Microstructural Changes in Noncoding Chloroplast DNA: Interpretation, Evolution, and Utility of Indels and Inversions in Basal Angiosperm Phylogenetic Inference

• Published in: International journal of plant sciences 2000-11, Vol.161 (S6), p.S83-S96
• Main Authors: Graham, Sean W., Reeves, Patrick A., Burns, Analiese C. E., Olmstead, Richard G.
• Format: Article
• Language: English
• Subjects: Angiosperms; Chloroplasts; Deoxyribonucleic acid; DNA; Evolution; Flowers & plants; Genetics; Genomes; Intergenic DNA; Introns; Nucleotides; Phylogenetics; Tandem repeat sequences; Taxa
• ISSN: 1058-5893; 1537-5315
• DOI: 10.1086/317583

Microstructural changes in several very slowly evolving chloroplast introns and intergenic spacers were characterized across a broad range of angiosperms, including most of the major basal lineages. Insertion/deletion events (indels) in the surveyed noncoding regions of the large inverted repeat (IR) region were shown to be rarer than nucleotide substitutions and thus constitute one of the slowest and least homoplastic types of data available to plant systematists. In our study we scored 180 indels in noncoding regions, of which 36 were parsimony informative within the angiosperms. Because they are relatively few in number, their general utility is currently limited. However, they provide support for specific major taxa, including the angiosperms as a whole, the water lilies, and Illiciaceae and relatives. Support for the basalmost angiosperm split is largely inconclusive, but a single indel supported a basal placement of the water lilies, notAmborella. We estimate that roughly double or triple the current amount (ca. 2.2 kb) of noncoding IR DNA would be required to obtain indel support for most of the deepest branches at the base of the angiosperms. A variety of molecular processes appear to be responsible for the observed indels. Indels are more frequently associated with tandem repeat sequences than not. Insertions are significantly more frequently associated with tandem repeats than are deletions. The latter finding may be, in part, a function of an ascertainment bias for insertions versus deletions. Single‐base indels were the most common size class, but there was an unexplained deficit of some other small indel size classes. Coding indels can be problematical, particularly when they overlap among taxa in an alignment. We favor one simple scheme for coding overlapping indels but argue that no existing scheme for coding overlapping indels for phylogenetic analysis, or dealing with them in alignment, is ideal. Several small inversions were observed. These included the most homoplastic microstructural character in the current study. Each inversion was associated with short flanking inverted repeats.