Direction of the expansion from pairwise psi matrix

[THccaa]

This is a rather basic question, but since I am confronted with a rather complicated scenario I have problems wrapping my head around it.
I have calculated pairwise psi values of 6 populations. The output looks like this:
0.000000000 | 0.0255514910 | 0.0019620907 | 0.0032395251 | 0.0270943481 | 0.10144639
-0.025551491 | 0.0000000000 | -0.0243305212 | -0.0230105421 | 0.0008357631 | 0.07573271
-0.001962091 | 0.0243305212 | 0.0000000000 | 0.0009405305 | 0.0249488201 | 0.09955641
-0.003239525 | 0.0230105421 | -0.0009405305 | 0.0000000000 | 0.0242193108 | 0.09973597
-0.027094348 | -0.0008357631 | -0.0249488201 | -0.0242193108 | 0.0000000000 | 0.07473885
-0.101446393 | -0.0757327092 | -0.0995564079 | -0.0997359748 | -0.0747388462 | 0.00000000

Now I wonder about the direction of expansion. E.g. there seems to be a signal of expansion between 1 and 6 (psi=0.10144639 / -0.101446393; these values are also significant based on permutation test), but I do not know where it started. Are the source populations (S1/S2) given in rows or the columns?

[esnielsen]

Hi Th-al, I'm struggling with the same question- did you ever figure out which are the source populations?

Thanks, Erica

[THccaa]

Hi Erica. I am fairly sure that the columns represent the source populations. I think so, because I calculated the most likely origin and if the source populations were in the rows, the direction of my psi values and the inferred origin would not fit together at all.