Main

In Africa, malaria spans the humid equatorial forest to the semi-arid zones in the north and south. In regions in which surface water—which is essential for larval development—is absent during the 3–8-month dry season, mosquito densities and disease transmission drop markedly3,4,5,6,7,8. However, shortly after the first rains, populations of vectors surge6, and transmission recommences. Recent studies suggest that in the Sahel A. coluzzii survives the long dry season by aestivation (a period of dormancy)3,6,9,10, whereas A. gambiae sensu stricto (hereafter, A. gambiae) and Anopheles arabiensis re-establish populations by migration from distant locations, at which larval sites are perennial3. However, direct evidence, such as the capture of aestivating adults in their shelters or the recapture of marked mosquitoes at sites that are hundreds of kilometres from their release sites, remains limited.

The dispersal of mosquitoes (hereafter referred to as migration11) has been extensively studied because it directly affects disease transmission, the spread of adaptations (for example, resistance to insecticides) and strategies for controlling mosquitoes (such as insecticide barriers)12,13. Although tracking mosquitoes over large scales has seldom been attempted12,13, the prevailing view is that the dispersal of malaria mosquitoes12,13,14,15 does not exceed 5 km and long-range movements16,17,18,19 represent ‘accidental events’ that are of minimal epidemiological importance12. Nonetheless, the prediction of long-distance migration of anopheline mosquitoes in the Sahel prompted us to question this view. To our knowledge, our study is the first to systematically sample insects that migrate at high altitudes over multiple seasons in Africa. We aimed to determine whether malaria vectors engage in wind-assisted movements and—if so—to assess the epidemiological relevance of this movement by addressing questions regarding the species involved, the frequency and heights of flights, how many mosquitoes migrate and how likely these mosquitoes are to carry Plasmodium. We then use simulations to estimate how far mosquitoes may have travelled, and from where.

During 617 nights on which aerial sampling took place, we caught 461,100 insects at heights that ranged between 40 and 290 m above ground level in four villages in the Sahel of Mali (Extended Data Fig. 1). These insects included 2,748 mosquitoes, of which 235 were anopheline mosquitoes (Table 1). These mosquitoes belonged to ten species: A. coluzzii, A. gambiae, Anopheles pharoensis, Anopheles coustani, A. squamosus, A. rufipes and A. namibiensis, as well as three distinct but currently undetermined Anopheles species (referred to here as Anopheles Mali species 1, Anopheles Mali species 2 and Anopheles species near (sp. nr) concolor) (Table 1). A. coluzzii and A. gambiae are the primary vectors of malaria in Africa, and A. pharoensis, A. coustani, A. squamous and A. rufipes are of secondary importance20. Mosquitoes were not among the 564 insects that were captured on 508 control nets (Table 1, Methods), which confirmed that these Anopheles mosquitoes were intercepted at altitude rather than near the ground during deployment. The maximum number of an anopheline species caught per night was five, which indicates that migration occurred over many nights. Consistent with Poisson distributions, the values of the variance:mean ratio were all near one (Table 1, Supplementary Discussion). Unless specified otherwise, the quantitative results presented here refer to the 5 most-abundant species, represented by more than 20 individuals (Table 1).

Table 1 Summary of mosquitoes collected in aerial samples in 2013–2015
Full size table

Females outnumbered males by more than 4:1 (Table 1). Critically, more than 90% of the anopheline females had taken a blood meal (2.9% were blood-fed, 87.5% were fully gravid and 0.7% were semi-gravid; blood feeding is required for the eggs of females to mature) before their high-altitude flights (Table 1), which suggests they were probably exposed to malaria and other pathogens. Although 31% of the blood meals came from humans, no Plasmodium-infected mosquitoes were detected among the 22 A. gambiae sensu lato (here represented by A. gambiae and A. coluzzii) that we tested, or the 174 secondary vectors (Table 1). Considering the typical rates of Plasmodium infections in primary (1–5%) and secondary (0.1–1%) vectors5,21,22,23, our results probably reflect the small sample size: the likelihood for zero infected mosquitoes was over 30% and over 18% (assuming the highest rates in each range) in the primary and secondary vectors, respectively (Supplementary Discussion). Therefore, unless infection reduces migratory capacity or migrants are resistant to parasites (and there is currently no evidence for either), Plasmodium and other pathogens are almost certainly transported by windborne mosquitoes that may infect people post-migration.

The mosquitoes were intercepted between 40 and 290 m above ground level (Fig. 1a). The mean panel density of mosquitoes per altitude and the corresponding aerial density (the panel density divided by the volume of air sampled) increased with altitude and there was a significant effect across species on mean panel density (P < 0.037, F1,24 = 4.9) (Extended Data Fig. 2b), which suggests that the migration of anopheline mosquitoes also occurs more than 290 m above ground level. The similar distribution of species across years and locations (Extended Data Fig. 2c; non-significant effects of year and location shown in Extended Data Table 1), combined with the marked seasonality of the migration or high-altitude flight activity (the aerial captures of mosquitoes occurred between July and November, and peaked between August and October) (Fig. 1b, Extended Data Table 1), attest to the regularity of the migration of windborne Anopheles mosquitoes.

Fig. 1: Flight altitude, seasonality, wind speed and abundance of migratory anopheline species.
figure 1

a, The relationship of altitude (height of the collecting panel) to panel density (blue) and aerial density (orange, mosquitoes per 106 m3 of air), for the five most-common anopheline species (Table 1). Bubble size is proportional to mosquito density (the value shown in the bubble × 103); when the value is zero, only a dot is shown. The number of sampling nights for each of the collecting-panel heights is shown on the left. b, Monthly panel density (n = 1,894 panels) for the five most-common species (Table 1), overlaid by the length of migration period (dashed lines). Values for A. squamosus were divided by three to preserve the scale. The sampling month for species that were collected only once or twice is shown by letters: c1, A. namibensis; g, A. gambiae; m1, Anopheles Mali species 1; m2, Anopheles Mali species 2; nC, A. sp. nr concolor. c, Distribution of the mean nightly wind speed at flight height on nights on which one or more anopheline mosquitoes were collected. Wind-speed data were taken from ERA5 database after matching the height of the collecting panel to the nearest vertical layer (Methods). The corresponding box and whisker plot (top) shows the median, mean, quartiles and extreme values overlaid by arrows indicating the mean and 10th and 90th percentiles (red). d, The number of mosquitoes per species that cross, at altitude (50–250 m above ground level), imaginary lines perpendicular to the prevailing wind direction. Migrants per night per 1 km (right y axis) are superimposed on the annual number of migrants per 100-km line (left y axis).

Full size image

Using mean aerial densities and wind speeds at altitude (4.8 m s−1) (Fig. 1c) and conservatively assuming that mosquitoes fly in a layer between 50 and 250 m above ground level, we estimated the nightly expected numbers of migrants crossing a 1-km line perpendicular to the wind direction. Nightly estimates ranged between 27 (for A. gambiae) and 3,719 (for A. squamosus) (Fig. 1d). When interpolated over a 100-km line that links our sampling sites (Extended Data Figs. 1a, 2c), annual migrations are estimated to exceed 80,000 A. gambiae, 6 million A. coluzzii and 44 million A. squamosus mosquitoes in that region alone (Fig. 1d). Thus, the migration of windborne mosquitoes in the Sahel occurs on a massive scale.

For each mosquito capture, we estimated the flight trajectories for 2- and 9-h-long flights using the hybrid single-particle Lagrangian integrated trajectory (HYSPLIT) model24 with the most accurate assimilated meteorological data available (ERA5), and assuming that mosquitoes ascend by their own flight but are passively carried by the wind at altitude (Methods). The mean nightly displacements were 30 and 120 km (maxima of 70 and 295 km) for 2-h and 9-h flights, respectively (Fig. 2, Table 2). Notably, the maximal 9-h nightly flight displacements ranged between 257 and 295 km for all anopheline species with sample size of more than 20 insects (Table 2). These backwards trajectories exhibited a southwestern origin (Rayleigh test; a mean bearing of 212°, r = 0.54, P < 0.0001) (Table 2), which corresponds to the prevailing winds during peak migration (August to September) (Fig. 2). The trajectories of most species originated from a broad arc (over 90°) (Fig. 2), which suggests migrants emanated from multiple sites across a large region. Migration from this direction is consistent with the presence of high-density populations due to perennial larval sites and earlier population growth following the monsoon rains. The backwards trajectories with a strong northerly component, which were observed during the sparsely sampled period of October to December (Fig. 2), might indicate southward ‘return flights’ on the Harmattan winds that prevail during this season.

Fig. 2: Backwards trajectories of the flights for each Anopheles capture event.
figure 2

Backward 9-h trajectories were estimated by HYSPLIT (Table 2), and are overlaid on a map showing parts of Mali and neighbouring countries. Map data are from Google, Landsat/Copernicus 2019. Each line represents one of four simulated trajectories of one (or more) mosquitoes intercepted at that location and on that night. The area encompassed by the four trajectories is shadowed. Migration season is shown by line colour. The Anopheles species is indicated above each panel. D, M, S and T are the balloon launch locations in the villages of Dallowere, Markabougou, Siguima and Thierola, respectively. Scale bars, 100 km. The seasonal-wind rose diagrams, reflecting wind conditions at 180 m above ground level averaged from 2013 to 2015, are shown at the right.

Full size image
Table 2 Summary of displacement distance (in a straight line) and source direction, based on 2-h and 9-h flight trajectories of mosquitoes
Full size table

Contrary to the conventional view that dispersal of African anopheline mosquitoes occurs over distances12,14,15,25 of less than 5 km, our results provide compelling evidence that primary and secondary malaria vectors regularly engage in high-altitude flights (or windborne) migrations that span tens to hundreds of kilometres per night. Because this migration includes large numbers of females that had taken at least one blood meal, it probably involves human Plasmodium (among other pathogens). Separate outbreaks of malaria in Egypt and Israel have previously been attributed16 to A. pharoensis travelling more than 280 km. Assuming a conservative22,26 1% infection rate in migrating females of A. coluzzii, A. gambiae, A. coustani and A. pharoensis and 0.1% in the remaining anopheline mosquitoes (excluding the unknown Anopheles Mali species 1 and Anopheles Mali species 2) (Supplementary Discussion), more than 286,000 infected migrant mosquitoes are expected to annually cross, at altitude, a 100-km line perpendicular to the prevailing wind direction. Accordingly, A. pharoensis, A. coustani and A. coluzzii would contribute 41%, 25% and 17%, respectively, to malaria transmission by infected windborne mosquitoes. Although these estimates are coarse, this suggests that migratory secondary vectors could be a major source of infections, and that they should be included in studies of transmission and in control programs.

A. coluzzii was more common than A. gambiae among the migrants, which was contrary to initial predictions3 that were based on data suggesting that A. coluzzii aestivates locally and therefore may not require migration to recolonize the Sahel. Indeed, migration occurs from the end of July to October—well after the surge of A. coluzzii populations in the Sahel following the first rain (May to June)3,6. Northward and southward oscillations of the intertropical convergence zone during the wet season continually reconfigure the better resource patches for mosquitoes, as the intensity of the rains shifts in location. Additionally, wet-season droughts endanger local mosquito populations every decade or two27. Thus, selection pressures to track freshwater resources by riding the winds that bring rain28 may explain why residents of the Sahel, such as Oedaleus senegalensis grasshoppers and A. coluzzii, have a mixed strategy of migration29 and local dormancy. A. gambiae, which presumably recolonizes the Sahel every wet season, is relatively rare in villages in the Sahel3—therefore, only one specimen of this species was captured by our nets. It may migrate on fewer nights and constitute a smaller fraction of windborne migrants (Supplementary Discussion).

In areas in which malaria is approaching elimination, cases of the disease that occur without travel history are presumed to represent indigenous transmission. We propose that a substantial fraction of such cases, especially those that occur within about 300 km of areas with high rates of malaria transmission, arise from the bites of exogenous, windborne infected mosquitoes. For example, northeastern South Africa has the highest incidence of persistent malaria in the country, with many cases not associated with human travel: these cases are concentrated in an arc that extends over about 150 km from the borders with Zimbabwe and Mozambique, where transmission is high. This area also includes the Kruger National Park, in which roads are scarce and vehicular transport of infected mosquitoes30 may be hampered. Testing the correlation of infection events such as these with the corresponding prevailing wind direction will help to assess this hypothesis. If confirmed, the incorporation of disease-control efforts in source populations to minimize or block migration is likely to be an essential element in strategies to eliminate malaria.

Methods

No statistical methods were used to predetermine sample size. The experiments were not randomized and investigators were not blinded to allocation during experiments and outcome assessment.

Study area

Aerial sampling stations were located in four villages in the Sahel in Mali (Extended Data Fig. 1): Thierola (13° 39′ 30.96′′ N, 7° 12′ 52.92′′ W) from March 2013 to November 2015, Siguima (14° 10′ 3.36′′ N, 7° 13′ 40.44′′ W) from March 2013 to October 2015; Markabougou (13° 54′ 51.84′′ N, 6° 20′ 37.68′′ W) from June 2013 to April 2015; and Dallowere (13° 36′ 56.88′′ N, 7° 2′ 12.84′′ W) from July 2015 to November 2015. This study area has previously been described in detail3,6,9,11,31,32,33. In brief, the region is rural, characterized by scattered villages with traditional mud-brick houses that are surrounded by fields. A single growing season (June to October) enables the farming of millet, sorghum, maize and peanuts, as well as subsistence vegetable gardens. Over 90% of the annual rains fall during this season (about 550 mm). Cattle, sheep and goats graze in the savannah that consists of grasses, shrubs and scattered trees. The rains form small puddles and larger seasonal ponds that usually are totally dry by the end of November. From November until May, rainfall is absent or negligible (total precipitation of less than 50 mm), and by December water is available only in deep wells.

Aerial sampling and specimen processing

Aerial sampling stations were placed about 0.5 km from the nearest house of the village in open areas away from large trees. The method of aerial collection of insects was adapted from a study on high-altitude mating flights in ants34. Rectangular 3 × 1-m nets (3 m2), cut from a roll of tulle netting (mesh of 8 holes per square centimetre, with hole diameter of 1.2 mm), were sewn to form 4 narrow sleeves, 1 m apart, along the net (Extended Data Fig. 3). A 1-m carbon rod was inserted into each sleeve and glued to the net using Duco Cement Glue (Devcon) (Extended Data Fig. 3). Three nets were spread over each other on a clean large wooden table topped by 3.5 × 1.5-m plywood and coated with a thin film of insect glue (Tanglefoot, Tropical Formula, Contech Enterprises) by rolling a PVC pipe smeared with this glue over them, while applying moderate pressure downward. The pipe was held at each end (from each side of the long table) by two persons and repeatedly rolled (and smeared) until a uniform thin layer of glue coated the net but did not block its holes. After coating, the sticky nets were immediately rolled individually, and kept in two tightly secured plastic bags indoors, to avoid accidental contact with insects before setup.

Before the launch, polyurethane balloons (3 m in diameter; from Mobile Airship & Blimps, or Lighter than Air) were inflated to full capacity with balloon-grade helium (>98.5%) and topped up to ensure full capacity as needed, usually every 1–3 days based on the balloon condition (Extended Data Fig. 3). Typically, balloons were launched over about ten consecutive nights per month. The balloon was kept stationary at about 200 m above ground level by a cord (AmSteelBlue, synthetic rope sling, Southwest Ocean Services) secured to a 1-m3 cement block inserted under the ground. The cord then passed through a horizontal, manually rotating drum made of a garden-hose reel that was used for reeling it in. A larger 3.3-m diameter balloon (Lighter than Air) was used between July and September 2015, and launched to about 300 m above ground level.

A team of five trained technicians operated each aerial-sampling station. During the launch of a balloon, one team member held the cord under the balloon with heavy-duty gloves and manually controlled its ascent and descent, another team member controlled the reel, and the other three team members added or removed the sticky nets to and from their specified positions on the cord. The nets were attached to Velcro panels that had previously been placed on the cord at desirable positions, and spaced to fit each of the matching Velcro pieces on the four carbon rods (Extended Data Fig. 3). A knot was made below the top-most Velcro panel and above the bottom-most Velcro panel, to ensure that the nets would remain stretched (rather than slip on the cord) even in strong winds. Additionally, the team secured the balloons over a ‘landing patch’ that was padded by tyres covered by a tarpaulin. The balloon was secured to the ground through its main cord by a central hook (at the middle of the landing patch), and by a large tarpaulin that covered it from the top, which was secured to the ground using 14 large stakes. Team members inspected the nets upon launch to verify that they were free of insects. Upon retrieval of the balloon, the team worked in reverse order and immediately rolled each sticky net and placed it into a clean labelled plastic bag, and inserted this into another bag; each of the bags was tightened with a cord until inspection.

Each balloon typically carried three sticky nets. Initially, the nets were suspended at 40, 120 and 160 m above ground level. From August 2013, the typical altitude was set to 90, 120 and 190 m above ground level. When the larger balloon was deployed in the Thierola station (August–September 2015), two additional nets were added at 240 and 290 m above ground level. Balloons were launched approximately 1 h before sunset (about 17:00) and retrieved 1 h after sunrise (about 07:30) the following morning. To control for insects trapped near the ground as the nets were raised and lowered, control nets were raised up to 40 m above ground level and immediately retrieved during the launch and retrieval operations; between September and November 2014, the control nets were raised to 120 m above ground level. The control nets spent 5 min in the air, or up to 10 min when raised to 120 m. Once retrieved, the control nets were processed in a manner identical to that of other nets. After retrieving the panels, inspection for insects was conducted between 09:00 and 11:30 in a dedicated clean area. The panel was stretched between two posts and scanned for mosquitoes, which were counted, removed using forceps and preserved in 80% ethanol before all other insects were similarly processed and placed in other tubes. Depending on their condition, the sticky panels were sometimes reused the subsequent night.

Species identification

Glue attached to the insects was washed off with 100% chloroform. The mosquitoes were gently agitated (for less than 30 s) to loosen them from one another. Individual mosquitoes were transferred into consecutive wells filled with 85% ethanol. Using a dissecting scope, the samples were morphologically sorted by mosquito subfamily (Anophelinae or Culicinae), and were tentatively identified to the Anopheles species or species-group level. All mosquitoes that were morphologically confirmed as A. gambiae sensu lato (and two insects that were identified on the basis of molecular barcode analysis) were identified to the species level on the basis of fragment-size differentiation after amplification of the nuclear ITS2 region and digestion of the product35. Validation was carried out in LSTM (laboratory of D.W.), in which each specimen was washed with 500 μl heptane followed by two further washes with ethanol. DNA was then extracted using the Nexttec (Biotechnologie) DNA isolation kit, according to the manufacturer’s instructions. Species identification used a standard PCR method, including all primers36, with products visualized on 2% agarose gel. Samples of A. gambiae sensu lato were further identified to species by short interspersed element insertion polymorphism37. In cases in which no species-specific bands were detected using the first method, an approximately 800-bp region of the mtDNA cytochrome oxidase I (COI) genes was amplified using the primers C1_J_2183 and TL2_N_301438. PCR products were purified using the QIAquick PCR-Purification kit (Qiagen) and sequenced in both directions using the original PCR primers by MacroGen. Sequences were aligned using CodonCode Aligner (CodonCode) and compared to existing sequences in GenBank to identify species. All other Anopheles mosquitoes were identified by the retrospective correlation of DNA barcodes, with morphologically verified reference barcodes compiled by Walter Reed Biosystematics Unit and the Mosquito Barcoding Initiative in the laboratory of Y.-M.L. Head and thorax portions of all samples were separated from abdomens and used for DNA extraction using the Autogen automated DNA extraction protocol. mtDNA COI barcodes were amplified using the universal LCO1490 and HCO2198 barcoding primers39, and amplified, cleaned and bi-directionally sequenced according to previously detailed conditions40. All DNA barcodes generated from this study are available under the project ‘MALAN – windborne Anopheles migrants in Mali’ on the Barcode of Life Database (www.boldsystems.org) and in GenBank under accession numbers MK585944–MK586043. Plasmodium infection status was tested for all available Anopheles on DNA extracts from the head and thorax portions (n = 190) and also from abdomens (n = 156) following previously described protocols41,42,43. Owing to the nature of the collections, all body parts were not available for each specimen, which accounts for any discrepancies in the numbers. Blood-meal identification was carried out following a published protocol44.

Data analysis

Although aerial collections started in April 2012, protocol optimization and standardization took most of that year; the data included in the present analysis cover only the period March 2013–November 2015. Nights on which operations were interrupted by storms or strong winds (for example, the balloon was retrieved during darkness) were also excluded.

The total number of mosquitoes per panel represents ‘panel density’ of each species. Aerial density was estimated on the basis of the panel density of the species, and total air volume that passed through that net that night: that is, aerial density = panel density/volume of air sampled, and volume of air sampled = panel surface area × mean nightly wind speed × sampling duration. The panel surface area was 3 m2. Wind-speed data were obtained from the atmospheric reanalyses of the global climate (ERA5). Hourly data were available at 31-km surface resolution, with multiple vertical levels that included ground, 2, 10, 32, 55, 85, 115, 180, 215, 255 and 300 m above ground level. Overnight records (18:00 through to 06:00) for the nearest grid centre were used to calculate the nightly direction and mean wind speed at each village (Siguima, Markabougou and Thierola). Dallowere, which is located 25 km south of Thierola, was included in the same grid cell as that of Thierola. The mean nightly wind speed at panel height was estimated on the basis of the nearest available altitude layer.

To evaluate clustering in mosquito panel density and the effects of season, panel height, year and locality, mixed linear models with either Poisson or negative-binomial error distributions were implemented by procGLIMMIX45. The clustering at the levels of the panel and night of sampling were evaluated as random effects, as was the case for the year of sampling and locality. These models accommodate counts as non-negative integer values. The ratio of the Pearson χ2 to the degrees of freedom was used to assess the overall ‘goodness of fit’ of the model, with values of more than two indicating a poor fit. The significance of the scale parameter that estimates k of the negative-binomial distribution was used to choose between Poisson and negative-binomial models. Sequential model fitting was used, starting with random factors before adding fixed effects. Lower Bayesian information criterion values, and the significance of the underlying factors, were also used to select the best-fitting model for each species.

The magnitude of migration of windborne migrants was expressed as the expected minimum number of migrants per species crossing an imaginary line of 1 km perpendicular to the wind direction at altitude. This commonly used measure of abundance assumes that the insects fly in a layer that is 1-km wide, and does not require knowledge of the distance or time the insects fly to or from the interception point46,47,48. We used the mean wind speed at altitude (4.8 m s−1; Fig. 1c), and assumed that mosquitoes fly in a layer depth of 200 m between 50 and 250 m above ground level, which conservatively reflects the fact that mosquitoes were captured between 40 and 290 m above ground level (Fig. 1a and Extended Data Fig. 2b). Accordingly, this nightly migration intensity was computed as the product of the mean aerial density across the year (conservatively including periods during which no migrants were captured) by the volume of air passing over the reference line during the night. The corresponding annual index was estimated by multiplying the nightly index by the period of migration of windborne migrants, estimated from the difference between the first and last day and month that a species was captured over the three years. Species that were captured once were assumed to migrate during a single month. The annual number of migrants per species that cross a 100-km line was used because of the similar composition of species across our sampling sites, which spanned 100 km (Extended Data Fig. 1a).

Similar to most insects in their size range47,49,50, the flight speed of mosquitoes51,52 does not typically exceed 1 m s−1. Because winds at panel altitude attain speeds considerably higher than the speed of the mosquito, flight direction and speed are governed by the wind46,47 and flight trajectories can therefore be simulated on the basis of the prevailing winds during the night of capture at the relevant locations and altitudes, as has previously been done53,54,55. Accordingly, backward trajectories of mosquito flights were simulated using HYSPLIT25 on the basis of ERA5 meteorological reanalysis data. Data that are available in ERA5 present the highest spatial and temporal resolution available for this region. Comparisons with data of lower spatial and temporal resolution that are available from the MERRA2 reanalysis data56, and with the Global Data Assimilation System (available at 0.5° spatial resolution), showed good agreement in trajectory direction and overall distance (data not shown). Trajectories of each captured mosquito were simulated starting at its capture location and altitude, and all multiple interception (full) hours during the night of the collection. Because anopheline mosquitoes are nocturnal, we conservatively assumed that flights started at or after 18:00 and ended by 06:00 the following morning, and computed trajectories for every hour that allowed for a total of a 2-h or 9-h flight. For example, to complete a 9-h flight by 06:00, a mosquito could have started at 18:00, 19:00, 20:00 or 21:00. The total flight duration of tethered female A. gambiae sensu lato and Anopheles atroparvus reached or exceeded 10 h, with an average speed51 of 1 km h−1 that is consistent with other studies52,57,58. Similarly, Anopheles vagus and Anopheles hyrcanus, caught 150 m above ground level after midnight over India, would have been migrating for more than 6 h, assuming they took off around dusk20. Thus, we conservatively assumed that anopheline mosquitoes at high altitude fly between 2 and 9 h per night, although longer durations are possible. Each trajectory consisted of the global positions of the mosquitoes at hourly intervals from the interception time. In addition to plotting trajectories59,60,61,62,63,64,65,66, the linear distance from the interception site and the simulated position of the mosquito, and the azimuth (angle between the interception site and the simulated position of the mosquito from the north, projected on a plane) were computed for all trajectories. To evaluate the distance range and dominant directions of flight, the mean and 95% confidence interval of the distance and azimuth (as a circular statistic) were computed for the 2-h and 9-h flight trajectories. The dispersion of individual angles (azimuths) around the mean was measured by the mean circular resultant length r, which can vary from 0 to 1, with higher values indicating tighter clustering around the mean. Rayleigh’s test was used to test whether there was a mean direction; no mean direction occurs when the angles form a uniform distribution over a circle67.

Reporting summary

Further information on research design is available in the Nature Research Reporting Summary linked to this paper.