What is a population?

This paper discusses what a population is in statistical genetics:

• Lawson (2014), "Populations in statistical genetic modelling and inference", submitted book chapter for the interdisciplinary book "What is a Population", OUP. (preprint) (Note: this is a newer version of a similar manuscript "What is a population? In statistical genetic modelling and inference". The text has changed, the pictures and overall message is the same).

Author Summary

A population is defined in STRUCTURE and FineSTRUCTURE as a group of individuals who cannot be statistically distinguished. We look at what this means in comparison to generative models of how populations are constructed - effectively, we need to assume that mating is sufficiently random over an appropriate timescale so that we cannot reconstruct familial relationships. The appropriate timescale is a function of the amount of data that we have, and with modern linked datasets, we can easily detect such relationships. We can still define a population however, but more thought must be put into it. We review how a population appears in PC (Principal Component) analysis and model-based approaches in the presence of genetic drift, historical and contemporary admixture, and familial structure.