Euphytica 118: 9–18, 2001. © 2001 Kluwer Academic Publishers. Printed in the Netherlands. 9 Effects of RNA editing on the coxI evolution and phylogeny reconstruction Alfred E. Szmidt 1* , Meng-Zhu Lu 2 & Xiao-Ru Wang 1 1 Department of Forest Genetics and Plant Physiology, Swedish University of Agricultural Sciences, 901 83, Umeå, Sweden; 2 Forest Research Institute, Chinese Academy of Forestry, Beijing 100091, China; ( * author for correspondence) Received 18 October 1999; accepted 31 May 2000 Key words: coxI, evolution, gymnosperms, phylogeny, RNA editing Summary CoxI genomic and cDNA sequences from gymnosperms and angiosperms were used to study the effects of RNA editing on gene evolution and phylogeny reconstruction. In six gymnosperms harboring edited coxI gene the number of nucleotide substitutions at 1 st ,2 nd and 3 rd codon positions was similar. In contrast, in angiosperms, the number of nucleotide substitutions at 1 st and 2 nd codon positions was much lower than at the 3 rd . The coxI gene in long-lived gymnosperms evolved much faster than in annual angiosperms. This accelerated rate of nucleotide substitution in gymnosperms is due to accumulation of T-C substitutions at edited sites that can randomly appear at all three codon positions. Editing predominantly occurred at 1 st and 2 nd codon positions as a result of selection against nonsynonymous T-C substitutions and other types of mutations. The tree topologies for the investigated species based on genomic DNA data were in concordance with their taxonomic positions. The trees based on polymorphic edited sites agreed with trees derived from complete sequence information. This indicates that edited sites are reliable sources of phylogenetic information especially for species that contain many edited sites. However, the fast evolution rate of coxI gene in gymnosperms has caused the long branches in the phylogenetic trees. The inclusion of the species with a processed paralog i.e., edited form of the coxI gene, affected the topology of phylogenetic trees, especially when the taxon with a processed paralog was closely related to the other species analyzed. Introduction Slow rate of evolution has made plant mitochondrial gene sequences not useful for phylogenetic studies at lower taxonomic levels (Laroche et al., 1995). Con- sequently, there are very few studies addressing vari- ation of mitochondrial coding regions among closely related species. In plants, mitochondrial and chloro- plast genetic information is modified by RNA editing resulting in mRNAs that are different from those en- coded by the corresponding genes in the genomic (g) DNA (Araya et al., 1994; Freyer et al., 1997; Han- son et al., 1996). In both organelles, specific cytidines (C’s) in the primary transcripts are changed to uridines (U’s) in the mature mRNAs (Maier et al., 1996; Stein- hauser et al., 1999). The occurrence of editing has raised concerns about the effect of this phenomenon on the evolution of edited sequences and their use- fulness for phylogenetic reconstruction (Bowe & de- Pamphilis, 1996; Hiesel et al., 1994; Lu et al., 1998; Pesole et al., 1996). Recently, we have sequenced and characterized ge- nomic and cDNAs of the mitochondrial coxI gene in several closely related gymnosperm species (Lu et al., 1998). Most species included in our study possessed a large number of edited sites ranging between 25 and 34 in a 648 base pairs (bp) region of the coxI gene. We also found that the coxI gene in gymnosperms evolves nearly five times faster than in angiosperms. In two taxa, Gingko biloba and Larix sibirica, the coxI gene did not contain any edited sites and resembled an edited transcript (Lu et al., 1998). Such sequences are usually regarded as processed paralogs and have been found to affect phylogeny reconstruction (Bowe