The phylogeny of land plants inferred from 18S rDNA sequences: Pushing thelimits of rDNA signal?

Previous studies of the phylogeny of land plants based on analysis of 18S r ibosomal DNA (rDNA) sequences have generally found weak support for the rel ationships recovered and at least some obviously spurious relationships, re sulting in equivocal inferences of land plant phylogeny. We hypothesized th at greater sampling of both characters and taxa would improve inferences of land plant phylogeny based on 18S rDNA sequences. We therefore conducted a phylogenetic analysis of complete (or nearly complete) 18S rDNA sequences for 93 species of land plants and 7 green algal relatives. Parsimony analys es with equal weighting of characters and character state changes and parsi mony analyses weighting (1) stem bases half as much as loop bases and (2) t ransitions half as much as transversions did not produce substantially diff erent topologies. Although the general structure of the shortest trees is c onsistent with most hypotheses of land plant phylogeny, several relationshi ps, particularly among major groups of land plants, appear spurious. Increa sed character and taxon sampling did not substantially improve the performa nce of 18S rDNA in phylogenetic analyses of land plants, nor did analyses d esigned to accommodate variation in evolutionary rates among sites. The rat e and pattern of 18S rDNA evolution across land plants may limit the useful ness of this gene for phylogeny reconstruction at deep levels of plant phyl ogeny. We conclude that the mosaic structure of 18S rDNA, consisting of hig hly conserved and highly variable regions, may contain historical signal at two levels. Rapidly evolving regions are informative for relatively recent divergences (e.g., within angiosperms, seed plants, and ferns), but homopl asy at these sites makes it difficult to resolve relationships among these groups. At deeper levels, changes in the highly conserved regions of small- subunit rDNAs provide signal across all of life. Because constraints impose d by the secondary structure of the rRNA may affect the phylogenetic inform ation content of 18S rDNA, we suggest that 18S rDNA sequences be combined w ith other data and that methods of analysis be employed to accommodate thes e differences in evolutionary patterns, particularly across deep divergence s in the tree of life.