9Recent history has provided us with two severe viral outbreaks (Ebola and Zika) and 10 one pandemic (Influenza A/H1N1). In all three cases, post-hoc analyses have given us 11 deep insights into what triggered these outbreaks, their timing, evolutionary dynamics, 12 and their phylogeography, but the genomic characteristics of outbreak viruses are still 13 unclear. To address this outstanding question, we searched for a common denominator of 14 these recent outbreaks, positing that genomes of outbreak viruses are in an unstable evolu-15 tionary state, while those of non-outbreak viruses are stabilized by a network of correlated 16 substitutions that have been found to be prevalent. Here, we show that during regular 17 epidemics, viral genomes are indeed stabilized by a dense network of weakly correlated 18 sites, and that these networks disappear during pandemics and outbreaks when rates of 19 evolution increase transiently. Post-pandemic, these evolutionary networks are progres-20 sively re-established. We finally show that destabilization is not caused by mutations 21 targeting epitopes, but more likely by changes in the environment sensu lato. Our results 22 prompt for a new interpretation of pandemics as being caused by, from an evolutionary 23 standpoint, destabilized, unhealthy viruses. 24 lution 26 29 situation, where both the virus and the hosts coexist. Such a stable evolutionary strategy 30 can however break down when the virus becomes extremely virulent, which can lead to a 31 severe outbreak or even a pandemic. Recent history is rich in such examples with an Ebola 32 virus outbreak in 2014 [2], a Zika outbreak in 2015 [3], and an Influenza pandemic in 2009 33 [4]. Despite all of these recent examples, in which the phylodymanics of these events were 34 meticulously reconstructed, we still do not know what characterizes the evolutionary 35 dynamics of outbreaks and pandemics. Here we address this outstanding question by 36contrasting the evolutionary dynamics of pandemic and non-pandemic viruses.
37As theory tells us that regular epidemics are the result of a dynamic equilibrium
38[5], we posit that outbreaks are associated with a disequilibrium at the genomic level.
39More specifically, we suggest that outbreaks involve destabilized viral genomes, where 40 evolutionary stability is maintained by compensatory mutations, that can be epistatic or 41 not, but that result in signals of correlated evolution. We predict that such signals are 42 severely weakened during an outbreak. As these signals often lead to complex networks 43 of interactions [6, 7], we test how the structure of these correlation networks is affected 44 during an outbreak. We show that during an outbreak, viral genes are destabilized.
45
Results
46Networks of correlated sites are destabilized during outbreaks. In search for 47 trasted the glycoprotein precursor (GP) sequences of the Ebola virus that circulated 49 4 before, and during 2014/2016 outbreak. For this, we identified the pairs of nucleotides 50 that show evidence for co...