Role for migratory wild birds in the global spread of avian influenza H5N8

See allHide authors and affiliations

Science  14 Oct 2016:
Vol. 354, Issue 6309, pp. 213-217
DOI: 10.1126/science.aaf8852

Migration of influenza in wild birds

Virus surveillance in wild birds could offer an early warning system that, combined with adequate farm hygiene, would lead to effective influenza control in poultry units. The Global Consortium for H5N8 and Related Influenza Viruses found that the H5 segment common to the highly pathogenic avian influenza viruses readily reassorts with other influenza viruses (see the Perspective by Russell). H5 is thus a continual source of new pathogenic variants. These data also show that the H5N8 virus that recently caused serious outbreaks in European and North American poultry farms came from migrant ducks, swans, and geese that meet at their Arctic breeding grounds. Because the virus is so infectious, culling wild birds is not an effective control measure.

Science, this issue p. 213; see also p. 174


Avian influenza viruses affect both poultry production and public health. A subtype H5N8 (clade virus, following an outbreak in poultry in South Korea in January 2014, rapidly spread worldwide in 2014–2015. Our analysis of H5N8 viral sequences, epidemiological investigations, waterfowl migration, and poultry trade showed that long-distance migratory birds can play a major role in the global spread of avian influenza viruses. Further, we found that the hemagglutinin of clade virus was remarkably promiscuous, creating reassortants with multiple neuraminidase subtypes. Improving our understanding of the circumpolar circulation of avian influenza viruses in migratory waterfowl will help to provide early warning of threats from avian influenza to poultry, and potentially human, health.

In 2014, highly pathogenic avian influenza (HPAI) virus of the subtype H5N8 caused disease outbreaks in poultry in Asia, Europe, and North America (13). Avian influenza viruses are a threat both to global poultry production and to public health; they have the potential to cause severe disease in people and to adapt to transmit efficiently in human populations (4). This was the first time since 2005 that a single subtype of HPAI virus had spread over such a large geographical area and the first time that a Eurasian HPAI virus had spread to North America. The rapid global spread of HPAI H5N8 virus outbreaks raised the question of the routes by which the virus had been transmitted.

The segment encoding for the hemagglutinin (HA) surface protein of the HPAI H5N8 viruses is a descendant of the HPAI H5N1 virus (A/Goose/Guangdong/1/1996), first detected in China in 1996 (5). Since then, HPAI H5N1 viruses have become endemic in poultry populations in several countries. The H5 viruses have developed new characteristics by mutation and by reassortment with other avian influenza (AI) viruses, both in poultry and in wild birds. In 2005–2006, HPAI H5N1 spread from Asia to Europe, the Middle East, and Africa during the course of a few months. Although virus spread traditionally had been attributed to transport of infected poultry, infected poultry products, or HPAI-virus–contaminated materials, several observations in the 2005–2006 epidemic suggested that wild birds also might have carried the virus to previously unaffected areas (6).

A HPAI H5N8 virus with genes from viruses of the influenza A (H5N1) A/Goose/Guangdong/1/1996 lineage was first detected in birds at live bird markets in China in 2010 (1). This HPAI H5N8 virus was a reassortant virus with the HA gene segment from HPAI H5N1 virus and other gene segments from multiple other AI viruses circulating in eastern China (1) and is now categorized as HPAI H5 virus clade (7). This clade is unusually promiscuous and has been found in combination with six different neuraminidase (NA) segments, and multiple H5Nx viruses may be circulating at the same time and in the same region (8, 9). The propensity of HPAI H5 virus clade to form novel subtypes capable of rapid, global spread is a major concern.

HPAI H5N8 virus caused a large avian influenza outbreak in poultry in South Korea in the winter of 2013–2014 and subsequently spread to Japan, North America, and Europe, causing outbreaks there between autumn 2014 and spring 2015 (table S1). However, it is not clear by which routes HPAI H5N8 virus spread so rapidly around the world. Although there have been reports on parts of these outbreaks (1, 2, 10) and speculation on possible routes of transmission (3), no comprehensive global analysis has yet been performed.

The goal of this study was to analyze the available genetic, epidemiological, and ornithological data for evidence of the relative contributions from poultry trade and from wild bird movements (3, 6) for the global spread of clade during 2014–2015. For this purpose, we performed phylogeographic analysis of HPAI H5N8 viruses detected in wild birds and poultry from this global outbreak. In addition, we analyzed migration patterns of wild birds found infected with HPAI H5N8 virus, epidemiological investigations of HPAI H5N8 virus outbreaks, and poultry-trade records from countries where HPAI H5N8 virus was reported (11).

Initial phylogenetic analysis was performed using HA sequences from HPAI H5 clade viruses of poultry and wild birds from around the world between 2004 and 2015, including subtypes H5N1, H5N2, H5N3, H5N5, H5N6, and H5N8. From 2004 to 2012, clade viruses were circulating predominantly in Eastern Asia (China), with some transmission to Southeastern Asia (Fig. 1 and fig. S1). During this period, transmission involving domestic anseriformes (ducks and geese) appears to dominate, although some contribution from domestic galliformes (chickens and turkeys) and short-distance migratory wild birds (e.g., mallard ducks) is also evident (Fig. 1). Unlike H5 segments from other clades, which are mostly found as H5N1, the HPAI H5 segment of the clade viruses reassorts frequently, acquiring NA segments from cocirculating low pathogenic avian influenza (LPAI) subtypes, including N5 (from 2006 to 2010), N2 (from 2008 to 2012), N8 (from 2010), and, more recently, N6 (from 2013) (8). To indicate the host species and regions in which the reassortments are thought to have occurred, a reassortment measure was calculated using the number of branches in the posterior set of phylogenetic trees for which the NA subtype changed while the host species and region traits remained the same (normalized by branch lengths). This measure suggests that most of the observed reassortants were generated in domestic anseriformes (fig. S2), and particularly domestic anseriformes in Eastern Asia (China) within the time period 2004 to 2012 (fig. S3).

Fig. 1 Maximum clade credibility (MCC) time-scaled phylogenetic tree of multisubtype HA sequences colored by subtype, region, and host-type traits.

The clades marked a and b contain H5N8 sequences, and c and d contain sequences from Europe and North America, respectively. The displayed MCC tree was obtained from a posterior set of trees inferred using the Bayesian Evolutionary Analysis Sampling Trees (BEAST) program (13) with the SRD06 nucleotide substitution model, uncorrelated relaxed clock model, and constant population size tree prior. The branches are colored according to the most probable ancestral trait, and ancestral traits were inferred by a symmetric (subtype and region) or asymmetric discrete trait model (host-type) upon the posterior tree set (14). Host types are Dom-Ans (red), domestic anseriform birds; Dom-Gal (green), domestic galliform birds; Wild-Long (blue), long-distance migratory wild birds; Wild-Short (purple), short-distance migratory wild birds.

The time to the most recent common ancestor (TMRCA) for the HA segment of all clade HPAI H5N8 sequences was estimated as June 2010 [95% highest posterior density (HPD), January to October 2010]; the TMRCA for the corresponding NA segments was similar (September 2010; 95% HPD, April to December 2010). Clade HA H5N8 sequences were found in two subclades (Fig. 1). The smaller and earlier subclade (a in Fig. 1) contained the first sequenced HPAI H5N8 virus [A/Duck/Jiangsu/k1203/2010(H5N8)]. The larger and more recent subclade (b in Fig. 1) contained sequences from outbreaks in South Korea and other countries included in this study and caused multiple HPAI outbreaks in 2014 and 2015 globally. The TMRCA of subclade b was September 2013 for both HA (95% HPD, July to November 2013) and NA (95% HPD, May to November 2013). Consistent with earlier findings (1, 10), the phylogeny indicates that HPAI H5N8 was introduced into South Korea by long-distance migrant wild birds that acquired it from the pool of HPAI H5 viruses circulating in domestic anseriformes in Eastern Asia (China), although we formally cannot exclude the possibility that HPAI H5 viruses were circulating in unsampled locations (Fig. 1).

Distinct, well-supported clades were identified in South Korea, likely originating in the transmission of HPAI H5N8 viruses from long-distance migrants to other wild and domestic birds (10). One clade (c in Fig. 1) was ancestral to the European outbreak and another (d in Fig. 1) was ancestral to the North American outbreak. Again, we cannot exclude the possibility that viruses from these subclades were present in unsampled locations.

More detailed phylogenetic analyses, using only clade H5N8 HA sequences with location coordinates (11), showed that the virus spread along two main long-distance migration routes: one from the east Asia coast/Korean peninsula, north to the Arctic coast of the Eurasian continent, then west to Europe; and the other north from the Korean peninsula, then east across the Bering Strait, and south along the northwest coast of the North American continent to Canada and the United States (Fig. 2 and movie S1). The reconstruction did not indicate any spread between Europe and North America. The TMRCA for European HA segments was August 2014 (95% HPD, July to October 2014), and September to October 2014 (95% HPD, August to November 2014) for the North American HA segments (table S2, a and b). Similar results were found from analysis of the NA segments (table S2, c and d). There were also four separate introductions into Japan, the first estimated around February 2014 (ancestral date of single virus A/Chicken/Kumamoto/1-7/2014), and then three more, all with TMRCAs in October and November 2014. The sequences from one Japanese introduction were most closely related to sequences from Taiwan and those from another introduction to the Russian (A/Wigeon/Sakha/1/2014) and European sequences.

Fig. 2 Reconstruction of the transmission routes using phylogenetic data only from H5N8 HA sequences.

At each time slice, the host-type and location coordinates on the branches of the posterior set of phylogenetic trees are inferred and plotted as a cloud of points. The host type was inferred by discrete trait model (as Fig. 1) (14), and the continuous location coordinates were inferred using a homogeneous Brownian motion diffusion model (15). The map projection used is the azimuthal equal areas projection, centered on the North Pole, which is marked with a + sign. Color key as for Fig. 1; see also movie S1.

The phylogenetic data were also used to infer the ancestral host categories of the most recent common ancestor of the European and North American outbreak sequences, thus providing evidence for which host type had introduced the viruses into those areas (Fig. 3, figs. S4 and S5, and table S2). The most likely ancestral host category for the North American outbreak for both HA and NA segments was long-distance migrants (HA, 66%; NA, 84%). A similar result was obtained for Europe (HA, 66%; NA, 70%).

Fig. 3 Posterior distributions of TMRCA of HA sequences from Europe and North America with H5N8 subtype only, including host-type reconstructions, based upon a posterior set of phylogenetic trees generated as in Fig. 1.

Color key as for Fig. 1.

Several wild bird species with known HPAI H5N8 sequences were long-distance migrants at different stages of their migratory cycle, depending on time and place found (table S3): Five of the nine species found in South Korea in winter 2013–2014 were long-distance migrants at their wintering sites or on spring migration. Both in North America and Europe, two of the four species found in winter 2014–2015 were long-distance migrants at their wintering sites or on autumn migration (11)(tables S4 and S5 and fig. S6).

The April 2014 HPAI H5N8 virus outbreak in Japan had different characteristics from the later outbreaks in North America and Europe. The Japan outbreak was the only one that was contemporaneous with the outbreak in South Korea, and no wild birds were found positive for HPAI H5N8 virus in Japan during that outbreak.

Qualitative analysis of data from outbreak investigations on affected poultry farms in North America, Europe, and Japan (11) (table S6) showed that the likelihood of virus introduction via contaminated water, feed, and poultry was negligible (Germany). Furthermore, no links between the outbreaks in one country and those in other countries could be attributed to personnel contacts or trade of live animals, feed, or products of animal origin (Germany, Netherlands, United Kingdom, and Hungary). Many affected poultry farms were located in areas where wild waterfowl are abundant (Germany, Netherlands, United Kingdom, Italy, and Canada). Direct contact with infected wild birds (United States) or indirect contact with materials (e.g., bedding material, boots, and wheels of vehicles) contaminated with wild-bird feces was considered the most likely route of introduction into poultry holdings (United States, Germany, The Netherlands, United Kingdom, and Italy). In some outbreaks, the source of infection was unknown or inconclusive (Japan and Hungary).

We reviewed data from the Food and Agriculture Organization of the United Nations (FAO) (12) for 2011 to 2013 on export and import of live domestic ducks and chickens of affected countries to estimate the risk of spread of HPAI virus from South Korea to other countries via the international poultry trade (table S7). Data on the export of live poultry from North Korea and Mongolia, also in East Asia, were not available from FAO. Although all countries (Japan, Canada, United States, Germany, Netherlands, United Kingdom, Italy, and Hungary) where HPAI H5N8 virus emerged between November 2014 and February 2015 imported live chickens and live domestic ducks in 2013, South Korea reported the export of a low number of live chickens and no export of live domestic ducks, although unreported cross-border trade cannot be excluded. Nevertheless, based on these data, it seems unlikely that international trade in live poultry played a major role in the long-distance spread of South Korean clade HPAI H5N8 virus in 2014–2015.

Our analysis, using four different sources of data, indicates that the main routes of large-scale geographical spread of HPAI H5N8 virus were most probably via long-distance flights of infected migratory wild birds, first in spring 2014 from South Korea or other unsampled locations in the region to northern breeding grounds and then in autumn 2014 from these breeding grounds along migration routes to wintering sites in North America and Europe.

Recognition of a likely role of wild birds in the spread of HPAI reinforces the need to improve biosecurity on poultry farms and to exclude wild birds from the immediate vicinity of poultry farms. Culling wild birds and draining or disinfecting wetlands would not be effective because these viruses disseminate on rapid time scales over very large distances, making reactive interventions of this kind impractical and ineffective, as well as contravening commitments made by signatory countries to the Convention on Migratory Species and the Ramsar Convention on Wetlands.

The potential role of wild birds in the circumpolar circulation of influenza viruses does point to the need to increase our knowledge about the connectedness at the vast circumpolar (sub)arctic breeding areas between migratory waterfowl populations originating from different wintering areas. Surveillance of waterfowl at the crossroads of migratory flyways to wintering areas in Europe, Asia, and North America would inform epidemiological risk analysis and provide early warning of specific HPAI threats to poultry, and potentially human, health.

The Global Consortium for H5N8 and Related Influenza Viruses

Samantha J. Lycett,1* Rogier Bodewes,2* Anne Pohlmann,3 Jill Banks,4 Krisztián Bányai,5 Maciej F. Boni,6,7 Ruth Bouwstra,8,9 Andrew C. Breed,10 Ian H. Brown,4 Hualan Chen,11 Ádám Dán,12 Thomas J. DeLiberto,13 Nguyen Diep,7 Marius Gilbert,14,15 Sarah Hill,16 Hon S. Ip,17 Chang Wen Ke,18 Hiroshi Kida,19 Mary Lea Killian,20 Marion P. Koopmans,21 Jung-Hoon Kwon,22 Dong-Hun Lee,23 Youn Jeong Lee,24 Lu Lu,25 Isabella Monne,26 John Pasick,27,28 Oliver G. Pybus,16 Andrew Rambaut,25 Timothy P. Robinson,29 Yoshihiro Sakoda,30 Siamak Zohari,31 Chang-Seon Song,22 David E. Swayne,23 Mia Kim Torchetti,20 Hsiang-Jung Tsai,32 Ron A. M. Fouchier,21 Martin Beer,3 Mark Woolhouse,25†Thijs Kuiken21† *These authors contributed equally to this work. These authors contributed equally to this work. 1The Roslin Institute, University of Edinburgh, Edinburgh EH25 9RG, UK. 2Department of Farm Animal Health, Faculty of Veterinary Medicine, University of Utrecht, 3584 CL Utrecht, Netherlands. 3Institute of Diagnostic Virology, Friedrich Loeffler Institut, D-17493 Greifswald-Insel Riems, Germany. 4Virology Department, Animal and Plant Health Agency, Woodham Lane, Addlestone KT15 3NB, UK. 5Institute for Veterinary Medical Research, Centre for Agricultural Research, Hungarian Academy of Sciences, H1143 Budapest, Hungary. 6Centre for Tropical Medicine, Nuffield Department of Medicine, University of Oxford, Oxford OX3 7FZ, UK. 7Oxford University Clinical Research Unit, Wellcome Trust Major Overseas Programme, Ho Chi Minh City, Vietnam. 8Department of Virology, Central Veterinary Institute, Wageningen University and Research Centre, 8221 RA Lelystad, Netherlands. 9Animal Health Service, 7400 AA Deventer, Netherlands. 10Department of Epidemiological Sciences, Animal and Plant Health Agency, Woodham Lane, Addlestone KT15 3NB, UK. 11Harbin Veterinary Research Institute, Chinese Academy of Agricultural Sciences, 150001 Harbin, China. 12Veterinary Diagnostic Directorate, National Food Chain Safety Office, H1149 Budapest, Hungary. 13National Wildlife Research Center, Wildlife Services, US Department of Agriculture, Fort Collins, CO 80521, USA. 14Spatial Epidemiology Laboratory (SpELL), Université Libre de Bruxelles, B-1050 Brussels, Belgium. 15Fonds National de la Recherche Scientifique, B-1000 Brussels, Belgium. 16Department of Zoology, University of Oxford, Oxford OX1 3PS, UK. 17Wildlife Disease Diagnostic Laboratories Branch, National Wildlife Health Center, US Geological Survey, Madison, WI 53711, USA. 18Institute of Microbiology, Center for Diseases Control and Prevention of Guangdong Province, 511430 Guangzhou, China. 19Research Center for Zoonosis Control, Hokkaido University, Sapporo, Hokkaido 001-0020, Japan. 20National Veterinary Services Laboratories, Veterinary Services, US Department of Agriculture, Ames, IA 50010, USA. 21Department of Viroscience, Erasmus University Medical Center, 3015 CN Rotterdam, Netherlands. 22Avian Disease Laboratory, College of Veterinary Medicine, Konkuk University, Seoul 143-701, Republic of Korea. 23Southeast Poultry Research Laboratory, US Department of Agriculture, Athens, GA 30605, USA. 24Avian Disease Division, Animal and Plant Quarantine Agency, Gimcheon, Republic of Korea. 25Centre for Immunity, Infection and Evolution, University of Edinburgh, Edinburgh EH9 3FL, UK. 26Research and Innovation Department, Istituto Zooprofilattico Sperimentale delle Venezie, 10-35020 Legnaro (Padova), Italy. 27National Centre for Foreign Animal Disease, Canadian Food Inspection Agency, Winnipeg, MB R3E 3M4, Canada. 28Canadian Food Inspection Agency, Guelph, ON N1G 4S9, Canada. 29Livestock Systems and Environment (LSE), International Livestock Research Institute (ILRI), Post Office Box 30709, 00100 Nairobi, Kenya. 30Graduate School of Veterinary Medicine, Hokkaido University, Sapporo, Hokkaido 060-0818, Japan. 31Department of Virology, Immunobiology and Parasitology, National Veterinary Institute, SE-751 89 Uppsala, Sweden. 32Animal Health Research Institute, Council of Agriculture, New Taipei City 25158, Taiwan.

Supplementary Materials

Materials and Methods

Figs. S1 to S6

Tables S1 to S10

Movie S1

References (1659)

References and Notes

  1. Materials and methods are available as supplementary materials on Science Online.
Acknowledgments: This study was financially supported by the European Commission H2020 program under contract number 643476 ( (to A.P., J.B., A.B., I.B., M.P.K., A.R., R.A.M.F., M.B., M.W., and T.K.), European Commission FP7 program under contract number 278433 (PREDEMICS) (to A.R.), the U. S. Geological Survey Ecosystems Mission Area (to H.S.I.), National Institutes of Health grant number 1R01AI101028-02A1 (to M.G.), United Kingdom Research Council Environmental and Social Ecology of Human Infectious Diseases UrbanZoo program (G1100783/1), Biotechnology and Biological Sciences Research Council (BBSRC) Zoonoses in Livestock in Kenya ZooLinK (BB/L019019/1) programs (to T.P.R. and M.W.), CGIAR Research Programme on Agriculture for Nutrition and Health (A4NH) (to T.P.R.), Canadian Food Inspection Agency (to J.P.), Hungarian Academy of Sciences Lendület (Momentum) program (to K.B.) and the Wellcome Trust (grant number 093724/B/10/Z) (to M.W. and A.R.). S.J.L. is supported by the University of Edinburgh Chancellor’s Fellowship scheme, the Roslin Institute BBSRC strategic program grant (BB/J004227/1), and the Centre of Expertise in Animal Disease Outbreaks (EPIC). We gratefully acknowledge the the originating laboratories, where specimens were first obtained, and the submitting laboratories, where sequence data were generated and submitted to the EpiFlu Database of the Global Initiative on Sharing All Influenza Data (GISAID), on which this research is based. All contributors of data may be contacted directly via the GISAID website ( The accession numbers (GenBank, GISAID, and/or workset identification numbers) of all genetic sequences used in this study are provided in table S9 and are accessible from the website of GISAID ( We acknowledge Y. Berhane and T. Hisanaga for sequencing the Canadian virus isolates and G. Koch for his technical advice on the poultry outbreaks in the Netherlands. The funders had no role in study design, data collection and interpretation, or the decision to submit the work for publication. Any use of trade products or firm names is for descriptive purposes and does not imply endorsement by the U.S. government.

Stay Connected to Science

Navigate This Article