METHODOLOGICAL ASPECTS OF MOLECULAR PHYLOGENY OF FISHES

Fish phylogeny has seen 2 major methodological changes during the last 30 years. First, the introduction of cladistics in Ichthyology in the late 1960s led to dramatic progress in fish classification and phylog eny. Second, molecular methods and especially DNA sequence data offere d new collections of discrete characters useful for phylogenetic inves tigations, especially for phylogenetic problems left unresolved by mor phological characters. But until now, their impact on fish phylogeny h as remained limited. Whatever the aim of a study, fundamental or appli ed, different kinds of molecular methods exist, among which those allo wing identification of molecular structures (such as sequencing) shoul d be preferred, in order to avoid ''experimental screens'' which are d escribed herein. The choice of genes, species, and tree-construction m ethods presents pitfalls that one should avoid. Robustness of phylogen etic trees should be considered. Differences exist between molecularis ts (geneticists) and morphologists as to their respective conceptions of phylogenetic trees. Distance-matrix methods are widely used in the Ist group, and a naive essentialist way to consider sequence alignment and trees is often encountered. Cladistics was born in the world of m orphologists, Most molecularists have not yet reached the correspondin g ''phylogenetic maturity''. Only parsimony methods allow researchers to identify in fine homologuous characters, and are, therefore, really phylogenetic. A phylogeny, as an inference on the history of life, mu st be performed with hypotheticodeductive methods. Parsimony methods s hould therefore be preferred over distance-matrix methods.