those exhibited by vector populations in the field. This suggested in turn ways to improve the design of laboratory vector competence studies as well as the extent to which the estimates obtained during previous studies might require adjustment. More accurate measures of the competence of field populations of vectors have increased our ability to predict the potential for establishment and spread of exotic arboviruses in Europe. We expect from transmission modelling risk maps for Rift Valley fever (RVF) and Culicoides-borne diseases (CBD) in Europe and northern Africa and a forecast model for RVF set up and formalized to provide near real-time predictions for regions at risk of RVF virus occurrence, thus forming the basis for an early-warning system for RVF and CBD in Sahelian, northern Africa and Europe.
Up to now, 20 papers have been published or are under the revision process. More than 30 talks were given by Vmerge partners. Many other papers are in preparation. So, these research and experimentation results will continued to be disseminated in the scientific community through both Vmerge websites in the aim to disseminate data and information to sister projects and the wider public, thereby drawing attention to the project itself.
Moreover, one of the major Vmerge aim was to translate research results to policy to contribute to setting up the surveillance methods and tools. We thus expected to provide national veterinary services with practical recommendations and standardized methods and tools for surveillance of vector-borne diseases, and to provide international organizations with science-based proposals to improve control of vector-borne diseases, resulting in safer livestock trade and better food security, as well as better protection of public health.
In this context, capacity building activities have been planned and performed for veterinary services: a first training on epidemiology was held in Tunisia from 8 to 10 October 2015 and a risk analysis course was held in Egypt from 23 to 25 February 2016. The two-day training in Tunis was organized by the Food and Agriculture Organization of the United Nations (FAO) and focused on preparedness for early detection and contingency planning of RVF. The training was attended by 30 participants, including all REMESA (REseau MEditerranéen de Santé Animale) network focal points from Algeria, Mauritania, Morocco, and Tunisia. REMESA was created in 2009 to generate a common framework for work and cooperation, having the necessary capabilities to assist and coordinate the development and implementation of animal health regional projects and programs. The specific objective of REMESA is the improvement of prevention and control against the major transboundary animal diseases and zoonoses through the strengthening of the national and regional resources and capacities, and the harmonization and coordination of surveillance and control activities. Also present was one representative from both the veterinary services of Senegal and the OIE in Tunis. The trainers came from Vmerge partner institutions, namely the FAO in collaboration with the Centre for Agricultural Research for Development (CIRAD) in France, the Centre de Recerca en Sanitat Animal (CReSA) in Spain, and the Istituto Zooprofilattico Sperimentale dell'Abruzzo e del Molise "Giuseppe Caporale" (IZSAM) in Italy.
This training, highly participative, covered the topics of RVF surveillance, disease recognition, establishing an early warning system, and contingency plan development. Participants had the opportunity to learn from the experience of both Mauritania and Senegal in their responses to RVF outbreaks and ongoing surveillance, prevention, and control programs. A desktop simulation exercise on RVF was also organized.
At the end of training, participants expressed need for:
(i) the development of contingency plans for RVF;
(ii) clear guidance on the vaccination strategy;
(iii) the availability of vaccines;
(iv) capacity building for conducting risk modelling to determine the main risk areas (Tunisia, Morocco, and Algeria).
FAO also reminded its commitment in assisting countries in disease surveillance. In this regard, the development of a RVF surveillance manual is underway.
A manual was produced to provide professional and para-professionals with information needed to design and implement effective animal health surveillance for RVF in a One Health context. The manual emphasizes a risk-based approach to surveillance and utilizes the principles outlined in the OIE Guide to Terrestrial Animal Health Surveillance for designing and implementing surveillance programs. The statements in the manual are intended to provide guidance and should not be treated as prescriptions. Appropriate surveillance procedures at the national level depend on the local epidemiological conditions, production systems, culture and institutional arrangements.
Currently FAO has a manual on Preparation for RVF Contingency Plans published in 2002 that provides information on the nature of RVF and the principles and strategic options for its prevention, control and elimination. However, with the increased risk of vector spread into new regions, the importance of surveillance cannot be over emphasised. Furthermore, the lessons learnt from RVF outbreaks since the manual’s publication in 2002 and the emergence of One Health both signify the need for an updated guide. The manual has the goal of disseminating prevention, surveillance, and control knowledge to all regions susceptible to RVFV (East, West, and North Africa plus the Middle East). By increasing the resilience of livelihoods to threats and crises, its goals were based upon the successful fulfilment of the following priorities:
- Priority 1 - Improving the capacity to formulate and promote risk reduction and crisis management policies, strategies and plans;
- Priority 2 – Setting up and/or improving the mechanisms which identify and monitor threats and assess risks, and to deliver integrated and timely early warning; and
- Priority 3 - Capacity development to support institutions at national and regional levels.
The purpose of the manual is to provide professional and para-professionals with information needed to design and implement effective animal health surveillance for RVF in a One Health context. As a contextual foundation for surveillance, the manual provides general information on the disease and the causative agent, RVF virus. The background data stresses epidemiology and determinants of risk.
The manual suggests appropriate surveillance objectives and activities for a range of countries according to their epidemiological status. The goal is to aid countries to develop risk-based surveillance systems that are fit-to-purpose and cost effective. In line with lessons learnt during recent outbreaks, the manual highly recommends the use of participatory epidemiology tools to engage all stakeholders to achieve improved disease surveillance. Participatory surveillance is an evidenced-based approach to surveillance that involves diverse stakeholder groups to identify problems and formulate solutions. The approach taps in to traditional knowledge systems and networks. The public often have local language names for common disease problems and are aware of when and where diseases are occurring. This knowledge frequently includes environmental, clinical and epidemiological characteristics of the disease. Participatory surveillance is action-oriented approach for the collection of epidemiological intelligence to inform decision-making. These approaches rely on cooperation among farmers, animal and human health care providers, and governments.
Given the zoonotic and vector-borne nature of the disease, RVF surveillance is ideally approached from a One Health perspective. Surveillance systems need to cover a range of human, animal and environmental issues and the risk of an RVF outbreak can only be assessed in a trans-disciplinary manner. Although the manual focuses on animal health surveillance, it attempts to do so in the context of an integrated surveillance and response system.
To be accessible to a large target audience (ex: both national level veterinarians and para-vets), it was decided that the wording and length should not be a barrier, and the scope broad (while remaining within the focus of surveillance). The manual is approximately 40 pages and contains the following sections (see attachment).
Proper dissemination of the manual is crucial to its relevance in combatting RVF. Initially, the manual will be available in only an electronic version, and will be translated from English into French, Swahili, Portuguese, and Arabic. The primary channel will be through the FAO website in the FAO Document Repository where all FAO public documents are available. A database to directly contact RVF stakeholders has also been created in order to distribute the RVF Newsletter. The Stakeholder database contains approximately 300 members, including all VMERGE partners, as well as individuals from various types of institutions, including research, international organisations, NGOs, governments, the private sector, and private individuals working directly on RVF or in RVF affected regions. After subsequent distribution of the newsletters, requests from colleagues of listserv members receiving the emails has kept the database updated and constantly expanding. Lastly, the manual will also be available on the Vmerge website. At this time, FAO is still exploring the need for a printed version of the manual.
A workshop on RVF concerning its transfer, modelling, forecasting, as well as animal mobility and vector surveillance was conducted as part of the project activities of the REMESA on the 6th October 2016 Tunis, Tunisia. A policy brief document “Enhancing Surveillance and Early Detection of RVF” was written based on the presentations and followed discussions. This policy brief recognized that differences in RVF risk of infection occur between (and within) infected and non-infected countries in the region. Furthermore, those countries are of different levels in managing the risk of RVR infection or incursion. For all of these issues, the work of Vmerge applied an approach in which each national member of REMESA is encouraged to develop a national RVF surveillance strategy that is supportive of the greater regional effort. It is not a top –down prescribed approach but a though mapping exercise on how to build a RVF taskforce in the REMESA region to adapt and mitigate the risk of RVF in a sustainable manner. The real work should begin in depth to start understanding all the outlines of the vector-borne animal diseases in the region, which has been greatly influenced under Vmerge.
The connectedness and interdependence of animal and human health demand full embrace of the One Health concept as a strategy to prevent the emergence and global spread of vector-borne pathogens. Any sustainable approach must include building relevant capacity, both human and technical, and implementing intervention programs at the source of potential infection and epidemics. Any decisions for dealing with these emerging diseases should take into account the major causes (rather than just the consequences) and also the types of re-emergence in question. There can be no single response. The purpose of this policy brief was to highlight recent advances in the fields of epidemiology, detection, and the prevention of diseases transmitted by arthropod vectors.
The design of control strategies for vector–borne diseases should be guided by research into emergence mechanisms in order to understand how a wild cycle can produce a pathogen that goes on to cause devastating urban epidemics. In addition to studying the emergence mechanisms, combining field work with basic research, we must also: improve surveillance techniques by developing tools for rapid pathogen detection; find alternative means of prevention by testing vaccines against arthropod bites; and propose new control strategies, without neglecting to improve existing anti-vector control techniques. Answers to such types of question are essential for gauging the present risks of disease, predicting future threats, and formulating methods for monitoring and control. Compared with many other diseases, the ever increasing threat of vector-borne diseases represents a great challenge to public and animal health managers. Complex life cycles, changing distribution ranges, a variety of potential vectors and hosts, and the possible role of reservoirs make surveillance for vector-borne diseases a grave concern in a changing environment with increasing technical constraints. Surveillance activities may have various specific objectives and may focus on clinical disease, pathogens, vectors, hosts and reservoirs, but ultimately such activities should improve our ability in a timely manner to predict, prevent and control the diseases of concern.
This surveillance therefore becomes a serious issue for local and national organizations and is being conducted more and more at the regional and international level through multidisciplinary networks. With financial and logistical constraints, tools for optimizing and evaluating the performance of surveillance systems are essential. The continuous development of mapping, analytical and modelling tools provides with an enhanced ability to interpret, visualize and communicate surveillance results. This also demonstrates the importance of the link between surveillance and research, with interactions and benefits in both directions.
Passive surveillance. The Purpose of RVF surveillance is intended for use in outbreak management, to inform disease control measures, to inform trade policy, and to prioritize further surveillance activities. The specific purpose of the surveillance system as a whole, and of each activity, should be carefully considered and expressly stated. The aims of surveillance for VBD may include one or more of the following:
- Early detection of outbreaks;
- Detection of cases to enable the timely implementation of control measures;
- Detection of vector-borne pathogens in hosts and /or vectors;
- Describing the distribution and level of disease or pathogen circulation;
- Monitoring space-time trends in levels of disease or pathogen circulation;
- Determining the geographic distribution of known or potential vector populations; and
- Mapping transmission risk.
In addition, surveillance activities may also contribute to improving our knowledge of the disease and to help inform and refine further surveillance by:
- identifying risk factors and determinants of pathogen occurrence and transmission;
- generating hypotheses regarding potential pathogen reservoirs or vectors; and
- identifying risk factors for the emergence in non –endemic areas.
Depending on the objectives of a surveillance program, data may be collected about one or more of the following parameters:
- Seroprevalence in host populations, including seasonal and annual changes;
- infection prevalence in hosts (and if warranted in vectors as well);
- Species composition, abundance and temporal/spatial variation in vector populations;
- Density of hosts and spatial and temporal variation in those populations;
- Incidence of clinical disease, mortality or other health –related outcomes in host populations;
- Presence and level of potential risk factors for disease, related to the agent, host, vector or environment.
Early warning surveillance. This is an important component of surveillance for RVF and it is widely used, since it potentially allows for the detection of multiple known and unknown emerging diseases. Early warning surveillance often forms part of comprehensive early warning contingency plans that also include many other components.
Sentinel surveillance. Sentinel surveillance employs repeated sampling subjects at selected sites, which act as proxies for the entire population. The sites may be randomly selected or located in areas where the probability of disease detection is increased (risk-based sampling). To detect recent seroconversion, young animals, free of maternal antibodies, are sampled at weekly to monthly intervals. Sentinel surveillance can also be deployed at a regional scale as demonstrated in West Africa for RVF, where a network of sentinel small ruminant herds was located in potential high-risk areas for RVF in Mauritania and Senegal.
Risk-based surveillance. Risk-based surveillance aims to increase the efficiency and cost-benefit ratio of surveillance by employing risk assessment methods to prioritize surveillance requirements in either the sampling or analysis of the surveillance data. Risk-based sampling methods are built into most surveillance systems for RVF, particularly those aimed at the early detection of disease, since preferentially sampling those strata of the population, or certain geographic areas, considered being at the greatest risk of disease occurrence increases the likelihood of early detection, and improves the cost-effectiveness of the system.
The risk assessment may be based on expert opinion; the presence of known risk factors and the outputs of qualitative or quantitative risk analysis that could be used to inform risk-based surveillance is an assessment of the risk of transporting pathogen. It could include for example some approaches to improve a risk-based surveillance system by selecting relevant animal movements for risk-based sampling. This significantly increased the sensitivity of the surveillance, when compared to completely random sampling.
Spatial analysis and modelling. Advances in mapping and geographic information system have made it easier to communicate surveillance data in various visual formats to aid understanding and facilitate decision-making and have also allowed more complex spatial and spatio-temporal analysis to identify risk factors for RVF occurrence (or prediction). Identifying high-risk areas is one of the important outputs of a surveillance programs that can then be fed into the planning and implementation of prevention and control measures, as well as used to refine further risk-based surveillance activities.
Information on vector distribution can be compiled into risk maps, indicating potential high-risk areas for disease. These risk maps can be further refined by incorporating data on other known risk factors, as well as previous disease outbreak data. A spatial risk model is used to estimate vector presence or abundance, or RVF presence or incidence, in a particular geographic area, with the outputs generally displayed in map format, as a risk surface.
Evaluation of surveillance systems. The development of new methods using up-to-date tools and scientific knowledge is crucial for RVF surveillance. We have to take into consideration that specific local conditions, rapid economic development and a plethora of management options in a financially restricted environment result in animal health managers facing situations where the wrong decisions could lead to hazardous consequences.
Surveillance and research. Surveillance is by nature interdisciplinary, hence collaborative networks and systems must be developed to facilitate the planning and co-ordination of surveillance activities, the integration of surveillance and research, efficient data management and analysis, and the communication of outputs to decision-makers and other stakeholders. It is also crucial to increase surveillance activities in regions where new vector-borne diseases are more likely to emerge.
Finally, dissemination actions were organized by FAO. These actions communicated project outcomes that are judged to be of interest to stakeholders outside the consortium, including disease updates, early warning messages, reports, scientific publications including those concerning Mauritania or Senegalese studies, training materials, manuals, guidelines, announcements of Vmerge project events and their outcomes, outbreaks (or their updates) including the 2015 Mauritania outbreak, and climate maps.
Audience. Stakeholder database contains approximately 300 members, including all VMERGE partners. They were emailed the newsletters using the FAO administered account rvf@fao.org. The database contains individuals from various types of institutions, including research, international organizations, NGOs, governments, the private sector, and private individuals working directly on RVF or in RVF affected regions. After subsequent distribution of the newsletters, requests from colleagues of listserv members receiving the emails kept the database updated and constantly expanding. The database may be used in the future to quickly access RVF stakeholders (see attachment).
Review process. Newsletters are timed according to the available information that needs to be shared. They are often timed around alerts or new RVF outbreaks (as reported by ProMED-mail or OIE) or when Vmerge partners requested specific news to be disseminated. Accompanying but less time sensitive RVF related news were then sourced, such as recent publications, before all topics of the newsletter were summarized and distributed to the Vmerge consortium for further input. After Vmerge partners provided input and or further news to share the newsletter was disseminated.
Route of dissemination. Stakeholders were sent the final newsletter in HTML format, which was subsequently advertised via EMPRES, FAO’s VPH newsgroup, and finally posted on the Vmerge website’s ‘publications’ section (http://www.vmerge.eu/publications/rvf-newsletters).
List of past newsletter items:
- FAO compiles past, ongoing, and future projects on Rift Valley fever and its vectors
- RVF training with REMESA network
- RVF human cases in Mauritania
- Euro-Aegis has produced a virtual species distribution map (SDM)
- Summary of climate update (19 Oct 2015)
- African climate forecast (5 Nov 2015)
- FAO and its partners are putting an effort together to improve Rift Valley fever preparedness in the Maghreb
- First RVF outbreaks in Mauritania reported to the OIE
- Update on FAO activities on RVF in East Africa
- Analysis of RVF early warning systems in place in the EU
- Africa climate forecast (15 Nov 2015)
- Rift Valley fever newsletter provides outbreak, forecasting, and project updates on RVF
- FAO helps countries prevent and control RVF
- RVF risk factors in central Sudan
- EMRPES Watch: Africa – El Nino and increased risk of Rift Valley fever – Warning to countries
- USDA Emerging Health Risk Notification, 20 Dec 2015. El Nino and Rift Valley fever, East Africa
- Climate update for RVF in East Africa (8 January 2016)
- RVF update for East Africa prepared by NASA, including rainfall hotspots, vegetation anomalies, and the RVF risk map (5 Feb 2016)
- Serological evidence of exposure to Rift Valley fever virus detected in Tunisia
- RVF update for East Africa prepared by NASA, including rainfall hotspots, veg anomalies, and the RVF risk map
- Chimpanzee Adenovirus vaccine provides multispecies protection against Rift Valley fever
- Climate update for RVF in East Africa (4 March 2016)
- Media reported RVF disease event in Uganda
- RVF paper: Current challenges and future prospects
- FAO compiles past, ongoing, and future projects on Rift Valley fever and its vectors
- OIE confirmed RVF event in Ugandan goat
- Rift Valley fever in kidney transplant recipient returning from Mali
- Complete genome sequence of Rift Valley fever virus strain Lunyo
- Geographically weighted discriminant analysis of environmental conditions associated with Rift Valley fever outbreaks in South Africa
- Culex pipiens and Stegomyia albopicta (=Aedes albopictus) populations as vectors for lineage 1 and 2 West Nile virus in Europe
- Possible Human RVF Case in Uganda
- Risk or Loss Function and Quality Measures for Rift Valley Fever Vaccine Production
- China confirms human case of RVF imported from Angola
- FAO updates RVF mapping tool
- Summary of RVF publications released as of June 2016
- RVF event in Niger
- Summary of RVF related publications released since August 2016
- Bulletin Epidémiologique sur la Fièvre de la Vallée du Rift : WP 6.5 Vmerge Project
- Two highlighted papers focused on RVF
- Likely spread of RVF outbreak from Niger to Mali
- Final Vmerge meeting
- 13th REMESA Joint Permanent Committee Meeting
- RVF risk maps
- Summary of presentations given at the final Vmerge meeting held
- Update on RVF Outbreak in Niger and Mali
List of Websites:
www.vmerge.eu
Coordinator: Thomas Balenghien, thomas.balenghien@cirad.fr
