modeling the split between non-sister populations?

[Mario Ernst]

Hi Ryan, 

currently, I am working on a project where we have around 10 populations (clusters identified using structure). They are allopatric or parapatric and inhabit different regions with very different landscape characteristics. I am testing three 2D models (secondary contact, isolation with migration, and strict isolation) and want to check whether we see differences in model choice depending on the region. 

Initially, I modeled only the splits between populations that are sister to each other. However, I would also like to add more comparisons to see how the support for competing models varies across space and phylogeny. Merging populations is not recommended, as you mention that structure within populations can strongly influence the results in unpredictive ways. But it also crossed my mind to compare populations that are not sisters to each other. So, for instance, given the following topology: (((pop1, pop2), pop3), pop4), I consider modeling the following comparisons:

• pop1 vs. pop3
• pop2 vs. pop3
• pop3 vs. pop4

Is it possible to do so or would it be strictly necessary to use 3D models in order to model more ancient splits?

Thanks very much for your support!!

All the best, Mario

[Ryan Gutenkunst]

Hello Mario,

It’s probably best to use 3D models, but you can certainly get started using 2D models. Be careful with interpretation, since indirect gene flow like 1 -> 2 -> 3 will show up as flow from 1 -> 3.

Best,
Ryan

> --
> You received this message because you are subscribed to the Google Groups "dadi-user" group.
> To unsubscribe from this group and stop receiving emails from it, send an email to dadi-user+...@googlegroups.com.
> To view this discussion on the web visit https://groups.google.com/d/msgid/dadi-user/af457716-ab4b-4d12-be53-006430c91146n%40googlegroups.com.

[Mario Ernst]

Thanks for the answer! That's a good point. I will keep it in mind.