PHYLOGENETIC POSITION Steve L. O'Kane, Jr.2 and Ihsan A. 3 AND GENERIC LIMITS OF Al-Shehbaz ARABIDOPSIS (BRASSICACEAE) BASED ON SEQUENCES OF NUCLEAR RIBOSOMAL DNA1 ABSTRACT The primary goals of this study were to assess the generic limits and monophyly of Arabidopsis and to investigate its relationships to related taxa in the family Brassicaceae. Sequences of the internal transcribed spacer region (ITS-1 and ITS-2) of nuclear ribosomal DNA, including 5.8S rDNA, were used in maximum parsimony analyses to construct phylogenetic trees. An attempt was made to include all species currently or recently included in Arabidopsis, as well as species suggested to be close relatives. Our ®ndings show that Arabidopsis, as traditionally recognized, is polyphyletic. The genus, as recircumscribed based on our results, (1) now includes species previously placed in Cardaminopsis and Hylandra as well as three species of Arabis and (2) excludes species now placed in Crucihimalaya, Beringia, Olimar- abidopsis, Pseudoarabidopsis, and Ianhedgea. Key words: Arabidopsis, Arabis, Beringia, Brassicaceae, Crucihimalaya, ITS phylogeny, Olimarabidopsis, Pseudoar- abidopsis. Arabidopsis thaliana (L.) Heynh. was ®rst rec- netic studies and has played a major role in un- ommended as a model plant for experimental ge- derstanding the various biological processes in netics over a half century ago (Laibach, 1943). In higher plants (see references in Somerville & Mey- recent years, many biologists worldwide have fo- erowitz, 2002). The intraspeci®c phylogeny of A. cused their research on this plant. As indicated by thaliana has been examined by Vander Zwan et al. Patrusky (1991), the widespread acceptance of A. (2000). Despite the acceptance of A. thaliana as a thaliana as a model organism is attributed to the model organism and the sequencing and mapping discovery that it has one of the smallest genomes of its nuclear genome (The Arabidopsis Genome of any ¯owering plant, a low chromosome number Initiative, 2000; Cooke et al., 1996), little is known (n 5 5), and that its genome contains few repetitive about the other species of Arabidopsis sensu lato, sequences and little intergenic spacer DNA. A sur- and their closest relatives. prising recent ®nding by Blanc et al. (2000), how- A small number of molecular phylogenetic stud- ever, showed that although A. thaliana has a re- ies have included a few members of Arabidopsis markably small genome, much of the DNA is Heynh. sensu lato (Price et al., 1994; O'Kane et present in more than one copy. In addition to these al., 1996; Galloway et al., 1998; Koch et al., 1999, important attributes, A. thaliana has a short gen- 2000, 2001; Yang et al., 1999). However, none of eration time (four to six weeks), a small size (dozens these studies attempted to examine all of the taxa can be grown in a small pot), and can easily be either currently or previously included in the ge- grown on synthetic media (Meyerowitz, 1989; Mey- nus, and they included only a small number of oth- erowitz & Pruitt, 1985). The species has been used er, sometimes distantly related, genera. The last extensively in developmental, evolutionary, and ge- comprehensive taxonomic account (Schulz, 1924), 1 Research and ®eldwork were supported by grants from the National Science Foundation (DEB-9208433), the National Geographic Society (NGS-5068-93), and the University of Northern Iowa. Special thanks to Barbara Schaal, in whose lab this work was initiated, and to the support of the Missouri Botanical Garden. We offer gratitude to the many hosts, ®eld companions, and herbarium curators who aided in this study, especially Abdulla Abbas, Nogman Aralbaev, Isa O. Baitulin, Alexandra Berkutenko, Ram Chaudhary, Gheorghe Dihoru, Vladimir Dorofeyev, Yang Guang, Sun Hang, Josef Holub, H. Kato, the late Sigizmund Kharkevich, Franta Krahulec, Hanna Kuciel, Karol Marhold, Zbigniew Mirek, Klaus Mummenhoff, Noriaki Murakami, Nonna Pavlova, and Boris Syomkin. Reviews by Donovan Bailey and an anonymous reviewer improved this paper substantially. 2 Department of Biology, University of Northern Iowa, Cedar Falls, Iowa 50614-0421, U.S.A. [email protected]. 3 Missouri Botanical Garden, P.O. Box 299, St. Louis, Missouri 63166-0299, U.S.A. [email protected]. ANN.MISSOURI BOT.GARD. 90: 603±612. 2003. 604 Annals of the Missouri Botanical Garden which recognized 11 species, is unsatisfactory. As Molecular-based results (O'Kane et al., 1996; many as 50 species have been placed in the genus, Kamm et al., 1995; Mummenhoff & Hurka, 1994) and, although many of these are now placed in oth- agree with LoÈve's (1961) and Hylander's (1957) hy- er genera (Al-Shehbaz et al., 1999), their phylo- pothesis in showing A. thaliana to be most closely genetic relationships remain unresolved. Monophy- related to species placed in Cardaminopsis. In an- ly of the genus has not yet been critically ticipation of results published here and to make the determined, and even basic biological information, names available for ¯oristic works in progress, we such as chromosome numbers, generation time, and previously published the needed nomenclatural in- breeding system of the members of the genus, is novations for the genus Arabidopsis (O'Kane & Al- lacking. Shehbaz, 1997) and have established several new Generic delimitation is perhaps one of the most genera to accommodate excluded species (Al-Sheh- dif®cult and frequently encountered problems in baz et al., 1999). In brief, Arabidopsis includes only the systematics of the Brassicaceae (Al-Shehbaz, A. thaliana and species previously included, or 1973; Rollins, 1993), and Arabidopsis clearly dem- suggested to be, in Cardaminopsis (Jones & Ake- onstrates this problem. There has been a lack of royd, 1993a, 1993b). Species now excluded from agreement among taxonomists on the number of Arabidopsis are placed in Thellungiella O. E. species that belong to Arabidopsis and on the char- Schulz (Al-Shehbaz & O'Kane, 1995), Neotorularia acters that indicate its generic boundaries (e.g., Hedge & J. LeÂonard (Al-Shehbaz & O'Kane, 1997), Ball, 1993; LoÈve, 1961; Rollins, 1993). The generic Ianhedgea Al-Shehbaz & O'Kane (Al-Shehbaz & limits of Arabidopsis have been highly unnatural, O'Kane, 1999), Crucihimalaya Al-Shehbaz et al., and there were no well-de®ned characters separat- Olimarabidopsis Al-Shehbaz et al., and Pseudoar- ing it from several closely associated genera (but abidopsis Al-Shehbaz et al. (Al-Shehbaz et al., see our taxonomic revision based on the results of 1999), and Beringia Price et al. (Price et al., 2001). this current work, O'Kane & Al-Shehbaz, 1997; Al- Our primary objectives are to determine the ge- Shehbaz & O'Kane, 2002a). Some individual Ara- neric limits of a morphologically coherent, mono- bidopsis species have been transferred among sev- phyletic Arabidopsis and to reconstruct a robust in- eral other genera. An example is A. thaliana, which terpretation of its phylogenetic neighborhood. A on the basis of Schulz's (1924) synonymy was pre- well-corroborated phylogeny of the group will allow viously placed in at least nine other genera, in- better evolutionary interpretations to be made of the cluding Arabis L., Conringia Adans., Crucifera E. massive amounts of data now accumulating for A. H. L. Krause, Erysimum L., Hesperis L., Nasturtium thaliana. Workers will know which species to com- R. Br., Pilosella Kostel., Sisymbrium L., and Sten- pare to A. thaliana when making interpretations of ophragma CÏ elak. evolutionary processes. Furthermore, these initial Arabidopsis has been closely associated with steps will provide a better understanding of mor- three different genera, Cardaminopsis (C. A. Mey.) phological character evolution in the Brassicaceae, Hayek, Arabis L., and Halimolobos Tausch. Schulz a family of great economic importance fraught with (1924, 1936) considered its nearest relative to be taxonomic problems related to an under-developed Halimolobos, and separated the latter as being understanding of character evolution and generic coarser herbs with the styles much narrower than delimitation. the fruit, as opposed to Arabidopsis, which were seen as slender herbs with the styles slightly nar- MATERIALS AND METHODS rower than the fruit. These alleged differences are TAXON SAMPLING not mutually exclusive, and species recognized by him in one genus can easily be accommodated in We included representatives of all taxa (at least the other. LoÈve (1961) and Hylander (1957) indi- at the generic level) that are now or have been in- cated a relationship with Cardaminopsis based on cluded in Arabidopsis (e.g., Schulz, 1924; Hedge, natural interspeci®c hybridization. Hedge (1968) 1965; Jafri, 1973; Al-Shehbaz, 1988; Ball, 1993). suggested a closer relationship between Arabidopsis Taxa shown to lie near Arabidopsis in other molec- and Arabis and indicated that the two differ only in ular studies have also been included (Price et al., the cotyledonary position. He further suggested that 1994; O'Kane et al., 1996; Galloway et al., 1998), Arabidopsis wallichii (Hook. f. & Thoms.) Busch as have a sampling of taxa from elsewhere in the probably represents the link between the two gen- Brassicaceae. Phylogenetic trees were initially root- era. An (1987) and Jafri's (1973) transfer of several ed by Cleome lutea Hook. of the Cleomaceae, a species from Arabis to Arabidopsis was probably in- family basal to the Brassicaceae (Rodman et al., ¯uenced by Hedge's view. 1993; Judd et al., 1994; Hall et al., 2002). Included Volume 90, Number 4 O'Kane & Al-Shehbaz 605 2003 Phylogenetic Position of Arabidopsis taxa, as well as voucher information and some no- strap support was obtained from 500 replicates using menclatural comments, are given in Table 1. Where a single round of simple taxon addition. Decay values possible, plant materials were collected in the ®eld (Bremer, 1988; Donoghue et al., 1992) were found and dried in powdered silica gel. In some cases using the AutoDecay program 4.01 of Eriksson tissue was obtained from plants grown from seeds.