Surveillance of invasive Aedes mosquitoes along Swiss traffic axes reveals different dispersal modes for Aedes albopictus and Ae. japonicus

Over the past three decades, Europe has witnessed an increased spread of invasive aedine mosquito species, most notably Aedes albopictus, a key vector of chikungunya, dengue and Zika virus. While its distribution in southern Europe is well documented, its dispersal modes across the Alps remain poorly investigated, preventing a projection of future scenarios beyond its current range in order to target mosquito control. To monitor the presence and frequency of invasive Aedes mosquitoes across and beyond the Alps we set oviposition and BG-Sentinel traps at potential points of entry with a focus on motorway service areas across Switzerland. We placed the traps from June to September and controlled them for the presence of mosquitoes every other week between 2013 and 2018. Over the six years of surveillance we identified three invasive Aedes species, including Ae. albopictus, Ae. japonicus and Ae. koreicus. Based on the frequency and distribution patterns we conclude that Ae. albopictus and Ae. koreicus are being passively spread primarily along the European route E35 from Italy to Germany, crossing the Alps, while Ae. japonicus has been expanding its range from northern Switzerland across the country most likely through active dispersal.

In addition to Ae. albopictus, two other invasive Aedes species, originating from East Asia, have previously been reported from Switzerland; Ae. japonicus japonicus (Theobald, 1901), the Asian or Japanese bush mosquito [30], hereafter called Ae. japonicus, and Ae. koreicus (Edwards, 1917), the Korean bush mosquito. In contrast to Ae. albopictus, these two species are generally considered to be less relevant in terms of public health because there is little evidence for disease transmission [1]. Nevertheless, under laboratory conditions, both species show vector competence for a range of human pathogenic viruses [31] and filarial nematodes [32,33]. Aedes japonicus and Ae. koreicus seem to prefer lower temperatures as compared to Ae. albopictus and, therefore, these species show a high invasive potential in the more temperate regions of Central Europe, including many areas across Switzerland [34,35].
In Switzerland Ae. japonicus was first identified in 2008 when it had already colonised an area of about 1,400 km 2 in the North across the border with southwestern Germany [36]. In Europe, it had previously only been found in smaller areas, including France [37] and Belgium [38], while until now Ae. japonicus has already been reported from 12 European countries [39]. Aedes koreicus, on the other hand, was first identified in ovitraps near the Swiss-Italian border in 2013 [40]. Before that time Ae. koreicus had also been reported from Belgium [41], north-eastern Italy [42] and European Russia [43]. Today, Ae. koreicus is present in at least seven European countries, including Austria, Belgium, Germany, Hungary, Italy, Slovenia and Switzerland [30,[40][41][42][44][45][46].
Until 2013 no Ae. albopictus had been reported from Switzerland outside the Canton of Ticino. Similarly, the other two invasive Aedes species, Ae. japonicus and Ae. koreicus had shown very localised distribution patterns [36,40]. The Alps potentially constitute a natural physical and climatic barrier for the migration of invasive Aedes species from South to North, and vice versa. Alternatively, Ae. albopictus and Ae. koreicus might have remained unnoticed outside Ticino in the absence of a nationwide surveillance programme.
While its distribution in southern Europe is well documented [26], its dispersal modes across the Alps remain poorly investigated, preventing a projection of future scenarios beyond its current range in order to target mosquito control. To monitor the presence and frequency of invasive Aedes mosquitoes beyond the Alps we set oviposition and BG-Sentinel traps at potential points of entry with a focus on motorway service areas across Switzerland. We set the traps between 2013 and 2018 from June to September and controlled them for the presence of mosquitoes every second week. We found both Ae. albopictus and Ae. koreicus being introduced primarily along the route from Italy to Germany across the Alps, while the pattern for Ae. japonicus implies a more active, radial range expansion away from its initial distribution in the North.

Mosquito sampling
From 2013 to 2018 we sampled invasive Aedes mosquitoes at potential points of entry (sites) along the major traffic axes in Switzerland, including motorway service areas (n = 35), commercial harbours (n = 3), international airports (n = 2) and the railway station in Chiasso at the Swiss-Italian border (Table 1). We deployed oviposition traps (ovitraps) to collect eggs and BG-Sentinel version 1 traps (Biogents, Regensburg, Germany) to catch host-seeking adults.
Each year the traps were set from end of June (i.e. calendar week 26 or 27) to mid-September (i.e. calendar week 36 or 37) as this corresponds to the activity peak of Ae. albopictus in northern Italy and in the infested areas of the Canton of Ticino [47]. We controlled the traps for the presence of mosquitoes every other week, leading to six sampling rounds per year. At the larger motorway service areas that are accessible from both driving directions (i.e. "A1 Gunzgen" and "A1 Rose de la Broye"), the airports, the commercial harbours and the Chiasso railway station we placed six ovitraps, while only three ovitraps were set at the smaller motorway service areas that can only be accessed in one direction (Table 1). To reduce competition in attraction between the traps we set them apart by at least 50 m. At the motorway service areas we placed one trap near the entrance, one next to the main service building and one near the lorry parking, usually situated near the exit. For the ovitraps we adopted the design previously described in Flacio et al. [28]. In brief, an ovitrap consists of a black 1.5 litre plastic container, filled with about 1.2 litres of tap water (Fig 1A), into which a short wooden slat is plunged that serves as a substrate for the female Aedes mosquitoes to lay their eggs ( Fig 1B). In each trap we added about 20-30 Vectobac granules (Valent BioSciences, Illinois, USA), containing the active Bacillus thuringensis var. israelensis (Bti), to prevent the trap from becoming an additional breeding site. We set the traps in hidden places on the ground that were protected from direct sunlight (e.g. under vegetation or near buildings).
At each site we also aimed at placing one BG-Sentinel trap. However, as the BG-Sentinel trap requires electricity to power its fan and because we visited each site once every other week we only placed BG-Sentinel traps at sites where we had direct access to a permanent power supply (Table 1). Each BG-Sentinel trap was fitted with a BG-Lure (Biogents, Regensburg, Germany) as an attractant. In addition, we equipped six BG-Sentinel traps with a cylinder that released CO 2 at a constant flow rate of 175 ml per minute (Table 1).

Species identification
Upon removing from the ovitraps, we wrapped each slat in cling film and stored the labelled and packaged slats at room temperature until they were inspected under a stereo microscope for the presence of Aedes eggs. Where present, we distinguished between the indigenous Ae. geniculatus and potentially invasive Aedes species. While Ae. geniculatus can be unambiguously identified morphologically, the eggs of invasive species are hard to identify to species level. Therefore, we first counted the eggs of invasive Aedes species and then took a subsample for molecular identification. For the molecular identification, we measured protein profiles using matrix-assisted laser desorption/ionization time-of-flight mass spectrometry (MALDI-TOF MS) with an AXIMA Performance spectrometer (Shimadzu, Kyota, Japan) following the protocol of Schaffner et al. [48]. Then we sent the resulting spectra to Mabritec AG (Riehen, Switzerland) to compare them against the company's validated mosquito species reference data base.
For the identification of the adult mosquitoes, we kept the specimens at -20˚C until morphological identification under a stereo microscope [49,50]. While we identified any mosquito potentially being an invasive Aedes species, all other specimens were not further analysed, with the exception of the collections in 2014 when we identified all specimens to species or species complex level. In some instances, the adult mosquitoes could not be identified morphologically because they were in a bad condition or members of closely related species that could morphologically not be distinguished (e.g. Culex pipiens/torrentium). In these cases, we identified the specimens also with MALDI-TOF MS following a similar protocol as used for the eggs but adapted for adults [51] and sent them to Mabritec AG.

Data analysis
Data were captured in an Access 2016 database (Microsoft, Washington, USA), then exported as separate text files for eggs and adults and finally imported into the open source software package R version 3.5.1 [52] for statistical analysis. The analysis comprised of descriptive summary statistics of the presence-absence and the numbers of eggs and adults of invasive Aedes species. For the ovitraps we assumed that all the eggs not identified morphologically to species level were the same species as those eggs identified by MALDI-TOF MS from the same slat. In a few cases we had more than one species on the same slat and, as we did not measure every single egg, for simplicity we accounted the total number of eggs to all species identified.
To investigate the relationship between the number of positive ovitraps and sampling year, we fitted generalised linear models (GLMs) with a logit link function and a binomial error distribution. Similarly, we modelled the number of eggs in the positive ovitraps as a function of year using GLMs with a log link function and a negative binomial error distribution in the R package "MASS" [53]. The level of significance was set at α = 0.05.
We plotted the graphs with the R package "ggplot" [54], while the maps were drawn with ArcGIS Desktop 10.6.1 (Environmental Systems Research Institute, Inc., California, USA).

Results
In total, we placed 5,294 wooden slats across 154 ovitraps over six years. The number of slats was slightly lower than theoretically possible (i.e. 5,544) because we could not set all traps in each year and trapping round. For example, ongoing construction work temporarily restricted access to some of the sites. From the wooden slats placed in the field, 13.1% were lost; either the traps or the slats were missing, or the traps have become dysfunctional (e.g. being tipped over, damaged or filled with rubbish). From the remaining slats 31.5% were positive. Among the positive slats (n = 1,448), 56.7% had Ae. japonicus eggs, 19.5% Ae. albopictus eggs, 1.9% Ae. koreicus eggs and 1.9% Ae. geniculatus. In 24.4% of the positive slats we had eggs that could not be identified to species level but were most likely one of the three invasive Aedes species. On a few occasions we also had slats with more than one Aedes species (4.4%). The raw data from the ovitraps are provided in S1 Data.
As with the ovitraps, we could not set all the 24 BG-Sentinel traps in Table 1 throughout the entire study and the number of traps in a year varied between 19 (2018) and 23 (2014). In contrast to the ovitraps, we could, however, reclaim and analyse all catch bags from the BG-Sentinel traps that were placed. Per trap we caught a maximum of five Ae. albopictus and a maximum of 10 Ae. japonicus adults. We had no Ae. koreicus in the BG-Sentinel traps. The raw data from the BG-Sentinel traps are provided in S2 Data.
Since 2013, the number of wooden slats that had eggs of an invasive Aedes species has increased continuously from one year to the next (odds ratio, OR = 1.23, 95% confidence interval, 95% CI = 1.19-1.28; Fig 2A). The overall annual increase in the proportion of positive traps was primarily due to a rise in the number of ovitraps with Ae. japonicus eggs (OR = 1.22, Traps that had not been set throughout the entire study were removed from the analysis. The error bars indicate the 95% confidence intervals (95% CI) around the average estimated using the generalised linear models. The large 95% CI's for Aedes koreicus are due to the low number of positive traps. "Aedes sp." denotes all counts associated with slats where the eggs were invasive Aedes species but could not be identified to species level.
https://doi.org/10.1371/journal.pntd.0008705.g002 95% CI = 1. 16-1.27). Nevertheless, the proportion of ovitraps with Ae. albopictus eggs has also slightly increased over the years (OR = 1.1, 95% CI = 1.02-1.18). It is also conceivable that the increased number of positive traps with unidentified eggs would, if known, further contribute to a rise in one or the other Aedes species. As an illustration, in 2013, the proportion of positive traps has been 17% (n = 710), while it rose to 47.5% (n = 750) in 2018, with an individual increase from 6.1% to 9.8% for Ae. albopictus and an almost three-fold increase from 8.5% to 25.2% for Ae. japonicus. In parallel to the number of positive traps the average egg count in the positive ovitraps has also increased between consecutive years (OR = 1.18, 95% CI = 1.13-1.23; Fig 2B). The ORs for Ae. japonicus and Ae. albopictus were 1.17 (95% CI = 1.11-1.23) and 1.22 (95% CI = 1.13-1.31), respectively. Since 2015, a small number of traps was sporadically also positive for Ae. koreicus (Fig 2).
In the collections from 2014, in addition to the invasive Aedes species, we also identified all other mosquito specimens to the lowest possible taxonomic level. We caught 7,424 specimens and identified 10 mosquito taxa, including six Aedes, one Anopheles and three Culex taxa ( Table 2). The most dominant taxon (97.1%) was the sibling species Culex pipiens/torrentium. Culex pipiens/torrentium was not only the most numerous but also the most widespread taxon being present across all BG-Sentinel traps with the exception of one single site (i.e. 22 out of 23). However, the BG-Sentinel trap placed at that site had no mosquitoes caught at all. A subset of 25 individuals from five different sites, one south ("Bellinzona-Sud") and three north ("Innenhof Swiss TPH", "A2 Neuenkirch-Nord" and "A1 Thurauen-Nord") of the Alps, measured with MALDI-TOF MS revealed that they were all Cx. pipiens s.s. Other more frequent taxa were Cx. hortensis, Anopheles plumbeus and Ae. japonicus (Table 2).
Aedes albopictus was mainly introduced along the national motorway A2 (i.e. European route E35) from south to north as highlighted by the increased frequency of both positive ovitraps ( Fig 3A) and positive BG-Sentinel traps (Fig 3B). The frequency of ovitraps also indicates a similar, though less dominant role, for the alternative south-north route over the San Bernardino ( Fig 3A). In contrast, both the ovitraps (Fig 4A) and the BG-sentinel traps ( Fig 4B) show a concentration of Ae. japonicus around the initially described distribution area in the Canton of Aargau from 2008 [36]. As we go further away from this area the frequencies are decreasing. For both Ae. albopictus and Ae. japonicus the traps in the West were negative throughout the study period (Figs 3 and 4). When comparing side-by-side the situation in terms of the presence of Ae. albopictus and Ae. japonicus between 2013 and 2018, the picture above becomes even more apparent (Fig 5). Aedes albopictus was primarily present in the Canton of Ticino and occasionally introduced to the North in 2013, while being more frequently introduced along the south-north axis in 2018. Aedes japonicus was almost exclusively present in the

PLOS NEGLECTED TROPICAL DISEASES
Surveillance of invasive Aedes mosquitoes along Swiss traffic axes midlands in 2013 and showed a wider distribution in 2018 when it was also found in the Canton of Ticino. In line with the increased proportion of positive ovitraps and egg numbers (Fig  2), the frequency of positive ovitraps in 2018 is a lot higher than it still was in 2013 (Fig 5). Intriguingly, we found both Ae. albopictus and Ae. japonicus at the highest site at an altitude of 1,064 m (i.e. A2 San Gottardo-Sud Stalvedro) as well as at the lowest site at 227 m (i.e. A2 Bellinzona-Sud), yet neither species was present in the traps set in the West of Switzerland. In addition to Ae. albopictus and Ae. japonicus we have also detected Ae. koreicus in ovitraps placed in the Canton of Ticino at the motorway service area "A2 Coldrerio" since 2015 and twice at the railway station in Chiasso in 2015 and 2017. In 2015 two sites north of the Alps, on the motorway service areas "A2 Teufengraben" and "A2 Neuenkirch-Süd", were also positive for Ae. koreicus (Fig 6). However, we did not find Ae. koreicus adults in the BG-sentinel traps.

Discussion
The aim of this study was to monitor the presence and frequency of invasive Aedes mosquitoes at potential points of entry in Switzerland with a focus on motorway service areas. We found that the frequency of positive ovitraps has markedly increased between 2013 and 2018, mainly because of a surge in Ae. japonicus. While Ae. albopictus is consistently introduced along the national motorway A2 from south to north, the pattern for Ae. japonicus implies a more active, radial range expansion. Since 2015 we have also been recording the presence of Ae. koreicus, both south and north of the Alps.
Before our study, Ae. albopictus was already widespread in the southern parts of the Canton of Ticino following its introduction from Italy [28], but had not been reported from any other region in Switzerland north of the Alps. While the Alps may constitute a natural barrier to the dispersal of mosquitoes, until 2013, Switzerland also lacked a national surveillance programme; and hence introductions might have simply remained unnoticed. Indeed, in 2007, Ae. albopictus eggs had already been detected in the Upper Rhine Valley in Germany at a motorway service area on the A5 near the Swiss border [55], followed by additional reports from the same region in 2011 [56] and in 2012 [57]. The German motorway A5 is the extension of the Swiss motorway A2. Both motorways constitute sections of the European Route E35 that runs from Rome, Italy, in southern Europe to Amsterdam, the Netherlands, in the north-western part of the continent. Because Italy is regarded as the primary source for Ae. albopictus in Europe [25], it does not come as a surprise that, in addition to the positive sites in Germany, multiple sites in Switzerland along the A2 were also frequently positive. Likewise, the passive spread of Ae. albopictus from south to north would explain our repeated findings along the San Bernardino route A13, an alternative south-north passage way through Switzerland. A similar picture emerges from a study in western Austria (Tyrol) where service stations along the motorways coming from Italy were repeatedly positive [44]. Together these observations support the general notion that Ae. albopictus is primarily being passively spread by motorised vehicles from south to north along the main road axes, and that the Alps represent no physical barrier for its expansion in Central Europe.
In contrast to the prominent south-north pattern we do not see much evidence in our data for an Ae. albopictus dispersal along the west-east axis. The motorway service areas in Western Switzerland near Geneva remained negative throughout the study, despite the mosquito had already been well established in southern France before the present study and has been reported from multiple sites along the Rhone Valley as close as Geneva [58]. An explanation could be lower mosquito densities in France as compared to Italy, reduced traffic volumes between the infested areas in France and Western Switzerland when compared to the southnorth circulation, fewer vehicles stopping at the motorway service areas included in our study, or a combination of these factors.
In contrast to Ae. albopictus, that shows a pattern consistent with passive dispersal through motorised vehicles, the distribution and frequency of sites positive for Ae. japonicus over the years imply a more active range expansion. In 2008, the initially colonised area was estimated at 1,400 km 2 , covering parts of Switzerland and bordering Germany [36]. Within that original range we had six sites in the present study, including the motorway service stations "A1 Kölliken-Nord", "A1 Kölliken-Süd ", "A1 Würenlos-Nord", "A1 Würenlos-Süd", "A1 Kempthal" and "Zürich Airport". Between 2013 and 2018, these sites were frequently positive for Ae. japonicus, while the sites at the periphery were positive depending on the distance away from the initial distribution area. Although passive distribution via traffic is generally assumed to be the chief mode of dispersal of invasive Aedes mosquitoes [59], our data for Ae. japonicus are more consistent with the hypothesis of active dispersal being the key driver. This would not exclude that passive dispersal along motorways may still take place as the more recent findings in the Canton of Ticino imply. Interestingly, a previous study of Ae. japonicus found that its larvae are more frequently present in rural than urban areas as compared to Ae. albopictus [60], which might reflect an adaptation allowing this species to disperse more actively across a heterogeneous landscape. Though population genetic studies would be more informative to test the hypothesis of active range expansion, the circumstantial evidence from this study supports the idea of active range expansion for Ae. japonicus.
In addition to Ae. japonicus we found Ae. koreicus in several of our ovitraps along the south-north Gotthard route. Given the few findings in our study the picture is still somewhat unclear, but this mosquito seems to be more present in the South of the country and, like Ae. albopictus, has likely been passively dispersed along the south-north axis through motorised vehicles driving up the motorway A2. This is in line with its first detection at the Swiss-Italian border in 2013 [40] and with previous reports from northern Italy [61]. Moreover, Ae. koreicus has also been repeatedly reported from Germany [30,62,63] and more recently from Austria [44]. Like the closely related Ae. japonicus, Ae. koreicus is a container-breeding mosquito species that is well adapted to a more temperate climate [34]; and hence we expect that this species, too, will become more widespread across Switzerland and the rest of Central Europe.
Although primarily used to trap host-seeking Ae. albopictus, the BG-Sentinel trap has also been found suitable for the surveillance of other adult mosquitoes in Europe [64]. With this in mind we also identified all other adult specimens that were collected in the BG-Sentinel traps to the lowest possible taxonomic level. However, we were only able to do this extra effort in 2014. In that year we identified 10 different taxa from about 40 known species in Switzerland [40,65,66]. Among the 10 taxa found in the BG-Sentinel traps the most dominant one was the sibling species pair Culex pipiens/torrentium. Culex pipiens/torrentium has frequently been described as the most abundant mosquito taxon in Switzerland [66,67]. Further MALDI-TOF MS analysis of a subset of specimens taken from one site in Ticino and three sites north of the Swiss Alps identified Cx. pipiens s.s. only and no Cx. torrentium. These results are in line with a more comprehensive Swiss study by Wagner et al. [68] who found that Cx. torrentium shows very low relative abundances across suburban and natural areas north of the Swiss Alps. Culex hortensis was the second most abundant species, followed by An. plumbeus and Ae. japonicus. These species have also been observed at higher frequencies in Switzerland in previous studies [36,67].
Both the ovitraps and the BG-Sentinel traps were set during the putative peak activity period of Ae. albopictus from June to September, and it is possible that we might have missed introductions of invasive Aedes mosquitoes before or after this period. Besides, we were not able to identify every single egg to species level, so that the actual numbers of eggs per species have some uncertainties. Another point of consideration, primarily for the adult trapping, is that choices for the positioning of the BG-Sentinel traps were limited because we had to rely on the power supply from the mains to operate them. Moreover, the traps had to be well concealed from the public to avoid vandalism. Similarly, only a small subset of BG-Sentinel traps could be fitted with a CO 2 bottle, meaning that they likely differed in their attraction to mosquitoes. Altogether, this might have introduced a sampling bias. Also, in 2013 we had to rely on ovitraps only. Therefore, a direct comparison between the two datasets is not warranted, however, where we have higher egg densities the BG-Sentinel traps are generally in agreement with the ovitraps.
As previously documented for Ae. albopictus [28,69] invasive Aedes mosquitoes escaping from vehicles upon stopping at motorway service areas may constitute critical points of entry for the early establishment of new populations, or potentially add to existing infestations, particularly if the points of entry are close to residential areas. In 2018 an attentive citizen reported an adult Ae. albopictus 300 m away from the motorway A1 service area "Gunzgen-Nord"where ovitraps were repeatedly positive-suggesting that the mosquito has either escaped from a vehicle stopping there or its offspring have made their way to the village. A recent study suggests that newly hatched adult Ae. albopictus mosquitoes fly several hundred metres away from their breeding site to seek hosts [70]. Therefore, unless effective control measures are implemented, several sites north of the Alps are at risk of becoming starting points for the establishment of new populations.
The regions north of the Alps are likely to be more suitable for the establishment of permanent Ae. japonicus rather than Ae. albopictus populations due to the former being better adapted to temperate climates [71,72] and because of its extended phenology [39]. Similarly, Ae. koreicus might also be more adapted to the colder climate north of the Alps [41]. However, while many areas might not provide ideal habitats for Ae. albopictus because of temperature limits during the winter months, warmer regions such as the area around the lake of Geneva, the Upper Rhine Valley, including Basel, or at the shores of Lake Constance might still be suitable enough for the establishment of locally reproducing populations as habitat suitability models [71], microclimatic studies [73] and recent reports from Basel [74] and southern Germany suggest [75,76]. To shed more light into this question, future work should consider population genetic approaches to identify the sources of the Ae. albopictus specimens found north of the Alps.
From our results we conclude that Ae. albopictus and Ae. koreicus are being passively spread primarily along the European route E35, while Ae. japonicus is most likely expanding its range largely through active dispersal. Because of increasing introduction of invasive Aedes mosquitoes, control measures should be put in place to reduce the likelihood of their establishment from points of entry such as motorway service stations.