Global dissemination of influenza A virus is driven by wild bird migration through arctic and subarctic zones

Abstract

Influenza A viruses (IAV) circulate endemically among many wild aquatic bird populations that seasonally migrate between wintering grounds in southern latitudes to breeding ranges along the perimeter of the circumpolar arctic. Arctic and subarctic zones are hypothesized to serve as ecologic drivers of the intercontinental movement and reassortment of IAVs due to high densities of disparate populations of long distance migratory and native bird species present during breeding seasons. Iceland is a staging ground that connects the East Atlantic and North Atlantic American flyways, providing a unique study system for characterizing viral flow between eastern and western hemispheres. Using Bayesian phylodynamic analyses, we sought to evaluate the viral connectivity of Iceland to proximal regions and how inter-species transmission and reassortment dynamics in this region influence the geographic spread of low and highly pathogenic IAVs. Findings demonstrate that IAV movement in the arctic and subarctic reflects wild bird migration around the perimeter of the circumpolar north, favouring short-distance flights between proximal regions rather than long distance flights over the polar interior. Iceland connects virus movement between mainland Europe and North America, consistent with the westward migration of wild birds from mainland Europe to Northeastern Canada and Greenland. Though virus diffusion rates were similar among avian taxonomic groups in Iceland, gulls play an outsized role as sinks of IAVs from other avian hosts prior to onward migration. These data identify patterns of virus movement in northern latitudes and inform future surveillance strategies related to seasonal and emergent IAVs with potential public health concern.

Keywords: Iceland; influenza A virus; phylodynamics; subarctic; transmission; wild birds.

Figure 1.

Markov Chain Monte Carlo (MCC) time-scaled phylogeographic tree and phylogeographic map of global downsampled Influenza A virus (IAV) sequences. (1.a) Phylogeographic tree, color-coded by geographic source region. Grey branches indicate historical sequences between 1978 and 2008 which contributed to the overall tree structure but were masked from phylogeographic analyses (Gass J.D., 2021). (1.b) phylogeographic map of global downsampled IAV sequences, color-coded by geographic source region. 95% HPD tree and posterior value support can be found in Supplementary figure 4.

Figure 2.

Global virus transitions and movement between regions. 2a) Discrete phylogeographic transitions between regional states. Transitions represent virus movement between regional states, measured by Bayes factor support (S.A.: South America, N.A.: North America). Cells with Bayes factor values above 3.0 are framed by a green border (Kass & Raftery, 1995). Only BFs > 3 with corresponding posterior probability (PP) estimates > 0.5 are presented as statistically supported. 2b) Markov jumps, which are represented as the percent of global transitions between regional states along phylogenetic tree branches. Global transitions <1% are not shown.

Figure 3.

Host migratory distribution and IAV circulation in the North Atlantic. Significant discrete phylogeographic transitions between Iceland, the rest of Europe, and North America are represented by arrows from mainland Europe to Iceland (Bayes Factor (BF)= 110,622), Northeastern USA to Iceland (BF= 414.87), and Iceland to northeastern Canada and Greenland (BF= 127.22). Only BFs > 3 with corresponding posterior probability (PP) estimates > 0.5 are presented as statistically supported. Arrows signify directionality and greater arrow width corresponds to higher BF support of phylogeographic transitions between intracontinental geographic states. BFs for state transitions between all global regions can be found in Supplementary Table 4. Host migratory ranges for populations of shorebirds (yellow), gulls (green), ducks (blue), and geese (red) were compiled from individual species-specific shape files acquired from IUCN Red List (IUCN, 2021). Species-specific migratory range maps can be found in the Supplementary figure 7.

Figure 4.

Discrete and continuous host-specific transmission dynamics and virus movement in Iceland. a) Transitions and Bayes Factors of IAV transmission between host taxa, and b) diffusion rates (km/year), by host taxa. Only BFs > 3 with corresponding posterior probability (PP) estimates > 0.5 are presented as statistically supported.

Figure 5.

Global and Iceland-specific reassortment patterns and Bayes factors (BF) among Hemagglutinin (HA) subtypes (5a and b), Multiple Correspondence Analysis of Host order and HA subtype (5c), and reassortment patterns and Bayes factors among Neuraminidase (NA) subtypes (5d and e). Only BFs > 3 with corresponding posterior probability (PP) estimates > 0.5 are presented as statistically supported. a) The global reassortment matrix of HA subtypes demonstrates the breadth and highly significant transitions between donor H1 subtypes and almost all other recipient subtypes. b) In Iceland, a strong association between donor H16 subtypes and recipient H2 subtypes were found, as well as other significant transitions between H3 and H10 and H11 subtypes and H2 and H10 subtypes. Both forward and reverse rates are visualized in these matrices, with subtypes labeled on the x-axis representing recipients and subtypes labeled on the y-axis representing donors. Bayes factors, signifying the level of significance, are represented as colors from beige (low) to dark red (high). c) Multiple correspondence analysis (MCA) of HA subtype by is presented host order. Size of blue rectangles for HA subtype corresponds to prevalence in the dataset. d) Global and e) Iceland-specific reassortment patterns and Bayes factors (BF) among Neuraminidase (NA) subtypes. The global reassortment matrix demonstrates the breadth and highly significant transitions between donor N8 subtypes and almost all other recipient subtypes. In Iceland, a strong association between donor N3 subtypes and recipient N7 subtypes were found, as well as other significant transitions between N8 and N9 and N7 subtypes. Both forward and reverse rates are visualized in these matrices, with subtypes labeled on the x-axis representing recipients and subtypes labeled on the y-axis representing donors. Bayes factors, signifying the level of significance, are represented as colors from white (low) to dark green (high).