How is migration incorporated into the SimBit program of Remi Matthey-Doret and Michael Whitlock?

Is an island model or stepping stone model assumed?


In SimBit, no specific model is assumed as the user can very much chose for himself/herself what migration scenario to be simulated. The migration scenario can be set with the option --m (or equivalently with --DispMat).

There are four ways (what is called "Mode" in the manual) to indicate migration. A, LSS, OnePatch, Island and LinearNormal.

  • A
    • This mode stands for "All". This mode allows the user to enter a matrix with the migration rate between any two patches. Any migration scenario including the ones achieved with the below modes can be achieved with A. The interest of the other modes is to simplify the user's life esp. when dealing with many patches. I rather recommend using the other modes when these apply.
  • LSS
    • This mode stands for "Linear Stepping Stones". It allows the user to set linear stepping stones with any number of stones and potential asymmetry.
  • OnePatch
    • This is the default. It means no migration scenario apply as there is only one patch (or one deme if you prefer)
  • Island
    • This is a classical island model. The user must input the probability of not migrating.
  • LinearNormal
    • This is just like a linear stepping stone model except that the dispersal kernel follows a normal distribution.

Note also that the migration scenario (as well as the number of patches) is a time specific argument and therefore it can vary through time at will. You can have a look at the manual for more information.

Note that I have been working quite a lot on new versions that I have not put online yet, mainly because I was too lazy to rewrite the manual! The last version online is 3.69 but I am currently working on version 3.139! I will upload a more recent version in the coming week(s). Most notable improvement with the new version include:

  1. Easier user interface; no need for all these quotation marks. Migration scenario will be given by the user as forward migration while it currently is backward migration which is a little less intuitive for the user.
  2. More freedom on demography.
  3. Possibility to simulate several species interaction with ecological relationships. One can simulate a whole ecosystem!
  • $\begingroup$ Thanks Remi for such a detailed response! Is there any justification for a user to have to input the probability of not migrating in the island model? $\endgroup$ Dec 30 '17 at 17:48
  • $\begingroup$ Would you prefer inputting the probability of migrating? It was just to avoid the possibility where the sum of the probability to migrate over all patches is greater than 1 (which would have thrown an error message). It felt more intuitive to me to go this way but it might not be for others. It was also quite intuitive to me to input backward migration rate but I already got a few people telling me forward migration rates are more intuitve so, I've changed it in more recent (not online yet) versions. $\endgroup$
    – Remi.b
    Dec 30 '17 at 19:31
  • $\begingroup$ I'm actually interested in how SimBit accounts for migration "under the hood". While there are many simulation tools already existing, I find that for my particular question, a new tool is needed. $\endgroup$ Jan 3 '18 at 21:16
  • $\begingroup$ At every generation, SimBit loops through each patch, ask how many individuals this patch will contain (this number can be affected by a fair number of different options) and then for each offspring look for where the father come from and where the mother come from. Select the mother and the father and then make them reproduce to get the offspring. I am not sure that really answered your question. If you have more specific questions about how SimBit works under the hood, I'll be happy to answer. $\endgroup$
    – Remi.b
    Jan 3 '18 at 22:51
  • $\begingroup$ Note that it is not so unusual to not have a simulation platform that does exactly everything we want for a particular project. But many of the things you might want to implement might be done quickly by the author and they / I may accept to help. If it is a big change, then the author of the simulation platform may expect authorship on the paper resulting from the project. Note that for the moment, you have not specifically indicated anything that either SimBit, Nemo or Slim could not do. $\endgroup$
    – Remi.b
    Jan 3 '18 at 22:53

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.