Hi all, Thanks to those of you who replied to my quesitons. Peter Beerli's book chapter (see below) was quite helpful and I recommend checking it out. Here is my orignial query: I am using Peter Beerli's MIGRATE program to estimate Ne for a number of populations for which I have microsatellite data. So far, I am getting reasonable Ne estimates (I assume microsat mutation rates between 10^-4 and 10^-3). I would like to get some feedback on how to properly interpret these coalescent-based estimates. Specifically, I am wondering how the estimates might be biased if: (1) the assumption of migration-drift equilibrium is violated (2) the assumption of constant population size has been violated (fluctuating pop size or declines) (3) the organisms have overlapping generations (which is the case with my study organism) Also, I have used ONeSAMP and LDNe to get estimates of Ne from the linkage-disequilibrium method. In general, the estimates I am getting from MIGRATE are about an order of magnitude larger than the LD-based estimates. Since the estimates from MIGRATE are "long-term" estimates and the LD estimates are "contemporary," does this difference provide evidence in support of population declines? Based on other analyses, I have reason to believe my study populations have undergone decline. Thanks for any input or suggestions you can provide! Ivan Here are the replies: Dear Ivan, > (2) In a recent book chapter (*) I explored the issue of violation of > assumption (concerning effects on population size) in MIGRATE a little. It turned out that the estimates are long-term averages > heavily influenced by the recent past. In terms of coalescent events this may be no surprise because many lineage > pairs give information recent coalescences and so seem to influence the results more than the > distant past. Do not forget that recent past the coalescence framework may be still far in the past for some other > estimators. > (1) If migration happened only during a short time in the past, MIGRATE > will fit an average migration rate but you may want to see whether the data supports such bursts in the past by > using the migration event options in the program. > (3) I assume all your populations have the same life history/generation > time schedule and so even if there is a bias in the estimates (although the coalescence seems rather robust to these) your comparison > would be unproblematic. > > You do not say how you did your analysis or whether your populations are > connected by migration or not, if they are then you should analyze them > together and not as single (independent) populations. > Peter > * P. Beerli. How to use migrate or why are Markov chain Monte Carlo > programs difficult to use? In G. Bertorelle, M. W. Bruford, H. C. Hauïe, A. Rizzoli, and C. Vernesi, > editors, Population Genetics for Animal Conservation, volume 17 of Conservation Biology, pages 42"79. > Cambridge University Press, Cambridge UK, 2009. Some of the problems you are thinking about are dealt with by the Lamarc program, which incorporates growth parameters. Also, use other software to estimate growth, isolantion-with-migration, and compare with your Migrate results. If your data has quite a straight-forward pattern all methods should give you similar results, and if they don't that is also interesting. > Also, my experience is that Peter Beerli himself is very approachable and helpful if you want to ask specific questions about Migrate -n. > Ramos-Orsins et al. have written a few papers about how different key tests are affected by population size change. > Best, Magdalena Hi Ivan, > I don't know if you've yet found a solution to your inquiry, but the > software BottleSim and information provided in the companion paper might > provide some insight for your purposes (see > http://chkuo.name/software.html). > Best wishes, > Fred Janzen Ivan C. Phillipsen Department of Zoology Oregon State University 3029 Cordley Hall Corvallis, OR 97331-2914 Email: philliiv@science.oregonstate.edu Website: www.science.oregonstate.edu/~philliiv/ phillipsen@gmail.com