Fig. 4

Susceptibility of maximum compatibility and maximum parsimony to the influence of suspect sequence data. The effect of adding 100, 300, or 1000 columns of suspect data is shown for each method. Three replicates (independent random selections of columns) are shown for each number of columns. The topological difference between trees reconstructed with and without the suspect columns is quantified by the number of columns on either tree that are absent from the other and the number on either tree that conflict with the other