[Molecular phylogenies and nucleotide insertion-deletion].

Molecular trees based on the analysis of sequence data can be obtained through parsimony procedures. Such approaches identify evolutionary events (homologies and homoplasies and their location in the tree). Nevertheless, it is not possible generally to analyze directly the information given by the aligned sequences. Noticeably, gaps ("indel") need special coding. Standard procedures for coding gaps offer alternatives which suffer of several weaknesses. In this article a new strategy of coding the sequences is defined in view to express the potential phylogenetic information contained in complex zones with internested insertion/deletion and substitutions, contrary to what is done up to the present. This strategy applies without loss or distortion of information to any case where gaps are present in aligned sequences. According to the hierarchy of internested states of characters (sites), this strategy introduces in the data matrix question marks, "?", which are optimized in fine in the cladogram based on all data. These "?" are not missing data, they are methodological codes, neutral to a priori phylogenetic hypotheses.