Constructing and testing phylogenetic trees

Described below is a basic, distance based Neighbour-Joining analysis using bootstrap statistical tests for robustness (Felsenstein, 1985). 

  1. From the main MEGA window, open the ‘Phylogeny’ pull down tab and select ‘Construct/Test Neighbour-Joining Tree’. 
  2. Browse to and open the .meg file you just created. Select the appropriate data type (nucleotide or protein sequence data).
    Note the defaults for missing data, alignment gaps and identity and make any changes if necessary then select ‘OK’. 
  3. A window will open asking you to identify your sequence data as protein-coding or not. 
  4. Another window will open asking for genetic code selection.
    For the example, provided here, using the cytochrome oxidase I (COI) gene, select ‘Invertebrate Mitochondrial’.
  5. A third window opens and allows you to select a number of parameters for your analysis. 
    A minimum of 100 and typically 1,000 iterations of bootstrapping are used to test the robustness of your phylogeny. For now, we will accept the default parameters for our simple analysis. 
  6. A progress window will open for you until the test is completed. 
  7. When complete, a window opens with two tabs showing the ‘Original tree’ generated, as well as a ‘Bootstrap consensus tree’, which is the tree you should refer to.
    Bootstrap support values show the percentage of iterations supporting the shown topology. 
  8. The tree image can be saved as a .pdf for good resolution for presentation (Fig. 3A), can be saved as a .mts Tree Session File for future viewing in MEGA, or exported and saved as a more general Newick (.nwk) format file that is readable by a variety of other tree viewing programs (e.g. FigTree http://tree.bio.ed.ac.uk/software/figtree/ ; TreeDyn http://www.treedyn.org/).

For comparison, a Maximum Likelihood (ML) analysis of the same alignment was performed in a similar manner and is shown in Fig. 3B.

Fig. 3. Phylogenetic reconstruction of cytochrome oxidase I gene from Apis and Bombus species (Hymenoptera; Apidae) using Neighbour-Joining method (A) and Maximum Likelihood (B). Topology of each was tested with 1,000 bootstrap iterations (consensus tree is shown) using Nasonia vitripennis (Hymenoptera; Pteromalidae) as outgroup. Scale represents the substitution rate per site from a total of 981 positions.  A) was computed using Maximum Composite Likelihood (Tamura et al., 2004) with uniform rates among sites and pairwise gap deletion. B) was computed using Tamura-Nei model (Tamura and Nei, 1993) with uniform rates at all sites.

Figure 3