Host-Specific and Segment-Specific Evolutionary Dynamics of Avian and Human Influenza A Viruses: A Systematic Review

Abstract

Understanding the evolutionary dynamics of influenza viruses is essential to control both avian and human influenza. Here, we analyze host-specific and segment-specific Tajima's D trends of influenza A virus through a systematic review using viral sequences registered in the National Center for Biotechnology Information. To avoid bias from viral population subdivision, viral sequences were stratified according to their sampling locations and sampling years. As a result, we obtained a total of 580 datasets each of which consists of nucleotide sequences of influenza A viruses isolated from a single population of hosts at a single sampling site within a single year. By analyzing nucleotide sequences in the datasets, we found that Tajima's D values of viral sequences were different depending on hosts and gene segments. Tajima's D values of viruses isolated from chicken and human samples showed negative, suggesting purifying selection or a rapid population growth of the viruses. The negative Tajima's D values in rapidly growing viral population were also observed in computer simulations. Tajima's D values of PB2, PB1, PA, NP, and M genes of the viruses circulating in wild mallards were close to zero, suggesting that these genes have undergone neutral selection in constant-sized population. On the other hand, Tajima's D values of HA and NA genes of these viruses were positive, indicating HA and NA have undergone balancing selection in wild mallards. Taken together, these results indicated the existence of unknown factors that maintain viral subtypes in wild mallards.

Fig 1. The selection process of systematic review of surveillance studies.

Fig 2. Tajima’s D values for gene segments sampled from the mallards, chickens and humans.

(a) shows Tajima's D values for the viruses isolated from wild mallards, (b) shows those from domestic chickens, and (c) shows those from humans. Black circles and error bars represent estimated medians and 95% confidence intervals for the median of Tajima's D across datasets using 1-sample Wilcoxon signed rank test. Gray circles represent Tajima's D values of each dataset. Asterisk denotes the significantly positive or negative Tajima's D based on the result of 1-sample Wilcoxon signed rank test.

Fig 3. The change of Tajima’s D with a sudden increase of population.

(a) shows the setting of time evolution of viral population size and (b) shows the result of time series change of mean Tajima’s D. Gray dot line represents 95% confidence interval of Tajima’s d value for D = 0.