Probabilistic historical biogeography: new models for founder-event speciation, imperfect detection, and fossils allow improved accuracy and model-testing

This is the authors' abstract. We don't add key points for this paper.

Historical biogeography has been characterized by a large diversity of methods and unresolved debates about which processes, such as dispersal or vicariance, are most important for explaining distributions. A new R package, BioGeoBEARS, implements many models in a common likelihood framework, so that standard statistical model selection procedures can be applied to let the data choose the best model. Available models include a likelihood version of DIVA (“DIVALIKE”), LAGRANGE’s DEC model, and BAYAREA, as well as “+J” versions of these models which include founder-event speciation, an important process left out of most inference methods. I use BioGeoBEARS on a large sample of island and non-island clades (including two fossil clades) to show that founder-event speciation is a crucial process in almost every clade, and that most published datasets reject the non-J models currently in widespread use. BioGeoBEARS is open-source and freely available for installation at the Comprehensive R Archive Network at http://CRAN.R-project.org/package=BioGeoBEARS. A step-by-step tutorial is available at http://phylo.wikidot.com/biogeobears.