Int. J. Plant Sci. 166(1):117–133. 2005. Ó 2005 by The University of Chicago. All rights reserved. 1058-5893/2005/16601-0011$15.00 PHYLOGENY OF TRAGOPOGON L. (ASTERACEAE) BASED ON INTERNAL AND EXTERNAL TRANSCRIBED SPACER SEQUENCE DATA Evgeny V. Mavrodiev,1 ,*,y Mark Tancig,* Anna M. Sherwood,z Matthew A. Gitzendanner,2 ,*,y Jennifer Rocca,*,y Pamela S. Soltis,3 ,y and Douglas E. Soltis4 ,* *Department of Botany, University of Florida, Gainesville, Florida 32611, U.S.A.; yFlorida Museum of Natural History, University of Florida, Gainesville, Florida 32611, U.S.A.; and zSchool of Biological Sciences, Washington State University, Pullman, Washington 99164, U.S.A. Tragopogon L. (Lactuceae, Cichorioideae, Scorzonerinae) is an Old World genus of ca. 150 species. To elucidate relationships in this poorly understood genus, we analyzed internal and external transcribed spacer sequences (ITS and ETS, respectively) from 59 species from 14 of the 17 sections recognized by Borisova. Podospermum and Lactuca were used as outgroups. ITS and ETS sequence data provide strong support for the monophyly of Tragopogon under parsimony optimality criteria. Sequence data also support the monophyly of several of the recognized sections within Tragopogon, including the sections Tragopogon, Majores, Hebecarpus, Chromopappus, and Collini. However, the circumscriptions of these groups are often altered from the taxonomic treatment of Borisova. Most members of the recognized sections Brevirostres and Profundisulcati appear within the Brevirostres and Profundisulcati clades, respectively. Species of section Nikitinia are also included in the Brevirostres clade. Sequence data also provide evidence for an Angustissimi clade of T. pusillus, T. filifolius, diploid T. latifolius (2n ¼ 12), T. segetum, T. acanthocarpus, T. serotinus, and T. charadzae, species not previously considered closely related. Our analyses also indicate possible relationships for 17 species that had not been previously placed in any section. The diploids T. porrifolius, T. pratensis, and T. dubius, which were introduced into North America and ultimately formed the recent alloploids T. mirus and T. miscellus, are not closely related within Tragopogon; T. dubius and T. porrifolius appear in sister clades, and T. pratensis is more distantly related. Characters often used to establish sectional groupings, such as the length of the achene beak, evolved multiple times in Tragopogon. Keywords: Tragopogon, internal transcribed spacer and external transcribed spacer sequence data, phylogeny. Introduction terized by a distinct ‘‘Tragopogon’’ type of pollen that has six abpolar, three equatorial, and six interapertural lacunae (Blackmore 1982; Nazarova 1997). Koelpinia, another genus of Scorzonerinae, has the same type of pollen (Blackmore 1982; Nazarova 1997), but this may represent an independent origin, because Koelpinia and Tragopogon are not sister taxa within subtribe Scorzonerinae (Mavrodiev et al. 2004). There are several systematic treatments of Tragopogon, although none of these is comprehensive, involving all recognized species. De Candolle (1838) split Tragopogon into two large groups that were not formally named. In one group he placed all species possessing peduncles thickened below the flowering capitula. In the second group the peduncles are not thickened below the capitula. Later, Boissier (1875) split Tragopogon into two large groups of unspecified rank based on flower color: (1) ‘‘species with yellow flowers’’ (Flaviflora group) and (2) ‘‘species with purplish flowers’’ (Rubriflora group). Following these initial treatments, other early investigators of Tragopogon (Nikitin 1930; Artemczhyk 1937; Regel 1937, 1955) described a large number of new species but did not provide any system of classification for the genus. Artemczhyk (1948) split all Ukrainian species of Tragopogon into three groups, Majores, Orientalis, and Dasyrhynchiformes, and recognized these groups as separate series, but gave descriptions for only the Ukrainian members of these series. The brief paper of Artemczhyk (1948) is also the first Tragopogon L. (Lactuceae, Cichorioideae, Scorzonerinae) is an Old World genus of ca. 150 species that occurs across Eurasia from the Atlantic to the Pacific Ocean, with a center of distribution in the Mediterranean region, the Middle East, and eastern Europe. Tragopogon includes biennial and perennial herbs with linear or linear-lanceolate leaves; solitary, simple, or sparingly branched stems; one or only a few capitula; and receptacles without scales. The monophyly of the genus was strongly supported in a recent phylogenetic analysis of Scorzonerinae based on internal transcribed spacer (ITS) sequence data (Mavrodiev et al. 2004). Species of Tragopogon are also united by several morphological characters. The achenes of Tragopogon are usually fusiform, with five to 10 more or less distinct ribs and a beak of varying length. The involucral bracts are always in one row, ligulate flowers are yellow or purplish, and the pappus is in one row of mostly plumose hairs (Richardson 1976). Tragopogon is also charac1 2 3 4 Author for correspondence; e-mail evgeny@ufl.edu. E-mail magitz@ufl.edu. E-mail psoltis@flmnh.ufl.edu. Author for correspondence; e-mail dsoltis@botany.ufl.edu. Manuscript received February 2004; revised manuscript received August 2004. 117 118 INTERNATIONAL JOURNAL OF PLANT SCIENCES evolutionary treatment of Tragopogon. Artemczhyk (1948) proposed that the center of origin of Tragopogon is the eastern Mediterranean, and he considered Tragopogon majus Jacq., T. orientalis L., and T. dasyrhynchus Artemcz. to be ancestral to all of the European species of the genus. Kuthatheladze (1957) proposed a classification of Tragopogon based on the analysis of 32 species, which she distributed among five sections: Brevirostres, Collini, Angustissimi, Majores, and Profundisulcati. According to Kuthatheladze (1957), Tragopogon is typically a Mediterranean genus, the oldest section in Tragopogon is Brevirostres, and the youngest is Profundisulcati. She also proposed that ancestral character states in Tragopogon include the perennial herbaceous habit, yellow ligules, and an achene having a short beak. In the Flora of the USSR, Borisova (1964) proposed a new system of classification for Tragopogon, based on the analysis of 79 species, all of which occurred in the territory of the former Soviet Union. This treatment included sections Angustissimi Kuth. (three species), Bessera Boriss. (four species), Brevirostres Kuth. (13 species), Chromopappus Boriss. (two species), Dasypogon Boriss. (one species), Dybjanskya Boriss. (one species), Hebecarpus Boriss. (one species), Kemularia Boriss. (one species), Krascheninnikovia Boriss. (one species), Macropogon (Kuth.) Boriss. (one species), Majores Kuth. (five species), Nikitinia Boriss. (four species), Profundisulcati Kuth. (11 species), Collini Kuth. (¼Rubriflori [Boiss.] Boriss.) (11 species), Sosnovskya Boriss. (four species), Tragopogon (14 species), and Tuberosi (Kuth.) Boriss. (two species). Rechinger (1977) followed Borisova’s (1964) system of classification for Tragopogon in his Flora Iranica. Klokov (1981) suggested that species of sections Nikitinia, Bessera, and Brevirostres, as well as some species of section Tragopogon, form one broad section, but he did not formally name this group. He also suggested that the members of this unnamed group derived from ancestral taxa having achenes with long beaks. Tzvelev (1985) proposed a taxonomic treatment of Tragopogon from the European portion of Russia based on 23 species. He accepted sections Brevirostres, Tragopogon, Majores, and Collini (¼Rubriflori) but disagreed with Borisova’s (1964) recognition of sections Hebecarpus, Nikitinia, and Bessera. Tzvelev (1985) suggested that many diploid species in the genus were of hybrid origin, but he did not suggest possible parents for most of the known polyploids. He proposed that the ancestral Tragopogon was a perennial herb with yellow ligules, achenes without beaks, and a chromosome number of 2n ¼ 14, with the more common chromosome number of 2n ¼ 12 as the derived state. Based on the study of Tragopogon in the Caucasus, Nazarova (1991) accepted sections Tragopogon, Tuberosi, Collini (¼Rubriflori), Angustissimi, Majores, Chromopappus, and Profundisulcati of Borisova (1964). In contrast to Tzvelev (1985), Nazarova (1991) did not believe that hybrid speciation was important in diploid Tragopogons. She also agreed that several species of Tragopogon were allopolyploids but did not suggest any parents. The chromosome number reported for almost all species of Tragopogon is 2n ¼ 12. However, for T. pratense (Ishikawa 1916 in Nazarova 1991; Clapman 1952 in Nazarova 1991) and T. crocifolius (Darlington 1956), both 2n ¼ 12 and 14 were reported. However, the validity of reports of 2n ¼ 14 for these two species is de- batable. The counts of 2n ¼ 14 may simply reflect the presence of B chromosomes in several populations (Nazarova 1991). Several polyploids have been reported in Tragopogon. Ownbey (1950) showed that two allotetraploid species of Tragopogon, T. mirus and T. miscellus, formed in North America in the 1900s following the introduction of three diploids from Europe. The parents of T. miscellus are T. dubius and T. pratensis; the parents of T. mirus are T. dubius and T. porrifolius (Ownbey 1950; Ownbey and McCollum 1953; reviewed in Soltis et al. 2004). Nazarova (1991) indicated that at least 10 species of Tragopogon from Europe are polyploids or include polyploid cytoptypes: T. castellanus Levier (2n ¼ 24), T. buphtalmoides (DC.) Boiss. (2n ¼ 24, 36), T. coloratus C. A. Mey. (2n ¼ 12, 24), T. cupani Guss. ex DC. (2n ¼ 12, 24), T. gracilis D. Don (2n ¼ 24), T. graminifolius DC. (2n ¼ 12, 24, 36), T. pusillus Bieb (2n ¼ 12, 24), T. reticulatus Boiss. et Huet. (2n ¼ 12, 24, 36, 56–58), T. latifolius Boiss. (2n ¼ 12, 24), and T. tuberosus C. Koch (2n ¼ 24). Nazarova (1991) further suggested that some of these polyploids may have subsequently been involved in hybridization events, yielding a complex of additional cytotypes: T: buphtalmoides 3 T: latifolius (2n ¼ 24), T: buphtalmoides 3 T: reticulatus (2n ¼ 24, 27–32, 35, 36). However, the parentage of all of these Eurasian polyploids is unknown, and they have been little studied. In a recent molecular phylogenetic investigation of subtribe Scorzonerinae based on ITS sequence data, we demonstrated that Tragopogon is monophyletic and distinct from other genera, including Geropogon (Mavrodiev et al. 2004), a monotypic genus sometimes included within Tragopogon (Richardson 1976). The independence of Geropogon was previously suggested on the basis of karyological and morphological data (D’Amato 1975 in Nazarova 1991; Wilson 1983). Relationships within Tragopogon are poorly understood. Many species of Tragopogon have not been placed in a section; most of these are narrow endemics that have been recognized and named but not treated taxonomically. Furthermore, the genus has never been the subject of a comprehensive monograph. Regional floras have provided treatments for species only in those geographic areas. The phylogenetic utility of ITS and external transcribed spacer (ETS) sequence data has been demonstrated in a number of studies of genera of Asteraceae (Clevinger and Panero 2000; Lee et al. 2002, 2003; Markos and Baldwin 2001; Saar et al. 2003; Urbatsch et al. 2003). In an initial effort to resolve relationships within Tragopogon, we conducted a phylogenetic analysis of 59 species, using ITS and ETS sequence data. We also hoped to elucidate the relationships among T. dubius, T. pratensis, and T. porrifolius, the diploid progenitors of the recently formed North American polyploids T. mirus and T. miscellus (Ownbey 1950). Material and Methods Species and Samples We followed the treatment of Tragopogon given in the Flora of the USSR (Borisova 1964; Rechinger 1979), because these are the most comprehensive treatments of the genus (table 1). We sampled broadly across the entire genus, including Table 1 Comparison of Treatments of Tragopogon Taxa T. acanthocarpus Boiss. T. afghanicus K.H. Rechinger et Koie T. armeniacus S. Kuthath. T. bornmuelleri M. Ownbey et K.H. Rechinger T. brevirostris DC. T. capitatus S. Nikit. T. charadzae S. Kuthath. T. collinus DC. T. coloratus C. A. Mey. (2n¼12) T. dasyrhynchus Artemcz. T. dubius Scop. T. T. T. T. T. T. T. T. T. T. dubjanskyi S. Niktin elongatus S. Nikit. filifolius Rehm. ex Boiss. floccosus Waldst. et Kit. heterospermus Schweigg. kemulariae S. Kuthath. ketzkhovelii S. Kuthath. kindingeri kotschyi Boiss. krascheninnikovii S. Nikit. T. latifolius Boiss. (2n¼12) T. longirostris Bischoff ex Sch. Bip. T. major Jacq. T. makaschwilii S. Kuthath. T. T. T. T. T. marginatus Boiss. et Buhse meskheticus S. Kuthath. montanus S. Nikit. orientalis L. paradoxus S. Nikit. T. T. T. T. T. podolicus Bess. ex DC. porrifolius L. pratensis Linn. pterocarpus DC. pusillus Bieb. (2n¼12) T. T. T. T. T. T. T. T. T. reticulatus Boiss. et Huet ruber S. T. Gmel. ruthenicus Bess. ex Claus segetum S. Kuthath. serotinus Sosn. sinuatus sosnowskyi S. Kuthath. trachycarpus S. Nikit. undulatus Jacq. Borisova 1964 Kuthatheladze 1957 Sect. Profundisulcati Kuthath. Sect. Majores (Artemcz.) Kuthath. Sect. Profundisulcati Kuthath. Sect. Profundisulcati Kuthath. Sect. Brevirostres Kuthath. Sect. Majores (Artemcz.) Kuthath. Sect. Sosnovskya Boriss. Sect. Rubriflori Boiss. Sect. Chromopappus Boriss. Sect. Brevirostres Kuthath. Rechinger 1977; Tzvelev 1985 Sect. Profundisulcati Kuthath. Sect. Majores (Artemcz.) Kuthath. Sect. Profundisulcati Kuthath. Sect. Brevirostres Kuthath. Sect. Majores (Artemcz.) Kuthath. Sect. Nikitinia Boriss. Sect. Rubriflori Boiss. Sect. Brevirostres Kuthath. Sect. Brevirostres Kuthath. Sect. Bessera Boriss. Sect. Kemularia Boriss. Sect. Profundisulcati Kuthath. Sect. Majores (Artemcz.) Kuthath. Sect. Brevirostres Kuthath. Sect. Collini Kuthath. Sect. Majores (Artemcz.) Kuthath. Sect. Brevirostres Kuthath. Sect. Majores (Artemcz.) Kuthath. Sect. Rubriflori Boiss. Sect. Chromopappus Boriss. Sect. Brevirostres Kuthath. Sect. Majores (Artemcz.) Kuthath. Sect. Brevirostres Kuthath. Sect. Brevirostres Kuthath. Sect. Profundisulcati Kuthath. Sect. Profundisulcati Kuthath. Sect. Brevirostres Kuthath. Sect. Brevirostres Kuthath. Sect. Kemularia Boriss. Sect. Sosnovskya Boriss. Sect. Krascheninnikovia Boriss. Sect. Profundisulcati Kuthath. Sect. Krascheninnikovia Boriss. Sect. Majores (Artemcz.) Kuthath. (as syn. T. dubius Scop.) Sect. Profundisulcati Kuthath. Sect. Sosnovskya Boriss. Sect. Profundisulcati Kuthath. Sect. Rubriflori Boiss. Sect. Tragopogon Sect. Majores (Artemcz.) Kuthath. Sect. Brevirostres Kuthath. Sect. Hebecarpus Boriss. Sect. Tragopogon Sect. Chromopappus Boriss. Sect. Tuberosi (Kuthath.) Boriss. Sect. Sosnovskya Boriss. Sect. Rubriflori Boiss. Sect. Nikitinia Boriss. Sect. Angustissimi Kuthath. Sect. Brevirostres Kuthath. Sect. Angustissimi Kuthath. Sect. Tragopogon Sect. Majores (Artemcz.) Kuthath. Sect. Profundisulcati Kuthath. Sect. Krascheninnikovia Boriss. Sect. Majores (Artemcz.) Kuthath. (as syn. T. dubius Scop.) Sect. Majores (Artemcz.) Kuthath. Sect. Collini Kuthath. Sect. Profundisulcati Kuthath. Sect. Sosnovskya Boriss. Sect. Rubriflori Boiss. Sect. Tragopogon Sect. Majores (Artemcz.) Kuthath. Sect. Profundisulcati Kuthath. Sect. Collini Kuthath. Sect. Brevirostres Kuthath. Sect. Sect. Sect. Sect. Sect. Brevirostres Kuthath. Majores (Artemcz.) Kuthath Tragopogon Chromopappus Boriss. Tuberosi (Kuthath.) Boriss. Sect. Sosnovskya Boriss. Sect. Collini Kuthath. Sect. Brevirostres Kuthath. Sect. Angustissimi Kuthath. Sect. Brevirostres Kuthath. Sect. Angustissimi Kuthath. Sect. Angustissimi Kuthath. Sect. Brevirostres Kuthath. Table 2 Voucher Data for Individuals Sequenced in This Study GenBank accession numbers Species T. T. T. T. T. T. T. T. T. T. T. T. T. T. T. T. acanthocarpus afghanicus albinerve armeniacus aureus australis balcanicum bornmuelleri brevirostris capitatus charadzae coelesyriacus collinus coloratus (2n¼12) crocifolius dasyrhynchus T. dubius T. dubjanskyi T. elongatus T. T. T. T. T. T. T. T. T. T. T. fibrosum filifolius floccosus hayekii heterospermus jesdianus kemulariae kindingeri kotschyi krascheninnikovii lamottei T. T. T. T. T. T. T. T. T. T. T. T. T. latifolius (2n¼12) longifolius longirostris major makaschwilii marginatus meskheticus minor montanus olympicus orientalis podolicus porrifolius T. pratensis T. pterodes T. pusillus T. reticulatus (2n¼12) T. ruber T. ruthenicus T. samaritani T. segetum Voucher ITS region of rDNA (ITS1, 5.8S, ITS2) ETS region of rDNA E. Nazarova 1988, s.n. (ERE) ‘‘Asia’’: K. H. Rechinger, 2193999 (USN) Turkey: V. A. Mattews (K) Armenia, 1972: A. Chechurov (LE) Turkey, 1993: A. Chochrjakov (MHA) 549760 (USN) Turkey, 1955: A. Chochrjakov (MHA) Iran, 1957: K. H. Rechinger, 1507 (K) M. Ownbey 274243 (WS) Kazachstan, 1930: S. Nikitin, 12 (MW) Georgia, 1948: S. Kuthatheladze (LE) M. Ownbey, 274106 (WS) Azerbajdzhan, 1952: N. N. Tzvelev (LE) Armenia, 1979: E. Nazarova, 856 (ERE) M. Ownbey 274740 (WS) Russia: Volgograd, 2002: E. Mavrodiev, field collections (Soltis lab.), det. N. N. Tzvelev M. Ownbey, 274197 (WS) Volgograd, 2002: E. Mavrodiev collections (Soltis lab.) Kazachstan, 1981: V. Bochkin, I. Rusanovich (MHA) Turkey, 1996: A. Khochrjakov (MHA) Azerbajdzhan, 1929: A. Grosshejm (LE) Hungary, 1990: 242630 (WS) Europe, 175685 (WS) Russia, Kaliningrad, 1973: A. Skvortzov (MHA) Afghanistan, 1974: I. Gubanov, V. Pavlov (MW) Armenia, 1951: E. Nazarova, 34564E (ERE) M. Ownbey, 251956 (WS) ‘‘Persia’’: K. H. Rechinger (K) Armenia, 1979: S. Kuthatheladze (LE) Spain, 1982: F. Valle, G. Blanca, field collections (Soltis lab.) Armenia, 1979: E. Nazarova, 132916 (ERE) M. Ownbey, 374431 (WS) Europe, 274106 (WS) 274198 (WS) Georgia, 1953: S. Kuthatheladze (LE) Azerbajdzhan, 1972: S. Kuthatheladze (LE) Georgia, 1945: D. Sosnovsky (LE) Europe: 1373086 (USN) ‘‘Persia,’’ Shared-Bustan: K. H. Rechinger (K) Bulgaria, 1853: (LE) Europe: M. Ownbey, 274728 (WS) Russia, Volgograd, 1989: V. Sagalaev (VOLG) Spain, 1982: G. Blanca et al., field collections (Soltis lab.) M. Ownbey, 208347 (WS) Bulgaria/Serbia, 1903: K. H. Rechinger, M. Ownbey (K) Azerbajdzhan, 1948: S. Lipshitz (LE) Armenia, 1979: E. Nazarova, 907 (ERE) Russia, Astrachan, 2002: E. Mavrodiev, field collections (Soltis lab.), det. N. N. Tzvelev Russia, Volgograd, 2002: E. Mavrodiev, field collections (Soltis lab.) M. Ownbey, 274420 (WS) Georgia, 1970: T. Popova (LE) AY645802 AY508175 AY508183 AY645803 AY645804 AY508166 AY645805 AY645806 AY508174 AY645807 AY645808 AY645809 AY645810 AY645811 AY508180 AY645812 AY645845 AY645846 AY645847 AY645848 AY645856 AY645857 AY645858 AY645813 AY645814 AY645859 AY645860 AY645815 AY645861 AY645816 AY645817 AY508182 AY645818 AY508168 AY645819 AY645820 AY508178 AY508181 AY645821 AY645823 AY645849 AY645850 AY645852 AY645851 AY645853 AY645854 AY645855 AY645862 AY645863 AY645864 AY645865 AY645866 AY645867 AY645868 AY645869 AY645871 AY645822 AY645824 AY508185 AY645825 AY645826 AY645827 AY645828 AY508184 AY508172 AY645829 AY508170 AY645831 AY645870 AY645872 AY645873 AY645874 AY645875 AY645878 AY645879 AY645880 AY508169 AY508167 AY508176 AY645881 AY645882 AY645883 AY645830 AY645832 AY645833 AY645884 AY645885 AY645886 AY645834 AY645887 AY645835 AY645836 AY645888 AY645889 AY645876 AY645877 MAVRODIEV ET AL.—PHYLOGENY OF TRAGOPOGON 121 Table 2 (Continued ) GenBank accession numbers Species T. T. T. T. serotinus sinuatus sosnowskyi stenophyllus T. tommasinii T. trachycarpus T. undulat us Podospermum jacquinianum Lactuca sp. Voucher ITS region of rDNA (ITS1, 5.8S, ITS2) ETS region of rDNA Armenia, 1967: S. Kuthatheladze (LE) Ownbey M., 274133 (WS) Caucasus, 1830: S. Kuthatheladze (LE) M. Ownbey, 274276 (WS) France: M. Ownbey, 274740 (WS) M. Ownbey, 274702 (WS) Mongolia: I. A. Gubanov (MW) Ukraina, Tauria: N. K. Schvedchikova (MW) AY645837 AY645838 AY645839 AY645840 AY645841 AY645842 AY508177 AY508171 AY645890 AY645891 AY645892 AY645893 AY645894 AY645895 AY645896 Turkey: P. H. Davis, 51673 K Gainesville: E. Mavrodiev, original collections AY508195 L13957 AY645844 AY645843 accessions of 59 species, representing 14 of the 17 sections of Borisova (1964). An important limitation in our sampling is that many species of Tragopogon are difficult to obtain. Most species occur in poorly collected areas and/or in countries experiencing political upheaval, including Iran, Iraq, Afghanistan, and Chechnya (Russian Federation). Hence, we relied largely on herbarium specimens and field collections of the first author (table 2). Plants collected in the field were either dried as pressed specimens or placed in silica gel. We sequenced ITS and ETS from one or more species from sections Angustissimi, Bessera, Brevirostres, Chromopappus, Hebecarpus, Krascheninnikovia, Majores, Nikitinia, Tuberosi, Profundisulcati, Collini (¼Rubriflori), Sosnovsky, and Tragopogon. We also sampled the single species of the monotypic section Kemularia (table 2). The only sections of Borisova (1964) not represented in this study are Macropogon (monotypic), Dasypogon (monotypic), and Dybjanskya (monotypic). We were unable to obtain plants of these three small sections. We also included 17 species of unknown systematic position in Tragopogon: T. albinerve Freyn. et Sint., T. aureus Boiss., T. balcanicum Velen. ex Nym., T. coelesyriacus Boiss., T. crocifolius L., T. fibrosum Freyn et Sint. ex Freyn, T. hayekii (Soo) I. B. K. Richardson, T. jesdianus Boiss. et Buhse, T. kindingeri Adamov, T. minor Miller, T. olympicus Boiss., T. samaritani Heldr. et Sart., T. sinuatus Ave-Lall., T. longifolius Heldr. et Sart., T. pterodes Panc, T. stenophyllus Jord., and T. tommasinii Sch. Bip. These species have never been attributed to any of the formally recognized sections of Tragopogon, and they are therefore not listed in table 1. We also attempted to include only diploid taxa in our analyses, because of the potential difficulties and uncertainties involved in using ITS and ETS sequences in polyploid taxa (Soltis and Soltis 1998; Álvarez and Wendel 2003). We therefore excluded from our analysis all known polyploids, such as the North American allotetraploids T. mirus and T. miscellus and the Eurasian polyploids T. buphtalmoides, T. castellanus, T. coloratus, T. cupani, T. gracilis, T. graminifolius, T. reticulatus, T. latifolius, T. pusillus, and T. tuberosus. In some cases (e.g., T. latifolius, T. reticulatus, T. coloratus, T. pusillus), we used only the diploid races of species for which several ploidal levels have been reported (Nazarova 1984, 1991). Both ITS and ETS sequences were obtained for 52 taxa. Because of the poor quality or small amount of some DNA extracts, we were unable to obtain ETS sequences for several species. Tragopogon krasheninnikovii, T. collinus, T. montanus, T. marginatus, T. fibrosus, T. aureus, and T. sosnowskyi are represented only by ITS sequences. We conducted two sets of analyses. In one, we analyzed the 52 taxa represented by both ITS and ETS sequences. The second analysis, of 59 taxa, also included the seven taxa for which only ITS was sequenced. As outgroups, we included the ITS (Mavrodiev et al. 2004) and ETS sequences of Podospermum jacquinianum Koch and Lactuca sp. (ITS of Lactuca sp. was obtained from GenBank; table 2). Lactuca represents Lactucinae, a subtribe that is the close relative of subtribe Scorzonerinae (Bremer 1994). DNA Amplification and Sequencing We isolated DNA following the general method of Doyle and Doyle (1987), as modified by Soltis et al. (1991). For ITS amplification and sequencing, we used forward primer N-nc18S10 (59-AGG AGA AGT CGT AAC AAG-39) (Wen and Zimmer 1996) and reverse primer ITS-4 (59-GGT AGT CCC GCC TGA CCT GG-39) (White at al. 1990) or, in some samples, reverse primer C26A (5-AGT TTC TTT TCC TCC GCT-39) (Wen and Zimmer 1996). PCR reactions (25 mL) contained 13 PCR Buffer (Promega, Madison, WI), 2.5 mM MgCl2, 200 mM each dNTP, 1 M betaine, 1 mM each of forward and reverse primers, 1.25 U Taq DNA polymerase (Promega), and ca. 10 ng of genomic DNA. The amplification profile consisted of an initial denaturation of 2 min at 95°C followed by a touch-down segment of 93°C for 1 min, then 53°C (decreasing 1°C per cycle) for 1 min and 72°C for 2 min for five cycles, followed by 35 cycles using a 48°C annealing temperature. A final extension of 5 min was added at the end of the cycle. The PCR products were purified using the QIAquick PCR purification kit (Qiagen, Valencia, CA) and the manufacturer’s protocol. We sequenced the 39 end of the ETS. For ETS amplification and sequencing, we used forward primer L-ETS (59-GGA TCG TTC GGT GCA TTC T-39) (Lee et al. 2002) and reverse primer 18S-ETS (59-AGA CAA GCA TAT GAC TAC TGG CAG G-39) (Baldwin and 122 INTERNATIONAL JOURNAL OF PLANT SCIENCES Markos 1998). The PCR conditions used for ETS were the same as those given above for ITS. We sequenced ITS and ETS PCR products in both directions with the PCR primers as sequencing primers, using BeckmanCoulter Dye Terminator Cycle Sequencing Quick-Start kits and a CEQ 8000 automated sequencer (Beckman Coulter, Fullerton, CA) following the manufacturer’s protocols, except that we conducted quarter-volume sequencing reactions. Alignment and Phylogenetic Analyses Sequences were initially aligned using the program Clustal X (Thompson et al. 1997). Following the initial alignment, the sequences were adjusted manually using the program SeAl, version 2.0 a11 (Rambaut 2003). Alignment of ITS and ETS was straightforward in these taxa. The ITS and ETS data sets for 59 and 52 taxa, respectively, were first analyzed separately and then combined. All phylogenetic analyses were conducted using PAUP*, version 4.0 b10 (Swofford 2002). Maximum parsimony (MP) analyses were conducted using heuristic searches with 1000 random addition replicates with no more than 100 trees saved per replicate and tree bisection reconnection (TBR) branch swapping, with the MulTrees option in effect; gaps were treated as ‘‘missing.’’ Internal support for clades was measured by bootstrap (BST; Felsenstein 1985) and jackknife (JK) analyses. BST estimates were obtained using 2500 replicates, each with 100 random addition replicates, saving no more than 1500 trees per BST replicate, TBR branch swapping, and the MulTrees option in effect. JK estimates were obtained using 2500 JK replicates with 36.0% of the characters deleted (Farris et al. 1996) in each replicate and 100 random addition sequences, saving no more than 1500 trees per replicate, TBR branch swapping, and the MulTrees option in effect. Character-State Reconstruction We reconstructed the evolution of two characters (table 3) that have been important in sectional diagnoses in Tragopogon: life history and the length of the beak of the achene. Morphological data were adopted from Kuthatheladze (1957), Borisova (1964), Matthews (1975), Richardson (1976), Rechinger (1977), Tzvelev (1985), and Nazarova (1991, 1995); in some cases, data were obtained from herbarium (LE, MW, MHA, ERE) and field collections (southeastern Russia, Armenia). We used MacClade, version 4.05 (Maddison and Maddison 2002), using one of the most parsimonious trees chosen at random and ‘‘all most parsimonious states’’ as the optimization method. To assess the impact of the topology on character-state reconstruction, we randomly selected 1000 of the 10,000 shortest trees and reconstructed character-state changes as described. In our reconstructions we also used Geropogon hybridus (L.) Sch. Bip., Scorzonera austriaca Willd., Scorzonera aristata Ram. ex DC., and Takhtajaniantha pusilla (Pall.) Nazarova as outgroups. These taxa were found to be close relatives of Tragopogon in our phylogenetic analysis of subtribe Scorzonerinae (Mavrodiev et al. 2004). The use of these alternative outgroups had no impact, however, on our characterstate reconstructions. Results The aligned data matrix of the ITS region of rDNA (ITS1, 5.8S, ITS2) consists of 701 characters (288 in ITS1, 168 in 5.8S, and 245 in ITS2), of which 459 are constant and 80 are parsimony-informative sites. We saved 10,000 most parsimonious trees of length ¼ 343, Consistency Index ðCIÞ ¼ 0:8426, Retention Index ðRIÞ ¼ 0:8402, and Rescaled Consistency Index ðRCÞ ¼ 0:7880 (fig. 1). The aligned data matrix of the ETS region of rDNA is 574 characters, of which 336 are constant and 82 are parsimony-informative. We saved 10,000 most parsimonious trees of length ¼ 344, CI ¼ 0:8256, RI ¼ 0:8729, and RC ¼ 0:7206 (fig. 2). The aligned data matrix of the ITS þ ETS regions for 52 taxa is 1274 characters, of which 882 are constant and 162 are parsimony-informative. In parsimony analyses we saved 10,000 most parsimonious trees of length ¼ 684, CI ¼ 0:8173, RI ¼ 0:8381, and RC ¼ 0:6849 (fig. 3). The aligned ITS þ ETS data set for 59 species (seven represented only by ITS sequences) is also 1274 characters in length, of which 896 are constant and 141 are parsimony informative. We saved 10,000 most parsimonious trees of length ¼ 616, CI ¼ 0:8101, RI ¼ 0:8484, and RC ¼ 0:6873 (fig. 4). Parsimony analysis of the ITS data set provided strong support (JK ¼ 100%, BST ¼ 100%) for the monophyly of the genus Tragopogon relative to the outgroups (Lactuca sp. and Podospermum jacquinianum), but phylogenetic relationships within Tragopogon were only partially resolved. ITS sequence data recovered a clade (JK ¼ 86%, BST ¼ 65%) that encompasses taxa from sections Majores, Chromopappus, and Hebecarpus and roughly corresponds to the Majores, Chromopappus, and Hebecarpus clades detected with ETS and ETS þ ITS sequences (see below). However, the Chromopappus and Hebecarpus clades were not obtained with ITS alone. With the Majores s. l. clade in the ITS tree is a clade of T. pterpodes, T. charadzae, T. dubius, T. capitatus, and T. major, but with weak support (JK ¼ 68%, BST ¼ 52%). However, the position of T. charadzae in this clade in the ITS tree is problematic. This species of section Sosnovskya or Brevirostres is morphologically very distinct from T. dubius and the other members of this small clade. In the ETS and ITS þ ETS trees, T. charadzae belongs to the Angustissimi clade. Similarly, T. floccosus, placed in section Brevirostres, is morphologically very distinct from all members of the Majores s. l. clade, where it appears on the basis of ITS sequence data. In contrast, in the ETS and ITS þ ETS trees, T. floccosus is a member of the Brevirostres clade. With ITS alone, a Collini clade with T. ruber, T. bronmuelleri, T. collinus, T. elongatus, T. jezdianus, T. montanus, and T. marginatus received moderate (JK ¼ 83%, BST ¼ 71%) support. The Profundisulcati clade of T. albinerve, T. meskheticus, T. makaschwilii, T. armeniacus, T. fibrosus, T. aureus, and T. kotschyi received weak support (JK ¼ 68%, BST ¼ 57%). In addition, with ITS alone there is strong support (JK ¼ 91%, BST ¼ 87%) for a sister-group relationship between clades Collini and Profundisulcati (fig. 1). ITS sequence data also reveal a weakly supported (JK ¼ 66%, BST ¼ 52%) Tragopogon clade of T. hayeki, T. tommasinii, T. longifolius, T. pratensis, T. orientalis, T. minor, and T. trachycarpus (fig. 1). However, the Brevirostres and MAVRODIEV ET AL.—PHYLOGENY OF TRAGOPOGON Table 3 Morphological Characteristics of Tragopogon Data Matrix for Character-State Reconstructions Using MacClade Species Geropogon hybridus Podospermum jacquinianum Scorzonera aristata Scorzonera austriaca Takhtajaniantha pusilla Tragopogon acanthocarpus T. afghanicus T. albinerve T. armeniacus T. australis T. balcanicum T. bornmuelleri T. brevirostris T. capitatus T. charadzae T. coelesyriacus T. coloratus T. crocifolius T. dasyrhynchus T. dubius T. dubjanskyi T. elongatus T. filifolius T. floccosus T. hayekii T. heterospermus T. jesdianus T. kemulariae T. kindingeri T. kotschyi T. lamottei T. latifolius T. longifolius T. longirostris T. major T. makaschwilii T. meskheticus T. minor T. olympicus T. orientalis T. podolicus T. porrifolius T. pratensis T. pterodes T. pusillus T. reticulatus T. ruber T. ruthenicus T. samaritani T. segetum T. serotinus T. sinuatus T. stenphyllus 1 T. stenphyllus 2 T. tommasinii T. trachycarpus T. undulatus Achene morphology Life history 1 2 0 0 0 0 1 1 1 1 1 1 1 0 1 1 1 1 1 0 1 0 1 1 0 1 0 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 1 1 1 1 1 1 0 1 1 1 1 1 1 1 0 0 1 1 1 1 1 1 0 1 1 1 0 0 1 0 1 1 1 1 1 1 0 0 0, 1 0 0, 1 3 1 0 0 1 1 1 1 1 1 0 1 0 1 1 0, 1, 2 1 1 0 0 0 1 1 1 0 ? 1 1 1 1 1 Note. Achene morphology: 0 ¼ beak of achene is very short, practically absent (usually #0.5 mm); 1 ¼ beak is clearly present (usually >2 mm). Life history: 0 ¼ perennial; 1 ¼ biennial; 2 ¼ annual; 3 ¼ other. 123 Angustissimi clades, which are both present in the ETS and ETS þ ITS trees, are not recovered in the ITS trees. With ETS sequence data alone a Majores clade of T. stenophyllus, T. coelesyriacus, T. capitatus, T. dubius, T. major, T. pterodes, T. afgh anicus, and T. kemulariae received moderate support (JK ¼ 83%, BST ¼ 2%) (fig. 2). Sister to the Majores clade is the Chromopappus clade, consisting of T. longirostris and T. coloratus (JK ¼ 71%, BST ¼ 66%), although support for the Majores þ Chromopappus sister group is less than 50%. A Hebecarpus clade (JK ¼ 89%, BST ¼ 82%) of T. crocifolius, T. australis, T. balcanicum, T. samaritani, T. sinuatus, and T. porrifolius is sister to Majores þ Chromopappus (JK ¼ 63%, BST ¼ 60%) (fig. 2). Analysis of ETS data also recovered a Profundisulcati clade (JK ¼ 73%, BST ¼ 61%) of T. albinerve, T. meskheticus, T. makaschwilii, T. armeniacus, and T. kotschyi. The Profundisulcati clade is sister (JK ¼ 65%, BST ¼ 50%) to the Collini clade (JK ¼ 75%, BST ¼ 68%) of T. ruber, T. bronmuelleri, T. elongatus, and T. jezdianus. Also receiving support with ETS sequences alone is an Angustissimi clade (JK ¼ 93%, BST ¼ 79%) of T. pusillus, T. filifolius, T. latifolius (2n ¼ 12), T. segetum, T. acanthocarpus, T. serotinus, and T. charadzae. This clade does not correspond to any formally named section but contains both of the species of section Angustissimi (T. segetum and T. sosnowskyi) (fig. 4; see below). A Brevirostres clade of T. kindingeri, T. brevirostris, T. dasyrhynchus, T. reticulatus, T. dubjanskyi, T. ruthenicus, T. heterospermus, T. undulatus, T. podolicus, and T. floccosus also received support (JK ¼ 90%, BST ¼ 76%) in analyses based only on ETS. The Brevirostres clade is part of a polytomy with T. lamottei, T. hayeki, T. tommasinii, T. longifolius, and species from section Tragopogon: T. pratensis, T. orientalis, T. minor, and T. trachycarpus. Tragopogon hayeki, T. tommasinii, T. longifolius, T. orientalis, T. pratensis, and T. minor form a moderately supported Tragopogon clade (JK ¼ 88%, BST ¼ 74%) in the ITS þ ETS tree, and T. lamottei is sister to this clade, but with weak support (JK ¼ 65%) (fig. 3). The ITS and ETS trees do not exhibit any major conflicts. In fact, there is close agreement between the ITS and ETS trees, although the ETS tree is better resolved and supported than the ITS tree (cf. figs. 1, 2). All major differences involve clades that are resolved and supported with ETS but unresolved with ITS. The primary differences are (1) the ETS tree detected the Chromopappus and Hebecarpus clades, whereas ITS did not; (2) species of section Brevirostres formed a moderately supported Brevirostres clade with ETS (JK ¼ 90%, BST ¼ 76%), whereas only a portion of this clade is recovered with ITS (and without support >50%); and (3) a moderately supported Angustissimi clade (JK ¼ 93%, BST ¼ 79%) with several species from the Caucasus is present in the ETS tree but not in the ITS tree. Given the lack of major conflict, we therefore combined the ITS and ETS data sets. Our analyses of the combined ITS þ ETS data set for 52 taxa produced a generally well-resolved tree for Tragopogon (fig. 3) that is similar to that produced when the seven species represented only by ITS sequences were added to the analysis (fig. 4). However, support for several clades is lower in the latter trees than in the 52-taxon ITS þ ETS tree, most likely as Fig. 1 One of 10,000 shortest trees saved of length 339 resulting from maximum parsimony analyses of the Tragopogon internal transcribed spacer data matrix. Arrows indicate branches that collapse in the strict consensus. Numbers above branches indicate branch lengths. Jackknife/ bootstrap values for nodes receiving >50% support are indicated in italics below the branches. Clade names roughly correspond to sections of Borisova (1964). 124 Fig. 2 One of 10,000 shortest trees saved of length 243 resulting from maximum parsimony analyses of the Tragopogon external transcribed spacer data matrix. Arrows indicate branches that collapse in the strict consensus. Numbers above branches indicate branch lengths. Jackknife/ bootstrap values for nodes receiving >50% support are indicated in italics below the branches. Clade names roughly correspond to sections of Borisova (1964). 125 126 INTERNATIONAL JOURNAL OF PLANT SCIENCES the result of more missing ETS data. We will first discuss the 52-taxon ITS þ ETS topology (fig. 3) and then return to the placement of the seven taxa for which only ITS data are present (fig. 4). Discussion Phylogeny of Tragopogon There are no well-supported differences between the ITS and ETS trees. Thus, there are no instances of hard incongruence (see Seelanan et al. 1997). Because of the lack of major conflict, the ITS and ETS data sets were combined. The ITS þ ETS tree provided strong support (JK ¼ 100%, BST ¼ 100%) for the monophyly of the genus Tragopogon relative to the outgroups (fig. 3). This result agrees with the findings of a recent molecular phylogenetic investigation of subtribe Scorzonerinae using ITS sequences alone (Mavrodiev et al. 2004). Sequence data also provide support for the monophyly of several sections of Tragopogon, although relationships among sections are not always clear or well supported. We provide our best inference of phylogenetic relationships within Tragopogon below, based on the combined ITS þ ETS tree. The Majores clade (JK ¼ 77%, BST ¼ 72%) contains T. stenophyllus, T. coelesyriacus, T. capitatus, T. dubius, T. pterodes, and T. afghanicus. Several of these species (i.e., T. dubius, T. capitatus, and T. afg hanicus) have been placed in section Majores by some authors (Borisova 1964; Rechinger 1977). Tragopogon stenophyllus was included in T. crocifolius by Richardson (1976), but our trees place T. crocifolius in the Hebecarpus clade, separate from the Majores clade (see below). Thus, molecular phylogenetic data indicate that T. stenophyllus is a species separate from T. crocifolius. This result is based on two samples of T. stenophyllus from different collections (fig. 3). The other species in the Majores clade, T. pterodes and T. coelesyriacus, have not been previously placed in any section of Tragopogon (fig. 3). All species of the Majores clade have achenes with a long beak and long peduncles, which are more or less plump below the flowering capitula. However, the color of the ligules varies within the clade. For example, T. stenophyllus, T. pterodes, T. coelesyriacus, and T. afghanicus have purple or violet ligules; in contrast, T. dubius and T. capitatus have yellow ligules. We agree with Tzvelev (1985) that the color of ligules is not an important character in determining sections within Tragopogon. Based on sequence data and morphology, section Majores could be expanded to include T. pterodes, T. stenophyllus, and T. coelesyriacus. However, Tragopogon remains so poorly understood that any formal changes should await a thorough molecular and morphological analysis of the entire genus. The Chromopappus clade is moderately supported (JK ¼ 80%, BST ¼ 70%) and contains T. coloratus and T. longirostris. Tragopogon coloratus is the type of section Chromopappus (Borisova 1964). However, T. longirostris is usually included in section Krascheninnikovia (Borisova 1964) and is morphologically similar to the type species of that section, T. krascheninnikovii (Borisova 1964). Also, T. krascheninnikovii is similar in appearance to T. coloratus (Nazarova 1995). However, T. krascheninnikovii (which is represented only by an ITS sequence) is placed in the Majores clade in the ITS þ ETS tree (fig. 4), as a part of a wellsupported (JK ¼ 90%, BST ¼ 87%) clade with T. coelesyriacus and T. stenophyllus. Our data indicate that the small section Krascheninnikovia (two species) is not monophyletic. Both species of the Chromopappus clade are morphologically very similar to T. dubius and other members of the Majores clade and in some treatments are actually included in section Majores (Kuthatheladze 1957). Hence, morphology and DNA data agree in suggesting a close relationship between Majores and Chromopappus. In our analyses, the Majores and Chromopappus clades are sisters, but with weak support (JK ¼ 68%, BST ¼ 51%; fig. 3). The Hebecarpus clade is strongly supported (JK ¼ 91%, BST ¼ 86%) and is sister to the Majores þ Chromopappus clade (JK ¼ 80%, BST ¼ 69%). This clade contains T. crocifolius, T. australis, T. balcanicum, T. samaritani, T. sinuatus, and T. porrifolius. In Borisova’s (1964) system, section Hebecarpus includes only T. porrifolius. The other species of our Hebecarpus clade have not been previously placed in any section of Tragopogon. However, T. australis is often included in T. porrifolius at the rank of subspecies (Richardson 1976). Similarly, Boissier (1875) indicated a very close relationship between T. samaritani and T. crocifolius, and Richardson (1976) considered T. samaritani to be a subspecies of T. crocifolius. However, our data indicate that both T. australis and T. samaritani are distinct species: T. australis is well supported, as distinct from T. porrifolius, and T. samaritani is distinct from T. crocifolius (fig. 3). The morphological description of section Hebecarpus (Borisova 1964) is very brief, and it is difficult to understand the differences between this section and sections Chromopappus or Krascheninnikovia. This may be the reason why Tzvelev (1985) included T. porrifolius (sect. Hebecaprus) in section Majores. The clade ðMajores þ ChromopappusÞ þ Hebecarpus received moderate (or low) support (JK ¼ 80%, BST ¼ 69%) and is consistent with a broad treatment of section Majores (i.e., including sections Majores, Chromopappus, and Hebecarpus), as proposed by Kuthatheladze (1957) and Tzvelev (1985). The morphological characters differentiating the Majores, Chromopappus, and Hebecarpus clades require additional study. The placement of T. kemulariae is problematic. In the ETS tree, T. kemulariae is a member of the Majores clade (JK ¼ 83%, BST ¼ 79%) (fig. 2), whereas in the ITS tree it occurs in the clade corresponding largely to section Angustissimi and Brevirostres. In the ITS þ ETS tree, the relationship of T. kemulariae is unclear (fig. 3); morphologically, T. kemulariae is very similar to T. dubius (Nazarova 1995) of the Majores clade. However, Kuthatheladze (1957) placed T. kemulariae in section Profundisulcati, and Borisova (1964) placed this taxon in a monotypic section Kemularia. The Profundisulcati clade is well supported (JK ¼ 99%, BST ¼ 98%) (fig. 3) and contains T. albinerve, T. meskheticus, T. makaschwilii, T. armeniacus, and T. kotschyi. Three of the taxa in the Profundisulcati clade (T. meskheticus, T. makaschwilii, T. armeniacus) were placed in section Profundisulcati (Borisova 1964). Kuthatheladze (1957) placed section Profundisulcati as a series within section Majores, Fig. 3 One of 10,000 shortest trees saved of length 591 resulting from maximum parsimony analyses of the Tragopogon internal þ external transcribed spacer data matrix for 52 taxa. Arrows indicate branches that collapse in the strict consensus. Numbers above branches indicate branch lengths. Jackknife/bootstrap values for nodes receiving >50% support are indicated in italics below the branches. Clade names roughly correspond to sections of Borisova (1964). Fig. 4 One of 10,000 shortest trees saved of length 616 resulting from maximum parsimony analyses of the Tragopogon internal þ external transcribed spacer data matrix with seven additional species represented only by ITS sequences (boldface). Arrows indicate branches that collapse in the strict consensus. Numbers above branches indicate branch lengths. Jackknife/bootstrap values for nodes receiving >50% support are indicated in italics below the branches. MAVRODIEV ET AL.—PHYLOGENY OF TRAGOPOGON 129 Fig. 5 Reconstruction of the diversification of achene morphology in Tragopogon. Character states were mapped onto one randomly selected maximum parsimony tree using MacClade and the all most parsimonious state optimization method. but based on our phylogenetic analysis the Profundisulcati clade is very distinct from the Majores clade. Tragopogon albinerve has not been previously placed in any section of Tragopogon, but morphologically it is very similar to another member of section Profundisulcati, the polyploid species T. buphtalmoides (Matthews 1975). Tragopogon kotschyi of the Profundisulcati clade was placed in section Sosnovskya (Rechinger 1977), a small section whose species are morphologically intermediate between species of sections Brevirostres and Profundisulcati (Borisova 1964). In our phylogenetic analyses, section Sosnovskya is not monophyletic. The four species of section Sosnovskya used in our analysis, T. reticulatus, T. charadzae, T. kotschyi, and T. marginatus, all belong to distinct clades (Brevirostres, Angustissimi, Profundisulcati, and Collini, respectively), each of which receives support >50% (figs. 1–4). Nazarova (1991) disagreed with the recognition of section Sosnovskya in her treatment of Tragopogon. Morphological investigations of section Profundisulcati are needed. Our data also indicate that the Profundisulcati clade includes T. aureus and T. fibrosus. Both of these taxa are placed in the Profundisulcati clade based only on their ITS sequences (fig. 4). These two species have not been placed in any section and are poorly understood morphologically. For example, the achenes of T. aureus are unknown (Matthews 1975). Tragopogon fibrosus is a rare, narrow endemic of North Anatolia known only from three specimens, two of which are very old (1895) and inaccessible (LD). Matthews (1975) remarked that this species is known only from the type collections, but in our treatment we used the unpublished collections of T. fibrosus obtained by A. Khokhriakov (MHA) from Turkey (table 1). The Collini clade is well supported (JK ¼ 98%, BST ¼ 97%) and contains T. ruber, T. bronmuelleri, T. elongatus, and T. jezdianus. Three additional species, T. montanus, T. collinus, and T. marginatus, for which we have only ITS sequences, are also placed in the Collini clade (fig. 4). Most of these taxa (T. ruber, T. montanus, T. elongatus, and T. collinus) are members of section Collini (¼Rubriflori) (Borisova 1964). Members of the Collini clade are perennial, with purplish ligules and with achene beaks that are equal to or shorter than the body of the achene. Tragopogon marginatus, with yellow ligules, has been placed in either section Sosnovskya (Borisova 1964) or section Collini (Kuthatheladze 1957). The rare endemic species T. bronmuelleri was placed in section Profundisulcati (Rechinger 1977). In our analysis, however, this species is sister to T. ruber, a species of section Collini (figs. 3, 4). The diagnostic attributes of the Collini clade require investigation. The sister-group relationship of the Profundisulcati and Collini clades is well supported (JK ¼ 99%, BST ¼ 96%). This result provides some support for a broader treatment of section Collini that combines species of sections Collini and Profundisulcati. However, our data also support the reciprocal monophyly of the Collini and Profundisulcati clades, so either a broad section Collini or smaller sections Collini and 130 INTERNATIONAL JOURNAL OF PLANT SCIENCES Fig. 6 Reconstruction of the diversification of life history in Tragopogon. Character states were mapped onto one randomly selected maximum parsimony tree using MacClade and the all most parsimonious state optimization method. Profundisulcati would be consistent with the phylogenetic tree. The Brevirostres clade is well supported (JK ¼ 95%, BST ¼ 86%) and consists of T. kindingeri, T. brevirostris, T. dasyrhynchus, T. reticulatus, T. dubjanskyi, T. ruthenicus, T. heterospermus, T. undulatus, T. podolicus, and T. floccosus (fig. 3). Many of the species of the Brevirostres clade (T. reticulatus, T. brevirostris, T. dasyrhynchus, T. undulatus, and T. podolicus) are from section Brevirostres (Kuthatheladze 1957; Tzvelev 1985; Nazarova 1991). The clade also contains two species (T. ruthenicus and T. dubjanskyi) placed in section Nikitinia by Borisova (1964). Tzvelev (1985) also included T. ruthenicus and T. dubjanskyi in section Brevirostres (see also Klokov 1981). The Brevirostres clade also contains T. heterospermus, a species of section Bessera (Borisova 1964). However, Tzvelev (1985) included all Bessera species in section Brevirostres, and the diagnostic morphological attributes of section Bessera are unclear (Tzvelev 1985). However, not all species of section Brevirostres appear in the Brevirostres clade. Tragopogon serotinus and T. filifolius, species that are usually included in section Brevirostres (Kuthatheladze 1957; Borisova 1964), both belong to a distinct, well-supported clade that does not correspond to a section, which we are calling the Angustissimi clade (JK ¼ 91%, BST ¼ 72%). The Tragopogon and Brevirostres clades are sisters in the ITS þ ETS tree, a relationship that receives conflicting levels of support (JK ¼ 90%, BST ¼ 67%) (fig. 3). The Tragopogon clade is fairly well supported (JK ¼ 88%, BST ¼ 74%) and is composed of T. hayeki, T. tommasinii, T. longifolius, T. pratensis, T. orientalis, T. minor, and T. trachycarpus (fig. 3). This clade contains the type species of the genus Tragopogon (T. pratensis). This species and three others (T. orientalis, T. trachycarpus, and T. minor) have been placed in section Tragopogon. The other taxa of the Tragopogon clade (T. hayeki, T. tommasinii, and T. longifolius) have not been placed in any section of Tragopogon. However, two of these taxa (T. hayeki and T. tommasinii) are morphologically very similar to T. pratensis s. l. (Richardson 1976; see also Boissier 1875). Some workers have included T. orientalis within T. pratensis (Richardson 1976). However, our results clearly indicate that T. pratensis is phylogenetically distinct from T. orientalis. A well-supported (JK ¼ 92%, BST ¼ 84%) subclade of T: hayeki þ T: tommasinii is sister to T: longifolius þ T: orientalis (JK ¼ 91%, BST ¼ 88%). These four species form a well-supported clade (JK ¼ 98%, BST ¼ 95%) that is in turn sister (JK ¼ 74%, BST ¼ 75%) to a subclade (JK ¼ 73%, BST ¼ 84%) of T: pratensis þ T: minor. Molecular data indicate that T. lamottei is a distinct species. This species appears as sister to all other species of the MAVRODIEV ET AL.—PHYLOGENY OF TRAGOPOGON Tragopogon clade (JK ¼ 65%; fig. 3). However, the phylogenetic position of T. lamottei is not well supported. Tragopogon lamottei has been considered closely related to T. pratensis (T. pratensis ssp. lamottei [Rouy] O. de Bolós & J. Vigo), but our results do not support this. Some species of section Tragopogon not sampled here (T. karelinii S. Nikit., T. tomentosulus Boriss., T. gracilis D. Don., T. subalpinus S. Nikit., and T. sibiricus Ganesh.) have purplish ligules, and the systematic placement of all of these Asian species is uncertain. For example, Nikitin (1937) placed all of these taxa (except T. sibiricus) with species from section Collini (¼Rubriflori). Species from section Tragopogon are mostly biennial plants, so the only clear morphological difference between sections Tragopogon and Brevirostres is the length of the achenes and the achene beak (Tzvelev 1985). All members of section Brevirostres and the Brevirostres clade have achenes with very short beaks (excluding T. reticulatus) and yellow ligulate flowers. Members of section Tragopogon and the Tragopogon clade have achenes with obvious beaks (excluding T. trachycarpus, with a very short beak) that usually are shorter than the body of the achene. Klokov (1981) suggested that species of section Brevirostres and some species of section Tragopogon be combined to form one broad section. Artemczhyk (1948) suggested that T . heterospermus and T. podolicus (sect. Brevirostres) are closely related to T. orientalis and T. pratensis from section Tragopogon. According to our topology, however, T. heterospermus and T. podolicus are in the Brevirostres clade, while T. orientalis and T. pratensis are in the Tragopogon clade. Tragopogon kindingeri is also morphologically similar to T. pratensis (Richardson 1976) but is sister to the rest of the Brevirostres clade and has not been previously placed in any section of Tragopogon. Although some have suggested that T. longifolius is a subspecies of T. brevirostris (Richardson 1976; also Boissier 1875), our results place T. longifolius as a part of a well-supported subclade (JK ¼ 98%, BST ¼ 95%), with T. orientalis, T. tommasinii, and T. hayeki, in the Tragopogon clade. We agree with the conclusions of Artemczhyk (1948) and Borisova (1964), who suggested that T. brevirostris is a narrow endemic of the Caucasus and that T. longifolius should not be included in T. brevirostris. We identified a previously unrecognized clade (JK ¼ 91%, BST ¼ 72%) that contains T. pusillus (sect. Tuberosi or Collini), T. filifolius (sect. Brevirostres), T. latifolius (2n ¼ 12) (sect. Profundisulcati or Majores), T. segetum (sect. Angustissimi), T. acanthocarpus (sect. Profundisulcati or Majores), T. serotinus (sect. Brevirostres), and T. charadzae (sect. Sosnovskya or Brevirostres) (fig. 3). We are calling this clade Angustissimi because it contains both species of section Angustissimi (T. segetum and T. sosnowskyi). Tragopogon sosnowskyi is included in the subclade based only on its ITS sequence (fig. 4). Within this clade, T. serotinus and T. charadzae form a clade (JK ¼ 89%, BST ¼ 71%). These two species are morphologically similar and were included in section Brevirostres by Kuthatheladze (1957). Diagnostic attributes of the Angustissimi clade are problematic. Most species of the Angustissimi clade have similar geographic distributions that center on the Caucasus. Tragopogon segetum, T. filifolius, T. serotinus, T. sosnowsky, and 131 T. charadzae are all restricted to the Caucasus. Tragopogon acanthocarpus occurs in a local area in the south Caucasus and in several provinces of northern and central Iran. Tragopogon latifolius is mostly a polyploid species, distributed in the south Caucasus (Nazarova 1995) and Turkey (Matthews 1975), but we used only the diploid race of T. latifolius. This rare race of T. latifolius is known only from a few places in Armenia (Nazarova 1984). Tragopogon pusillus is reported from Turkey (Matthews 1975), Iran (Borisova 1964), Crimea (Tzvelev 1985), and Turkmenistan (‘‘Turkomania’’) (Borisova 1964). However, Matthews (1975, p. 667) indicated that the Turkish record of T. pusillus ‘‘needs confirmation.’’ Rechinger (1977) described an Iranian population of T. pusillus as a new species (T. gongylorrhizus Rech.). The distribution of Fig. 7 Types of achenes in Tragopogon. A, T. ruber: achene with short beak (b). B, T. ruthenicus: achene without beak. C, T. capitatus: achene with a long beak (b). Scale bar ¼ 10 mm. Original photographs. 132 INTERNATIONAL JOURNAL OF PLANT SCIENCES T. pusillus in Crimea and Turkmenistan is very local. For example, in Turkmenistan, T. pusillus is known only from one location (Borisova 1964). The close relationship of the flora of the Crimea/Caucasus region to the Turkmenistan flora is also seen in the moss flora (Abramov et al. 1987). In Crimea, T. pusillus is known from just a narrow area on the southern coast (Tzvelev 1985), and the nature of this disjunct area of T. pusillus is the subject of discussion (Tzvelev 1985). Tzvelev (1985) suggested that many species from Crimea, especially T. pusillus, with clear ‘‘roots’’ in the Caucasus, are the result of the previous existence of a mountain region between Crimea and the modern Caucasus before the origin of the modern Black Sea. Additional morphological investigations are needed to clarify the diagnostic attributes of this geographically well-demarcated clade of Tragopogon. Character Evolution The length of the beak of the achene has been an important taxonomic character in Tragopogon (Borisova 1964; Richardson 1976; Tzvelev 1985), differing among species and sections. It is difficult to describe all variants of the ratio between the length of the beak and the length of the body of the achene. Following Klokov (1981) and Tzvelev (1985), we split all Tragopogon species into two large groups based on achene morphology: the first group contains species without a beak, and the second group contains species with a beak, either long or short (fig. 7). The outgroups have achenes without beaks. Using the 1000 randomly selected shortest trees with the all most parsimonious states option, the ancestral state of Tragopogon is an achene with a beak (fig. 5; fig. 7A, 7C). The ancestral state of each clade (Majores, Chromopappus, Hebecarpus, Profundisulcati, Collini, Brevirostres, Tragopogon, and Angustissimi) is also an achene with a beak. Achenes without beaks appear to have evolved two times, once in the ancestor of the Brevirostres clade and again in T. trachycarpus in the Tragopogon clade. Within the Brevirostres clade there was a subsequent reversal to a beaked achene in T. reticulatus (fig. 5). Mavrodiev et al. (2004) discussed the evolution of habit (annual, biennial, or perennial) across the entire subtribe Scorzonerinae, using 20 representatives of Tragopogon (biennial or perennial). That broad analysis revealed that the an- cestral state of Tragopogon is the perennial habit, with multiple changes to biennial. In this article, we have repeated the reconstruction of habit, using many more species (59) of Tragopogon. Our results confirm our earlier reconstruction while providing new insights within Tragopogon and with close relatives of Tragopogon as outgroups. The ancestral habit of the genus Tragopogon is again reconstructed as perennial (fig. 6). However, the ancestral state of some clades, such as Majores, Chromopappus, Hebecarpus, and Tragopogon, is biennial. The ancestral state of the Brevirostres and Angustissimi clades is equivocal. Our phylogenetic analysis has provided the first assessment of phylogenetic relationships within the large, taxonomically complex genus Tragopogon. Many of the sections recognized by morphology are supported as monophyletic by analyses of gene sequence data. In addition, possible relationships are indicated for numerous species of Tragopogon that have never been assigned to any section within the genus. We observed a general correlation between phylogeny and geography, with closely related taxa occurring in the same geographic region. For example, most species of the Angustissimi clade have identical geographic distributions that center on the Caucasus. This phylogenetic analysis should be considered only a first step toward resolving relationships within this poorly understood genus. Not only are additional molecular phylogenetic analyses required that employ additional genes and more taxa, but in addition, the genus requires a thorough morphological analysis. Acknowledgments This research was supported by National Science Foundation (NSF)-North Atlantic Treaty Organization grant DGE0209500, an REU supplement to NSF grant DEB-9707868, the University of Florida Research Foundation, and the School of Biological Sciences, Washington State University, for support of undergraduate research. We thank the Royal Botanic Gardens, Kew, for access to specimens of Tragopogon and for permission to remove samples for DNA analysis. We also thank Professor Estella A. Nazarova (Botanical Institute of Armenia [Erevan]) for important consultations and for the collecting of material. Literature Cited Abramov II, AL Abramova, IV Sirotina 1987 The Crimea-Caucasian relationships of the moss flora of the Turkomania. Nov Sist Nizshikh Rast 24:179–184. (In Russian.) Álvarez I, JF Wendel 2003 Ribosomal ITS sequences and plant phylogenetic inference. Mol Phylogenet Evol 29:417–434. Artemczhyk IV 1937 Tragopogon in Ukrainian flora. Trans Bot Inst Charkov State Univ 2:23–65. (In Ukrainian.) ——— 1948 On the origin of sand-land Tragopogon species of the European part of the USSR. Sci Trans Biol Fac Chernigov State Univ 1:85–91. (In Ukrainian with Russian abstract.) Baldwin BG, S Markos 1998 Phylogenetic utility of the external transcribed spacers (ETS) of 18S–26S rDNA: congruence of ETS and ITS trees of Calycadenia (Compositae). Mol Phylogenet Evol 10:449–463. Blackmore S 1982 Palynology of subtribe Scorzonerinae (Compositae: Lactuceae) and its taxonomic significance. Grana 21: 149–160. Boissier EP 1875 Tragopogon. Pages 744–755 in EP Boissier, ed. Flora orientalis. Vol 3. H. Georg, Geneva. Borisova AG 1964 Tragopogon. Pages 115–196 in BK Shishkin, ed. Flora of the USSR. Vol 29. Nauka, Moscow. (In Russian.) Bremer K 1994 Asteraceae, cladistics and classification. Timber, Portland, OR. Clevinger JA, JL Panero 2000 Phylogenetic analysis of Silphium and subtribe Engelmanninae (Asteraceae: Heliantheae) based on ITS and ETS sequence data. Am J Bot 37:565–572. Darlington CD 1956 Chromosome botany and the origins of cultivated plants. 3rd ed. Hafner, New York. MAVRODIEV ET AL.—PHYLOGENY OF TRAGOPOGON De Candolle AP 1838 Prodromus systematis naturalis regni vegetabilis. Vol 7. Treuttel & Wurtz, Paris. Doyle JJ, JL Doyle 1987 A rapid DNA isolation procedure from small quantities of fresh leaf tissue. Phytochem Bull 19:11–15. Farris JS, VA Albert, M Kallersjo, D Lipscomb, AG Kluge 1996 Parsimony jackknifing outperforms neighbor-joining. Cladistics 12: 99–124. Felsenstein J 1985 Confidence limits on phylogenetics: an approach using the bootstrap. Evolution 39:783–791. Klokov MV 1981 Ukrainian sand’s floristic complex. Pages 90–133 in MV Klokov, ed. News of higher and lower plant systematics. Naukova Dumka, Kiev. (In Russian.) Kuthatheladze SI 1957 Caucasian representatives of Tragopogon genus. Monogr Bot Inst Acad Sci Georgia, ser. A, 2:3–79. (In Russian with Georgian abstract.) Lee J, BG Baldwin, LD Gottlieb 2002 Phylogeny of Stephanomeria and related genera (Compositae–Lactuceae) based on analysis of 18S–26S nuclear rDNA ITS and ETS sequences. Am J Bot 89: 160–168. ——— 2003 Phylogenetic relationship among the primary North American genera of Cichoriae (Compositae) based on analysis of 18S–26S nuclear rDNA ITS and ETS sequences. Syst Bot 28: 616–626. Maddison WP, DR Maddison 2002 MacClade, version 4.05: analysis of phylogeny and character evolution. Sinauer, Sunderland, MA. Markos S, BG Baldwin 2001 Higher-level relationship and major lineages of Lessingia (Compositae, Astereae) based on nuclear rDNA internal and external transcribed spacer (ITS and ETS) sequences. Syst Bot 26:168–183. Matthews VA 1975 Tragopogon. Pages 657–668 in PH Davis, ed. Flora of Turkey and East Aegean Islands. Vol 5. Edinburgh University Press, Edinburgh. Mavrodiev EV, CE Edwards, DC Albach, MA Gitzendanner, PS Soltis, DE Soltis 2004 Phylogenetic relationships in subtribe Scorzonerinae (Asteraceae: Cichorioideae: Cichorieae) based on ITS sequence data. Taxon 53:699–712. Nazarova EA 1984 Chromosome numbers in the Caucasian representatives of the families Asteraceae, Brassicaceae, Fabaceae, Linoniaceae. Bot Zh 63:972–975. (In Russian.) ——— 1991 Karyotypical evolution in genus Tragopogon L. (Lactuceae, Asteraceae). Pages 116–134 in AL Takhtajan, ed. Flora, vegetation, and vegetable resources of Armenia. Academy of Science of Armenia, Erevan. (In Russian with Armenian abstract.) ——— 1995 Tragopogon. Pages 111–131 in AL Takhtajan, ed. Flora of Armenia. Vol. 9. Koeltz, Koeningstein. (In Russian.) ——— 1997 Karyosystematic investigation of the genus Scorzonera L. s. l. (Lactuceae, Asteraceae). Caryologia 50:250–261. Nikitin SA 1930 On systematic of sand races of Tragopogon. Proc Main Bot Gard USSR 29:484–491. (In Russian.) ——— 1937 New species of the genus Tragopogon. Bot Mater Herb Bot Inst USSR Acad Sci 7:257–274. (In Russian.) Ownbey M 1950 Natural hybridization and amphiploidy in the genus Tragopogon Am J Bot 37:487–499. 133 Ownbey M, GB McCollum 1953 Cytoplasmic inheritance and reciprocal amphiploidy in Tragopogon. Am J Bot 40:788–796. Rambaut A 2003 Se-Al: a manual sequence alignment editor. http:// evolve.zoo.ox.ac.uk/software.html?id=seal. Rechinger KH 1977 Tragopogon. Pages 83–120 in KH Rechinger, ed. Flora Iranica. Vol 122. Akademische Druck und Verlagsanstalt, Graz. Regel C 1937 Beitrag zur Kentnis einiger Tragopogon. Scr Horti Bot Univ Vytauti Magni 5:19–31. ——— 1955 Die Verbreitung einiger europaeisch-asiatischer Tragopogon-Arten. Ber Schweiz Bot Ges 65:251–262. Richardson IBK 1976 Tragopogon. Pages 322–326 in TG Tutin, VH Heywood, NA Burges, CH Valentine, eds. Flora Europaea. Vol 4. Cambridge University Press, Cambridge. Saar DE, NO Polans, PD Sorensen 2003 A phylogenetic analysis of the genus Dahlia (Asteraceae) based on internal and external transcribed spacer regions of nuclear ribosomal DNA. Syst Bot 28: 627–639. Seelanan TH, H Schnabel, JF Wendel 1997 Congruence and consensus in the cotton tribe. Syst Bot 22:259–290. Soltis DE, TG Collier, ML Edgerton 1991 Chloroplast DNA variation within and among genera of the Heuchera group (Saxifragaceae): evidence for chloroplast transfer and paraphyly. Am J Bot 78:1091–1112. Soltis DE, PS Soltis 1998 Choosing and approach and an appropriate gene for phylogenetic analysis. Pages 1–42 in DE Soltis, PS Soltis, JJ Doyle, eds. Molecular systematics of plants. 2nd ed. Kluwer, Boston. Soltis DE, PS Soltis, JC Pires, A Kovarik, JA Tate 2004 Recent and recurrent polyploidy in Tragopogon (Asteraceae): cytogenetic, genomic, and genetic comparisons. Bot J Linn Soc 82:485–501. Swofford DL 2002 PAUP*: phylogenetic analysis using parsimony (*and other methods). Version 4.10b. Sinauer, Sunderland, MA. Thompson JD, TJ Gibson, F Plewniak, F Jeanmougin, DG Higgins 1997 The Clustal X Windows interface: flexible strategies for multiple sequence alignment aided by quality analysis tools. Nucleic Acids Res 24:4876–4882. Tzvelev NN 1985 Genus Tragopogon L. (Asteraceae) in the European part of the USSR. Pages 238–250 in T Egorova, ed. News in higher plant systematics. Vol 22. Nauka, Leningrad. (In Russian.) Urbatsch LE, RP Roberts, V Karaman 2003 Phylogenetic evaluation of Xylothamia gundlachia and related genera (Asteraceae, Astereae) based on ETS and ITS nrDNA sequence data. Am J Bot 90: 634–649. Wen J, EA Zimmer 1996 Phylogeny and biogeography of Panax L. (the ginseng genus, Araliaceae): inference from its sequences of the nuclear ribosomal DNA. Mol Phylogenet Evol 6:167–177. White TJ, T Brunce, S Lee, J Taylor 1990 Amplification and direct sequencing of fungal ribosomal RNA genes for phyloginetics. Pages 315–322 in M Innis, D Gelfand, J Sinsky, T White, eds. PCR protocols: a guide to methods and applications. Academic Press, San Diego, CA. Wilson DF 1983 Karyotypes of Tragopogon (Compositae: Lactuceae). Brittonia 35:341–350.