ABSTRACT
Identifying the most parsimonious tree is computationally demanding. This chapter discusses the most common techniques used, as well as their application in morphological datasets. Only for low numbers of taxa is it possible to guarantee finding the most parsimonious trees by implicitly enumerating all trees without examining the full set of possible trees. For larger numbers of taxa, only heuristic (i.e. trial-and-error) methods can be used; these do not guarantee finding most parsimonious trees but are very effective in practice. The basic methods for heuristic searches are discussed, as well as criteria to evaluate whether the most parsimonious tree is likely to have been found. Explicit criteria for comparing the efficiency of different algorithms are described and used to compare the different search algorithms included in this chapter. General rules for the applicability of different search strategies in morphological datasets are discussed. The implementation of these methods in the computer program TNT allows reliably analyzing most morphological datasets; the details of the implementation in TNT are described.
