Figure 6.

Pairwise comparison of inference methods when gapsare coded as distinct evolutionary events. values are compared, in a pairwise fashion, for the inference methods: MP, Bayesian analysis, and the MLε analysis, when the gaps were treated as binary characters or by the DNAMLε method. As in Figure 4 and 5, the average phylogenetic accuracy, , for one method is plotted against that of another in each panel. For each pairwise comparison between the inference methods, the left panel shows the results for the insertion-deletion rate ratio is 1:1 and the right panel when it is 1:3. The dots in each graph are color coded to reflect the gap percentage (G/S) against which the values have been measured, ranging from light blue (for the lowest G/S values) to red (for the highest G/S values). Each value reflects one of all possible combinations of values of substitution rate, r and indel rate, λ, (see Additional file 1), sequence length, l = 500, transition-transversion rate ratio, κ = 2, and the gamma among-site rate variation shape parameter, α = 0.5, averaged over 100 replicates, for a total of 110 data points in each graph. The paired t-test (p < 0.05) results are shown with a letter (within the dot) that signifies if a particular method is statistically better than the other in a given comparison (B – Bayesian analysis, L -MLε, and M – Maximum Parsimony) for the parameter combination. The paired t-test results that were not significant are presented as dot with no symbol. The results of the Z test (p < 0.001) over 110 data points for each method-method comparison is shown with a letter followed by an asterisk (B* – Bayesian analysis, L* – MLε, and M* – Maximum Parsimony).