A trans-oceanic flight of over 4,200 km by painted lady butterflies

The study addresses whether small insects can complete trans-oceanic long-distance dispersal and how frequently such events might occur. Despite many anecdotal observations of insects over open ocean and on remote coasts, robust tracking over multi-day, continent-scale distances is limited by technology and the small body size of most insects. The authors report and investigate a flock of Vanessa cardui (painted lady) butterflies observed arriving on a French Guiana beach on 28 October 2013, outside the species’ typical South American range. They aim to reconstruct the butterflies’ dispersal route and natal origins, assess the role of winds (notably trade winds and the Saharan Air Layer) in enabling the crossing, and evaluate the energetic feasibility of such a journey. The work’s significance lies in clarifying the potential for wind-assisted, multi-day, transoceanic dispersal by an iconic migratory butterfly and its implications for insect biogeography and ecosystem connectivity.

The paper synthesizes prior literature documenting insect presence far offshore and on remote islands, as well as early and modern attempts to track insect migration (ship/platform observations, aerial netting, radar, and miniaturized VHF tags). It notes that technological methods typically capture short-distance or short-duration movements and are biased toward larger insects. The authors reference known long-distance insect migrants (e.g., Pantala flavescens across the Indian Ocean; monarch butterfly migration between Canada and Mexico) and historical instances of African desert locusts reaching the Caribbean, arguing that transoceanic windborne dispersal has likely been underappreciated. They discuss the Saharan Air Layer and easterly trade winds as potential, recurrent aerial highways with documented impacts on trans-Atlantic dust transport and biogeochemical cycles, suggesting they may similarly facilitate insect movement. The distribution and occurrence records of V. cardui in regions like Hawaii and historical observations in Australia are cited to contextualize extreme dispersal events.

The authors used an integrative, multi-source approach: (1) Field sampling: During October 2013 coastal surveys in French Guiana, three live V. cardui individuals from a larger flock were collected at ~6:00 am on 28 October, preserved for molecular and pollen analyses. (2) Wind trajectory modeling: Hourly backward wind trajectories from the capture site were reconstructed over 200 hours (~8.3 days) using NOAA’s HYSPLIT model at altitudes of 500, 1000, and 2000 m AGL for dates spanning 21–31 October. Trajectories were processed and visualized with R packages (splitr/opentrq/sp/raster/geosphere/ggplot2/viridis), and intersections with the African coastline were quantified; wind speeds along trajectories were summarized. (3) Genomics: ddRAD-seq data were generated/compiled for 126 V. cardui from North America (n=40), Europe (n=56), and Africa (n=34), plus the French Guiana samples. Data were processed (trimming, mapping, SNP calling), with population structure assessed via PCA (PLINK) and co-ancestry (fineRADstructure). Filtering steps addressed missingness and linkage disequilibrium. (4) Pollen metabarcoding: Pollen from butterfly bodies was extracted, ITS2 amplicons prepared (triplicate PCRs, indexing), and sequenced on Illumina MiSeq. Reads were merged, adapters/primers trimmed, quality-filtered, clustered/denoised (VSEARCH/UNOISE), and taxonomically classified with SINTAX against curated plant databases (Sickel et al.; PLAINTS). Potential contaminants and erroneous entries were removed. (5) Isotope geolocation: Dual-isotope assignments combined δ2H (wing tissue calibrated to precipitation isoscapes using known-origin butterfly datasets across North America, Europe, and Africa) and 87Sr/86Sr (mapped to global bioavailable Sr isoscapes). δ2H was measured via TC/EA-IRMS after cleaning and equilibration; 87Sr/86Sr was measured by MC-ICP-MS following digestion and chemical preparation, with standards and corrections applied. Individual probability surfaces (single and joint isotopes) were computed in R (assign package), then combined across individuals and constrained by ecological niche models (ENM) of V. cardui breeding suitability for September–October to refine likely natal regions. (6) Energetic flight modeling: The team modeled energy requirements under three strategies: (i) continuous active flight without winds, (ii) continuous active flight with wind assistance (ground speeds derived from HYSPLIT winds; ~4.7–8.8 m/s wind components yielding ~10.7–17.9 m/s ground speeds when flying), and (iii) alternating phases of active flight and minimal-energy flight (assumed 15% active, 85% minimal-effort) with metabolic rates scaled from monarch butterfly data and assumed body mass/fat reserves for V. cardui. Fat consumption was calculated using oxygen consumption-to-calorie and calorie-to-fat conversions; crossing times were estimated given a minimum over-water distance of 4200 km.

- Wind trajectories: For the 48 hours immediately before the observation (26–28 October 2013), backward trajectories were highly consistent across altitudes and indicated trans-Atlantic flow from West Africa; 83% of these trajectories intersected the African coastline, with crossings occurring roughly 6.6 days prior to capture, supporting a recent Africa-to-South America transport window. - Genomics: PCA and fineRADstructure analyses grouped the French Guiana butterflies with the Afro-European population, clearly excluding a North American origin. - Pollen metabarcoding: Among 58–155 identified plant taxa per method, two Sahelian endemics, Guiera senegalensis (especially abundant) and Ziziphus spina-christi, were detected on butterfly bodies; both bloom Aug–Nov in West Africa, indicating recent contact with Sahelian flora near the western African coast. - Isotope geolocation: δ2H and 87Sr/86Sr signals were similar across individuals, indicating a shared natal origin. Dual-isotope probability surfaces, constrained by ENM breeding suitability for Sept–Oct, highlighted Western Europe (Portugal, France, Ireland, UK) and West Africa (Mali; coastal Senegal/Guinea) as the most probable natal regions, with low suitability discounting much of North Africa. - Dispersal distance and duration: The documented trans-Atlantic crossing covered at least 4200 km, consistent with a 5–8 day journey. If individuals developed in Western Europe and moved via West Africa, the total Europe–Africa–South America dispersal could exceed 7000 km. - Energetics: Nonstop active flight (with or without wind assistance) exceeded plausible maximum fat reserves. An alternating strategy with approximately 15% active flight and 85% minimal-effort windborne transport rendered the crossing feasible, requiring fat reserves as low as ~13.7% of body mass and consistent with the 5–8 day timeframe. - Broader inference: The results suggest that wind-assisted, transoceanic insect dispersal may be more common than recognized, facilitated by recurrent easterly trade winds and the Saharan Air Layer.

The multi-line evidence—wind back-trajectories during a favorable trade-wind window, genomic clustering with Afro-European populations, Sahelian pollen on the butterflies, and dual-isotope geolocation constrained by ecological suitability—converges on a scenario in which V. cardui crossed the Atlantic from West Africa to French Guiana over 5–8 days. The energetics analysis indicates that such a crossing is plausible when largely leveraging wind assistance with intermittent active flight. This case demonstrates that persistent atmospheric circulation features, notably the Saharan Air Layer and easterly trades, can serve as predictable aerial highways enabling intercontinental dispersal by small insects. Despite the absence of established V. cardui populations in South America, such events may still occur with some frequency, potentially seeding temporary occurrences. Collectively, the findings challenge assumptions about the rarity of transoceanic dispersal in insects and underscore its potential importance in shaping distributions, gene flow, and biogeographic patterns over evolutionary and ecological timescales.

This study provides the first integrative documentation of a trans-Atlantic crossing by painted lady butterflies to South America, combining atmospheric modeling, genomics, pollen metabarcoding, isotope geolocation, and ecological niche models. The results show that small, gram-scale insects can traverse oceans by exploiting trade winds, with an inferred Europe–Africa–Americas pathway possibly exceeding 7000 km. The work highlights atmospheric circulation as a key driver of insect dispersal and suggests such events may be underdetected yet consequential for ecology and biogeography. Future research should quantify the frequency and seasonality of such crossings, extend multi-proxy approaches to other taxa and routes, refine energetic and flight behavior models (including altitude selection and minimal-effort flight costs), and improve isoscape calibrations and real-time atmospheric monitoring for predictive dispersal forecasting.

- Sample size and observation constraints: Only three individuals were captured from a single observed flock and date; lack of continuous tracking introduces uncertainty in precise routes and behavior aloft. - Modeling assumptions: Energetic models borrow metabolic and fat-reserve parameters primarily from monarchs and lab-reared butterflies, with unmeasured species-specific values for V. cardui; minimal-effort flight metabolic costs and active/minimal time proportions are approximations. - Isotope geolocation uncertainty: δ2H calibrations and 87Sr/86Sr isoscapes carry spatial uncertainty; dual-isotope assignments provide probabilistic, not definitive, origins and depend on the quality of baseline datasets. - Atmospheric reconstruction: HYSPLIT back-trajectories are subject to reanalysis data resolution and modeling uncertainty; inferred timings (e.g., coastline crossing) may have error margins. - Lack of South American establishment data: The absence of established populations limits assessment of demographic consequences; detection bias may obscure transient colonization events.