How to distinguish selection vs population demography from Tajima's D test

Question

[nitinra]

Hi all,

I have calculated genomewide Tajima's D using VCFtools --Tajima option and got a positive Tajima value (diploid organism, butterfly species). This could either mean balancing selection or population contraction. Is there a way to distinguish between the two? I have also done pairwise Fst and did find outlier loci in my populations. Can I use that as a proxy for selection along with positive Tajima's D?

Answer 1

[h.mon]

I am not to speed with the latest literature in the are, so take my answer with a grain of salt.

Generally (and superficially) speaking, selection is thought to be "local", that is, it will affect the site under selection and its close neighborhood, while demographic events will affect the whole genome more or less similarly. Thus, one way of differentiating selection from demography is by analysing the distribution of Tajima's D over the genome, e.g. loci by loci, or by sliding-windows. In practice, it is not so simple, because the data is noisy and there are many unknown parameters, thus alternative models can fit equally good the data.

There are many examples in the literature, e.g. Genomic regions exhibiting positive selection identified from dense genotype data.