Figure 27.10. Parsimony analysis. Two possible trees relating the sequences from Fig. 27.9 are compared. (Left column) Possible tree #1; (right column) possible tree #2. (A,B) Two possible trees. (C,D) Character states from alignment column #1 are overlaid onto the trees. (E,F) Possible ancestral character state reconstructions are shown for each tree. Character states are indicated on the tree for ancestral nodes. State changes are indicated by arrows. For tree #1, one reconstruction (Ea) requires only one character state change, whereas the other (Eb) requires two. Thus for this tree we would infer that only one character state change is required to fit the data in alignment column 1 to the tree. For tree #2, on reconstruction requires two state changes (Fa), whereas the other requires three (Fb). Thus for this tree we would infer that two character state changes are required to fit the data in alignment column 1 to the tree. Since tree #1 only requires a single state change, this tree would be favored as more parsimonious over tree #2.