ddRadTag Fragment Size Selection for EcoRI

[Shann_rk]

Hi everyone,
I'm currently trying to optimise a ddradtag protocol and after we received our first set of sequences back, it appears we've had a few issues. (We are using Stacks to analyse our data)
My first question is, when using EcoRI as one of the digest enzymes, does anyone know if there should be a difference in fragment numbers depending on which size you select? We are going mostly off the Peterson paper and according to them, the fragment distribution appears to be fairly uniform.
We have size selected for ~400bp (+/- 50bp) which I know is quite a large window, but we are getting fairly excessive numbers of stacks (upwards of 75000) with quite a low depth.
I'm thinking its fairly likely that this is due to our wide size selection range, but as other people seem to be selecting smaller fragments, I was wondering if this is likely to have any effect?
Any advice would be appreciated.
Cheers

[SNPsaurus]

What species are you genotyping and what is your second enzyme? If the GC content differs from the GC of the second enzyme, then that will shift the size distribution of the fragments. Are you size selecting on a Pippin?

[Shann_rk]

We are doing Koala and Black lip oyster (and a few other species which we haven't looked at properly yet). Our second enzyme is MspI and we chose to use the pippin prep for size selection as we figured it would be more accurate than a gel cut.
I'm just trying to work out why we ended up with so many fragments compared to what others have documented.

[SNPsaurus]

Hmmm, MspI is GC rich, and pearl oyster (is that a comparable reference?) is AT-rich, so that would significantly increase the size of ddRAD fragments, and EcoRI is AT-rich, so that would increase the number of loci. But if you are seeing the same pattern in different species with different GC contents, that wouldn't really explain it.

If there are lots of polymorphisms then reads that should be allelic might be kept as separate loci as well, so that could contribute.