Inferring natural influenza transmission bottleneck sizes from sequencing data*

Pathogen populations typically go through a bottleneck when they pass from host to host. These bottlenecks control the genetic diversity (and therefore evolutionary potential) of single infections and determines the extent to which multiple pathogen types can reliably co-transmit. They are, however, essentially impossible to measure directly in nature. We introduce a simple method for inferring bottleneck sizes from sequences of pathogen populations. Applying this method to data from influenza A virus transmission chains, we find dramatic differences between different populations, with some consistently having bottlenecks of hundreds of viruses while others are consistently bottlenecked to a single virus. We consider several possible explanations for this difference and suggest how they could be tested.