Correct mosquito species identification is essential for mosquito and disease control programs. However, this is complicated by the difficulties in morphologically identifying some mosquito species. In this study, variation of a partial sequence of the cytochrome c oxidase unit I (COI) gene was used for the molecular identification of British mosquito species and to facilitate the discovery of cryptic diversity, and monitoring invasive species. Three DNA extraction methods were compared to obtain DNA barcodes from adult specimens. In total, we analyzed 42 species belonging to the genera Aedes Meigen, 1818 (21 species), Anopheles Meigen, 1818 (7 species), Coquillettidia Theobald, 1904 (1 species), Culex Linnaeus, 1758 (6 species), Culiseta Felt, 1904 (7 species), and Orthopodomyia Theobald, 1904 (1 species). Intraspecific genetic divergence ranged from 0% to 5.4%, while higher interspecific divergences were identified between Aedes geminus Peus, 1971/Culiseta litorea (Shute, 1928) (24.6%) and Ae. geminus/An. plumbeus Stephens, 1828 (22.5%). Taxonomic discrepancy was shown between An. daciae Linton, Nicolescu & Harbach, 2004 and An. messeae Falleroni, 1828 indicating the poor resolution of the COI DNA barcoding region in separating these taxa. Other species such as Ae. cantans (Meigen, 1818)/Ae. annulipes (Meigen, 1830) showed similar discrepancies indicating some limitation of this genetic marker to identify certain mosquito species. The combination of morphology and DNA barcoding is an effective approach for the identification of British mosquitoes, for invasive mosquitoes posing a threat to the UK, and for the detection of hidden diversity within species groups.

DNA extraction methods, hidden genetic diversity, molecular identification, vector species

The family Culicidae includes approximately 112 genera and 3,547 described species (Harbach 2017, 2018). Several species are biting pests playing an important role as vectors of pathogens of humans and livestock (Becker et al. 2010). These include chikungunya, dengue, Japanese encephalitis, yellow fever, West Nile, Rift Valley fever, and Zika viruses, as well as several nematodes and protozoans such as Plasmodium Marchiafava & Celli, 1885 (Becker et al. 2010, Medlock et al. 2007). In addition to their medical and veterinary importance, mosquitoes are significant nuisance biters of humans and within the environment may serve additional roles such as key indicators of landscape degradation (Dorvillé 1996, Guedes and Navarro-Silva 2014, Montagner et al. 2018). As a result, mosquitoes are one of the principal target groups within surveillance and control programs worldwide (Hernández-Triana et al. 2017).

Current approaches to species identification still rely heavily upon morphology-based procedures, which typically require substantial training and may not always provide a good resolution on a specimen’s identity due to homogeneity between life stages of different species and the presence of species complexes (Cywinska et al. 2006, Hernández-Triana et al. 2012, 2014, 2015, Linton et al. 2005, Packer et al. 2009, Versteirt et al. 2015). To overcome this obstacle, a small region (658 bp) of the mitochondrial cytochrome c oxidase unit I (COI) gene was proposed as a standardized DNA marker in support of species identification for animal barcodes, in a process commonly referred to as DNA barcoding (Hebert et al. 2003a, b).

Until recently, thirty-four mosquito species have been recorded in the United Kingdom (UK) (Medlock et al. 2015, Medlock and Vaux 2009, 2015). However, Medlock et al. (2017a) detected the presence of Ae. albopictus (Skuse, 1895) in southern England, and Dallimore et al. (2017) collected a single male Ae. aegypti (Linnaeus, 1762) in Merseyside in north west England, although these two invasive species are not believed to be locally established. Nonetheless, these findings demonstrated that the UK is at risk of introduction by invasive species of Aedes (Dallimore et al., 2017). In addition, Harbach et al. (2017) discovered Ae. nigrinus (Eckstein, 1918) in the New Forest, southern England, which brings the total count of named species to 37 [35 native species plus two records of invasive species]. In addition, the occurrence of certain species has been very sporadic as in the case of Ae. vexans (Meigen, 1830); however, Medlock et al. (2017b) reported a resident population of this species at Marston Marshes, Norwich. Although no mosquito-borne pathogen affecting humans or livestock is presently thought to circulate in the UK, there is potential for future pathogen emergence (Medlock and Leach 2015; Medlock et al. 2017a, b; Vaux et al. 2015) and there remains continuing mosquito nuisance in various parts of the country (Brugman et al. 2017a, b). Collectively, these discoveries highlight the need for continued research on the native mosquito fauna of the UK, taking into account the potential incursion of invasive species.

There is, however, a paucity of data on the utility of molecular methods for species identification of the British mosquito fauna. During the first development of a molecular assay for the identification of hybrids and sibling species within Culex pipiens s.l., Smith and Fonseca (2004) used specimens from England and Scotland. Golding et al. (2012) subsequently employed the COI marker to compare sequences of Cx. modestus Ficalbi, 1890 with other Culex Linnaeus, 1758 species in southeast England, and Danabalan et al. (2012) employed a combination of the internal transcribed spacer gene-2 (ITS-2) and COI markers in their assessment of molecular identification tools to determine the status of Cx. pipiens s.l. The same approach was used by Danabalan et al. (2014) to confirm the occurrence of species within the Anopheles maculipennis complex Theobald, 1911 in England and Wales. Recently, Hernández-Triana et al. (2015) employed an integrated approach to determine mosquito host feeding preferences (via identification of blood meal origin), as well as the molecular characterization of mosquito species carrying pathogens such as myxoma virus (Brugman et al. 2015, 2017a, b) and Theileria orientalis Yakimoff & Soudatschenkoff, 1931 within their bloodmeal (de Marco et al. 2016).

In the present paper, we apply the COI DNA barcoding approach in support of the identification of native British mosquitoes and known invasive species in continental Europe. In addition, we assessed the DNA barcode variability using genetic distance methods to detect cryptic diversity across the taxa.

This study assessed minimally destructive approaches that retained a significant part of the sample as a referenced voucher and the development of a COI DNA barcoding library in mosquitoes, and assessed the use of the variability within COI in support of species identification. Overall, the three extraction methods used provided sufficient DNA for subsequent analysis; however the modified Hotshot technique of Montero-Pau et al. (2008) proved to be the most efficient and inexpensive method for obtaining COI DNA barcode sequences. This method has been applied to other groups such as the Hymenoptera (Guzmán-Larralde et al. 2016) with good results as assessed by DNA yield and PCR amplification success. The amplification of vertebrate DNA from engorged abdomens using the Qiagen DNeasy Blood and Tissue extraction kit highlights the need to use insect specific primers, for example LepF/LepR (see www.boldsystems.org/index.php/ Public_Primer_PrimerSeach) instead of the standard Folmer primers (Folmer et al. 1994). In terms of cost, considering that we did not have to buy a DNA extraction kit to perform the DNA extraction for processing the legs, the Hotshot technique represented savings of around 500 GBP per 96-well plate to our laboratory, making it a cost-effective method for performing DNA extractions.

The majority of morphologically identified species in this study formed defined groups in the NJ analysis based on DNA barcodes (Fig. 2), supporting the use of COI DNA barcoding in combination with morphological observation as a suitable approach for species identification. Genetic divergence between morphospecies ranged from 0.19% to 24.6%, whereas intraspecific genetic divergences within distinct species ranged from 0% to 5.4% (average 1.30%; Table 1, Suppl. material 2). Most of the specimens within a morphospecies were resolved in the NJ tree (Fig. 2). However, some individuals in certain taxa such as Ae. annulipes/Ae. cantans, Ae. cinereus/Ae. geminus, Ae. sticticus/Ae. nigrinus, An. daciae/An. messeae, and Cx. quinquefasciatus/Cx. pipiens s.l. clustered together (Fig. 2), indicating some limitations of the COI gene as a marker to separate these species. This finding is not surprising as these taxa are phylogenetically, as it has been highlighted in the literature (Harbach 2017, Harbach et al. 2017).

With regard to Anopheles maculipennis s.l., although some morphological traits in egg structure provide an effective method to separate some members of this group, there is some dispute regarding the taxonomic status of An. daciae (e.g. Linton et al. 2002, 2005, Kronefeld et al. 2012). This species was described by Nicolescu et al. (2004) based on all life stages collected from the Black Sea region in Romania. The authors stated that An. daciae and An. messeae have been misidentified in the past because of similar morphology. However, they showed that An. daciae eggs are generally smaller than those of An. messeae, with patches of larger deck tubercles that contrast more sharply with patches of smaller tubercles to impart greater definition to the mottled surface of the deck (see Nicolescu et al. 2004; fig. 3 A, C). In contrast, the deck of An. messeae eggs has a more diffuse or weakly mottled appearance (see Nicolescu et al. 2004; fig. 3 B, D). In the same study, molecular analysis of ITS-2 sequences identified single nucleotide polymorphisms and unique haplotypes which, in the authors’ views, confirmed the specific status of An. daciae. However, other authors have queried the specific status of An. daciae, and stated that COI offered poor resolution and advocated for further work to determine the status of An. daciae. Similarly, in their study of Belgian mosquitoes, Versteirt et al. (2015) reported that specimens identified as Ae. annulipes and Ae. cantans grouped together in their NJ tree, and stated that COI cannot separate these two species; the same results have been obtained in our study (Fig. 2).

Moreover, COI DNA barcoding highlighted mis-identifications within the genus Culiseta (Cs. fumipennis, Cs. litorea and Cs. morsitans). These species are placed in the subgenus Culicella, and the females are difficult to identify because of their morphological similarity and absence of reliable diagnostic characteristics (Becker 2010, Cranston et al. 1987, Medlock and Vaux 2010). All of our females were collected in traps and were in relatively poor condition. In addition, no COI sequences of reliably identified material of Cs. litorea were available in the NCBI or BOLD databases to compare with our specimens (Ruiz-Arrondo et al. 2017). Because the sequences of females of Cs. morsitans matched the two males identified as Cs. litorea from Spain, we considered all our specimens to be Cs. litorea. In the literature (Medlock et al. 2007), Cs. morsitans is considered a widespread species in the UK, found in permanent waters, while Cs. litorea has a more restricted distribution in coastal regions of southern England. Both species feed mainly on birds, but can also bite humans, and they are considered bridge vectors of arboviruses (Medlock et al. 2007). Our findings highlight the need for careful examination of material obtained from traps in combination with the application of molecular techniques for a reliable identification of these species. Even though in our dataset Ae. vexans showed a low genetic divergence of approximately 2%, specimens of this species separated into two distinct groups in the NJ tree (Fig. 2), which may be an indication of some genetic differentiation within the population. This agrees with the findings of Lilja et al. (2018) in which the authors reported a distinct genotype of Ae. vexans in Europe.

Regarding non-indigenous mosquito species, although the adults of certain species are easily identified using morphological keys, for example Ae. aegypti and Ae. albopictus (Becker 2010, Cranston et al. 1987), the development of a molecular library for species identification is important, in particular when specimens are found in a poor stage of preservation. This is an essential step for the establishment of control measures (Versteirt et al. 2015) in the event of a recent introduction, as in the case of the detection of Ae. albopictus in the UK (see Medlock et al. 2017a, b).

In our dataset, Ae. koreicus and Cs. litorea showed higher intraspecific genetic divergences (Table 1, Fig. 2), which may indicate the presence of cryptic diversity. For all other species, the variation in intra- and interspecific genetic values reported in this study fall within the range for DNA barcoding studies of European mosquitoes (Engdahl et al. 2014, Gunay et al. 2015, Versteirt et al. 2015) or other zoogeographical regions such as the Nearctic and Oriental Regions (Cywinska et al. 2006, Kumar et al. 2007, Murugan et al. 2016). Nonetheless, we advocate the combination of the COI DNA barcoding with other genetic markers such as the Elongator Complex Protein 1 gene (ECP1) (Low et al. 2016, Pangjanda and Pramual 2016, Senatore et al. 2014) and ITS-2 sequences from a larger number of specimens across the species distribution range in order to resolve some of the taxonomic problems highlighted in this study.

This study provides COI DNA barcoding data to support the molecular identification of mosquito species in the UK as well as invasive mosquito species, many of which are currently expanding their geographical range in continental Europe. We augment the barcoding data for anthropophilic species such as Ae. cinereus, Ae. detritus, Ae. sticticus, Ae. vexans, and Cx. modestus, as well as other species of veterinary importance such as the bridge vector Cs. annulata. Even though the majority of specimens were separated by COI, certain taxa could not be distinguished using this genetic marker within the genera Aedes, Anopheles and Culex. The use of COI also underlined identification problems in Culiseta species (Cs. fumipennis, Cs. litorea and Cs. morsitans) within the BOLD and NCBI databases. This finding supports the need for continuing research combining the use of molecular methodologies with morphological traits for species delineation in the Culicidae.

We thank the kind assistance of each of the field site employees without whom the work could not have been completed. Thanks are given to our colleagues at Public Health England (Jolyon Medlock, Alex Vaux, and Ben Cull), The Pirbright Institute (Simon Carpenter) and the Natural History Museum (Ralph Harbach) for discussions on mosquito ecology and distribution. Luis M. Hernández-Triana gives special thanks to Quetzaly Siller (Juarez University, Durango Estate) for the preparation of the distribution map. In addition, we also thank Luke Alphey (The Pirbright Institute, UK), Aleksandra Ignjatović-Ćupina (University of Novi Sad, Faculty of Agriculture, Serbia), Jenny Hesson (Liverpool School of Tropical Medicine, UK), Gale Chapman (School of Veterinary Medicine and Science, University of Nottingham, UK), Shahida Begum (London School of Hygiene and Tropical Medicine, UK), Norbert Becker and Daniel Hoffman (University of Heidelberg, Heidelberg, Germany), Renke Lühken (Bernhard Nocht Institute for Tropical Medicine, WHO Collaborating Centre for Arbovirus and Hemorrhagic Fever Reference and Research, Hamburg, Germany), Adolfo Justicia-Ibáñez (National Centre for Monitoring of Vector, Netherlands Food and Consumer Product Safety Authority, Ministry of Economic Affairs, Wageningen, Holland), and Filiz Gunay (Hacettepe University, Faculty of Science, Department of Biology, Ecology Section, ESRL Laboratories Turkey) for providing material for invasive species analyzed in this study. Funding for LMHT was provided by the UK Department for Environment Food and Rural Affairs (DEFRA), Scottish Government and Welsh Government through grants SV3045, and SE4113, and the EU Framework Horizon 2020 Innovation Grant, European Virus Archive (EVAg, grant no. 653316). Funding for VAB was additionally provided by the Biotechnology and Biological Sciences Research Council (BBSRC), grant BB/F016492/1 and the Pirbright Institute as part of his PhD project. All German collections between 2011 and 2014 were part of projects conducted by the Bernhard Nocht Institute for Tropical Medicine, Hamburg, Germany, and collaborators. This work was funded by the Leibnitz Association, grant no. SAW-2011-BNI-3, and the German Federal Ministry for Environment, Nature Conservation, Building and Nuclear Safety (BMUB) through the Federal Environment Agency (UBA), grant no. FKZ371148404.