Credits

Mesquite was begun in 1997, but its roots go deeper, to the initial development in 1985 of MacClade, which even from the start allowed interactive manipulation of trees and interpretation of character evolution. As features were added to MacClade through the years, most from Wayne Maddison in early versions (1 and 2) and from David Maddison in later versions (2, 3 and 4), MacClade developed an exploratory approach to phylogenetic calculations with a distinctive user interface. After MacClade version 3 was released in 1992, the coauthors worked to ready version 4. Almost all of the many new features in version 4, including the new facilities for molecular sequence editing, were the results of David's efforts. Wayne's efforts on MacClade 4 involved an attempt to graft a modular architecture onto MacClade to allow plug-ins so that its capabilities could be extended. This, we hoped, would allow us and other programmers to add many new tree-based analyses to MacClade. After about a year of work on this, it became clear that grafting this new architecture on to an existing program was not going to work. MacClade was then returned to its original, non-modular state, and it was within this more traditional framework that David completed MacClade 4.

In order to build the desired modular architecture, Wayne had to start from scratch, and so a new project was born in July of 1997. The very first prototype, after one day of work, can be seen here (for the first few days it was called BeanTree, before it became known as Mesquite). Mesquite's vision, exploratory nature, and its user interface borrow extensively from ideas developed in MacClade, but the underlying architecture is quite different. Thus, Mesquite contains a mix of features borrowed directly from MacClade, features we had wanted to put into MacClade but couldn't (e.g., coordinated selection of objects, Trace Character over Trees, likelihood reconstructions), and newly conceived features.

The subsequent chronology of the Mesquite project is:

• August 1998 - first public demonstration (Cambridge University)
• July 1999 - limited seeding to a few developers (prototype version 0.9.2)
• August 1999 - project web page on-line (currently at http://mesquiteproject.org/)
• 29 September 1999 - broader release to developers (prototype version 0.9.5)
• early 2000 - passed 100,000 lines of total code, and 200 total modules.
• 26 June 2000 - Mesquite introduced at Evolution meetings, Bloomington, Indiana (prototype version 0.9.28)
• 14 March 2001 - first public beta version (version 0.95.80)
• 2 April 2001 - public beta version (version 0.96)
• 24 July 2001 - version 0.98 with source code released
• 21 August 2002 - version 0.99 released
• 14 September 2002 - version 0.991 released
• 27 September 2002 - version 0.992 released (internal version (build) d24)
• 10 January 2003 - version 0.993 released (build d42)
• 7 February 2003 - version 0.994 released (build d51)
• 21 May 2003 - version 0.995 released (build e23)
• 21 June 2003 - version 0.996 released (build e30)
• 22 September 2003 - version 1.0 released (build e58)
• 14 January 2004 - version 1.01 released (build e80; build e81 released 17 January to fix cosmetic bug)
• 6 May 2004 - version 1.02 released (build g6; build g7 released 12 May to fix non-substantive bug)
• 1 July 2004 - version 1.03 released (build g19)
• 1 September 2004 - version 1.04 released (build g21)
• 24 September 2004 - version 1.05 released (build g24)

From July 1997 through October 2000, the architectural design, programming and documentation for the basic Mesquite libraries and modules was done by Wayne Maddison, with occasional input from David Maddison. David entered the project in earnest in November 2000. (Other packages of modules for the Mesquite system are due to other authors: for instance, the Rhetenor package of morphometrics modules is by Eric Dyreson and Wayne Maddison.)

Acknowledgments

Mesquite was developed with the assistance of a Fellowship to WPM from the David and Lucile Packard Foundation, and the patience of our families.

The following are gratefully acknowledged for direct assistance with Mesquite's programming or source code. David Swofford helped with the implementation of the likelihood calculations, and supplied code for the optimization routines. John Huelsenbeck gave us portions of the code used in calculation of change probabilities for the Ti/Tv rate matrix model. Maureen O'Leary has given extensive feedback on the use of Mesquite for large morphological data matrices. Image-containing Annotations were inspired by Morphobank, and were introduced into version 1.02 in preparation for planned communication between Mesquite and Morphobank. Lars Rosengreen helped us prepare the code for compilation by means other than Codewarrior (see his website).

For feedback, including bug reports, and other assistance we thank Peter Midford, Michel Laurin, Lars Rosengreen, Cymon Cox, Mario Cozzuol, Katherine St. John, Korbinian Strimmer, Alexei Drummond, Lacey Knowles, Nikolaj Scharff, Gustavo Hormiga, Robin Allaby, Maureen O'Leary, Jonathan Coddington, Chris Hardy, Alan de Queiroz, Stephanie Diezmann, François Lutzoni, Daniel Rafael Miranda-Esquivel, Cam Webb, Jolanta Miadlikowska, Frank Kauff, Matt Hare, Beatrice Kondo, Erin Maxwell, Jennifer Steinbachs, Margaret Thayer, Lukas Ruber, Olga Zhaxybayeva, Galina Glazko, and Steve Kembel.