Standard models of DNA substitution are not appropriate for the phylogenetic analysis of ribosomal-encoding DNA (rDNA) sequences because of the non-independence of pairing-nucleotides maintaining higher order structure in these molecules. Several models of rDNA evolution (that are dependent on predicted secondary structure) have recently been proposed, and 2 programs now exist to simultaneously combine pairing and non-pairing regions using more appropriate likelihood models for phylogenetic analysis. Specifically, we present an approach that incorporates both a standard 4x4 substitution model for non-pairing regions and a more complex rRNA model that treats pairing-regions as linked characters in the substitution matrix. We explore a wide range of existing rRNA models and report on the efficiency of each model as it pertains to the helical regions of the nuclear 28S expansion segments D2 and D3 in 230 rootworm beetles (Coleoptera: Chrysomelidae). The results of these rRNA-specific models are compared to standard methods of phylogeny reconstruction. In general, phylogenetic methods that do not account for the dependence of pairing-nucleotides in rRNA data yield trees with inflated resolution and branch support measures (bootstrap, Bremer support, posterior probabilities). Thus, the accommodation of character non-independence is more suitable for the phylogenetic analysis of rRNA data than standard methods (i.e., parsimony, weighting, DNA likelihood models) that ignore the coevolution (hence dependence) of pairing-nucleotides.
Keywords: rRNA secondary structure, rRNA likelihood models
Back to Ten-Minute Papers, Section A. Systematics, Morphology, and Evolution
Back to Ten-Minute Papers, Section A. Systematics, Morphology, and Evolution
Back to The 2003 ESA Annual Meeting and Exhibition