Divergence Models - One pop diverging from the rest?

[Sean Canfield]

Aloha all, and happy holidays!

I have four populations, and I want to run a model where Pop 1 has diverged from Pops 2,3,4 (with continued migration), and want to model full migration among pops 2,3,4.

Do I need to specify divergence of Pop 1 from all of the other populations? Like so:

*DDD

****

****

****

In the case of divergence without migration:

*ddd

0***

0***

0***

Just trying to wrap my head around this a bit. Divergence models in Migrate are definitely new territory for me.

I know it's a holiday (and I've been monopolizing this space lately), so please don't feel obligated to respond right away.

Many thanks,

Sean

[Peter Beerli]

Sean

Your model may work but seems overkill, all models with divergence and migration will have issues because divergence and migration are interconnected (e.g. high migration rate will affect the divergence time estimation and bias it in simulation to too low values.)

your model seems odd because 1 is colonized from 2, 3, 4; I guess a system where 1 only comes from one of the
the others seems more appropriate
*D**
****
****
****
but if you think that 1 is split off early from the others and they split from a common ancestor you could
1. add
0 ancestor
to your infile and increase the number of populations on the first line of the infile
and then
use

****0
****d
****d
****d
d000*

I also suggest to see whether divergence only without migration will do well.

Whatever you do the program needs to be able to form a tree for the likelihood calculation; i.e. when you draw
the connection graph a population need to be connected to at least one other one.

Peter

> --
> You received this message because you are subscribed to the Google Groups "migrate-support" group.
> To unsubscribe from this group and stop receiving emails from it, send an email to migrate-suppo...@googlegroups.com.
> To view this discussion on the web visit https://groups.google.com/d/msgid/migrate-support/486528a7-d82d-42f2-8463-99f5a1d5c5a0n%40googlegroups.com.

[Sean C]

Thanks Peter! I'll do some more thinking about these models. It's an analysis of the mitochondrial control region (just the one locus), so maybe I shouldn't overdo/overthink things.

Happy holidays,

Sean

[Sean C]

Thanks Peter! I'll do some more thinking about this system and these models.

Happy holidays,

Sean

On Tuesday, December 28, 2021 at 10:51:37 AM UTC-10 Peter wrote: