I am interested in creating Bayesian Skyline Plots to look at demographic changes in certain population groups. However, these populations have very little within population variation. Around 5 polymorphic sites within each population and one dominating haplotype that comprises almost 90% of the population (the rest being rare haplotypes). Because, BSP uses Coalescent processes for the calculations, the sample sizes would have a effect on the results. I wanted to know if sample sizes of around 30 (for DNA alignments of 700bp) would be ideal for carrying out the analysis and expect reasonably good results ?

  • 1
    $\begingroup$ What software are you using? $\endgroup$ Mar 15, 2015 at 10:32
  • $\begingroup$ Note that any plot is just a visualization method. You should be asking about sample sizes for reliable analysis. I am not an expert at this but according to wikipedia, Coalescence time and heterozygosity are dependent on population size. You have to clarify what you mean by "results". $\endgroup$
    Mar 17, 2015 at 12:49

1 Answer 1


I found a PLOS one article that might help. In this they used similar numbers you're planning to use (if i get your question and the article's data right).

The influence of the sampling scheme was investigated by drawing 40 samples in three different ways: 1) all 40 samples from a single deme, 2) 4 samples from each of 10 demes and 3) one sample from each of the 40 demes in the structured population. These correspond respectively to the local, pooled and scattered

(taken from the above linked article)

So depending on your sampling scenario your 30 samples seems to okay.

Also this is a link for the BEAST FAQ that might be helpful.


You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .