Ancient origin of an urban underground mosquito

Abstract

Understanding how life is adapting to urban environments represents an important challenge in evolutionary biology. Here we investigate a widely cited example of urban adaptation, Culex pipiens form molestus, also known as the London Underground Mosquito. Population genomic analysis of ~350 contemporary and historical samples counter the popular hypothesis that molestus originated belowground in London less than 200 years ago. Instead, we show that molestus first adapted to human environments aboveground in the Middle East over the course of >1000 years, likely in concert with the rise of agricultural civilizations. Our results highlight the role of early human society in priming taxa for contemporary urban evolution and have important implications for understanding arbovirus transmission.

Main

The rise of modern cities is rapidly reshaping our planet and imposing novel selective pressure on the living organisms around us. Many species have begun to adapt to these unique challenges. A review of the literature highlights at least 130 examples of animals, plants, and microbes that have evolved responses to dense urban environments (1). Yet how such adaptations occur and the amount of time they require remain poorly understood. As urbanization accelerates over the coming decades (2), there is a pressing need to better understand the mechanisms and timescale of urban adaptation.

One of the most widely cited examples of urban adaptation involves the northern house mosquito Culex pipiens Linnaeus, 1758 (Fig. 1A). Cx. pipiens is common in temperate zones across the world (3, 4). In Europe and North America, an ancestral form has long been appreciated as a bird-biting mosquito that requires open space for mating and pauses reproduction (i.e. diapauses) during the cold northern winter (Fig. 1B–C, blue) (3). However, a derived, human-biting form thrives in urban belowground environments, such as subways, cellars, and cesspits, and differs from its aboveground counterpart in ways that seem perfectly suited to subterranean life (Fig. 1B–C, red) (3). The belowground mosquitoes are able to mate in confined spaces and remain active in winter. Adult females readily bite humans (and other mammals). Yet if hosts are scarce, they can develop a first clutch of eggs without taking any blood, a trait known as autogeny. Strikingly, despite this array of genetically based behavioral and physiological differences, the two mosquitoes show no consistent morphological differences (3). They are formally considered distinct forms: the bird-biting Cx. pipiens f. pipiens Linnaeus, 1758 and the human-biting Cx. pipiens f. molestus Forskål, 1775 (5), hereafter referred to as simply pipiens and molestus, respectively.

Figure 1. Culex pipiens form molestus behavior, ecology, and hypothetical origin.

(A) Female Cx. pipiens complex mosquito. (B) Behavioral and physiological characteristics of Cx. pipiens forms in northern Eurasia. At warmer latitudes molestus can breed aboveground. (C) Example microhabitats: a city park (pipiens) and the flooded basement of an apartment complex (molestus). (D) Two hypotheses describing molestus’ origin. Hypothesis 1 (left) posits that belowground molestus evolved from local aboveground pipiens in situ within the past 100–200 years. Hypothesis 2 (right) posits that molestus first evolved in an aboveground context thousands of years ago, possibly in association with early agricultural societies of the Mediterranean basin, with colonization of belowground habitats (dotted arrow) occurring much later (22, 23). [Image credits: Lawrence Reeves (mosquito); Yuki Haba (city park); Colin Malcolm (flooded basement)]

The sophisticated adaptations of molestus to urban belowground environments have led to much speculation over when and where it originated. A widely cited hypothesis suggests that molestus evolved in the London Underground subway system, where it first became famous in the 1940s during World War II (1, 6–11). At that time, many Londoners took nightly refuge in the city’s subway system to escape intense Nazi bombing. Sleeping on subway platforms protected people from bombs but made them easy targets for molestus, which became known as the ‘London Underground Mosquito’ and was hypothesized to have evolved there during the ~100 year period between subway tunnel construction and mosquito discovery (Fig. 1D, left) (6, 12). Form molestus was reported in cellars and cesspits in France, Denmark, Germany, and the former USSR 10–25 years before its discovery in London (3, 13–15), but an urban, belowground origin in northern Europe within the past few hundred years remains possible. Recent reviews have pointed to molestus as one of the best candidates for rapid urban adaptation (1, 7–11), and major science news outlets treat this hypothesis as fact (16–21). The idea that such an array of traits could emerge de novo in just a few hundred years is striking and sets a new bar for the number and complexity of changes we might expect to occur in modern cities over short timescales.

An alternative hypothesis, which is mentioned but less prominent in the literature, posits that molestus first adapted to humans in an aboveground context, long before the rise of modern cities (Fig. 1D, right) (22, 23). While molestus is confined to belowground habitats in cold regions, it thrives aboveground in warmer climates, particularly in the Mediterranean basin (23). Moreover, early records document molestus-like mosquitoes breeding and biting humans aboveground in Egypt, Croatia, and Italy 50–100 years before they were discovered in basements and subways (24–26). According to this alternative scenario, many of the traits that allow molestus to thrive in urban belowground environments would represent exaptations, or traits that first arose in a different time and context (27). An aboveground Mediterranean origin could also push the timing of molestus’ origin back thousands of years, to an era when humans first started forming dense agricultural communities. Early allozyme and microsatellite studies indicate that contemporary molestus populations from aboveground and belowground habitats are genetically related (22, 28), but the validity and timing of a putative aboveground origin remain to be tested.

Here we leverage the first large population genomic dataset for Cx. pipiens to infer when, where, and in what ecological context molestus first evolved. Beyond its enigmatic origins, molestus is a competent disease vector, implicated in the transmission of West Nile virus and other arboviruses across Eurasia and North America over the past several decades (29, 30). Solving the mystery of molestus’ origins thus has important implications for understanding both rapid urban adaptation and emerging threats to human health.

Form molestus is genetically isolated from pipiens across the Western Palearctic

Multiple lines of evidence indicate that molestus first split from pipiens somewhere in the Western Palearctic (a region that includes Europe, North Africa, and western Asia) (31), before spreading to other parts of the world. However, the structure of populations across this region has been difficult to decipher due to the absence of morphological differences. Analysis of one or a small number of genetic loci shows that the two forms are isolated in northern Europe, where harsh winters confine molestus to belowground environments (22, 23). However, molestus and pipiens appear to be more genetically similar in southern Europe, where both breed aboveground, and may even collapse into a single panmictic population in North Africa (22, 23). To better resolve the situation with high-resolution genomic data, we sequenced the whole genomes of 357 Cx. pipiens individuals collected in 77 locations scattered across the Western Palearctic (Fig. 2A; n = ~5 individuals per population at 12.9X median coverage). These data are part of a larger collection of 840 genomes to be presented in a companion study of the deeper evolutionary history of Cx. pipiens across its entire global range (32).

Fig. 2. Form molestus is genetically isolated from pipiens across the Western Palearctic.

(A) Sampled populations, colored by average PC1 value. Circles and triangles represent aboveground and belowground locations, respectively. Half-circles indicate that both pipiens and molestus were collected in the same or nearby aboveground sites. (B) PCA of genetic variation across all samples in (A) (n = 357). (C) PC1 values plotted against latitude, with marginal frequency histogram at top. The gray dashed line indicates a natural break in the histogram, inferred to separate pipiens and molestus (PC1 = 0.04). Thick outlines mark individuals from North Africa and the Middle East. Asterisk marks a putative F1 hybrid from southwest Russia (Stavropol). Inset shows position of historical London samples in a combined PCA with contemporary mosquitoes (n=22, collected 1940–1985; see also fig. S1). (D) PC1 values for pipiens and molestus individuals collected in the exact same day and trap (green lines) or in the same general location (within 5–45 km; grey lines).

We used a Principal Component Analysis (PCA) to assess variation across the Western Palearctic using 504,000 high quality SNPs (32) (Fig. 2B). The first major axis (PC1) accounted for by far the most variation (39.5%, fig. S2A) and was thus likely to represent divergence between pipiens and molestus. Consistent with this prediction, belowground and aboveground samples from northern latitudes were clustered at opposite ends of the PC1 axis (Fig. 2C). Sequenced mosquitoes with known biting or egg-laying behavior (n = 13), including those from lower latitudes, were also arrayed across PC1 according to expected form (fig. S1A). PC2 explained ~4% of genetic variation across the sample (Fig. 2B) and was strongly correlated with longitude (fig. S2B).

Our sample included aboveground mosquitoes from London, which clustered with other northern European pipiens, but we were not given permission to collect mosquitoes in the London Underground. To confirm that the genetic picture today reflects the one present when iconic WWII populations were first discovered, we used a minimally destructive approach (33) to extract and sequence DNA from 22 museum specimens collected at 15 sites in London between 1940 and 1985 (table S2, mean genome-wide coverage = 5.8X). Metadata for most samples did not specify microhabitat, but the sampling locations included the sites of major underground stations, including Paddington, Monument, and Barking. A joint PCA with contemporary samples placed the historical London specimens in the same two genetic clusters that characterize mosquitoes at that latitude today (Fig. 2C, inset; fig. S1B). We conclude that the genetic character of pipiens and molestus populations in northern Europe has been stable for the past 75 years.

Form pipiens and molestus are genetically well separated in the north, but our data confirm that they are less distinct at southern latitudes. Mosquitoes on both the molestus and pipiens ends of the PC1 axis have increasingly intermediate values as one moves from northern Europe towards Africa, creating a striking U-shaped pattern when PC1 is plotted against latitude (Fig. 2C). Importantly, however, they never completely merge; even southern populations fall into two discrete genetic clusters with a break at PC1 ~0.04 (Fig. 2C, dashed line). Moreover, individuals from these two clusters were frequently collected in the same traps, highlighting the absence of microgeographic barriers (Fig. 2D). Our whole genome data thus show unequivocally that pipiens and molestus are able to coexist in sympatry across the region (34, 35). They are genetically closer in the south, and several individuals in our sample may represent early generation hybrids (e.g. see asterisk in Fig. 2C). Despite this, we see no evidence of collapse into a panmictic population.

Ancestral latitudinal gradient within pipiens suggests molestus arose at the southern edge of the Western Palearctic

The genetic similarity of molestus and pipiens at southern latitudes is believed to result from increased gene flow (22, 23); hybridization should be rare in the north where the two forms occupy different microhabitats, but increasingly common in the south where both breed aboveground (Fig. 3A). To test this hypothesis we examined the latitudinal cline within pipiens, which is much stronger than that within molestus (Fig. 2C). More specifically, we used genome-wide f3 statistics (36) to model each pipiens population as a mix of ‘pure’ pipiens and molestus reference populations taken from their northern extremes. Many European and west Asian populations showed evidence of mixing (Fig. 3B), but the signal was not latitudinal (Fig. 3C; Pearson’s r = −0.022, P = 0.90). Moreover, North African pipiens populations, which are genetically closest to molestus, showed no signs of admixture (Fig. 3B). These results cast doubt on the longstanding hypothesis that latitudinal variation within pipiens is driven by hybridization with molestus.

Fig. 3. Ancestral latitudinal gradient within pipiens suggests molestus arose at the southern edge of the Western Palearctic.

(A) Hybridization gradient hypothesis: the genetic gradient within pipiens may result from increasing levels of gene flow with molestus as one moves from north to south. (B) Z scores of genome-wide f3 values for each pipiens population when modeled as a mixture of northern pipiens (Sweden) and northern molestus (Belgium). Significantly negative f3 values (Z < −3, green outlines) are consistent with the presence of admixture. (C) f3 statistics of pipiens populations with significant signs of admixture, plotted against latitude. (D) Ancestral gradient hypothesis: the genetic gradient within pipiens may be ancestral, with molestus evolving from southern pipiens populations. (E) Fraction of derived alleles shared by each pipiens population with northern pipiens versus northern molestus. Culex torrentium was used as the outgroup. Light brown in the map shows the Mediterranean climate zone. (F) Fraction of derived alleles shared with northern pipiens vs. molestus, plotted against latitude. Both (C) and (F) include linear regression line with 95% confidence interval and Pearson’s correlation test statistics. Across all analyses, only populations with four or more individuals were included.

An alternative hypothesis, which has not been explored in the literature, is that the latitudinal gradient within pipiens predates the evolution of molestus. In this case, southern pipiens could be genetically closer to molestus not because they mix with molestus, but because they gave rise to molestus (Fig. 3D). Consistent with this idea, we found that southern pipiens—and especially pipiens populations in the Mediterranean basin—share as many, or more, derived alleles with a reference molestus population from the north than they do with a reference pipiens population from the north (Fig. 3E). Moreover, the overall signal of relative allele sharing was strongly latitudinal (Fig. 3F, Pearson’s r = 0.88, P = 2.5 × 10⁻¹⁴). Taken together, we conclude that the latitudinal gradient within pipiens is ancestral—perhaps reflecting adaptation to variation in temperature and/or precipitation—and that molestus is most likely derived from populations in the south.

Form molestus evolved thousands of years ago in the Middle East

We further explored the geography of molestus’ origin by constructing a distance-based (Dxy) tree for Cx. pipiens individuals from the full global sample (32). Contemporary gene flow can obscure ancestral relationships in phylogenetic trees. We were therefore careful to exclude any individual or population that showed signs of recent introgression from the other form (fig. S3–S4, S8) or from Culex quinquefasciatus, a tropical sibling species that hybridizes with Cx. pipiens in the Americas and Asia (32). We also excluded low coverage samples (<10X), leaving a total of 205 individuals.

The resulting tree provided strong support for a southern origin of molestus. First, all molestus samples formed a monophyletic clade that was nested within Mediterranean pipiens (Fig. 4A, fig. S5). Second, the earliest branching lineages within molestus corresponded to aboveground mosquitoes from the eastern Mediterranean—specifically Egypt, Israel, and Greece (Fig. 4A). Egyptian and Israeli samples were also among the most genetically diverse, together with two populations from the Caucasus region (Armenia and southern Russia) (Fig. 4B). Finally, while belowground molestus from northern latitudes formed tight, derived clades, aboveground molestus populations from North Africa, the Middle East, and southern Europe were scattered across the base of the tree (Fig. 4A). These results support the hypothesis that molestus first evolved in an aboveground context in the greater Mediterranean basin, and more specifically in the Middle East.

Fig. 4. Form molestus evolved thousands of years ago in the Middle East.

(A) molestus clade excerpted from neighbor-joining tree based on pairwise genetic distance (Dxy) among putatively unadmixed pipiens and molestus individuals from the global sample (32). Terminal branches are collapsed at the root of each population, with a symbol and number indicating microhabitat and sample size, respectively (32). Map inset shows distribution of two subgroups of molestus from the tree (orange/red) and pipiens (dark grey). Only molestus is present in Egypt, marked by an asterisk. Black circles mark nodes with >95% bootstrap support. See fig. S5 for the full tree. (B) Genome-wide nucleotide diversity (π) of populations shown in (A). (C) Relative cross-coalescence (rCC) rate between Moroccan pipiens (MAK) and Egyptian molestus (ADR) inferred from phased, whole genome sequences (32). rCC rate is expected to plateau at 1, going backwards in time, when populations have merged into a single ancestral population. Rapid divergence is observed between ~10K and 1K years ago, with a split time (T; rCC rate=50%) of 2,141 years (black dashed arrow) based on estimates of generation time and mutation rate (32). Biologically reasonable upper and lower bounds for these parameters give minimum and maximum split times of 1,298 and 12,468 years (grey arrowheads). Thick black line shows genome-wide result, and light grey lines show 100 bootstrap replicates. Illustration credits: Wheat, Freepik.com. Pharaoh and Glacier, Vecteezy.com.

The Middle East is a compelling location for the emergence of molestus not only because it harbors diverse, early branching populations (Fig. 4A–B), but also because it is the only place within the Western Palearctic where molestus is known to occur on its own, in the absence of pipiens (Fig. 2A) (28, 37). The Middle East was also home to some of the earliest agricultural societies, which were thriving in Mesopotamia and Egypt by 3000 BCE (38). A Middle Eastern origin thus raises the possibility that molestus first adapted to human hosts and habitats in isolation from pipiens and on a timescale of thousands, rather than hundreds, of years (Fig. 1D, right) (22, 23).

We explored the timing of molestus’ origin using a cross-coalescent analysis of DNA haplotypes from Middle Eastern molestus (Egypt) and Mediterranean pipiens (Morocco) (n = 2 individuals with ~50X coverage from each population) (39). As expected, the relative cross coalescence (rCC) rate starts near zero in the recent past, when the two populations are isolated, but rises monotonically and eventually plateaus near one, going backward in time, when they merge into a single ancestral population (Fig. 4C). Accurate assignment of dates to this rCC curve requires knowledge of the de novo mutation rate (μ) and generation time (g), neither of which has been directly measured for Cx. pipiens in nature. However, plausible literature estimates for μ (4.85×10^–9) and g (20 days) (32) suggest that peak rates of divergence occurred ~2,000 years ago (Fig. 4C, dashed arrow), while minimum and maximum reasonable values (32) lead to split times anywhere between 1,300 and 12,500 years ago (Fig. 4C, grey arrowheads). We obtained a similar range of split times when using an alternative pipiens population from the southern Caucasus region (Armenia; fig. S7). Taken together, these results are inconsistent with a post-industrial origin for molestus in northern Europe (Fig. 1D, left) and instead support an ancient origin associated with early agricultural civilizations of the Middle East (Fig. 1D, right).

Introgression from molestus into aboveground pipiens is associated with human density

Recent urbanization did not drive initial evolution of molestus, but it may have driven its expansion and increased contact with pipiens across the northern hemisphere—contact that is thought to have contributed to the emergence of West Nile virus (WNV) in human populations over the last several decades (29, 30). WNV is a mosquito-borne virus that primarily infects birds and is effectively amplified within avian populations by bird-biting pipiens (Fig. 5A). Spillover to dead-end human hosts can only occur if local pipiens mosquitoes are also willing to bite humans, a broadening of biting behavior that may be driven by gene flow from molestus in urban areas (40–42). This idea has spurred efforts to detect and quantify admixture between pipiens and molestus in natural populations (23), yet we show above that much of the genetic signal previously attributed to mixing between forms instead represents ancestral variation (Fig. 3). Form pipiens likely receives genetic input from molestus in some places, but exactly where and to what extent are not known.

Fig. 5. Introgression from molestus into pipiens is associated with human density.

(A) Schematic of West Nile virus transmission dynamics. Form pipiens-molestus hybrids, which have intermediate biting preference (42), are implicated in the spillover of WNV from birds to humans (40, 41). (B) Base tree used to simultaneously estimate three potential sources of introgression into focal pipiens populations (X pip): northern pipiens (Sweden, SWE), Middle Eastern molestus (Egypt, ADR), and northern molestus (Belgium, BVR). (C–D) Population-specific estimates of ‘gene flow’ from northern pipiens (C), which accounts for the ancestral latitudinal gradient, and northern molestus (D). We did not detect gene flow into any population from Middle Eastern molestus (fig. S8). (E) Correlation between gene flow from northern molestus (D) and human population density within circles of varying radius around each collection site. (F) Gene flow from northern molestus as a function of human density within a 3-km radius of each collection site. Three outliers in grey (Cook’s distance > 4) were excluded, but regression remains significant if included (P = 0.005, R² = 0.18). [Image credit: Phylopic.com (bird and human)]

We used f-branch statistics to reassess levels of gene flow from molestus into pipiens while controlling for ancestral variation. f-branch statistics allow simultaneous quantification of gene flow among multiple branches in a tree (43). To account for the sister relationship between forms in the south (Fig. 3D), we specified a fixed tree in which focal pipiens populations were genetically closer to molestus than to a reference pipiens population from the far north (Fig 5B). Deviations from this topology (i.e. for focal populations from the north) can then be modeled as significant ‘gene flow’ into the focal pipiens population from the northern reference (Fig. 5B, arrow 1). As expected, the resulting signal was strongly latitudinal (Fig. 5C, fig. S8A). The tree also included two potential sources of molestus introgression, allowing us to distinguish them. We could not detect gene flow into any pipiens population from the early branching molestus lineage in Egypt (Fig. 5B arrow 2; fig. S8C), consistent with its isolated location at the southern edge of the contemporary range. However, we detected substantial gene flow into some pipiens populations from a derived molestus lineage present in the north (Fig. 5B arrow 3, Fig. 5D, fig. S8B).

Levels of gene flow from northern molestus into pipiens did not covary with latitude (Fig. 5D, fig. S8B), but showed a positive association with human population density (44–46). The more humans living within 1–10 km of each sampling location, the more likely we were to observe introgression (Fig. 5E). This relationship was most significant when averaging human density across an area with 3 km radius (linear regression P = 0.003, R² = 0.21, Fig. 5E–F), suggesting that levels of urbanization immediately around collection sites are most predictive of hybridization. Moreover, this signal was driven primarily by the consistent presence of ~5% introgression in truly ‘urban’ areas, defined by the European Commission as having >1500 people per km² (Fig. 5F) (47). Introgression was less predictable at rural sites (P = 0.07, R² = 0.11 excluding urban centers). Inclusion of three statistical outliers in this analysis (Fig. 5F, grey dots) slightly weakened the trend, but still indicated a strong association (regression P = 0.005, R² = 0.18). Interestingly, a site in Paris showed ~15% introgression, as one might expect based on its density, but we could not detect any genetic input from molestus in London. Such geographic variability in levels of gene flow may in part reflect whether local pipiens and molestus populations are infected by compatible or incompatible strains of Wolbachia pipientis bacteria (48, 49). Taken together, our results counter the longstanding idea that gene flow between pipiens and molestus is greatest at southern latitudes (where both forms breed aboveground) and instead reveal a complex landscape of introgression that is modestly associated with levels of human activity.

Discussion

Understanding how life can adapt to rapid urbanization is an important challenge in evolutionary biology. As examples accumulate in the literature, each new case provides a reference for the potential speed and character of adaptation. Here, we revisit one of the most iconic such examples using high resolution population genomic data. Instead of evolving in the subway system of a northern European city over the course of 100–200 years, our results suggest that Cx. pipiens f. molestus first adapted to human hosts and habitats in the Middle East over the course of 1000 or more years, possibly in association with early agricultural societies (Fig. 6). Early agricultural settlements would have provided a novel ‘human’ niche in a region that might otherwise have been too arid to support robust Cx. pipiens populations. Irrigation systems offer rich breeding sites for larval stages, while abundant humans and domestic animals offer a reliable source of blood for adult females. We cannot say where exactly within the Middle East adaptation first occurred, as our sampling in the region is limited and ranges shift over time. However, molestus is particularly abundant in Egypt’s Nile basin, and ancient pharaonic artifacts and papyrus are consistent with the idea that molestus was spreading filarial worms among humans there as many as 2000 years ago (23).

Fig. 6. Inferred evolutionary history of molestus.

Three sequential panels show the inferred history of molestus. Left: The latitudinal genetic gradient characterizing extant pipiens populations in the Western Palearctic predates molestus. Middle: The rise of dense, settled, agricultural communities in the Middle East 2000–10,000 years ago, including along the Nile River in Egypt, would have provided new ecological opportunities for mosquitoes that could adapt to human hosts and habitats—driving the evolution of molestus. Right: Eventually molestus must have spread north, where it became established alongside pipiens in the warm Mediterranean region by the 1800s (but possibly earlier) and in belowground microhabitats of cold, northern cities by the early 1900s. Our distance tree (Fig. 4A) suggests that molestus was further spread (most likely by humans) overland all the way to East Asia and from the Mediterranean region overseas to America and Australia.

Rather than benchmarking the speed and complexity of urban evolution, this updated history highlights the role of preexisting traits, or exaptations, in adaptation to urban environments (27). Three of the key behaviors that allow molestus to thrive belowground are present in contemporary Egyptian populations and almost certainly arose in ancient times: mammal-biting, the ability to mate in confined spaces, and the ability to lay a first clutch of eggs without a blood meal (37, 50). A fourth trait, lack of diapause, which limits molestus to belowground environments at northern latitudes, is also present in the Middle East today (37). Form molestus was thus primed to take advantage of northern, belowground environments before they arose. Indeed, it joins a host of other ‘urban’ taxa that first became dependent on humans thousands of years ago, including brown rats (51), house mice (52), cockroaches (53), house sparrows (54), and the dengue mosquito Aedes aegypti (55).

Ancient origins do not preclude additional, contemporary evolution (51, 53). Once established belowground, molestus was likely exposed to a new suite of challenges. For example, many belowground habitats lack vertebrate hosts altogether, providing a competitive edge to females that can develop eggs without a bloodmeal. This trait, called autogeny, is present in contemporary Middle Eastern molestus, but only at low frequency (37, 50). In contrast, it occurs at high frequency in some aboveground molestus from southern Europe (34) and is nearly fixed in northern belowground populations (Fig. 1B). Future work should explore whether increased autogeny may provide a bona fide example of rapid, urban evolution in belowground environments and whether this change arose just once, or many times in parallel (56). Our distance tree (Fig. 4A) suggests that belowground populations from northern Eurasia are all closely related, but that those from the east coast of North America (and possibly Australia) represent independent colonization events.

Our findings also carry public health implications. The emergence and spread of WNV over the last two decades has triggered an intense interest in quantifying admixture between molestus and pipiens, as hybridization is thought to drive spillover from birds to humans (40, 41). Yet we show that true patterns of admixture are obscured by ancestral relationships at southern latitudes. In particular, it is currently standard practice to identify ‘hybrids’ using a single locus marker called CQ11 (57). Pure pipiens and molestus were thought to be fixed for different alleles at this locus such that heterozygotes must be ‘hybrids’. Our data instead suggest that pipiens harbors ancestral variation at this locus. The ‘molestus’ allele was likely present at moderate frequency in the Mediterranean basin before molestus arose, and remains present in ‘pure’ Mediterranean pipiens today. Accurate inferences of gene flow will require more substantial genomic data and relatively complex analytical methods (e.g. Fig. 5). Future work should also consider the possibility that Mediterranean pipiens populations are somewhat intermediate between canonical northern forms at the phenotypic, as well as genetic, level (58). They may be effective bridge vectors even in the absence of genetic input from molestus. WNV represents an increasing threat to public health across the northern hemisphere, with many of the most severe outbreaks occurring within the last 5 years (59, 60). Taken together, we hope that our work opens the door to more incisive investigation of the potential links between urbanization, gene flow, ancestral variation, and viral spillover.

Funding Statement

This work was supported by grants or fellowships from the Masason Foundation (YH), Honjo International Scholarship Foundation (YH), Pacific Southwest Center of Excellence in Vector-borne Diseases (CSM), Society for the Study of Evolution (Rosemary Grant Advance Award to YH), Princeton High Meadows Environmental Institute (Walbridge Fund Graduate Award to YH), the American Philosophical Society (Lewis and Clark Field Scholar Award to YH), and the New York Stem Cell Foundation. CSM was a New York Stem Cell Foundation—Robertson Investigator. MKNL and PK were supported by core funding from Wellcome (220540/Z/20/A), which also supported the historic mosquito sequencing costs. LL was funded by the French Government’s Investissement d’Avenir program, Laboratoire d’Excellence Integrative Biology of Emerging Infectious Diseases (ANR-10-LABX-62-IBEID). AdT and BC were supported by Ministero della Ricerca (Italy), Piano Nazionale di Ripresa e Resilienza, and the EU’s Extended Partnership initiative on Emerging Infectious Diseases (project number PE00000007). EF was supported by grants from Consejería de Economía, Comercio e Innovación of the Extremadura regional Governments, Spain (IB10044 and IB16135). CM was supported by grants by Fundação para a Ciência e a Tecnologia (Starting Grant IF/01302/2015; GHTM UID/Multi/04413/2020; LA-REAL LA/P/0117/2020). JB was supported by the US National Institute of Allergy and Infectious Diseases (R01AI148551). Sampling and species identification in Belgium was supported by the Flemish, Walloon, and Brussels regional governments and the Federal Public Service in the context of NEHAP (MEMO project, CES-2016-02) as well as the Belgian Science Policy Office and the Department of Economy, Science and Innovation of the Flemish government. MOA, APGA, CM and MTN were funded by Global Health and Tropical Medicine, a unit of the Instituto de Higiene e Medicina Tropical of the Universidade NOVA de Lisboa.

Data and materials availability:

All genome resequencing data associated with this study is available in the NCBI SRA (790 contemporary genomes PRJNA1209100, 22 genomes from archival museum specimens ERS10924505–ERS10924526). Associated scripts will be available on GitHub (github.com/YukiHaba).