it is possible not to test a model if it does not make sense with the biological group

[Daniel Eusebio Quintero Vallejo]

Hi Nick, I hope you are ok. I have run six different models (DEC, DIVALIKE, BAYAREALIKEA and their J+ versions) and I got that BAYAREALIKE+J was the best fit model. However, this one shows quite strange results. 

A large number of nodes have as their most probable ancestral range ranges composed of 3 areas. This seems strange to me because many of the extant species do not have that range of 3 areas (only a few). In addition, these are areas that, with the geographic information that we currently have, do not make much sense. I thought that this would be a result of the BAYAREALIKE model allowing only widespread sympatry and not other processes such as subset sympatry and vicariance, therefore, it is assumed that the ancestral ranks are identical to those of the descendants (although they are few, there are some that have a range of 3 areas).

I work with species that are distributed on different islands (each species on no more than one island) and that have probably dispersed by sea between them. Before testing the BAYARELIKE (y +J) model, I performed the model comparison between DEC (y +J) and DIVALIKE (y +J). In this case, the DEC model had the best fit, but both DEC and the other models yielded results that did make a lot of sense, taking into account the geographical and biological history of the group and its distribution. My question is, is it possible to exclude a model because, based on how it's built, it doesn't "fit" my biological group? Is there any information regarding this about the BAYAREALIKE?

Thank you so much!

[Nick Matzke]

Hi! This sounds like maybe the BAYAREALIKE model results are being plotted, but are being called BAYAREALIKE+J.  This can happen if some other step of the script is missed or some such. Typically BAYAREALIKE+J looks very similar to DEC+J and DIVALIKE+J.

If you send me the inputs/outputs & script I can take a look, probably on Monday though...

Cheers,

Nick

--
You received this message because you are subscribed to the Google Groups "BioGeoBEARS" group.
To unsubscribe from this group and stop receiving emails from it, send an email to biogeobears...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/biogeobears/2c44e9e2-7b7c-4637-a987-4f57f6c9a3b0n%40googlegroups.com.

[Daniel Eusebio Quintero Vallejo]

Thanks! I have send you all the files by mail (because they are large files and I just could do it by drive). I forgot to tell you something important: The results that surprise me are both bayearealike and bayarealike+j, they are quite similar and that is why I think that the possible error is not in the parameter J.

I am available for anything you need