Coding of insertion-deletion events of the chloroplastic intergene atpbeta-rbcL for the phylogeny of the Valerianeae tribe (Valerianaceae)

A preliminary analysis of the sequence alignment of the chloroplast intergene atpbeta-rbcL in tribe Valerianeae reveals that insertion-deletion evolutionary events ('indels'), combined with nucleotide substitutions, have occurred in large zones in some of the studied taxa. Due to the frequent occurrence and large size of indels within this tribe, intergene length varies from 531 to 788 base pairs within the studied species. This situation poses gap coding problems that we had to tackle before phylogenetic analysis. Four methods of gap coding were used: elimination of gapped sites ('complete omission'), 'missing data', 'fifth base' and Barriels coding method, which translates indels into new multistate characters in the data matrix. After application of these four methods of data treatment, phylogenetic analyses (maximum parsimony) did not lead to very different results. Three robust clades emerged in each case, corresponding to the Centranthinae subtribe (genus Centranthus), the Fediinae subtribe (genera Fedia and Valerianella), and the American species of Valeriana. The theoretical basis and biological significance of these four methods are discussed in order to apply the best ones in future studies.