Thermal limits for flight activity of field-collected Culicoides in the United Kingdom defined under laboratory conditions

Background Culicoides biting midges (Diptera: Ceratopogonidae) are biological vectors of internationally important arboviruses and inflict biting nuisance on humans, companion animals and livestock. In temperate regions, transmission of arboviruses is limited by temperature thresholds, in both replication and dissemination of arboviruses within the vector and in the flight activity of adult Culicoides. This study aims to determine the cold-temperature thresholds for flight activity of Culicoides from the UK under laboratory conditions. Methods Over 18,000 Culicoides adults were collected from the field using 4 W down-draught miniature ultraviolet Centers for Disease Control traps. Populations of Culicoides were sampled at three different geographical locations within the UK during the summer months and again in the autumn at one geographical location. Activity at constant temperatures was assessed using a bioassay that detected movement of adult Culicoides towards an ultraviolet light source over a 24-h period. Results The proportion of active adult Culicoides increased with temperature but cold temperature thresholds for activity varied significantly according to collection season and location. Populations dominated by the subgenus Avaritia collected in South East England had a lower activity threshold temperature in the autumn (4 °C) compared with populations collected in the summer (10 °C). Within the subgenus Avaritia, Culicoides scoticus was significantly more active across all temperatures tested than Culicoides obsoletus within the experimental setup. Populations of Culicoides impunctatus collected in the North East of England were only active once temperatures reached 14 °C. Preliminary data suggested flight activity of the subgenus Avaritia does not differ between populations in South East England and those in the Scottish Borders. Conclusions These findings demonstrate seasonal changes in temperature thresholds for flight and across different populations of Culicoides. These data, alongside that defining thresholds for virus replication within Culicoides, provide a primary tool for risk assessment of arbovirus transmission in temperate regions. In addition, the study also provides a comparison with thermal limits derived directly from light-suction trapping data, which is currently used as the main method to define adult Culicoides activity during surveillance.

Culicoides biting midges (Diptera: Ceratopogonidae) are biological vectors of a range of internationally important arboviruses of companion animals, livestock and wild mammals including bluetongue virus (BTV), African horse sickness virus (AHSV) and Schmallenberg virus (SBV) [1,2]. In addition, Culicoides transmit Oropouche virus (OROV), which causes a febrile illness in humans, and play a poorly de ned role in the transmission of a range of other zoonotic arboviruses [3,4]. Within northern Europe, both BTV and SBV have a signi cant impact on livestock production through clinical disease and also ruminant movement controls and trade restrictions imposed to reduce virus spread [5,6]. The risk of major shifts in the global distribution of Culicoides-borne arboviruses is currently considered to be high [7,8], as illustrated by the recent unprecedented emergence of AHSV in Thailand [9].
In temperate regions, such as North-Western Europe, temperature has a fundamental in uence on the transmission and persistence of Culicoides-borne arboviruses [10]. The rate of arbovirus replication and dissemination within a biological vector is usually highly dependent on environmental temperature, as most arthropods are poikilothermic. In the case of Culicoides: BTV interactions, the threshold for infection and replication lies within a range of 11-15°C [11]. It is clear, however, that the temperature threshold for Culicoides active ight is below this range and therefore could allow survival of the virus through winter periods when replication does not occur but when infected adults are able to survive, y and feed on hosts [12,13]. It has also been demonstrated that Culicoides infected with AHSV at 25°C and then immediately held at a constant temperature of 10°C for 35 days develop full infections following a return to 25°C for 3 days [14]. Hence, resumption of temperatures conducive to virus replication could also lead to transmission during transient warmer periods within winter.
Temperature-related thresholds to activity in adult Culicoides can be assessed through direct analysis in the laboratory and through a variety of eld-based methods. Direct studies of activity use laboratorybased bioassays based on a phototactic response while they are held at a known temperature. A recent study in the Republic of South Africa (RSA) examined the major afrotropical vector of arboviruses, Culicoides imicola Kieffer, and measured ight response using a white uorescence light stimulus and horizontal ight over a distance of approximately 15cm [15]. At 10°C and 13°C, C. imicola phototactic response was less than 5% of the insects introduced, while at 17°C, more than 25% responded. At temperatures of more than 17°C, the phototactic response in C. imicola did not exceed 28% of those introduced into the bioassay system. An earlier study, based in Japan, examined activity using a dedicated bioassay system based on a requirement for 8cm of horizontal ight in response to an ultraviolet light source [16]. In both Culicoides oxystoma Kieffer and Culicoides maculatus Shiraki, active ight was recorded in less than 10% of individuals at temperatures below 16-17°C, although a few individuals were recorded as being active even at the minimum temperature of 6°C. In contrast to the study in RSA, over 80% of individuals exhibited a phototactic response at temperatures of greater than 25°C.
Activity of adult Culicoides can also be measured in the eld by collection in parallel with measurement of environmental temperature. The most epidemiologically relevant method of assessing thermal limits to activity is to measure adult Culicoides blood-feeding behaviour on a relevant host. A key issue is that these studies are generally carried out at times of peak adult seasonal abundance, where thermal limits to activity are not usually reached, and are also limited by the availability of daylight [17]. As an example, collections of adult Culicoides from sheep made in the United Kingdom (UK) using a drop-trap design reported all species within the subgenus Avaritia as active across the entire temperature sampling range of 11.8-29.0°C between the daylight hours of 18:00 and 21:30 [17]. Truck trapping has also been carried out in the UK using a vehicle driven over a set course with a net mounted on its roof [18]. This samples populations of adult Culicoides actively ying at a set height of approximately 1.5m without an attractant. Culicoides of the subgenus Avaritia were collected over a range of 8.6-30.0°C, with larger catches associated with higher temperatures, although reductions in activity were recorded when the temperature exceeded 21°C [18]. Both of these approaches are usually only deployed at single sites on a local scale due to their complex logistics.
Thermal limits to Culicoides adult activity can also be inferred from less labour-intensive trapping approaches. Semiochemical-based traps for livestock-associated vector Culicoides in northern Europe have received only limited attention [19], with the majority of research targeting the nuisance biting species Culicoides impunctatus Goetghebuer [20,21]. In contrast, ultraviolet (UV) light-suction traps are used in Europe for both research activities and surveillance of Culicoides during arbovirus outbreaks [22,23,24]. While the intrinsic biases in Culicoides populations sampled through the use of light-suction traps are well known [17,25,26], they are by far the most commonly deployed means of detecting activity due to their ease of use, although data collected from winter periods examining temperature thresholds for activity remains scant.
In France, adult Culicoides activity was inhibited, as measured by an absence of individuals in trap collections, on nights when the maximum temperature fell below 10°C [27], and a similar result was found in Germany where activity was limited to days with a maximum temperature of ≥10.8°C [28]. A eld study in the UK was able to collect adult Culicoides, using UV light-suction traps, in February and March when the maximum temperature for both months was 10.1°C [29]. Results from a large dataset of Culicoides collections across Europe indicate that the onset of the vector season can occur at mean temperatures as low as 1°C in the most northern latitude tested, whereas mean temperatures of 10°C are required for Culicoides activity in countries in the southern latitudes [24]. Key variation in experimental methods include the method by which temperature is inferred (e.g. in situ, using national weather stationbased monitoring or remotely sensed data) and the fact that light-suction trap collections are not representative of biting rate on hosts in this region [17].
Surveillance of adult Culicoides ight periods is currently adopted in Europe to determine the time period during which the probability of arbovirus transmission is very low, de ned as the seasonal vector-free period (SVFP) [30]. Following the BTV-8 outbreak in northern Europe in 2006, the European Commission enforced bluetongue surveillance programs in member states which involved entomological surveys of Culicoides vectors using light-suction traps [31]. Using this surveillance data, a SVFP was identi ed in the winter of 2006/2007 in the UK, enabling over 70,000 susceptible livestock movements [32]. The determination of the SVFP is dependent on the use of light-suction trap surveillance despite their limitations which include the underestimation of host-seeking activity compared to other methods [33,34].
Understanding the temperature threshold for ight activity in Culicoides is critical in evaluating the risk of transmission following an incursion, in addition to understanding their role in arbovirus overwintering in the UK [12]. The implementation of trap surveillance, if not already in place, is time-consuming resulting in a delay of active surveillance from the point of initial incursion. Applying easily accessible temperature data, however, to evaluate the activity rates of Culicoides populations and thus infer the risk of transmission by active Culicoides could overcome the issues associated with post-incursion surveillance. The aim of this study was therefore to determine a cold temperature threshold for ight activity under experimental conditions for eld-caught UK Culicoides species, collected at different seasons and geographical locations.

Collection methods
Insects were collected using down-draught miniature blacklight (UV) Centers for Disease Control (CDC), model 912 traps (John W Hock Co, Gainesville, FL, USA). Each CDC trap uses a 4W UV tube and is powered by a 12V lead acid sealed battery (Yuasa, Japan). At each site, between four and eight traps were used in order to catch su cient Culicoides for the trials and were positioned at least 50m apart [35]. Traps were suspended at a height of approximately 1.5m above ground. Traps were set up at least two hours before sunset and run overnight before collection within two hours of sunrise the following day. Insects were collected live into a 340ml cardboard collection cup containing a cotton pad soaked in 10% sucrose to provide a sugar source and paper towel cut into long, thin strips to provide shelter from the trap fan downdraft. Cardboard collection cups were secured to the CDC trap using a CDC sleeve and elastic bands (Additional File 2: Figure S1). The following morning, sleeves from each trap were removed and tied with a clip-lock tie to contain the insects within the collection cup and sleeve.

Flight activity experiments
Insects collected were transferred and released into a large plastic, black box (42cm (l) x 35cm (w) x 32cm (h)) with a translucent funnel (14cm in length) attached on one side to allow for insects to exit.
Clear plastic tubing (6cm in length) was secured to the translucent funnel to visualise and count approximately 150-250 active, phototactic adult Culicoides exiting the box and entering a ' ight activity pot' (Figure 2). The ight activity pots were modi ed 340ml cardboard collection cups with a small plastic funnel (66mm top diameter; 7.4cm in length) covered in aluminium foil tted at the top, and a ne mesh covering the bottom of the pot. The total height of the ight activity pot from the base of the cardboard cup to the tip of the funnel was 14cm. A cotton wool bung within the funnel retained all insects within the ight activity pot until the start of the experiments.
Once eight ight activity pots had been lled, they were numbered at random from one to eight. Pots one to six were placed into a 'test' temperature-controlled incubator (MIR-254 Panasonic, UK) at an initial temperature of 20±1°C, while pots seven and eight were placed in a second 'maintenance' incubator also set at 20±1°C as a control for activity of Culicoides in the sample. Both incubators contained a UV light source (4W blue-blacklight tube; John W Hock Co, Gainesville, FL, USA) suspended within the incubator that acted as an attractant and no other light was provided within the incubators. The test incubator was then set to the desired temperature (2°C -14°C) and insects were allowed to acclimatise for two hours.
Following the acclimatisation period, the cotton wool bung at the top of the ight activity pot funnel was removed and a 7.5cm diameter pill box (Watkins and Doncaster, UK) with a ne mesh lid was secured on top of the funnel allowing active insects to y towards the UV light stimulus into the attached pill box ( Figure 2E). All ight activity pots were placed with the mesh lid faced downwards onto a tray to ensure darkness, so that the UV light was the only light source available above the ight activity pots. A small cotton pad soaked in 10% sucrose was placed underneath each ight activity pot on the tray, rather than within the pot, so that insects could still access the sugar-meal through the ne mesh but were prevented from sticking to the pad.
At regular intervals the collection pill boxes were replaced (0.5, 1, 2, 4, 6, 8, 10-and 24-hour intervals) and the collected Culicoides were killed by freezing at -20°C for a minimum of 2 hours. All remaining inactive individuals in the ight activity pots were also killed by freezing after 24 hours. The percentage of Culicoides ying after 24 hours was determined by calculating the mean cumulative proportion of active Culicoides across the six ight activity pots maintained at each test temperature. In each temperature trial, collections were also made at the same intervals from pots seven and eight, which were maintained at 20±1°C throughout the experiment to ensure the population tested were active without the constraint of cold temperatures. A ow diagram of the full experimental design is provided in Additional File 3: Figure   S2. In the SES cohort, ve temperatures were tested (6, 8, 10, 12 and 14°C) and the SEA cohort a further ve temperatures were tested (2, 4, 6, 8 and 10°C). Four temperatures were tested in the NES cohort (8, 10, 12 and 14°C) and only one temperature of 12°C was tested in the SBS cohort.

Sample identi cation
Only individuals that were tested in activity experiments were identi ed, all surplus insects that remained in the initial black plastic sorting box were not used for experimental study. Culicoides from the activity experiments were sorted morphologically under a dissecting microscope using characteristic wing patterns with the aid of an identi cation key [36]. Adult Culicoides were grouped into six categories: subgenus Avaritia, Culicoides pulicaris Linnaeus, Culicoides punctatus Meigen, Culicoides achrayi Kettle and Lawson, C. impunctatus and other Culicoides. Female Culicoides were also identi ed to physiological state by examination of the abdomen [37] and assigned to one of the following categories: unpigmented, pigmented, gravid and blood-fed.
For both South East England cohorts (SES and SEA), a sub-sample of females within the subgenus Avaritia, were identi ed further to species level (C. obsoletus and C. scoticus only) using an adapted multiplex polymerase chain reaction (PCR) method targeting the internal transcribed spacer (ITS) 1-5.8S-ITS2 region [38]. All individuals belonging to the subgenus Avaritia from one ight pot from each temperature trial was chosen at random for molecular analysis and used as a representative sample for each population at each temperature trial. Culicoides were transferred to individual reaction tubes with 200µl of tissue digest solution containing 100nM Tris-HCl pH8 (Thermo Fisher, UK), 200mM NaCl (Sigma-Aldrich, UK), 0.2%(w/v) SDS (Thermo Fisher), 5mM EDTA (Thermo Fisher), 200µg/mL proteinase K (Thermo Fisher), and nuclease-free water (Thermo Fisher). Following an overnight incubation in tissue digest solution at 37°C, individual Culicoides specimens were transferred to tubes containing 70% ethanol for storage. The Culicoides DNA was then extracted from 100µl of tissue digest solution and eluted into 100µl buffer using the KingFisher Flex automated extraction platform and the MagMAX™ CORE Nucleic Acid Puri cation Kit (Thermo Fisher) according to manufacturer's instructions.
Two microliters of each sample DNA was added to each well on a PCR plate (Life Technologies, UK) each containing 8µl of mastermix which consisted of 1x TaqMan Fast Advanced MasterMix (Thermo Fisher), 0.3µM of each primer [38], 0.2µM of each probe [38] and diluted to a total reaction volume of 10µl using nuclease free water (Thermo Fisher). Negative extraction controls consisted of elution from wells which did not contain any Culicoides specimen in the extraction plate and negative PCR controls contained just nuclease free water. At least three negative controls and at least three positive controls, using DNA extracted from males from the same study morphologically identi ed as either C. obsoletus or C. scoticus, were added to each plate.
The PCR thermal pro le used consisted of 2 mins at 50°C for activation of uracil-DNA-glycosylases (UDG), an initial denaturation step of 2 mins at 95°C, followed by 40 cycles of 95°C for 3 secs and 60°C for 30 secs and was carried out using an Applied Biosystems 7500 Fast instrument (Thermo Fisher). Each plate was analysed using the ViiA7 Real Time PCR system software (Thermo Fisher). Determination of species for each individual specimen was based on the cycle threshold (C t ) value for each speciesspeci c primer-probe pairing. Negative samples were de ned as having a C t ≥35 and positive samples were de ned as having a C t ≤25. Samples with a C t between >25 and <35 were regarded as undetermined and were repeated. If samples remained undetermined following re-examination, samples were de ned as unknown and were removed from analysis.

Statistical modelling
Generalised linear mixed models (GLMMs) were used to investigate the relationship between temperature and the proportion of Culicoides ying and how this relationship differed amongst cohorts. Speci cally, a binomial family GLMM with a logit link function was constructed with the proportion of Culicoides ying as the response variable. Model selection proceeded by stepwise deletion of non-signi cant (P>0.05) terms (as judged by likelihood ratio tests), starting from a model including temperature (°C), cohort and an interaction between them as xed effects and pot as a random effect (to allow for between-pot variation). The models were implemented using the lme4 package [39] in R (version 3.6.1) [40].
Separate models were constructed for total Culicoides (all Culicoides, unpigmented females and pigmented females), the Avaritia subgenus (all Culicoides, unpigmented females and pigmented females) and C. impunctatus (all Culicoides and pigmented females). Sample sizes were insu cient to examine relationships for: (i) C. pulicaris, C. achrayi or other Culicoides; or (ii) blood fed females, gravid females or males for any species/groups. In addition, the SBS cohort was excluded from this analysis as activity was only assessed at a single temperature (12°C) for this cohort.
Further models were constructed for Culicoides ight activity at 12°C to compare activity amongst the SES cohort, the NES cohort and the SBS cohort. A similar approach to that described above was used, except the model included cohort as a xed effect and pot as a random effect. Flight activity in the cohorts were compared using Tukey multiple comparisons.
For the SES and SEA cohorts, there was a su cient number of Culicoides caught to allow two further analyses. First, to compare ight activity of unpigmented and pigmented females of the Avaritia subgenus a GLMM was constructed including pigmentation state (i.e. unpigmented or pigmented) as a xed effect, as well as two-and three-way interactions between it and the other xed effects (i.e. temperature and cohort). Model selection was carried out as described above. Second, to compare ight activity of C. obsoletus and C. scoticus, a binomial family generalised linear model (GLM) with a logit link function was constructed. The proportion of Culicoides ying was the response variable and temperature, cohort and species were xed effects. Model selection proceeded as described above.

Results
For both the SES and SEA cohorts, a total of ve collections were made to test ve temperatures in each cohort. Each collection, made across multiple sites (sites one to three), was used to test one temperature at a time. In the SES cohort, a total of 5,586 Culicoides were collected across the ve collections to test ve temperatures (6, 8, 10, 12 and 14°C). In the SEA cohort, a total of 5,309 Culicoides were collected across the ve collections to test ve temperatures (2, 4, 6, 8 and 10°C). A total of 7,228 Culicoides were tested in the NES cohort from a total of four collections from site 4 only, to test four temperatures (8, 10, 12, 14°C). One collection was made in the SBS cohort to test one temperature (12°C) in which 585 Culicoides were tested (Additional File 4: Table S2).
In the South East of England, >97% of Culicoides in both the summer (5,463 of the 5,586 individuals) and autumn (5,259 of the 5,309 individuals) cohorts belonged to the subgenus Avaritia. In contrast, >99% of Culicoides in North East England were identi ed as C. impunctatus (6,611 of the 6,643 individuals). There were insu cient numbers collected of other species (C. pulicaris, C. punctatus, C. achrayi and other Culicoides) from all cohorts for further analyses to be conducted on these species. (Additional File 4: Table S2). Collections were also dominated by unpigmented and pigmented female Culicoides and no further analysis was conducted on males, gravid or blood fed females (Additional File 4: Table S2).
In all cohorts where multiple temperatures were tested, the proportion of active Culicoides increased as temperature increased within the range tested ( Figure 3). This relationship was observed for total Culicoides (all adults, unpigmented females and pigmented females), the subgenus Avaritia (all adults, unpigmented females and pigmented females) and C. impunctatus (all adults and pigmented females) in both South East England cohorts (SES and SEA) and the NES cohort ( Figure 3). Moreover, the rate at which Culicoides activity increased with temperature was the same amongst the three cohorts tested (i.e. there was no signi cant (P>0.05) interaction between temperature and cohort) (Additional File 5: Table  S3).
The season and geographical location of Culicoides collection in uenced the temperature at which Culicoides activity began and the levels of activity reached at each temperature ( Figure 3, Additional File 5: Table S3). The cold temperature threshold for activity, as measured by the minimum temperature at which >5% of the Culicoides population is active, was signi cantly higher for Culicoides from the NES cohort compared with Culicoides from the two south east England cohorts, SES and SEA. Collections made from North East England primarily comprised of C. impunctatus, whereas collections made in South East England were dominated by Culicoides of the Avaritia subgenus. In the NES cohort, the cold temperature threshold was 14°C, in which only 9% of the Culicoides population were active (Figure 3). The two South East of England cohorts had substantially higher activity rates at all temperatures compared to NES (Figure 3). Yet, adult Culicoides in the SEA cohort had a lower cold temperature threshold for activity (4°C) compared with those collected in the SES cohort (10°C), though the two populations demonstrated similar activity levels at 14°C (Figure 3). This pattern was similar for both the unpigmented and pigmented females in these cohorts ( Figure 3). Indeed, there were no signi cant (P = 0.66) differences in ight activity between unpigmented and pigmented females of the Avaritia subgenus for both the SES and SEA cohorts. In all temperature trials, Culicoides from the same populations maintained at 20±1°C within control pots were active, with average activity levels in control pots of 86% and 78% for the SES and SEA cohorts respectively, indicating that any reduction in activity is a result of the colder temperatures in temperature trial pots.
A single collection was subsequently made at a site in the Scottish Borders in the summer (SBS) which consisted of a greater proportion of livestock-associated species of Culicoides compared with collections made in the NES cohort, collected less than 5 miles away; a total of only16 of the 6,643 individuals tested from the NES cohort were classi ed in the subgenus Avaritia. Adult Culicoides collected in the SBS cohort had a signi cantly greater ight activity at 12°C, the only temperature tested, compared with Culicoides collected nearby in the NES cohort, when measuring total Culicoides and pigmented females (P<0.001). Flight activity for C. impunctatus speci cally at 12°C, however, was not signi cantly different between the NES and SBS cohorts (P=0. 34). Flight activity at 12°C was not signi cantly different between the SES cohort and the SBS cohort for total Culicoides (P=0.99), total unpigmented females (P=0.73) and total pigmented females (P=0.35) (Figure 4, Additional File 6: Table S4). For adults belonging to the subgenus Avaritia speci cally, ight activity at 12°C was higher for the SBS cohort compared with the SES cohort, for pigmented females only (P=0.007). For all other physiological states within the subgenus Avaritia, activity levels between the SES and SBS cohorts were not signi cantly different.
Within the South East of England, the two cryptic species within the subgenus Avaritia, C. obsoletus and C. scoticus differed in their ight activity (P<0.001), with a greater proportion of C. scoticus than C. obsoletus active at all temperatures ( Figure 5, Additional File 7: Table S5). As seen in the previous analyses, adults of each species from the SEA cohort were more active at lower temperatures than those from the SES cohort ( Figure 5). Additionally, there was a greater number of individuals from the subgenus Avaritia subsample identi ed as C. scoticus later in the season in the SEA cohort (84%) when compared to collections made earlier in the season in the SES cohort (37%).

Discussion
This study has identi ed signi cant differences in temperature thresholds for ight activity of adult Culicoides according to species and season under laboratory conditions. In the populations examined in South East England, which were dominated by Culicoides of the subgenus Avaritia, the cold temperature activity threshold was higher for populations tested earlier in the season (SES: 10°C), in comparison to those collected later in the season (SEA: 4°C), despite similar activity rates being recorded at higher temperatures. In addition, species speci c differences in activity were recorded at an interspeci c level with greater activity being recorded across the temperatures tested for C. scoticus relative to C. obsoletus. In addition, C. impunctatus collected in North East England (NES: 14°C) demonstrated signi cantly reduced activity across all temperatures despite an identical process of acclimatisation and test of ight tness prior to use in the bioassay.
Studies carried out in South East England utilised two separate populations taken from summer (June-August) and autumn (September-October). The number of generations occurring each year in Culicoides in Northern Europe remains unclear, although bi-or trivoltinism has been suggested in populations of C. impunctatus [41] and the subgenus Avaritia [42]. This study highlights a signi cant difference in thermal tolerance of populations sampled in the same year and site using identical methods of selection, acclimatisation and measurement of activity. The underlying mechanism enabling this adaptation is unknown, although previous studies carried out in the United States of America (USA) demonstrated rapid cold hardening in adult Culicoides sonorensis through short term exposure to temperatures of less than 5°C [43]. This study highlights the lack of knowledge that currently exists in the drivers of differences in behaviour, biology and ecology between generations of Culicoides.
Despite this apparent rapid adaptation to activity at cooler temperatures observed in populations of the subgenus Avaritia in South East England, testing of C. impunctatus at a more northern latitude led to equivocal results. While cited as a putative vector of arboviruses [44], this species differs signi cantly in biology and ecology from individuals belonging to the subgenus Avaritia. Firstly, it is autogenous [41,45], mating facultatively and laying its rst egg batch a few days following emergence from development sites without host seeking [17,46]. This results in a ying adult female population that is dominated by parous individuals for the majority of the season, as re ected in the results of the current study where this life stage comprised over 95% of individuals tested. The high temperature threshold for activity in C. impunctatus was primarily driven by a limited ight response in populations tested within the bioassay. This was despite the fact that adults capable of ight towards light following trapping were pre-selected by the experimental design and conditions used were identical to the trials in South East England. The average ight activity response at 20°C for total Culicoides collections across the SES and SEA cohorts was 86% and 78% respectively. In contrast, a greatly reduced ight activity level of 20% was observed in Culicoides populations collected in the NES cohort maintained at 20°C. In addition, it was demonstrated through the trap collection conducted in the nearby Scottish Borders that this was a species-speci c response, rather than a broader effect of sampling location. The reasons for this lack of response remain unclear but could represent a reduced phototactic response in this species, or a speciesspeci c impact on behaviour caused by initial capture using light-suction trapping.
The temperature thresholds for ight of subgenus Avaritia populations in this study (4°C in SEA and 10°C in SES) were similar or slightly above those recorded for the temperate C. oxystoma in Japan (6°C) [16], but were lower than the afrotropical C. imicola (14°C) in RSA [15]. One major difference in experimental design between the current study and these was the avoidance of a sorting step that necessitated cold anaesthesia prior to testing activity. This was viewed as preferable to pre-exposing the Culicoides to cold, given previous evidence of cold hardening [43]. The experimental trade-off for avoiding this step was a lack of harmonisation of density in pots due to the di culties of estimating numbers of Culicoides introduced. While this could have led to variation in density-dependent disturbance between ight activity pots, this was not evident in the study and would have been minimal in cooler temperatures where ight was restricted to a small proportion of the midges introduced. In addition, the requirement for vertical ight, rather than horizontal, enabled a more straightforward measure of activity, potentially reducing this factor.
In terms of wider policy implications of the study, the European Food Safety Authority (EFSA) has reported previously that, based on light-suction trapping the threshold temperature required to initiate C. obsoletus complex adult activity was 10°C, with temperatures of ≤4°C for longer than 10 days leading to an end to adult activity [47], although differentiation between survival and activity in this report is not well de ned. The disparity between predictions of activity made from light-suction trapping and direct examination of ight could result from a range of factors including the fact that light-suction traps are usually set at 1.5-2m from ground level and have a limited effective range [35]. In addition, the period where the temperature range most often limits adult ight occurs during late autumn, winter and early spring, when risk of arbovirus incursion, and hence systematic trapping, is limited.
The SVFP is determined from collections made through vector surveillance programs which, when limited, can underestimate the vector activity. The use of temperature to determine optimal times to deploy surveillance trapping would provide a better assessment of transmission risk, particularly in the event of an incursion of an arbovirus. In this scenario, the deployment of entomological surveillance in affected areas can take considerable time. In the period where there is an absence of active vector surveillance, predicting the activity of Culicoides achieved using temperature-related baseline parameters contributes signi cantly to arbovirus risk assessments and policy decision making in the immediate aftermath of an outbreak.
Vector surveillance using UV light suction traps can actively select for host seeking female Culicoides due to their increased attractiveness to light, therefore underestimating the number of Culicoides from the total population [48,49] and also underestimate the levels of activity particularly of non-host seeking Culicoides. This study con rmed that there was no signi cant difference in the ight response towards a UV light stimulus between unpigmented and pigmented females of the subgenus Avaritia collected in either the summer or autumn from the same geographical location in the South East of England. This conclusion has implications when determining the SVFP, which is de ned as <5 parous (pigmented) Culicoides speci cally caught in a UV light-suction trap collection over one night, despite both unpigmented and pigmented females being equally as active and both host seeking.
Whilst both collections from South East England consisted predominately of individuals belonging to the subgenus Avaritia, there was a far greater abundance of C. scoticus present in the SEA cohort (84%) compared with the SES cohort (37%). In this study we show, for the rst time, that C. scoticus are signi cantly more active, under these laboratory conditions, at any temperature compared with C. obsoletus in both the summer and autumn, although this increased activity of C. scoticus could also have been in uenced by differences in the species attractiveness to light. Due to di culties separating these two species by morphological methods [50] differences in their behaviour as well as their vector competence have not yet been fully explored. Nevertheless, the differences in phototactic activity and seasonality between C. obsoletus and C. scoticus observed in this study could have implications for disease management especially when determining the transmission season which could vary between species.
It has been suggested that Culicoides adults are able to survive the winter in an inactive, dormant form, and following an increase in temperature are able to become active and subsequently transmit virus, albeit in climates far warmer than those in northern Europe [14,51]. As demonstrated for C. imicola, activity may cease at sub-optimal temperatures (<4°C), but complete mortality is only observed at much lower temperatures (-3°C) [52] with one study showing that Culicoides can survive 15 days at -1.5°C [53]. Hence, temperatures below the activity threshold will render Culicoides inactive, but individuals may still be able to survive. These individuals will go undetected during surveillance, and if fully infected represent a risk of transmission with consequences for arbovirus overwintering. The accidental import of viraemic ruminants during trade may also lead to uncertainty regarding the infection of active and host-seeking Culicoides at temperatures above the threshold for ight activity (4°C), but below that required for virus replication (11-15°C). Therefore, studies of initial infection at the range of 4-11°C, followed by a realistic increase in temperature simulating a change to warmer conditions would be useful to quantify this risk pathway. Furthermore, it is not known whether the small proportion of active individuals at low temperatures found in this study can remain active for longer than 24 hours, although Culicoides have previously been shown to recover after 10 days at 4°C [54]. Further investigations into the effect of prolonged exposure to cold temperatures on adult Culicoides activity are required to assess the effect of cold temperatures at the population level in northern Europe.

Conclusions
The data presented here de ne cold temperature thresholds for ight activity in a range of UK Culicoides species and populations. Variation was observed in the subgenus Avaritia populations tested at different times of the year suggesting an effect of season on the activity of adult Culicoides whereby populations emerging later in the season possess a greater degree of cold tolerance. Further investigations are required to determine the effect of prolonged exposure to cold temperatures on Culicoides activity and survival to fully understand the consequence of cold winter temperatures on UK Culicoides populations and their ability to act as vectors of arboviruses.

Availability of data and materials
All data generated or analysed during this study are included in this published article and its additional les.

Competing interests
The authors declare that they have no competing interests.