Questions about the models and parameters to use, and about the statistics to employ to interpret the results

[victor SUAREZ SANTIAGO]

Dear Peter,

I have two questions that I would appreciate you clarifying:

1. Model Specification in Migrate: I am studying a rare, high-mountain plant species endemic to a small area, with only three populations. These populations occupy specific niches and have a disjunct distribution. My data (8 microsatellite loci) show isolation by distance. Additionally, structure analyses reveal a primary structure of genetic diversity between the two major population clusters and a substructure within one differentiating the other two populations. I am attempting to use Migrate to test how divergence occurred among these three populations, but I am unsure if I am defining the models correctly. For 3 populations, there are 6 possible genealogies. Furthermore, I am considering the possibility of both recurrent (D) and no (d) gene flow, which increases the number of models to 24. What I am unclear about is whether I should specify the possibility of migration towards the potential source population. For example, one of the models would be that 2 diverges from 1 with recurrent migration (D) and 3 diverges from 2 also with recurrent migration: x00 Dx0 0Dx. Or the one that considers that 2 diverges from 1 with recurrent migration (D) and 3 diverges from 2 without migration (d): x00 Dx0 0dx, etc... And so on for all 24 models. Would specifying the models in this way be sufficient when testing divergence, or should I consider the possibility of migration from the diverged populations to the source populations, so that I have to use "s" where I now put 0?

2. Interpreting Migrate Results: To draw conclusions from the results with Migrate, what statistic do you recommend using: the Mode (as you do in your Current Protocols paper), the Median (which is robust to outliers), or the Mean?