Through recent technological and bioinformatic advances in virus genetic analysis and internationally coordinated surveillance, the 2009 influenza A pandemic has presented a unique opportunity to analyse the global spread of a new virus strain and to explore different transmission models, such as local spread and introductions from outside, in the dissemination of infections in a community. We have previously shown using highly specific diagnostic assays that the virus had infected approximately 20
% of the adult population of Scotland by 1 December 2010 (McLeish et al., 2011
). From the same study, it is possible to estimate a clinical attack rate of approximately 10
% for this virus, rendering classical epidemiological contact tracing almost impossible. Here we have added to the epidemiological information a phylogenetic analysis based on newly generated sequences of seven out of eight segments of the viral genome from a total of 70 strains from Scotland. These strains were distributed equally between the summer and winter phases and sampled representatively with respect to geographical location. Comparison of these strains with 128 strains from RUK and 292 from other global sources revealed strikingly different epidemic patterns within Scotland in the summer and winter phases. The initial phase was characterized by multiple independent introductions from both international and other UK sources, with significant local expansion of one clade that had a clear link to a possible progenitor in England, whilst the winter phase was more diverse genetically with several clades of similar size, some of which had no particularly close phylogenetic affinity to strains circulating in England. Bayesian phylogeographical analysis confirmed that the links between the epidemic in other parts of the UK to Scotland were stronger in the summer than in the winter phase.
Early studies of the diversification of A H1N1 2009 revealed seven distinguishable clades of virus (Nelson et al., 2009
), six of which were identified in North America in the summer phase (Nelson et al., 2011b
). As expected, the samples from Scotland included only a subset of the global diversity, with just two clades, 3 and 7, being represented. However, whilst clade 3 virtually disappeared from the USA by winter (Nelson et al., 2011b
), it still comprised 13
% (5/39) of the winter-phase sample in Scotland.
In comparison to England, Scotland had much lower diversity in the summer phase, with a smaller number of introductions and extensive local spread in the greater Glasgow area, leading to an epidemic dominated by one large clade. However, this clade did not lead into the winter epidemic, which appeared to arise from separate introductions. To put this on a quantitative basis, we used the beast output files to evaluate the extent to which the epidemic in Scotland developed independently (). This demonstrated clearly that local spread was dominant in both phases, but that the winter phase () appeared to be completely distinct from the epidemic in England. Unfortunately, the representation of the two phases among strains from England was quite uneven, with only 26 (out of 147) strains sequenced coming from the winter phase. This low sampling density means that we cannot say more. Nevertheless, it appears that the gravity model alone does not fully explain the phylogenetic pattern in the two phases of the epidemic.
A surprising feature of the results that may be relevant, however, is the clear links between the UK-Early sequences and both North America and Asia, particularly in the summer phase, strongly suggesting an epidemiological link from North America to Asia through the UK in the global spread of the pandemic. This suggests that air-travel routes influenced the initial spread of the pandemic between continents, as there was a much weaker direct connection between the sequences from these continents in contrast to the results obtained by Jombart et al. (2009)
from haemagglutinin and neuraminidase sequences obtained in the summer phase. In a phylogeographical analysis of the summer and winter phases of the 2009 pandemic in the USA, Nelson et al. (2011b)
found evidence for founder effects leading to phylogenetically distinct epidemics among different cities in the summer phase, whereas the winter phase was almost completely homogeneous across the USA. The high degree of mixing occurring within the USA was illustrated very clearly by the analysis of a single university campus, implying >20 phylogenetically distinct introductions to that one site in the winter phase (Holmes et al., 2011
). Earlier studies of seasonal influenza suggested seeding from the tropics as an important influence in the establishment of annual epidemics (Rambaut et al., 2008
), but the phylogeographical analysis here does not suggest that any particular link outside the UK was important for the winter epidemic in Scotland. Further and more detailed sample collections are clearly required to provide adequate answers to the questions of persistence through summer and the source of winter epidemics within the UK.