# genetic difference subpopulations vs movement rate

Someone told me that if a geneticist finds no significant difference between 2 subpopulations that have temporal spatial overlap it might be that they are still almost closed (no connection). Is this right?

Can I deduce anything about the individual exchange rate between populations and the genetic similarity/difference? For instance, one would need 5% of individuals to switch populations to keep them almost genetically identical. Does this differ between species (let's say fish and mammals)?

Thanks.

ps: any references to this would be great

[..] if a geneticist finds no significant difference between 2 subpopulations that have temporal spatial overlap

This isn't clear. Significance difference between what? What statistical hypothesis are you testing for which one would found no signicant difference (accepted the null hypothesis). Are you maybe thinking about a t-test between testing difference between the average phenotypic traits?

it might be that they are still almost closed (no connection). Is this right?

Do you mean no gene flow (aka. no migration)?

Can I deduce anything about the individual exchange rate between populations and the genetic similarity/difference?

"Individual exchange rate" is typically "called migration" or "gene flow".

At some extend, yes you can. But the question is really broad and will be hard to answer in the general sense. I will just provide a very common methodology here.

You can compute $F_{ST}$, a statistic of population divergence. A very good paper explaining the math of $F_{ST}$ is Nei (1973). I recommend computing $F_{ST}$ following Weir and Cockerham (1984). Then, if you can assume a finite island model (equal migration rate $m$ between any two population) and a constant and known population size $N$, then

$$F_{St} = \frac{1}{1 + 4N(m+\mu) \frac{d}{d-1}}$$

, where $d$ is the number of demes (number of islands)This equation can be found at many places. Slatkin (1995) is a very good paper but note that its final solution differs a little bit from the one I presented (by a square term) for reasons explained in Charlesworth (1998). You can solve for $m$ and plug the observed $F_{ST}$ and $N$. Note that depending on the markers used, $\mu$ will be negligible in comparison to $m$.

You might want to have a look at the very related post How could one calculate the gene flow between two populations?

• @Wave Did it answer your question? – Remi.b Dec 5 '17 at 21:41