Live adult mosquitoes were captured at dusk in the summers of 2019 and 2020 using a sweep net. Sampling took place in the North American Arctic, namely in the United States of America (Toolik, Alaska: 64°54.91'N, 147°57.96'W) and in Canada (Cambridge Bay, Nunavut: 62°7.22'N, 105°2.7'W; Karrak Lake, Nunavut: 67°14.15'N, 100°15.42'W; Kuujjuaq, Nunavik, Québec: 58°7.63'N, 68°23.08'W). The mosquitoes collected in the field were placed in a labelled plastic container and frozen at –18°C until they were shipped to the Faculté de médecine vétérinaire of the Université de Montréal (Saint-Hyacinthe, Québec, Canada). In this study, morphological terminology follows that of Wood (1979). Female mosquitoes were identified in a chilled Petri dish using a stereomicroscope (Acuter, Model T1A) using external morphological characteristics and established taxonomic keys (Carpenter and LaCasse 1955; Wood et al. 1979; Thielman and Hunter 2007). The species terminology employed follows the classification outlined by Wilkerson et al. (2015). A leg of each specimen was removed and stored in RNAlater® (Sigma-Aldrich), and the rest of the specimen was double-mounted on pins Photographs of each specimen were taken using a digital microscope (Keyence Canada Inc., Model VHX-71000). The specimens were then deposited as vouchers in the Ouellet-Robert’s entomological collection of the Université de Montréal (http://qmor.umontreal.ca).

Genomic DNA was obtained from one leg of each sample using the QIAGEN DNeasy Blood & Tissue Kit with slight modifications to the manufacturer’s protocol. Each leg was macerated with a sterile micro pestle and digested in a thermomixer (1000 rpm at 56°C for 120 min). To increase the DNA yield, 50 μl of elution buffer was added to the center of the spin column on two separate occasions and spun down for a total final volume of 100 μl. The elution buffer was heated to approximately 37°C before adding it to the spin column. Invitrogen™ Platinum™ Taq DNA polymerase was used for amplification using specific primers (Table 1). Each 25 μl reaction consisted of 2.5 μl 10 × buffer, 1 μl MgSO₄, 0.5 μl of dNTP, forward and reverse primers, 0.1 μl high fidelity DNA polymerase, 10 μl DNA template and 9.9 μl ddH₂O. The following PCR conditions were used for amplification: initial denaturation at 94°C for 60 s, followed by 35 cycles of denaturation at 94°C for 15 s, annealing at 51°C (COI) / 55°C (ITS2) for 30 s, and extension at 68°C for 60 s. The final extension occurred at 68°C for 300 s. The amplified products were visualised on a 1.5% agarose gel and purified with the QIAGEN QIAquick PCR Purification Kit according to the manufacturer’s protocol. Purified products were sent for Sanger sequencing at the National Research Council in Saskatoon (Saskatchewan, Canada). The forward and reverse sequences were assembled using the QIAGEN CLC Main Workbench. All the assembled sequences were uploaded to BOLD. GenBank accession numbers were obtained via BOLD.

The ingroup consisted of 26 species of North American black-legged Aedes (Table 2). Species were selected for this study based on two criteria: they needed to (1) have a distribution across North America (but not limited to the Arctic or subarctic regions) and (2) have DNA sequences of the COI and ITS2 barcoding genes available from captured female mosquitoes in BOLD. Mansonia uniformis Theobald, 1901 was used as the outgroup taxon. If a species met the selection criteria, up to five sequences were randomly selected and extracted from BOLD, and compiled into a FASTA file alongside the other sequences generated in this study.

Using the extracted sequences, three datasets were generated: one dataset containing COI sequences, one dataset containing ITS2 sequences, and one concatenated dataset regrouping COI sequences (partitioned as per codon position and not partitioned), ITS2 sequences, and 28S sequences. All datasets were aligned using the online version of MAFFT7 (https://mafft.cbrc.jp/alignment/server) under default parameters. The aligned matrix was viewed, trimmed and edited using MEGA7 (Kumar et al. 2016). The most appropriate substitution model was determined for each dataset using the Akaike Information Criterion (AIC) in MEGA7, which was GTR+G+I for the COI dataset, and GTR+G for the ITS2 and concatenated datasets.

Phylogenetic analyses were conducted using Maximum Likelihood (ML) with RAxML (Stamatakis 2014) and Bayesian inference (BI) with MrBayes version 3 (Ronquist and Huelsenbeck 2003). For ML analysis, robustness of the tree was tested with 1000 bootstrapped datasets, with bootstrap support values shown on each node. For BI analysis, the Markov chain Monte Carlo (MCMC) simulation was run for one million generations (which resulted in an average standard deviation of split frequencies below 0.01). All phylogenetic trees were visualised using Figtree software version 1.4.4 (http://tree.bio.ed.ac.uk/software/figtree). Bootstrap values (B) of ML analysis higher than 70 and posterior probabilities (PP) of BI analysis higher than 0.95 were considered strong support values.

A total of 25 black-legged Aedes females were collected from five localities (Table 3) (Villeneuve 2023; see ­Table S1 for the vouchers’ complete information). Morphologically, the specimens were in near-perfect condition and all specimens resembled the global description of females (Wood et al. 1979).

The final aligned matrices used for analyses had a total length of 658 bp for the COI sequences and 381 bp for the ITS2 sequences, comprising 25 and 14 sequences, respectively. The sequences were deposited in BOLD under GenBank accession numbers OR367073–OR367086 for ITS2, OR367052–OR367072 for the putative 28S section amplified with the chosen primers and OR367027–OR367051 for COI ((Table 3; see Table S2 for all the sequences used to perform these analyses and see ­Figure S1 (trees) for the results of the BI and ML analyses of all datasets)).

Analysis of concatenated COI, ITS2, and 28S gene regions generated three distinct morphology-based groups (Fig. 7): one monophyletic group (Ae. impiger), one paraphyletic group (Ae. nigripes), and one polyphyletic group (Ae. punctor and Ae. hexodontus). As expected, the molecular identification of the two isomorphic species, Ae. punctor and Ae. hexodontus, was not conclusive, further confirming the challenges associated with the Punctor subgroup.

Aedes hexodontus and Ae. punctor were recovered as a polyphyletic group in the BI and ML analyses on the concatenate dataset (Fig. 7: PP = 1.00), the ITS2 dataset (Fig. 8: B = 20, PP = 0.75), and the COI datasets, both partitioned (Fig. 9: B = 52, PP = 0.76) and unpartitioned (Fig. 9: B = 91, PP = 1.00).

Aedes impiger was recovered as a monophyletic group in the BI and ML analyses in the concatenated dataset (Fig. 7: PP = 0.96) and the COI datasets, both partitioned (Fig. 9: B = 38, PP = 0.80) and unpartitioned (Fig. 9: B = 73, PP = 0.98). The same results can be observed with Ae. nigripes for the BI and ML analyses on the COI datasets, both partitioned (Fig. 9: B = 61, PP = 0.95) and unpartitioned (Fig. 9: B = 62, PP = 0.87). Regarding the BI analysis on the concatenated dataset (Fig. 7: PP = 0.96 and PP = 0.60), Ae. nigripes was rendered paraphyletic. Aedes impiger and Ae. nigripes formed a polyphyly in the BI and ML analyses on the ITS2 dataset (Fig. 8: B = 73, PP = 0.52).

Concatenation of the gene sequences increased the power of our BI analysis for the Punctor subgroup (Fig. 7: PP = 1, vs Fig. 8: PP = 0.75 and Fig. 9: PP = 0.92), but not for Ae. nigripes (Fig. 9: PP = 0.97, vs Fig. 7: PP = 0.96 and 0.96) and Ae. impiger (Fig. 9: PP = 0.98, vs Fig. 7: PP = 0.96). Partitioning substantially decreased the support of our BI and ML analyses for the Punctor subgroup (Fig. 9: B = 91 vs 52, and PP = 1 vs 0.76) and Ae. impiger (Fig. 9: B = 73 vs 38, and PP = 0.98 vs 0.80), but slightly increased the support of our BI and ML analyses for Ae. nigripes (Fig. 9: B = 62 vs 61, and PP = 0.87 vs 0.95).

Aedes impiger and Aedes nigripes are the only species of mosquitoes found in the high Arctic of North America (Vockeroth 1954; Wood et al. 1979) with Ae. nigripes being the dominant species north of the tree line (i.e. the highest latitude at which trees can grow) (Wood et al. 1979). These two species share striking morphological similarities throughout their life stages, necessitating the development of a specific key to aid in their identification (Danks and Corbet 1973). Although the female mosquitoes of these species exhibit a noticeable hairy appearance compared to other black-legged Aedes (Vockeroth 1954; Wood et al. 1979; Thielman and Hunter 2007), this criterion alone is insufficient to confirm accurate identification.

The arrangement of postpronotal setae serves to differentiate these two species from other black-legged Aedes mosquitoes (Carpenter and LaCasse 1955; Wood et al. 1979; Thielman and Hunter 2007). However, it can be quite challenging to observe this criterion since setae are frequently absent and scales on the postpronotum can conceal the setae attachment sites (Thielman and Hunter 2007). Consequently, this criterion is easily overlooked, especially by an untrained observer, leading to misidentification. This issue primarily affects Ae. impiger, as it inhabits subarctic environments alongside multiple other Aedes species (Wood et al. 1979; Ward and Darsie 2005). In situations where the arrangement of the postpronotum setae is challenging to assess, examination of the tarsal claw should confirm the identification of Ae. impiger since no other northern black-legged species of Aedes possesses claws with a similar shape (Knight 1951; Vockeroth 1954; Danks and Corbet 1973).

Our study represents one of the first attempts to assess the performance of the COI and ITS2 barcoding regions in identifying Ae. impiger and Ae. nigripes. The results of our analysis revealed that Ae. impiger and Ae. nigripes form two distinct monophyletic clades when using the COI barcoding region, thus supporting the use of this barcoding region as an effective means of identification for these two species. Since Ae. nigripes is rendered paraphyletic in the concatenated analysis, this may be a sign of species complexes or even cryptic species. Still, the number of samples would have to be considerably increased to explore those possibilities. However, when employing the ITS2 barcoding region, Ae. impiger and Ae. nigripes grouped together in a single polyphyly, indicating that this region is not reliable for distinguishing between the two species. Based on our findings, the COI, but not ITS2, barcoding region can be used for the identification of Ae. impiger and Ae. nigripes.

The Punctor subgroup belongs to a distinct subdivision of the Aedes communis group (group G) of Edward’s classification (Edwards 1932). It was first mentioned by Dyar and revised by Knight, primarily based on the morphology of male genitalia (Dyar 1922; Knight 1951). The species belonging to the Punctor subgroup are Ae. punctor, Ae. hexodontus, Aedes aboriginis Dyar, 1917, Aedes abserratus Felt and Young, 1904 and Aedes punctodes Kirby, 1837 (Dyar 1922). Among these species, Ae. hexodontus and Ae. punctor are the only ones with distributions that reach the Arctic (Wood et al. 1979; Ward and Darsie 2005).

In subarctic regions, the morphological characters usually used to differentiate between Ae. hexodontus and Ae. punctor become obscure. In Ae. punctor, the striped scutum starts fading and white scales can sometimes be observed at the base of the costa – in these cases, Ae. punctor is scarcely distinguishable from Ae. hexodontus (Wood et al. 1979; Ward and Darsie 2005).

To help differentiate between Ae. hexodontus and Ae. punctor morphologically, the number of scales on the probasisternum can be used. Typically, female Ae. hexodontus have a heavily scaled probasisternum, whereas female Ae. punctor have a bare probasisternum (Wood et al. 1979). However, even this morphological characteristic can be doubtful, as some individuals of Ae. punctor can also exhibit scattered pale scales on the probasisternum. In our case, the number of scales on the probasisternum proved to be a dubious criterion for identifying specimens of Ae. punctor and Ae. hexodontus. Specimens with a heavily scaled probasisternum did not fit into either group. These specimens exhibited morphological characteristics associated with both Ae. punctor (a few pale scales at the base of the costa) and Ae. hexodontus (a heavily scaled probasisternum), making it impossible to identify them beyond the Punctor subgroup. Gimnig (2000) reported comparable findings for Ae. hexodontus where the majority of the specimens displayed extensive scaling on the probasisternum. However, all of their Ae. punctor specimens had completely bare probasisternum (Gimnig 2000). It is worth noting that the specimens were collected exclusively from the West coast, and none of the sampling locations were in the Arctic or subarctic regions, which may explain the morphological differences. Gimnig (2000) concluded that in females, the presence of scales on the probasisternum and the number of white scales at the base of the costa did not provide a satisfactory separation of any member of the Punctor subgroup (Gimnig 2000). Due to the limited number of specimens in our study, we cannot confirm or refute this statement. The number of scales on the probasisternum and the number of white scales at the base of the costa have potential as a differentiating characteristic between the two species, but our findings indicate that these identification criteria cannot be universally applied to all the Punctor subgroup. There are two explanations for this discrepancy: either the number of scales on the probasisternum and at the base of the costa is not a reliable diagnostic feature, or the outlier specimens collected at the same sampling location (Kuujjuaq) are hybrids. Further analyses are necessary before confirming the inclusion of the number of scales on the probasisternum and at the base of the costa as definitive identification criteria for the Punctor subgroup.

Female morphology may not help to distinguish Ae. hexodontus and Ae. punctor, but it is possible to identify the species of the Punctor subgroup based on larval morphology (Knight 1951). The numbers and the length of the comb scales were first used as a criterion, since Ae. hexodontus have noticeably fewer and larger scales (n<9; >0.1 mm) than Ae. punctor (n>10; <0.08 mm) (Vockeroth 1954). However, it was later shown that there is a much greater overlap in the number of comb scales; both species can have 5 to 12 comb scales (hex n = 4 to 12; pun n = 5 to 25) (Wood 1977). There is also an overlap in scale length in Ae. hexodontus larvae (as short as 0.089 mm) and Ae. punctor larvae (as long as 0.108 mm) (Gimnig 2000). However, other characteristics can be used, such as the shape of the cranial setae 5-C and 6-C, where Ae. hexodontus is double-branched and Ae. punctor is single-branched. However, these larval criteria are not useful as an identification tool for females captured in the field.

In addition to the morphological characteristics, the known geographic distribution could potentially help distinguish the two species. The distribution of Ae. punctor is practically confined to the boreal forest, seldom occurring to the south or north onto the tundra, whereas Ae. hexodontus is the dominant mosquito above the tree line (Wood et al. 1979). However, both species are present in subarctic locations. We also need to consider that, based on the difficulty of identification, the extent of their range is not truly known – specimens of Ae. hexodontus and Ae. punctor could have been confused with other members of the Punctor subgroup (Knight 1951; Carpenter and LaCasse 1955).

Our phylogenetic analysis showed that Ae. hexodontus and Ae. punctor form a single polyphyletic cluster when using the COI or ITS2 barcoding regions. A previous study on Canadian mosquitoes has also indicated the limitations of the COI barcoding region in delineating species within complex groups (Namin et al. 2014). For instance, a phylogenetic analysis of the Aedes communis complex using the COI barcoding region revealed comparable findings, with the complex appearing paraphyletic in relation to Ae. abserratus and Aedes implicatus Vockeroth, 1954 (Namin et al. 2014). These results underscore the need for expanded genetic and taxonomic sampling to accurately infer the phylogenetic relationships of the Punctor subgroup. Based on our findings, neither the COI nor the ITS2 barcoding regions can be used effectively for the identification of Ae. hexodontus and Ae. punctor.

Given these findings and the striking morphological similarities observed throughout the life stages of Ae. hexodontus and Ae. punctor, the question arises of whether they are truly distinct species. Some authors considered them two closely related species based mostly on larval morphology (Knight 1951; Vockeroth 1954), but these conclusions were reached using unreliable criteria for larval identification. A more recent study based on genetic and morphological comparison of Ae. hexodontus, Ae. abserratus, and Ae. punctor suggests three distinct species, but only in the southern part of their distribution (Gimnig 2000). This finding highlights the fact that, given their broad distribution, Ae. hexodontus and Ae. punctor might be separate species outside of the studied area. Overall, the question of whether Ae. hexodontus and Ae. punctor truly represent distinct species requires further investigation using comprehensive morphological and molecular approaches, especially for specimens in the northern part of their distribution.

To what extent can we trust the publicly available DNA sequences attributed to the northern black-legged Aedes species? The difficulty of using morphology to distinguish among species could easily populate the molecular libraries with sequences bearing inaccurate species designations. Furthermore, we demonstrated that the COI and ITS2 barcoding regions do not effectively differentiate among Ae. impiger, Ae. nigripes, Ae. hexodontus and Ae. punctor. Therefore, the accuracy of the sequence identification might be questionable – especially for specimens that have not been morphologically pre-identified by a taxonomist familiar with the Punctor subgroup.

For example, due to challenges associated with accessing northern sampling locations, there are not many sequences of Ae. nigripes and Ae. impiger available online on GenBank (nig n = 103 ; imp n = 46) or BOLD (nig n = 150, imp n = 145). Accuracy of identification is also a challenge. On the BOLD database (Table 4; Table 5), both Ae. impiger and Ae. nigripes share the same Barcode Index Number (BIN; i.e., AAA3750). However, these two species account for only a fraction of the specimens associated with this BIN (9%); 86% are classified as unknown specimens, while 3% are identified as Aedes cata­phylla Dyar, 1916, 1% as Ae. implicatus, and less than 1% as Aedes niphadopsis Dyar & Knab, 1918, Aedes sp., Aedes leucomelas Meigen, 1804, Aedes ventrovittis Dyar, 1916, Ae. hexodontus, Aedes intrudens Dyar, 1919, and Aedes communis DeGeer, 1776. This can be explained by misidentification of specimens at the species level. Therefore, it is not possible to use the associated BIN (AAA3750) for identifying Ae. impiger and Ae. nigripes, considering the diverse array of species linked to the same BIN. While our phylogenetic analyses indicate that Ae. impiger and Ae. nigripes form monophyletic clades when using the COI barcode, it is important to combine this identification method with other molecular and morphological traits.

Due to their wider geographic distribution, there are twice as many sequences for Ae. hexodontus and Ae. punctor on online databases such as GenBank (hex n = 201; pun n = 194) or BOLD (hex n = 327; pun n = 361). Unfortunately, the precision of BINs associated with the Punctor subgroup species is even more questionable (Table 4; Table 5). Aedes hexodontus and Ae. punctor are linked to seven and six different BINs, respectively, with three BINs shared between the two species. However, the majority of sequences associated with these species fall under BIN AAA3748, where Ae. hexodontus and Ae. punctor account for only one third of the specimens, with 58% classified as unknown specimens, 7% as Ae. abserratus, and less than 3% as Ae. aboriginis, Aedes diantaeus Howard, Dyar & Knab, 1913, Aedes sp., Ae. ventrovittis, Ae. communis, Ae. nigripes, Aedes pionips Dyar, 1919, Aedes fitchii Felt & Young, 1904, Ae. intrudens, and Aedes decticus Howard, Dyar & Knab, 1917. Given the numerous BINs associated with Ae. hexodontus and Ae. punctor and the diversity of associated species, the use of BINs is futile. Northern specimens without clear distinguishing features such as a scutum distinctly striped with a bare probasisternum cannot be identified as Ae. punctor with certainty (Wood et al. 1979). Furthermore, the absence of accompanying morphological observations associated with public sequences creates uncertainty regarding the basis of identification of morphospecies. Therefore, we propose that public sequences labelled as Ae. hexodontus and Ae. punctor and identified through COI and ITS2 molecular barcodes, be regarded as the Punctor subgroup. This relabelling should only apply if the specimen meets at least one of the following criteria: (1) sampled near the tree line, or at any Arctic and subarctic locations, (2) no morphological identification prior to sequencing, (3) unstriped (or bare) scutum and/or probasisternum displaying scales.

Accurate species identification is crucial for an effective arbovirus surveillance program, especially in northern locations where Ae. hexodontus and Ae. punctor frequently carry Jamestown Canyon virus (JCV) and Snowshoe hare virus (SSHV) (Iversen et al. 1973; McLean et al. 1977; Campbell et al. 1991; Hardy et al. 1993; Carson et al. 2017; Snyman et al. 2023). SSHV has also been isolated from Ae. nigripes (McLean and Lester 1984). Vector competence studies have also confirmed that both Ae. hexodontus and Ae. punctor can transmit one or both of these viruses (Boromisa and Grayson 1990; Heard et al. 1991; Kramer et al. 1993; Walker et al. 1993). Serological studies have shown that antibodies for both JCV and SSHV are present in wildlife and people in the Arctic (Zarnke et al. 1983; Walters et al. 1999; Miernyk et al. 2019; Buhler et al. 2023). Although human cases are rare, JCV and SSHV infections can cause symptoms ranging from fever and headache, to altered mental state ranging from confusion to coma, with or without additional signs of brain dysfunction, and death (Snyman et al. 2023). In the context of the Arctic, where the consequences of climate change are alarming, accurate species identification becomes especially significant (Hoberg and Brooks 2015). As warming trends persist, there is potential for additional vector and virus species to invade and survive in more northern regions, as well as an increase in abundance of existing species (Alto and Juliano 2001). These changes can disrupt the barriers that prevent the emergence of arboviruses by substantially affecting the population dynamics of viruses, mosquito vectors, and wildlife hosts (Høye 2020; Koltz and Culler 2021). There is a need to monitor these changes due to the burden it could impose on northern communities (Villeneuve et al. 2021). However, to ensure an effective surveillance program, accurate vector species identification is indispensable, both to detect newly introduced species as well as to avoid mistaking newly detected species as newly introduced. Given that the Punctor subgroup constitutes a substantial portion of the mosquito population and is known to act as vectors for arboviruses, it becomes crucial to address the identification challenges associated with this subgroup.

In establishing an effective surveillance program, another crucial aspect is ensuring the reliability of the vector status of northern Aedes species. Most of our information is based on dated studies, which poses a problem, as their accuracy relies on the correct identification of the specimens. Before 1977, Ae. hexodontus and Ae. punctor were regarded as separate forms, namely the “normal” type and the “tundra” type variety, based on female and larval characteristics (Knight 1951). It was only after Wood (1977) re-examined Knight’s specimens that he concluded the “tundra” type of Ae. punctor was, in fact, Ae. hexodontus (Wood 1977). Due to the challenges in identifying females of northern black-legged Aedes, we cannot be sure that both Ae. hexodontus and Ae. punctor are competent vectors for JCV and SSHV based on previous reports. Consequently, a considerable amount of work needs to be done to confirm the true Arctic vectors of JCV and SSHV and their transmission efficacy to better inform existing and future surveillance programs in the North.