Tree length

Tree length is the criterion used by the Maximum Parsimony method to search for the best tree. It is defined as the sum of the minimum numbers of substitutions over all sites for the given topology.

To compute the tree length for the unweighted parsimony method, we use the procedure described in Fitch (1971), which is based on the two rules described below. For a given site these rules are applied to each node and the sum of substitutions over all nodes and over all sites are taken. Note that the estimation of the minimum number of substitutions is not affected by the position of the root.

Rule 1. When the two descendent nodes of an ancestral node have some states (nucleotides or amino acids) in common, the ancestral node is assigned to the set of common states. In this case, the most parsimonious explanation does not require any substitutions.

Rule 2. When the two descendant nodes have no states in common, then all states in the descendent nodes are combined to form the set of possible states at the ancestral node. In this case, one substitution is required.