Aerial collection stations were established in rural open areas in Ghana and Mali (Fig. 1A). In southern Ghana two stations were set up in the moist-semi-deciduous forest near the town of Agogo (6.961, -0.960), and in the Guinea woodland ecozone near the town of Wenchi (7.781, -2.162). In southern Mali one station was set in the Sudano-Guinean ecozone near the village Kenieroba (12.112, -8.332). Agogo, Wenchi, and Kenieroba receive 1200–1400, 1000–1200, and 900–1000 mm of rain annually, respectively. However, in Kenieroba, the rainy season is confined to May-November whereas the Ghanian sites receive rains most months of the year (Siebert 2014). These areas are interspersed with farmland along rivers and grassland. The main crops grown near Agogo are cocoa, coffee, oil palm, citrus, cashew, mango, cassava, yam, among others. The main crops in Wenchi are corn, rice, sorghum, yam, plantain, groundnuts, and cowpea, among others. In Kenieroba, farmers grow mainly rice, sorghum, millet, corn, beans, among other crops; cereals are grown during the wet season (May-October) and vegetables in smaller plots during dry season (November-April).

Figure 1.

Study area showing collection sites with wind patterns (during inferred migration season) and panel and aerial densities of C. insularis A map showing sampling sites in Ghana and Mali (balloon symbols) with the port of Abidjan, Ivory Coast (blue anchor). The arrow shows the predominant wind direction B nightly wind at altitude (300 m agl) by month during 2020-2021 at each sampling station (color) showing the direction (arrow) the insects will be carried towards, from the aerial sampling site (origin). The wind speed is indicated by the vector length (source: ERA5) C panel density of C. insularis at the different sampling stations (regardless of altitude) with corresponding color fringe (bottom) indicates sampling night used in this study. Blue shade indicates period of interception of C. insularis D aerial densities of C. insularis at Agogo and Wenchi at altitude (based on panels at 160-190 m agl). Fringe (bottom) indicates sampling dates at altitude in Ghana.

The aerial collection methods were described in detail previously (Huestis et al. 2019; Florio et al. 2020), Briefly, insect sampling was conducted using sticky nets (panels, each 1 m wide × 3 m long) attached to the tethering line of 3 m diameter helium-filled balloons. Each balloon typically carried three panels, set at 120 m, 160 m, and 190 m agl (above ground level). Balloons were launched around 17:30 before sunset and retrieved around 07:00, after sunrise the following morning. To control for insects trapped near the ground as the panels were raised and lowered, comparable control panels were raised up to 100 m agl and immediately retrieved during each balloon launch and retrieval operation. Following panel retrieval, inspection for insects was conducted in a dedicated clean area. Individual insects were removed from the nets with forceps, counted, and stored in vials containing 80% ethanol. Concurrent samples were collected from a 1 m² panel mounted 1 m above ground to represent ground insects. This net was attached to a line from a frame in such a way that allowed it to orient perpendicular to the wind direction like the nets suspended from the helium balloon.

Using a dissecting microscope, African specimens were identified morphologically to order and to morphospecies, counted, and recorded. Specimens of the selected morphotypes were identified by expert taxonomist who narrowed the identification down to species or genus. All African specimens of C. insularis were confirmed by Dr. James Zahniser (USDA-APHIS, National Museum of Natural History [USNM], Smithsonian Institution, Washington, DC, USA). Voucher specimens are deposited at USNM.

Publicly available records of observations of C. insularis from BugGuide.net (n = 22; (BugGuide 2024)), iNaturalist (n = 827; (iNaturalist 2024)) that met “Research grade” standard (verified by SVH) were downloaded and added to records available from the Florida Department of Agriculture and Consumer Services, Division of Plant Industry (FDACS-DPI) (N = 266) after the specimens were identified by FDACS-DPI hemipterists. Vouchers for FDACS-DPI specimens are deposited in the Florida State Collection of Arthropods. The data from Florida also included 3 interception records in trucks that originated from Arizona (2) and Mexico (1). Border interception records of C. insularis (since 2019) from ports of entry in the USA (identified by Dr. James Zahniser, USDA) were provided from USDA/APHIS (N = 9) included the port of entry and date of collection. Of 1,124 observations collected until December 31, 2023, from the USA, 11 represented insects during transit, and thus were excluded from distribution data that were used to plot C. insularis range in the USA. This compilation of data is included in Suppl. material 1.

Subsamples of the Mali collections from March to December 2019 and the Ghana collections from May to October 2021 were evaluated for the presence of C. insularis. Every month of collection at each study site was represented by at least 6 panels at altitude and at least 4 panels at ground level. The total number of insects per panel represents the ‘panel density’. Aerial density was estimated as the panel density of the species divided by the total air volume that passed through that net that night (i.e., 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 re-analyses of the global climate (ERA5). Hourly data consistent of the eastward and northward components (horizontal vectors) of the wind were available at 31-km surface resolution at 2 and 300 m agl (1000 and 975 mbar pressure levels). Overnight records (19:00 through to 06:00) were averaged to calculate the nightly mean direction and mean wind speed over each African sampling station and select locations in the USA (below) based on standard formulae using code written in SAS (SAS software 2019).

The intensity of migration was expressed as the expected number of migrants crossing a line of 1 km perpendicular to the wind direction at altitude, which reflect their direction of movement (Drake and Reynolds 2012; Hu et al. 2016; Reynolds et al. 2017; Florio et al. 2020). We used the mean wind speed at altitude during the migration season (Table 1) and assumed that the leafhoppers fly in a layer depth of 200 m above ground level (Florio et al. 2020). The nightly migration intensity was computed across the flight season (including sampling nights during which no migrants were captured). The corresponding annual index was estimated by multiplying the nightly index by the period of migration, estimated from the first and last month that the species was captured.

To assess the likelihood of windborne movement to a new locations in the USA, we identified new sites where C. insularis was observed for the first time outside its previous year’s range, defined by connecting all the extreme points of its cumulative distribution in the previous year. For each new site, we consider its nearest known site—where C. insularis was previously reported—as a putative source. Underlying our approach is the assumption that the missing data due to low sampling in certain localities and/or certain years would generate noisier data-patterns rather than as systematic pattern. Therefore, finding a consistent biological trend (signal) in these data, relevant to the process in question is likely produced by a biological process rather than by variation in sampling intensity. Virtually all datasets on geographic expansion at these scales of time and space would present similar “imperfections”, inviting inquiries to better assess and address their limitations. However, large Citizen-Scientist databases provide compelling advantages as pointed out by (Kittelberger et al. 2021).

The annual distributions of nightly (19:00–06:00) winds during the year of the new record were plotted as vectors pointing to the direction the insects would be carried if they flew 8 hours from that site on that night’s wind at 300 m agl. The self-propelled flight speed of leafhoppers does not typically exceed 1 m s⁻¹ (Zhou et al. 2003) and therefore at altitude with winds exceeding 4 m/s, the direction of the movement of the insects will be determined primarily by the wind. Only nights with at least 8 hours when temperatures were > 16 °C at altitude and on the ground were included (assuming no flight occurred below 16 °C (Shields and Testa 1999). The total flight duration of tethered leafhoppers reached over 7 h (Zhou et al. 2003) and additionally, inferences about windborne leafhoppers in the US traveling hundreds of kilometers per night (or tens of kilometers/night over successive nights) have been made based on their phenology (Carlson et al. 1992; Shields and Testa 1999; Reynolds et al. 2017). Thus, we assumed that C. insularis could fly at altitude between 2 and 8 h per night. The wind trajectories were projected on maps based on the wind direction, speed, and 8 hours of flight using SAS 9.4 (SAS software 2019). Windborne dispersal from the putative source to its destinations was assessed based on visual inspection of the trajectory distributions.

Overall, 25,431 insect specimens collected in West Africa on 308 panels (157 panels at 120–290 m agl, 84 panels at 1 m agl, and 67 control panels) were sorted and evaluated for the presence of Curtara insularis. Interception of C. insularis at altitude and at ground level occurred between May and August (Fig. 1) but not in September and October (sampling in other months were only performed in Mali). Although we cannot rule out high-altitude flights in the rest of the year, we consider the period between May and August as its migration period and unless otherwise stated, confined the summary statistics to this period (Table 1, Fig. 1). A total of 18 specimens of C. insularis were identified among samples collected in Agogo and Wenchi (Ghana) during May to August (9,113 insect specimens, Table 1), but none was found in Kenieroba (Mali) during these months (3,897 insect specimens, Table 1) or throughout the year (6,151 insect specimens). Most C. insularis (N = 14) were intercepted at altitude (160–190 m agl), 3 were collected at ground level, and 1 on a control panel (Table 1). A single specimen on the control panels (N = 45 control panels) as opposed to 14 specimens intercepted in these locations at altitude (N = 61 panels, Table 1) indicate that “contamination” near the ground (<50 m agl) is unlikely to account for the capture of so many specimens at altitude. At altitude, similar numbers of males (8) and females (6) were collected, indicating that both sexes equally engage in high-altitude flight unlike mosquitoes, in which females consist of > 80% from the aerial collection (Huestis et al. 2019; Yaro et al. 2022). All specimens were macropterous (having fully formed fore- and hindwings), albeit no polymorphism in wing development has been noted for this species.

The higher numbers of C. insularis at altitude and its relative larger fraction of the total insects on the panels among all specimens collected (Table 1) attest for its propensity to engage in high-altitude windborne dispersal. The scale of C. insularis movement at altitude was estimated as the average number of individuals flying between 50 and 250 m agl crossing a 1 km line perpendicular to the wind during a single night. With tens of thousands flying nightly at altitude across 1 km (Table 1), the expected number over the four-months season (Fig. 1) was in the millions, reflecting that this species had established robust populations in its source sites. Radar studies have shown that windborne insects fly nightly over swathes that are tens or even several hundreds of kilometers wide (Drake and Reynolds 2012; Hu et al. 2016). With a mean wind speed of 4.5 m/s, individual insects flying 2–8 hours would readily cover 30–150 km per night. High-altitude windborne dispersal has been reported in leafhoppers including a flight > 3,000 km over the ocean (Balclutha salturella [Previously: B. pauxilla]), culminating in the invasion of Ascension Island (Ghauri 1983), and seasonal migrations over hundreds of kilometers in North America (Empoasca fabae, Macrosteles quadrilineatus, and Circulifer tenellus), and Asia (Nilparvata lugens and Sogatella furcifera) (Glick 1939; Taylor 1974; Carlson et al. 1992; Reynolds et al. 2017; Hu et al. 2019; Wu et al. 2019). In tropical West-Africa, temperature and humidity rarely restrict flight activity even at altitude (Table 1; Sanogo et al. 2021). Altogether, high-altitude dispersal of C. insularis is a potent strategy for spreading rapidly over large areas. These large number of migrants account for massive propagule pressure (Lockwood et al. 2005; Simberloff 2009) that could well explain the importance of this modality of dispersal compared with vehicular spread, which involves small number of specimens that often arrive in inhospitable spaces, e.g., storage facilities surrounded by urban environment, which limit reproduction success and establishment of populations.

Given the predominant wind directions in this region of West Africa (Fig. 1, Table 1), C. insularis is probably carried by the winds inland, encroaching up to at least ~300 km (Wenchi) from its supposed landing sites in one or more West Africa ports (Fig. 1). If it had been introduced through the largest port of West Africa, Abidjan, Ivory Coast, the predominant southwesterly winds would have carried it straight to the Agogo and Wenchi areas (Fig. 1), whereas Kenieroba is clearly off the predominant wind trajectories from the main Gulf of Guinea ports. Thus, invasion of locations such as Kenieroba may require over-land vehicular transport or considerably longer time using windborne dispersal. Whether C. insularis can establish populations in Kenieroba, Mali is an open question. Located within the Sudano-Guinean zone along the Niger river, the overall climatic conditions in Kenieroba are similar to those in Wenchi (Table 1) or other sites where C. insularis is found (below), albeit with a longer dry season (December-April). Other explanations in which C. insularis has arrived earlier at other region(s) of Africa cannot be rule out without additional data on its distribution on the continent.

The current distribution of C. insularis in the USA is based on 1,109 records (excluding 11 “in-transit” records) spanning the period of 2004–2023 (Suppl. material 1, 2: fig. S2). The records cover much of the southeastern and south-central USA (25^°—36^°N latitudes and 77^°— 102^°W longitude, Suppl. material 2: fig. S2). There was minimal seasonal variation considering the month of observation and latitude or longitude (not shown), suggesting that C. insularis populations are perennial in that range. Given the number of the records, this range suggests a habitat-suitability space, which appears bounded to the north by annual temperature minima isotherm of -12 °C, described by vegetation zone 8 (USDA 2023) and to the west by annual precipitation isohyet of 510–630 mm (NOAA 2023). Despite their apparent correlation between these boundaries and the current limits on the C. insularis distribution, studies are needed to establish causal relationships of these hypothetical factors. Permissive combinations of temperatures and precipitation also occur along the Northwest coast (e.g., in California) and the Southwest (e.g., in Arizona). Indeed, C. insularis was intercepted in Florida on two trucks that originated from Arizona in 2022 and 2024 (FDACS-DPI records), although the species has not yet been reported from any western state. Future records of this species from those territories may reveal the stability of the current range. We consider the range of records available up to December 2023 as the space in which range expansion could have occurred, evaluating how C. insularis “filled” it.

The interception of C. insularis in international ports (9 records since 2019) and on trucks entering Florida (3 records since 2004) provide evidence for the role of the maritime trade as well as vehicles overland in transporting this species. The port interceptions were in Florida (5), Houston Texas (2), Georgia (1), and Puerto Rico (1). These records substantiate that C. insularis can spread by all these means as well as by wind at altitude (above) as other invasive pests, such as Helicoverpa and Spodoptera moths (Jones et al. 2019). Given the large distances of transport by truck (daily average of ~1,000 km), airplane, or ships, the expected pattern of spread would be by long and irregular leaps in all directions (Suarez et al. 2001; Ahn et al. 2023; Lehmann et al. 2023), whereas windborne spread would predict a continuous incremental spread that resembles “diffusion process” that follows predominant winds.

The first record of C. insularis in the USA dates to January 2004 (Hillsborough County, Florida, Suppl. material 1). However, 42 specimens from 5 central-south Florida counties were identified that year, indicating the original introduction took place into that region earlier (Fig. 2A, Suppl. material 2: fig. S3). Over the next 5 years, it has spread throughout the state, reaching the panhandle by 2008, and expanding outside the state only in 2012 (Suppl. material 2: fig. S3). The relatively slow spread northward, given the numerous sightings in Florida and compared with subsequent years is incompatible with vehicular transport but can be attributed to “unfavorable” predominant winds which have a strong east-west component in relation to the narrow peninsula (~200 km wide), resulting in most windborne journeys ending at sea (Fig. 2A). That Florida is a major producer of most nursery stocks and diverse produce that are shipped by trucks efficiently all over the US, highlights the contrast between this mode of transport and the pattern of spread observed. In 2012, C. insularis was reported near Houston, Texas, and a year later near Lafayette, Louisiana (Suppl. material 2: fig. S3). With average nightly wind speeds 5.6 m/s (maximum of 16.9 m/s) we expect displacement ranges of 40–150 km/night over 2–8 hours of flight with up to 500 km during maximum recorded (nightly) windspeed. The closest C. insularis record was from Florida, > 900 km away from Houston (500 km away from Lafayette). The scarcity of westward winds from Florida and the large distance suggests that C. insularis was transported to Houston by a ship, truck, or airplane rather than by the wind (Fig. 2B, and below). Nonetheless, its arrival near Lafayette (Louisiana, Suppl. material 2: fig. S3) the following year is consistent with frequent eastward winds from Houston (Fig. 2B). In 2016, C. insularis was observed around Austin and Dallas (Texas) as well as near Columbus, Georgia (Suppl. material 2: fig. S3). Wind trajectories from Houston to Dallas were common and those towards Austin are modestly common, as are the trajectories from Tallassee to Columbus (Fig. 2B). The slower arrival into these destinations (~4 years) agrees with the modest frequency of corresponding wind trajectories. By 2017 and 2018, C. insularis was reported from South Carolina and in new counties of Texas, including a remote region of western Texas > 50 km from a major highway (Suppl. material 2: fig. S3). Fifteen of the new records were located consistent with wind trajectories from previously established populations (Fig. 2C), however 4 of 20 would require 9–11 hours windborne flight (at average nightly wind speed) and a single site near Matamoros, Mexico, would require ~16 hours flight or an additional landing by ship at the nearby port of Brownsville Texas (Fig. 2C, Suppl. material 2: fig. S3). In 2019, C. insularis was reported from its northern-most point near Raleigh, North Carolina (~170 km from Florence, SC, Suppl. material 2: fig. S3), Tuscaloosa, Alabama (~260 km from Crestview FL), and Jackson, Mississippi (~130 km from Pensacola FL), as well as to other new localities, closer to previously established populations (Fig. 2D, Suppl. material 2: fig. S3). All these sites were consistent with common wind directions and distances from previously established populations (Fig. 2D, Suppl. material 2: fig. S3). Similarly, new counties in 2020 and 2021–2023 were also < 100 km from previously established sites and often from multiple such sites (Fig. 2D, Suppl. material 2: fig. S3), consistent with windborne spread. This approach is conservative because it assumes that the source populations of news localities are known, although unknown source sites may be closer to the new location, resulting in overestimating the distance these migrants actually passed. The average annual maximum distance of range expansion was 229 km (range: 0–950 km, Suppl. material 2: fig. S4). The arrival into Houston (recorded on 2012) presented an extreme outlier, indicating a different mode of range expansion from the rest (Fig. 2, Suppl. material 2: figs S3, S4). Excluding this value, yields an average of annual maximum range expansion of 176 km (Suppl. material 2: fig. S4: inset). Both values appear considerably lower than the average maximum distances of truck transporting produce in the US.

Figure 2.

Maps showing range expansion of Curtara insularis in relation to projection of annual wind trajectories at altitude (300 m asl) from putative source(s) reported previously assuming 8-hour windborne flight and linear winds (broken lines) A expansion of C. insularis (2004—2011) with annual nightly projections of wind trajectories from Tampa 2005 (see text) B expansion of C. insularis (2012—2016) with annual nightly wind trajectories from Tallahassee and Houston 2012 (see text) C expansion of C. insularis (2017—2018) with projections of annual nightly wind trajectories from Jacksonville, Tallahassee, Columbus, Lafayette, Houston, Dallas, and Austin. Expansion of C. insularis (2018—2023) with projections of annual nightly wind trajectories from Florence, Columbia, Columbus, Crestview, Lafayette, Dallas, Austin, Matamoros (Mexico).

The authors have declared that no competing interests exist.

No ethical statement was reported.

This study was supported by the Division of Intramural Research, National Institute of Allergy and Infectious Diseases, National Institutes of Health, Bethesda MD (ZIA AI001196-06), the Bill & Melinda Gates Foundation, Grand Challenges Explorations grant (OPP1217659; awarded to TL), the Florida Department of Agriculture and Consumer Services, the United Kingdom Biotechnology and Biological Sciences Research Council to Rothamsted Research, Division of Plant Industry, and the U.S. Department of Agriculture, Agricultural Research Service.

The study was conceived by TL and RN and the study design was shaped with inputs from JWC, DR, SEH, SH, YA, MB, LC, CB, RB, and KOD. Field work, data management, and initial specimen processing were carried out by AD, ASY, YA, ARM, CMO-A, GA-B, CMP-Z, MD, ZLS, and SD. Laboratory analysis was done by RN, and LC, data compilation from different sources were carried out by SEH, SH, CEN, MB, TL and RN. Data analysis was carried out by TL with inputs from MB, LC, RN, RB, DRR, JWC, SEH. The first draft was written by TL and RN and all authors commented on previous versions of the manuscript. All authors read and approved the final manuscript.

Rita Nartey https://orcid.org/0009-0003-3846-4927

Lourdes Chamorro https://orcid.org/0000-0002-3536-9372

Matt Buffington https://orcid.org/0000-0003-1900-3861

Abdul Rahim Mohammed https://orcid.org/0000-0002-4514-6914

Susan E. Halbert https://orcid.org/0000-0003-4341-5196

Charles R. Bartlett https://orcid.org/0000-0001-9428-7337

Don R. Reynolds https://orcid.org/0000-0001-8749-7491

Tovi Lehmann https://orcid.org/0000-0002-8142-3915

All of the data that support the findings of this study are available in the main text or Supplementary Information.