In South Africa, bluetongue (BT) occurs annually during late summer and autumn in areas with high rainfall (Verwoerd & Erasmus 2004). Outbreaks of BT usually stopped abruptly after the first frosts, resulting in apparent BT-free winters during which no clinical cases are observed (Verwoerd & Erasmus 2004). The causative virus, bluetongue virus (BTV), is therefore believed to be absent from the area or dormant in either its vector or vertebrate host. Seasonal recurrences of BTV raise the question whether the virus is re-introduced into an area or re-emerges as a result of an increase in vector capacity. Apparent over-wintering (or absence of cases during winter) of BTV is documented in South Africa (Nevill 1971) as well as Europe (Maan et al. 2008; Osmani et al. 2006; Velthuis et al. 2010), although the exact mechanism is still unknown (Singer, MacLachlan & Carpenter 2001).

Once infected, Culicoides vectors, although relatively short-lived, are usually infectious for the rest of their life span of up to 3 months (Purse et al. 2015; Wilson, Darpel & Mellor 2008). Several arboviruses such as Ross River virus (Lindsay et al. 1993), West Nile virus (Goddard et al. 2003; Reisen et al. 2006) and Orungo virus (Cordellier et al. 1982) have demonstrated the ability to be transmitted vertically by infected arthropod females. Apparent vertical transmission of BTV in the genus Culicoides has been demonstrated in Culicoides sonorensis, the primary North American vector, after BTV nucleic acid was detected in field-collected larvae (White et al. 2005). This study, however, did not demonstrate whether BTV RNA-positive larvae would have emerged and been able to transmit infectious virus as adults and to date there is still lack of evidence for the natural vertical transmission of BTV in Culicoides vectors (Osborne et al. 2015).

In most parts of South Africa climatic conditions are suitable for adult Culicoides midges to remain active throughout the year as temperatures seldom remain at sub-zero levels for extended periods. It is therefore possible that small populations of virus-infected adult midges might survive long enough to bridge the gap between transmission seasons during mild winters. To ensure continuous transmission of BTV, vector-free periods must be of shorter duration than the maximum period of viraemia in the vertebrate population (Mellor 1994). BTV viraemia can last up to 50 and 60 days in sheep and cattle, respectively (Koumbati et al. 1999; MacLachlan et al. 1990; Sellers & Taylor 1980). Cattle have been incriminated as amplifying and maintenance host or cycling hosts in certain parts of the country where more than 95% of cattle had antibodies specific to BTV (Steyn et al. 2015). Most species of wild ruminants and camelids are also susceptible to BTV infection, although they are frequently asymptomatic and are regarded as important reservoir hosts (Maclachlan et al. 2009). Midge-free periods must therefore be longer than 60 days to effectively break the cycle in transmission. Surveys have demonstrated the presence of large numbers of Culicoides midges throughout the year in areas where temperatures rarely drop below 0 °C (Meiswinkel, Venter & Nevill 2004; Venter, Nevill & Van der Linde 1997). In cooler parts of the country, adult midges have been collected during nights where the minimum temperature was as low as -4.5 °C (Venter et al. 2014). Throughout such periods, the ambient temperature is probably too low for midge emergence or viral replication within the vector, which will only occur when the temperature becomes favourable (Carpenter et al. 2011; Mellor et al. 1998; Paweska, Venter & Mellor 2002; Purse et al. 2015; Wittmann 2000).

Adult Culicoides midges have a lifespan of 10–20 days (Mellor, Boorman & Baylis 2000), although studies have demonstrated that their lifespan may extend during mild winter months with some individuals surviving for up to 3 months at 10 °C (Lysyk & Danyk 2007; Purse et al. 2015). It is generally accepted that survival will increase, despite lower relative humidity, with a decrease in temperature (Gerry & Mullens 2000; Mullens et al. 2004). The rate of virus replication and the minimum temperature required for replication are generally consistent for different orbiviruses across different Culicoides vector species (Carpenter et al. 2011). In Europe members of the Obsoletus complex can be active at temperatures as low as 3.5 °C (Glukhova 1989). In the laboratory, Culicoides obsoletus can survive for > 90 days and 10 days at 17 °C and 4 °C, respectively (Goffredo et al. 2004). In South Africa it was shown that Culicoides imicola adults can survive for > 15 days and Culicoides pycnostictus up to 54 days at -1.5 °C (Nevill 1971). Previous studies have also demonstrated the capability of C. imicola and Culicoides bolitinos to harbour BTV for more than 20 days at 10 °C (Paweska et al. 2002; Wellby et al. 1996). Culicoides bolitinos seems to be adapted to cooler temperatures than C. imicola (Verhoef, Venter & Weldon 2014) and able to survive the relatively mild winter temperatures as adults in Gauteng Province.

It was shown that Culicoides midges can, despite relatively low temperatures, occur throughout the colder winter months in some parts of South Africa (Steyn et al. 2015; Venter et al. 1997, 2014). It was also shown that the detection of viable African horse sickness virus (AHSV) in field-collected midges and in C. imicola in particular was linked to relatively large insect collections since no virus was detected in any pool < 1000 midges (Venter et al. 2014). Even during outbreak situations of AHSV the field infection in C. imicola can be as low as 0.003% (Venter, Koekemoer & Paweska 2006a). In this study the objective was to determine whether BTV could overwinter in Culicoides populations in the Onderstepoort area, Gauteng Province, South Africa.

Materials used in the experiment posed no health risk to researchers, and no vertebrate animals were harmed. The project was approved by the research committee of University of Pretoria and funded by the Tshwane Animal Health Innovation Cluster (TAHIC) – Technology Innovation Agency (TIA) (E H Venter Grant number TAHIC12-00035). The collection of midges was financially supported by the Gauteng Department of Agriculture and Rural Development (project number: OV07/23/C231).

This was an extension of a study done by Venter et al. (2014) on the closely related AHSV that demonstrated a relatively low virus isolation rate in Culicoides midges collected at the ARC-OVI (Venter et al. 2014). No AHSV was detected during the winter although continuous adult Culicoides activity indicated that transmission could potentially occur. The detection of AHSV was linked to high midge numbers collected and the absence of AHSV in these midges during winter months was ascribed to the relatively low numbers of midges collected (Venter et al. 2014). Similar to AHSV, the over-wintering mechanism for BTV is unknown. The present study focused on the possibility that Culicoides that survived throughout winter in an area can harbour BTV and represent a possible over-wintering mechanism for the virus.

In total more than 500 000 midges were collected during the 15-month survey with species-level identification revealing 27 different species. As discussed by Venter et al. (2014), the presence of both parous and freshly blood-fed females throughout the year indicated that feeding continued throughout winter. Similar to what was concluded for AHSV, this suggests that transmission of BTV throughout the winter might be possible. The decrease in midge numbers, as determined by light trap collections during winter months, can be ascribed partly to a shift in flight activity. Midges will feed earlier (late afternoon rather than after dusk) to avoid the colder night temperatures, which would influence the numbers collected as the light traps will be less efficient during the day (Meiswinkel & Elbers 2015; Scheffer et al. 2012).

Real-time RT-qPCR was used to screen the samples for the presence of BTV RNA in the Culicoides population. BTV RNA was detected in 91.2% of the midge pools tested compared to 40% AHSV RNA in the same samples (Venter et al. 2014). This higher detection rate of BTV could be due to BTV outcompeting AHSV within the midge, or that BTV is predominant at the ARC-OVI due to the presence of a higher number of BTV host species such as sheep and cattle. Similar results were observed during a 6-year country-wide survey where BTV and AHSV were detected with conventional virus isolation methods (embryonated chicken eggs, baby mice, cell culture) in 11.7% and 1.5% of the midge population, respectively (Nevill, Erasmus & Venter 1992a). The difference in midge pool infection with BTV and AHSV is evident, although lower than the figures for our study. This could be because of the difference in the sensitivity of conventional viral isolation (VI) methods versus real-time PCR, where the latter is much more sensitive.

All but two of the pools that tested positive for AHSV RNA (12/35) also tested positive for BTV RNA (52/57). This demonstrates co-circulation of BTV and AHSV within the midge population and could be indicative of a possible co-infection of Culicoides midges.

Negative midge pools were present in all four seasons (August 2010, October 2010, February 2011 and May 2011). No distinct seasonal pattern in BTV RNA negative Culicoides was observed. Virus or viruses isolated in three of the four midge samples were not identified at the time, as the PCR assay yielded negative results for AHSV in those samples (Venter et al. 2014). All four VI positive samples consequently tested positive for BTV RNA (C_q values ranging from 9.8 to 34.7). Virus isolation positives were only detected during summer and beginning of autumn (December 2010, January to March 2011) and in only one midge pool per month. This could indicate that although BTV RNA is present throughout winter, viruses capable of replicating to infectious levels are only present in summer and early autumn.

Monitoring BTV infection in Culicoides midges trapped at various sites in South Africa from 1979 to 1984, Nevill et al. (1992a) showed that between 14–18 different serotypes will be encountered every season, although at varying frequencies (Gerdes 2004). Usually, three to five serotypes represented more than 60% of the total number of isolates during a specific season. These dominant serotypes are replaced by others the following season, only to regain dominance 3–4 years later (Gerdes 2004). Although this phenomenon does not support the over-wintering in adult midges, it emphasised that a number of factors will play a role in the occurrence and outbreaks of BTV. The frequency of occurrence of specific serotypes will not only depend on the Culicoides midges but will also be influenced by the availability of susceptible mammal hosts. The number of susceptible mammal hosts will depend on previous infection and vaccine use.

Comparing real-time RT-qPCR and VI illustrates the importance of diagnostic techniques used when determining the presence or absence of viruses. Only 7.0% of the samples tested positive with cell culture isolation techniques in comparison with 91.2% that tested positive with real-time RT-qPCR. Virus isolation detects infectious virus if the viral titre is high enough and capable of showing CPE. In contrast, real-time RT-qPCR detects viral RNA even if the virus is non-infectious. Real-time RT-qPCR is therefore important to establish whether the virus is present in a specific area or population even if it is not infectious.

It is evident that Culicoides midges are present throughout the year even though temperatures can fall well below the normal activity range of these insects and that viral RNA is detectable in midges throughout the year. One hypothesis is that BTV overwinters in the midges, and as soon as summer approaches and temperatures rise above 15 °C, BTV starts replicating. BTV then builds up in the midge population to sufficient levels for transmission to occur during summer and clinical disease only occurs in late summer and early autumn (Mellor et al. 2000). The increase in temperature also shortens the period between blood meals, increase feeding rate and leads to larger population sizes. Although detecting nucleic acid does not necessarily imply presence of viable viruses, our findings suggest that BTV may possibly overwinter in midges in the Onderstepoort region. Further studies are required to demonstrate conclusively that this is indeed the case.

Culicoides midges were present throughout the 15 month survey period demonstrating that the insects are able to survive mild winter temperatures recorded at the ARC-OVI, Gauteng Province. The presence of males, freshly blood-feds and parous females during the winter months also indicate that breeding and blood feeding continues throughout winter. BTV RNA was detected in most of the midge pools and viable virus was detected in summer and autumn collections. This study demonstrates that vector-free periods are not present at Onderstepoort and therefore possible overwintering of BTV within midges in this area.