Figure 5

Phylogeny inference simulations. Phylogeny inference simulations performed on 30-leaf trees are shown. An inference is considered successful at the first visit to the true tree, i.e., the tree generating the sequence data, or to a state with log-likelihood value at least as good as the true tree. In the left figure, we plot percent successes as a function of number of iterations for 100 runs on one tree with, respectively, MAP using the combined method, i.e., with DP (solid green line with stars indicating that at least one run has reached the success zone during the last 1000 steps), MAP using the r × t-method, i.e., without DP (solid red line with circles) and MCMC with (dotted green line with stars) and without DP (dotted red line with circles). We also plot results obtained with MAP and the combined method but without the SAL-method (solid green line with triangles). In the right figure, the comparison is between the same MAP runs with (again solid green line with stars) and without DP (again red solid line with circles) when a success is recorded as before, and the same methods (dashed green and red lines with stars and circles respectively) where a success is recorded when optimum for the true tree is reached.