Extremely high migration rate parameters

[Sarah]

Hi!

I am a new user and am running IMa with 2 populations of 57 and 118 individuals using 10 microsat loci and 1 mtDNA locus (2673 bp). One population is an isolated island subspecies of the other and I am trying to estimate migration and population splitting time.

I am running into problems trying to set my priors. I started initially with the manual prior suggestions which showed that the priors were way too low for -m1 and -m2 with values leaving zero close to the upper bound and rising to an unseen peak off the scale. So time and time again I have been increasing the migration priors and I keep getting too low priors for -m1 and -m2. I am now on -q1 0.0005 -q2 0.02 -qA 10 -m1 3000 -m2 500. With these values I am getting a peak for m1 with a long tail extending to the upper bound, and -m2 is still too low with a peak presumably past the upper bound! I am also getting very low values of t peaking around 0.003.

I am pretty sure my migration priors are way too high at 3000, which seem implausible? Is this telling me that it is simply just one population? Due to the remoteness of the subspecies population and FST values between it and the mainland of 0.19 for microsats and 0.45 for mtDNA I would not expect these results.

Preliminary phylogenies show that there may have been a second more recent colonisation of the island. Would this explain the high migration results and low splitting time? Or are these results simply an indication that I do not have enough data to estimate migration?

Finally, I have good chain-swapping but am struggling to get my ESS values high enough. When I try to increase the number of chains above 40, IMa crashes before starting the Markov chain. Would poor mixing be contributing to my strange results?

Any advice on what to try next or what I might be doing wrong would be really appreciated as I am stumped! My command line prompt settings are below.

Thanks for all your help,

-q1 0.0005 -q2 0.02 -qA 10 -m1 3000 -m2 500 -t 0.008 -b500000 -L10000 -n40 -fg -g10.9 -g20.8 -s123 -p45 -z1000

[Katie Everson]

Hi,

I'm having the same problem. I'm working with a 2-population, 19-nuclear-gene dataset, and my migration rate is consistently off the charts. This is true even when I raise -m over 1000 (which I believe is unreasonably high). The analysis is also recovering a huge ancestral population size and extremely recent divergence times.

Is this a "true" result (i.e., the two localities -- one mainland and one island -- actually represent a single population), or is it possible that this is an error with my analysis? I have tried raising the number of generations to 10 million (-l1000000) and my ESS values seem OK. Attached is an example of what I see in IMfig (migration not shown because it is non-significant in the analysis).

I know that this is an old post, but it appears that it was never resolved. Any help would be much appreciated,

Katie Everson