We consider the problem of estimating the evolutionary history of
a collection of organisms in terms of a phylogenetic tree. This is
a hard combinatorial optimization problem for which different EA
approaches are proposed and evaluated. Using two problem instances
of different sizes, it is shown that an EA that directly encodes
trees and uses ad-hoc operators performs better than
several decoder-based EAs, but does not scale well with the
problem size. A greedy-decoder EA provides the overall best
results, achieving near 100%-success at a lower computational
cost than the remaining approaches.