Bayesian stochastic mapping for estimating biogeographic history on phylogenies
1 NIMBioS, Univ. of Tennessee
in Historical and Paleo-Biogeography
10.01.2015, 13:30-13:45, H 22, RW II
Traditional likelihood methods in historical biogeography estimate the probability of each geographic range at each node. Usually the most-probable range at each node is plotted, and this is taken to be the approximate history. This is not technically accurate and might be badly misleading in some cases. A solution is stochastic mapping of possible histories on the phylogeny. This has been widely applied in phylogenetics for sequence data and discrete characters, but these character models are inappropriate in historical biogeography, where the state space is much more complex, and geographic range changes through both anagenetic and cladogenetic events. I present a novel algorithm that enables stochastic mapping on any biogeographic model available in BioGeoBEARS, as well as graphical display and statistical summary of the timing and frequency of dispersal and vicariance events. An animation of realizations of possible histories under the DEC and DEC+J models is demonstrated for Hawaiian Psychotria shrubs. R functions and an example script performing stochastic mapping are available at http://phylo.wikidot.com/biogeobears. The functions build upon on the R package BioGeoBEARS, available for all platforms at CRAN.
Keywords: BioGeoBEARS, historical biogeography, biogeographical stochastic mapping, ancestral state inference
Export as iCal: