Analyses.knit

Principal component analysis of all sequenced samples.

We can see that SC2 ans NS9 cluster on their own, turns out they are smoky shrews.

Principal component analysis of Sorex cinereus samples.

Ran PCAngsd again but without the Sorex fumeus samples. Each population clusters well together now, as expected.

NGSadmix with all samples and different K values

Here again NS9 and SC2 are clearly geneticlaly different from all other samples.

Site frequency spectrum analyses for each population using angsd.

All genome scans done with 50Kb sliding windows with 10Kb steps.

Waterson’s theta

Tajima’s D

Fst

Did not include vs ON plots because they look horrible

Dxy

Did not include vs ON plots because they look horrible

Top righ quadrant represents top 5%.

MSMC2 analyses for each population

Used all the samples I had that were over ~15x. BPI (n=5), SC (n=3), ON (n=1). LI did not have any individuals with high enough coverage.

Ran with the standard 1x2+25x1+1x2+1x3 split. I ran the MSMC on each phased individual and then per population but with unphased data.

Pink samples are BPI, yellow are SC, and blue are ON.

Thin lines are individual runs, thick lines are population runs.

Bootstrapping

100 bootstraps - unphased BPI

Cross-coalescence rate for BPI (n=5) and SC (n=3)

Don’t think I need to run bootstraps for this?

GONE analysis for Bon Portage Island samples

Turns out we can’t use this type of analysis with our data. Need more than 10 samples per population.

Here’s what it looks like with BPI (n=10), which is our population with the most samples.