Oscillations in dengue incidence and serotype prevalence are a characteristic feature of the epidemiology of this virus. Although changes in genotype composition through time are similarly a relatively common observation in studies of DENV evolution, to date there has been generally insufficient data to determine (i) whether changing patterns of genotype prevalence have any association with viral phenotype including the virological manifestation of DENV infection, and (ii) the evolutionary basis of these lineage replacement events, and specifically the respective roles of natural selection versus genetic drift.
Here we demonstrate a rapid and apparently complete lineage (genotype) replacement event within DENV-2 in southern Viet Nam that was temporally associated with an increase in disease incidence. Strikingly, a functional basis for the displacement of resident Asian/American lineage viruses was suggested by higher viremia levels in pediatric patients with Asian 1 DENV-2 infections. The presence of higher viremias in children hospitalized with Asian 1 DENV-2 infections relative to Asian/American DENV-2 infections would likely increase the probability of human-to-mosquito transmission and hence facilitate greater population diffusion. Another possible outcome of higher viraemia levels is an increased incidence of more severe disease. We did not detect significant differences in the extent of capillary permeability or thrombocytopaenia between patients with Asian 1 or Asian/American viruses in the cohort of hospitalized patients (n
389) here. This suggests that Asian 1 DENV-2 infections were not overtly associated with more severe disease. However, this was a relatively small sample size to detect differences in clinical outcomes and a larger cohort of symptomatic patients, including non-hospitalised individuals, would most likely be needed to answer this question definitively.
Our best estimates suggest the Asian 1 genotype was first introduced into southern Viet Nam in the late 1990's. Our sampling illuminated the replacement of the previously dominant Asian/American genotype by Asian 1 viruses during 2003–2007, a period in which there was an almost doubling of dengue incidence, mostly associated with DENV-2.
A large number of susceptible hosts in the population, and an associated increased force of infection, could help explain the seemingly short period in which genotype replacement occurred. Whether the apparent fitness advantage of Asian 1 viruses could be attributable to “antigenic fitness” in the face of the population-wide immune landscape during this period is unknown. However, there is a precedent for biologically relevant antigenic differences between genotypes of DENV-2. For example, South-East Asian DENV-2 viruses are less susceptible than American lineage viruses to cross-neutralization by antibodies elicited by DENV-1 infection 
. Population wide seroepidemiology, coupled with a better understanding of correlates of immunity, are clearly needed to understand serotype and genotype replacement in all endemic regions.
Virus traits in the mosquito host might also explain the difference in fitness between the Asian I and Asian/American genotypes. As an example, previous studies have demonstrated “SE Asian genotype” viruses (there was no analysis of differences between Asian 1 or Asian/American genotypes) are more infectious and disseminate faster in Ae. aegypti
mosquitoes, and replicate more efficiently in human dendritic cells, than American genotype DENV-2 viruses 
. However, we were unable to detect a measurable difference between Asian 1 or Asian/American viruses in terms of overall replication rates in C6/36 mosquito cells, or infectiousness for local Ae. aegypti
mosquitoes. Other features of the virus-mosquito interaction (e.g. extrinsic incubation time) could be equally or more important than the infectious dose. However, published data from Armstrong et al.
suggested that Asian 1 and Asian/American viruses had similar dissemination rates in Ae. aegypti
. Further studies are therefore needed to understand the importance of the mosquito as a site for the expression of the fitness differences between these two virus lineages.
Intriguingly, the displacement of Asian/American lineage viruses by Asian 1 viruses has also seemingly occurred in Thailand and Cambodia. In Thailand, the Asian/American genotype most likely co-circulated with the Asian 1 genotype for at least a decade prior to 1991, but is then absent from amongst the 139 Thai DENV-2 viruses sampled between 1992 and 2006, which all belong to the Asian 1 lineage. In Cambodia, despite a smaller sample size, lineage replacement appears to have occurred along a remarkably similar time-frame to that seen in Viet Nam, with only Asian 1 viruses being sampled after 2005. Both Thailand and Cambodia have considerable transport links with Viet Nam and it's conceivable these are sources for the introduction of Asian 1 viruses into Viet Nam.
In sum, we show that a lineage replacement event in DENV is highly likely to be linked to an underlying difference in fitness. This suggests that natural selection may play a more important role in shaping viral dynamics than previously realized. The virus genetic traits associated with the fitness of Asian 1 viruses are difficult to identify definitively on the basis of sequence data alone. Elucidation of the possible functional consequences of the 16 amino acid differences that characterize the Asian I viruses will clearly require complex reverse genetic experiments. However, we predict that the amino acid changes at E226 and E228 will be of particular importance given that they occur at sites that are invariant across all DENV sequences sampled to date (and which suggest that the vast majority of mutations at these sites are strongly deleterious because they have a major impact on fitness). Our documentation of a major lineage replacement event, coupled with the current dominance of Asian I viruses, suggests that this genotype will continue to dominate DENV-2 infections in Thailand, Cambodia and Viet Nam unless there is a major change in the host environment, such as that brought on by changes in serotype (and which themselves exhibit complex population dynamics 
The prevalence of Asian I DENV-2 viruses has multiple implications. First, it is paramount that the DENV-2 component of future dengue vaccines (reviewed in 
) be competent at eliciting immunity to viruses belonging to this genotype. Similarly, programs to develop anti-viral drugs for dengue should include Asian 1 DENV-2 viruses in their pre-clinical discovery and development programs 
. Furthermore, we would predict that Asian 1 viruses will continue to outcompete Asian/American DENV-2 viruses. A likely future setting for this event is in the Americas where currently Asian/American DENV-2 viruses predominate, having themselves displaced the American genotype.