BEAST: BAYESIAN ECOLOGICAL ANALYSIS OF STATISTICAL TRENDS
                                                                                                                                          

Overview
Get the Software
Installation Instructions
For Developers
FAQ
Contact Information
Version History & Bugs
Acknowledgements

Overview

This is the homepage for the BEAST software, developed by Eric Ward at the University of Washington (Seattle). BEAST is free software, designed to provide a flexible modelling tool for theoretical biologists to evaluate population models in a Bayesian framework. A variety of models are supported: single stage / age models, stage-structured population models, and age-structured population models. For the age- and stage- structured models, density dependence may be incorporated into any stage / age, and may assume one of several density-dependent functions. All models support variable change points (broken stick models), time lags, and error may be incorporated in the form of either observation error or process error. A wide range of symmetric and asymmetric likelihoods are supported, and any one of the ~ 25 prior distributions may be placed on any parameter in the model. BEAST also allows the user to do MCMC parameter estimation, population projections, and to conduct formal Bayesian decision analyses.

All models are currently limited to single-species models (future versions may contain multi-species models). A further limitation of BEAST is that the only data currently supported are abundance estimates (which may contain error) and optional additional mortality / harvest data (which does not contain error).


Get the Software

Download the most recent version (2.0.1) here:

BEAST.exe

BEAST source code

BEAST Instruction Manual

Download the previous version, 1.1.3 here:

Download BEAST 1.1.3.zip and the 1.1.3 instruction manual (different from the insructions below).


Installation Instructions
1)    Nothing needs to be installed if you're just wanting to run the executable file on a PC. If you want to run the Beast.jar file, or run the program on another platform, you need to download Sun's JDK.

For Developers
If you're interested in contributing to the project, or learning more about Java, visit the BEAST Developer's Page.

FAQ
1)   Is there a citation?
Yes. A full description of the software, including an example with the theta-logistic model is available in the October 2006 issue of Wildlife Society Bulletin. Please cite as follows:.

Ward, E.J. 2006. A new BEAST: Bayesian software tools for ecological trend analysis. Wildlife Society Bulletin, xx:xx-xxx.

2)    What language is BEAST written in?
BEAST 1.* was written in both Java and C.  Java's object-oriented approach was used to design the majority of the classes used, creating a modular program that is easily expandable.  Java's JNI (Java Native Interfaces) was used to interface with C code to perform all mathematical operations. To make the program clearer and easier for users to modify, I released BEAST 2.0 as a 100% Java application.   

3)    What platforms are supported? 
BEAST was written to be compatible with Windows XP, and although it hasn't been tested, the 2.0 application should be able to run on Unix and Mac platforms. 

4)    Is the source code available?
Yes.  This software was designed to be shared / improved, but is not intended for commercial purposes.

5)    Would it be possible to add more distributions and / or density functions?
Yes.  If you want to do it yourself, add a new function in the densityFunction class, and include the function in the model class (specifically the getValue method). If you don't want to do it yuorself, please send all requests to ericward@beastsoftware.org .

6)    What if I want to incorporate additional data, besides mortality and abundance estimates?
Please send all requests to ericward@beastsoftware.org .

7) What current modifications are being done to BEAST?
I am working on expanding the collection of models to include some hierarchical GLMs, which have seen widespread use in biology.    


Contact Information / Troubleshooting
Please direct all comments, questions, or problems to
Eric Ward: ericward@beastsoftware.org.

Version History & Bug Fixes

Sept 3 2006: there is a small bug in the MCMC sampling of discrete distributions - this

isn't critical, but will make sampling very inefficient. Will be fixed soon.

v. 2.0.1 released, allowing model histories to be saved as text files (August 3 2006)

v. 2.0.0 released as a 100% pure Java program (June 12 2006)

v. 1.0.5 released, fixing a bug in the quadratic allee population model

v. 1.0.4 released, enabling the jumping parameters to be updated during the MCMC sampling process

v. 1.0.3 released, fixing the problem of negative population sizes for some projections

v. 1.0.2 released, making sampling with gamma distributions more efficient.

v. 1.0.1 released (July 2005)


Acknowledgements
Funding for this software was provided by the National Marine Fisheries Service, Washington State Sea Grant, and additional support from the University of Washington .  Additional support provided by the School of Aquatic and Fishery Sciences at the University of Washington (Ray Hilborn's lab) and the Alaska Fisheries Science Center. A big thanks also to Mike Gilpin for introducing me to programming and Java while I was an undergraduate at UCSD.