Computational Evolution Group

Dr Quentin Atkinson, Professor Russell Gray and I have just had a paper published in Molecular Biology and Evolution in which we use coalescent theory to estimate relative human population sizes through time in different regions of the world using the Bayesian skyline plot in the BEAST software package:

Mitochondrial DNA Variation Predicts Population Size in Humans and Reveals a Major Southern Asian Chapter in Human Prehistory

Our estimates of the relative size of “modern” populations are remarkably concordant (PNG/Australia being the exception) with pre-colonial population size estimates based on anthropological evidence. Strikingly this concordance persists across a number of different definitions of region population boundaries in Eurasia. I interpret this to suggest that the real population history is more of a continuous, isolation-by-distance process, so that, if you draw a boundary around a region of the world, the coalescent information will roughly reflect the relative population size of the people in that demarcated area (assuming that rates of movement is roughly equal everywhere – which is clearly not the case – and might be one of the reasons for the inconsistencies we see in Papua New Guinea and Australia).

One should contrast this with papers such as Wang et al (2007; discussed in previous post) which suggest that genetic diversity (as measured by heterozygosity) is correlated with the age of the population rather than its size. While both of these statements are likely to be true in certain circumstances, the dominant effect will depend on whether the population is near equilibrium. If a population has been at equilibrium for a long time then genetic diversity will no longer reflect the population’s age, but will rather be dominated by the population’s size.

In this respect the coalescent analysis has some advantage over methods that just look at heterozygosity in that it is able to dissect the population diversity out across time by splitting it up into coalescent events. In this way coalescent theory can also be used to track changes in population size through time. We use this fact to estimate relative population sizes back through time and in doing so uncover evidence that from about 45,000 to 20,000 years ago, the majority of humans were in Asia. The method we use in this paper suffers from too many simplifying assumptions to list, but it nevertheless hints at some attractive directions for research, both in development of methods (explicit inclusion of geographic information and models of isolation-by-distance) and in furthering our understanding of human history: why was everyone in Asia 30,000 years ago?

In this regard you may be interested in a related bit of research by Garrigan et al (2007) that has just come out in Genetics:

Inferring human population sizes, divergence times and rates of gene flow from mitochondrial, X and Y chromosome resequencing data

Merry Christmas (or Happy holidays, or whatever is the appropriate well-wish is in your culture at this time of year!),

Alexei