A 20-state empirical amino-acid substitution model for green plant chloroplasts.

We introduce a new empirical amino-acid substitution model intended for use with for green plant (Viridiplantae) chloroplast protein data, which we call gcpREV (green chloroplast reversible model). Relative exchange rates and base composition frequencies among amino-acids were calculated using a Markov-chain Monte Carlo analysis on a combined data set of 34 proteins sampled among 27 taxa of green plants. The gcpREV model is a better fit than the commonly-used cpREV model to five previously published chloroplast protein test data sets of green plants, but is not a better fit to test data consisting mostly of non-green plant chloroplasts. Consequently, our analyses suggest a degree of specificity of the new model to green plant chloroplast data. The new model is implemented in the software P4, and model description files are available for other popular phylogenetic analysis software.