Ecological characterization and molecular differentiation of Culex pipiens complex taxa and Culex torrentium in eastern Austria

Background Culex pipiens complex taxa differ in behaviour, ecophysiology and epidemiologic importance. Despite their epidemiologic significance, information on genetic diversity, occurrence and seasonal and spatial distribution patterns of the Cx. pipiens complex is still insufficient. Assessment of seasonal and spatial distribution patterns of Culex pipiens forms and their congener Cx. torrentium is crucial for the understanding of their vector–pathogen dynamics. Methods Female mosquitoes were trapped from April–October 2014 twice a month for a 24-h time period with BG-sentinel traps at 24 sampling sites in eastern Austria, using carbon dioxide as attractant. Ecological forms of Cx. pipiens s.l. and their hybrids were differentiated using the CQ11 locus, and Cx. pipiens forms and their congener Cx. torrentium using the ACE-2 gene. Differential exploitation of ecological niches by Cx. pipiens forms and Cx. torrentium was analysed using likelihood ratio tests. Possible effects of environmental parameters on these taxa were tested using PERMANOVA based on distance matrices and, if significant, were modelled in nMDS ordination space to estimate non-linear relationships. Results For this study, 1476 Culex spp. were sampled. Culex pipiens f. pipiens representing 87.33 % of the total catch was most abundant, followed by hybrids of both forms (5.62 %), Cx. torrentium (3.79 %) and Cx. pipiens f. molestus (3.25 %). Differences in proportional abundances were found between land cover classes. Ecological parameters affecting seasonal and spatial distribution of these taxa in eastern Austria are precipitation duration, air temperature, sunlight and the interaction term of precipitation amount and the Danube water level, which can be interpreted as a proxy for breeding habitat availability. Conclusions The Cx. pipiens complex of eastern Austria comprises both ecologically different forms, the mainly ornithophilic form pipiens and the mainly mammalophilic and anthropophilic form molestus. Heterogeneous agricultural areas as areas of coexistence may serve as hybridization zones, resulting in potential bridge vectors between birds and humans. Occurrence, seasonal and spatial distribution patterns of the Cx. pipiens complex and Cx. torrentium and the presence of hybrids between both forms were quantified for the first time in Austria. These findings will improve the knowledge of their vector–pathogen dynamics in this country.


Background
Species of the Culex (Culex) pipiens complex are hardly distinguishable due to their morphological similarity, but differ extensively in behaviour, physiology and host preference. The complex is considered to consist of seven taxa: Cx. quinquefasciatus Say, Cx. pipiens pallens Coquillet, Cx. australicus Dobrotworsky & Drummond, Cx. globocoxitus Dobrotworsky and the nominal species, Cx. pipiens L., comprising two genetically and ecologically distinct forms: Culex pipiens f. pipiens, which is repeatedly described and examined as ornithophilic, diapausing, anautogenous, and eurygamous, whereas Culex pipiens f. molestus is interpreted as mammophilic (and especially anthropophilic), autogenous, and stenogamous [1][2][3][4]. Both forms are known to hybridize in areas of coexistence [5,6], potentially resulting in bridge vector populations presumed to feed mostly on birds, but also on humans [4]. However, hybridization events in central and northern Europe potentially are rare due to selective and exclusive habitat use of both forms [7][8][9][10][11]. Genetically distinct non-members, but often indistinguishable species from those in the Cx. pipiens complex are Cx. torrentium Martini in northern Europe, Cx. restuans Theob., Cx. nigripalpus Theob. and Cx. salinarius Coquillet in North America, Cx. pervigilans Von Bergroth in New Zealand and Cx. vagans Wiedemann in central and eastern Asia [12].
Culex pipiens complex taxa are key vectors succouring the transmission of a variety of pathogens such as avian malaria (Plasmodium spp.) and filarioid helminths (Dirofilaria spp.) [13,14] and are considered to be main vectors of West Nile virus [15][16][17][18] and highly competent vectors for Usutu virus [19]. The presumed ornithophilic Cx. pipiens f. pipiens and Cx. torrentium are additionally vectors of Sindbis virus and differ in vector competence. Different transmission efficiencies for Sindbis virus (the causative agent of Ockelbo disease and Karelian fever [20,21]) were found in Cx. torrentium and Cx. pipiens f. pipiens by transmission experiments in which the former species was found to be a more efficient vector [22]. Sindbis virus seroprevalence linked to migratory bird infection rates was found to fluctuate in the human population in northern Europe [23], and recently the Sindbis virus was isolated in Germany from Cx. torrentium [24]. Furthermore, Cx. torrentium is under discussion as a potentially better vector of West Nile virus than Cx. pipiens f. pipiens [25]. In addition, Cx. pipiens f. molestus was identified as vector of West Nile virus with the potential for vertical transmission within local populations [26]. Precise data on the distribution and ecology of these taxa are necessary to assess potential risks for local human populations [23] and biodiversity (e.g. [27]). The absence of stable morphological determination characters and sympatric occurrence in some Culex species however resulted in incomplete or erroneous data on spatial and seasonal distribution of these taxa [28]. This clearly necessitates detailed scrutinization of the ecology of Culex species, including Cx. pipiens complex taxa. However, only males or larvae of Cx. pipiens forms and Cx. torrentium can be distinguished reliably (by either the structure of the phallosome on the hypopygium [29,30] or chaetotaxy [31]), but neither males nor larvae are of major interest in strictly epidemiological studies.
Species identification of morphologically hardly separable females of Culex spp. is nonetheless feasible by means of molecular methods. The mitochondrial cytochrome oxidase c subunit I can be used to distinguish Cx. torrentium from Cx. pipiens complex taxa [32], but attempts to identify hybrids between these taxa based on a length variation in the di-nucleotid microsatellite locus CQ11 potentially lead to some molecular misidentification of Cx. torrentium and Cx. pipiens forms [33]. Even so, utilization of two different protocols [1,34] enables indisputable identification of Cx. pipiens f. pipiens, Cx. pipiens f. molestus, and their hybrids, Cx. torrentium. To our knowledge this is the first study to quantitatively examine the seasonal and spatial distribution patterns of the Culex pipiens forms and their sister taxon Cx. torrentium in Austria. This knowledge is crucial for the understanding of vector-pathogen dynamics in Austria and is furthermore essential for the implementation of appropriate mosquito surveillance and control strategies.

Mosquito sampling and identification
In our study 24 permanent sampling sites distributed across Lower Austria, of which two were located in Vienna (only on artificial surfaces), 14 in Lower Austria (representing predominantly artificial surfaces) and eight in Burgenland (representing predominantly agricultural areas), were monitored from April to October 2014. Mosquito communities including members of the Culex pipiens complex were sampled every second week for a 24-h time period using mosquito traps (Biogents®, Regensburg, Germany) baited with carbon dioxide as an attractant and were stored at −80°C. Female mosquitoes were identified by morphological determination characters using the key of Becker et al. [31]. Identification of morphologically cryptic mosquito females of the Cx. pipiens complex and Cx. torrentium was performed following Smith & Fonseca [34] (exploiting an intron length polymorphism in the ace-2 gene to differentiate Cx. torrentium from Cx. pipiens forms) and Bahnck & Fonseca [1] (exploiting a length polymorphism of the CQ11 gene to distinguish Cx. pipiens f. pipiens and Cx. pipiens f. molestus).

Habitat preferences
Proportional differences in abundance of Culex taxa in eastern provinces of Austria (corrected against number of sampling sites per province) were assessed using a Williams-corrected likelihood ratio test (G-test) of independence [35]. Additionally, sampling sites were assigned to specific habitat types using the CORINE [Co-ordinated Information on the Environment] Land Cover [36] database to investigate potential habitat type preferences of Culex taxa. CLC was selected because of its high spatial resolution (1:100,000) and the methodological homogeneity used for the land cover classification. To detect differences in the abundance of Culex spp. in land cover classes a Williams-corrected likelihood ratio test (G-test) of goodness of fit was used.
Seasonal and spatial variations in Culex spp. communities Differential abundance of Cx. pipiens complex taxa per month and province was assessed using linearized graphical representations of Bray-Curtis distances produced by means of non-metric multidimensional scaling (nMDS). Meteorological data, such as temperature, air pressure, humidity, amount and duration of precipitation of 15 weather stations distributed across the sampling area were provided by the Austrian Central Institute for Meteorology and Geodynamics (ZAMG). These parameters, together with water levels of the aquatic habitats, primarily influence egg-laying, larval development and availability of larval habitats [37,38]. Thus, 14-day means prior to the sampling date were computed to account for effects of meteorological parameters on abundance and occurrence of sampled Culex taxa. Effects of meteorological parameters were assessed using permutational multivariate analysis of variance (PERMANOVA) on dissimilarity matrices as implemented in the 'vegan' package [39], based on raw mosquito data pooled per month and province. Parameters contributing significantly to the observed patterns were subsequently modelled in ordination space using the function "ordisurf()" to estimate non-linear relationships. Species abundance in relation to communities was assessed by computing and plotting weighted average species scores. All statistical analyses were performed in the R statistical environment (R Development Core Team, 2011).
Based on an analysis of differential distribution of Culex spp. in habitat types of CORINE Land Cover Level 1 habitat types, Cx. pipiens forms occur predominately in agricultural areas and Cx. torrentium on artificial surfaces. Significant differences in proportional abundances were found exclusively for Cx. pipiens f. pipiens in CLC level 2 habitat types (highest abundances in arable land; G = 13.312, X-squared df = 5, P = 0.02062) and in CLC level 3 habitat types (highest abundances in nonirrigated arable land; G = 19.766, X-squared df = 7, P = 0.006098). The other taxa examined did not differ significantly in their proportional abundances. Highest abundances of Cx. pipiens f. pipiens were recorded in complex cultivation patterns and non-irrigated arable land (CLC level 3), whereas Cx. pipiens f. molestus was most abundant in complex cultivation patterns (CLC level 3). Hybrids of Cx. pipiens f. pipiens and Cx. pipiens f. molestus were most abundant in non-irrigated arable land (CLC level 3) and Cx. torrentium in discontinuous urban fabric regions (CLC level 3).

Seasonal and spatial variations in Culex pipiens complex communities
PERMANOVA indicated effects of some meteorological parameters on spatial and seasonal variations of Culex spp.: air temperature, sunshine duration and the amount of precipitation are possible drivers of the seasonal and spatial differentiation. Further, the interaction term of amount of precipitation and Danube water level, a proxy of breeding habitat availability, was found to structure communities ( Table 2).
Culex pipiens complex communities differed from one another in relation to sampling province and time.
Viennese, Burgenland and Lower Austrian samples were depicted with a large spread, indicating differentiated communities throughout the sampling period (Figs. 1,  2). Viennese June and August, Lower Austrian June, July, August and September, and Burgenland August, September and October communities were loosely arranged around average weighted species scores. Communities characterized by low abundances (Viennese April, May and September, and Lower Austrian April) were more similar to one another and to all other communities sampled. May communities sampled in Burgenland and Lower Austria were relatively similar to one another and to Viennese July communities. Viennese September and Burgenland June communities were somewhat separate from the rest.
Generalized additive modelling of environmental parameters identified by PERMANOVA as potential drivers of community composition (Table 2) in ordination space suggests high abundances of all taxa to be associated with high air temperatures (Fig. 1), intermediate sunshine duration per day (Fig. 2), high duration of precipitation, and combined high Danube water levels and amounts of precipitation.

Discussion
Taxonomy, abundance and habitat preference of Cx. pipiens forms and Cx. torrentium in eastern Austria Our results indicate that all taxa of the Cx. pipiens complex are common and widely distributed in eastern Austria. Additionally, sympatric co-occurrence of Cx. torrentium and Cx. pipiens f. pipiens was observed at 14 sampling localities throughout eastern Austria, predominately in discontinuous urban fabric habitats (CLC level 3). While formerly considered a rare species [40], recent surveys in Germany suggest Cx. torrentium to be one of the most abundant mosquitoes in Europe [41]. Previous under-representation of this species is potentially linked to cryptic taxonomic characters thwarting morphological identification (cf. [40,41]). Distribution of adult Cx. pipiens f. pipiens and Cx. torrentium assessed in the present study suggests shared habitats, confirming prior studies on co-occurrence of larvae in a variety of breeding habitats (e.g. [3,30,[41][42][43][44][45][46]). Further, the distribution of Cx. torrentium suggests a preference for discontinuous urban fabric, corroborating a preference for anthropogenic habitats (e.g. [25,30,47]).
Sympatric co-occurrence of Cx. pipiens forms was observed in 11 out of 24 sampling localities in Lower Austria. Culex pipiens forms were most abundant in agricultural areas, especially arable land and heterogeneous agricultural areas (CLC level 2), where Cx. pipiens f. pipiens and Cx. pipiens f. molestus are associated with complex cultivation patterns in co-occurrence, and Cx. pipiens f. pipiens and hybrids predominately co-occur in non-irrigated arable land (CLC level 3). Furthermore, a high proportion of hybrids in these habitat types, particularly non-irrigated arable land, characterizes them as hybridization zones. While a certain rate of hybridization of Cx. pipiens f. pipiens and Cx. pipiens f. molestus is common, these forms supposedly occur in distinct habitats types linked to their ecological peculiarities: whereas Cx. pipiens f. pipiens is reported to be ubiquitous, Cx. pipiens f. molestus is described as restricted to so-called (mostly anthropogenic) 'underground' habitats (e.g. in Morocco [48], Spain [7], Portugal [49] and Netherlands [5]). Also, the occurrence of Cx. pipiens f. molestus and co-occurrence of Cx. pipiens forms was observed in Germany and Portugal in Fig. 1 Configuration of spatial and temporal fluctuations of Culex spp. communities in eastern Austria in a two dimensional NMDS representation of Bray-Curtis distances. Isolines represent air temperature [°C × 10 −1 ] variability throughout the sampling period modelled in ordination space using a generalized additive modelling approach as implemented in ordisurf(), deviance in ordination space explained = 44.9 %; R 2 = <0.01 Asterisks indicate significant effects of certain environmental parameters (*, P ≤ 0.05; **, P ≤ 0.01) highly urbanized areas [3,4]. Distribution patterns of Cx. pipiens f. pipiens and Cx. pipiens f. molestus recovered in this study contrarily indicate an association of both Cx. pipiens f. pipiens, Cx. pipiens f. molestus and their hybrids with arable land and agricultural areas. This is particularly noteworthy, as an increased potential for hybridization and the formation of bridge-vectors for West Nile virus or other mosquito-borne pathogens can be expected under such circumstances. Additionally, these results clearly necessitate intensified, focused investigations on habitat preference and the distribution of Cx. pipiens complex taxa throughout Europe.

Spatial and temporal variation of Culex taxa
Temporal variation (as pattern of Bray-Curtis distances recovered by nMDS) of Cx. pipiens complex taxa was observed throughout the sampling period with the most similar communities and highest abundances observed in the summer months. Meteorological parameters found to structure abundance and occurrence mostly affect developmental time and abundance (via larval survival and larval habitat availability) generation cycles (cf. [38,50]). Furthermore, Cx. pipiens f. pipiens occurred earlier in the year than Cx. torrentium and Cx. pipiens f. molestus. This suggests somewhat higher temperature optima of these latter taxa and additionally supports the hypothesis of a more southern distribution and genesis of Cx. pipiens f. molestus [15,51,52] and Cx. torrentium [41]. However, the latter species is reported as widely absent in the Mediterranean region [41]. Putatively, this supposedly strict ornithophilic taxon is more strongly affected by other environmental parameters that also control abundance of its preferred host, e.g. habitat heterogeneity. However, the interpretability of results presented in this work is limited by the low number of specimens collected, a problem encountered by several other studies which focussed on adults (cf. [25,53]). Nevertheless, these results indicate that a wide range of suitable larval habitats is present in eastern Austria, and, considering the low effectiveness of carbon-dioxide baited traps on certain culicid taxa, a large population of potential vectors for mosquito-borne pathogens must exist.

Comments on efficacy of trapping methods
Interestingly, a small proportion of all collected specimens were identified as Cx. torrentium in our study. Under-representation of Cx. torrentium in carbon dioxide baited traps is commonly observed (e.g. [41,54]), and potentially leads to an under-estimation of the proportional abundance of Cx. torrentium. This is particularly noteworthy, as carbon dioxide baited traps are described as attracting and catching a broad range of different mosquito species compared to other commonly used traps [55]. Differential efficacy of carbon dioxide baited traps was related to a relatively lower attractiveness of carbon dioxide to strictly ornithophilic species in comparison to anthropophilic or mammalophilic species [31,41]. However, the supposedly mammalophilic and anthropophilic Cx. pipiens f. molestus (e.g. [31]) should consequently have been collected in higher numbers. While a large number of studies suggest strong attraction of culicidas to pure CO 2 (e.g. [56,57]), attractiveness of host odours seems to be controlled by prior sensitization through brief CO 2 exposure in Ae. aegypti [58]. In addition, flight velocity, orientation of flight and source finding was found to increase post-CO 2 exposure compared to naive specimens [58]. Potentially, CO 2 acts rather as primary attractant and mediator eliciting higher sensitivity and thus the more specific response to host-odours [58]. Thus, the efficacy of carbon-dioxidebaited traps may be reduced in taxa requiring further cues to locate potent hosts. In particular, volatile substances from, for example, preen glands of birds (cf. [59,60]) or serous glands of amphibia (cf. [61,62]) might be more important for host localization in ornithophilic or herpetophilic taxa such as Cx. torrentium or Culiseta longiareolata [31]. Furthermore, additional host localization cues currently not considered (e.g. thermic signatures, CO 2 concentration, optical cues) could be relevant for differential trap efficacy. As sympatric occurrence with roughly equal abundances of Cx. torrentium and Cx. pipiens complex taxa is mainly demonstrated in larval surveys [3,30,46], reliable estimates of abundance of Cx. pipiens forms requires the analysis of larval community composition and exposition of gravid traps and ovitraps (cf. [41]). Putatively, such surveys will also improve data on Cx. pipiens f. molestus distribution and habitat preference as this taxon is currently potentially under-represented in the majority of studies, similar to Cx. torrentium.

Conclusions
In eastern Austria both ecological forms of Culex pipiens exist, the mainly ornithophilic form pipiens and the mainly mammalophilic and anthropophilic form molestus. Cx. pipiens form pipiens is predominant in eastern Austria. Areas of co-occurrence are agricultural, peri-urban regions that may serve as areas of hybridization, resulting in bridge vectors between birds and humans. An assessment of occurrence, as well as seasonal and spatial distribution patterns of the Cx. pipiens forms and Cx. torrentium as well as the identification of hybrids was attained for the first time in Austria. These findings will contribute to the understanding of their vector-pathogen dynamics in this country.