Bluetongue disease virus (BT) is a vector-borne, infectious but non-contagious animal pathogen. This emerging disease affects all ruminants and has been responsible for an unprecedented continuing European epidemic which has been occurring for the past decade 1. Belonging to the Orbivirus genus and Reoviridae family, there are currently 24 recognized serotypes transmitted globally by a multitude of Culicoides midge species, each with their own habitat preferences and geographical distribution albeit with some overlapping occurrence. Several serotypes, mostly affecting sheep, have been circulating in the Balkan and Mediterranean areas since the late 90's, which could be predicted by the advance of its Old World vector C. imicola. Yet, an outbreak of BT serotype 8 (BTV-8) in 2006, which was last recorded in the African and the Caribbean region 2, suddenly occurred in Northern Europe, an area previously free of BT infection 3. Preceding this event, outbreaks had been reported on a regular seasonal basis in southern Europe, mainly the Mediterranean region involving several serotypes namely BTV -1, -2, -4, -6, -9 and -16 1.

At the time of writing, the first occurrence of BTV-8 in Switzerland was reported in northern Switzerland in the canton of Basel-City in late October 2006. This was shortly followed by cases in the canton of Solothurn, and Basel-Land respectively. Further cases of BTV-8 were consecutively detected in Basel-Land and Solothurn. In January 2007, a surprising case was discovered in the southern canton of Valais. In the first outbreaks of BT ever recorded in Switzerland dating from October 2006 to February 2007, a total of 12 cattle and two goats tested positive for BTV-8 originating from seven different farms. Although no firm conclusion have yet arisen as to the cause of these cases, wind direction patterns along with temperature records of the affected areas suggest likely intrusion of infected vectors originating from the surrounding BT-reporting areas.

Due to the dynamics of the pathogen, combined with the fact that the geographical limits of other vector-borne diseases are also expanding, Switzerland conducted a nation-wide survey in 2003 to determine the status of BT disease and the presence of its vectors 4. Additionally since 2005 5, studies conducted have recorded the presence of the following potential BT vectors: C. obsoletus (Meigen), C. pulicaris, C. scoticus and C. dewulfi, species which have all been implicated in BT outbreaks in several countries located in Europe and the Balkan areas 367

Although this resulted in proving freedom of infection, the presence of vector species competent of transmitting BT were found to be abundant in various areas of the country. This in turn prompted the establishment of a sentinel herd surveillance system through serological and entomological monitoring, focusing on certain areas of the country (Fig. 1) 89. Due to the nature of vector-borne diseases, and the fact that BT was still absent in Switzerland, risk based sampling was implemented 9. This risk-based design involved identifying geographical areas which matched habitat criteria for the presence, survival and establishment of a vector species 58, which were identified through GIS mapping techniques 10. However, these maps were limited to annual analysis, and did not differentiate different levels of vector activity throughout the year, nor did they consider rainfall and snow data in certain months. Hence a more informative and detailed mapping method alongside a mathematical model was created.

The aim of this study was to combine GIS maps of data collected in the field with maps of predicted basic reproduction number estimates for potential BT outbreaks in Switzerland, in order to explore the spatial and temporal regions more prone to 1) the establishment of important vector populations, and 2) enabling the spread of the disease due to the nature of geographical and climatic features.

Our results showed that the months of July and September 2006 were the most suitable period in Switzerland for vector presence in terms of climatic conditions (suitability maps), while the highest R₀ value indicating vector activity and therefore spread, occurred in June. This is an interesting finding in relation to the BT incursion and subsequent outbreak in Northern Europe, which was first detected in August 2006, possibly indicating similar conditions in the affected countries, such as The Netherlands, Belgium and Germany. The R₀ values estimated in this study were used as a risk indicator levels for vector activity in different geographic areas (Fig. 4 middle) in terms of BT transmission rates. They were then combined with maps indicating areas with predicted higher vector suitability likelihood (Fig. 4 left). The actual cases of BT correspond to the maps generated for 2006, yet due to the data originating from this year, cannot be replicated for future outbreaks of BT. Similarly to the mapping methods for climate, R₀ values were classified into 4 risk levels, signifying most rapid spread of disease in areas presenting higher range of R₀ values.

The basic reproduction number (R₀) for vector-borne diseases is a more complex number to calculate due to the influence of seasonal fluctuations 16, local climate and environmental features as well as the abundance of breeding sites available near hosts which affect vector dynamics 14. The transmission rates of the disease will also change depending on temperature factors affecting vector to host or host to vector interaction, along with the extrinsic incubation time, biting rates and vector mortality rates 15. In this study, vector density numbers along with temperature values were used from field data collected throughout the project in order to produce R₀ values specific to the areas and time frame studied. The values which were most affected in terms of location and time period were those of vector density, vector death rates and extrinsic incubation rates. These were mainly affected by differences in temperature and humidity levels which in turn are very specific to the various microclimates present in Switzerland targeted in the sampling. Concerning the effect of host availability on transmission rates, since data was taken from sentinel farms where minimal transhumance activity took place, cattle density was considered as stable, with a Swiss average size of 30 cattle per farm.

A recent study by Gubbins et al. 15 assessed the risk of BT in the United Kingdom using the basic reproduction number R₀. Similarly to their findings, the R₀ values were highest when temperatures were between 15°C–25°C. Whilst the mentioned study took into account both the ratio of vectors to cattle and sheep in the formula, our study only considered the relation of BTV-8 in cattle farms and not sheep due to much larger cattle population of larger ruminants present in Switzerland. Additionally in financial terms, the cattle industry and more specifically the milk sector is an important part of the Swiss economy http://www.blw.admin.ch.

For Switzerland, the peak of R₀ can be explained by the fact that the largest amount of Culicoides midges in 2006 were recorded in June, probably due to optimal breeding and hatching conditions as May was a mild and humid month. July held the record for maximum temperature, which leads to higher vector activity and successful virus transmission. However high temperatures also increase vector mortality rate and thereby lower R₀ values for this month. Due to the very different meteorological patterns in the past three years, we would therefore expect that maps and R₀ for 2007 (comparable to conditions in 2003) to be quite different compared to 2005 and 2006 in terms of increased vector activity and BT transmission rates. http://www.meteosuisse.admin.ch/web/fr/climat/climat_des_derniers_mois/Jahresbilanzen.html.

Affected by the initial high temperatures recorded in July 2006, transmission values as well as vector density decreased significantly following a notably milder August. In terms of the Northern European BTV8 outbreak data, this would match reports stating that the maximum number of cases occurred in October which would originate from a high number of vectors present a few weeks previously. Due to the nature of Culicoides development, it has been suggested that cases occur at a time lag of circa four weeks from peak vector density periods, which corroborate the evidence from the trapping data and the transmission values 17. As seen in The Netherlands which recorded high temperatures in July, this climatic event could have primed the vectors in terms of their competence and capacity levels, translating in an effective transmission period, as observed in many other affected countries 17. As mentioned in a study conducted by Murray 18, rapid changes in climatic conditions can affect vector population age structure, along with vector density and alter transmission rates of disease.

A limiting factor in our model is that vector density data resulted from trapping sites representing high risk areas for entomological surveillance, located in southern parts of Switzerland. In these locations, high numbers of vectors were expected. Therefore, the transmission values may be overestimated in some parts of the country, yet due to the targeted nature of setting up sentinel herds in high risk areas this limitation was acceptable in this scenario.

Another factor to consider is that only the dynamics between BTV-8 and the vectors belonging to the Culicoides obsoletus group were studied, and considering that a majority of the biological parameters used in calculating values for R₀ were derived from the available literature, this should be taken into account when applying such methods to different countries, especially when micro-climates or other BT vectors are present. Outbreaks involving other BT serotypes and their dynamics in Culicoides vectors produce different R₀ values 19. They may also have a different preference for distinct geographic and climatic conditions. Such differences have been described for the behaviour of BTV-2 and Culicoides imicola in southern France 20, and the role of C. imicola in South Africa 21.

The recent cases of BT in Switzerland in the month of October 2007 occurred in the northern part of the country, where temperatures for the affected region were similar to those of 2006, and the number of cattle affected on the seven farms correspond to the R₀ figures calculated for that area and time period. Out of a total of 608 susceptible cattle, 16 sheep and six goats, originating from the farms located in Basel-City, Solothurn, Basel-Land, Valais and Jura, BT was detected in a total of 12 cattle and 2 goats, with prevalence rates ranging in chronological terms of 10.3% to 0.6% in cattle and 33% in goats http://www.bluetongue.ch. The R₀ values provided by the model were similar to the actual prevalence rates for the time of the year, with the exception of goats, which were not included in the study.

Our findings highlight the potential for establishing a flexible surveillance system taking into account environmental factors. In a targeted surveillance system, this could mean increased serological testing during a specific warmer period or in specific geographical areas. Given the scarcity of epidemiological data available for Switzerland concerning BT cases prior to this study, the creation of thematic and risk maps on a monthly and annual basis, illustrated the variability in the behaviour of vector borne diseases and the possible consequences of virus introduction. It also provided the basis for creating a surveillance system targeting higher BT-risk regions and months. In the case of Switzerland, the maps and R₀ values were used as input parameters for the creation of a BT surveillance Scenario Tree model 25, with the aim of comparing the efficacy of alternative surveillance system designs. A risk-based surveillance program was implemented in July 2007 consisting of three surveillance system components; serological bulk milk testing of 200 sentinel herds located in areas considered of higher risk to BT occurrence, as well as clinical surveillance programs for cattle and sheep farmers throughout the country.

GIS mapping techniques combined with statistical and mathematical models can help improve disease surveillance and control methods by providing a basis for targeting monitoring efforts. Targeted surveillance in this scenario meant focusing financial and monitoring efforts to pre-defined areas originating from the maps created in this study.

An advantage of GIS methods includes the ability to improve prediction maps once more comprehensive field data has been collected, and adjust surveillance efforts in a timely and accurate manner. Flexible surveillance programs should be used in order to attribute financial and human resources to high-risk areas considering temporal and spatial factors.

In this study, maps were used to highlight areas in Switzerland which presented higher likely risk for vector habitat, which incorporated climatic elements, (Figure 4. left margin), combined with maps highlighting different vector activity levels based on R₀ values for BT (Figure 4. middle column), in order to produce a set of final combination maps representing areas in Switzerland most likely to have outbreaks of BT due to vector presence in relation to vector activity levels on a monthly basis for 2006. As mentioned, the different risk zone distribution originating from the final combination maps were used as input parameters in a Scenario Tree model to aid in decision making processes concerning BT surveillance in Switzerland http://www.bluetongue.ch.

Approaches using GIS maps, and the combination of spatial analysis and mathematical modelling as predictive tools have been used in other countries such as Italy, Spain, and France concerning BT and its vectors 222324, and stress the advantages of using different technological methods in supporting surveillance efforts, as well as for other purposes aiding in the improvement of health and prevention of disease both in public and veterinary terms.

The authors declare that they have no competing interests.

VR and KS conceived the idea for the study. VR conducted all analytical work and wrote the manuscript. GV provided expert opinion on entomological aspects, CG provided expert advice on BT disease. All authors read, edited and approved the final version of the manuscript.