Phylodynamics of H5N1 Highly Pathogenic Avian Influenza in Europe, 2005-2010: Potential for Molecular Surveillance of New Outbreaks

Abstract

Previous Bayesian phylogeographic studies of H5N1 highly pathogenic avian influenza viruses (HPAIVs) explored the origin and spread of the epidemic from China into Russia, indicating that HPAIV circulated in Russia prior to its detection there in 2005. In this study, we extend this research to explore the evolution and spread of HPAIV within Europe during the 2005-2010 epidemic, using all available sequences of the hemagglutinin (HA) and neuraminidase (NA) gene regions that were collected in Europe and Russia during the outbreak. We use discrete-trait phylodynamic models within a Bayesian statistical framework to explore the evolution of HPAIV. Our results indicate that the genetic diversity and effective population size of HPAIV peaked between mid-2005 and early 2006, followed by drastic decline in 2007, which coincides with the end of the epidemic in Europe. Our results also suggest that domestic birds were the most likely source of the spread of the virus from Russia into Europe. Additionally, estimates of viral dispersal routes indicate that Russia, Romania, and Germany were key epicenters of these outbreaks. Our study quantifies the dynamics of a major European HPAIV pandemic and substantiates the ability of phylodynamic models to improve molecular surveillance of novel AIVs.

Keywords: Bayesian inference; Europe; H5N1; Russia; highly pathogenic avian influenza; phylodynamic models; phylogeography; surveillance.

Figure 1

(A,B) Bayesian Skyline plots (BSP) illustrating temporal changes in the relative genetic diversity of H5N1 HPAIV isolates from outbreaks in Europe and the Russia between May, 2005 and June, 2010 estimated from the HA and NA gene sequences. Line plots summarize estimates of the effective population size (N_eT), a measure of genetic diversity, for HA (above) and NA (below) gene regions; the shaded regions correspond to the 95% HPD; (C) Temporal distribution of H5N1 HPAI outbreaks (per year) in domesticated poultry and wild birds in Europe and Russia from between 2005 and 2010.

Figure 2

Maximum clade credibility (MCC) trees for H5N1 HPAIV hemagglutinin (HA; A) and neauraminidase (NA; B) gene regions, respectively, Branch lengths are rendered proportional to absolute time (see timescales), and branches are colored according to the most probable host type (wild birds, domestic birds, or other). Branches where the host state is uncertain (where the posterior probability of any hosts < 0.5) are colored black. The posterior probabilities for the ancestral host states are shown in the upper left panel for each tree.

Figure 3

Maximum clade credibility (MCC) tree for H5N1 HPAIV hemagglutinin (HA; A) and neauraminidase (NA; B) gene regions. Branch lengths are rendered proportional to absolute time (see timescales), and branches are colored according to the most probable geographic location.

Figure 4

Posterior probabilities of ancestral areas of H5N1 HPAIV hemagglutinin (HA; A) and neauraminidase (NA; B) gene regions collected in Europe and the Russia.

Figure 5

Bayes factor (BF) test for significant non-zero dispersal rates in H5N1 HPAIV. Only rates supported by a BF greater than six are indicated. The color gradient of lines correspond to the probability of the inferred dispersal routes; blue lines and red lines indicate relatively weak and strong support, respectively. The maps are based on satellite pictures available in Google Earth (Available online: http://earth.google.com). (A) HA gene; (B) NA gene. The maps are based on satellite images sourced from the NASA World Wind Java SDK (Available online: http://worldwind.arc.nasa.gov/java/).