R Tip: Labeling Trees w/ Posterior Probability and Bootstrap Support

What's the worst part about making figures for phylogeny papers? When I sat down yesterday to make some new figures I realized that my answer was the same as it was ten years ago: adding support values to each node by hand using programs like Adobe Illustrator. I decided I was never going to go through this archaic error-prone process again. In sharing my solution here, I hope that none of you will either! The text below is an annottated R script that does most of the hard work for you (you'll still have to tweak a few things in Illustrator). Of course, you'll need to get familiar with basic R functions for phylogentics before attempting to use these scripts. The best way to do this is to pick up a copy of Paradis' nice little introductory text^{abe/half/amzn/bn}. Don't be lazy and tell yourself you can label your nodes by hand quicker than you could learn R. This might be true if you only had to label one tree, but what are you going to do for your next paper, or when the reviewers advise you to re-run your analyses? If you know how to use the scripts below you'll just be able to plug any tree in and have your labeled tree almost instantly!

#Fist we need to open some necessary libraries
library(ape)
library(geiger)
#The getAllSubTrees function below is a necessary subfunction that atomizes a tree into each individual subclade and was provided compliments of Luke Harmon.
getAllSubtrees<-function(phy, minSize=2) {
res<-list()
count=1
ntip<-length(phy$tip.label)
for(i in 1:phy$Nnode) {
l<-node.leaves(phy, ntip+i)
bt<-match(phy$tip.label, l)
if(sum(is.na(bt))==0) {
st<-phy
} else st<-drop.tip(phy, phy$tip.label[is.na(bt)])
if(length(st$tip.label)>=minSize) {
res[[count]]<-st
count<-count+1
}}res}

#The plotBayesBoot function below plots both posterior probability and bootstrap values on each node of the consensus tree obtained from your Bayesian analysis. Bootstrap values will appear in bold text immediately below and to the left of the node they support, whereas Bayesian posterior probabilies will appear in regular face above and to the left of the node.

plotBayesBoot <- function(bayesTree,bootTree) {
getAllSubtrees(bayesTree)->bayesSub
getAllSubtrees(bootTree)->bootSub
bootList<-matrix("<50",nnode(bayestree),1)
#The commands below compare all the subclades in the Bayes tree to all the subclades in the bootstrap tree, and vice versa, and identifies all those clades that are identical.
for(i in 1:Nnode(bayesTree)) {
for(j in 1:Nnode(bootTree)) {
match(bayesSub[[i]]$tip.label[order(bayesSub[[i]]$tip.label)], bootSub[[j]]$tip.label[order(bootSub[[j]]$tip.label)])->shared
match(bootSub[[j]]$tip.label[order(bootSub[[j]]$tip.label)], bayesSub[[i]]$tip.label[order(bayesSub[[i]]$tip.label)])->shared2
if(sum(is.na(c(shared,shared2)))==0) {
bootTree$node.label[j]->bootList[i]
}}}
plot(bayesTree, cex=1, lwd=0.5) #Plots your Bayesian consensus tree
nodelabels(bayesTree$node.label, adj=c(1.2, -0.3), frame="n", cex=1, font=1) #Adds posterior probability values to the tree. Change the 'cex' value to make the fond smaller or larger. A value of 1 will give you a readable result in the R quartz window, but a value closer to 0.25 might be better for publication)
nodelabels(bootList, adj=c(1.4, 1.3), frame="n", cex=1, font=2) #Adds bootstrap values.
}

#Now you're ready to read your trees in and make your figure!
read.nexus("yourBayesTree.con")->bayesTree #Reads in the .con file that results from analyses in MrBayes.
bayesTree[[1]]->bayesTree #Extracts one of the two trees in the .con file.
read.nexus("yourBootTree.nex")->bootTree #Reads in the consensus tree from a bootstrap analysis in PAUP.
plotBayesBoot(bayesTree, bootTree)