The Winners Take It All? Evolutionary Success of H5Nx Reassortants in the 2020–2024 Panzootic

The Winners Take It All? Evolutionary Success of H5Nx Reassortants in the 2020–2024 Panzootic | bioRxiv /* */ /* */ <!-- <!-- /*! * yepnope1.5.4 * (c) WTFPL, GPLv2 */ (function(a,b,c){function d(a){return"[object Function]"==o.call(a)}function e(a){return"string"==typeof a}function f(){}function g(a){return!a||"loaded"==a||"complete"==a||"uninitialized"==a}function h(){var a=p.shift();q=1,a?a.t?m(function(){("c"==a.t?B.injectCss:B.injectJs)(a.s,0,a.a,a.x,a.e,1)},0):(a(),h()):q=0}function i(a,c,d,e,f,i,j){function k(b){if(!o&&g(l.readyState)&&(u.r=o=1,!q&&h(),l.onload=l.onreadystatechange=null,b)){"img"!=a&&m(function(){t.removeChild(l)},50);for(var d in y[c])y[c].hasOwnProperty(d)&&y[c][d].onload()}}var j=j||B.errorTimeout,l=b.createElement(a),o=0,r=0,u={t:d,s:c,e:f,a:i,x:j};1===y[c]&&(r=1,y[c]=[]),"object"==a?l.data=c:(l.src=c,l.type=a),l.width=l.height="0",l.onerror=l.onload=l.onreadystatechange=function(){k.call(this,r)},p.splice(e,0,u),"img"!=a&&(r||2===y[c]?(t.insertBefore(l,s?null:n),m(k,j)):y[c].push(l))}function j(a,b,c,d,f){return q=0,b=b||"j",e(a)?i("c"==b?v:u,a,b,this.i++,c,d,f):(p.splice(this.i++,0,a),1==p.length&&h()),this}function k(){var a=B;return a.loader={load:j,i:0},a}var l=b.documentElement,m=a.setTimeout,n=b.getElementsByTagName("script")[0],o={}.toString,p=[],q=0,r="MozAppearance"in l.style,s=r&&!!b.createRange().compareNode,t=s?l:n.parentNode,l=a.opera&&"[object Opera]"==o.call(a.opera),l=!!b.attachEvent&&!l,u=r?"object":l?"script":"img",v=l?"script":u,w=Array.isArray||function(a){return"[object Array]"==o.call(a)},x=[],y={},z={timeout:function(a,b){return b.length&&(a.timeout=b[0]),a}},A,B;B=function(a){function b(a){var a=a.split("!"),b=x.length,c=a.pop(),d=a.length,c={url:c,origUrl:c,prefixes:a},e,f,g;for(f=0;f<d;f++)g=a[f].split("="),(e=z[g.shift()])&&(c=e(c,g));for(f=0;f<b;f++)c=x[f](c);return c}function g(a,e,f,g,h){var i=b(a),j=i.autoCallback;i.url.split(".").pop().split("?").shift(),i.bypass||(e&&(e=d(e)?e:e[a]||e[g]||e[a.split("/").pop().split("?")[0]]),i.instead?i.instead(a,e,f,g,h):(y[i.url]?i.noexec=!0:y[i.url]=1,f.load(i.url,i.forceCSS||!i.forceJS&&"css"==i.url.split(".").pop().split("?").shift()?"c":c,i.noexec,i.attrs,i.timeout),(d(e)||d(j))&&f.load(function(){k(),e&&e(i.origUrl,h,g),j&&j(i.origUrl,h,g),y[i.url]=2})))}function h(a,b){function c(a,c){if(a){if(e(a))c||(j=function(){var a=[].slice.call(arguments);k.apply(this,a),l()}),g(a,j,b,0,h);else if(Object(a)===a)for(n in m=function(){var b=0,c;for(c in a)a.hasOwnProperty(c)&&b++;return b}(),a)a.hasOwnProperty(n)&&(!c&&!--m&&(d(j)?j=function(){var a=[].slice.call(arguments);k.apply(this,a),l()}:j[n]=function(a){return function(){var b=[].slice.call(arguments);a&&a.apply(this,b),l()}}(k[n])),g(a[n],j,b,n,h))}else!c&&l()}var h=!!a.test,i=a.load||a.both,j=a.callback||f,k=j,l=a.complete||f,m,n;c(h?a.yep:a.nope,!!i),i&&c(i)}var i,j,l=this.yepnope.loader;if(e(a))g(a,0,l,0);else if(w(a))for(i=0;i (function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start':new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0];var j=d.createElement(s);var dl=l!='dataLayer'?'&l='+l:'';j.src='//www.googletagmanager.com/gtm.js?id='+i+dl;j.type='text/javascript';j.async=true;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-M677548'); Skip to main content Home About Submit ALERTS / RSS Search for this keyword Advanced Search New Results The Winners Take It All? Evolutionary Success of H5Nx Reassortants in the 2020–2024 Panzootic View ORCID Profile James Baxter , View ORCID Profile Jing Yang , View ORCID Profile Will Harvey , View ORCID Profile Simon Dellicour , View ORCID Profile Anne Pohlmann , View ORCID Profile Mingxiao Ma , Wenjun Liu , View ORCID Profile Yuhai Bi , View ORCID Profile Paul Digard , View ORCID Profile Martin Beer , View ORCID Profile Marion Koopmans , View ORCID Profile Samantha Lycett , View ORCID Profile Lu Lu doi: https://doi.org/10.1101/2025.07.19.665680 James Baxter 1 Roslin Institute, The University of Edinburgh, Edinburgh , United Kingdom Find this author on Google Scholar Find this author on PubMed Search for this author on this site ORCID record for James Baxter Jing Yang 2 Institute of Microbiology, Chinese Academy of Sciences , Beijing, People’s Republic of China Find this author on Google Scholar Find this author on PubMed Search for this author on this site ORCID record for Jing Yang Will Harvey 1 Roslin Institute, The University of Edinburgh, Edinburgh , United Kingdom Find this author on Google Scholar Find this author on PubMed Search for this author on this site ORCID record for Will Harvey Simon Dellicour 3 Spatial Epidemiology Lab (SpELL), Université Libre de Bruxelles , Brussels, Belgium 4 Department of Microbiology, Immunology and Transplantation, KU Leuven , Leuven, Belgium 5 Interuniversity Institute of Bioinformatics in Brussels, Université Libre de Bruxelles , Brussels, Belgium Find this author on Google Scholar Find this author on PubMed Search for this author on this site ORCID record for Simon Dellicour Anne Pohlmann 6 Institute of Diagnostic Virology, Friedrich-Loeffler-Institut , Greifswald-Riems, Germany Find this author on Google Scholar Find this author on PubMed Search for this author on this site ORCID record for Anne Pohlmann Mingxiao Ma 7 Collaborative Innovation Center for Prevention and Control of Zoonoses, Jinzhou Medical University , Jinzhou, People’s Republic of China Find this author on Google Scholar Find this author on PubMed Search for this author on this site ORCID record for Mingxiao Ma Wenjun Liu 2 Institute of Microbiology, Chinese Academy of Sciences , Beijing, People’s Republic of China Find this author on Google Scholar Find this author on PubMed Search for this author on this site Yuhai Bi 2 Institute of Microbiology, Chinese Academy of Sciences , Beijing, People’s Republic of China Find this author on Google Scholar Find this author on PubMed Search for this author on this site ORCID record for Yuhai Bi Paul Digard 1 Roslin Institute, The University of Edinburgh, Edinburgh , United Kingdom Find this author on Google Scholar Find this author on PubMed Search for this author on this site ORCID record for Paul Digard Martin Beer 6 Institute of Diagnostic Virology, Friedrich-Loeffler-Institut , Greifswald-Riems, Germany Find this author on Google Scholar Find this author on PubMed Search for this author on this site ORCID record for Martin Beer Marion Koopmans 8 Erasmus MC, Viroscience and Pandemic and Disaster Preparedness Centre , Rotterdam, the Netherlands Find this author on Google Scholar Find this author on PubMed Search for this author on this site ORCID record for Marion Koopmans Samantha Lycett 1 Roslin Institute, The University of Edinburgh, Edinburgh , United Kingdom Find this author on Google Scholar Find this author on PubMed Search for this author on this site ORCID record for Samantha Lycett Lu Lu 1 Roslin Institute, The University of Edinburgh, Edinburgh , United Kingdom Find this author on Google Scholar Find this author on PubMed Search for this author on this site ORCID record for Lu Lu For correspondence: lu.lu{at}ed.ac.uk Abstract Full Text Info/History Metrics Preview PDF Abstract Avian influenza viruses undergo frequent genetic reassortment, which can coincide with phenotypic changes in transmission, pathogenicity, and host species niche. Since 2020, clade 2.3.4.4b H5 high pathogenicity avian influenza viruses (HPAIVs) have driven a global panzootic, causing mass mortality in wild birds, poultry, and, for the first time, repeated spillover infections in a variety of mam-malian species. This resurgence of H5 HPAIV has coincided with a dramatic increase in the number of circulating reassortant strains; however, the scale, impact and drivers of these reassortants remain unknown. Here, we combined statistical and phylodynamic modelling to reconstruct the global evolutionary dynamics of H5Nx viruses across four epizootic seasons (2020-2024). We identified 209 genetically distinct reassortants, stratified into three transmission categories based on their phylogenetic and epidemiological profiles. Accounting for sampling depth and HPAIV incidence, we estimated that reassortants emerged most frequently in Asia, but ‘major’ reassortants associated with increased host range, inter-seasonal persistence, and long-range dissemination, more frequently emerged from Europe. Altogether, reassortant emergence followed an episodic pattern in which most reassortants were transient, but 3% seeded large clusters of secondary reassortants soon after their own emergence. Statistical modelling revealed that reassortant success was strongly shaped by ecological factors, including circulation in specific wild bird orders and the ability to infect a wider range of host niches. Reassortant dispersal was linked to poultry trade intensity, particularly in North America. Collectively, our findings reveal reassortment dynamics in H5 HPAIVs and identify key virological and ecological drivers underpinning the emergence and global spread of successful reassortants. These insights support the importance of enhanced surveillance to track evolution of H5 HPAIV and identify traits relevant for consideration in pandemic risk assessment. 1 Introduction The Influenza A virus (IAV) genome comprises eight single-stranded negative sense RNA segments, which can reassort when two or more strains coinfect the same cell ( Kawaoka et al. 1989 ). Genetic reassortment can alter IAV evolutionary dynamics, if it leads to phenotypic ‘shifts’ that play a critical role in the emergence and crossspecies transmission of novel viral strains ( Lowen 2017 ). Notably, reassortment of IAVs has been implicated in several past pandemics including 1918 H1N1, 1957 H2N2 and 1968 H3N2, where gene segments from avian influenza viruses combined with human-adapted viruses ( Taubenberger et al. 2005 ; Schäffr et al. 1993 ; Fang et al. 1981 ). While inter-stubype reassortment between IAVs circulating in the human population is relatively infrequent, the coexistence of multiple AIV subtypes in the wild avian population coincides with a remarkably high rate of reassortment ( Dugan et al. 2008 ; Lu et al. 2014 ). H5 High Pathogenicity Avian Influenza Virus (HPAIV) was first isolated from a poultry outbreak in Aberdeenshire, Scotland in 1959 ( Pereira et al. 1965 ); however, it was not until the H5N1 goose/Guangdong (Gs/Gd) lineage emerged in China in 1996 that sustained transmission in domestic poultry was established ( Chen et al. 2006 ). Since 2005, repeated spillovers from domestic poultry to wild Anseriformes (geese, ducks and swans) established a broad diversity of H5Nx lineages across Asia, Europe, and Africa ( Salzberg et al. 2007 ; Lycett et al. 2019 ; Smith et al. 2015 ). In 2014, the emergence of a lineage with reduced virulence in wild Anseriformes (clade 2.3.4.4), led to the dispersal of Gs/Gd HPAIV H5Nx to North America for the first time ( THE GLOBAL CONSORTIUM FOR H5N8 AND RELATED INFLUENZA VIRUSES 2016 ; Lee et al. 2015 ). From 2016 onwards, clade 2.3.4.4b H5N8 viruses originating from East Asia repeatedly caused outbreaks in wild Anseriformes, coinciding with a relative increased virulence for Anatidae spp and a marked increase in reassortment frequency compared with the 2014/2015 seasons and earlier European H5N1 outbreaks ( Hesterberg et al. 2009 ; Lycett et al. 2020 ; Pohlmann et al. 2018 ). Since 2020, clade 2.3.4.4b H5N1 HPAIV have surged to cause unprecedented outbreaks in wild-bird species worldwide, displacing contemporary H5N8 lineages, and showing remarkable persistence in wild birds alongside a broadened host range ( Xie et al. 2023 ; Lewis et al. 2021 ; Zeng et al. 2024 ). By 2024, descendants of the 2020 H5N1 lineage had caused severe mortality in wild birds and domestic poultry across Europe, North America, and Africa. Moreover, sustained avian-to-mammal transmission for the first time became apparent across multiple settings including fur farms ( Agüero et al. 2023 ), as well as marine and terrestrial mammals ( Elsmo et al. 2023 ; Leguia et al. 2023 ; Uhart et al. 2024 ; Tomás et al. 2024 ; Peacock et al. 2025 ). Circulation in mammals, notably cattle in the US, has resulted in sporadic mammal-to-human transmissions, typically following prolonged exposure ( Morse et al. 2024 ; Uyeki et al. 2024 ). The 2020-2024 panzootic was characterised by extensive genetic reassortment and circulation of reassortant viruses across a wide diversity of host species worldwide. The reassortment dynamics of clade 2.3.4.4b H5Nx viruses from Europe in 2020–2022 ( Fusaro et al. 2024 ) and from North America ( Youk et al. 2023 ; Signore et al. 2025 ) have been reported; however, a systematic investigation of this complex evolutionary picture at the global scale has yet to be undertaken. In this study, we applied a previously described global reassortant classification system ( Lycett et al. 2020 ) to characterise H5 HPAIV reassortants of clade 2.3.4.4b that circulated between 2020 and 2024. We used phylogenetic approaches to ascertain spatio-temporal patterns of reassortant emergence, before combining phylodynamic and statistical modelling to elucidate drivers of reassortant persistence and dispersal. We show that a small minority (5/209) of reassortants dominated the 2020-2024 panzootic, and reveal how short inter-reassortment generation times led to a clustered, episodic pattern of reassortant emergence. Averaged across all reassortants, we find viral dispersal was maintained in Anseriformes spp, and accelerated by circulation in Charadriiformes spp. Collectively, we discuss the role of virological and ecological factors on the formation of distinct reassortment patterns throughout the 2020-2024 panzootic. 2 Methods 2.1 Sequence Data Curation and Assembly We assembled a dataset of all available HPAIV H5 clade 2.3.4.4b full genome sequences from the GISAID EpiFlu database collected between 1st Jan 2019 and 1st May 2024 (Downloaded 28th May 2024) ( Shu and McCauley 2017 ). Altogether, these data included 9,935 genomes, including viruses of the subtypes H5N8 ( n = 1,587), H5N1 ( n = 8,143), H5N5 ( n = 94), H5N6 ( n = 70) and other H5Nx ( n = 41). 93.86% of these sequences were sampled from birds ( n = 9,325), with the remainder sampled from mammals ( n = 457) and the environment ( n = 153). These data also include early sequences sampled from H5N1 infected cattle in the United States of America ( n = 12, collected in March 2024). Details of data and data providers are summarised and acknowledged (SI Appendix) . 2.2 Reassortant Classification We determined reassortment profiles for each influenza genome (SI Appendix) according to a previously described reassortant classification algorithm for global clade 2.3.4.4b H5 AIVs ( Lycett et al. 2020 ). Briefly, we aligned the nucleotide sequences of each of the eight gene segments using MAFFT v7.511 ( Katoh and Standley 2013 ); excluded noncoding regions and removed insertions present in fewer than 10% of sequences. We defined clusters using nearest neighbour hierarchical clustering for genomes containing all 8 segments, each with at least 80% nucleotide coverage. Specifically, we clustered pairwise nucleotide distances, setting a 0.5% threshold for the longest three gene segments (PB2, PB1 and PA) and a 1% threshold for the remaining segments using “bioseq” package in R v4.1.2 ( Keck 2020 ). We selected different thresholds to account for differences in segment length, resulting in monophyletic clusters distinguished by approximately 8-15 nucleotides. Our reassortant profiles were broadly consistent with genotyping schemes established by the European Union Reference Laboratory for Avian Influenza and Newcastle Disease ( Fusaro et al. 2019 ) and the United States Department of Agriculture ( Youk et al. 2023 ). For each gene segment cluster, we assigned a number that corresponds to the ranked cluster size (1-n, largest to smallest), resulting in a unique 8-number barcode based on the cluster combination of eight genomic segments for each genome (e.g 21111111). For ease of reference, the label for each reassortant profile is given by the virus HN subtype, collection year of the first sampled genome, and the chronological order of discovery among all reassortant profiles identified within that year (e.g., 2020/H5N1/R1 - see supplementary table for details). 2.3 Time and location-stratified reassortant diversity We investigated the diversity of H5Nx reassortants through time, stratified by their sampling continent. Using a sliding time window of one year, we calculated the proportional abundances of reassortants sampled in each affected continent, excluding Antarctica (Africa, Asia, Europe, North America and South America). From these proportional abundances, we calculated diversity as the Hill number of order 1 ( Hill 1973 ). This measure accounts for both the number of reassortants present within a discrete place and time and the evenness of the spread or distribution of reassortants; it represents the effective number of reassortants present if they were evenly distributed and is equal to the natural exponent of Shannon entropy. To mitigate the naïve assumption implicit in the above computation, that all reassortant profiles are equally distinct, we additionally calculated a similarity-sensitive measure of diversity ( Leinster and Cobbold 2012 ). We calculated a similarity matrix from pairwise nucleotide distances between consensus sequences for each reassortant using an exponential transformation using the “rdiversity” package in R v4.1.2 ( Mitchell et al. 2022 ). To account for differences in sampling intensity between continents, we repeatedly and randomly down-sampled overrepresented continents to match the number of profiles present in the least well represented continent. For each metric, we quantified patterns over time using a univariate generalised additive model (GAM) (using “mgcv” v1.9-1 package in R v4.1.2) with a penalised cubic regression spline fitted to the midpoints of the sliding windows ( Wood 2011 ). 2.4 Phylodynamic analyses To provide a contextual background for further analyses, we identified the 500 most genetically similar sequences for each gene segment of every reassortant profile using the Basic Local Alignment Search Tool (BLAST) with the default parameters provided by the GISAID database ( Altschul et al. 1997 ). To ensure our results reflected the origins of each clade 2.3.4.4.b reassortant, we tested only the earliest sequences of each reassortant. Our BLAST search results were not constrained by time, host, location, or subtype. We curated the accompanying data for the contextual sequences as described for our whole genome data. To infer key evolutionary parameters of all reassortant profiles identified in this study, we first grouped genomes into reassortants that were found within a given continent (intracontinental) and reassortants found in more than one continent (intercontinental). We analysed intercontinental reassortants ( n = 13) individually and analysed intracontinental reassortants ( n = 196) together within each continent. For each dataset, we analysed each segment and NA serotype separately. We inferred maximum likelihood trees using IQTREE v.2.1.3, assuming an HKY (Hasegawa Kishino Yano) substitution model with a four-category gamma distribution model for among-site rate heterogeneity ( Hasegawa et al. 1985 ; Yang 1994 ), and confirmed the heterochronicity of each alignment using Tempest v1.5.3 ( Max Carvalho et al. 2016 ). To balance the distribution of sequence samples across each phylogeny, we sub-sampled the combined dataset. Specifically, for each alignment, we calculated pairwise Hamming distances (HD) and inferred sequence clusters with a maximum difference of 5 SNPs. We then selected one genome for each unique combination of location (defined as most refined administrative subdivision available), host order, reassortant, and HD cluster. We estimated Bayesian time-resolved phylogenetic trees for each alignment using BEAST v1.10.4 coupled with the BEAGLE v3.1 library to enhance computational performance ( Suchard et al. 2018 ; Ayres et al. 2019 ). We assumed a SRD06 nucleotide substitution model, in which codon positions 1+2 and 3 are partitioned and a separate HKY model with a four-category gamma distribution model for among-site rate heterogeneity, is fitted to each ( Hasegawa et al. 1985 ; Yang 1994 ; Shapiro et al. 2006 ). Additionally, we assumed an uncorrelated relaxed molecular clock, with evolutionary rates sampled from a lognormal distribution ( Drummond et al. 2006 ). We specified a lognormal prior for mean evolutionary rate ( X ∼ LogNormal(−8.537, 1.805)), a uniform prior for relative rates amongst partitions ( X ∼ Uniform(0, 100)) and a nonparametric skygrid coalescent tree prior ( Gill et al. 2013 ). For each gene-segment alignment, we ran two independent Markov Chain Monte Carlo (MCMC) simulations, each comprising 2 × 10 8 iterations with sampling every 20,000 iterations. We assessed convergence and satisfactory effective sample size (ESS >200) using Tracer 1.7.2 ( Rambaut et al. 2018 ). Next, we created a detailed phylodynamic profile for each reassortant. Briefly, we estimated the most recent common ancestor (MRCA) for every reassortant in each phylogeny, reconstructing the ancestral reassortant ‘state’ backwards in time using an asymmetric continuous-time Markov chain ( Lemey et al. 2009 ). For each reassortant-specific clade, we then estimated: i) the median posterior evolutionary rate, ii) the ancestral states of influenza subtype, iii) continent, iv) host, and v) the number and identity of the segments changed relative to the immediately ancestral reassortant. We further calculated for each reassortant-specific clade: vi) the frequency of host transitions, vii) the frequency of avian-to-mammal transitions, and viii) the median persistence time in each host. Finally, we modelled the spatio-temporal diffusion patterns for each reassortant, using a gamma relaxed random walk (RRW) diffusion model incorporating the geospatial location (latitude and longitude) of tip sequences ( Lemey et al. 2010 ). For all reassortants, we estimated the weighted diffusion coefficient, which reflects the land area invaded per unit of time ( Dellicour et al. 2024 , Dellicour et al. 2016 ). To understand underlying factors related to the five major H5 reassortants of clade 2.3.4.4b and H5 HPAIV spread among sub-continent regions, we extended our discrete trait analysis with a phylogenetic generalised linear model (PGLM) ( Lemey et al. 2014 ). Informed by our previous work ( Yang et al. 2019a ,b), we chose six candidate predictors complementary to wild bird movement for inclusion in the PGLM analyses: i) live poultry trade, ii) poultry production, iii) geographic distance, iv) the integrated free-range duck farming style, v) the coastline/area ratio, and vi) sample size. Prior to model fitting, we tested for multicollinearity between the variables. For continuous variables, we standardised the variables to a mean value of 0 and a standard deviation of 1. Due to limited sequencing in some countries, we aggregated geographically close countries into sub-regions. More details of predictor selection are provided in the supplementary methods (Appendix A.1, p35). 2.5 Statistical analyses 2.5.1 Reassortant Clustering We classified reassortants using K-means clustering according to their dispersal scales in space, time, and host range. Specifically, we fitted to i) the number of different bird orders from which the reassortant had been sampled (bird richness), ii) whether the reassortant had been isolated in mammals or not, iii) the maximum geographic distance between samples within each reassortant, iv) the number of genomes, v) weighted diffusion coefficient, vi) evolutionary rate, vii) persistence time (time between the latest sampling date and the estimated date for the MRCA), vii) the number of ancestral host transitions and ix) the number of ancestral host transitions to mammals. For data extracted from phylodynamic analyses (v-ix), we included estimates inferred from HA and PB2 segments only. Prior to model fitting, we normalised numeric variables and encoded presence in mammals using one-hot encoding. We iteratively fitted clusters for K in 1, 2, …, 20, selecting the optimal value of K according to the ‘Elbow’ method. For the optimal value of K, we inferred the relative importance of each variable to cluster assignment with a permutation analysis. Specifically, we sequentially excluded one variable from the dataset, refitted the model, and quantified the difference between each permutation and the original dataset using the adjusted rand index (ARI) ( Rand 1971 ; Hubert and Arabie 1985 ). We implemented these models using the “stats” v4.4.2 package in R v4.1.2 ( R Core Team 2018 ). Extensive use was made of the Tidyverse suite v2.0.0 for data handling, Recipes v1.09, and Broom v0.2.9.6 in the model pipeline ( Robinson et al. 2025 ; Wickham et al. 2019 ). 2.5.2 Number of Reassortants To understand global patterns of reassortant emergence, we fitted a hierarchical model to predict the number of novel reassortants, y , stratified by year-month, i , and continent, j . Our model consisted of three components, inspired by ecological models ( Royle 2004 ; Knape et al. 2018 ). First, we hypothesised that reassortants could only emerge at a fraction of time points, perhaps due to epidemiological or ecological suitability. We assumed conditions at each time point were either permissible or not for reassortment, z ij ∈ {0, 1}, with continent-specific probability, θ j ∈ (0, 1). Next, we assumed that only a proportion of reassortants are ever observed due to incomplete sampling, p ij ∈ (0, 1). Finally, we assumed that a ‘true’ latent number of reassortants per month per continent, N ij , follows a Poisson distribution with continent-stratified rate λ ij . We modelled additional linear predictors for each of θ j , N ij and y ij . Specifically, we included continent-specific intercepts for all components, in addition to global coefficients for HPAIV incidence and the number of GISAID whole genomes for y ij and N ij , respectively. We approximated HPAIV incidence as the number of (exposed) subunits from all biannual reports submitted to the World Organisation for Animal Health (WOAH) related to H5 HPAIV in wild birds, mammals and poultry between 1st January 2017 and 31st December 2024 (Accessed 2025-Feb-25) ( Figure S1 , p42). We log-transformed continuous predictors prior to fitting the model. We fitted non-centred random intercepts for the year in which the reassortant was estimated to have emerged for N ij and y ij . A full description of the statistical model is provided in the supplementary methods (Appendix A.2, p35). 2.5.3 Reassortant Classification In a separate model, we estimated the probability that a novel reassortant, y i , is assigned a class, c , from the ordered set C = {minor, moderate, major}. We assumed that the probability a reassortant is assigned a given class follows a cumulative distribution, with classes increasing from minor to moderate to major ( Samejima 1997 ). We modelled each class as the discretisation of a latent (unobserved) continuous variable, , via threshold parameters, τ c , which partition the distribution into classspecific intervals. For each threshold, we modelled four linear predictors. Specifically, we included i) the class identity of the reassortant immediately ancestral to reassortant i (with respect to HA), ii) origin continent, iii) the number of segments changed relative to the immediately ancestral reassortant, and iv) the time interval between the MRCA of reassortant i and that for the most recent major reassortant. We modelled a penalised thin plate regression spline to smooth the time interval between the MRCA of reassortant i and that for the most recent major reassortant ( Wood 2003 ). A full description of the statistical model is provided in the supplementary methods (Appendix A.3, p39). 2.5.4 Reassortant Diffusion Coefficient To evaluate determinants of novel reassortants dispersal, we fitted a mixed model to predict the weighted diffusion coefficients (km 2 year -1 ), y , calculated from our phylogeographic analysis for each novel reassortant, i . We restricted our analysis to reassortants with a clade size greater than 1, since we cannot confidently distinguish between reassortants that truly exist at a single locus and reassortants with limited (but non-zero) circulation and incomplete sampling. For all y i > 0, we assumed a gamma distribution parametrised such that, y i ∼ Gamma( κ i , θ i ), with rate parameter, , shape parameter κ i and mean µ i . We modelled a linear combination of predictors for each of µ i and κ i . For both µ i and κ i , we included continent-specific intercepts. For µ i only, we included predictors for the number of host state transitions across the phylogeny, the proportion of evolutionary time (sum of branch lengths) in Anseriformes spp (including ducks, geese and swans), Charadriiformes spp (including waders, gulls and auks) and Galliformes spp (including turkeys, chickens and quail); and persistence time. We log-transformed the number of host state transitions prior to model fitting. We allowed persistence time and the proportion of branching time in either Anseriformes spp to vary by continent (i.e continent is the grouping factor for non-centred multivariate random effects). We also included uncorrelated random intercepts for the date of the reassortant MRCA, grouped by calendar year. A full description of the statistical model is provided in the supplementary methods (Appendix A.4, p40). 2.5.5 Computation Due to limited prior knowledge of the relationship between the explanatory and predictor variables, we specified weakly informative priors for all models. Exact prior specifications for each model are described in the supplementary methods (Appendix A, p35). To fit each model, we ran four parallel MCMC simulations, each of 5000 iterations. The first 500 iterations of each chain were discarded. We assessed the convergence of posterior chains and adequate mixing against criteria of ESS > 1, 000 and rank-normalised Potential Scale Reduction Factor (PSRF) of ( Vehtari et al. 2021 ). We conducted further visual checks for within-chain autocorrelation, parameter identifiability and the fit of the posterior predictive distribution to our data ( Figures S5 - S7 , S11 - S13 , & S16 - S18 ). The ‘number of reassortants’ model was fitted in Stan using cmdstanr v0.8.1 and the remaining models were fitted in Stan using BRMS v2.20.4 in R v4.1.2 ( Gabry et al. 2025 ; Bürkner 2017 ; Carpenter et al. 2017 ; R Core Team 2018 ). Extensive use was made of the Tidyverse suite v2.0.0 for data handling, and tidybayes v3.0.6 and marginaleffects v0.25.0 for post-processing ( Wickham et al. 2019 ; Arel-Bundock et al. 2024 ; Kay 2024 ). 3 Results 3.1 Patterns of Novel Reassortant Emergence We identified 209 unique reassortants across the clade 2.3.4.4b H5Nx genomes assembled for this study ( Figure 1A ). Importantly, these data encompass critical changes in HPAI H5 epidemiology, including the near-total replacement of H5N8 by H5N1 in 2021, the H5N1 panzootic and the initial phase of H5N5 resurgence in early 2024 ( Figures S1 & S2A ). Concomitantly, H5 HPAIVs expanded to a broader range of new hosts in addition to the increasing number of infections in major reservoir species ( Figure S2B ). This resulted in sequential epizootics in European seabirds, marine mammals in South America, and more recently, dairy cattle in North America. Reconstructing the ancestral patterns of reassortant profiles for each continent, we inferred the number of unique reassortants emerging from each ( Figure 1 ). Asia harboured the largest number of unique reassortants ( n = 95), followed by Europe ( n = 58), North and central America ( n = 44) and Africa ( n = 12). We did not infer the origins of any reassortant to be located in South America, Southern Ocean and Antarctica, or Australasia and Oceania during this time period, highlighting these regions as sinks for newly emerged reassortants that diffuse southwards from northern and central America and Asia, respectively. Download figure Open in new tab Fig. 1. HPAI H5 clade 2.3.4.4b reassortment diversity in multiple continents. (A) Time-resolved maximum clade credibility (MCC) phylogenetic tree, summarising the global Haemagglutinin (HA) H5 2.3.4.4b diversity in our data. Coloured tips denote five of the most numerous reassortants in our dataset. Inset is the number of reassortants estimated to have originated from each continent, host and season. Densities correspond to the posterior estimates of the date of the most recent common ancestor of each reassortant. B) the distribution of whole genomes per reassortant over time We also compared reassortant emergence across different host groups ( Figure 1C ). Most reassortants emerged in wild Anseriformes spp ( n = 125; 59.8%), Galliformes spp ( n = 31; 14.8%) and other wild birds ( n = 25; 12.0%). Stratifying these estimates across continents, we found wild Anseriformes spp to be the primary origin host for clade 2.3.4.4b H5Nx reassortants in Asia, Europe, and North America. In contrast, reassortants emerging from Africa were likely to have originated in domestic Galliformes spp ( n = 12); however, sampling bias and divergent evolutionary trajectories likely constrains the accuracy of the host origin estimation. While novel H5Nx reassortants occur year-round, these data also reveal clear temporal trends in reassortment ( Figure 1D ). Specifically, novel H5Nx reassortant types were more often detected during the autumn migration season (approximately between September–November; n = 78), followed by the breeding season (between June–August; n = 56), the wintering period (December–February; n = 54), and the spring migration season (March–May; n = 21). 3.2 Reassortant Emergence and Diversity Through Time We quantified the rate of reassortant emergence using a hierarchical model designed to disentangle ecological and anthropogenic effects. After accounting for variation in sampling and virus ecology, we estimated the true (latent) number of reassortants per month. Assuming a fixed number of HPAIV cases and marginalising over calendar year, we found that reassortants most frequently emerged from Asia at a rate of 2.01 (95% Highest Posterior Density (HPD): 1.24 - 2.78) per month, followed by Central and Northern America (1.78 [1.04 - 2.46]), Europe (1.70 [0.868 - 2.47]), and Africa (1.16 [0.320 - 1.99]) ( Figure 2A ). Download figure Open in new tab Fig. 2. Number and diversity of H5 reassortants over time (A) Posterior distribution of the inferred true (latent) number of reassortants, stratified by continent. (B) Numbers-equivalent diversity over time, stratified by continent. Calculated as the exponent of Shannon entropy, the numbers-equivalent diversity represents the effective number of equally likely states required to generate the estimated Shannon entropy. (C) Mean pairwise genetic distance between reassortants, stratified by continent. In (B) & (C), shaded regions correspond to 95% confidence intervals We anticipated a priori that variation in sampling and virus ecology might introduce variation in the reporting of reassortment events. To address this issue, we explicitly assumed that only a proportion of observation windows could be permissive for reassortment and that detection of reassortments is incomplete. Our results indicate that the probability of conditions conducive to reassortment was lowest in Africa (0.182 [95% HPD: 0.100 - 0.280]), followed by Central and Northern America (0.462 [0.324 - 0.604]), Europe (0.477 [0.350 - 0.598]), and Asia (0.503 [0.381 - 0.617]). Conditional on reassortment having occurred, an increasing number of sequences per month was intuitively associated with a higher probability of detection. Averaged across continents, as the number of sequences per month increased fourfold from 3 to 12, the probability of detection rose from 0.855 (95% HPD:0.475 - 1.00) to 0.953 (0.654 - 1.00). However, with continuing increases in sequencing volume, the marginal gain in detection probability was diminished ( Figure S4 ). Since the rate of reassortant emergence varied through time and across continents, we sought to understand the role of viral genetic diversity in patterns of reassortment. First, we estimated the effective number of distinct reassortant profiles occurring in each continent over time using the exponent of Shannon entropy ( Figure 2B ). Under this approach, a diversity score of 7 implies a system with diversity equivalent to 7 distinct reassortant profiles of equal frequency. In Asia, we observed consistently high reassortant diversity, largely due to co-circulation of several reassortant types with balanced frequencies. In contrast, Europe exhibited lower diversity before 2023, with a single reassortant (H5N1/2021/R1 (AIV09)) dominating. By late 2023 to early 2024, following the disappearance of the H5N1/2022/R10 reassortant (AIV48), reassortant diversity in Europe increased, even though fewer sequences were detected. The increase in reassortant diversity can be attributed a more even spread of profiles among the that remained during this period, with an absence of a single dominant type. These patterns were robust to repeatedly downsampling of regions at a uniform frequency ( Figure S8 ). Second, we estimated the reassortant diversity score over time based on genetic distances between reassortant profiles in each continent ( Figure 2C ). A higher diversity score suggests (i) more distinct reassortant profiles, (ii) more balanced frequencies, and (iii) greater genetic distances between reassortants, reflecting not only the number of reassortants but also their genetic distinctiveness. In Europe and Asia, despite temporal fluctuations in reassortant diversity, the genetic distance between circulating strains remained relatively small. This is because each new reassortant that emerged differed from the major reassortant of the previous season by only a single gene segment. Consequently, despite the observed rise in profile diversity, the overall genetic distinctiveness among these reassortants remained relatively low ( Figure 2C ). In contrast, North America exhibited a sharp rise in reassortant diversity between mid-2021 and mid-2022, which correlates to the introduction of European H5Nx. Subsequent reassortment occurred between the introduced European strains and the genetically distinct low pathogenicity avian influenza virus (LPAIV) gene pools. Thereafter, the reassortant diversity score decreased as local North American strains gradually outcompeted the European-originated gene segments. 3.3 Episodic Nature of Reassortant Emergence We further explored the reassortment dynamics of clade 2.3.4.4b H5 HPAI by inferring the ancestral reassortant network relative to the HA time-scaled phylogenetic tree ( Figure 3A ). The resulting graph is highly structured and reveals an episodic pattern of reassortant emergence. Five pivotal reassortants (H5N1/2021/R1, H5N1/2022/R32, H5N1/2020/R1, H5N8/2019/R3, and H5N8/2019/R7) were the source of over half of all reassortants, each giving rise to a median of 17 (IQR: 16–28) new reassortants. Collectively, these highly connected reassortants formed a backbone that sustained the 2020–2024 panzootic across multiple seasons, facilitating global dispersal. Download figure Open in new tab Fig. 3. Episodic nature of H5 reassortant emergence A) With respect to the HA phylogeny, we plot the network of 2.3.4.4b reassortants between 2019 and 2024. Edge lengths are scaled by the number of segments changed between connected nodes and node size is scaled by out-degree (the number of immediate offspring nodes). B) For the eight reassortants with the most immediate descendants, we inferred the emergence time of their ‘offspring’ relative to the MRCA of the ‘parent’. Collectively, these inter-reassortant generation times followed a log-normal distribution with real-space mean 0.196 and standard deviation 0.923 (Log-likelihood = − 105.20), reflecting a bias towards early reassortment. Across our data, segments forming part of the viral ribonucleoprotein complex (vRNP) were exchanged with greater propensity than other segments (C). We quantified the tendency of each segment-pair (A and B) to be exchanged together using a Jaccard index (D). The colour and size of the dots plotted from the resulting symmetrical matrix are scaled by the magnitude of the index score Among all reassortants with at least one ‘offspring,’ the number of descendants was positively correlated with persistence time (Spearman’s Rank Correlation (SPC), ρ = 0.742, p ≤ 0.001), intuitively suggesting that longer circulation increases the likelihood of further reassortment. Under this scenario, one might also expect the inter-reassortant generation time to increase concomitantly with persistence time; however, in our dataset no correlation was present (SPC, ρ = 0.00, p = 0.9861) with a median inter-reassortant generation time of 3 months and 26 days ( Figure 3B ). Most reassortants arose early in the lifespan of their ‘parent’ reassortants, possibly due to initial ecological opportunities or the higher fitness of early variants, rather than accumulating through prolonged circulation. To investigate patterns of individual gene exchange, we mapped changes in genome composition from one reassortant to the next across the network. Segments comprising the viral ribonucleoprotein (vRNP) complex were exchanged most frequently ( Figure 3C ). Half of all reassortants ( n = 104) acquired a new PB2 segment relative to their immediate ancestor, followed by PB1 ( n = 86), PA ( n = 77), and NP ( n = 76), which were the second, third, and fourth most frequently exchanged segments, respectively. An exact binomial test showed that the number of reassortants where all vRNP segments were replaced (17 out of 209) was significantly greater than expected by chance ( p < 0.001, one-tailed). To identify segment pairs that frequently reassorted together, we calculated the ratio of the intersection to the union (Jaccard Index) of segment-switching events. This analysis revealed a propensity for internal segments to reassort in combination ( Figure 3D ). Notably, PA and PB2 were significantly more likely to be exchanged together than individually (permutation test, p = 0.042). We also identified a significant, though less frequent, association between the reassortment of NA and M segments ( p = 0.028). 3.4 Phylodynamic Reassortant Classes To characterise differences between reassortants, we classified each of the 209 reassortant profiles according to their phylodynamic profiles. These data included posterior estimates of evolutionary rate, host-state transition rates, diffusion coefficient, and persistence times. We identified three main clusters of reassortant, corresponding to levels of circulation across spatio-temporal, ecological, and epidemiological scales. Specifically, group A (minor) comprised 180 (86.12 %) tightly clustered reassortants profiles (median pairwise Euclidean distance = 1.98), group B (moderate) comprised 24 (11.48%) reassortants, and group C (major) comprised five (2.39%) weakly clustered reassortants (median pairwise Euclidean distance = 11.44 ( Figure S9 ). We calculated that ‘minor’ reassortants generally persisted for a median of 0.209 years (IQR: 0.0897 – 0.499), rarely switched hosts (median: 0.00; IQR: 0.00 – 0.00), and included a mammalian sample in only a small proportion of cases (0.060; 95% CI: 0.031 – 0.110). In contrast, ‘moderate’ reassortants circulated for a median 1.24 years (IQR: 0.426 - 1.94), which coincided with semi-regular host-state transitions (median 10; IQR: 0.00 - 19). Approximately half (0.541, 95% CI: 0.332 - 0.738) of ‘moderate’ reassortants included samples obtained from mammals. Finally, ‘major’ reassortants circulated the longest, with a median persistence time of 3.08 years (IQR: 2.25 - 3.78). All ‘major’ reassortants included samples obtained from mammals (95% CI: 0.462 - 1.00) and with a median of 215 host switches each (IQR: 145 - 292) (Summarised in Figure 4A ). Download figure Open in new tab Fig. 4. Reassortant Class (A) Summary of reassortant classes with defining characteristics. (B) Posterior estimates the probability that a given reassortant is of class, k, stratified by continent. The average marginal effect of the number of segments exchanged (C) and the time since the last major reassortant (D) on the reassortant class probability. (D) The ‘last’ reassortant is defined according to the ancestral state reconstruction (ie a phylogenetic order rather than chronological order) Out of the five major reassortants, three emerged in Europe (H5N1/2020/R1, H5N1/2021/R1, and H5N1/2022/R10), one in Africa (H5N8/2019/R7), and one in North America (H5N1/2022/R7). Major reassortants that emerged during the early phase of clade 2.3.4.4b resurgence (such as H5N1/2020/R1) were characterised by markedly increased persistence, creating the opportunity for extended circulation across Europe, Africa, and Asia. H5N1/2020/R1 (AB) was formed from the reassortment of H5N8/2019/R7 segments HA and MP with low pathogenicity avian influenza virus (LPAIV) H5N1 detected in Europe and Africa between 2020–2021 ( Figure S10 ). The extent of global circulation of H5N1/2020/R1 means it is the ancestral variant of almost all subsequent clade 2.3.4.4b reassortants during 2020-2024 ( Byrne et al. 2023 ). The following season, novel reassortant H5N1/2021/R1 (AIV09) was formed from the exchange of PB2 and PA from LPAIV H5N3 circulating in Northern Europe with the remaining segments of H5N1/2020/R1. H5N1/2021/R1 was the dominant reassortant throughout 2021-2022, during which time it acquired new PA, NP NS segments from LPAIV H13, typically infecting gulls, giving rise to H5N1/2022/R10. Unusually, this major reassortant circulated extensively during the summer of 2022, causing unprecedented outbreaks in Charadriiformes spp ( Fusaro et al. 2024 ). Notwithstanding the scale of the ‘gull-flu’ outbreak, H5N1/2022/R10 was the ‘major’ reassortant with the lowest number of ‘offspring’ reassortants ( n = 3), all of which were minor with very limited circulation. Separately, the extended persistence of H5N1/2020/R1 within wild birds facilitated dispersal from Europe to Northern America through pelagic migratory routes over the Atlantic Ocean. Colonisation of Northern America exposed Eurasian clade 2.3.4.4b HPAIVs to locally circulating LPAIVs, triggering a new wave of reassortment between the two ( Kandeil et al. 2023 ). The first ‘North American’ reassortant, H5N1/2022/R2, exchanged PB2 and NP segments before undergoing further reassortment exchanging PB2, PB1 and NS to result in H5N1/2022/R7 (B3.2). H5N1/2022/R7 was a major reassortant that dispersed widely across northern and South America causing repeated spillovers into terrestrial and marine mammals, finally reaching the Antarctic region by October 2023 ( Banyard et al. 2024 ). Moderate reassortants also played a pivotal role in shaping the evolution of clade 2.3.4.4b H5, albeit at a smaller scale relative to major reassortants. For example, H5N1/2021/R3 comprised 70% of genomes isolated from wild birds in China since 2021, having originated from wild birds in southern Africa in the autumn of 2020. H5N1/2021/R3 spread among wild migratory birds in China and also infected poultry and wild birds in eastern (mainly South Korea and Japan) and south-eastern Asia during the 2022/2023 season, even leading to human infections in eastern China in 2023. Similarly, H5N1/2023/R29 (B3.13) is the moderate reassortant responsible for the initial outbreak in US dairy cattle and over 20 subsequent human infections ( Caserta et al. 2024 ; Nguyen et al. 2025 ; Garg et al. 2025 ). Notably, six moderate reassortants had five or more ‘offspring’ reassortants (H5N1/2022/R32 (17), H5N8/2019/R3 (9), H5N1/2021/R4 (7), H5N1/2022/R2 (6), H5N1/2021/R3 (5) and H5N1/2023/R6 (5)), highlighting the capacity of these ‘moderately fit’ variants to further influence viral evolutionary dynamics. We analysed whether patterns of our reassortant classes varied through time and by continent. Marginalised over the empirical data distribution, we estimated that ‘major’ reassortants were proportionally most likely to have originated in Central and Northern America (with probability 0.057 [95% HPD: 0.011 - 0.121]), followed by Europe (0.026 [0.003 - 0.062]), Africa (0.014 [0.003 - 0.062]), and Asia (0.003 [0.000 - 0.012]) ( Figure 4B ). Similarly, we found that moderate reassortants were also more likely to originate from Central and Northern America (0.215 [0.120-0.314]) and Europe (0.134 [0.067 - 0.210]) than either Africa or Asia (0.014 [0.00 - 0.099] and 0.003 [0.00 - 0.012], respectively). Perhaps unsurprisingly, we found minor reassortants occurred at high frequency across all continents. For all reassortant classes, our predictions for Africa were least certain, potentially due to reduced sampling relative to other continents. Averaged across the empirical data distribution, major and moderate reassortants were most likely to follow a major reassortant with probabilities 0.035 (95% HPD: 0.008 - 0.071), and 0.143 (0.084 - 0.209), respectively. Specifically, H5N8/2019/R7 is immediately ancestral to H5N1/2020/R1, and H5N1/2020/R1 is itself immediately ancestral to H5N1/2021/R1 and H5N1/2022/R10. H5N1/2022/R7 is the only major reassortant not immediately descended from another major reassortant, instead interspersed by moderate reassortant H5N1/2022/R2. If a minor reassortant was ancestral to any reassortant at all, it was highly likely to also be minor (0.946 [0.873 - 0.995]). We also identified differences in the reassortant class-identity according to the number of new segments relative to the ‘parent’ reassortant and the time since the last major reassortant. If two segments were exchanged relative to the parent reassortant, we estimated the probability of a major reassortant to be 0.0223 (95% HPD: 0.004 - 0.047), which increases three-fold when the number of segments exchanged was raised to four (0.063 [0.016 - 0.128]) ( Figure 4C ). Marginalised over the empirical data distribution, the average affect of each additional segment increased the probability of a reassortant being ‘moderate’ or ‘major’ by 3.84 (1.21 - 6.84) and 1.67 (0.251 - 3.79) percentage points, respectively. Finally, at shorter inter-major-reassortant intervals, the probability of major and moderate reassortants was relatively high (0.0446 [0.006 - 0.109] and 0.166 [0.076 - 0.272], respectively), decreasing rapidly as the interval increased ( Figure 4D ). Conversely, the probability of a minor reassortant increased with time increasing from 0.786 (0.650 - 0.9145) at a six month interval to 0.944 (0.869-0.998) with a 3-year interval. 3.5 Estimating the diffusion velocity of reassortant spread To understand factors associated with the circulation of emerging reassortants, we inferred the dispersal history of all reassortant profiles for which we obtained two or more spatially distinct whole genomes ( Figure S3 ). Next, we fitted a generalised linear mixed model to disentangle the effects of host richness, proportion of total branch time in Anseriformes spp and Charadriiformes and persistence time on the weighted diffusion coefficients (an estimation of the area invaded per unit of time) estimated for each reassortant ( Figure 5 ). Download figure Open in new tab Fig. 5. Diffusion coefficient estimates of H5Nx reassortants stratified by time, location, and host persistence. (A) Expected values of the posterior distribution diffusion coefficient, strati-fied by continent of origin; (B) Posterior predictions of diffusion coefficient, predicted for the number of host state transitions and marginalised over the empirical data distribution; (C) Marginal mean posterior distribution of diffusion coefficient, stratified by year of reassortant MRCA. The point and bar correspond with the median value of predicted diffusion coefficients and 95% highest posterior density, respectively. Marginalising over all observations in our data, the median posterior prediction was a diffusion coefficient of 2,807.66 km 2 day -1 (95% Highest Posterior Density (HPD): 1,232.76 - 8,037.34 km 2 day -1 ). Stratified by continent ( Figure 5A ), reassortants emerging from Europe diffused fastest (3,661.08 km 2 day -1 [653.20 - 16,483.15 km 2 day -1 ]) followed by reassortants emerging from Africa (2,473.32 km 2 day -1 [573.70 - 9,724.24) and Central and Northern America (2,472.08 km 2 day -1 [1,130.59 - 4,845.96]). Reassortants that emerged from Asia were predicted to be the slowest to occupy new geographies (1,677.87 km 2 day -1 [959.59 - 2,761.41]). Stratified by the year of reassortant MRCA, reassortants that emerged in 2019 had the highest average marginal diffusion coefficient (5,304.82 km 2 day -1 [617.51 - 27,218.02]). Next, we investigated the effect of host state on the magnitude of the weighted diffusion coefficient. We estimated that the highest number of host-state transitions amongst all species per 100 whole genomes was in Southern America (2.13 [IQR:1.70 - 2.13]), followed by Central and Northern America (2.11 [1.96 - 2.21]), Africa and Europe (0.683 [0.505 - 0.761] and 0.256 [0.229 - 0.279], respectively) ( Figure S15 ). We identified a positive association between the number of host-state transitions and the diffusion coefficient, rising from 1,944.13 km 2 day -1 (95% HPD: 1,198.85 - 3,097.40) for 2 host state transitions to 4,753.95 km 2 day -1 (1,689.95 - 11,744.54) with ten host class switches ( Figure 5B ). The diffusion coefficient estimated for each reassortant also varied by bird host order. For a fixed persistence of one year, the weighted diffusion coefficient for a reassortant with 50% of circulation Charadriiformes spp was 3,398.74 km 2 day -1 (440.5 - 16,337.32), decreasing to 2413.63 km 2 day -1 (797.89 - 11,146.57) for 50% of circulation in Galliformes spp ( Figure S14 ). Averaged across our data, we estimated that a 0.1 increase in circulation of either Charadriiformes spp or Galliformes spp would result in a 96.16 km 2 day -1 (−913.31 - 1,498.42) increase or -110.49 km 2 day -1 (−680.56 - 414.28) decrease, respectively. Additionally, we calculated the effect of circulation in Anseriformes spp; the archetypical host order associated with long-range dispersal of H5 HPAIV. We estimated a diffusion coefficient of 2,650.65 km 2 day -1 (973.38 - 11,535.58), assuming 50% of circulation Anseriformes spp and a single year of persistence. For this dataset, however, the average effect of circulation in Anseriformes spp on the reassortant diffusion coefficient was variable ( Figure S14 ). 3.6 Underlying Drivers of Reassortant Spread Previous studies have suggested that wild bird migration has driven the long-distance spread, especially cross-continent dispersal, of H5Nx AIVs during the recent world-wide epizootics ( Olsen et al. 2006 ; THE GLOBAL CONSORTIUM FOR H5N8 AND RELATED INFLUENZA VIRUSES 2016 ; Yang et al. 2024 ). We further investigated the underlying environmental and anthropogenic drivers related to the spread of the five major reassortants of H5Nx AIV between regions ( Figure 6 ). Using a GLM extension of the discrete phylogeographic inference method, we identified geographic distance between regions as a negative driver for the dispersal of three major reassortants (H5N1/2020/R1, H5N8/2019/R7 and H5N1/2022/R7), suggesting small-scale circulation patterns between neighbouring regions were the driving force behind the global spread of these reassortants, rather that sudden long range migratory movements. We identified land-based poultry production was as a driver for virus spread for the same major reassortants (H5N1/2020/R1, H5N8/2019/R7 and H5N1/2022/R7), whereas the land-based poultry trade was supported as a positive driver for the interregional dispersal of reassortants, H5N1/2021/R1, H5N1/2022/R10, and H5N1/2022/R7. In addition, the integrated free-range duck farming style was negatively correlated with the spread of H5N8/2019/R7 reassortant, emphasising a role for farming practices and poultry production to sustain the circulation of H5N1 in the current panzootic. Download figure Open in new tab Fig. 6. Contributions of predictors for five dominant reassortants of H5Nx AIV spread between regions. The virus dispersal is inferred using HA genes by GLM-extended Bayesian discrete phylogeographic inference. The coefficients (left panel) represent the mean size estimate (on a log scale) of the contribution of the predictors with credible intervals. The indicators (right panel) represent the estimated inclusion probability of the predictors. We also investigated the underlying factors related to the spread of clade 2.3.4.4b H5 HPAIV reassortants between regions within each continent ( Figure S19 ). In Asia, the geographic distance between regions was identified as a negative predictor for virus diffusion frequency between regions, meaining clade 2.3.4.4b H5 HPAIV are more likely to spread between two neighbouring regions. In North America, live land-based poultry trade and geographic distance were supported as two factors related to H5Nx spread. In Europe, except for the sample sizes, the live land-based poultry trade was suggested as a positive driver for H5 spread, although the coefficient for this predictor is relatively small. In Africa, the live poultry trade was identified as a positive predictor to virus spread in the GLM analysis. Indeed, the live poultry trade was not frequent between African countries, and this predictor could potentially have been identified by the GLM analysis due to poultry trade from Europe to Africa. Although we included the coastline/area ratio to approximate the role of seabirds in facilitating H5 spread, it was not identified as a supportive predictor in the GLMs for any continents. We suggest this arises from the high-level spatial scale (within the continent) for these analyses. 4 Discussion The resurgence of clade 2.3.4.4b coincided with step-changes in the epidemiological dynamics of H5 HPAIV. At first, novel reassortant H5N1/2020/R1 not only out-competed contemporary HPAIV H5N8 lineages, but triggered the start of a global epizootic and ecological crisis in birds and marine mammals. Pervasive reassortment has facilitated novel genetic combinations, concurrent with an expansion of host range, long-term persistence and worldwide dispersal ( Xie et al. 2023 ). In this study, we systematically characterised global patterns of reassortment, identifying 209 unique H5Nx reassortants circulating during the 2020-2024 panzootic. We estimated that a greater number of reassortants originated from Asia than any other continent, and revealed a structured, intermittent pattern of reassortant emergence dominated by five ‘major’ reassortants. We estimated that reassortants from Europe spread most quickly, and reconfirm a critical role for wild and domestic birds in the H5 HPAIV dispersal and persistence. Specifically, our results highlight an emerging role for Charadriiformes spp. in increasing the diffusion coefficient of newly emerged reassortants, alongside well-established roles of Anseriformes spp. Five out of 209 reassortants were classified as major. Of these, three emerged in Europe (H5N1/2020/R1, H5N1/2021/R1, and H5N1/2022/R10), one in Africa (H5N8/2019/R7), and one in North America (H5N1/2022/R7). Major reassortants that emerged during the early phase of clade 2.3.4.4b H5Nx virus resurgence were characterised by markedly increased persistence, creating the opportunity for extended circulation across Europe, Africa, and Asia. Crucially, persistence within wild birds facilitated dispersal from Europe to Northern America through pelagic migratory routes over the Atlantic Ocean ( Caliendo et al. 2022 ). Colonisation of Northern America exposed Eurasian 2.3.4.4b H5 HPAIVs to locally circulating LPAIVs, triggering a new wave of reassortment between the two ( Kandeil et al. 2023 ; Signore et al. 2025 ). Our analyses indicated the advantage of long-term persistence in Charadriiformes spp, which accelerated the invasion rate (diffusion coefficient) of newly emerged reassortants. In previous H5Nx outbreaks, Anseriformes spp have been widely linked to extending persistence and geospatial range ( Trovão et al. 2015 ); however, in North America between 2008 and 2018, Charadriiformes spp were also associated with faster migration of H5Nx, with dispersal rates greater than either Anseriformes spp or domestic Galliformes spp by 1650 km year −1 and 4496 km year −1 , respectively ( Hill et al. 2022 ). Our findings support the notion that Charadriiformes spp may act as key facilitators of the widespread viral dispersal, probably through long-distance pelagic routes. Combined with our findings that inter-regional distance was negatively correlated with reassortant dispersal, while the number of host jumps was positively correlated with geographic area expansion, we anticipate a key role of Charadriiformes spp colonies and stopover sites in the maintenance of clade 2.3.4.4b H5N1 circulation ( Hicks et al. 2022 ). The contribution of domestic birds in the spread of clade 2.3.4.4b H5 HPAIV reassortants, however, is less clear. In our study, the rate of reassortant invasion showed a weak negative association with the proportion of persistence in domestic Galliformes spp, whereas land-based poultry production and trade were identified as drivers of interregional H5 HPAIV dispersal. Intensive poultry farming has long since been considered a risk factor driving H5 HPAIV spread, particularly in eastern and southeastern Asia where an integrated rice-duck farming system is commonplace ( Li et al. 2014 ; Chen and Bu 2009 ). Interfaces between domestic ducks and wild birds, in addition to interfaces between different classes of wild birds, can increase potential for sustained transmission and reassortment ( Hicks et al. 2022 ; Kwon et al. 2020 ; Barman et al. 2017 ), yet a mechanistic role for Galliformes spp in clade 2.3.4.4b H5 HPAIV transmission remains undetermined. Clade 2.3.4.4b H5 HPAIVs, such as major reassortant H5N1/2020/R1, have been revealed to have strong host preferences for wild Anseriformes and comparatively poor infectiousness or transmissibility in Galliformes spp ( James et al. 2023 ). Nonetheless, previous analyses of host dynamics and continuous spatial diffusion ( Fusaro et al. 2019 ; Awada et al. 2025 ) are consistent with our conclusions that poultry trade as well as wild bird migration has contributed to viral spread. We also rarely observed reassortment between poultry-adapted AIV subtypes and 2.3.3.4.b H5 HPAIVs. Previously, in situ measurements of reassortment between H5N1 and poultry-specific AIVs highlighted relatively low rates of reassortment, not explained by high mortality of H5 HPAIVs ( Lu et al. 2014 ). In China, poultry-specific BJ/94 lineage H9N2 virus has sustained a dominant presence among poultry since 2016 and previously reassorted with H7N9 and H3N8 to give rise to humaninfective AIV lineages that are a concern for human public health ( Zhou et al. 2024 ; Sun et al. 2023 ; Lam et al. 2015 ). Notwithstanding the endemicity of H9N2 across Asia, Europe and the Middle East, as well as the role of intensive poultry farming in driving HPAIV spread across eastern and southeastern Asia, few reassortants involving H9N2 and panzootic H5 HPAIVs have been reported ( Fusaro et al. 2011 ; Chen and Bu 2009 ; Gilbert et al. 2014 ; Barman et al. 2025 ; El-Shesheny et al. 2025 ). Currently, the reasons for this pattern remain unclear. Previous lineages of H5N1 (A/chicken/Egypt/CL69/2013) were found to be compatible in vitro with H9N2 for reassortment ( Arai et al. 2019 ), and short-lived H5NX-H9N2 reassortants have been briefly detected in wild birds in Bangladesh and China prior to the resurgence of clade 2.3.4.4b ( Barman et al. 2019 ; Aji et al. 2021 ). Ancestral clade 2.3.4.4b H5Nx (A/tufted duck/Germany/8444/2016) was also shown to reassort with H9N2 in vitro ; however, compatibility was limited to the integration of the H9N2 PB2, NA, NS segments and coincided with a reduction in between-poultry transmissibility ( Mostafa et al. 2020 ). Whether a H5N1-H9N2 reassortant can competently sustain transmission between poultry and/or wild birds will remain a critical area of interest. If a trade-off between wild-bird adapted (such as H5Nx) and poultry-adapted (such as H9N2) lineages exists, not only is the probability of acquiring compatible segments likely to be restricted, but initial descendants are also likely to suffer a fitness disadvantage relative to ‘pure’ poultry-adapted subtypes, attenuating transmission potential. Across our data, we observed a repeated episodic pattern of reassortant emergence in which a transient period of elevated reassortment occurs early in the lifecycle of certain reassortants. Localised HPAIV outbreaks are typically preceded by an increase in LPAIV circulation which then declines upon the introduction of HPAIV ( Tuncer et al. 2016 ). LPAIV and HPAIV can be said to compete for susceptible birds, meaning the invasion of HPAIV with a relative transmission advantage will deplete the susceptible pool for both LPAIV and HPAIV due to cross-protection or death. Therefore, there is likely only a brief window during the early phase of the HPAIV epidemic, in which both LPAIV and HPAIV co-circulate in the same population at levels where the probability of co-infection is non-negligible. The extent to which LPAIV and HPAIV can coexist for evenly distributed reassortment depends on their relative fitness advantages, immune cross-protection and migration patterns ( Saucedo and Martcheva 2017 ). If previous infection by LPAIV provides partial protection, subsequent HPAIV mortality may be reduced and crucially the infectious period of HPAIV will be extended relative to naive infection ( Bourouiba et al. 2011 ; Nick-bakhsh et al. 2016 ). In our data, major reassortant H5N1/2022/R10 (also known as BB) was uniquely concentrated in Charadriiformes spp, and had very low number of ‘offspring’. In contrast to other wild bird populations, prior to 2022 Charadri-iformes spp in Europe were mostly immunologically naive to H5; ‘gull-like’ LPAIV subtypes provide little cross-protection with H5 HPAIV ( Hill et al. 2022 ). Reduced cross-protection would be expected to increase the relative transmission advantage of HPAIV, reducing the opportunity for reassortment relative to other epidemic systems. Likewise, we estimated the highest number of reassortants emerged from Asia, where high-level circulation of LPAIV in Anseriformes spp is widely thought to contribute to the ‘cryptic’ spread of HPAIV ( Fereidouni et al. 2009 ). Collectively, these findings reveal the intertwined dynamics of transmission, immunology and reassortment that determine the trajectory of the 2020-2024 H5N1 panzootic. Throughout this study, we must be cognizant to the impact of sampling bias. Regional variation in sentinel avian influenza surveillance efforts may mean some reassortment events in less monitored areas may be under-represented. Reactive sequencing efforts, stood-up across the globe in response to the escalating panzootic, likely exacerbated the spatio-temporal covariance of these sampling regimes. This can present challenges when assigning reassortment clusters. For example, a poorly sampled viral lineage could be misclassified as a novel reassortant simply because it appears to be relatively distant from its neighbours due to the absence of unsampled evolutionary intermediates. Similarly, analyses of host class switching will be responsive to both sampling intensity and specificity, which likely varies between countries and/or regions. In describing the high-level global reassortant dynamics, we may also not capture fine-scale patterns in reassortant emergence and migration. For example, low-level heteroscedasticity in viral migration may inflate the diffusion coefficients estimated for each reassortant ( Layan et al. 2023 ). Likewise, we also recommend a conservative interpretation of the continent-specific detection probabilities in our ‘number of reassortants’ model which were only weakly identifiable ( Kéry 2018 ). In summary, this study provides valuable insights into the mechanisms driving reassortment success and interspecies transmission of AIVs; providing a comprehensive analysis of worldwide patterns of reassortants in the clade 2.3.4.4b H5 HPAIV panzootic 2020-2024. Intermittent reassortment was a key evolutionary mechanism that fuelled the four-year panzootic of clade 2.3.4.4b H5N1 HPAIVs, which was predominantly driven by five major reassortants responsible for intercontinental virus dispersal and the derivation of almost all novel reassortant types. Enhanced AIV surveillance is needed for the early identification of the novel and major reassortants, leveraging known phenotypic-related mutations and lab-based evidence to evaluate risk potential. As clade 2.3.4.4b H5N1 continues to persist globally, ongoing vigilance will be required to mitigate the risk to both wild birds and domestic poultry. Alongside strengthened biosafety management and animal husbandry practices in the poultry industry, intensified collaboration and data sharing within the global AIV surveillance and monitoring network is essential. Expanding surveillance efforts and promoting a more balanced approach to data collection between regions will enhance our understanding of global avian influenza dynamics. Improved access to genomic and epidemiological data will lead future efforts to predict and control future out-breaks, including the rational design of vaccines. 6 Data Sharing Code pertaining to data extraction, reassortment clustering and statistical models fitting will be made available in due course. 7 Funding This work was supported by an Ecology and Evolution of Infectious Diseases collab-orative grant with LL, SL and PD funded by the UK Biotechnology and Biological Sciences Research Council (BBSRC, UK) (grant no. BB/V011286/1); and JY, WJL and YHB funded by the National Natural Science Foundation of China (NSFC, China) (grant no. 32061123001). JY and YHB received additional support from the NSFC (grant nos. 32425053 and 32200416) and the National Key R&D Program of China (grant no. 2023YFC2307500). PD, SL and WH acknowledge support from a UK research consortium on avian influenza research gaps, funded by the BBSRC, Medical Research Council (MRC, UK), and Department for Environment Food and Rural Affairs (DEFRA, UK) as FluMAP’ (grant nos. BB/X006204/1, BB/X006166/1), ‘FluTrailMap’ (grant nos. BB/Y007271/1, BB/Y007298/1) and FluTrailMap-One Health (grant no. MR/Y03368X/1). PD was also supported by a European Union (EU) Horizon 2020 award (grant agreement no. 727922 [DELTA-FLU]). JB, LL and SL were supported by an EU Horizon 2020 award (grant agreement no. 874735 [VEO]) and JB, LL, PD, SL and WH were additionally supported by BBSRC Institute Strategic Grants to the Roslin Institute (grant nos. BBS/E/RL/230002C and BBS/E/RL/230002D). SD acknowledges support from the Fonds National de la Recherche Scientifique (F.R.S.-FNRS, Belgium; grant no. F.4515.22), from the Research Foundation — Flanders ( Fonds voor Wetenschappelijk Onderzoek — Vlaan-deren , FWO, Belgium; grant no. G098321N), and from EU Horizon 2020 awards (grant agreement no. 874850 [MOOD], and grant agreement no. 101094685 [LEAPS]). MB acknowledges funding from an EU Horizon 2020 award (grant agreement No. 101084171 [KAPPA-FLU]). Views and opinions expressed are however those of the author(s) only and do not necessarily reflect those of the European Union. Neither the European Union nor the granting authority can be held responsible for them. 8 Competing Interests The authors declare no competing interests. 5 Acknowledgements This work has made use of the resources provided by the Edinburgh Compute and Data Facility (ECDF) ( http://www.ecdf.ed.ac.uk/ ). We gratefully acknowledge all data contributors, i.e., the Authors and their Originating laboratories responsible for obtaining the specimens, and their Submitting laboratories for generating the genetic sequence and metadata and sharing via the GISAID Initiative, on which this research is based. We thank PhD students, Kunpeng Yuan and Lixia Wang; and graduate students, Haoyu Wen, Dongjia, and Mingjia Wu, for their contribution to data curation. Appendix A Supplementary Methods A.1 Predicators related to H5 virus spread We used GLM analyses as an extension of discrete phylogeographic inference to understand the underlying environmental factors of H5 virus spread. According to previous studies, we chosen live poultry trade, poultry production, geographic distance, the integrated rice-animal farming style, the coastline/area ratio, and sample size as the predictors. The live poultry trade and poultry production in each country during 2019–2022 are obtained from the FAOSTAT database ( https://www.fao.org/faostat/ ). The annually import quantity and export quantity between countries for the land-based poultry (chicken and turkey) and waterfowl poultry (duck and goose) are included, and the poultry trade are calculated from the average value between import quantity and export quantity between countries. The geographic distance between countries was calculated as the great-circle distance given latitudes and longitudes of their centroids. The coastline/area ratio ( https://wikimili.com/en/List of countries by length of coastline) was chosen to show the coastline length due to seabirds considered as a host facilitating the H5 spread. The integrated rice-animal farming style ( Li et al. 2023 ) was selected as a 0 or 1 variable to indicate the potential interaction between domestic and wild waterfowl. Before the GLM analyses, we first tested the multicollinearity between the variables and included the independent variables in the regression analysis. Except for the 0 or 1 variable, we standardized the variables with mean value of 0 and standard deviation of 1 to avoid the impact of the different order of magnitude of the predictor variables. A.2 Number of Reassortants For each year-month observation, i ∈ {1, 2, …, I }, taken in continent, j ∈ {africa, asia, americas, europe}, let y ij ∈ ℤ ≥0 be the observed number of reassortants. We assume y ij can be modelled as a mixture of three components: a detection model, an abundance model, and a zero-inflation model. A.2.1 Detection Model First, we consider that only a proportion, p ij ∈ (0, 1), of true (latent) reassortants, , are ultimately observed: where , is the continent-stratified proportion of reassortants detected and , is the log-scale quantity of HPAIV full genomes present on GISAID. We assume that the interval between sequence collection and the most recent common ancestor of each reassortant is of sufficiently short duration that no lag is required to be accounted for. We specify the following weakly informative prior distributions for the detection model: A.2.2 Abundance Model Second, we model the true number of reassortants, , as a discrete latent variable that follows a Poisson distribution: where λ ij is the expected number of reassortants per observation on the log scale, , is the continent-stratified baseline abundance, , is the log-scale HPAIV incidence estimate, and and are the zero-centred random intercepts and standard deviation of calendar year. We specify the following weakly informative prior and hyperprior distributions for the abundance model: A.2.3 Zero-Inflation Model Third, we consider that ecological or epidemiological conditions may not always be conducive for reassortment/reassortant emergence. We assume this process is fundamentally distinct from a structural absence of reassortment (i.e situations where reassortment/reassortant emergence is feasible but does not occur). We model a Bernoulli zero-inflation component, z ij ∈ {0, 1}, parametrised by a continent-specific probability that a conditions are not permissive for reassortment/reassortant emergence, θ i : We specify the following weakly informative prior distributions for the zero-inflation model: A.2.4 Joint Probability We combined the abundance, detection and zero inflation components to calculate the joint probability that y reassortants are observed each month i per continent j s: where, We fit this model using probabilistic programming language Stan ( Carpenter et al. 2017 ). Since the Hamiltonian Monte Carlo sampler used by Stan cannot handle discrete latent variables, we integrate the joint probability over plausible values of N ij ∈ { y i , y i + 1, …, y + K − 1} where K = 12 ( Dennis et al. 2015 ): Essentially, our model describes a thinned zero-inflated Poisson distributed variable. To avoid the necessary computational approximations, one could alternatively calculate the exact probability, resulting in a Poisson-distributed variable, scaled by p ij . We chose our explicit definition to clearly separate the ecological and anthropological determinants of the true and observed number of reassortants, respectively. A.3 Reassortant Class Let C = {minor, moderate, major} be the ordered set of all possible reassortant classes. Each reassortant, i is assigned a class y i = c ∈ C . To calculate the probability that a reassortant is assigned a particular class, we assume y i follows a cumulative distribution in which minor < moderate < major ( Samejima 1997 ). We assume y i originates from the categorisation of a latent (unobserved) continuous variable, , via thresh-old parameters, τ , that partition the latent scale into intervals corresponding to each class. Specifically, we define τ minor−1 = −∞ and τ moderate+1 = ∞: We modelled the latent variable as the sum of four linear predictors, η i , and random error, ϵ i : i) the class identity of the reassortant immediately ancestral (with respect to HA) to reassortant i , ii) origin continent, iii) the number of segments changed relative to the immediately ancestral reassortant, and iv) the time interval between the MRCA of reassortant i and that for the most recent major reassortant. In addition, we modelled a penalised thin plate regression spline to smooth the time interval between the MRCA of reassortant i and that for the most recent major reassortant ( Wood 2003 ). Briefly, for each knot, κ k , where k = 1, 2, …, K , we computed the distance, , evaluated the radial basis function, φ ( r ) = r 2 log r , and scaled the result by γ k . Altogether, we estimated η i as: Let Φ be the cumulative distribution function of the Normal distribution. The probability that reassortant, i , is of class, c , is therefore given by: We specify the following weakly informative prior and hyperprior distributions for the reassortant class model: A.4 Diffusion Coefficient For each reassortant, i , we estimated a weighted diffusion coefficient, y . We assume all y i > 0 follow a Gamma distribution with shape, κ i , and rate, θ i , such that We model µ i as the exponent of linear predictors for continent, ; host richness, x cross-species i ; the proportion of evolutionary time in Anseriformes spp, , Charadriiformes spp, , and domestic Galliformes spp, ; and total persistence time, . We allowed persistence time and the proportion of time in Anseriformes spp to vary by continent (i.e continent is the grouping factor for multivariate random effects, with a non-centred parametrisation, γ ). We also included non-centred random intercepts for the date of the reassortant MRCA, grouped by calendar year, δ year[ i ] : The group-level effects, γ , are correlated across the six continent-level slopes. To estimate γ , we multiplied the vectors of group-level standard deviations, σ continent , standard normal latent variables, z 1 , and the Cholesky factor of the group-level correlation matrix, L 1 : We assume shape parameter, κ i , is also predicted as the exponent of continent, indexed as ρ continent[ i ] : Finally, we specify the following prior and hyperprior distributions for the diffusion coefficient model: Appendix B Supplementary Figures Download figure Open in new tab Fig. S1. Highly Pathogenic H5 reports from World Organisation for Animal Health. We downloaded all biannual reports concerning highly pathogenic H5 avian influenza between January 2019 and July 2024. Stratified by reporting continent, we compare the number of reported cases, deaths and susceptibles (both wild and domestic), relative to the number of GISAID whole genome sequences. In our statistical models, we selected the number of susceptibles to represent H5 HPAIV incidence due to inconsistencies in deaths and cases (e.g deaths sometimes exceed the number of cases). Bars correspond to the number of whole genomes and their sampling date Download figure Open in new tab Fig. S2. Epidemic transitions throughout 2020-2024 2.3.4.4b Panzootic Our data encompasses key changes in the epidemiology of H5 highly pathogenic avian influenza virus. In particular, these data include distinct changes in subtype (A) and variation in host composition (B) as the panzootic progressed Download figure Open in new tab Fig. S3. Continuous Phylogeography for Major Reassortants For each reassortant, we inferred global circulation patterns throughout time. Solid dots are the location and time of tip sequences, while hollow dots correspond to the estimated location of intermediate nodes. Shaded polygons show the 95% highest posterior density estimated for each internal node. Download figure Open in new tab Fig. S4. Conditional effect of the number of sequences per month on the probability that a true reassortant is detected Download figure Open in new tab Fig. S5. Prior (orange) and posterior distributions (green) for select parameters of the ‘numbers of reassortant’ model. In most cases the two distributions show minimal overlap, which indicates model ‘learning’ from the data. For ρ parameters, there is a close overlap between the posterior and prior distributions, indicating these parameters are only weakly identifiable Download figure Open in new tab Fig. S6. Posterior predictive check for the ‘numbers of reassortant’ model. The underlying bar plot represents the empirical data distribution. Each point corresponds to the median estimate of the posterior predictive distribution, with the associated error bars displaying the 95% Highest Posterior Densities (HPD). Download figure Open in new tab Fig. S7. Quantile-Quantile residual diagnostic plot for the numbers model. Scaled residuals were calculated using DHARMa ( Hartig 2024 ) Download figure Open in new tab Fig. S8. Downsampling of Numbers-Equivalent Shanon Entropy. In our dataset, Africa was the least frequently sampled continent. We repeatedly resampled all other continents by the number of whole genomes sampled from Africa, to evaluate the susceptibility of numbers-equivalent Shanon entropy to sampling bias. Across 100 resamples, the patterns within each continent remained broadly consistent Download figure Open in new tab Fig. S9. K-Means Clustering of Reassortant Phylodynamic Profiles. A) Selection of k = 3 as optimal number of clusters for further analysis. B) Two dimensional representation of our phylo-dynamic profiles, where each point represents a single reassortant and is coloured by assigned cluster. C) Relative influence of phylodynamic variables on cluster allocation. We iteratively exclude one variable and calculate the similarity in clustering patterns between the original and iterative result using the Adjusted Rand Index (ARI). We define variable importance as 1-ARI. Download figure Open in new tab Fig. S10. Progression of Major reassortants across the 2020-2024 2.3.4.4b H5 panzootic Download figure Open in new tab Fig. S11. Prior (orange) and posterior distributions (green) for select parameters of the reassortant class model. In all cases the two distributions show minimal overlap, which suggests the posterior distributions are well-informed by the model likelihood. Download figure Open in new tab Fig. S12. Posterior predictive check for reassortant class model. Stratified by continent, each underlying bar plot represents the empirical data distribution. Each point corresponds to the median estimate of the posterior predictive distribution, with the associated error bars displaying the 95% Highest Posterior Densities (HPD). Download figure Open in new tab Fig. S13. Quantile-Quantile residual diagnostic plot for the reassortant class model. Scaled residuals were calculated using DHARMa ( Hartig 2024 ) Download figure Open in new tab Fig. S14. Average marginal effect of the proportion of total branch-time in which the ancestral host was estimated to be a) Anseriformes spp, (B) Charadriiformes spp or (C) domestic Galliformes, stratified by origin continent and marginalised over the empirical data distribution. The black line is the median of the posterior distribution and shaded regions (lightest to darkest) correspond to 95%, 90% and 50% highest posterior density regions. spp. Download figure Open in new tab Fig. S15. Scaled frequencies of host class transitions for H5 reassortants, stratified by continent. Each plot shows the number of observed host-state transitions for each reassortant, scaled by (A) the total number of sequences and (B) the number of sequences from mammals only. Since the exact number of state-transitions can vary according to each segment, we report the median number of state-transitions for each reassortant and interquartile ranges. Download figure Open in new tab Fig. S16. Prior (orange) and posterior distributions (green) for select parameters of the diffusion coefficient model. In all cases the two distributions show minimal overlap, which suggests the posterior distributions are well-informed by the model likelihood. Download figure Open in new tab Fig. S17. Posterior predictive check for the diffusion coefficient model. Stratified by continent, each thick purple line represents the empirical data distribution. Each thine trace line corresponds to a posterior predictive density, extracted from a single draw. Download figure Open in new tab Fig. S18. Quantile-Quantile residual diagnostic plot for the diffusion coefficient model. Scaled residuals were calculated using DHARMa ( Hartig 2024 ) Download figure Open in new tab Fig. S19. Contributions of predictors for H5Nx AIV spread between regions within each continent. The virus dispersal is inferred using HA genes by GLM-extended Bayesian discrete phylogeographic inference. The coefficients (left panel) represent the mean size estimate (on a log scale) of the contribution of the predictors with credible intervals. The indicators (right panel) represent the estimated inclusion probability of the predictors. Funder Information Declared Biotechnology and Biological Sciences Research Council , BB/V011286/1 , BB/X006204/1 , BB/X006166/1 , BB/Y007271/1 , BB/Y007298/1 Biotechnology and Biological Sciences Research Council - Institute Strategic Grants , BBS/E/RL/230002C , BBS/E/RL/230002D Medical Research Council , MR/Y03368X/1 National Natural Science Foundation of China, https://ror.org/01h0zpd94 , 32061123001 , 32425053 , 32200416 National Key Research and Development Program of China , 2023YFC2307500 European Union , 727922 , 874850 , 101094685 , 101084171 , 874735 Fonds National de la Recherche Scientifique , F.4515.22 Fonds voor Wetenschappelijk Onderzoek — Vlaanderen , G098321N References ↵ Agüero M , Monne I , Sánchez A , et al. ( 2023 ) Highly pathogenic avian influenza A(H5N1) virus infection in farmed minks, Spain, October 2022 . Eurosurveillance 28 ( 3 ): 2300001 . 1560-7917.ES.2023.28.3.2300001, URL https://www.eurosurveillance.org/content/10.2807/1560-7917.ES.2023.28.3.2300001 , publisher: European Centre for Disease Prevention and Control OpenUrl PubMed ↵ Aji D , Chang N , Zhang C , et al. ( 2021 ) Rapid Emergence of the Reassortant 2.3.4.4b H5N2 Highly Pathogenic Avian Influenza Viruses in a Live Poultry Market in Xinjiang, Northwest China . Avian Diseases 65 ( 4 ): 578 – 583 . doi: 10.1637/aviandiseases-D-21-00075 , URL https://bioone.org/journals/avian-diseases/volume-65/issue-4/aviandiseases-D-21-00075/Rapid-Emergence-of-the-Reassortant-2344b-H5N2-Highly-Pathogenic-Avian/10.1637/aviandiseases-D-21-00075.full , publisher: American Association of Avian Pathologists OpenUrl CrossRef PubMed ↵ Altschul SF , Madden TL , Schäffer AA , et al. ( 1997 ) Gapped BLAST and PSI-BLAST: a new generation of protein database search programs . Nucleic Acids Research 25 ( 17 ): 3389 – 3402 . doi: 10.1093/nar/25.17.3389 , URL https://doi.org/10.1093/nar/25.17.3389 OpenUrl CrossRef PubMed Web of Science ↵ Arai Y , Ibrahim MS , Elgendy EM , et al. ( 2019 ) Genetic Compatibility of Reassortants between Avian H5N1 and H9N2 Influenza Viruses with Higher Pathogenicity in Mammals . Journal of Virology 93 ( 4 ): 10 .1128/jvi.01969–18. doi: 10.1128/jvi.01969-18 , URL https://journals.asm.org/doi/10.1128/jvi.01969-18 , publisher: American Society for Microbiology OpenUrl CrossRef ↵ Arel-Bundock V , Greifer N , Heiss A ( 2024 ) How to Interpret Statistical Models Using marginaleffects for R and Python . Journal of Statistical Software 111 : 1 – 32 . https://doi.org/10.18637/jss.v111.i09 , xURL doi: 10.18637/jss.v111.i09 OpenUrl CrossRef ↵ Awada L , Vrancken B , Thézé J , et al. ( 2025 ) Quantifying Time-Dependent Predictors for the International Spatial Spread of Highly Pathogenic Avian Influenza H5NX: Focus on Trade and Surveillance Efforts . Transboundary and Emerging Diseases 2025 ( 1 ): 2020766 . doi: 10.1155/tbed/2020766 , URL https://onlinelibrary.wiley.com/doi/abs/10.1155/tbed/2020766 , xeprint: https://onlinelibrary.wiley.com/doi/pdf/10.1155/tbed/2020766 OpenUrl CrossRef ↵ Ayres DL , Cummings MP , Baele G , et al. ( 2019 ) BEAGLE 3: Improved Performance, Scaling, and Usability for a High-Performance Computing Library for Statistical Phylogenetics . Systematic Biology 68 ( 6 ): 1052 – 1061 . doi: 10.1093/sysbio/syz020 , URL https://doi.org/10.1093/sysbio/syz020 OpenUrl CrossRef PubMed ↵ Banyard AC , Bennison A , Byrne AMP , et al. ( 2024 ) Detection and spread of high pathogenicity avian influenza virus H5N1 in the Antarctic Region . Nature Communications 15 ( 1 ): 7433 . doi: 10.1038/s41467-024-51490-8 , URL https://www.nature.com/articles/s41467-024-51490-8 , publisher: Nature Publishing Group OpenUrl CrossRef PubMed ↵ Barman S , Marinova-Petkova A , Hasan MK , et al. ( 2017 ) Role of domestic ducks in the emergence of a new genotype of highly pathogenic H5N1 avian influenza A viruses in Bangladesh . Emerging Microbes & Infections 6 ( 1 ): 1 – 13 . doi: 10.1038/emi.2017.60 , URL https://doi.org/10.1038/emi.2017.60 , publisher: Taylor & Francis eprint: https://doi.org/10.1038/emi.2017.60 OpenUrl CrossRef ↵ Barman S , TJasmine C. M. , HM. Kamrul , et al. ( 2019 ) Continuing evolution of highly pathogenic H5N1 viruses in Bangladeshi live poultry markets . Emerging Microbes & Infections 8 ( 1 ): 650 – 661 . doi: 10.1080/22221751.2019.1605845 , URL https://doi.org/10.1080/22221751.2019.1605845 , publisher: Taylor & Francis eprint: https://doi.org/10.1080/22221751.2019.1605845 OpenUrl CrossRef PubMed ↵ Barman S , TJasmine C. M. , HM. Kamrul , et al. ( 2025 ) Reassortment of newly emergent clade 2.3.4.4b A(H5N1) highly pathogenic avian influenza A viruses in Bangladesh . Emerging Microbes & Infections 14 ( 1 ): 2432351 . doi: 10.1080/22221751.2024.2432351 , URL https://doi.org/10.1080/22221751.2024.2432351 , publisher: Taylor & Francis eprint: https://doi.org/10.1080/22221751.2024.2432351 OpenUrl CrossRef PubMed ↵ Bourouiba L , Teslya A , Wu J ( 2011 ) Highly pathogenic avian influenza outbreak mitigated by seasonal low pathogenic strains: Insights from dynamic modeling . Journal of Theoretical Biology 271 ( 1 ): 181 – 201 . doi: 10.1016/j.jtbi.2010.11.013 , URL https://www.sciencedirect.com/science/article/pii/S0022519310005989 OpenUrl CrossRef PubMed ↵ Byrne AMP , James J , Mollett BC , et al. ( 2023 ) Investigating the Genetic Diversity of H5 Avian Influenza Viruses in the United Kingdom from 2020–2022 . Microbiology Spectrum 11 ( 4 ): e04776 – 22 . doi: 10.1128/spectrum.04776-22 , URL https://journals.asm.org/doi/full/10.1128/spectrum.04776-22 , publisher: American Society for Microbiology OpenUrl CrossRef ↵ Bürkner PC ( 2017 ) brms: An R Package for Bayesian Multilevel Models Using Stan . Journal of Statistical Software 80 : 1 – 28 . doi: 10.18637/jss.v080.i01 , URL https://doi.org/10.18637/jss.v080.i01 OpenUrl CrossRef PubMed ↵ Caliendo V , Lewis NS , Pohlmann A , et al. ( 2022 ) Transatlantic spread of highly pathogenic avian influenza H5N1 by wild birds from Europe to North America in 2021 . Scientific Reports 12 ( 1 ): 11729 . doi: 10.1038/s41598-022-13447-z , URL https://www.nature.com/articles/s41598-022-13447-z , publisher: Nature Publishing Group OpenUrl CrossRef PubMed ↵ Carpenter B , Gelman A , Hoffman MD , et al. ( 2017 ) Stan: A Probabilistic Programming Language . Journal of Statistical Software 76 : 1 – 32 . doi: 10.18637/jss.v076.i01 , URL https://doi.org/10.18637/jss.v076.i01 OpenUrl CrossRef PubMed ↵ Caserta LC , Frye EA , Butt SL , et al. ( 2024 ) Spillover of highly pathogenic avian influenza H5N1 virus to dairy cattle . Nature 634 ( 8034 ): 669 – 676 . doi: 10.1038/s41586-024-07849-4 , URL https://www.nature.com/articles/ s41586-024-07849-4, publisher: Nature Publishing Group OpenUrl CrossRef ↵ Chen H , Bu Z ( 2009 ) Development and application of avian influenza vaccines in China . Current Topics in Microbiology and Immunology 333 : 153 – 162 . doi: 10.1007/978-3-540-92165-37 OpenUrl CrossRef PubMed ↵ Chen H , Smith GJD , Li KS , et al. ( 2006 ) Establishment of multiple sublineages of H5N1 influenza virus in Asia: Implications for pandemic control . Proceedings of the National Academy of Sciences 103 ( 8 ): 2845 – 2850 . doi: 10.1073/pnas.0511120103 , URL https://www.pnas.org/doi/10.1073/pnas.0511120103 , publisher: Proceedings of the National Academy of Sciences OpenUrl Abstract / FREE Full Text ↵ Dellicour S , Rose R , Faria NR , et al. ( 2016 ) SERAPHIM: studying environmental rasters and phylogenetically informed movements . Bioinformatics 32 ( 20 ): 3204 – 3206 . URL doi: 10.1093/bioinformatics/btw384 OpenUrl CrossRef PubMed ↵ Dellicour S , Bastide P , Rocu P , et al. ( 2024 ) How fast are viruses spreading in the wild? PLOS Biology 22 ( 12 ): e3002914 . doi: 10.1371/journal.pbio.3002914 , URL https://journals.plos.org/plosbiology/article?id=10.1371/journal . pbio.3002914, publisher: Public Library of Science OpenUrl CrossRef PubMed ↵ Dennis EB , Morgan BJ , Ridout MS ( 2015 ) Computational Aspects of N-Mixture Models . Biometrics 71 ( 1 ): 237 – 246 . doi: 10.1111/biom.12246 , URL https://doi.org/10.1111/biom.12246 OpenUrl CrossRef PubMed ↵ Drummond AJ , Ho SYW , Phillips MJ , et al. ( 2006 ) Relaxed Phylogenetics and Dating with Confidence . PLOS Biology 4 ( 5 ): e88 . URL doi: 10.1371/journal.pbio.0040088 , publisher: Public Library of Science OpenUrl CrossRef PubMed ↵ Dugan VG , Chen R , Spiro DJ , et al. ( 2008 ) The Evolutionary Genetics and Emergence of Avian Influenza Viruses in Wild Birds . PLoS Pathogens 4 ( 5 ): e1000076 . doi: 10.1371/journal.ppat.1000076 , URL https://www.ncbi.nlm.nih.gov/pmc/ articles/PMC2387073/ OpenUrl CrossRef PubMed ↵ El-Shesheny R , Gomaa M , Sayes ME , et al. ( 2025 ) Emergence of a novel reassortant highly pathogenic avian influenza clade 2.3.4.4b A(H5N2) Virus, 2024 . Emerging Microbes & Infections 14 ( 1 ): 2455601 . doi: 10.1080/22221751.2025.2455601 , URL https://doi.org/10.1080/22221751.2025.2455601 , publisher: Taylor & Francis eprint: https://doi.org/10.1080/22221751.2025.2455601 OpenUrl CrossRef PubMed ↵ Elsmo EJ , Wünschmann A , Beckmen KB , et al. ( 2023 ) Highly Pathogenic Avian Influenza A(H5N1) Virus Clade 2.3.4.4b Infections in Wild Terrestrial Mammals, United States, 2022 . Emerging Infectious Diseases 29 ( 12 ): 2451 – 2460 . doi: 10.3201/eid2912.230464 , URL https://www.ncbi.nlm.nih.gov/pmc/articles/PMC10683806/ OpenUrl CrossRef PubMed ↵ Fang R , Min Jou W , Huylebroeck D , et al. ( 1981 ) Complete structure of A/duck-/Ukraine/63 influenza hemagglutinin gene: Animal virus as progenitor of human H3 Hong Kong 1968 influenza hemagglutinin . Cell 25 ( 2 ): 315 – 323 . doi: 10.1016/0092-8674(81)90049-0 , URL https://www.sciencedirect.com/science/article/ pii/0092867481900490 OpenUrl CrossRef PubMed Web of Science ↵ Fereidouni SR , Starick E , Beer M , et al. ( 2009 ) Highly Pathogenic Avian Influenza Virus Infection of Mallards with Homo- and Heterosubtypic Immunity Induced by Low Pathogenic Avian Influenza Viruses . PLOS ONE 4 ( 8 ): e6706 . doi: 10.1371/journal.pone.0006706 , URL https://journals.plos.org/plosone/article? id=10.1371/journal.pone.0006706, publisher: Public Library of Science OpenUrl CrossRef PubMed ↵ Fusaro A , Monne I , Salviato A , et al. ( 2011 ) Phylogeography and Evolutionary History of Reassortant H9N2 Viruses with Potential Human Health Implications . Journal of Virology 85 ( 16 ): 8413 – 8421 . doi: 10.1128/JVI.00219-11 , URL https://www.ncbi.nlm.nih.gov/pmc/articles/PMC3147996/ OpenUrl Abstract / FREE Full Text ↵ Fusaro A , Zecchin B , Vrancken B , et al. ( 2019 ) Disentangling the role of Africa in the global spread of H5 highly pathogenic avian influenza . Nature Communications 10 ( 1 ): 5310 . doi: 10.1038/s41467-019-13287-y , URL https://www.nature.com/articles/s41467-019-13287-y , publisher: Nature Publishing Group OpenUrl CrossRef PubMed ↵ Fusaro A , Zecchin B , Giussani E , et al. ( 2024 ) High pathogenic avian influenza A(H5) viruses of clade 2.3.4.4b in Europe—Why trends of virus evolution are more difficult to predict . Virus Evolution 10 ( 1 ): veae027 . doi: 10.1093/ve/veae027 , URL https://doi.org/10.1093/ve/veae027 OpenUrl CrossRef PubMed ↵ Gabry J , Češnovar R , Johnson A , et al. ( 2025 ) cmdstanr: R Interface to ‘CmdStan’ . URL https://mc-stan.org/cmdstanr/ ↵ Garg S , Reinhart K , Couture A , et al. ( 2025 ) Highly Pathogenic Avian Influenza A(H5N1) Virus Infections in Humans . New England Journal of Medicine 392 ( 9 ): 843 – 854 . doi: 10.1056/NEJMoa2414610 , URL https://www.nejm.org/doi/full/10.1056/NEJMoa2414610 , publisher: Massachusetts Medical Society eprint: https://www.nejm.org/doi/pdf/10.1056/NEJMoa2414610 OpenUrl CrossRef ↵ Gilbert M , Golding N , Zhou H , et al. ( 2014 ) Predicting the risk of avian influenza A H7N9 infection in live-poultry markets across Asia . Nature Communications 5 ( 1 ): 4116 . doi: 10.1038/ncomms5116 , URL https://www.nature.com/articles/ncomms5116 , publisher: Nature Publishing Group OpenUrl CrossRef PubMed ↵ Gill MS , Lemey P , Faria NR , et al. ( 2013 ) Improving Bayesian Population Dynamics Inference: A Coalescent-Based Model for Multiple Loci . Molecular Biology and Evolution 30 ( 3 ): 713 – 724 . doi: 10.1093/molbev/mss265 , URL https://doi.org/10.1093/molbev/mss265 OpenUrl CrossRef PubMed Web of Science ↵ Hartig F ( 2024 ) DHARMa: Residual Diagnostics for Hierarchical (Multi-Level / Mixed) Regression Models . URL https://github.com/florianhartig/dharma ↵ Hasegawa M , Kishino H , Yano Ta ( 1985 ) Dating of the human-ape splitting by a molecular clock of mitochondrial DNA . Journal of Molecular Evolution 22 ( 2 ): 160 – 174 . https://doi.org/10.1007/BF02101694 , xURL doi: 10.1007/BF02101694 OpenUrl CrossRef PubMed Web of Science ↵ Hesterberg U , Harris K , Stroud D , et al. ( 2009 ) Avian influenza surveillance in wild birds in the European Union in 2006 . Influenza and Other Respiratory Viruses 3 ( 1 ): 1 – 14 . doi: 10.1111/j.1750-2659.2008.00058.x , URL https://onlinelibrary.wiley.com/doi/abs/10.1111/j.1750-2659.2008.00058.x , xeprint: https://onlinelibrary.wiley.com/doi/pdf/10.1111/j.1750-2659.2008.00058.x OpenUrl CrossRef PubMed Web of Science ↵ Hicks JT , Edwards K , Qiu X , et al. ( 2022 ) Host diversity and behavior determine patterns of interspecies transmission and geographic diffusion of avian influenza A subtypes among North American wild reservoir species . PLOS Pathogens 18 ( 4 ): e1009973 . doi: 10.1371/journal.ppat.1009973 , URL https://journals.plos.org/plospathogens/article?id=10.1371/journal . ppat.1009973, publisher: Public Library of Science OpenUrl CrossRef PubMed ↵ Hill MO ( 1973 ) Diversity and Evenness: A Unifying Notation and Its Consequences . Ecology 54 ( 2 ): 427 – 432 . doi: 10.2307/1934352 , URL https://www.jstor.org/stable/1934352 , publisher: [Wiley, Ecological Society of America] OpenUrl CrossRef Web of Science ↵ Hill NJ , Bishop MA , Trovão NS , et al. ( 2022 ) Ecological divergence of wild birds drives avian influenza spillover and global spread . PLOS Pathogens 18 ( 5 ): e1010062 . doi: 10.1371/journal.ppat.1010062 , URL https://journals.plos.org/plospathogens/article?id=10.1371/journal.ppat.1010062 , publisher: Public Library of Science OpenUrl CrossRef PubMed ↵ Hubert L , Arabie P ( 1985 ) Comparing partitions . Journal of Classification 2 ( 1 ): 193 – 218 . doi: 10.1007/BF01908075 , URL https://doi.org/10.1007/BF01908075 OpenUrl CrossRef Web of Science ↵ James J , Billington E , Warren CJ , et al. ( 2023 ) Clade 2.3.4.4b H5N1 high pathogenicity avian influenza virus (HPAIV) from the 2021/22 epizootic is highly duck adapted and poorly adapted to chickens . Journal of General Virology 104 ( 5 ): 001852 . doi: 10.1099/jgv.0.001852 , URL https://www.microbiologyresearch.org/content/journal/jgv/10.1099/jgv.0.001852 , publisher: Microbiology Society , OpenUrl CrossRef ↵ Kandeil A , Patton C , Jones JC , et al. ( 2023 ) Rapid evolution of A(H5N1) influenza viruses after intercontinental spread to North America . Nature Communications 14 ( 1 ): 3082 . doi: 10.1038/s41467-023-38415-7 , URL https://www.nature.com/articles/s41467-023-38415-7 , publisher: Nature Publishing Group OpenUrl CrossRef PubMed ↵ Katoh K , Standley DM ( 2013 ) MAFFT Multiple Sequence Alignment Software Version 7: Improvements in Performance and Usability . Molecular Biology and Evolution 30 ( 4 ): 772 – 780 . doi: 10.1093/molbev/mst010 , URL https://doi.org/10.1093/molbev/mst010 OpenUrl CrossRef PubMed Web of Science ↵ Kawaoka Y , Krauss S , Webster RG ( 1989 ) Avian-to-human transmission of the PB1 gene of influenza A viruses in the 1957 and 1968 pandemics . Journal of Virology 63 ( 11 ): 4603 – 4608 . doi: 10.1128/jvi.63.11.4603-4608.1989 , URL https://journals.asm.org/doi/10.1128/jvi.63.11.4603-4608.1989 , publisher: American Society for Microbiology OpenUrl Abstract / FREE Full Text ↵ Kay M ( 2024 ) tidybayes: Tidy Data and Geoms for Bayesian Models . URL http://mjskay.github.io/tidybayes ↵ Keck F ( 2020 ) Handling biological sequences in R with the bioseq package . Methods in Ecology and Evolution 11 ( 12 ): 1728 – 1732 . doi: 10.1111/2041-210X.13490 , URL https://onlinelibrary.wiley.com/doi/abs/10.1111/2041-210X.13490 , xeprint: https://onlinelibrary.wiley.com/doi/pdf/10.1111/2041-210X.13490 OpenUrl CrossRef ↵ Knape J , Arlt D , Barraquand F , et al. ( 2018 ) Sensitivity of binomial N-mixture models to overdispersion: The importance of assessing model fit . Methods in Ecology and Evolution 9 ( 10 ): 2102 – 2114 . doi: 10.1111/2041-210X.13062 , URL https://onlinelibrary.wiley.com/doi/abs/10.1111/2041-210X.13062 , xeprint: https://onlinelibrary.wiley.com/doi/pdf/10.1111/2041-210X.13062 OpenUrl CrossRef ↵ Kwon JH , Bahl J , Swayne DE , et al. ( 2020 ) Domestic ducks play a major role in the maintenance and spread of H5N8 highly pathogenic avian influenza viruses in South Korea . Transboundary and Emerging Diseases 67 ( 2 ): 844 – 851 . doi: 10.1111/tbed.13406 , URL https://onlinelibrary.wiley.com/doi/abs/10.1111/tbed . 13406, eprint: https://onlinelibrary.wiley.com/doi/pdf/10.1111/tbed.13406 OpenUrl CrossRef ↵ Kéry M ( 2018 ) Identifiability in N-mixture models: a large-scale screening test with bird data . Ecology 99 ( 2 ): 281 – 288 . doi: 10.1002/ecy.2093 , URL https://onlinelibrary.wiley.com/doi/abs/10.1002/ecy.2093 , xeprint: https://onlinelibrary.wiley.com/doi/pdf/10.1002/ecy.2093 OpenUrl CrossRef PubMed ↵ Lam TTY , Zhou B , Wang J , et al. ( 2015 ) tDissemination, divergence and establishment of H7N9 influenza viruses in China . Nature 522 ( 7554 ): 102 – 105 . doi: 10.1038/nature14348 , URL https://www.nature.com/articles/nature14348 , publisher: Nature Publishing Group OpenUrl CrossRef ↵ Layan M , Dacheux L , Lemey P , et al. ( 2023 ) Uncovering the endemic circulation of rabies in Cambodia . Molecular Ecology 32 ( 18 ): 5140 – 5155 . doi: 10.1111/mec.17087 , URL https://onlinelibrary.wiley.com/doi/abs/10.1111/mec.17087 , xeprint: https://onlinelibrary.wiley.com/doi/pdf/10.1111/mec.17087 OpenUrl CrossRef ↵ Lee DH , Torchetti MK , Winker K , et al. ( 2015 ) Intercontinental Spread of Asian-Origin H5N8 to North America through Beringia by Migratory Birds . Journal of Virology 89 ( 12 ): 6521 – 6524 . doi: 10.1128/jvi.00728-15 , URL https://journals.asm.org/doi/10.1128/jvi.00728-15 , publisher: American Society for Microbiology OpenUrl Abstract / FREE Full Text ↵ Leguia M , Garcia-Glaessner A , Muñoz-Saavedra B , et al. ( 2023 ) Highly pathogenic avian influenza A (H5N1) in marine mammals and seabirds in Peru . Nature Communications 14 ( 1 ): 5489 . doi: 10.1038/s41467-023-41182-0 , URL https://www.nature.com/articles/s41467-023-41182-0 , publisher: Nature Publishing Group OpenUrl CrossRef PubMed ↵ Leinster T , Cobbold CA ( 2012 ) Measuring diversity: the importance of species similarity . Ecology 93 ( 3 ): 477 – 489 . doi: 10.1890/10-2402.1 , URL https://onlinelibrary.wiley.com/doi/abs/10.1890/10-2402.1 , xeprint: https://onlinelibrary.wiley.com/doi/pdf/10.1890/10-2402.1 OpenUrl CrossRef PubMed Web of Science ↵ Lemey P , Rambaut A , Drummond AJ , et al. ( 2009 ) Bayesian Phylogeography Finds Its Roots . PLOS Computational Biology 5 ( 9 ): e1000520 . doi: 10.1371/journal.pcbi.1000520 , URL https://journals.plos.org/ploscompbiol/article?id=10.1371/journal.pcbi.1000520 OpenUrl CrossRef PubMed ↵ Lemey P , Rambaut A , Welch JJ , et al. ( 2010 ) Phylogeography Takes a Relaxed Random Walk in Continuous Space and Time . Molecular Biology and Evolution 27 ( 8 ): 1877 – 1885 . doi: 10.1093/molbev/msq067 , URL https://doi.org/10.1093/molbev/msq067 OpenUrl CrossRef PubMed Web of Science ↵ Lemey P , Rambaut A , Bedford T , et al. ( 2014 ) Unifying Viral Genetics and Human Transportation Data to Predict the Global Transmission Dynamics of Human Influenza H3N2 . PLOS Pathogens 10 ( 2 ): e1003932 . doi: 10.1371/journal.ppat.1003932 , URL https://journals.plos.org/plospathogens/article?id=10.1371/journal.ppat.1003932 OpenUrl CrossRef PubMed ↵ Lewis NS , Banyard AC , Whittard E , et al. ( 2021 ) Emergence and spread of novel H5N8, H5N5 and H5N1 clade 2.3.4.4 highly pathogenic avian influenza in 2020 . Emerging Microbes & Infections 10 ( 1 ): 148 – 151 . doi: 10.1080/22221751.2021.1872355 , URL https://doi.org/10.1080/22221751.2021.1872355 , publisher: Taylor & Francis eprint: https://doi.org/10.1080/22221751.2021.1872355 OpenUrl CrossRef PubMed ↵ Li C , Bu Z , Chen H ( 2014 ) Avian influenza vaccines against H5N1 ‘bird flu’ . Trends in Biotechnology 32 ( 3 ): 147 – 156 . doi: 10.1016/j.tibtech.2014.01.001 OpenUrl CrossRef PubMed ↵ Li Y , Wu T , Wang S , et al. ( 2023 ) Developing integrated rice-animal farming based on climate and farmers choices . Agricultural Systems 204 : 103554 . doi: 10.1016/j.agsy.2022.103554 , URL https://www.sciencedirect.com/science/article/pii/ S0308521X22001901 OpenUrl CrossRef ↵ Lowen AC ( 2017 ) tConstraints, Drivers, and Implications of Influenza A Virus Reassortment . Annual Review of Virology 4 (Volume 4 , 2017): 105 – 121 . doi: 10.1146/annurev-virology-101416-041726 , URL https://www.annualreviews.org/content/journals/10.1146/annurev-virology-101416-041726 , publisher: Annual Reviews OpenUrl CrossRef PubMed ↵ Lu L , Lycett SJ , Leigh Brown AJ ( 2014 ) Reassortment patterns of avian influenza virus internal segments among different subtypes . BMC Evolutionary Biology 14 ( 1 ): 16 . doi: 10.1186/1471-2148-14-16 , URL https://doi.org/10.1186/1471-2148-14-16 OpenUrl CrossRef PubMed ↵ Lycett SJ , Duchatel F , Digard P ( 2019 ) A brief history of bird flu . Philosophical Transactions of the Royal Society B: Biological Sciences 374 ( 1775 ): 20180257 . https://doi.org/10.1098/rstb.2018.0257 , URL https://royalsocietypublishing.org/doi/10 . 1098/rstb.2018.0257, publisher: Royal Society OpenUrl CrossRef PubMed ↵ Lycett SJ , Pohlmann A , Staubach C , et al. ( 2020 ) Genesis and spread of multiple reassortants during the 2016/2017 H5 avian influenza epidemic in Eurasia . Proceedings of the National Academy of Sciences 117 ( 34 ): 20814 – 20825 . doi: 10.1073/pnas.2001813117 , URL https://www.pnas.org/doi/10.1073/pnas.2001813117 , publisher: Proceedings of the National Academy of Sciences OpenUrl Abstract / FREE Full Text ↵ Max Carvalho L , Rambaut A , Lam TT , et al. ( 2016 ) Exploring the temporal structure of heterochronous sequences using TempEst (formerly Path-O-Gen) . Virus Evolution 2 ( 1 ). doi: 10.1093/ve/vew007 , URL https://doi.org/10.1093/ve/vew007 OpenUrl CrossRef PubMed ↵ Mitchell S , Reeve R , White T , et al. ( 2022 ) rdiversity: Measurement and Partitioning of Similarity-Sensitive Biodiversity . URL 10.32614/CRAN.package.rdiversity ↵ Morse J , Coyle J , Mikesell L , et al. ( 2024 ) Influenza A(H5N1) Virus Infection in Two Dairy Farm Workers in Michigan . New England Journal of Medicine 391 ( 10 ): 963 – 964 . doi: 10.1056/NEJMc2407264 , URL https://www.nejm.org/doi/full/10.1056/NEJMc2407264 , publisher: Massachusetts Medical Society eprint: https://www.nejm.org/doi/pdf/10.1056/NEJMc2407264 OpenUrl CrossRef PubMed ↵ Mostafa A , Blaurock C , Scheibner D , et al. ( 2020 ) Genetic incompatibilities and reduced transmission in chickens may limit the evolution of reassortants between H9N2 and panzootic H5N8 clade 2.3.4.4 avian influenza virus showing high virulence for mammals . Virus Evolution 6 ( 2 ): veaa077 . doi: 10.1093/ve/veaa077 , URL https://doi.org/10.1093/ve/veaa077 OpenUrl CrossRef PubMed ↵ Nguyen TQ , Hutter CR , Markin A , et al. ( 2025 ) Emergence and interstate spread of highly pathogenic avian influenza A(H5N1) in dairy cattle in the United States . Science 388 ( 6745 ): eadq0900 . doi: 10.1126/science.adq0900 , URL https://www.science.org/doi/full/10.1126/science.adq0900 , publisher: American Association for the Advancement of Science OpenUrl CrossRef PubMed ↵ Nickbakhsh S , Hall MD , Dorigatti I , et al. ( 2016 ) Modelling the impact of co-circulating low pathogenic avian influenza viruses on epidemics of highly pathogenic avian influenza in poultry . Epidemics 17 : 27 – 34 . doi: 10.1016/j.epidem.2016.10.005 , URL https://www.sciencedirect.com/science/article/pii/S1755436516300354 OpenUrl CrossRef PubMed ↵ Olsen B , Munster VJ , Wallensten A , et al. ( 2006 ) Global Patterns of Influenza A Virus in Wild Birds . Science 312 ( 5772 ): 384 – 388 . doi: 10.1126/science.1122438 , URL https://www.science.org/doi/10.1126/science.1122438 , publisher: American Association for the Advancement of Science OpenUrl Abstract / FREE Full Text ↵ Peacock TP , Moncla L , Dudas G , et al. ( 2025 ) The global H5N1 influenza panzootic in mammals . Nature 637 ( 8045 ): 304 – 313 . doi: 10.1038/s41586-024-08054-z , URL https://www.nature.com/articles/s41586-024-08054-z , publisher: Nature Publishing Group OpenUrl CrossRef ↵ Pereira HG , Tůmová B , Law VG ( 1965 ) Avian influenza A viruses . Bulletin of the World Health Organization 32 ( 6 ): 855 – 860 . URL https://www.ncbi.nlm.nih.gov/pmc/articles/PMC2555286/ OpenUrl PubMed Web of Science ↵ Pohlmann A , Starick E , Grund C , et al. ( 2018 ) Swarm incursions of reassortants of highly pathogenic avian influenza virus strains H5N8 and H5N5, clade 2.3.4.4b, Germany, winter 2016/17 . Scientific Reports 8 ( 1 ): 15 . doi: 10.1038/s41598-017-16936-8 , URL https://www.nature.com/articles/s41598-017-16936-8 , publisher: Nature Publishing Group OpenUrl CrossRef PubMed ↵ R Core Team ( 2018 ) R: A language and environment for statistical computing . R Foundation for Statistical Computing , Vienna, Austria . ↵ Rambaut A , Drummond AJ , Xie D , et al. ( 2018 ) Posterior Summarization in Bayesian Phylogenetics Using Tracer 1.7 . Systematic Biology 67 ( 5 ): 901 – 904 . doi: 10.1093/sysbio/syy032 , URL https://doi.org/10.1093/sysbio/syy032 OpenUrl CrossRef PubMed ↵ Rand WM ( 1971 ) Objective Criteria for the Evaluation of Clustering Methods . Journal of the American Statistical Association 66 ( 336 ): 846 – 850 . doi: 10.2307/2284239 , URL https://www.jstor.org/stable/2284239 , publisher: [ American Statistical Association, Taylor & Francis, Ltd.] OpenUrl CrossRef Web of Science ↵ Robinson D , Hayes A , Couch S ( 2025 ) broom: Convert Statistical Objects into Tidy Tibbles . URL https://broom.tidymodels.org/ ↵ Royle JA ( 2004 ) N-Mixture Models for Estimating Population Size from Spatially Replicated Counts . Biometrics 60 ( 1 ): 108 – 115 . doi: 10.1111/j.0006-341X.2004.00142.x , URL https://doi.org/10.1111/j.0006-341X.2004.00142.x OpenUrl CrossRef PubMed Web of Science ↵ Salzberg SL , Kingsford C , Cattoli G , et al. ( 2007 ) Genome Analysis Linking Recent European and African Influenza (H5N1) Viruses . Emerging Infectious Diseases 13 ( 5 ): 713 – 718 . doi: 10.3201/eid1305.070013 , URL https://www.ncbi.nlm.nih.gov/pmc/articles/PMC2432181/ OpenUrl CrossRef PubMed Web of Science ↵ van der Linden WJ , Hambleton RK Samejima F ( 1997 ) Graded Response Model . In: van der Linden WJ , Hambleton RK (eds) Handbook of Modern Item Response Theory . Springer , New York, NY, p 85 – 100 , doi: 10.1007/978-1-4757-2691-65 , URL https://doi.org/10.1007/978-1-4757-2691-65 OpenUrl CrossRef ↵ Saucedo O , Martcheva M ( 2017 ) Competition between low and high pathogenicity avian influenza in a two-patch system . Mathematical Biosciences 288 : 52 – 70 . doi: 10.1016/j.mbs.2017.02.012 , URL https://www.sciencedirect.com/science/article/pii/S0025556417300986 OpenUrl CrossRef PubMed ↵ Schäffr JR , Kawaoka Y , Bean WJ , et al. ( 1993 ) Origin of the Pandemic 1957 H2 Influenza A Virus and the Persistence of Its Possible Progenitors in the Avian Reservoir . Virology 194 ( 2 ): 781 – 788 . doi: 10.1006/viro.1993.1319 , URL https://www.sciencedirect.com/science/article/pii/S004268228371319X OpenUrl CrossRef PubMed Web of Science ↵ Shapiro B , Rambaut A , Drummond AJ ( 2006 ) Choosing Appropriate Substitution Models for the Phylogenetic Analysis of Protein-Coding Sequences . Molecular Biology and Evolution 23 ( 1 ): 7 – 9 . doi: 10.1093/molbev/msj021 , URL https://doi.org/10.1093/molbev/msj021 OpenUrl CrossRef PubMed Web of Science ↵ Shu Y , McCauley J ( 2017 ) GISAID: Global initiative on sharing all influenza data– from vision to reality . Eurosurveillance 22 ( 13 ): 30494 . 1560-7917.ES.2017.22.13.30494, URL https://www.eurosurveillance.org/content/10.2807/1560-7917.ES.2017.22.13.30494 , publisher: European Centre for Disease Prevention and Control OpenUrl PubMed ↵ Signore AV , Giacinti J , Jones MEB , et al. ( 2025 ) Spatiotemporal reconstruction of the North American A(H5N1) outbreak reveals successive lineage replacements by descendant reassortants . Science Advances 11 ( 28 ): eadu4909 . doi: 10.1126/sciadv.adu4909 , URL https://www.science.org/doi/10.1126/sciadv.adu4909 , publisher: American Association for the Advancement of Science OpenUrl CrossRef PubMed ↵ Smith GJD , Donis RO , World Health Organization/World Organisation for Animal Health/Food and Agriculture Organization (WHO/OIE/FAO) H5 Evolution Working Group ( 2015 ) Nomenclature updates resulting from the evolution of avian influenza A(H5) virus clades 2.1.3.2a, 2.2.1, and 2.3.4 during 2013–2014 . Influenza and Other Respiratory Viruses 9 ( 5 ): 271 – 276 . doi: 10.1111/irv.12324 , URL https://onlinelibrary.wiley.com/doi/abs/10.1111/irv.12324 , xeprint: https://onlinelibrary.wiley.com/doi/pdf/10.1111/irv.12324 OpenUrl CrossRef PubMed ↵ Suchard MA , Lemey P , Baele G , et al. ( 2018 ) Bayesian phylogenetic and phylodynamic data integration using BEAST 1.10 . Virus evolution 4 ( 1 ): vey016 – vey016 . doi: 10.1093/ve/vey016 , URL https://www.ncbi.nlm.nih.gov/pubmed/29942656 https://www.ncbi.nlm.nih.gov/pmc/PMC6007674/ , publisher: Oxford University Press OpenUrl CrossRef PubMed ↵ Sun H , Li H , Tong Q , et al. ( 2023 ) Airborne transmission of human-isolated avian H3N8 influenza virus between ferrets . Cell 186 ( 19 ): 4074 – 4084.e11 . doi: 10.1016/j.cell.2023.08.011 , URL https://www.sciencedirect.com/science/article/pii/S0092867423008917 OpenUrl CrossRef PubMed ↵ Taubenberger JK , Reid AH , Lourens RM , et al. ( 2005 ) Characterization of the 1918 influenza virus polymerase genes . Nature 437 ( 7060 ): 889 – 893 . doi: 10.1038/nature04230 , URL https://www.nature.com/articles/nature04230 , publisher: Nature Publishing Group OpenUrl CrossRef PubMed Web of Science ↵ THE GLOBAL CONSORTIUM FOR H5N8 and RELATED INFLUENZA VIRUSES ( 2016 ) Role for migratory wild birds in the global spread of avian influenza H5N8 . Science 354 ( 6309 ): 213 – 217 . doi: 10.1126/science.aaf8852 , URL https://www.science.org/doi/10.1126/science.aaf8852 , publisher: American Association for the Advancement of Science OpenUrl Abstract / FREE Full Text ↵ Tomás G , Marandino A , Panzera Y , et al. ( 2024 ) Highly pathogenic avian influenza H5N1 virus infections in pinnipeds and seabirds in Uruguay: Implications for bird–mammal transmission in South America . Virus Evolution 10 ( 1 ): veae031 . https://doi.org/10.1093/ve/veae031 , xURL doi: 10.1093/ve/veae031 OpenUrl CrossRef ↵ Trovão NS , Suchard MA , Baele G , et al. ( 2015 ) Bayesian Inference Reveals Host-Specific Contributions to the Epidemic Expansion of Influenza A H5N1 . Molecular Biology and Evolution 32 ( 12 ): 3264 – 3275 . doi: 10.1093/molbev/msv185 , URL https://doi.org/10.1093/molbev/msv185 OpenUrl CrossRef PubMed ↵ Tuncer N , TJuan , MMaia , et al. ( 2016 ) Dynamics of low and high pathogenic avian influenza in wild and domestic bird populations . Journal of Biological Dynamics 10 ( 1 ): 104 – 139 . doi: 10.1080/17513758.2015.1111449 , URL https://doi.org/10.1080/17513758.2015.1111449 , publisher: Taylor & Francis eprint: https://doi.org/10.1080/17513758.2015.1111449 OpenUrl CrossRef PubMed ↵ Uhart MM , Vanstreels RET , Nelson MI , et al. ( 2024 ) Epidemiological data of an influenza A/H5N1 outbreak in elephant seals in Argentina indicates mammal-to-mammal transmission . Nature Communications 15 ( 1 ): 9516 . doi: 10.1038/s41467-024-53766-5 , URL https://www.nature.com/articles/s41467-024-53766-5 , publisher: Nature Publishing Group OpenUrl CrossRef PubMed ↵ Uyeki TM , Milton S , Hamid CA , et al. ( 2024 ) Highly Pathogenic Avian Influenza A(H5N1) Virus Infection in a Dairy Farm Worker . New England Journal of Medicine 390 ( 21 ): 2028 – 2029 . doi: 10.1056/NEJMc2405371 , URL https://www.nejm.org/doi/full/10.1056/NEJMc2405371 , publisher: Massachusetts Medical Society eprint: https://www.nejm.org/doi/pdf/10.1056/NEJMc2405371 OpenUrl CrossRef PubMed ↵ Vehtari A , Gelman A , Simpson D , et al. ( 2021 ) Rank-Normalization, Folding, and Localization: An Improved R^ for Assessing Convergence of MCMC (with Discussion ). Bayesian Analysis 16 ( 2 ): 667 – 718 . doi: 10.1214/20-BA1221 , URL https://projecteuclid.org/journals/bayesian-analysis/volume-16/issue-2/ Rank-Normalization-Folding-and-Localization--An-Improved-R%cb%86-for/10.1214/20-BA1221.full, publisher: International Society for Bayesian Analysis OpenUrl CrossRef ↵ Wickham H , Averick M , Bryan J , et al. ( 2019 ) Welcome to the Tidyverse . Journal of open source software 4 ( 43 ): 1686 OpenUrl CrossRef ↵ Wood SN ( 2003 ) Thin plate regression splines . Journal of the Royal Statistical Society: Series B (Statistical Methodology) 65 ( 1 ): 95 – 114 . doi: 10.1111/1467-9868.00374 , URL https://onlinelibrary.wiley.com/doi/abs/10.1111/1467-9868.00374 , xeprint: https://onlinelibrary.wiley.com/doi/pdf/10.1111/1467-9868.00374 OpenUrl CrossRef ↵ Wood SN ( 2011 ) Fast Stable Restricted Maximum Likelihood and Marginal Likelihood Estimation of Semiparametric Generalized Linear Models . Journal of the Royal Statistical Society Series B: Statistical Methodology 73 ( 1 ): 3 – 36 . doi: 10.1111/j.1467-9868.2010.00749.x , URL https://doi.org/10.1111/j.1467-9868.2010 . 00749.x OpenUrl CrossRef PubMed ↵ Xie R , Edwards KM , Wille M , et al. ( 2023 ) The episodic resurgence of highly pathogenic avian influenza H5 virus . Nature 622 ( 7984 ): 810 – 817 . https://doi.org/10.1038/s41586-023-06631-2 , URL https://www.nature.com/articles/s41586-023-06631-2 , xnumber: 7984 Publisher: Nature Publishing Group OpenUrl CrossRef PubMed ↵ Yang J , Müller NF , Bouckaert R , et al. ( 2019a ) Bayesian phylodynamics of avian influenza A virus H9N2 in Asia with time-dependent predictors of migration . PLOS Computational Biology 15 ( 8 ): e1007189 . doi: 10.1371/journal.pcbi.1007189 , URL https://journals.plos.org/ploscompbiol/article?id=10.1371/journal.pcbi.1007189 , publisher: Public Library of Science OpenUrl CrossRef Yang J , Xie D , Nie Z , et al. ( 2019b ) Inferring host roles in bayesian phylodynamics of global avian influenza A virus H9N2 . Virology 538 : 86 – 96 . doi: 10.1016/j.virol.2019.09.011 , URL https://www.sciencedirect.com/science/article/pii/S0042682219302764 OpenUrl CrossRef PubMed ↵ Yang Q , Wang B , Lemey P , et al. ( 2024 ) Synchrony of Bird Migration with Global Dispersal of Avian Influenza Reveals Exposed Bird Orders . Nature Communications 15 ( 1 ): 1126 . doi: 10.1038/s41467-024-45462-1 , URL https://www.nature.com/articles/s41467-024-45462-1 , publisher: Nature Publishing Group OpenUrl CrossRef PubMed ↵ Yang Z ( 1994 ) Maximum likelihood phylogenetic estimation from DNA sequences with variable rates over sites: Approximate methods . Journal of Molecular Evolution 39 ( 3 ): 306 – 314 . doi: 10.1007/BF00160154 , URL https://doi.org/10.1007/BF00160154 OpenUrl CrossRef PubMed Web of Science ↵ Youk S , Torchetti MK , Lantz K , et al. ( 2023 ) H5N1 highly pathogenic avian influenza clade 2.3.4.4b in wild and domestic birds: Introductions into the United States and reassortments, December 2021–April 2022 . Virology 587 : 109860 . doi: 10.1016/j.virol.2023.109860 , URL https://www.sciencedirect.com/science/article/pii/S0042682223001733 OpenUrl CrossRef ↵ Zeng J , Du F , Xiao L , et al. ( 2024 ) Spatiotemporal genotype replacement of H5N8 avian influenza viruses contributed to H5N1 emergence in 2021/2022 panzootic . Journal of Virology 98 ( 3 ): e01401 – 23 . doi: 10.1128/jvi.01401-23 , URL https://journals.asm.org/doi/full/10.1128/jvi.01401-23 , publisher: American Society for Microbiology OpenUrl CrossRef PubMed ↵ Zhou Y , Li Y , Chen H , et al. ( 2024 ) tOrigin, spread, and interspecies transmission of a dominant genotype of BJ/94 lineage H9N2 avian influenza viruses with increased threat . Virus Evolution 10 ( 1 ): veae106 . doi: 10.1093/ve/veae106 , URL https://doi.org/10.1093/ve/veae106 OpenUrl CrossRef PubMed View the discussion thread. Back to top Previous Next Posted July 23, 2025. Download PDF Email Thank you for your interest in spreading the word about bioRxiv. NOTE: Your email address is requested solely to identify you as the sender of this article. Your Email * Your Name * Send To * Enter multiple addresses on separate lines or separate them with commas. You are going to email the following The Winners Take It All? Evolutionary Success of H5Nx Reassortants in the 2020–2024 Panzootic Message Subject (Your Name) has forwarded a page to you from bioRxiv Message Body (Your Name) thought you would like to see this page from the bioRxiv website. Your Personal Message CAPTCHA This question is for testing whether or not you are a human visitor and to prevent automated spam submissions. Share The Winners Take It All? Evolutionary Success of H5Nx Reassortants in the 2020–2024 Panzootic James Baxter , Jing Yang , Will Harvey , Simon Dellicour , Anne Pohlmann , Mingxiao Ma , Wenjun Liu , Yuhai Bi , Paul Digard , Martin Beer , Marion Koopmans , Samantha Lycett , Lu Lu bioRxiv 2025.07.19.665680; doi: https://doi.org/10.1101/2025.07.19.665680 Share This Article: Copy Citation Tools The Winners Take It All? Evolutionary Success of H5Nx Reassortants in the 2020–2024 Panzootic James Baxter , Jing Yang , Will Harvey , Simon Dellicour , Anne Pohlmann , Mingxiao Ma , Wenjun Liu , Yuhai Bi , Paul Digard , Martin Beer , Marion Koopmans , Samantha Lycett , Lu Lu bioRxiv 2025.07.19.665680; doi: https://doi.org/10.1101/2025.07.19.665680 Citation Manager Formats BibTeX Bookends EasyBib EndNote (tagged) EndNote 8 (xml) Medlars Mendeley Papers RefWorks Tagged Ref Manager RIS Zotero Tweet Widget Facebook Like Google Plus One Subject Area Evolutionary Biology Subject Areas All Articles Animal Behavior and Cognition (7635) Biochemistry (17691) Bioengineering (13892) Bioinformatics (41937) Biophysics (21452) Cancer Biology (18588) Cell Biology (25504) Clinical Trials (138) Developmental Biology (13378) Ecology (19899) Epidemiology (2067) Evolutionary Biology (24320) Genetics (15609) Genomics (22506) Immunology (17736) Microbiology (40394) Molecular Biology (17181) Neuroscience (88605) Paleontology (666) Pathology (2832) Pharmacology and Toxicology (4824) Physiology (7641) Plant Biology (15156) Scientific Communication and Education (2045) Synthetic Biology (4294) Systems Biology (9825) Zoology (2271)