Bayesian-based trees

© Joseph T. & Suzanne L. Collins / Photo Researchers, Inc.

The paper:
M.C. Brandley et al., "Partitioned Bayesian analyses, partition choice, and the phylogenetic relationships of scincid lizards," Systematic Biol, 54:373-90, 2005. (Cited in 73 papers)

The finding:
Tod Reeder's group at San Diego State University and a Swiss colleague suggested a revised taxonomy for lizards called skinks. By testing novel methods for partitioning genetic data into discrete segments used in Bayesian analysis, they found, among other results, that one skink genus, Eumeces, was actually composed of lizards belonging to several genera.

The significance:
"No one had ever put together a comprehensive DNA phylogeny of the group," says Matt Brandley, now at the University of California, Berkeley, and the first author of the paper. "Anyone who builds phylogenies, whether you're studying bacteria or apes, can benefit from using that technique."

The applications:
David Marshall, a University of Connecticut postdoc studying the phylogenetics of cicadas, says that Brandley's paper is one of the few papers that carefully considers the effects of data partitioning in Bayesian analysis. "It has definitely guided my work," says Marshall, who has used the technique to model evolutionary relationships in his insects.

The future:
Brandley is further revising Eumeces and is using partitioned data and Bayesian analysis to answer "molecular clock" questions regarding the timing of biogeographic events in the evolution of the lizards, such as speciation.