FEEMS with river network

[dmacguigan]

Hello,

I work on freshwater fish population genomics. I'm curious about using FEEMS with grid confined to a river network. To that end, I generated a resolution 14 ISEA4T grid with DGGRID and cropped it using "outer coordinates" created by buffering a river lines shapefile.

The resulting grid looks like this. The black line indicates the watershed boundary. The dark gray lines within the watershed are actually a very fine scale triangular grid that follows the river course.

Does using this sort of structured grid violate any of the assumptions of FEEMS?

Thank you,
Dan

[jnovembre]

Hi Daniel,

1. EEMS/FEEMS fits a model of symmetric migration. Do you expect in this system that gene flow will be mainly downstream and thus asymmetric? In those cases, you can run feems, but it can be trickier to interpret the results. Erik Lundgren and Peter Ralph have a good paper looking into the challenge of asymmetric migration (https://onlinelibrary.wiley.com/doi/abs/10.1111/1755-0998.13035) and it's on our wishlist to add fitting asymmetric migration to feems, but we're not there yet! If migration is symmetric - you should be ok in principle to run feems.

2. You say the rivers are a very fine triangular grid - It might be more natural to have the rivers be linear stepping stones? I'm pretty sure the code can handle that - it need not be triangular grids.

3. On the side, I'd love to hear more about the dataset you're generating - feel free to email me if you'd like to share more!

All the best,
John

[cypayne]

Just lurking here because I'm also interested in running feems for a river fish system that experiences asymmetric migration. Has there been progress on fitting asymmetric migration to feems?

Thanks so much!
Cheyenne