bayarea

Bayesian inference of historical biogeography for many discrete areas

This is the software implementation of the method described in our recent Systematic Biology paper:

Landis, M. J., Matzke, N. J., Moore, B. R., & Huelsenbeck, J. P. (2013). Bayesian analysis of biogeography when the number of areas is large. Systematic Biology, 62(6), 789-804. [link]

Quick guide

Download the most recent zip file. Unzip it. Enter the unzipped folder in the command line. Just type

chmod +x bayarea my_run.sh ./my_run.sh

to run a pre-loaded example analysis.

Slow guide

If the included command-line program doesn't work for your computer, you will need to compile the source code to produce an executable. In general, this is an easy process -- just follow the instructions in the manual. If you have gcc installed, type

g++ -O3 *.cpp -o bayarea

from within the source code folder. If you're stumped, feel welcome to email mlandis (at) gmail (dot) com for help.

Source code

BayArea is tested on Mac OS X 10.8 and compiled using gcc-4.2. C++ source code is available under Downloads (stable) and Source (unstable). An executable is provided with the source code, but you may have to compile it yourself.

Contact

All feedback is appreciated -- share your thoughts, questions, and comments regarding the software and model by emailing mlandis (at) gmail (dot) com.

====

Updates

June 29, 2016

Migrate from GoogleCode to GitHub.

This program is no longer being actively supported. Almost all functionality has been ported to RevBayes.

v1.0.3

• Improved MCMC starting states for better chances of the chain reaching stationarity.
• Add support for tuning proposal windows for rate and distance power parameter proposals, via -rateProposalTuner and -distanceProposalTuner.
• Add support for using standard path sampling versus auxiliary variable path sampling as defined in Landis et al. (2013), toggled with the -useAuxiliarySampling flag. The standard path sampling algorithm is guaranteed to work, which roughly follows stochastic mapping algorithm proposed by Nielsen (2002), while the auxillary path sampling algorithm lacks a mathematical proof for convergence but appears to work in practice for datasets I've examined. Credit to Jan Irvahn and Vladimir Minin for identifying this issue.

March 13, 2014

New tutorial hosted at http://treethinkers.org/tutorials/biogeography-with-bayarea/. Give it a try!

September 11, 2013

(no BayArea version update)

• Numerous updates to bayarea-fig, in particular you may now assign colors to partitions of areas.

July 9, 2013

v1.0.2

• File handling now works for all end line symbols.
• Initialization output block now shows the random seed when left uninitialized.
• Added setting to constrain the distance power parameter to be positive (set “-geoDistancePowerPositive=T”, default False). This is useful for global and sparse distributions with high rates of dispersal, which may send the distance power to very large negative values.
• Added setting to approximate distances computations to speed up analyses for large numbers of areas (set “-geoDistanceTruncate=T”, default False).

June 28, 2013

v1.0.1

• Fixed issue where analyses with small numbers of areas (e.g. N<10) would randomly freeze during analysis (thanks to Julien Vien for reporting this).
• Improved input file handling.

May 7, 2013

v1.0.0

• Manual now available
• Default settings set to be conservative for general analysis
• Added settings for prior and proposal densities
• Added lnL scores to area_probs.txt
• Example command-line script added
• Example vireya files added

January 8, 2013

v0.0.1

• Initial code clean-up for reviewers