This repository includes Bayesian implementations of State Speciation and Extinction models, including MuSSE, GeoSSE and ClaSSE.
The use of (half-Cauchy or Exponential) hyper-priors on the rate parameters allows to control for over-parameterization.
The mcmc-SSE.R script uses exponential priors on the rate parameters with a gamma hyper-prior.
The script requires a tree file (NEXUS format), a text file with the trait data (see example files), and specifying a model. The models currently available are "musse", "classe", "geosse".
RScript mcmc-SSE.R example_files/bromelioieae_consensus.tre example_files/traitGeoSSE.txt geosse
The taxon sampling (fraction of sampled species out of the total) is provided for each character state using the flag --rho. For example --rho "0.5 0.4 1" in a geosse model specifies a 50%, 40% sampling for species in area 1 and 2, respectively, and a complete sampling for widespread species.
RScript mcmc-SSE.R example_files/bromelioieae_consensus.tre example_files/traitGeoSSE.txt geosse --rho "0.5 0.4 1"
Additional options are:
--i: number of MCMC iterations
--s: sampling frequency
--p: print frequency
R libraries: ape, optparse, picante, diversitree all available on CRAN.
Silvestro et al. 2014 Evolution
Burin et al. 2016 Nature Comm