7.6.3. Building trees using distance and character based methods

To assess the reliability of the tree topology, users should be aware of the phylogenetic tree construction and tree analysis methods according to the data and algorithmic strategy used.  Each method has different assumptions that may or may not be valid for the evolutionary process of the given sequence data. For example, the distance based method UPGMA (Unweighted Pair Group Method with Arithmetic mean) assumes a neutral mutation rate proportional to time (a molecular clock). Therefore, it is important to be aware of this fact when evaluating tree topology generated by each method. It is encouraged that one run a variety of distance based methods (Neighbour-Joining, UPGMA, Minimum Evolution) that calculate evolutionary distance between sequences, and character based methods (Maximum Parsimony, Maximum Likelihood, Bayesian) that determine the most probable evolutionary event history between sequences (Table 1).

Table 1
.  Classification of phylogenetic analysis methods and strategies.

Tree building strategy    



Distance based

Character based

Clustering algorithm

Neighbour Joining    


Optimality criterion

Minimum Evolution    

Maximum Parsimony     
Maximum Likelihood
Bayesian Analysis

Statistical testing of topology should also be performed where possible, i.e. bootstrapping analysis.  Altering the substitution model, rates and patterns, and treatment of gaps/missing data may also be warranted, though the varying justifications for each of these tests is beyond the detail provided here.  Low branch support for any topology shown in the final tree or conflicts in topology determined by multiple testing should be addressed when presenting any phylogenetic data.