There's a nice review in the most recent Trends in Genetics on the use of spatially explicit models in human population genetics. As everyone knows, classic population genetic theory generally makes very restrictive assumptions about the amount of structure in a population-- that is, that there is none. Or maybe a couple populations that exchange migrants. These are all nice assumptions for making the math less hairy, but the recent influx of data (largely generated for medical genetic purposes) on populations from around the globe has revived an interest in modeling variation in a more flexible manner. Along with every person comes their approximate latitude and longitude-- how does genetic data vary on those axes?

The review is a nice summary of recent work in the area, which has shown that a progressive two-dimensional stepping stone model is, for the moment, a decent approximation for human variation (the bottom part of the figure shows parts of the world shaded according to heterozygosity. The largest values are in Africa, where humans originated, and gradually go down, through the serial bottleneck, the further you go from the origin). But this field is ripe for new models and theory. And exploring new models, of course, leads to overturning old ones.

Labels: Genetics, Population genetics