Phylodynamic analysis of ~13,000 hemagglutinin sequences from 2002–2007 showed that seasonal influenza A(H3N2) circulates continuously in an East–Southeast (E-SE) Asian network of temporally overlapping epidemics that seeds each temperate epidemic season, but it could not determine why this network dominates global circulation or how surveillance should be redesigned to exploit the finding. Later work complicated the strict source–sink picture (persistence and temperate hubs outside Asia), established that circulation patterns are subtype- and context-dependent, implicated mobility-mediated lineage competition and population-immunity structure as candidate drivers, and produced increasingly skillful evolutionary forecasts — yet no study has quantitatively discriminated the competing mechanistic hypotheses (regional population-immunity structure versus higher introduction rates along the mobility network versus climatic/demographic drivers of year-round circulation) or translated any mechanism into a surveillance design with a demonstrated gain in prediction horizon. The problem asks for exactly that: a mechanistic, predictive account of the global circulation network of seasonal influenza and the surveillance density, geographic allocation, and real-time analytics it implies for vaccine strain selection.

Background

Seasonal influenza A(H3N2) — the subtype that entered humans in 1968 and causes the most severe annual epidemics among circulating seasonal influenza viruses — evades prior immunity through antigenic drift: the stepwise accumulation of amino-acid changes, mainly in hemagglutinin (HA), the viral surface protein that is the principal target of neutralizing antibodies. Because drift steadily erodes vaccine effectiveness, the World Health Organization (WHO) Global Influenza Surveillance Network (later the Global Influenza Surveillance and Response System, GISRS), a network of national laboratories that contributes virus isolates and sequence and antigenic data, must update the vaccine composition roughly annually, choosing strains months before the season they must protect against. Antigenic properties are still classically measured by hemagglutination-inhibition (HI) assays, and antigenic cartography — the geometric mapping of HI titers into a two-dimensional antigenic space — visualizes the drift of circulating strains relative to the vaccine strain. The recurring practical failure mode is vaccine mismatch: the recommended strain is antigenically behind the strains that dominate the next season. A mechanistic question sits underneath this operational one. Genetic and antigenic analysis of HA from viruses collected on six continents during 2002–2007 (Russell, Jones, Barr, Cox and colleagues, Science 2008; summarized for vaccine strain selection in Russell et al., Vaccine 2008, the source of this problem) showed that H3N2 circulates continuously within an E-SE Asian network of temporally overlapping epidemics, that epidemics in temperate regions are seeded from this network each year (reaching Oceania, North America and Europe first, and South America last), and that strains leaving the network rarely contribute to long-term evolution. Phylogeography — the inference of viral migration routes from dated phylogenetic trees of viral sequences — and phylodynamics, the broader study of how epidemiological, immunological and evolutionary processes shape viral phylogenies, are the methodological backbone of this line of work. The 2008 analysis left the mechanism undetermined: it did not establish whether locally emerging E-SE Asian strains outcompete strains returning into the network predominantly because population immunity within the network is more advanced (so locally adapted variants are pre-equipped to escape immunity elsewhere), because introduction rates among the well-connected countries inside the network are higher, or because of other factors such as climate-driven year-round transmission. Discriminating among these candidate mechanisms — and between a strong E-SE Asian source–sink metapopulation and a more diffuse global migrating metapopulation — remained an open question contested by subsequent work, as did the fine-scale structure and the demographic and climatic drivers of the network, whether the pattern generalizes across time and subtypes, and how surveillance would have to be densified and re-allocated to extend the prediction horizon of antigenic variant emergence; the latter elaborates the 2008 finding that, if the trends observed during the study period are an accurate representation of overall patterns of spread, the antigenic characteristics of A(H3N2) viruses outside E-SE Asia may be forecast each year based on surveillance within E-SE Asia, with consequent improvements to vaccine strain selection. Subsequent research has both qualified the empirical picture and deepened the mechanistic puzzle. Bedford, Cobey, Beerli and Pascual (2010), reconstructing global transmission histories for 1998–2009, found China and Southeast Asia central to the network but also multi-season persistence of lineages outside Asia, with the USA acting as the primary temperate hub. Bahl and colleagues (2011), analyzing 2003–2006 samples including newly sequenced Hong Kong viruses, concluded that tropical Southeast Asia did not maintain the global source and that multiple lineages and multiple regions can seed temperate epidemics. Volz, Koelle and Bedford (2013) summarized the resulting state of the field as an open debate between a strong E-SE Asian source–sink metapopulation model and a more diffuse global metapopulation — a metapopulation being a set of geographically separated local viral populations connected by migration. Zinder, Bedford, Gupta and Pascual (2013) showed that a mutation–competition model with the lower and mixed seasonality of the tropics reproduces the higher E-SE Asian contribution to globally circulating lineages. Bedford, Riley, Barr and colleagues (2015), analyzing 9,604 HA sequences from 2000–2012, confirmed reseeding of H3N2 from E-SE Asia but showed that pre-2009 seasonal H1N1 and the two influenza B lineages circulate differently, persisting across seasons with a limited E-SE Asian role, and argued from epidemic models that differences in age of infection combined with the less frequent travel of children drive these subtype differences. The COVID-19 pandemic provided a natural perturbation experiment: Chen and colleagues (2024) showed pandemic interventions reshaped global lineage dispersal (South Asia serving as the phylogenetic trunk location during the pandemic, possible extinction of B/Yamagata), with dispersal intensity returning to pre-pandemic patterns after air-traffic recovery; Chen and colleagues (2025) additionally characterized the internal movement structure within Southeastern Asia during 2007–2023 and its disruption during two pandemics. At national scale, de Jong, Conlan, Han and Russell (2025) showed for the USA (2014–2023) that competition between co-circulating transmission lineages, mediated by the human mobility network, shapes yearly epidemics while the timing and explosiveness of initial epidemic sparks limit predictability. On the immunity side, Eales, Shearer and McCaw (2025) modeled how lifetime exposure histories and antigenic seniority (the tendency of early-childhood imprinting infections to dominate later immune responses) shape long-term H3N2 dynamics, and Kikawa and colleagues (2026) showed that high-throughput neutralization measurements of population antibody landscapes correlate strongly with the evolutionary success of H3N2 strains. In parallel, forecasting methodology has advanced from the predictive fitness model of Łuksza and Lässig (2014) through evolution-informed epidemiological forecasting (Du, King, Woods and Pascual, 2017) to scalable phylodynamic fitness estimation (Lefrancq and colleagues, 2025) and to explicit quantification of the forecast-accuracy gains attainable by shortening the vaccine-selection forecast horizon and the sequence-submission lag (Huddleston and Bedford, 2025). What none of these achievements supplies is the discriminating, quantitative mechanistic account — the relative contributions of immunity structure, introduction rates, and climatic/demographic drivers — or the operational surveillance design that the 2008 authors anticipated such an account would enable.

Problem Statement

Determine the mechanistic determinants of the global circulation of seasonal influenza A(H3N2), and from them the surveillance design required for vaccine strain selection. Concretely, using global genetic, antigenic, and where possible serological and mobility data: (1) quantify the relative contributions of the candidate proximate mechanisms debated in the literature following the source work — more advanced or differently structured population immunity within the East–Southeast (E-SE) Asian circulation network, higher rates of viral introductions among the well-connected countries inside the network, and climatic or demographic drivers of the network's year-round circulation — and establish which of them actually sustains the network's dominant role in producing the antigenic variants that seed temperate epidemics, resolving the long-standing debate between a strong E-SE Asian source–sink metapopulation and a more diffuse global migrating metapopulation; (2) characterize the fine-scale spatial topology and connectivity of the network (including the relative contributions of tropical, subtropical and temperate locations); and (3) on that mechanistic basis, determine the surveillance density, geographic allocation, and real-time analytical practices that would extend the prediction horizon for emerging antigenic variants, and demonstrate retrospectively that the proposed surveillance configuration predicts the documented emergence of antigenic variants accurately over a longer horizon than current practice, benchmarked against known vaccine-mismatch seasons in routine WHO strain selection, and report the quantified horizon gain. All three deliverables are mandatory and jointly required; they share one overarching objective: to replace the descriptive source–sink picture with a predictive, mechanism-based account of global seasonal influenza circulation that is validated against observed phylodynamic patterns and translates into actionable surveillance design. The question of whether the pattern generalizes across subtypes and longer timeframes has been substantially answered by later work (the patterns are subtype- and context-dependent) and is not a required deliverable, though a mechanistic account should be consistent with those findings.

References

  1. Russell CA, Jones TC, Barr IG, Cox NJ, Garten RJ, Gregory V, Gust ID, Hampson AW, Hay AJ, Hurt AC, de Jong JC, Kelso A, Klimov AI, Kageyama T, Komadina N, Lapedes AS, Lin YP, Mosterín A, Obuchi M, Odagiri T, Osterhaus AD, Rimmelzwaan GF, Shaw MW, Skepner E, Stöhr K, Tashiro M, Fouchier RA, Smith DJ. Influenza vaccine strain selection and recent studies on the global migration of seasonal influenza viruses. Vaccine 26(S4):D31–D34, 2008. DOI: 10.1016/j.vaccine.2008.07.078. https://doi.org/10.1016/j.vaccine.2008.07.078
  2. Russell CA, Jones TC, Barr IG, Cox NJ, Garten RJ, Gregory V, Gust ID, Hampson AW, Hay AJ, Hurt AC, de Jong JC, Kelso A, Klimov AI, Kageyama T, Komadina N, Lapedes AS, Lin YP, Mosterín A, Obuchi M, Odagiri T, Osterhaus AD, Rimmelzwaan GF, Shaw MW, Skepner E, Stöhr K, Tashiro M, Fouchier RA, Smith DJ. The Global Circulation of Seasonal Influenza A (H3N2) Viruses. Science 320(5874):340–346, 2008. DOI: 10.1126/science.1154137. https://doi.org/10.1126/science.1154137
  3. Bedford T, Cobey S, Beerli P, Pascual M. Global migration dynamics underlie evolution and persistence of human influenza A (H3N2). PLoS Pathogens 6(5):e1000918, 2010. DOI: 10.1371/journal.ppat.1000918. https://doi.org/10.1371/journal.ppat.1000918
  4. Bahl J, Nelson MI, Chan KH, Chen R, Vijaykrishna D, Halpin RA, Stockwell TB, Lin X, Wentworth DE, Ghedin E, Guan Y, Peiris JS, Riley S, Rambaut A, Holmes EC, Smith GJ. Temporally structured metapopulation dynamics and persistence of influenza A H3N2 virus in humans. Proceedings of the National Academy of Sciences 108(48):19359–19364, 2011. DOI: 10.1073/pnas.1109314108. https://doi.org/10.1073/pnas.1109314108
  5. Volz EM, Koelle K, Bedford T. Viral phylodynamics. PLoS Computational Biology 9(3):e1002947, 2013. DOI: 10.1371/journal.pcbi.1002947. https://doi.org/10.1371/journal.pcbi.1002947
  6. Zinder D, Bedford T, Gupta S, Pascual M. The Roles of Competition and Mutation in Shaping Antigenic and Genetic Diversity in Influenza. PLoS Pathogens 9(1):e1003104, 2013. DOI: 10.1371/journal.ppat.1003104. https://doi.org/10.1371/journal.ppat.1003104
  7. Bedford T, Riley S, Barr IG, Broor S, Chadha M, Cox NJ, Daniels RS, Gunasekaran CP, Hurt AC, Kelso A, Klimov A, Lewis NS, Li X, McCauley JW, Odagiri T, Potdar V, Rambaut A, Shu Y, Skepner E, Smith DJ, Suchard MA, Tashiro M, Wang D, Xu X, Lemey P, Russell CA. Global circulation patterns of seasonal influenza viruses vary with antigenic drift. Nature 523(7559):217–220, 2015. DOI: 10.1038/nature14460. https://doi.org/10.1038/nature14460
  8. Łuksza M, Lässig M. A predictive fitness model for influenza. Nature 507(7490):57–61, 2014. DOI: 10.1038/nature13087. https://doi.org/10.1038/nature13087
  9. Du X, King AA, Woods RJ, Pascual M. Evolution-informed forecasting of seasonal influenza A (H3N2). Science Translational Medicine 9(413):eaan5325, 2017. DOI: 10.1126/scitranslmed.aan5325. https://doi.org/10.1126/scitranslmed.aan5325
  10. Lefrancq N, Duret L, Bouchez V, Brisse S, Parkhill J, Salje H. Learning the fitness dynamics of pathogens from phylogenies. Nature 637(8046):683–690, 2025. DOI: 10.1038/s41586-024-08309-9. https://doi.org/10.1038/s41586-024-08309-9
  11. Chen Z, Tsui JL-H, Gutierrez B, Busch Moreno S, du Plessis L, Deng X, Cai J, Bajaj S, Suchard MA, Pybus OG, Lemey P, Kraemer MUG, Yu H. COVID-19 pandemic interventions reshaped the global dispersal of seasonal influenza viruses. Science 386(6722):eadq3003, 2024. DOI: 10.1126/science.adq3003. https://doi.org/10.1126/science.adq3003
  12. Chen Z, Tsui JL-H, Cai J, Su S, Viboud C, du Plessis L, Lemey P, Kraemer MUG, Yu H. Disruption of seasonal influenza circulation and evolution during the 2009 H1N1 and COVID-19 pandemics in Southeastern Asia. Nature Communications 16:475, 2025. DOI: 10.1038/s41467-025-55840-y. https://doi.org/10.1038/s41467-025-55840-y
  13. de Jong SPJ, Conlan AJK, Han AX, Russell CA. Competition between transmission lineages mediated by human mobility shapes seasonal influenza epidemics in the US. Nature Communications 16:4605, 2025. DOI: 10.1038/s41467-025-59757-4. https://doi.org/10.1038/s41467-025-59757-4
  14. Eales O, Shearer FM, McCaw JM. How immunity shapes the long-term dynamics of influenza H3N2. PLOS Computational Biology 21(3):e1012893, 2025. DOI: 10.1371/journal.pcbi.1012893. https://doi.org/10.1371/journal.pcbi.1012893
  15. Kikawa C, Loes AN, Huddleston J, Figgins MD, Steinberg P, Griffiths T, Drapeau EM, Peck H, Barr I, Englund JA, Hensley SE, Bedford T, Bloom JD. High-throughput neutralization measurements correlate strongly with evolutionary success of human influenza strains. eLife 14:RP106811, 2026. DOI: 10.7554/eLife.106811. https://doi.org/10.7554/eLife.106811
  16. Huddleston J, Bedford T. Timely vaccine strain selection and genomic surveillance improve evolutionary forecast accuracy of seasonal influenza A/H3N2. eLife 14:RP104282, 2025. DOI: 10.7554/eLife.104282. https://doi.org/10.7554/eLife.104282
  17. Shi W, Wohlwend J, Wu M, Barzilay R. Influenza vaccine strain selection with an AI-based evolutionary and antigenicity model. Nature Medicine 31(11):3862–3870, 2025. DOI: 10.1038/s41591-025-03917-y. https://doi.org/10.1038/s41591-025-03917-y
  18. de Rooij AJH, Lempers VJC, Park Y, Vicic N, Han AX, Russell CA, Rudin D. Reproducible and later vaccine strain selection can improve vaccine match to A/H3N2 seasonal influenza viruses. npj Vaccines 10:243, 2025. DOI: 10.1038/s41541-025-01292-w. https://doi.org/10.1038/s41541-025-01292-w