SBV seroconversion was greatest in the south and south east of Ireland. In the initial 2012 survey, 100 % herd-level seroprevalence was recorded in four counties and >75 % in a further five counties within this region. In the former four counties, animal-level seroprevalence ranged from 85 to 99 %. Both animal and herd-level seroprevalences tended to diminish in the more north western parts of the country (Fig. 3a). Factors such as local vector populations, vector species competence, topography and ambient temperature are known to influence the infection rate among farm animal species. Midge biting rate, extrinsic incubation period and vector mortality rates are all known to be temperature dependent, and all influence the transmission of a vector borne disease [18]. The availability of suitable vector habitats and suitable hosts and the characteristics of the pathogen are key determinants in facilitating the transmission of a vector born disease [19]. One possible explanation as to why seroconversion was confined to the south and east was that SBV had been introduced relatively late in the vector season. The limited geographic extent of spread of SBV in Ireland at the end of 2012 may be attributable to the timing of incursion, which most probably occurred in the latter half of the vector active season. At the end of the first vector season after SBV was introduced into England, a similar pattern was evident in that seropositive sheep flocks and clinical cases were confined to the south and east of the country [8, 18], whereas in the Netherlands, herd prevalences in all regions ranged from 90 to 100 % at the end of the first vector season [4]. A Scottish model has shown that SBV introduction late in the vector season in climatic conditions similar to Ireland markedly reduces the spread of infection compared to an introduction earlier in the vector season [14]. The reasons for the reduced transmission include the shorter duration of the remaining vector active period, the bimodal distribution of Culicoides activity, and the lower temperature later in the vector season.

There was little evidence of further spread of SBV over the course of the second vector season after viral introduction (Fig. 2a, b, c). Indeed, the results of the serological surveys conducted during the latter half of summer 2013 and late autumn 2013 were little different from those obtained from the initial 2012 serosurvey. Based on the experience in several other European countries, it had been anticipated that SBV transmission would resume with the commencement of the 2013 vector active season, leading to an extension of the geographic distribution of exposed herds in Ireland towards the north and west. However, this anticipated spread never materialised, and suggested reasons for this relate to adverse weather conditions in the spring of 2013, the virus failing to reactivate after the winter of 2012 and herd immunity. The spring and early summer of 2013 were very cold, wet and windy in Ireland [20], which is likely to have delayed the resumption of midge activity [22]. SBV has successfully overwintered, despite lengthy periods of minimal vector activity [21]. Data from Germany has shown that midge activity during cold weather is minimal [23]. Another possible explanation for the lack of spread in 2013 is “herd immunity” attributable either to previous exposure to the virus or the use of SBV vaccines. The high rate of seroconversion achieved in the first vector season in affected counties would have resulted in a high frequency of immunity in the second season [24]. If there was a further incursion of SBV in 2013 from either the UK, or mainland Europe, into a previously infected region, transmission could be limited due to widespread pre-existing immunity. It is unlikely that vaccination had any significant impact as vaccine uptake among Irish cattle farmers was low, because the vaccine only became available after the end of the spring breeding season (and the second serosurvey presented here) and could not have affected seroconversion in the first and second serosurveys.

The remote geographical location of Ireland is one of the most important factor in limiting the incursion and spread of vector-borne viral diseases. For example, while Bluetongue spread rapidly and widely throughout the ruminant populations of north-western Europe in 2006/2007, including into the south-east of England [25] there was no evidence of incursion into Ireland during that outbreak. Ireland’s remote location means complex rare meteorological events are likely to be required to facilitate the wind borne introduction of a vector borne disease into the country. A Scottish SBV model indicates that introduction of SBV must occur relatively early in the vector season to bring about extensive spread [14]. The model also showed that mean Scottish summer temperatures (similar to Irish summer temperatures) facilitate only limited spread as SBV is so temperature dependent. Assuming this model is applicable to Ireland, for extensive spread of SBV by a wind borne vector to occur requires a relatively rare meteorological event to occur early in the vector season in a year with above average temperatures. While these events can occur, the fact that all three are necessary reduces the probability of SBV being successfully established in Ireland.

The spatial distribution of SBV seroconversion was correlated with the spatial distribution of confirmed SBV cases in malformed ovine and bovine foetuses (Fig. 4). The pathological surveillance strategy was active and risk-based, with the objective of detecting cases, from a biased set of samples where SBV was suspected at gross post mortem examination. While the serological based surveillance was also active, it was not risk based, but was designed with the objective of estimating prevalence and/or proving freedom from infection from a truly random sample of the population. The pathology-based surveillance had correctly identified the regions most affected by SBV (Fig. 3). The key concern for pathological based surveillance is the ability to detect cases of SBV malformation, while the purpose of the serological surveillance was to provide an accurate quantification of the extent of SBV exposure to be extrapolated nationally. While the two strategies were complimentary, they were not directly comparable.

The serological sampling strategy involved the opportunistic real-time selection of serum samples submitted for the national brucellosis serological surveillance programme, as this was considered the most cost effective and efficient way to access serum from cattle throughout the country. The parameters set out in the sampling plan were conservative from a herd-level perspective. In designing the sampling frame, it was assumed that if SBV was present in a herd, there would be relatively high within-herd prevalence as had been indicated by serological data from several European countries. The within herd prevalence was set at 70 % which is relatively low compared to the levels of seroconversion seen in exposed herds in other countries. The herd level prevalence was set at 1 % to detect a low level herd prevalence. While there is always the potential for bias with an opportunistic sampling strategy, this was addressed in part by selecting herds which had already been randomly selected for the national brucellosis eradication programme. The consistency in the results between the three serological surveys, in itself, suggests that there was limited bias in sample selection. Where single seropositive animals were identified in herds within those regions with low herd prevalence this may have been the result of moving animals that had already been exposed from the high prevalence areas of the south and south-east rather than local vector-borne spread. A previous study has shown there is considerable movement of cattle within Ireland [26]. In separate serological studies to assess SBV exposure in sheep flocks, follow-up of a single seropositive sheep in a county with low levels of seroconversion, showed that it had been purchased from the south east [27]. Such movements rather than “false positive” test results are a more likely explanation of singletons given the reported specificity of the serological test method that was used.

In future serological studies on SBV in Ireland, it would be prudent to sample animals born in or after 2013 in the high prevalence areas disclosed in this study, to determine if infection is present and active in those areas. Several studies have shown that cattle are more attractive to biting midges than other ruminant species, which makes them the ideal sentinel species for a midge borne disease. In one study 83 % of midges collected originated from cattle and 17 % originated from sheep, and almost 50 % of the midges originating from cattle were blood engorged versus 7 % of the midges which originated from sheep [28]. The present study has shown marked regional variation in SBV seroprevalences. Any future prevalence studies of vector borne disease could be informed by this gradient information and adjust/set their sampling frames on an intelligent regional basis rather than on a single national basis. It would be prudent to continue national serosurveys and the targeting of animals expected to be immunologically naïve, i.e. 2013 and 2014 born animals in the high prevalence areas in this study, to monitor SBV in the future. SBV is a relatively low impact disease, but shares a vector with pathogens that may potentially be of much greater significance. The patterns revealed and understanding gained with SBV should prove transferrable to other arboviruses. The epidemiology of SBV in its Culicoides vector is also worthy of further investigation in an Irish context. Understanding the ecology of various midge species and their competence as viral vectors will be very important in assessing the risk of viral introduction into and spread within Ireland in the face of any future midge-borne viral epizootic in Europe.