1 Nuclear and plastid DNA phylogeny of the tribe Cardueae 2 (Compositae) with Hyb-Seq data: A new subtribal classification and a 3 temporal framework for the origin of the tribe and the subtribes 4 5 Sonia Herrando-Morairaa,*, Juan Antonio Callejab, Mercè Galbany-Casalsb, Núria Garcia-Jacasa, Jian- 6 Quan Liuc, Javier López-Alvaradob, Jordi López-Pujola, Jennifer R. Mandeld, Noemí Montes-Morenoa, 7 Cristina Roquetb,e, Llorenç Sáezb, Alexander Sennikovf, Alfonso Susannaa, Roser Vilatersanaa 8 9 a Botanic Institute of Barcelona (IBB, CSIC-ICUB), Pg. del Migdia, s.n., 08038 Barcelona, Spain 10 b Systematics and Evolution of Vascular Plants (UAB) – Associated Unit to CSIC, Departament de 11 Biologia Animal, Biologia Vegetal i Ecologia, Facultat de Biociències, Universitat Autònoma de 12 Barcelona, ES-08193 Bellaterra, Spain 13 c Key Laboratory for Bio-Resources and Eco-Environment, College of Life Sciences, Sichuan University, 14 Chengdu, China 15 d Department of Biological Sciences, University of Memphis, Memphis, TN 38152, USA 16 e Univ. Grenoble Alpes, Univ. Savoie Mont Blanc, CNRS, LECA (Laboratoire d’Ecologie Alpine), FR- 17 38000 Grenoble, France 18 f Botanical Museum, Finnish Museum of Natural History, PO Box 7, FI-00014 University of Helsinki, 19 Finland; and Herbarium, Komarov Botanical Institute of Russian Academy of Sciences, Prof. Popov str. 20 2, 197376 St. Petersburg, Russia 21 22 *Corresponding author at: Botanic Institute of Barcelona (IBB, CSIC-ICUB), Pg. del Migdia, s. n., ES- 23 08038 Barcelona, Spain. E-mail address: [email protected] (S. Herrando-Moraira). 24

25 Abstract 26 Classification of the tribe Cardueae in natural subtribes has always been a challenge due to the lack of 27 support of some critical branches in previous phylogenies based on traditional Sanger markers. With the 28 aim to propose a new subtribal delimitation, we applied a Hyb-Seq approach to a set of 76 Cardueae 29 species representing all the subtribes and informal groups defined in the tribe, targeting 1061 nuclear 30 conserved orthology loci (COS) designed for Compositae and obtaining chloroplast coding regions as by- 31 product of off-target reads. For the extraction of target nuclear data, we used two strategies, PHYLUCE 32 and HybPiper, and 776 and 1055 COS loci were recovered with each of them, respectively. Additionally, 33 87 chloroplast genes were assembled and annotated. With the three datasets, phylogenetic relationships 34 within the tribe were reconstructed under approaches of concatenation (using supermatrices as input for 35 maximum likelihood analysis with RAxML) and coalescence (species tree estimated with ASTRAL 36 based on the individual gene trees of each COS locus). The phylogenetic analyses of the nuclear datasets 37 fully resolved virtually all nodes with very high support. Although nuclear and plastid tree topologies are 38 highly congruent, they still present some incongruences, which are shortly discussed. On the basis of the 39 phylogenies obtained, we propose a new taxonomic scheme of 12 monophyletic and morphologically 40 consistent subtribes: Carlininae, Cardopatiinae, Echinopsinae, Dipterocominae (new), Xerantheminae 41 (new), Berardiinae (new), Staehelininae (new), Onopordinae (new), Carduinae (redelimited), Arctiinae 42 (new), Saussureinae (new), and Centaureinae. Another main key result of the study was the high 43 resolution recovered at the backbone of the Cardueae tree, which led to obtain better inter-subtribal 44 relationships. Using as tree base the nuclear HybPiper phylogeny, we updated the temporal framework for 45 the origin and diversification of the tribe and subtribes. Overall, the power of Hyb-Seq is demonstrated to 46 solve relationships traditionally suggested by morphology but never recovered with support using Sanger 47 sequencing of a few DNA markers. 48 49 Keywords 50 Asteraceae 51 COS targets 52 Subtribes 53 Systematics 54 Phylogenomics 55 Target enrichment 56 57 58 59 1. Introduction 60 61 The Cardueae is one of the largest tribes of the 43 described in the Compositae (Asteraceae; Funk et al., 62 2009), with almost 10% of the species of the whole family: 2400 species in 73 genera (Susanna and 63 Garcia-Jacas, 2009). The tribe belongs to the subfamily Carduoideae, which is composed by four tribes 64 (Panero and Funk, 2008; Ortiz et al., 2009): Dicomeae (97 spp.), Tarchonantheae (13 spp.), 65 Oldenburgieae (4 spp.), and Cardueae, the latter representing the 95% of the subfamily’s diversity 66 (Susanna and Garcia-Jacas, 2009). The Carduoideae is a very successful evolutionary lineage within 67 Compositae, which was estimated by Panero and Crozier (2016) to have the second highest 68 diversification rate in the family and a moderate rate of extinction. The number of species is not uniform 69 across the genera of the tribe Cardueae, being six of them highly diversified (ca. 200–600 spp.) and with a 70 high endemism rate (Carduus, Cirsium, Centaurea, Cousinia, Jurinea, and Saussurea), while, on the 71 other extreme, 22 (ca. 30%) are monotypic. Geographically, the tribe is distributed mainly in the 72 Mediterranean and the Irano-Turanian regions, but it is reported from all continents except the Antarctica. 73 The ecosystems where the Cardueae species inhabit are very variable, e.g. Mediterranean landscapes, 74 steppes, semiarid areas, deserts, alpine meadows, or tropical savannahs (Susanna and Garcia-Jacas, 2009). 75 Indeed, in many aspects the tribe Cardueae is an extremely heterogeneous group in terms of habit 76 (perennial, biennial, monocarpic or annual herbs, shrubs, treelets, often spiny), karyology (high variability 77 in chromosome numbers from x=6 to x=18, frequent disploidy), or pollen structure (caveate or ecaveate, 78 smooth, scabrate, or spiny). This complexity summed to its high diversity have greatly contributed to the 79 turbulent taxonomic history of Cardueae at several ranks, e.g. tribal and subtribal (see Table 1 for a 80 historical overview) or even misclassifications of some genera in different tribes or subtribes (see 81 examples in Table 2). 82 In the first tribal classification of the family, Cassini (1819) divided the present Cardueae into three 83 tribes: Echinopseae, Carlineae, and Cardueae, the latter comprising the subtribes Carduinae and 84 Centaureinae. Years later, Bentham (1873) and Hoffmann (1894) proposed grouping the three former 85 tribes in a single tribe Cardueae, with four subtribes: Carlininae Dumort., Echinopsinae [“Echinopinae”] 86 Dumort., Carduinae Cass., and Centaureinae Dumort. This conservative approach was accepted over long 87 time, until Wagenitz (1976) reinstated the tribe Echinopseae. Soon after, Dittrich (1977) returned to 88 Cassini's earlier views, proposing the restoration of tribes Echinopseae, Carlineae, and Cardueae. Bremer 89 (1994) favored the conservative classification of a single tribe Cardueae, which is nowadays generally 90 accepted (Susanna and Garcia-Jacas, 2007, 2009), with the recent reinstauration of the subtribe 91 Cardopatiinae (Susanna et al., 2006). 92 Not surprisingly, Carlininae and Echinopsinae, two of the two basalmost subtribes in molecular 93 phylogenies, have been considered at some time as independent tribes, mainly due to their clear 94 diagnostic characters with respect to the rest of the Cardueae assembly (Susanna and Garcia-Jacas, 2009). 95 However, the Cardueae are morphologically consistent as a whole entity at the tribal level, sharing a 96 unique synapomorphic morphological character within the Compositae: style with a papillose-pilose 97 thickening below the branches and the stigmatic areas confined to the inner surface (Susanna and Garcia- 98 Jacas, 2009). Additionally, the tribe Cardueae has also been broadly confirmed empirically as 99 monophyletic, first by cladistic analyses based on morphology (Bremer, 1987, 1994; Karis et al., 1992), 100 and later by molecular phylogenies (Jansen et al., 1990, 1991; Kim et al., 1992; Susanna et al., 1995, 101 2006; Garcia-Jacas et al., 2002; Barres et al., 2013). 102 In the most recent and accepted treatments (Susanna et al., 2006; Susanna and Garcia-Jacas, 2007, 103 2009), the authors pointed out that four subtribes are natural groups with clear limits (Cardopatiinae, 104 Carlininae, Centaureinae, and Echinopsinae); however, subtribe Carduinae is an unnatural, artificial, and 105 problematic group. The Carduinae have been a dumping ground of genera that do not fit in any of the 106 other subtribes and totals 1700 species, near 70% of the whole tribe diversity (Susanna and Garcia-Jacas, 107 2007). The fact that the group is a questionable and heterogeneous assemblage of genera was also 108 reflected in several phylogenetic studies, which have reported the subtribe as paraphyletic (Susanna et al., 109 1995, 2006; Häffner and Hellwig, 1999; Garcia-Jacas et al., 2002; Barres et al., 2013). The alternate 110 solution of combining subtribes Carduinae and Centaureinae in one enormous subtribe was discarded, 111 owing to the impractical constitution of a huge subtribe encompassing 2300 species and 90% of the 112 species of the whole tribe (Susanna and Garcia-Jacas, 2007). 113 With the intention of fragmenting subtribe Carduinae, some informal morphological groups have 114 been described within it (Susanna and Garcia-Jacas, 2007), which could on the long run be considered 115 subtribes for a more natural classification: “[the alternative is] splitting present Carduinae in at least seven 116 new subtribes (many of them presently unsupported): Xerantheminae, Staehelininae, Berardininae, 117 Onopordinae, Carduinae, Arctiinae, and Saussureinae […]” (Susanna and Garcia-Jacas, 2009). In view of 118 the lack of support for the monophyly of all segregate subtribes, especially for the most important in 119 terms of species number (core Carduinae, Arctiinae, and Saussureinae), no formal proposal was 120 performed in the past. A well-resolved and supported phylogenetic hypothesis for the major lineages 121 within Carduinae is still lacking, and it would be very useful to confirm the morphological alliances or 122 generic complexes proposed by Susanna and Garcia-Jacas (2009). 123 An additional taxonomic problem within the tribe Cardueae is that some genera or complex of genera 124 have been classified within one subtribe or another depending on the classification followed (see some of 125 the cases in Table 2). Moreover, in extreme cases, some genera have been independently classified in 126 tribe Cardueae or in other tribes of Compositae (e.g. Berardia in tribe Mutisiae, or Dipterocome in tribe 127 Calenduleae). A great proportion of subtribal misplacements were reported for uncertain cases between 128 subtribes Carduinae and Centaureinae, which could be attributed to the inconspicuous morphological 129 differences between both subtribes (based on microcharacters of achene and pappus) that are sometimes 130 lacking in herbarium specimens, or are immature structures during field collections (Susanna et al., 2006). 131 Even though many cases have been successfully resolved with the aid of the molecular phylogenies 132 obtained with Sanger sequencing (see references in Table 2), more changes in the adscription of genera 133 could occur in a new subtribal framework for the Cardueae. 134 Since to date, all previous molecular studies of the tribe have been based on Sanger sequencing data 135 (Susanna et al., 1995, 2006; Häffner and Hellwig, 1999; Garcia-Jacas et al., 2002; Barres et al., 2013), 136 being the largest datasets constructed with four chloroplast regions (trnL-trnF, matK, ndhF, rbcL) and one 137 nuclear ribosomal marker (ITS; Barres et al., 2013). Although subtribal clades of Carlininae, 138 Echinopsinae, Cardopatiinae, and Centaureinae are well supported, Carduinae remain paraphyletic when 139 Centaureinae are removed, and not all the informal groups have been recovered as monophyletic. 140 Moreover, relationships between subtribes have not been resolved, and the backbone of the phylogenetic 141 tree remains undefined. Accordingly, the available divergence time estimation of Cardueae is based on a 142 partially resolved phylogenetic tree that was performed on a combined dataset of chloroplast markers, and 143 this dating could be significantly improved in terms of the methodological approach and the sequence 144 data used. 145 In the last decade, next generation sequencing (NGS) has emerged as an important methodological 146 advance for solving phylogenetic problems (see review of Harrison and Kidner, 2011). Plant phylogenies 147 of historical taxonomically complex groups are becoming resolved at different taxonomic levels, e.g. 148 order (Ranunculales, cf. Lane et al., 2018; or Zingiberales, cf. Carlsen et al., 2018), family 149 (Goodeniaceae, cf. Gardner et al., 2016), tribe (Shoreeae, cf. Heckenhauer et al., 2018), or species 150 complexes as Claytonia (Stoughton et al., 2018) or Amaranthus (Viljoen et al., 2018). For the 151 Compositae, Mandel et al. (2014) designed a probe set that hybridizes with 1061 nuclear conserved 152 orthology loci (hereafter COS), which in combination with genome skimming allowed also to recover 153 other parts of the genome such as chloroplast regions (defined as Hyb-Seq technique; Weitemier et al., 154 2014). This methodological workflow has been proved successfully in Compositae-wide studies (Mandel 155 et al., 2015, 2017) and was tested in a recent research focused on highly radiated genera within Cardueae 156 (Herrando-Moraira et al., 2018). However, the COS locus set has not been yet applied to taxonomic 157 delimitation within the family. 158 Thereafter, we here apply the Hyb-Seq method to a sample of 76 species representing all the 159 subtribes and informal suprageneric groups of the Cardueae with the main goals of: (1) to obtain a well- 160 defined phylogeny of the high-level lineages in the tribe with nuclear and chloroplast data; on the basis of 161 this phylogeny, (2) to propose a new subtribal classification, especially focused on testing the splitting of 162 subtribe Carduinae in smaller and more practical natural subtribes; (3) to examine previously unresolved 163 phylogenetic relationships between subtribal lineages at the backbone of the tree; and (4) to update the 164 temporal framework of tribe and subtribes origin and diversification, based on a better resolved 165 phylogeny. 166 167 168 2. Materials and methods 169 170 2.1. Taxon sampling 171 172 To obtain a compete sampling of the tribe Cardueae, we included representatives of all described 173 subtribes based on the taxonomic treatment of Susanna and Garcia-Jacas (2009): (1) Carlininae, 6 species; 174 2) Cardopatinae, 2 species; 3) Echinopsinae, 3 species; 4) Carduinae, 34 species; 5) Centaureinae, 31 175 species. The sampling strategy was based on maximizing the number of species of unresolved informal 176 groups, especially within Carduinae and Centaureinae, and was proportional to the total number of genera 177 and species included in each group. In total, 76 different species of the tribe Cardueae were included. We 178 did not include all the genera within the tribe because the adscription of many of the ones with taxonomic 179 problems had already been confirmed in previous works (Table 2). In addition, we incorporated 5 species 180 as representatives of the subfamily Carduoideae outside of the Cardueae; 5 species from other subfamilies 181 within Compositae (2 from the Mutisioideae, 2 from the Barnadesioideae, 1 from the Famatinanthoideae); 182 and finally, Nastanthus patagonicus from the Calyceraceae, which has been recognized as the sister 183 family to Compositae (Mandel et al., 2017). Overall, 58 species were newly sequenced for this study, and 184 the data of the remaining 29 were directly obtained from raw reads from Mandel et al. (2014, 2017) or 185 Herrando-Moraira et al. (2018). See Supplementary Table S1 for details of each sampled species. 186 187 2.2. DNA extraction, library and capture preparation, and sequencing 188 189 Around 10–30 mg of dried plant material per sample was weighed and homogenized with a Mixer Mill 190 MM 301 (Retsch®, Haan, Germany). The DNA was extracted with the DNeasy plant mini kit (Qiagen, 191 Valencia, CA, USA) or the E.N.Z.A SP Plant DNA Mini Kit (Omega Bio-Tek Inc., Norcross, Georgia, 192 USA) according to the manufacturer’s protocol. The total genomic DNA quantity was measured with the 193 Qubit™ 3.0 Fluorometer (Thermo Scientific, Waltham, MA, USA). The standardized DNA (1 µg in 70 194 µl) was sheared in the Genomics Unit of the Centre for Genomic Regulation (CRG, Barcelona, Spain) 195 using a Covaris S2 System (Covaris, Woburn, MA, USA) in microTUBEs with sample volume of 50 μl 196 and a target peak set to 400 bp. Sequencing libraries and subsequent sequence capture were conducted as 197 specified in Herrando-Moraira et al. (2018). Additionally, for the species newly sequenced for this study 198 (see Supplementary Table S1), we conducted a library spiking with the following proportions: 40% of 199 unenriched solution and 60% of enriched libraries. The final spiked library pools were sequenced (pair- 200 end 100 bp) in the DNA Sequencing Core CGRC/ICBR of the University of Florida on one lane of an 201 Illumina HiSeq 3000 (Illumina, USA). 202 203 2.3. Raw data processing 204 205 Demultiplexed raw sequence reads were checked with FastQC v.0.10.1 206 (https://www.bioinformatics.babraham.ac.uk/projects/fastqc/) in order to perform a first quality control 207 step. Subsequently, raw reads were trimmed by a quality based assessment (sliding-window set to 5:20), 208 and were also subjected to an adapter trimming using Trimmomatic v.0.36 (Bolger et al., 2014). In 209 addition, we specified the software to remove those cleaned reads that after trimming were less than 36 210 bp, and those with a missing forward or reverse pair. 211 212 2.4. Nuclear data extraction 213 214 Following the methodological workflow proposed in Herrando-Moraira et al. (2018), the 1061 target COS 215 loci were extracted with two orthology-detection pipeline packages: PHYLUCE v.1.5 (Faircloth, 2015) 216 and HybPiper v.1.1 (Johnson et al., 2016). One of main differences between both approaches is the 217 treatment of potential paralog loci (Herrando-Moraira et al., 2018). We extracted the COS loci with the 218 two pipelines to be sure that the distinct paralog processing strategy was not affecting phylogenetic 219 reconstructions in terms of topology and branch support values. 220 For the PHYLUCE method, the trimmed reads were de novo assembled into contigs with the 221 software SPAdes v.3.9.0 (Bankevich et al., 2012), specifying different predefined k-mer lengths (21, 33, 222 55, and 77). Then, the recovered contigs were mapped back to target references using LASTZ (Harris, 223 2007), and were extracted following the methodological specifications detailed in Herrando-Moraira et al. 224 (2018). For the HybPiper method, the trimmed reads were firstly mapped to the targets using BWA (Li 225 and Durbin, 2009), and secondly were assembled into contigs also using SPAdes. 226 The multi-fasta files obtained for each target locus both with PHYLUCE and HybPiper were aligned 227 with MAFFT v.7.266 (Katoh and Standley, 2013) using the auto setting mode. The ambiguously aligned 228 regions were removed with trimAl v.14 (Capella-Gutiérrez et al., 2009) applying the automated1 flag. 229 126 out of the total 902 loci extracted with PHYLUCE, and 1 locus out of the total 1057 loci extracted 230 with HybPiper, were removed from each dataset because they were obtained for less than three species. 231 One additional locus was removed from the HybPiper dataset due to its short length recovered after the 232 alignment trimming (3 bp). Finally, for each target extraction method, two datasets were constructed in 233 order to perform two different phylogenetic inference analyses. One consisted of the trimmed aligned 234 sequences of each locus separately, and the other consisted of a supermatrix obtained by concatenating all 235 trimmed aligned loci with FASconCAT-G v.1.02 (Kück and Longo, 2014). The summary statistics of the 236 supermatrices were calculated with AMAS (Borowiec, 2016). 237 238 2.5. Chloroplast data extraction 239 240 For the reconstruction of the chloroplast genomes, we used the off-target reads also recovered with the 241 Hyb-Seq approach. The chloroplast extraction was conducted using the pipeline package MITObim v.1.9 242 (Hahn et al., 2013) with the default conditions. The package has incorporated the module MIRA v.4.0.2 243 (Chevreux et al., 1999), which is used in mapping mode. In the first step, it identifies the more conserved 244 regions between the total readpool and a phylogenetically related reference genome (in this case, Cirsium 245 arvense NCBI accession number NC_036965.1) by mapping the trimmed interleaved reads to this initial 246 reference. Then, the mapped reads are assembled into contigs and a new reference sequence or bait is 247 created. To extend the reference close the gaps, two more steps are performed. The second consists of 248 separating from the total read pool the reads that overlap with the reference, and the third recovers the 249 reads separated to map them back to the bait, thus building a new more extended reference sequence 250 through an assembly process of new mapped reads. These two last steps are iteratively repeated until a 251 stationary stage of the mapped and assembled reads is reached (Hahn et al., 2013). 252 The chloroplast genomes recovered were annotated with the web tool application GeSeq (Tillich et 253 al., 2017), which uses a customizable reference database to annotate the genomes using BLAT-driven 254 best-match approach. As a database, we selected all the available annotations in NCBI of Cardueae 255 representatives (Carthamus tinctorius NC_030783.1, Centaurea diffusa NC_024286.1, Cirsium arvense 256 NC_036965.1, Cirsium eriophorum NC_036966.1, Cirsium vulgare NC_036967.1, Cynara baetica 257 NC_028005.1, Cynara cornigera NC_028006.1, Cynara humilis NC_027113.1, Saussurea 258 chabyoungsanica NC_036677.1, Saussurea involucrata NC_029465.1, Saussurea polylephis 259 NC_036490.1, and Silybum marianum NC_028027.1). To extract the coding regions of genes (CDS) in 260 separate files from the global multi-fasta matrix generated with GeSeq, we used the script 261 “Phyluce_assembly_explode_get_fastas_file” of PHYLUCE package with slight modifications. The 262 recovered CDS were individually aligned with the MACSE codon aligner (Ranwez et al., 2011). The 263 alignments were visualized in SeaView v.4.7 (Gouy et al., 2009), and 10 species from the initial sampling 264 were excluded due to the high presence of frameshifts and preceding codon stops. To remove the poorly 265 aligned regions from the alignments, we used the program Gblocks v.0.91b (Castresana, 2000) with the 266 option “codon” to trim entire codon sets. A total 87 CDS regions were recovered. We concatenated all 267 CDS in a single supermatrix (chloroplast dataset) with FASconCAT-G v.1.02, assuming that the 268 chloroplast genome is considered to be haploid, nonrecombinant, generally uniparentally inherited, and 269 “single copy” (Small et al., 1998, 2004). 270 271 2.6. Phylogenetic inference analyses 272 273 The phylogeny reconstruction of the tribe Cardueae was performed under two different approaches for 274 nuclear data: the concatenation approach (using the supermatrix dataset) and the coalescence approach 275 (using the individual gene trees for inferring the species tree). For chloroplast data, the phylogenetic tree 276 inference was only performed under the concatenation approach. 277 For the concatenation approach, Maximum Likelihood (ML) analyses were run using the software 278 RAxML v.8.2.9 (Stamatakis, 2014), which were conducted on XSEDE in the CIPRES Science Gateway 279 v.3.1 (Miller et al., 2010). We selected the algorithm of simultaneously searching the best ML tree (from 280 10 randomized maximum parsimony starting trees) and performing a rapid bootstrapping (1000 281 replicates). Branches were considered as statistically supported for bootstrap (BS) values > 70% (Hillis 282 and Bull, 1993). In relation to the partition scheme, each locus (in the nuclear datasets) or gene (in the 283 chloroplast dataset) was considered a unit of partition, using the evolution model of GTRGAMMA 284 following recommendations of Stamatakis (2006). Output trees were visualized and exported in FigTree 285 v.1.4.3 (Rambaut, 2016). 286 For the coalescence approach, gene trees were inferred also using RAxML, under the same conditions 287 as for the concatenation based analyses but with a bootstrap resampling of 200 replicates. Resulting 288 unrooted gene trees were inputted into ASTRAL-III v.5.5.3 (Zhang et al., 2018) to infer the species tree. 289 The support values were calculated using local posterior probabilities (LPP; Sayyari and Mirarab, 2016). 290 Branches with LPP > 0.95 were considered as strongly supported. 291 292 2.7. Gene trees concordance analyses 293 294 Topological conflicts among gene trees and the species tree were explored using the software Phyparts 295 (Smith et al., 2015; https://bitbucket.org/blackrim/phyparts) on the nuclear HybPyper dataset. This 296 methodological approach consists of mapping, for each node or bipartition of interest, the level of 297 concordance/discordance between the different individual gene trees on supporting the reference tree 298 topology, being this the ASTRAL species tree in this case. First, gene and species trees were rooted with 299 the online tool STRAW (Shaw et al., 2013; 300 http://bioinformatics.publichealth.uga.edu/SpeciesTreeAnalysis/). The outgroup selection was based on 301 the species divergence rank recovered by the species ML tree inferred under concatenation. When the 302 first preferred outgroup was missing, STRAW searches the next specified species. Then, Phyparts was 303 run, considering only the branches of gene trees with more than 33% BS support. Finally, the script 304 phypartspiecharts.py (https://github.com/mossmatters/MJPythonNotebooks) was used to summarize and 305 map the Phyparts output on the ASTRAL species tree. Owing to computing limitations of Phyparts, we 306 could not run the analysis for the entire set of 1055 nuclear gene trees of HybPiper dataset, therefore a 307 subset of 750 random gene trees was selected. 308 309 2.8. Divergence time analysis 310 311 The divergence time analysis was performed on the phylogenetic tree obtained under concatenation from 312 the nuclear HybPiper dataset. This tree was time-calibrated rescaling the branch lengths using the 313 penalized likelihood method (Sanderson, 2002), implemented in the software treePL (Smith and 314 O’Meara, 2012). This approach is reported to produce similar results than the software BEAST (e.g. 315 Lagomarsino et al., 2016; Stubbs et al., 2018), but, as an advantage, is capable of computing larger 316 datasets in a lower running time. The dating procedure was divided in two main stages, which consisted 317 on: (1) selection of the optimal model parameters; and (2) running the analysis with the optimal 318 parameters selected and, additionally, accounting for the uncertainty in calibration points to obtain 319 confidence intervals (95% CI) in the estimated node ages. 320 For both stages, five calibration points (CP) were used as node age constrains. One was a secondary 321 dated node corresponding to the origin of the Compositae family (69.56 Myr with a 95% CI 59.0280.17; 322 coded here as CP1) reported in Panero and Crozier (2016). The other four points were based on fossil 323 records: (CP2) the capitulescence of Raiguenrayun cura that was dated at 47.5 Myr (Barreda et al., 2012), 324 which was used to constrain the clade of Subfamily Mutisioideae + Subfamily Carduoideae as in the most 325 recent interpretation by Panero and Crozier (2016), instead of placing it at the crown Compositae as in 326 Barres et al. (2013); (CP3) the achenes identified as belonging to Cirsium with an age of 14 Myr (Mai, 327 1995), which were placed at the stem node of the Carduus-Cirsium clade following Barres et al. (2013); 328 (CP4) the achenes assigned to Arctium (López-Vinyallonga et al., 2009) and dated at 8 Myr (Mai, 2001), 329 which were placed at the stem node of Arctium, correcting the misplacement of this fossil in Barres et al. 330 (2013), where it was used at the split of A. minus and A. lappa; and (CP5) the pollen of Centaurea type 331 Cyanus dated at 6 Myr (Wagenitz, 1955; Ivanov et al., 2007), which was placed at the stem node of 332 Centaurea subgenus Cyanus. This is a new, more precise and better resolved Centaurea fossil that the 333 pollen and achene records of Mai (1995) and Popescu (2002) that were used in Barres et al. (2013), where 334 they were placed at stem age of Centaurea due to their uncertain placement within the genus. 335 In the first dating stage, an initial “priming” run (prime command) was carried out to detect the 336 optimal parameter settings (opt = 4, optad = 2, and optcvad = 2). Subsequently, the best smoothing rate 337 was also evaluated through a random subsample and replicate cross-validation procedure (selecting 338 thorough and randomcv commands), allowing varying it from 0.001 (cvstart) to 1000 (cvstop). The best 339 smoothing value resulted in 0.1 after cross-validation analysis, which was selected from the lowest value 340 of a Chi-Square test. The two runs of the first stage were run using the range of 95% CI value for the CP1 341 (minimum = 59.02 and maximum = 80.17) and the lower bound of fossil age estimations (minimum = 342 fossil age) for CP2-CP5. 343 In the second stage, we used our own R (R Core Team 2014) script (Appendix A) to obtain 10,000 344 random values for each CP in order to account for the uncertainty of the node age estimations. 345 Specifically, the random values were generated under a normal distribution for CP1 (based on 95% CI 346 estimations of Panero and Crozier, 2016; mean = 69.56 and sd = 3) and a lognormal distribution for CP2- 347 CP5 (mean = fossil age and sd = 1.1). Once values were extracted, a new data frame was created in a 348 spreadsheet (Appendix B), which was then used to randomly select 100 unique combinations of five 349 numbers, i.e. one for each CP from the random pool of 10,000. Finally, 100 independent treePL analyses, 350 each one with CP constrained to a single value (minimum = maximum), were ran using the same 351 phylogenetic input tree and the optimized parameters found in the first stage. The resultant 100 dated tree 352 files were modified to fit the format of TreeAnnotator v.1.7.5 (Drummond et al., 2012), and were 353 introduced in this software to obtain a maximum clade credibility tree chronogram with median node 354 heights and corresponding CIs. 355 356 357 3. Results and discussion 358 359 3.1. Performance of Hyb-Seq at the subtribal level within the tribe Cardueae 360 361 The Hyb-Seq NGS technique designed for the Compositae (Mandel et al., 2014) is confirmed here as a 362 powerful tool to reconstruct highly resolved phylogenies at deep taxonomic levels, in line with previous 363 studies more focused on the methods (Mandel et al., 2015, 2017; Herrando-Moraira et al., 2018). As far 364 as we known, this is the first study in Compositae that uses the COS loci workflow to obtain taxonomic 365 conclusions, specifically aimed to clarify the entangled delimitation of subtribes within Cardueae. The 366 transition from traditional Sanger datasets, up to 5 markers, to NGS datasets, with 1142 markers (Table 367 3), has resulted in obtaining the most helpful dichotomous and confidently supported divergence pattern 368 reported to date for Cardueae (Figs. 1, 2, and 3, Supplementary Figs. S1 and S2). 369 In average 5,420,504 reads (sd = 3,461,361) were sequenced per species (Supplementary Table S2). 370 From the total 1061 COS loci, 776 were finally recovered and used with PHYLUCE and 1055 with 371 HybPiper, which conformed matrices of trimmed aligned sequences of 492,549 bp (189,716 parsimony 372 informative [PI] sites) and 332,260 bp (123,73 PI sites), respectively (Table 4). When parameters related 373 to missing data were compared between PHYLUCE and HybPiper (see parameters 4 and 11 in Table 4), 374 it was remarkable that PHYLUCE is less efficient than HybPiper in the sequence extraction process from 375 on-target reads, as documented in Herrando-Moraira et al. (2018). However, phylogenetic trees did not 376 differ in topology or branch supports between both extraction methods at the subtribal level (Figs. 1 and 377 3, Supplementary Figs. S1 and S2), even though PHYLUCE uses a more restrictive procedure to remove 378 potential paralogs than HybPiper (see Fig. 1 in Herrando-Moraira et al., 2018). As no differences were 379 observed, results from HybPiper method were analyzed in deep to support taxonomic proposals. 380 Interestingly, this lack of differences could be reflecting that paralogy incidence is relatively low in 381 Cardueae. In agreement, the study of Jones et al. (in prep.), which tests the COS loci applicability for 382 seven tribes of Compositae, found that Cardueae was the tribe with the lowest number of paralogs 383 (average = 140, similar to the value found here, average = 130), in comparison with others like 384 Vernonieae (average = 256). Also in concordance, Huang et al. (2016) did not detect any recent whole 385 genome duplication event in Cardueae, in contrast to other Compositae tribes, which could involve 386 paralogy issues. 387 As other studies have shown, Hyb-Seq can provide nearly complete datasets of chloroplast genomes 388 (e.g. Weitemier et al., 2014; Folk et al., 2015; Carlsen et al., 2018). Regarding the present study, we were 389 able to retrieve 87 protein coding regions of the chloroplast genome from off-target reads, which atfter 390 concatenation resulted in a matrix of 78,531 bp (4290 PI sites; Table 4). Indeed, the increase of 391 phylogenetically informative characters is responsible of notably higher topological resolution and clade 392 support in comparison with the previous results reported with Sanger sequencing data for the tribe 393 (Susanna et al., 1995, 2006; Häffner and Hellwig, 1999; Garcia-Jacas et al., 2002; Barres et al., 2013; 394 Table 3). In a similar way, this effect has been already documented for other plant groups (e.g. Pinus, 395 Parks et al., 2009; Inga, Nicholls et al., 2015), which have benefited tremendously from the advent of 396 high-throughput sequencing techniques. 397 398 3.2. The new subtribal classification of the tribe Cardueae 399 400 We can confidently re-confirm the monophyly of subfamily Carduoideae as currently defined (tribes 401 Dicomeae, Oldenburgieae, Tarchonantheae and Cardueae; cf. Funk et al., 2009). The tribe Cardueae 402 results again a robust taxonomic unit, in agreement with former studies based on morphology and 403 molecular data (Bremer, 1987, 1994; Jansen et al., 1990, 1991; Karis et al., 1992; Kim et al., 1992; 404 Susanna et al., 1995, 2006; Garcia-Jacas et al., 2002; Barres et al., 2013). Specifically, the clade of 405 Cardueae presented high support values in all the analyses (Figs. 1, 2, and 3). The alternative option of 406 rising up some subtribes to a tribal rank (see Table 1) is once again discarded. As suggested by Susanna et 407 al. (2006), it is unpractical the fragmentation of a natural group that is strongly phylogenetically 408 consistent and can be easily recognized by macromorphological characters. In this way, the classification 409 criteria for the definition of Compositae tribes are also maintained homogeneously along the family (Funk 410 et al., 2009). 411 At subtribal level, the high resolution of phylogenies provided by the Hyb-Seq data makes possible 412 the proposal of a new classification. Subtribal limits established here are based on the application of the 413 integrative taxonomy principles (Dayrat, 2005; Schlick-Steiner et al., 2010): firstly, considering the 414 morphological entities to be recognized (Susanna and Garcia-Jacas, 2007, 2009), and then verifying their 415 monophyly with molecular data. Following this model, a total of 12 subtribes are suggested. Five of them 416 had been already previously recognized and are also confirmed with our results (Carlininae, 417 Echinopsinae, Cardopatiinae, Carduinae, and Centaureinae; Susanna et al., 2006); seven new subtribes 418 result from partitioning the former wide and paraphyletic subtribe Carduinae into new monophyletic 419 subtribes based on previously defined informal morphological groups (Dipterocominae, Xerantheminae, 420 Berardiinae, Staehelininae, Onopordinae, Arctiinae, and Saussureinae). These new subtribes are highly 421 heterogenic in terms of species richness, encompassing from relict single lineages to highly recently 422 radiated groups. For instance, Dipterocominae and Berardiinae are defined as monotypic subtribes, 423 comprising only Dipterocome pusilla and Berardia subacaulis, respectively. In an opposed way, 424 Arctiinae and Saussureinae harbor ca. 600 and 550 species, respectively (Susanna and Garcia-Jacas, 425 2007). For a complete description of subtribes see below Taxonomic proposal. The new phylogenetic 426 reconstruction of high-level lineages presented here will be used as a classification basis for future 427 botanical studies of Cardueae. 428 Certainly, the historical overview of the tribe (Table 1) reveals that Carlininae, Echinopsinae, and 429 Centaureinae are the best morpho-molecular defined entities, which have been long recovered with 430 statistical support since the first phylogenetic studies (Table 3). In accordance, we obtained high clade 431 support for these subtribes (BS = 100 and LPP = 1; Figs. 1, 2, and 3). The Carlininae initially showed a 432 monophyletic pattern in Garcia-Jacas et al. (2002). However, the incorporation of Tugarinovia mongolica 433 broke the monophyly of the subtribe in Susanna et al. (2006). This was probably caused by the fact that it 434 is a dioecious species, a trait unique in the tribe (Susanna and Garcia-Jacas, 2009) and relatively 435 uncommon in angiosperms (6% of species; Renner and Ricklefs, 1995). It has been suggested that dioecy 436 has the negative impact of decreasing diversification rates (Heilbuth, 2000; Kay et al., 2006) with 437 consequent phylogenetic misplacements (Vamosi et al., 2003). In Barres et al. (2013) and our present 438 study, the increase in the number of molecular markers aided to recover the monophyly of Carlininae, 439 including Tugarinovia mongolica within this subtribe as a sister lineage to the rest of members. The 440 Echinopsinae presented a persistent monophyly across all molecular studies (Table 3), except for the 441 chloroplast-based phylogeny of Garcia-Jacas et al. (2002), in which the limited sequence variation found 442 in chloroplast matK gene could have been the responsible for the low branch support of Echinopsinae. 443 Similarly, the monophyly of subtribe Centaureinae has been fully supported in former phylogenetic 444 studies (Table 3), with an also a unique exception of the chloroplast tree obtained with the matK gene by 445 Garcia-Jacas et al. (2002). 446 Although Cardopatiinae is a relatively newly recovered subtribe (Susanna et al., 2006), it has resulted 447 also in a strong-supported clade in recent phylogenies with both nuclear and chloroplast sequence data 448 (Table 3), besides its morphological singularity (Susanna and Garcia-Jacas, 2007, 2009). In agreement, 449 we can confirm here that the two species of Cardopatiinae (Cardopatium corymbosum and Cousiniopsis 450 atractyloides) constitute a well-defined evolutionary segregation (BS = 100 and LPP = 1; Figs. 1, 2, and 451 3), therefore its taxonomic rank as subtribe seems appropriate. 452 Conversely, the new seven subtribes have not always been well delineated in Sanger-based 453 phylogenies, at least with enough confidence to propose a formal subtribal splitting. From the 454 morphological informal groups of Carduinae described in Susanna and Garcia-Jacas (2009), the 455 Xeranthemum group is one of the best supported of all datasets and analyses (BS = 100 and LPP = 1; 456 Figs. 1, 2, and 3). Thus, its consideration as subtribe Xerantheminae can be corroborated. Our results are 457 not surprising, given that since first phylogenies the Xerantheminae species were recovered as a 458 monophyletic and statistically highly supported group, both with nuclear and chloroplast data (Table 3). 459 Related to the former Xeranthemum group, we found one unexpected finding, which is the 460 phylogenetic isolated position of Dipterocome pusilla. This species had not been incorporated to 461 Cardueae phylogenies until the study of Anderberg et al. (2007), confirmed by Barres et al. (2013), where 462 it was clustered with the rest of Xerantheminae members in the combined cpDNA (matK, ndhF, rbcL, and 463 trnL-trnF) tree, but without support in a phylogeny of the ITS nuclear region. However, in the present 464 study we found that in four of the five reconstructed phylogenies (Figs. 2 and 3, Supplementary Figs. S1 465 and S2), Dipterocome is placed in a distinct clade from Xerantheminae. Only in the tree reconstructed 466 under concatenation with the nuclear HybPiper dataset Dipterocome was grouped with Xerantheminae 467 (BS = 97; Fig. 1), in agreement with the results of Barres et al. (2013). This may probably be caused by 468 the bias produced by concatenation analysis of NGS datasets with a large number of loci (Kubatko and 469 Degnan, 2007; Edwards et al., 2016), that can result on long branch attraction artefacts (Liu et al., 2015). 470 Morphologically, Susanna and Garcia-Jacas (2009) pointed out that floral characters of Dipterocome 471 show close relationship with the Xeranthemum complex. Nevertheless, a histological analysis of the 472 capitula structure of Dipterocome has shown that the purported “florets” of Dipterocome are actually 473 uniflowered capitula (Susanna, pers. comm.) grouped in second-order synflorescences, an extremely 474 unfrequent structure in the tribe shared only by the genus Echinops. Overall, it seems that there is enough 475 molecular and morphological evidence to classify Dipterocome in a separate monotypic subtribe. 476 Two other controversial genera considered to belong to the former Carduinae are Berardia and 477 Staehelina. Although both genera do not present obvious morphological affinities (Susanna and Garcia- 478 Jacas, 2009), some molecular phylogenies suggested that they could be evolutionary closely related. The 479 genus Staehelina has long been reported as a genetically isolated group within the tribe, presenting 480 always a monophyletic pattern throughout all phylogenetic reconstructions (Table 3). Regarding its 481 relationships with Berardia, the combined nuclear (ITS) plus chloroplast (trnL-trnF and matK) in 482 Susanna et al. (2006) grouped both genera in an unsupported clade, while only considering nuclear data 483 (ITS), they appeared not directly linked but as adjacent lineages, as detected in nuclear and chloroplast 484 phylogenies of Barres et al. (2013). Here, we also identified the same two phylogenetic signals. Even 485 though, only one of five phylogenetic analyses (chloroplast dataset analyzed under the concatenation 486 approach) resulted in a moderately-high grouping of both genera (BS = 76; Fig. 2), and the rest separated 487 Berardia and Staehelina (Figs. 1 and 3, Supplementary Figs. S1 and S2). As we hypothesized for the case 488 of Dipterocome, Berardia could be misplaced in some chloroplast phylogenies by a long branch attraction 489 effect in concatenated analysis due to its relict character without any known alive direct congener. 490 Conclusively, Berardia and Staehelina could be considered as separate new subtribes Berardiinae and 491 Staehelininae based on the new molecular evidences presented. 492 The Onopordum group is another lineage of former Carduinae with a consistent morpho-genetic 493 historical division (Susanna and Garcia-Jacas, 2009). Except for the chloroplast matK phylogeny of 494 Garcia-Jacas et al. (2002), probably with limited resolution power as mentioned above for other groups, 495 the rest of molecular studies (Table 3) and our genomic phylogenies perfectly outlined the Onopordum 496 assembly (BS = 100 and LPP = 1; Figs. 1, 2, and 3). Hence, it seems reasonably to assign the subtribal 497 rank as Onopordinae. 498 One of most traditionally difficult alliances to delimit within the tribe, and one of the largest 499 complexes of former Carduinae, is the Carduus group (ca. 500 species; Susanna et al., 2006). The main 500 reason for its complicate delimitation was attributed by Susanna et al. (2006) to the fact that the group 501 encompasses a great variety of life strategies; annual or biennial species (e.g. Carduus, Galactites, 502 Picnomon, Silybum, or Tyrimnus) or perennials (as many Cirsium, Cynara, Lamyropsis, or Ptilostemon). 503 Certainly, different generational times that affect mutation rates may hinder a proper phylogenetic 504 comparison among species. The overview of previous phylogenetic studies reflects the difficulty of 505 reconstructing the Carduus group as monophyletic, which was only achieved in three from the total eight 506 Sanger-based phylogenies (Table 3). In contrast, we found a strong phylogenetic signal with Hyb-Seq 507 data (BS = 92–100 and LPP = 1; Figs. 1, 2, and 3), which reinforces the morphological hypothesis of 508 Susanna et al. (2006) and favor its proposal as subtribe Carduinae. Susanna and Garcia-Jacas (2009) 509 suggested a posterior division of the Carduus group into Cynara and Carduus-Cirsium complexes, but 510 our phylogenies did not show a strictly bifurcating divergence pattern for these species complexes (Figs. 511 1, 2, and 3), hindering to formalize their segregation at subtribal rank. 512 The Arctium-Cousinia and Jurinea-Saussurea complexes have been other of classically entangled 513 groups of former Carduinae. Although both conform well defined assemblies by morphology (Susanna 514 and Garcia-Jacas, 2009), there has not been any Sanger-based phylogeny able to correctly recover the 515 morphological pattern as monophyletic lineages (Table 3). Only two isolated molecular studies, Häffner 516 and Hellwig (1999) and Susanna et al. (2006), reported Arctium-Cousinia and Jurinea-Saussurea as 517 monophyletic, respectively. In recent times, the application of Hyb-Seq technique and the COS loci set of 518 Mandel et al. (2014) for Compositae, aided to shed light on phylogenetic connections of these four 519 genera. Phylogenomic trees performed in Herrando-Moraira et al. (2018) showed a clear sister 520 relationship between Arctium-Cousinia and Jurinea-Saussurea, which also formed two monophyletic 521 supported clades. In agreement, on a wider taxonomic sampling of the tribe, we found the same 522 monophyletic pattern for both complexes under concatenation and coalescence approaches with nuclear 523 data (BS = 100 and LPP = 1; Fig. 1 and 3A). However, the chloroplast phylogenies analyzed under 524 concatenation recovered an unsupported clade of Saussurea-Cousinia-Arctium (Fig. 2) with Jurinea as a 525 separate lineage, hence leaving the complex Jurinea-Saussurea as a paraphyletic assembly (Fig. 2). 526 Similarly, a morphologically incongruent grouping pattern has been also recovered in the Sanger-based 527 chloroplast phylogenies of Barres et al (2013), where the genus Saussurea-Jurinea-Cousinia were 528 clustered together with support and Arctium was placed as sister clade. 529 Overall, it seems that even with massive Hyb-Seq data, a clear cyto-nuclear discordance exists 530 between evolutionary histories inferred for Arctium-Cousinia and Jurinea-Saussurea groups. Two factors 531 could explain such pattern of incongruent genealogies (Morales-Briones et al., 2018), one more related 532 with tree inference methods (phylogenetic error), and the other with inherent evolutionary history of the 533 genera (incomplete lineage sorting [ILS] or hybridization). Firstly, it is stated that using gene fragments 534 with limited phylogenetic signal may derive in an increasing of both stochastic and systematic errors in 535 reconstructed phylogenies (Lemmon et al., 2009, Lemmon and Lemmon, 2013; Carlsen et al., 2018). In 536 agreement, if we compared our nuclear vs. chloroplast datasets, a large difference exists in terms of 537 phylogenetic informativeness between both, for instance, in PI sites (ca. 38% vs. 5.5%; Table 4) or the 538 average branch length of trees (0.04 vs. 0.004). Chloroplast data from protein coding regions, in 539 comparison with hundreds of nuclear loci, could be providing limited phylogenetic information, which 540 may explain the differences in the resulting tree topologies. 541 Secondly, processes like ILS or reticulation events remain as possible alternatives to explain the 542 nuclear-chloroplast conflicting topologies, as well as gene tree incongruences. The short internal branches 543 observed at the base of genera in chloroplast tree from concatenation analysis (Fig. 2B) could match the 544 signals left by ILS, i.e. a scenario of a rapid diversification of genera in a short timeframe (rapid 545 bifurcating cladogenesis) or a simultaneous polychotomous divergence (hard polytomy). Past 546 hybridization events could also have taken place as suggested by the concatenation chloroplast tree 547 topology (Fig. 2). The chloroplast genome of ancestral Saussurea lineage could have been lost due to an 548 organelle capture, thus, the currently observed is that retained from an ancestral hybridization with the 549 Arctium-Cousinia lineage. Cyto-nuclear discordances due to introgressive hybridization have been 550 suggested in other plant groups like Ficus through chloroplast NGS phylogenies (Bruun-Lund et al., 551 2017). But more explicit methods to address these processes should be performed (e.g. Folk et al., 2017; 552 García et al., 2017; Mitchell et al., 2017; Morales-Briones et al., 2018; Knowles et al., 2018), in addition 553 to a wider species sampling. 554 Taking into account all shortcomings of chloroplast data, it would be reasonable to give more weight 555 to nuclear results, which were also in agreement with the morphological hypothesis, and accept two new 556 subtribes: Arctiinae (comprising Arctium-Cousinia) and Saussureinae (Saussurea-Jurinea). 557 558 3.3. Phylogenetic relationships among subtribes 559 560 Undoubtedly, one of major improvements that we obtained using Hyb-Seq data is the high phylogenetic 561 resolution recovered along the backbone of Cardueae tree, which is very useful to disentangle subtribal 562 relationships (Fig. 4). Although in most recent Sanger-based phylogenies, 8 of the total 12 circumscribed 563 subtribes were already monophyletically well defined (see Table 3; Barres et al., 2013), their 564 interrelationships remained unresolved since to date. In the phylogenetic trees of Barres et al. (2013), 565 those inferred under maximum parsimony were completely undefined for the subtribal nodes, and under 566 Bayesian inference most of the subtribe branches were anchored in polytomic structures, especially in the 567 nuclear ITS phylogeny (Fig. 4A). In contrast, our phylogenetic reconstructions showed a significantly 568 supported bifurcating pattern (Figs. 4C, D, and E). 569 The subtribes Carlininae, Cardopatiinae, and Echinopsinae were placed as the first diverging linages 570 within tribe Cardueae near the root of the tree. Unfortunately, we did not find which is the sister subtribe 571 to the rest with confidence, due to a cyto-nuclear discordance. In the nuclear tree inferred under 572 concatenation, the Carlininae were sister to the rest (Fig. 4C); but individual gene trees show conflicting 573 topologies at this part of the tree (Figs. 3 and 4E); and in the chloroplast tree inferred under concatenation 574 the Cardopatiinae were sister to the rest (Fig. 4D). Successive sister clades recovered with all 575 reconstruction methods occur in order: Dipterocominae, Xerantheminae, Berardiinae, Staehelininae, and 576 Onopordinae. Note that, as mentioned above, Dipterocominae appeared nested with Xerantheminae and 577 Berardiinae with Staehelininae in the nuclear and chloroplast concatenated analyses, respectively. The 578 grouping pattern of the rest of subtribes was slightly different on nuclear and plastid phylogenies. In 579 nuclear trees, Carduinae was placed as sister to Arctiinae-Saussureinae-Centaureinae clade (BS = 100 and 580 LPP = 0.99), which was subsequently bifurcated in Arctiinae-Saussureinae (BS = 100 and LPP = 1) and 581 Centaureinae clade (Figs. 4C, E). In the chloroplast tree, Carduinae-Arctiinae-Saussureinae-Centaureinae 582 were clustered in a supported clade (BS = 100) that internally formed an unsupported group of Carduinae- 583 Arctiinae-Saussureinae (BS = 63) with Centaureinae as sister to the rest (Fig. 4D). 584 Here, despite Hyb-Seq data offered bifurcating reconstructions, the short branch lengths recovered 585 (Figs. 1B, 2B) may confirm the previous suspicion of a rapid subtribal lineage divergence, which would 586 explain the traditional difficulty to resolve this part of the tree. The Phyparts analysis performed allowed 587 us to go one step further and inspect the degree of gene tree discordance along nodes of Cardueae tree. 588 Results of Phyparts indicated that most of gene tree topologies are in conflict with the species tree in 589 internal subtribal nodes (great red proportion; Fig. 3), in contrast to shallow nodes of genera or species 590 relationships that showed more congruent gene tree histories (more proportion of blue; Fig. 3). We also 591 detected that a considerable proportion of gene trees were poorly supported and uninformative (gray color 592 in Fig. 3 for gene tree branches with BS < 30). Considering all results, possible sources of gene tree 593 discordance at the backbone could derive from: (1) insufficient phylogenetic signal in sampled loci (e.g. 594 short loci length, high missing positions or taxa as a result of capture failure, or poorly variable loci; 595 Villaverde et al., 2018); (2) extinction of critical lineages; (3) rapid splitting of major subtribal clades; or 596 (4) gene flow among ancestral lineages or ILS effect (Morales-Briones et al., 2018). It is important to 597 notice that although individual gene tree bipartitions were mainly in conflict at the subtribal level, the 598 coalescence species tree was, in general, not weakly supported (Fig. 3). Thus, this limitation is not 599 necessary problematic to the aim of subtribal delimitation intended here. 600 601 3.4. Dating framework for the tribe Cardueae 602 603 Figure 5 shows the divergence time estimates of Cardueae based on the nuclear tree obtained with 604 HybPiper and the concatenation approach (1055 COS loci). Median age values and 95% CI are presented 605 for each node in Supplementary Table S3. Our dating analysis revealed that the tribe Cardueae could have 606 originated between Late Eocene and Early Oligocene 34.12 Myr (29.97–40.04 95% CI). Generally, this 607 time frame is in a middle range of the estimated age origin for the tribe in previous dated phylogenies that 608 included several Cardueae members. On the one hand, the studies conducted by Kim et al. (2005) and 609 Panero and Crozier (2016) reported a younger age of the tribe, 24–29 Myr and 22 Myr (11–33 95% CI), 610 respectively. On the other hand, Barres et al. (2013) and Huang et al. (2016) found an older tribe median 611 age with values of 40.15 Myr (35.80–44.26 95% CI) and ca. 42 Myr, respectively. Taxon sampling, 612 calibration points, and molecular markers used could be responsible of dating differences among studies. 613 The subtribes originated in two different main time periods. The subtribes Carlininae, Cardopatiinae, 614 Dipterocominae, Xerantheminae, and Berardiinae probably originated during the Oligocene (see Fig. 5 615 and Supplementary Table S3 for age of subtribes), whereas Echinopsinae, Staehelininae, Onopordinae, 616 Carduinae, Arctiinae, Saussureinae, and Centaureinae originated during Middle-Late Miocene (Fig. 5 and 617 Supplementary Table S3). 618 If we draw in parallel dating results obtained here and the ones obtained by Barres et al. (2013), 619 which is the most comparable study in terms of taxon sampling, we can observe that tribe and subtribe 620 age estimates were broadly older in the latter than those we found here (see Fig. 6). At tribal level, the 621 different placement of the fossil Raiguenrayun cura (dated to 47.5 Myr, placed here to constrain 622 subfamily Mutisioideae + subfamily Carduoideae; CP2 in Fig. 6) does not explain why we obtained a 623 most recent tribal origin than Barres et al. (2013), who placed it in a more basal node as it is the crown 624 node of Compositae family. Probably, the fact that we now have a more complete sampling at the basal 625 part of the tree, including more representatives of subfamilies other than Carduoideae and more 626 representatives of tribes within Carduoideae other than Cardueae, results in greater time distances 627 between the origin of the family Compositae and the tribe Cardueae. At subtribal level, the greatest 628 differences between non-overlapping age ranges were detected for subtribes Onopordinae and 629 Centaureinae, for which the median age differed among studies in 9.32 Myr and 15.38 Myr, respectively. 630 The different number and nature of molecular markers used would be one of main causes of the distinct 631 dating results. We employed a set of 1055 low-copy nuclear loci biparentally inherited that potentially 632 provides more sequence variation, while Barres et al. (2013) used four chloroplast regions that are usually 633 maternally inherited and have low mutation rates, which would explain the older age values reported in 634 their estimations. In addition, it should be also noted that for those nodes constrained with minimum ages 635 based on fossil calibration points, we obtained a better fitting between minimum age assigned and final 636 dating result recovered in comparison with Barres et al. (2013); for instance, our CP5 constrained at 6 637 Myr resulted in 5.91 Myr (5.21–7.01 95% CI), and in Barres et al. (2013) the CP used at the stem node of 638 Centaurea constrained at 5 Myr resulted in 15.54 Myr (12.10–19.96 95% CI). 639 In a general overview of the dated phylogenetic tree it is noteworthy that CI interval limits are wider 640 and more overlapped among nodes at the base of the tree than those at the tips (Fig. 5). Accordingly, 641 lower precision and resolution was obtained for estimated ages of non-Cardueae and early diverging 642 lineages of the Cardueae (e.g. Carlininae or Cardopatiinae). Certainly, the fact that the standard deviation 643 of our most basal and secondary dated calibration point used (CP1) can be higher than the other fossil- 644 based CPs may cause larger uncertainty at the tree base. Moreover, the number of species included as 645 Cardueae outgroups in our study is relatively poor, particularly compared with Panero and Crozier (2016) 646 who used a wide representation of Compositae subfamilies but few representatives for each tribe. The 647 different taxon sampling used could be the reason for differences in the chronograms obtained, resulting 648 the poorly sampled nodes with a more uncertain dating value. As Linder et al. (2005) suggested, taxon 649 sampling is one of most influential factors in molecular clock dating analysis. For this reason, resulting 650 node estimates at the tree base obtained here should be taken with caution, and undoubtedly, a wider 651 sampling of other family members could improve lineages placement and dating resolution at the base of 652 Cardueae. 653 654 3.5. Taxonomic proposal 655 656 Subtribe Carlininae Dumort. 1827. 657 Perennial herbs or shrubs, less often annual plants, monoecious or exceptionally dioecious (Tugarinovia). 658 Leaves usually spiny, deeply pinnatisect, rarely entire. Capitula frequently subtended by pectinate leaf- 659 like bracts, homogamous, rarely heterogamous with radiate florets (Fig. 7A). Inner involucral bracts often 660 showy, radiant and coloured. Receptacle densely covered with large scales fused into a honeycombed 661 structure, absent only in Tugarinovia. Anther filaments glabrous, appendages long and sericeous. Corolla 662 and style very short. Achenes with parenchymatous pericarp, densely sericeous with twin hairs. Pappus of 663 plumose bristles, often connate into stout scales, persistent or deciduous. 664 Genera included in the subtribe: Atractylodes DC., Atractylis L., Carlina L., Thevenotia DC., and 665 Tugarinovia Iljin. 666 Geographic distribution (Fig. 8A): Atractylis, Carlina, and Thevenotia grow in Eurasia, North Africa, and 667 the Irano-Turanian region; Tugarinovia is endemic to the deserts of Mongolia and NW China; 668 Atractylodes is distributed in East Asia (China, Japan, and Korea). 669 670 Subtribe Cardopatiinae Less. 1832. 671 Spiny perennial or annual herbs. Leaves spiny-dentate or pinnatisect. Capitula few-flowered, 672 homogamous, densely (Cardopatium) or loosely (Cousiniopsis) clustered in corymbs (Fig. 7B). 673 Involucral bracts with spiny pectinate-fimbriate appendages. Anther filaments glabrous. Florets bright 674 blue below, deep blue above; petals linear. Style very shortly bilobed. Achenes with parenchymatic 675 pericarp, densely sericeous; pappus double, of two rings of short scales. 676 Genera included in the subtribe: Cardopatium Juss. and Cousiniopsis Nevski. 677 Geographic distribution (Fig. 8B): disjunct area; Cardopatium is endemic in the south of the 678 Mediterranean Region, from Algeria to Greece and Turkey, and Cousiniopsis grows in the deserts of 679 Middle Asia. 680 681 Subtribe Echinopsinae Cass. ex Dumort. 1829. 682 Perennial herbs, rarely annuals, usually very spiny. Capitula one-flowered, aggregated in spherical (rarely 683 hemispherical) secondary inflorescences (Fig. 7C). Bracts spiny. Anther filaments glabrous, basal 684 appendages short, laciniate. Achenes with parenchymatous pericarp, densely sericeous. Pappus of broad, 685 short scales directly attached to the apical plate. 686 This subtribe tribe includes only the genus Echinops L. 687 Geographic distribution (Fig. 8C): the genus grows in the Mediterranean and the Irano-Turanian regions, 688 with the main center of speciation in Iran. It is one of the genera that extends more to the south, reaching 689 tropical Africa. 690 691 Subtribe Dipterocominae Garcia-Jacas & Susanna, new subtribe. Type of the subtribe: Dipterocome 692 Fisch. & C. A. Mey., Index Seminum Horti Petropolitani 1: 26 1835. 693 Annual desert plants to 20 cm. Leaves entire, linear-spatulate. Capitula uniflowered and unisexual, 694 clustered in second-order synflorescences. Bracts of the synflorescence in several rows, lanceolate, with a 695 membranous, shortly ciliate margin. Female capitula 6–8 in the periphery of the synflorescence; 696 involucral bracts fused into a spiny-horned structure acrescent with the achene; corollas shortly bilabiate. 697 Male florets ca. 4 in the center of the synflorescence; bracts of the involucre fused ino a basal sheat; 698 corolla tubulose with very short, equal petals; anthers shortly caudate. 699 Comprises only the monotypic genus Dipterocome Fisch. & C. A. Mey. 700 Geographic distribution (Fig. 8D): Deserts in the East Mediterranean and Irano-Turanian regions: Middle 701 East (Israel, Jordan, Lebanon and Syria), Transcaucasia (Armenia, Azerbaijan and Georgia) and Iran. 702 We have interpreted the minute heads of Dipterocome pusilla as second-order inflorescences, formed by 703 female and male uniflowered capitula (Fig. 7D). The bracts of the involucre of the female florets are 704 fused into a spiny-horned structure, and the achenes are enclosed in the arcuated structure at maturity 705 (Fig. 4C). 706 707 Subtribe Xerantheminae Cass. ex Dumort. 1829. 708 Unarmed annual herbs (Chardinia, Siebera, Xeranthemum), rarely dwarf shrubs (Amphoricarpos), 709 exceptionally rhizomatous perennials (Shangwua). Leaves always entire, velvety underneath. Capitula 710 usually heterogamous (with the exception of Shangwua); central florets discoid, peripheral florets shorly 711 bilabiate, fertile, bisexuate or female. Receptacle with large scarious scales. Anther filaments glabrous, 712 appendages short, laciniate. Corolla lobes very short. Achenes often dimorphic; pappus of long tapering 713 or subulate scales, rarely reduced to a corona in Chardinia, sparsely sericeous with twin hairs (Fig. 7E). 714 Comprises five genera: Amphoricarpos Vis., Chardinia Desf., Shangwua Yu J. Wang, Raab-Straube, 715 Susanna & J. Quan Liu, Siebera J. Gay and Xeranthemum L. 716 Geographic distribution (Fig. 8E): extremely diverse and complicated. The genus Xeranthemum is a 717 widespread in waste places in Eurasia, from Middle Asia to the Iberian Peninsula, more frequently in the 718 Mediterranean region. Two other closely related genera, Chardinia and Siebera, occupy similar habitats 719 in the Irano-Turanian region, from the Tian Shan to Turkey. Amphoricarpos shows a disjunct distribution 720 in the mountains of the East Mediterranean: two species grow in the Balkans, two more in Anatolia, and 721 one in the Caucasus. Shangwua is endemic to the mountains of Middle Asia and the Himalayas, from 722 Tadjikistan to China. 723 724 Subtribe Berardiinae Garcia-Jacas & Susanna, new subtribe. Type of the subtribe: Berardia Vill., 725 Prosp. Hist. Pl. Dauphiné 27. 1779. 726 Acaulescent, unarmed perennial herb (Fig. 7F). Leaves rounded, entire or denticulate, densely woolly, 727 with veins prominent beneath, white above. Capitula solitary, sessile, homogamous. Involucral bracts 728 subulate, scarious, woolly, ending in a slender flattened point. Receptacle areolate. Florets yellowish or 729 pinkish. Staminal connective very long, apiculate. Achenes oblong, glabrous, slightly sulcate. Pericarp 730 straw-coloured. Pappus of scabrid cylindric bristles retrorsely twisted, directly attached to the apical plate. 731 Comprises only the monotypic genus Berardia Vill. 732 Geographic distribution (Fig. 8F): Berardia is a relict genus endemic to the southern Alps (SE France and 733 NW Italy). 734 735 Subtribe Staehelininae Garcia-Jacas & Susanna, new subtribe. Type of the subtribe: Staehelina L., Sp. 736 Pl. 2: 840. 1753. 737 Unarmed dwarf shrubs or subshrubs. Leaves entire or dentate-pinnatifid, linear to obovate, dark green 738 above, white-woolly beneath. Capitula corymbose or rarely solitary, homogamous. Involucral bracts 739 ovate to lanceolate, mucronate, sometimes minutely hirsute. Receptacle with wide, basally connate scales. 740 Florets whitish or pink-purple. Corolla lobes very long. Anther filaments glabrous; basal appendages very 741 long, sericeous. Achenes linear-oblong, glabrous or sericeous, with minute apical coronula. Pappus of 742 bristles basally connate into broader paleae, more or less divided apically into pinnulate fimbriae (into 743 capillary hairs in Staehelina dubia L. and S. baetica DC.), always overtopping involucre, sometimes 744 deciduous in a ring. 745 Comprises only the genus Staehelina L. (Fig. 7G). 746 Geographic distribution (Fig. 8G): stony places on limestone or rarely serpentines in the Mediterranean 747 region, from the Iberian Peninsula to the limits between Turkey and Irak (Staehelina kurdica). Most 748 diverse in the Eastern Mediterranean region (six species, in contrast with only two species in the Western 749 Mediterranean). 750 751 Subtribe Onopordinae Garcia-Jacas & Susanna, new subtribe. Type of the subtribe: Onopordum L., Sp. 752 Pl. 2: 827. 1753. 753 Robust biennial or perennial herbs, usually spiny-toothed, rarely weakly unarmed. Leaves lanceolate to 754 oblong, variously dissected. Capitula cupuliform or broadly ovoid, homogamous, discoid, most often 755 solitary or on long peduncles. Involucral bracts usually spiny, innermost sometimes scarious. 756 Receptacular bracts very often absent; receptacle foveolate, margins of foveoles with short scales, rarely 757 setose. Achenes obovoid or fusiform, surface usually transversely rugulose, wrinkled or foveolate, apical 758 plate without caruncle. Insertion areole straight or lateral-abaxial. Pappus bristles in a few rows, 759 differentiated or not, basally connate in a ring, always deciduous as a single piece (Fig. 7H). 760 Comprises seven genera: Alfredia Cass., Ancathia DC., Synurus Iljin, Syreitchikovia Pavlov, 761 Lamyropappus Knorring & Tamamsch., Olgaea Iljin, Onopordum L., and Xanthopappus C. Winkl. 762 Geographic distribution (Fig. 8H): the genus Onopordum is native in the Mediterranean and Irano- 763 Turanian regions, from the Iberian Peninsula to Middle Asia, and it has been introduced as a weed in 764 USA (California), Argentina, Chile, South Africa, and Australia. The remaining genera have narrow 765 distributions in Middle Asia, the Tian Shan, and west China, with the exception of Synurus, endemic in 766 east China, Korea, and Japan. 767 768 Subtribe Carduinae Dumort. (1827). 769 Perennial, biennial or annual spiny herbs or subshrubs, rarely unarmed. Leaves mostly lanceolate to 770 oblong, entire or variously dissected. Capitula globose or cupuliform, homogamous, discoid, very rarely 771 peripheral florets sterile and radiant (Galactites). Involucral bracts usually spiny, innermost 772 exappendiculate or with rudimentary appendages. Receptacle densely setose. Anther filaments papillose, 773 rarely glabrous. Achenes obovoid-fusiform, laterally compressed, with insertion areole straight or lateral- 774 abaxial; apical plate very often slanted, inclined adaxially, usually with caruncle. Pappus inserted on a 775 parenchymatous ring in the apical plate, with bristles in many weakly differentiated or undifferentiated 776 rows, deciduous as a single piece (Fig. 7I). 777 Comprises nine genera: Carduus L., Cirsium Mill., Cynara L., Galactites Moench, Lamyropsis 778 (Kharadze) Dittrich, Notobasis Cass., Ptilostemon Cass., Silybum Vaill., and Tyrimnus Cass. 779 Geographic distribution (Fig. 8I): Widespread in all Eurasia, especially in the Mediterranean, extending 780 southwards to Tropical Africa, introduced elsewhere as very noxious weeds. One genus, Cirsium, native 781 in America, from Canada to Guatemala. 782 783 Subtribe Saussureinae Garcia-Jacas & Susanna, new subtribe. Type of the subtribe: Saussurea DC., 784 Ann. Mus. Natl. Hist. Nat. 16: 156. 1810. 785 Unarmed perennial herbs or subshrubs, very rarely annual herbs. Leaves entire or pinnatisect, often silver- 786 white below and glabrous above, sometimes hirsute-scabrid. Capitula cylindrical or globose, often 787 paniculate, homogamous discoid. polycarpic or rarely monocarpic, Anther filaments glabrous, rarely 788 papillose. Achenes cylindrical, slightly obconical or obpyramidal, indistinctly ribbed to costate, surface 789 smooth or transversally rugose, very rarely with spines or scales, with or without a crown; apical plate 790 with a persisting style base. Pappus of very long (overtopping involucral bracts), showy, scabrid or 791 plumose, strongly differentiated bristles, the inner ones basally connate in a ring, persistent or detachable 792 as a single piece, outer bristles shorter and freely deciduous (Fig. 7J). 793 Comprises four genera: Dolomiaea DC., Jurinea Cass., Polytaxis Bunge, and Saussurea DC. 794 Geographic distribution (Fig. 8J): Native in Eurasia, Jurinea grows especially in the eastern 795 Mediterranean and West Asia with three species reaching the mountains of the Iberian peninsula and 796 Morocco; Saussurea shows its most important centre of speciation in the Himalayas and the Hengduan 797 mountains (more than 300 species); two species in Western Europe (Alps and Pyrenees), six species in 798 North America. Dolomiaea is circumscribed to the Himalayas and adjacent mountains. Polytaxis is 799 limited to the Tian Shan in Middle Asia. 800 801 Subtribe Arctiinae Garcia-Jacas & Susanna, new subtribe. Type of the subtribe: Arctium L., Sp. Pl. 2: 802 816. 1753. 803 Biennial or perennial herbs, polycarpic or rarely monocarpic, very rarely annuals, spiny or less often 804 unarmed. Leaves entire, lyrate or pinnatisect, usually with arachnoid pubescence. Capitula homogamous, 805 cylindrical or globose, discoid. Outer and middle involucral bracts spiny or hooked, inner bracts ended in 806 a scarious, unarmed or spiny appendage. Receptacle with long, twisted fimbrillae. Anther filaments 807 slightly papillose or glabrous. Achenes cylindric to narrowly obovoid, usually laterally compressed or 4- 808 costate, rarely shortly winged, usually longitudinally inconspicuously ribbed and stripped, apex glabrous 809 or coronulate; insertion arelole lateral-adaxial; apical plate truncate, without caruncle. Pappus in three 810 rows of freely deciduous undifferentiated bristles, rarely reduced (Fig. 7K). 811 Comprises two genera: Arctium L. and Cousinia Cass. 812 Geographic distribution (Fig. 8K): some species of Arctium are widespread in Eurasia, but most of the 813 species of the genus are endemic to the mountains of Middle Asia, especially the Tian Shan. Cousinia is 814 confined to the Irano-Turanian floristic region, where it has radiated explosively (600 species!). 815 816 Subtribe Centaureinae Dumort. (1827). 817 Perennial, biennial or annual unarmed herbs, shrubs or very rarely treelets, rarely spiny. Leaves oblong to 818 linear, entire or variously dissected. Capitula ovoid, cupuliform or cylindric, often heterogamous with 819 sterile radiant florets, less often homogamous, discoid. Involucral bracts usually ended in a diversely 820 scarious, fimbriate, pectinate, spiny or unarmed appendage; innermost bracts always with a scarious 821 appendage. Receptacle setaceous. Anther filaments glabrous or papillose. Achenes obovoid, laterally 822 compressed, with sclerified pericarp, usually glabrous and smooth, less often hirsute, rugose, pitted, or 823 ridged; insertion areole concave, lateral-adaxial, very rarely (Crupina) straight, often with an elaiosome; 824 apical plate straight, without caruncle. Pappus inserted on a parenchymatous ring in the apical plate, 825 biseriate, outer series formed by several rows of long, differently pinnulate bristles or rarely scales, 826 basally connate in a ring or free, deciduos as a whole of separately, often persistent, inner series of some 827 short or longer bristles, sometimes reduced; rarely pappus single or missing by abortion (Fig. 7L). 828 Comprises 32 genera: Amberboa Vaill., Callicephalus C. A. Mey. Carduncellus Adans., Carthamus L., 829 Centaurea1 L., Centaurodendron Johow, Centaurothamnus Wagenitz & Dittrich, Cheirolophus Cass., 830 Crocodilium Vaill., Crupina (Pers.) DC., Femeniasia Susanna, Goniocaulon Cass., Karvandarina Rech. 831 f., Klasea Cass., Mantisalca Cass., Myopordon Boiss., Ochrocephala Dittrich, Oligochaeta (DC.) K.

1 Including Cyanus Mill., a genus that is sometimes segregated from Centaurea. Our analysis (Fig. 1) confirms that it should be considered a subgenus of Centaurea. 832 Koch, Phalacrachena Iljin, Phonus Hill, Plagiobasis Schrenk, Plectocephalus D. Don, Psephellus Cass., 833 Rhaponticoides Vaill., Rhaponticum Vaill., Russowia C. Winkl., Schischkinia Iljin, Serratula L., 834 Stizolophus Cass., Tricholepis DC., Volutaria Cass., and Zoegea L. 835 Geographic distribution (Fig. 8L): mainly Mediterranean and Irano-Turanian regions, with some species 836 of Centaurea extended widely in north Eurasia as far as Scotland, and the easternmost representatives 837 reaching Middle Asia; exceptionally, Tricholepis has gone beyond the mountains of Middle Asia and 838 reaches Burma; one species of Rhaponticum grows in Australia. In the south of the area, a few species 839 grow in subtropical and tropical Africa, from Senegal to Somalia and southwards to Zimbabwe. 840 Plectocephalus has a disjunct distribution in Ethiopia, U.S.A., and South America (Brazil, Argentina and 841 Chile). Some relict genera grow far away from the center of the distribution of the subtribe, namely 842 Centaurodendron in the Juan Fernández archipelago, Centaurothamnus in Yemen, Goniocaulon in India, 843 and Ochrocephala in Etiopia and Sudan; all of them but Centaurodendron are monotypic. Some noxious 844 weeds are widespread in all the globe, especially in regions with a Mediterranean-type climate (Australia, 845 California, Chile, and South Africa): Carthamus lanatus, Centaurea diffusa, C. solstitialis, C. stoebe, 846 Rhaponticum repens, and Volutaria tubuliflora. 847 848 Key to subtribes of tribe Cardueae 849 850 1. Achenes enclosed in an arcuate, spiny-toothed diaspore; heads 851 minute, unisexual, clustered in the axilles of leaf verticils Dipterocominae 852 – Achenes free, not enclosed in a spiny-toothed diaspore; heads 853 bisexual, not clustered in the axilles of leaf verticils 2 854 855 2. Capitula one-flowered, clustered in second-order spherical 856 or hemispherical compound synflorescences Echinopsinae 857 – Capitula not one-flowered, sometimes clustered in flat-topped 858 corymbs but never forming second-order heads 3 859 860 3. Acaulescent perennials with gray-green, entire leaves; 861 achenes with the pappus bristles retrorsely twisted Berardiinae 862 – Caulescent or acaulescent annuals or perennials; achenes 863 with pappus of straight bristles 4 864 865 4. Leaves always entire. Achenes with pappus of 5-15 866 subulate, dentate or plumose scales, rarely reduced to a short corona Xerantheminae 867 – Leaves entire or divided. Achenes with pappus of many bristles 5 868 869 5. Capitula usually subtended by pectinate-pinnatisect leaf-like 870 bracts. Corolla lobes 1–3 mm long. Pappus of long, plumose 871 bristles usually connate at the base into broader, robust scales Carlininae 872 – Capitula subtended by entire or dentate bracts or on leafless 873 pedicels. Corolla lobes longer than 3 mm. Pappus of free bristles 874 or scales not connate at the base into broader scales 6 875 876 6. Capitula with 8-12 florets, rarely many-flowered. Pappus 877 of two equal rows of short lanceolate scales. Corollas bright blue 878 below, deep blue above, with linear lobes Cardopatiinae 879 Capitula many-flowered. Pappus of one or several rows of 880 pinnulate or plumose long scales or bristles. Corollas not 881 bright blue below and deep blue above; lobes not linear 7 882 883 7. Achenes rugulose, pitted or faintly velvety, never smooth, without apical 884 caruncle or basal elaiosome 8 885 – Achenes usually smooth or rarely ridged, either with apical caruncle 886 or basal elaiosome 9 887 888 8. Receptacle fimbrillate with long, twisted fimbrillae. Achenes longitudinally 889 stripped; pappus bristles individually deciduous Arctiinae 890 – Receptacle usually naked and foveolate, rarely with straight bristles. 891 Achenes transversally rugulose; pappus bristles basally connate 892 in a ring, detachable as a single piece Onopordinae 893 894 9. Plants unarmed. Leaves always white-woolly beneath. Pappus 895 very long, often overtopping invloucral bracts at anthesis 10 896 – Plants unarmed or spiny. Leaves woolly or glabrous below. Pappus 897 not overtopping involucral bracts 11 898 899 10. Shrubs or subshrubs. Corolla lobes very long, narrowly triangular. Style 900 lobes very short, straight. Achenes linear, faintly sulcate; pappus in one row Staehelininae 901 – Perennial herbs, rarely annuals. Corolla lobes short, broadly 902 triangular. Style lobes long, reflexed. Achenes cylindrical or slightly obconical or 903 obpyramidal, smooth, echinate or rugulose; pappus in several rows Saussureinae 904 905 11. Insertion areole of the achenes usually straight, basal or basal-adaxial. 906 Apical plate of the achenes slanted, with elaiosome (except. Notobasis). 907 Pappus undifferentiated, deciduous as a whole Carduinae 908 – Insertion areole of the achenes usually concave, lateral-abaxial, often with 909 an elaiosome. Apical plate straight, without caruncle. Pappus usually 910 differentiated in two rows, persistent or rarely deciduous as a single piece, 911 rarely absent by abortion Centaureinae 912 913 914 915 Acknowledgments 916 917 Authors thank the herbaria that provided material for the study: BC, DUSH, E, ERE, FRU, GDA, H, LE, 918 MJG, TK, W, and A. Pavlenko for kindly supplying the photograph of Dipterocome pusilla. Financial 919 support from the Ministerio de Ciencia e Innovación (Project CGL2015-66703-P and Ph.D. grant to Sonia 920 Herrando-Moraira) and the Catalan government (“Ajuts a grups consolidats” 2017-SGR1116) is 921 gratefully acknowledged. 922 923 924 Supplementary material and Appendices 925 926 All alignments and tree files for each dataset are deposited in Mendeley Data repository 927 (http://dx.doi.org/xxxxx). Supplementary data associated with this article can be found, in the online 928 version, at https://doi.org/ xxxxx 929 930 931 References 932 933 Anderberg, A., Ghahremaninejad, F., Källersjö, M., 2007. The enigmatic genus Dipterocome. 934 Compositae Newslett. 45, 23–36. 935 Bankevich, A., Nurk, S., Antipov, D., Gurevich, A.A., Dvorkin, M., Kulikov, A.S., Lesin V.M., 936 Nikolenko, S.I., Pham, S., Prjibelski A.D., Pyshkin, A.V., Sirotkin A.V., Vyahhi, N., Tesler, G., 937 Alekseyev, M.A., Pyshkin, A.V., 2012. SPAdes: a new genome assembly algorithm and its 938 applications to single-cell sequencing. J. Comput. Biol. 19, 455–477. 939 https://doi.org/10.1089/cmb.2012.0021 940 Barreda, V.D., Palazzesi, L., Katinas, L., Crisci, J.V., Tellería, M.C., Bremer, K., Passala, M.G., 941 Bechis, F., Corsolini, R., 2012. An extinct Eocene taxon of the daisy family (Asteraceae): 942 evolutionary, ecological and biogeographical implications. Ann. Bot., London, 109, 127–134. 943 https://doi.org/10.1093/aob/mcr240 944 Barres, L., Sanmartín, I., Anderson, C.L., Susanna, A., Buerki, S., Galbany-Casals, M., 945 Vilatersana, R., 2013. Reconstructing the evolution and biogeographic history of tribe Cardueae 946 (Compositae). Am. J. Bot. 100, 867–882. https://doi.org/10.3732/ajb.1200058 947 Bentham, G., 1873. Compositae. In: Bentham, G., Hooker, J.D. (Eds.), Genera plantarum, vol. 2. 948 Spottiswoode & Co., London, UK, pp. 163–533. 949 Boissier, P.E., 1849. Diagnoses plantarum orientalium novarum, series 1, vol. 11. Ducloux, Paris, 950 France. 951 Bolger, A.M., Lohse, M., Usadel, B., 2014. Trimmomatic: A flexible trimmer for Illumina sequence 952 data. Bioinformatics 30, 2114–2120. https://doi.org/10.1093/bioinformatics/btu170 953 Borowiec, M.L., 2016. AMAS: a fast tool for alignment manipulation and computing of summary 954 statistics. PeerJ 4, e1660. https://doi.org/10.7717/peerj.1660 955 Bremer, K., 1987. Tribal interrelationships of the Asteraceae. Cladistics 3, 210–253. 956 https://doi.org/10.1111/j.1096-0031.1987.tb00509.x 957 Bremer, K., 1994. Compositae: Cladistics and classification. Timber Press, Portland, Oregon, USA. 958 Bruun-Lund, S., Clement, W. L., Kjellberg, F., Rønsted, N., 2017. First plastid phylogenomic study 959 reveals potential cyto-nuclear discordance in the evolutionary history of Ficus L. (Moraceae). Mol. 960 Phylogenet. Evol. 109, 93–104. https://doi.org/10.1016/j.ympev.2016.12.031 961 Candolle, A.P. de., 1838. Prodromus systematis naturalis regni vegetabilis, vol. 7. Treuttel & Würtz, 962 Paris, France. 963 Capella-Gutiérrez, S., Silla-Martínez, J.M., Gabaldón, T., 2009. trimAl: a tool for automated 964 alignment trimming in large-scale phylogenetic analyses. Bioinformatics 25, 1972–1973. 965 https://doi.org/10.1093/bioinformatics/btp348 966 Carlsen, M.M., Fér, T., Schmickl, R., Leong-Škorničková, J., Newman, M., Kress, W.J., 2018. 967 Resolving the rapid plant radiation of early diverging lineages in the tropical Zingiberales: pushing 968 the limits of genomic data. Mol. Phylogenet. Evol. 128, 55–68. 969 https://doi.org/10.1016/j.ympev.2018.07.020 970 Cassini, H., 1819. [Different articles]. Dictionnaire de Sciences Naturelles. Paris. In: King, R., Dawson, 971 H.W. (Eds., 1975), Cassini on Compositae. Oriole, New York, New York, USA. 972 Castresana, J., 2000. Selection of conserved blocks from multiple alignments for their use in 973 phylogenetic analysis. Mol. Biol. Evol. 17, 540–552. 974 https://doi.org/10.1093/oxfordjournals.molbev.a026334 975 Chevreux, B., Wetter, T., Suhai, S., 1999. Genome sequence assembly using trace signals and 976 additional sequence information. In: Computer Science and Biology: Proceedings of the German 977 Conference on Bioinformatics (GCB), vol. 99, pp. 45–56. 978 Dayrat, B., 2005. Towards integrative taxonomy. Biol. J. Linn. Soc. 85, 407–415. 979 https://doi.org/10.1111/j.1095-8312.2005.00503.x 980 Dittrich, M., 1977. Cynareae–systematic review. In: Heywood, V.H., Harborne, J.B., Turner, B.L. (Eds.), 981 The Biology and Chemistry of the Compositae. Academic Press, London–New York–San Francisco, 982 UK and USA, pp. 999–1015. 983 Dittrich, M., 1996a. Bemerkungen zur Tribuszugehörigkeit von Berardia. Ann. Naturhist. Mus. Wien 99 984 (B, suppl.), pp. 329–342. 985 Dittrich, M., 1996b. Die Bedeutung morphologischer and anatomischer Achänen―Merkmale für die 986 Systematik der Tribus Echinopeae Cass. and Carlineae Cass. Boissiera 51: 9–102. 987 Drummond, A.J., Suchard, M.A., Xie, D., Rambaut, A., 2012. Bayesian phylogenetics with BEAUti 988 and the BEAST 1.7. Mol. Biol. Evol. 29, 1969–1973. https://doi.org/10.1093/molbev/mss075 989 Edwards, S.V., Xi, Z., Janke, A., Faircloth, B.C., McCormack, J.E., Glenn, T.C., Zhong, B., Wu, S., 990 Lemmon, E.M., Lemmon, A.R., Leache, A.D., Liu, L., Davis, C.C., 2016. Implementing and 991 testing the multispecies coalescent model: A valuable paradigm for phylogenomics. Mol. Phylogenet. 992 Evol. 94, 447–462. https://doi.org/10.1016/j.ympev.2015.10.027 993 Faircloth, B.C., 2015. PHYLUCE is a software package for the analysis of conserved genomic loci. 994 Bioinformatics 32, 786–788. https://doi.org/10.1093/bioinformatics/btv646. 995 Fischer, F.E.L. von, Meyer, C.A. von., 1835. Index Seminum Hortus Petropolitanus 1. St. Petersburg, 996 Russia. 997 Folk, R.A., Mandel, J.R., Freudenstein, J.V., 2015. A protocol for targeted enrichment of intron‐ 998 containing sequence markers for recent radiations: A phylogenomic example from Heuchera 999 (Saxifragaceae). Appl. Plant Sci. 3, 1500039. https://doi.org/10.3732/apps.1500039 1000 Folk, R.A., Mandel, J.R., Freudenstein, J.V., 2017. Ancestral gene flow and parallel organellar genome 1001 capture result in extreme phylogenomic discord in a lineage of angiosperms. Syst. Biol. 66, 320–337. 1002 https://doi.org/10.1093/sysbio/syw083 1003 Funk, V. A., Susanna, A., Stuessy, T.F., Robinson, H., 2009. Classification of Compositae. In: Funk, 1004 V.A., Susanna, A., Stuessy, T.F., Bayer, R.J. (Eds.), 2009. Systematics, evolution, and biogeography 1005 of Compositae. International Association for Plant Taxonomy, Vienna, pp. 171–189. 1006 García, N., Folk, R.A., Meerow, A.W., Chamala, S., Gitzendanner, M.A., de Oliveira, R.S., Soltis, 1007 D.E., Soltis, P.S., 2017. Deep reticulation and incomplete lineage sorting obscure the diploid 1008 phylogeny of rain-lilies and allies (Amaryllidaceae tribe Hippeastreae). Mol. Phylogenet. Evol. 111, 1009 231–247. https://doi.org/10.1016/j.ympev.2017.04.003 1010 Garcia-Jacas, N., Garnatje, T., Susanna, A., Vilatersana, R., 2002. Tribal and subtribal delimitation 1011 and phylogeny of the Cardueae (Asteraceae): a combined nuclear and chloroplast DNA analysis. 1012 Mol. Phylogenet. Evol. 22, 51– 64. https://doi.org/10.1006/mpev.2001.1038 1013 Gardner, A.G., Sessa, E.B., Michener, P., Johnson, E., Shepherd, K.A., Howarth, D.G., Jabaily, 1014 R.S., 2016. Utilizing next-generation sequencing to resolve the backbone of the Core Goodeniaceae 1015 and inform future taxonomic and floral form studies. Molec. Phylogenet. Evol. 94, 605–617. 1016 https://doi.org/10.1016/j.ympev.2015.10.003 1017 Gouy, M., Guindon, S., Gascuel, O., 2009. SeaView version 4: a multiplatform graphical user interface 1018 for sequence alignment and phylogenetic tree building. Mol. Biol. Evol. 27, 221–224. 1019 https://doi.org/10.1093/molbev/msp259 1020 Häffner, E., Hellwig, F.H., 1999. Phylogeny of the tribe Cardueae (Compositae) with emphasis on the 1021 subtribe Carduinae: an analysis based on ITS sequence data. Willdenowia 29, 27–39. 1022 https://doi.org/10.3372/wi.29.2902 1023 Hahn, C., Bachmann, L., Chevreux, B., 2013. Reconstructing mitochondrial genomes directly from 1024 genomic next-generation sequencing reads—a baiting and iterative mapping approach. Nucleic Acids 1025 Res. 41, e129. https://doi.org/10.1093/nar/gkt371 1026 Harris, R.S., 2007. Improved pairwise alignment of genomic DNA. PhD Thesis, Pennsylvania State 1027 University, Pennsylvania. 1028 Harrison, N., Kidner, C.A., 2011. Next-generation sequencing and systematics: what can a billion base 1029 pairs of DNA sequence data do for you? Taxon 60, 1552–1566. 1030 Heckenhauer, J., Samuel, R., Ashton, P.S., Salim, K.A., Paun, O., 2018. Phylogenomics resolves 1031 evolutionary relationships and provides insights into floral evolution in the tribe Shoreeae 1032 (Dipterocarpaceae). Mol. Phylogenet. Evol. 127, 1–13. https://doi.org/10.1016/j.ympev.2018.05.010 1033 Heilbuth, J.C., 2000. Lower species richness in dioecious clades. Am. Nat. 156, 221–41. 1034 https://doi.org/10.1086/303389 1035 Herrando-Moraira, S., The Cardueae Radiations Group (Calleja, J.A., Carnicero-Campmany, 1036 Fujikawa, K., Galbany-Casals, M., Garcia-Jacas, N., Im, H.-T., Kim, S.-C., Liu, J.-Q., López- 1037 Alvarado, J., López-Pujol, J., Mandel, J.R., Massó, S., Mehregan, I., Montes-Moreno, N., Pyak, 1038 E., Roquet, C., Sáez, L., Sennikov, A., Susanna, A., Vilatersana, R.), 2018. Exploring data 1039 processing strategies in NGS target enrichment to disentangle radiations in the tribe Cardueae 1040 (Compositae). Mol. Phylogenet. Evol. 128, 69–87. https://doi.org/10.1016/j.ympev.2018.07.012 1041 Hillis, D.M., Bull, J.J., 1993. An empirical test of bootstrapping as a method for assessing confidence in 1042 phylogenetic analysis. Syst. Biol. 42, 182–192. https://doi.org/10.1093/sysbio/42.2.182 1043 Hoffmann, O., 1894. Compositae. In: Engler, A., Prantl, K. (Eds.), Die natürlichen Pflanzenfamilien, 1044 vol. 5, part 4. Wilhelm Engelmann, Leipzig, Germany, pp. 24–333. 1045 Huang, C.H., Zhang, C., Liu, M., Hu, Y., Gao, T., Qi, J., Ma, H., 2016. Multiple polyploidization 1046 events across Asteraceae with two nested events in the early history revealed by nuclear 1047 phylogenomics. Mol. Biol. Evol. 33, 2820–2835. https://doi.org/10.1093/molbev/msw157 1048 Ivanov, D., Ashraf, A.R., Utescher, T., Mosbrugger, V., Slavomirova, E., 2007. Late Miocene 1049 vegetation and climate of the Balkan region, palynology of the Beli Breg Coal Basin sediments. Geol. 1050 Carpath. 58, 367–381. 1051 Jansen, R.K., Holsinger, K.E., Michaels, H.J., Palmer, J.D., 1990. Phylogenetic analysis of chloroplast 1052 DNA restriction site data at higher taxonomic levels: an example from the Asteraceae. Evolution 44, 1053 2089–2105. https://doi.org/10.1111/j.1558-5646.1990.tb04314.x 1054 Jansen, R.K., Michaels, H.J., Palmer, J.D., 1991. Phylogeny and character evolution in the Asteraceae 1055 based on chloroplast DNA restriction site mapping. Syst. Bot. 16, 98–115. 1056 https://doi.org/10.2307/2418976 1057 Jaubert, H.F., Spach, E., 1850. Illustrationes plantarum orientalium, vol. 3. Roret, Paris, France. 1058 Johnson, M.G., Gardner, E.M., Liu, Y., Medina, R., Goffinet, B., Shaw, A.J., Zerega, N.J.C., 1059 Wickett, N.J., 2016. HybPiper: Extracting coding sequence and introns for phylogenetics from high- 1060 throughput sequencing reads using target enrichment. Appl. Plant Sci. 4, 1600016. 1061 https://doi.org/10.3732/apps.1600016. 1062 Jones, K.E., Fér, T., Schmickl, R., Dikow, R.B., Funk, V.J., Herrando-Moraira, S., Johnston, P.R., 1063 Siniscalchi, C.M., Susanna, A., Slovák, M., Thapa, R., Watson, L., Mandel, J.R., in preparation. 1064 An empirical assessment of a single family-wide target capture locus set at multiple evolutionary 1065 timescales within the largest family of flowering plants. 1066 Karis, P.O., Källersjö, M., Bremer, K., 1992. Phylogenetic analysis of the Cichorioideae (Asteraceae), 1067 with emphasis on the Mutisieae. Ann. Mo. Bot. Gard. 79, 416–427. https://doi.org/10.2307/2399778 1068 Katoh, K., Standley, D.M., 2013. MAFFT multiple sequence alignment software version 7: 1069 improvements in performance and usability. Mol. Biol. Evol. 30, 772–780. 1070 https://doi.org/10.1093/molbev/mst010 1071 Kay, K.M., Voelckel, C., Yang, J.Y., Hufford, K.M., Kaska, D.D., Hodges, S.A., 2006. Floral 1072 characters and species diversification. In: Harder, L.D., Barrett, S.C.H. (Eds.), Ecology and evolution 1073 of flowers. Oxford University Press, Oxford, UK, pp. 311–325. 1074 Kim, K.-J., Choi, K.-S., Jansen, R.K., 2005. Two chloroplast DNA inversions originated 1075 simultaneously during the early evolution of the sunflower family (Asteraceae). Mol. Biol. Ecol. 22, 1076 1783–1792. https://doi.org/10.1093/molbev/msi174 1077 Kim, K.J., Jansen, R.K., Wallace, R.S., Michaels, H.J., Palmer, J.D., 1992. Phylogenetic implications 1078 of rbcL sequence variation in the Asteraceae. Ann. Mo. Bot. Gard. 428–445. 1079 https://doi.org/10.2307/2399779 1080 Knowles, L.L., Huang, H., Sukumaran, J., Smith, S.A., 2018. A matter of phylogenetic scale: 1081 Distinguishing incomplete lineage sorting from lateral gene transfer as the cause of gene tree discord 1082 in recent versus deep diversification histories. Am. J. Bot. 105, 376–384. 1083 https://doi.org/10.1002/ajb2.1064 1084 Kubatko, L.S., Degnan, J.H., 2007. Inconsistency of phylogenetic estimates from concatenated data 1085 under coalescence. Syst. Biol. 56, 17–24. https://doi.org/10.1080/10635150601146041. 1086 Kück, P., Longo, G.C., 2014. FASconCAT-G: extensive functions for multiple sequence alignment 1087 preparations concerning phylogenetic studies. Front. Zool. 11, 81. https://doi.org/10.1186/s12983- 1088 014-0081-x 1089 Lagomarsino, L.P., Condamine, F.L., Antonelli, A., Mulch, A., Davis, C.C., 2016. The abiotic and 1090 biotic drivers of rapid diversification in Andean bellflowers (Campanulaceae). New Phytol. 210, 1091 1430–1442. https://doi.org/10.1111/nph.13920 1092 Lane, A.K., Augustin, M.M., Ayyampalayam, S., Plant, A., Gleissberg, S., Di Stilio, V.S., 1093 Depamphilis C.W., Wong, G.K.-S., Kutchan, T.M., Leebens-Mack, J.H., 2018. Phylogenomic 1094 analysis of Ranunculales resolves branching events across the order. Bot. J. Linn. Soc. 187, 157–166. 1095 https://doi.org/10.1093/botlinnean/boy015 1096 Lemmon, A.R., Brown, J.M., Stanger-Hall, K., Lemmon, E.M., 2009. The effect of ambiguous data 1097 on phylogenetic estimates obtained by maximum likelihood and Bayesian inference. Syst. Biol. 58, 1098 130–145. https://doi.org/10.1093/sysbio/syp017 1099 Lemmon, E.M., Lemmon, A.R., 2013. High-throughput genomic data in systematics and phylogenetics. 1100 Ann. Rev. Ecol. Evol. Syst. 44, 99–121. https://doi.org/10.1146/annurev-ecolsys-110512-135822. 1101 Linder, H.P., Hardy, C.R., Rutschmann, F., 2005. Taxon sampling effects in molecular clock dating: 1102 an example from the African Restionaceae. Mol. Phylogenet. Evol. 35, 569–582. 1103 https://doi.org/10.1016/j.ympev.2004.12.006 1104 Liu, L., Xi, Z., Davis, C.C., 2015. Coalescent methods are robust to the simultaneous effects of long 1105 branches and incomplete lineage sorting. Mol. Biol. Evol. 32, 791–805. 1106 https://doi.org/10.1093/molbev/msu331 1107 López-Vinyallonga, S., Mehregan, I., Garcia-Jacas, N., Tscherneva, O., Susanna, A., Kadereit, J.K., 1108 2009. Phylogeny and evolution of the Arctium-Cousinia complex (Compositae, Cardueae- 1109 Carduinae). Taxon 58, 153–171. 1110 Mai, D.H., 1995. Tertiäre Vegetationsgeschichte Europas: Methoden und Ergebnisse. Gustav Fischer 1111 Verlag, Jena, Germany. 1112 Mai, D.H., 2001. Die mittelmiozänen und obermiozänen Floren aus der Meuroer und Raunoer Folge in 1113 der Lausitz, Teil II, Dicotyledonen. Palaeontogr. Abt. B 257, 35–174. 1114 Mandel, J.R., Barker, M.S., Bayer, R.J., Dikow, R.B., Gao, T., Jones, K.E., Keely, S., Kilian, N., 1115 Ma, H., Siniscalchi, C.M., Susanna, A., Thapa, R., Watson, L., Funk, V.A., 2017. The 1116 Compositae Tree of Life in the age of phylogenomics. J. System. Evol. 55, 405–410. 1117 https://doi.org/10.1111/jse.12265. 1118 Mandel, J.R., Dikow, R.B., Funk, V.A., 2015. Using phylogenomics to resolve mega-families: an 1119 example from Compositae. J. System. Evol. 53, 391–402. https://doi.org/10.1111/jse.12167. 1120 Mandel, J.R., Dikow, R.B., Funk, V.A., Masalia, R.R., Staton, S.E., Kozik, A., Michelmore, R.W., 1121 Rieseberg, L.H., Burke, J.M., 2014. A target enrichment method for gathering phylogenetic 1122 information from hundreds of loci: an example from the Compositae. Appl. Plant Sci. 2, 1300085. 1123 https://doi.org/10.3732/apps.1300085. 1124 Miller, M.A., Pfeiffer, W., Schwartz, T., 2010. Creating the CIPRES Science Gateway for inference of 1125 large phylogenetic trees. In: Proceedings of the Gateway Computing Environments Workshop 1126 (GCE), 14 Nov. 2010. New Orleans, pp. 1–8. https://doi.org/10.1109/GCE.2010.5676129. 1127 Mitchell, N., Lewis, P.O., Lemmon, E.M., Lemmon, A.R., Holsinger, K.E., 2017. Anchored 1128 phylogenomics improves the resolution of evolutionary relationships in the rapid radiation of Protea 1129 L. Am. J. Bot. 104, 102–115. https://doi.org/10.3732/ajb.1600227 1130 Morales-Briones, D.F., Liston, A., Tank, D.C., 2018. Phylogenomic analyses reveal a deep history of 1131 hybridization and polyploidy in the Neotropical genus Lachemilla (Rosaceae). New Phytol. 218, 1132 1668–1684. https://doi.org/10.1111/nph.15099 1133 Nicholls, J.A., Pennington, R.T., Koenen, E.J., Hughes, C.E., Hearn, J., Bunnefeld, L., Dexter, K.G., 1134 Stone, G.N., Kidner, C.A., 2015. Using targeted enrichment of nuclear genes to increase 1135 phylogenetic resolution in the neotropical rain forest genus Inga (Leguminosae: Mimosoideae). 1136 Front. Plant Sci. 6, 710. https://doi.org/10.3389/fpls.2015.00710. 1137 Ortiz, S., Bonifacino, J.M., Crisci, J.V., Funk, V.A., Hansen, H.V., Hind, D.J.N., Katinas, L., Roque, 1138 N., Sancho, G., Susanna, A., Tellería, M.C., 2009. The basal grade of the Compositae: Mutisieae 1139 (sensu Cabrera) and Carduoideae. In: Funk, V.A., Susanna, A., Stuessy, T.F., Bayer, R.J. (Eds.), 1140 Systematics, evolution, and biogeography of Compositae. International Association for Plant Taxonomy, 1141 Vienna, Austria, pp. 193–213. 1142 Panero, J.L., Crozier, B.S., 2016. Macroevolutionary dynamics in the early diversification of 1143 Asteraceae. Mol. Phylogenet. Evol. 99, 116–132. https://doi.org/10.1016/j.ympev.2016.03.007 1144 Panero, J.L., Funk, V.A., 2008. The value of sampling anomalous taxa in phylogenetic studies: Major 1145 clades of the Asteraceae revealed. Mol. Phylogenet. Evol. 47, 757–782. 1146 Parks, M., Cronn, R., Liston, A., 2009. Increasing phylogenetic resolution at low taxonomic levels 1147 using massively parallel sequencing of chloroplast genomes. BMC Biol. 7, 84. 1148 https://doi.org/10.1186/1741-7007-7-84 1149 Petit, D.P., 1988. Le genre Echinops L. (Compositae, Cardueae). 1. Position phylétique et interprétation 1150 de l’incapitulescence. Candollea 43, 467–481. 1151 Petit, D.P., 1997. Generic interrelationships of the Cardueae (Compositae): a cladistic analysis of 1152 morphological data. Plant Syst. Evol. 207, 173–203. https://doi.org/10.1007/BF00984388 1153 Petit, D.P., Mathez, J., Qaid, A., 1996. Early differentiation of the Cardueae sensu lato: Morphology 1154 and pollen. In: Hind, D.J.N., Beentje, H.J. (Eds.), Compositae: Systematics. Proceedings of the 1155 International Compositae Conference, Kew 1994, pp. 79–93. 1156 Popescu, S.M., 2002. Repetitive changes in Early Pliocene vegetation revealed by high-resolution pollen 1157 analysis: Revised cyclostratigraphy of southwestern Romania. Rev. Palaeobot. Palynol. 120, 181– 1158 202. https://doi.org/10.1016/S0034-6667(01)00142-7 1159 Rambaut, A. 2016. FigTree ver. 1.4.3. Department of Zoology, University of Oxford, Oxford. Available 1160 at http://tree.bio.ed.ac.uk/software/figtree/ 1161 Ranwez, V., Harispe, S., Delsuc, F., Douzery, E.J., 2011. MACSE: Multiple Alignment of Coding 1162 SEquences accounting for frameshifts and stop codons. PloS One 6: e22594. 1163 https://doi.org/10.1371/journal.pone.0022594 1164 Renner, S.S., Ricklefs, R.E., 1995. Dioecy and its correlates in the flowering plants. Am. J. Bot. 82, 1165 596–606. https://doi.org/10.1002/j.1537-2197.1995.tb11504.x 1166 Sanderson, M.J., 2002. Estimating absolute rates of molecular evolution and divergence times: a 1167 penalized likelihood approach. Mol. Biol. Evol. 19, 101–109. 1168 https://doi.org/10.1093/oxfordjournals.molbev.a003974 1169 Sayyari, E., Mirarab, S., 2016. Fast coalescent-based computation of local branch support from quartet 1170 frequencies. Mol. Biol. Evol. 33, 1654–1668. https://doi.org/10.1093/molbev/msw079 1171 Schlick-Steiner, B.C., Steiner, F.M., Seifert, B., Stauffer, C., Christian, E., Crozier, R.H., 2010. 1172 Integrative taxonomy: A multisource approach to exploring biodiversity. Ann. Rev. Entomol. 55, 1173 421–438. https://doi.org/10.1146/annurev-ento-112408-085432 1174 Shaw, T.I., Ruan, Z., Glenn, T.C., Liu, L., 2013. STRAW: Species TRee Analysis Web server. Nucleic 1175 Acids Res. 41 (Web Server issue), W238–W241. 1176 Small, R.L., Cronn, R.C., Wendel, J.F., 2004. Use of nuclear genes for phylogeny reconstruction in 1177 plants. Aust. Syst. Bot. 17: 145–170. 1178 Small, R.L., Ryburn, J.A., Cronn, R.C., Seelanan, T., Wendel, J.F., 1998. The tortoise and the hare: 1179 choosing between noncoding plastome and nuclear Adh sequences for phylogeny reconstruction in a 1180 recently diverged plant group. Am. J. Bot. 85, 1301–1315. 1181 Smith, S.A., Moore, M.J., Brown, J.W., Yang, Y., 2015. Analysis of phylogenomic datasets reveals 1182 conflict, concordance, and gene duplications with examples from animals and plants. BMC Evol. 1183 Biol. 15, 150. https://doi.org/10.1186/s12862-015-0423-0 1184 Smith, S.A., O’Meara, B.C., 2012. treePL: divergence time estimation using penalized likelihood for 1185 large phylogenies. Bioinformatics 28, 2689–2690. https://doi.org/10.1093/bioinformatics/bts492 1186 Stamatakis, A., 2006. RAxML-VI-HPC: maximum likelihood-based phylogenetic analyses with 1187 thousands of taxa and mixed models. Bioinformatics 22, 2688–2690. 1188 https://doi.org/10.1093/bioinformatics/btl446. 1189 Stamatakis, A., 2014. RAxML version 8: a tool for phylogenetic analysis and post-analysis of large 1190 phylogenies. Bioinformatics 30, 1312–1313. https://doi.org/10.1093/bioinformatics/btu033. 1191 Stoughton, T.R., Kriebel, R., Jolles, D.D., O'Quinn, R.L., 2018. Next‐generation lineage discovery: A 1192 case study of tuberous Claytonia L. Am. J. Bot. 105, 536–548. https://doi.org/10.1002/ajb2.1061 1193 Stubbs, R.L., Folk, R.A., Xiang, C.L., Soltis, D.E., Cellinese, N., 2018. Pseudo-parallel patterns of 1194 disjunctions in an Arctic-alpine plant lineage. Mol. Phylogenet Evol. 123, 88–100. 1195 https://doi.org/10.1016/j.ympev.2018.02.016 1196 Susanna, A., 1987. Femeniasia, novus genus Carduearum. Collectanea Botanica 17, 83–88. 1197 https://doi.org/10.3989/collectbot.1988.v17.158 1198 Susanna, A., Garcia-Jacas, N., 2009. Cardueae (Carduoideae). In: Funk, V.A., Susanna, A., Stuessy, 1199 T.F., Bayer, R.J. (Eds.), Systematics, evolution, and biogeography of Compositae. IAPT, Vienna, 1200 Austria. pp. 293–313. 1201 Susanna, A., Garcia-Jacas, N., Hidalgo, O., Vilatersana, R. Garnatje, T., 2006. The Cardueae 1202 (Compositae) revisited: Insights from ITS, trnL-trnF, and matK nuclear and chloroplast DNA 1203 analysis. Ann. Mo. Bot. Gard. 93, 150–171. https://doi.org/10.3417/0026- 1204 6493(2006)93[150:TCCRIF]2.0.CO;2 1205 Susanna, A., Garcia-Jacas, N., Soltis, D.E., Soltis, P.S., 1995. Phylogenetic relationships in tribe 1206 Cardueae (Asteraceae) based on ITS sequences. Am. J. Bot. 82, 1056–1068. 1207 https://doi.org/10.1002/j.1537-2197.1995.tb11571.x 1208 Susanna, A., Garcia-Jacas. N., 2007. Tribe Cardueae. In: Kadereit, J.W., Jeffrey, C. (Eds.), Flowering 1209 Plants. Eudicots. Asterales, vol. 8. In: Kubitzki, K. (Ed.), The families and genera of vascular plants. 1210 Springer Verlag, Berlin, Germany, pp. 123–146. 1211 Susanna, A., Garnatje, T., Garcia-Jacas, N., Vilatersana, R., 2002. On the correct subtribal placement 1212 of the genera Syreitschikovia and Nikitinia (Asteraceae, Cardueae): Carduinae or Centaureinae? Bot. 1213 J. Linn. Soc. 140, 313–319. https://doi.org/10.1046/j.1095-8339.2002.00104.x 1214 Tillich, M., Lehwark, P., Pellizzer, T., Ulbricht-Jones, E. S., Fischer, A., Bock, R., Greiner, S., 2017. 1215 GeSeq–versatile and accurate annotation of organelle genomes. Nucleic Acids Res. 45(W1), W6– 1216 W11. https://doi.org/10.1093/nar/gkx391 1217 Vamosi, J.C., Otto, S.P., Barrett, S.C., 2003. Phylogenetic analysis of the ecological correlates of 1218 dioecy in angiosperms. J. Evol. Biol. 16, 1006–1018. https://doi.org/10.1046/j.1420- 1219 9101.2003.00559.x 1220 Viljoen, E., Odeny, D.A., Coetzee, M.P., Berger, D.K., Rees, D.J., 2018. Application of chloroplast 1221 phylogenomics to resolve species relationships within the plant genus Amaranthus. J. Mol. Evol. 86, 1222 216–239. https://doi.org/10.1007/s00239-018-9837-9 1223 Villaverde, T., Pokorny, L., Olsson, S., Rincón-Barrado, M., Johnson, M.G., Gardner, E.M., 1224 Wickett, N.J., Molero, J., Riina, R., Sanmartín, I., 2018. Bridging the micro- and 1225 macroevolutionary levels in phylogenomics: Hyb-Seq solves relationships from populations to 1226 species and above. New Phytol. 220, 636–650. https://doi.org/10.1111/nph.15312 1227 Wagenitz, G., 1955. Pollenmorphologie und Systematik in der Gattung Centaurea L. s.l. Flora 142, 213– 1228 279. 1229 Wagenitz, G., 1958. Die Gattung Myopordon Boiss. (Compositae - Cynareae). Ber. Deutsch. Bot. Ges. 1230 71, 271–277. 1231 Wagenitz, G., 1976. Systematics and phylogeny of the Compositae (Asteraceae). Plant Syst. Evol. 125, 1232 29–46. 1233 Weitemier, K., Straub, S.C., Cronn, R.C., Fishbein, M., Schmickl, R., McDonnell, A., Liston, A. 1234 2014. Hyb‐Seq: Combining target enrichment and genome skimming for plant phylogenomics. App. 1235 Plant Sci. 2, 1400042. https://doi.org/10.3732/apps.1400042 1236 Zhang, C., Rabiee, M., Sayyari, E., Mirarab, S., 2018. ASTRAL-III: polynomial time species tree 1237 reconstruction from partially resolved gene trees. BMC Bioinformatics 19, 153. 1238 https://doi.org/10.1186/s12859-018-2129-y 1239 1240 1241 1242 1243 1244 1245 1246 1247 1248 1249 1250 1251 1252 1253 1254 1255 1256 1257 1258 1259 1260 1261 1262 1263 1264 1265 1266 1267 1268 1269 1270 1271 1272 TABLES 1273 1274 Table 1. Historical outline of the classification of the tribe Cardueae. 1275 Classification Taxonomic treatment references Tribe Echinopeae Tribe Carlineae Cassini (1819), Tribe Cardueae Dittrich (1977) Subtribe Carduinae Subtribe Centaureinae Tribe Cardueae Subtribe Echinopsinae Bentham (1873), Subtribe Carlininae Hoffmann (1894), Subtribe Carduinae Bremer (1994) Subtribe Centaureinae Tribe Echinopeae Tribe Cardueae Wagenitz (1976), Subtribe Carlininae Petit (1988, 1997) Subtribe Carduinae-Centaureinae Tribe Cardueae Subtribe Cardopatiinae Susanna et al. (2006), Subtribe Carlininae Susanna and Garcia-Jacas (2007, Subtribe Echinopsinae 2009) Subtribe Carduinae Subtribe Centaureinae Tribe Cardueae Subtribe Carlininae Subtribe Cardopatiinae Subtribe Echinopsinae Subtribe Dipterocominae Subtribe Xerantheminae New treatment proposed here Subtribe Berardiinae Subtribe Staehelininae Subtribe Onopordinae Subtribe Carduinae Subtribe Arctiinae Subtribe Saussureinae Subtribe Centaureinae

1276 1277 1278 1279 1280 1281 1282 1283 1284 1285 1286 1287 1288 1289 1290 1291 1292 1293 1294 1295 1296 1297 1298 1299 1300 Table 2. Some cases of genera with historical conflicts of adscription to subtribes (or even tribes) of the 1301 tribe Cardueae resolved with phylogenies based on Sanger sequencing data. The tribes and treatment 1302 references in bold represent the most recent accepted taxonomic proposal and their corresponding source, 1303 respectively. The last column indicates the taxonomic treatment proposed in the present study for the 1304 traditional taxonomic conflict cases based on new subtribal delimitation. *Xeranthemum group, including 1305 Amphoricarpos, Chardinia, and Siebera. 1306 Taxonomic Taxonomic Genus rank of Included in References of treatments treatment conflict proposed here Mutisiae Bremer (1994), Dittrich (1996a) Cardueae Berardia tribal Cardueae (Berardiinae) (Carduinae) Garcia-Jacas et al. (2002) Cichorieae Boissier (1849) Calenduleae Bentham (1873), Hoffmann (1894) Cardueae Dipterocome tribal Cardueae Susanna and Garcia-Jacas (2009) (Dipterocominae) (Carduinae) Carlininae Dittrich (1977, 1996b), Bremer (1994), Cardopatium subtribal Echinopsidinae Petit (1997) Cardopatiinae Cardopatiinae Susanna et al. (2006) Carlininae Dittrich (1977, 1996b), Bremer (1994), Cousiniopsis subtribal Echinopsidinae Petit (1997) Cardopatiinae Cardopatiinae Susanna et al. (2006) Carduinae Susanna (1987) Femeniasia subtribal Centaureinae Centaureinae Susanna et al. (1995) Carduinae Wagenitz (1958), Dittrich (1977) Myopordon subtribal Centaureinae Centaureinae Susanna et al. (2006) Carduinae Dittrich (1977), Bremer (1994) Nikitinia subtribal Centaureinae Centaureinae Susanna et al. (2002, 2006) Carlininae Bentham (1873), Hoffmann (1894), Staehelina subtribal Carduinae Dittrich (1977, 11996b), Bremer (1994) Staehelininae Petit (1997), Susanna et al. (2006) Centaureinae Dittrich (1977), Bremer (1994) Synurus subtribal Onopordinae Cardueae Häffner and Hellwig (1999) Centaureinae Dittrich (1977), Bremer (1994) Syreitschikovia subtribal Onopordinae Carduinae Susanna et al. (2002) Xeranthemum Carlininae Dittrich (1977, 1996b), Bremer (1994), subtribal Xerantheminae (all the group)* Carduinae Petit (1997), Susanna et al. (2006) 1307 1308 1309 1310 1311 1312 1313 1314 1315 1316 1317 1318 1319 1320 1321 1322 1323 1324 1325 1326 1327 1328 1329 1330 1331 1332 Table 3. Overview and results from previous phylogenetic studies (performed with Sanger sequencing 1333 data) and present study (performed with NGS data) focused on tribe Cardueae with emphasis in the 1334 subtribal classification. Subtribes are divided following the classification proposed here. In the subtribes 1335 rows it is indicated the number of species sampled in the corresponding study. Statistically supported 1336 subtribes are highlighted in green, considering supported subtribes those with a monophyletic pattern and 1337 bootstrap support values ≥ 70% (in ML and MP analyses) and Bayesian posterior probabilities ≥ 0.95 (in 1338 BI analysis). In case that the study used two phylogenetic inference methods, supported subtribes are 1339 considered when one or both methods showed a significantly supported monophyly. The nuclear dataset 1340 summarized corresponds to nuclear HybPiper dataset. Abbreviations used: BI = Bayesian Inference; C = 1341 chloroplast marker; H, H = Häffner and Hellwig; ML = Maximum Likelihood; MP = Maximum 1342 Parsimony; N = nuclear marker; Nº = number; S = Susanna et al. 1343 Systematic studies of the tribe Cardueae at subtribal level S H, H Garcia-Jacas et Susanna et al. Barres et al. (1995) (1999) al. (2002) (2006) (2013) Present study Nº DNA markers 1 1 1 1 1 1+2 1 4 1055 87 DNA marker type N N N C N N+C N C N C MP, MP, MP, Inference method MP MP MP MP BI BI BI BI ML ML Total sampling 36 32 58 42 187 108 116 124 76 67 Carlininae 1 1 10 9 11 9 1 9 6 5 Echinopsidinae 1 1 5 3 7 5 7 7 3 2 Cardopatiinae - - 2 1 2 2 2 2 2 2 Dipterocominae ------1 1 1 1 Berardiinae - 1 1 1 1 1 1 1 1 1 Staehelininae - - 1 - 5 4 3 3 4 4 Xerantheminae - 1 3 3 8 7 7 7 4 3 Onopordiinae - 5 3 4 11 9 11 11 4 4 Carduinae 3 14 15 11 19 15 24 24 7 6 Saussureinae 1 3 8 2 25 5 5 5 7 6 Arctiinae 1 2 5 3 28 13 10 10 6 2 Centaureinae 29 4 5 5 70 38 44 44 31 31 1344 1345 Table 4. Summary metrics of sequence extraction performance for nuclear datasets (targeting 1061 COS 1346 targets; Mandel et al., 2014) using PHYLUCE (Faircloth, 2015) and HybPiper (Johnson et al., 2016) 1347 methods, and chloroplast dataset (targeting coding regions). Parameters from 4 to 11 are calculated based 1348 on dataset “3. Nº of used loci”. Parameter values were extracted with FASconCAT-G v.1.02 (Kück and 1349 Longo, 2014) and AMAS (Borowiec, 2016) programs. Abbreviations used: bp = base pairs; max = 1350 maximum; min = minimum; N° = number; sd = standard deviation. *loci recovered for more than three 1351 species or higher than 3 bp length (see text for details) 1352 Nuclear datasets Chloroplast Parameters PHYLUCE HybPiper dataset method method 1. Nº of species included 87 87 77 2. Nº of recovered loci (% respect COS targets) 902 (85.0) 1057 (99.6) 87 3. Nº of used loci* (% respect COS targets) 776 (73.1) 1055 (99.4) 87 4. Nº of captured loci in ≥ 90% of species (%) 13 (1.7) 729 (69.1) 85 (97.7) 363 (59; 924 (174; 84 (2; 5. Average of recovered loci per species (sd; min-max) 169455) 211021) 7286) 74 (12; 6. Average of species recovered per loci (sd; min-max) 41 (26; 482) 76 (14; 486) 477) 634 (360; 314 (184; 902 (1331; 7. Mean alignment length per locus in bp (sd; min-max) 903825) 231589) 896851) 8. Length of concatenated matrix in bp 492,549 332,260 78,531 10,067 9. Nº of variable sites in the concatenated matrix (%) 296,609 (60.2) 169,903 (51.1) (12.8) 10. Nº of parsimony informative sites in the concatenated 189,716 (38.5) 123,731 (37.2) 4290 (5.5) matrix (%) 11. Proportion of missing data in the concatenated matrix 62.6 16.6 18.2 (%) 1353 1354 1355 FIGURE CAPTIONS 1356 1357 Fig. 1. Phylogenetic reconstruction obtained with HybPiper target extraction method and under the 1358 concatenation approach (maximum likelihood analysis performed with RAxML), showing the 1359 evolutionary relationships among 12 newly defined subtribes within the tribe Cardueae. (A) cladogram, 1360 (B) phylogram. Branch labels indicate bootstrap (BS) support values, those < 70% (in red) are considered 1361 statistically unsupported. 1362 1363 Fig. 2. Phylogenetic reconstruction inferred from the chloroplast dataset (87 coding regions) under the 1364 concatenation approach (maximum likelihood analysis performed with RAxML), showing the 1365 evolutionary relationships among 12 newly defined subtribes within the tribe Cardueae. (A) cladogram, 1366 (B) phylogram. Branch labels indicate bootstrap (BS) support values, those < 70% (in red) are considered 1367 statistically unsupported. 1368 1369 Fig. 3. Phylogenetic trees inferred under coalescence approach (estimated with ASTRAL) for the nuclear 1370 HybPiper, composed by 1055 COS loci. Branch labels indicate local posterior probabilities (LPP), values 1371 < 0.95 (in red) are considered statistically low supported. Results from PhyParts of conflicting and 1372 concordant gene trees are also outlined for the nuclear HybPiper dataset (see text for details). The pie 1373 charts represent the proportion of the four categories (blue, green, red, and gray) for each node. For the 1374 subtribal clades, the number above and below branches indicate the number of gene trees concordant and 1375 in conflict, respectively, with that clade respect the species tree. 1376 1377 Fig. 4. Summary phylogenetic trees recovered in Barres et al. (2013) and present study showing 1378 evolutionary relationships among the 12 subtribes of the tribe Cardueae. On the left, we show phylogenies 1379 reconstructed with nuclear sequence data (A, C, E), and on right (B, D, F), those with plastid data. Trees 1380 redrawn from Barres et al. (2013) were obtained under maximum parsimony (MP) and Bayesian 1381 inference (BI) approaches, using a nuclear dataset (A) composed by sequences of internal transcribed 1382 spacer marker (ITS) and the chloroplast dataset (B) composed by a combination of sequence data from 1383 genes matK, ndhF, rbcL and intergenic spacer trnL-trnF. Phylogenetic trees inferred here derived from 1384 the nuclear HybPiper dataset composed by 1055 COS loci (C, E) and the chloroplast dataset composed by 1385 87 coding regions (D, F), both datasets analyzed under the concatenation approach (app.; maximum 1386 likelihood [ML] analysis performed with RAxML; C, D) and coalescence approach (estimated with 1387 ASTRAL; E). Branch labels indicate: bootstrap support (BS) values in case of MP and ML analysis; 1388 posterior probabilities (pp) in BI; and local posterior probabilities (LPP) in case of under coalescence. 1389 Values with BS < 70% and pp or LPP < 0.95 (in red) are considered statistically low supported. Note that 1390 in trees of Barres et al. (2013), the low supported branches were collapsed in the redrawn version. The 1391 genus Saussurea and Jurinea belong to Saussureinae, and Arctium and Cousinia to Arctiinae. *not 1392 monophyletic subtribe. 1393 1394 Fig. 5. Time-calibrated phylogeny of the 12 newly defined subtribes within the tribe Cardueae. The 1395 phylogenetic tree estimated with nuclear HybPiper dataset (1055 COS loci) and under concatenation 1396 approach (maximum likelihood analysis performed with RAxML) was used as the input tree for the 1397 penalized likelihood dating analysis using treePL. Gray bars on nodes show the 95% of confidence 1398 intervals (CI). Black circles on nodes represent the calibration points (CP) used in dating analysis. 1399 Number above branches are node codes. See Supplementary Table S3 for numeric details of median 1400 estimated ages and Lower 95% CI and Upper 95% CI for each node. In the scale axis, “P” and “Q” 1401 correspond to Pleistocene and Quaternary, respectively. 1402 1403 Fig. 6. Comparison of dating results of age node estimates (in Myr) for tribe Cardueae and subtribes 1404 obtained in the present study, with 1055 COS loci from nuclear genome (upper line in each subtribe), and 1405 those from Barres et al. (2013), with four chloroplast regions: trnL-trnF, matK, ndhF, rbcL (lower line in 1406 each subtribe). Colored boxes comprise the 95% of confidence interval limits and white dots represent the 1407 median age of lineages outlined. Note that subtribes Carduinae, Arctiinae, and Saussureinae are not 1408 represented due to their lack of monophyly in dated phylogeny of Barres et al. (2013). 1409 1410 Fig. 7. Representation of capitula diversity for the 12 newly subtribes recognized here within the tribe 1411 Cardueae: (A) subtribe Carlininae, Atractylis humilis L, Spain; (B) subtribe Cardopatiinae, Cardopatium 1412 corymbosum Pers., Barcelona Botanical Garden; (C) subtribe Echinopsinae, Echinops viscosus Rchb., 1413 Turkey; (D) subtribe Dipterocominae, Dipterocome pusilla Fisch. & C. A. Mey.; (E) subtribe 1414 Xerantheminae, Siebera pungens J. Gay, Turkey; (F) subtribe Berardiinae, Berardia subacaulis falta info 1415 Cristina; (G) subtribe Staehelininae, Staehelina uniflosculosa Sm., Greece; (H) subtribe Onopordinae, 1416 Onopordum nervosum Boiss., Spain; (I) subtribe Carduinae, Cynara baetica (Spreng.) Pau, Spain; (J) 1417 subtribe Arctiinae, Cousinia lanata C. Winkl., Tadjikistan; (K) subtribe Saussureinae, Saussurea 1418 (JORDI), China; and (L) subtribe Centaureinae, Centaurea cephalariifolia Willk., Spain. (A) (B) (C) (E) 1419 (G) (H) (I) (J) (L) Photos by A. Susanna; (D) Photo by A. Pavlenko; (F) Photo by C. Roquet; (K) Photo 1420 by J. López-Pujol. 1421 1422 Fig. 8. Geographical representation of main distribution ranges of the 12 subtribes of the tribe Cardueae 1423 defined in the present study. (A) Nuclear (concatenation approach, 1055 loci) (B)

Family Calyceraceae Nastanthus patagonicus Subfamily Barnadesioideae 28 Chuquiraga sp. Subfamily Famatinanthoideae Fulcaldea stuessyi Famatinanthus sp. 100 Family 100 Subfamily Mutisioideae Chaetanthera sp. Compositae Gerbera hybrida 15 100 Tribe Dicomeae Pleiotaxis pulcherrima 100 Macledium salignum Dicoma anomala 100 100 Tribe Tarchonantheae Brachylaena discolor Oldenburgia papionum Subfamily Tribe Oldenburgieae 100 Carlininae Tugarinovia mongolica Carduoideae 100 100 Atractylodes japonica 100 Thevenotia persica 100 Atractylis echinata 100 100 Carlina diae Carlina biebersteinii Tribe Cardueae 100 Cardopatiinae Cardopatium corymbosum (12 subtribes) Cousiniopsis atractyloides 100 Echinopsinae Echinops strigosus 100 100 Echinops karatavicus Echinops onopordum Dipterocome pusilla 100 97 Dipterocominae 100 Shangwua jacea Xerantheminae 100 Xeranthemum annuum 100 Amphoricarpos exsul 100 Berardiinae Amphoricarpos autariatus Berardia subacaulis 100 Staehelininae Staehelina lobelii 100 Staehelina petiolata 100 100 Staehelina baetica Staehelina dubia 100 Onopordinae Onopordum nervosum 100 100 Olgaea petriprimi 100 Syreitschikovia tenuis Alfredia acantholepis 100 Carduinae Ptilostemon diacantha 95 Cynara cardunculus Galactites tomentosa 100 100 100 Carduus nutans 100 Carduus pycnocephalus 100 Cirsium acaulon Cirsium sairamense 100 Cousinia pusilla 100 Cousinia ninae 100 100 Arctiinae Cousinia sogdiana 100 Arctium minus 100 Arctium abolinii 100 Arctium aureum 100 Saussurea elegans 100 Saussurea leucophylla 100 Saussurea controversa 100 Jurinea orientalis Saussureinae 100 Jurinea fontqueri 100 100 Jurinea carduicephala Jurinea karategini 100 Volutaria canariensis 100 Plagiobasis centauroides Amberboa moschata 100 Klasea quinquefolia 100 80 Klasea flavescens Centaureinae 100 Klasea coriacea 100 Rhaponticum acaule 100 Rhaponticum lyratum 92 Rhaponticum integrifolium 100 Cheirolophus arbutifolius Cheirolophus crassifolius 100 100 Plectocephalus tweediei 100 Plectocephalus cachinalensis 100 Centaurodendron dracaenoides 100 Rhaponticoides centaurium Rhaponticoides alpina 100 100 Psephellus simplicicaulis Psephellus mucroniferus 100 100 Crupina crupinastrum 100 Crupina vulgaris Phalacrachena inuloides 100 100 Carthamus tinctorius 100 Phonus arborescens 100 Carduncellus dianius 100 Centaurea lanigera 100 Centaurea triumfettii 89 Centaurea aspera 61 Centaurea babylonica 50 Centaurea benedicta 100 Centaurea tauromenitana Centaurea clementei

0.04 0.04 (A) Plastid (concatenation approach, 87 genes) (B)

Family Calyceraceae Nastanthus patagonicus Subfamily Barnadesioideae Fulcaldea stuessyi Subfamily Barnadesioideae Chuquiraga sp. 100 Subfamily Famatinanthoideae Family Famatinanthus sp. Compositae 100 Subfamily Mutisioideae Gerbera hybrida 97 100 Tribe Tarchonantheae Brachylaena discolor Tribe Oldenburgieae Oldenburgia papionum 100 100 Tribe Dicomeae Pleiotaxis pulcherrima 100 Dicoma anomala Subfamily Macledium salignum Carduoideae 83 100 Cardopatiinae Cardopatium corymbosum Cousiniopsis atractyloides 100 Tugarinovia mongolica 100 Carlininae 100 Atractylodes japonica Tribe Cardueae 100 Thevenotia persica 93 (12 subtribes) Carlina biebersteinii 100 Atractylis echinata 100 Echinopsinae Echinops strigosus Dipterocominae Echinops karatavicus 76 Dipterocome pusilla 100 Xerantheminae Shangwua jacea 100 Xeranthemum annuum 100 Amphoricarpos autariatus 76 Berardiinae Berardia subacaulis 100 Staehelina lobelii 81 Staehelininae 100 Staehelina petiolata 100 Staehelina dubia Staehelina baetica 100 Onopordinae Onopordum nervosum 100 100 Alfredia acantholepis 62 Olgaea petriprimi Syreitschikovia tenuis Carduinae 92 Galactites tomentosa 96 Cynara cardunculus 88 Ptilostemon diacantha 100 Cirsium sairamense 100 63 100 Carduus nutans Cirsium acaulon 100 Jurinea fontqueri 100 Jurinea karategini 100 Jurinea carduicephala 100 Saussurea leucophylla 100 Saussureinae 65 Saussurea elegans 100 Saussurea controversa Arctiinae 99 Arctium minus Cousinia ninae 100 Volutaria canariensis 100 Amberboa moschata Plagiobasis centauroides 100 Cheirolophus crassifolius Centaureinae 100 92 Cheirolophus arbutifolius 100 Rhaponticum acaule 100 Rhaponticum integrifolium 100 Rhaponticum lyratum 100 Klasea quinquefolia 100 Klasea flavescens Klasea coriacea 100 100 Plectocephalus tweediei 100 Plectocephalus cachinalensis 100 Centaurodendron dracaenoides 100 Psephellus simplicicaulis Psephellus mucroniferus 100 100 Rhaponticoides alpina Rhaponticoides centaurium 100 Phalacrachena inuloides 100 100 Centaurea lanigera Centaurea triumfettii 98 100 Crupina crupinastrum Crupina vulgaris 100 100 Carthamus tinctorius 100 Phonus arborescens 100 Carduncellus dianius 100 Centaurea babylonica 100 Centaurea aspera 100 Centaurea benedicta 100 Centaurea tauromenitana Centaurea clementei

0.004 0.004 Nuclear (coalescence approach, 750 loci)

Nastanthus patagonicus Chuquiraga sp. Fulcaldea stuessyi 0.5 Famatinanthus sp. 1 Chaetanthera sp. 1 Gerbera hybrida 1 Brachylaena discolor 0.77 Oldenburgia papionum 1 1 Pleiotaxis pulcherrima 1 Dicoma anomala Macledium salignum 1 1 77 Tugarinovia mongolica 341 Carlininae 1 Atractylodes japonica 1 Thevenotia persica 1 Atractylis echinata 1 1 Carlina biebersteinii Carlina diae 1 172 217 Cardopatiinae Cardopatium corymbosum Cousiniopsis atractyloides 0.83 1 486 Echinops strigosus Echinopsinae 1 124 Echinops karatavicus 0.78 79 Echinops onopordum 421 Dipterocominae Dipterocome pusilla 1 60 Shangwua jacea 1 372 Xerantheminae 1 Xeranthemum annuum 1 Amphoricarpos autariatus 0.99 44 Amphoricarpos exsul 463 Berardiinae Berardia subacaulis 1 543 Staehelina lobelii 103 Staehelininae 1 Staehelina petiolata 1 1 Staehelina dubia Staehelina baetica 165 Onopordum nervosum 1 1 229 Onopordinae Alfredia acantholepis 1 1 Syreitschikovia tenuis Olgaea petriprimi 1 19 Ptilostemon diacantha 389 Carduinae 0.89 Cynara cardunculus Galactites tomentosa 1 1 1 1 Carduus nutans Carduus pycnocephalus 1 Cirsium acaulon Cirsium sairamense 1 Arctium minus 1 175 Arctium aureum 1 280 Arctiinae 1 Arctium abolinii 1 Cousinia pusilla 1 Cousinia sogdiana 1 Cousinia ninae 1 Saussurea elegans 1 Saussurea leucophylla 1 Saussurea controversa 1 115 1 Jurinea orientalis Jurinea fontqueri 0.99 337 Saussureinae 1 Jurinea karategini Jurinea carduicephala 0.98 Volutaria canariensis 1 Plagiobasis centauroides Amberboa moschata 1 Klasea flavescens 172 0.87 1 Klasea quinquefolia 338 1 Centaureinae Klasea coriacea 1 Rhaponticum acaule 1 0.81 Rhaponticum lyratum Rhaponticum integrifolium 1 Cheirolophus arbutifolius Cheirolophus crassifolius 1 1 Plectocephalus tweediei 1 Centaurodendron dracaenoides 1 Plectocephalus cachinalensis 1 Rhaponticoides centaurium Rhaponticoides alpina 0.99 1 Psephellus mucroniferus 1 Psephellus simplicicaulis 1 Crupina crupinastrum 1 Crupina vulgaris Phalacrachena inuloides 1 1 Carthamus tinctorius 1 Phonus arborescens 1 Carduncellus dianius 1 Centaurea lanigera Centaurea triumfettii 0.98 Centaurea benedicta 0.95 0.43 Centaurea aspera Centaurea babylonica 0.64 1 Centaurea tauromenitana Centaurea clementei 0.9

Blue: gene trees supporting that clade Green: gene trees supporting the main alternative conflicting topology for that clade Red: gene trees supporting the remaining conflicting topologies for that clade Gray: gene trees supporting or confliting that clade with BS < 30% Nuclear phylogenies Plastid phylogenies

outgroup outgroup Carlininae 1/ Cardopatiinae Cardopatiinae 81 Carlininae 1/ 1/- Echinopsinae 100 Echinopsinae Dipterocominae 1/79 Dipterocominae 0.96/- Xerantheminae Xerantheminae 1/- Berardiinae 1/ Berardiinae Staehelininae 99 Staehelininae 1/- Onopordinae Onopordinae Carduinae Carduinae* 1/85 1/- Saussureinae Arctium 1/-

Barres et al. (2013) Jurinea 0.99/- Arctiinae 1/- 1/- Saussurea (A) ITS Centaureinae (B) 4 regions concatenation app. Cousinia Centaureinae

outgroup outgroup Carlininae Cardopatiinae 100 100 Cardopatiinae Carlininae 100 100 Echinopsinae Echinopsinae 76 Dipterocominae 100 97 Dipterocominae Xerantheminae 100 Xerantheminae 100 Berardiinae 81 76 Berardiinae 100 Staehelininae Staehelininae 100 100 Onopordinae Onopordinae Carduinae 100 Carduinae 100 63 Jurinea 100 Saussureinae 100 (C) 1055 loci 100 (D) 87 genes Saussurea 100 Arctiinae 100 65 concatenation app. Centaureinae concatenation app. Arctiinae Centaureinae

outgroup Present study Carlininae 1 Cardopatiinae 0.83 Echinopsinae 0.78 Dipterocominae 1 Xerantheminae 0.99 Berardiinae 1 Staehelininae 1 Onopordinae 1 Carduinae 1 1 Saussureinae (E) 1055 loci 0.99 Arctiinae coalescence app. Centaureinae Nastanthus patagonicus (F. Calyceraceae) 1 3 Chuquiraga sp. (F. Compositae, S. Barnadesioideae) 2 Fulcaldea stuessyi (F. Compositae, S. Barnadesioideae) CP1 Famatinanthus sp. (F. Compositae, S. Famatinanthoideae) 6 4 Chaetanthera sp. (F. Compositae, S. Mutisioideae) Gerbera hybrida (F. Compositae, S. Mutisioideae) 5 8 Pleiotaxis pulcherrima (F. Compositae, S. Carduoideae, T. Dicomeae) 9 Macledium salignum (F. Compositae, S. Carduoideae, T. Dicomeae) Dicoma anomala (F. Compositae, S. Carduoideae, T. Dicomeae) CP2 7 11 Brachylaena discolor (F. Compositae, S. Carduoideae, T. Tarchonantheae) Oldenburgia papionum (F. Compositae, S. Carduoideae, T. Oldenburgieae) Subfamily Carduoideae 13 Tugarinovia mongolica 10 14 Atractylodes japonica 15 Thevenotia persica 16 Atractylis echinata Carlininae 17 12 Carlina diae Carlina biebersteinii 19 Cardopatium corymbosum Tribe Cardueae Cousiniopsis atractyloides Cardopatiinae 21 Echinops strigosus 18 22 Echinops karatavicus Echinopsinae Echinops onopordum 20 24 Dipterocome pusilla Dipterocominae 25 Shangwua jacea 26 Xeranthemum annuum 27 Amphoricarpos exsul Xerantheminae 23 Amphoricarpos autariatus Berardia subacaulis Berardiinae 30 Staehelina lobelii 31 Staehelina petiolata 28 32 Staehelina baetica Staehelininae Staehelina dubia 34 Onopordum nervosum 29 35 Olgaea petriprimi 36 Syreitschikovia tenuis Onopordinae Alfredia acantholepis 38 Ptilostemon diacantha 39 Cynara cardunculus Galactites tomentosa 33 40 42 Carduus nutans Carduinae 41 Carduus pycnocephalus CP3 43 Cirsium acaulon Node Median age [95% CI] Cirsium sairamense 47 Cousinia pusilla Subfamily level 48 46 Cousinia ninae 37 Cousinia sogdiana 40.09 [33.68-46.67] 49 7 CP4 Arctium minus Arctiinae 50 Arctium abolinii 45 Tribal level Arctium aureum 52 Saussurea elegans 53 Saussurea leucophylla 12 34.12 [29.94-40.04] 51 Saussurea controversa 55 Jurinea orientalis 54 Saussureinae Subtribal level Jurinea fontqueri 44 56 Jurinea carduicephala Jurinea karategini 13 31.09 [27.31-36.45] 58 Volutaria canariensis 59 Plagiobasis centauroides 19 26.30 [23.10-30.88] Amberboa moschata 62 Klasea quinquefolia 57 63 14.58 [12.00-18.17] 61 Klasea flavescens 21 Klasea coriacea 64 Rhaponticum acaule 24 26.57 [23.62-30.24] 65 Rhaponticum lyratum 60 Rhaponticum integrifolium 67 24.56 [21.82-27.93] Cheirolophus arbutifolius 25 Cheirolophus crassifolius 66 69 Plectocephalus tweediei 28 24.54 [21.76-27.40] 70 Plectocephalus cachinalensis Centaurodendron dracaenoides 68 72 7.26 [6.50-8.00] Rhaponticoides centaurium 30 Rhaponticoides alpina 71 Centaureinae 74 Psephellus simplicicaulis 15.59 [13.65-17.57] Psephellus mucroniferus 34 73 76 Crupina crupinastrum 16.33 [14.55-18.73] 75 Crupina vulgaris 38 Phalacrachena inuloides 77 79 Carthamus tinctorius 46 7.86 [6.63-9.01] 80 Phonus arborescens 78 Carduncellus dianius 82 12.48 [11.40-13.68] Centaurea lanigera 51 81 Centaurea triumfettii CP5 84 Centaurea aspera 57 12.39 [11.39-13.66] 83 Centaurea babylonica 85 Centaurea benedicta 86 Centaurea tauromenitana Centaurea clementei 7.0

Cretaceous Paleocene Eocene Oligocene Miocene P Q

80 70 60 50 40 30 20 10 0 Age (Myr)

Carlininae Echinopsinae Cardopatiinae Staehelininae 1 2 Onopordinae Two different main time periods of subtribes origin Dipterocominae Xerantheminae Carduinae Berardiinae Arctiinae Saussureinae Centaureinae

(Myr) 10 20 30 40

Centaureinae ● Subtribe

● Onopordinae ● Subtribe

● Staehelininae

● Subtribe Subtribe

● Berardiinae

● Subtribe Subtribe

●

● Xerantheminae Subtribe Subtribe

●

● Dipterocominae Subtribe Subtribe

●

● Echinopsinae

● Subtribe

● Cardopatiinae

● Subtribe

● Carlininae

● Subtribe

Barres et al. (2013) al. et Barres

e ● Carduea

Results obtained in the present study present the in obtained Results

e ib r ● T A B C D

E F G H

I J K L (A) (B)

Carlininae Cardopatiinae

(C) (D)

Echinopsinae Dipterocominae

(E) (F)

Xerantheminae Berardiinae

(G) (H)

Staehelininae Onopordinae

(I) (J)

Carduinae Saussureinae

(K) (L)

Arctiinae Centaureinae (A) Nuclear PHYLUCE (concatenation approach, 776 loci) (B)

Family Calyceraceae Nastanthus patagonicus Subfamily Barnadesioideae 100 Chuquiraga sp. Subfamily Famatinanthoideae Fulcaldea stuessyi Famatinanthus sp. 100 Family 100 Subfamily Mutisioideae Chaetanthera sp. Gerbera hybrida Compositae 100 39 Tribe Tarchonantheae Brachylaena discolor Tribe Oldenburgieae Oldenburgia papionum 100 Pleiotaxis pulcherrima 100 Tribe Dicomeae 100 Macledium salignum Subfamily Dicoma anomala Tugarinovia mongolica Carduoideae 65 100 Carlininae 100 Atractylodes japonica 100 Thevenotia persica 100 Atractylis echinata 100 100 Carlina biebersteinii Tribe Cardueae Carlina diae 100 Cardopatiinae Cardopatium corymbosum (12 subtribes) Cousiniopsis atractyloides 100 100 Echinopsinae Echinops strigosus 100 Echinops karatavicus 80 Dipterocominae Echinops onopordum Dipterocome pusilla 100 Xerantheminae Shangwua jacea 100 100 Xeranthemum annuum 100 Amphoricarpos exsul 64 Berardiinae Amphoricarpos autariatus Berardia subacaulis 100 Staehelininae Staehelina lobelii 100 Staehelina petiolata 100 100 Staehelina dubia Staehelina baetica 100 Onopordinae Onopordum nervosum 100 100 Alfredia acantholepis 100 Syreitschikovia tenuis Olgaea petriprimi 100 Carduinae Ptilostemon diacantha 100 Cynara cardunculus 100 Galactites tomentosa 100 100 Carduus pycnocephalus 100 Carduus nutans 100 Cirsium acaulon Cirsium sairamense 100 Arctium minus 100 100 Arctium abolinii Arctiinae Arctium aureum 100 100 Cousinia pusilla 100 Cousinia ninae 100 Cousinia sogdiana 100 Saussurea elegans 100 Saussurea controversa 100 Saussurea leucophylla 100 Jurinea orientalis Saussureinae 100 Jurinea fontqueri 100 Jurinea carduicephala 100 Jurinea karategini 100 Volutaria canariensis 100 Plagiobasis centauroides 100 Amberboa moschata 100 Rhaponticum acaule 100 100 Rhaponticum lyratum Rhaponticum integrifolium 100 Klasea flavescens 100 100 Klasea coriacea Klasea quinquefolia Centaureinae 100 Cheirolophus crassifolius Cheirolophus arbutifolius 100 100 Plectocephalus tweediei 100 Centaurodendron dracaenoides 100 Plectocephalus cachinalensis 100 Rhaponticoides centaurium Rhaponticoides alpina 100 100 Psephellus mucroniferus Psephellus simplicicaulis 100 100 Crupina vulgaris Crupina crupinastrum 100 97 Phalacrachena inuloides 100 Centaurea lanigera 100 Centaurea triumfettii 100 Carthamus tinctorius 100 Phonus arborescens 97 Carduncellus dianius 100 Centaurea clementei 100 Centaurea tauromenitana 100 Centaurea aspera 100 Centaurea benedicta Centaurea babylonica

0.07 0.07 Nuclear PHYLUCE (coalescence approach, 776 loci)

Nastanthus patagonicus 1 Fulcaldea stuessyi Chuquiraga sp. Famatinanthus sp. 1 Chaetanthera sp. 0.99 Gerbera hybrida 1 Oldenburgia papionum 0.99 1 Brachylaena discolor 1 Pleiotaxis pulcherrima 1 Dicoma anomala Macledium salignum 1 1 Carlininae Tugarinovia mongolica 1 Atractylodes japonica 1 Thevenotia persica 1 Atractylis echinata 1 1 Carlina diae Carlina biebersteinii 1 Cardopatiinae Cardopatium corymbosum Cousiniopsis atractyloides 0.84 1 Echinopsinae Echinops strigosus 1 Echinops karatavicus 0.32 Dipterocominae Echinops onopordum Dipterocome pusilla 1 Xerantheminae Shangwua jacea 1 1 Xeranthemum annuum 1 Amphoricarpos exsul 0.9 Berardiinae Amphoricarpos autariatus Berardia subacaulis 1 Staehelininae Staehelina lobelii 1 Staehelina petiolata 1 1 Staehelina dubia Staehelina baetica 1 Onopordinae Onopordum nervosum 1 1 Alfredia acantholepis 0.71 Olgaea petriprimi Syreitschikovia tenuis 1 Carduinae Ptilostemon diacantha 0.91 Cynara cardunculus Galactites tomentosa 1 1 1 Carduus nutans 1 Carduus pycnocephalus 1 Cirsium sairamense Cirsium acaulon 1 Arctium minus 1 1 Arctium abolinii 1 Arctiinae Arctium aureum 1 Cousinia pusilla 1 Cousinia ninae 1 Cousinia sogdiana 1 Saussurea elegans 1 Saussurea leucophylla 1 Saussurea controversa 1 Jurinea karategini Saussureinae 1 Jurinea carduicephala 1 1 Jurinea fontqueri Jurinea orientalis Centaureinae 1 Volutaria canariensis 1 Plagiobasis centauroides Amberboa moschata 1 Klasea flavescens 1 0.98 1 Klasea coriacea Klasea quinquefolia 1 Rhaponticum acaule 1 Rhaponticum lyratum 0.96 Rhaponticum integrifolium 1 Cheirolophus crassifolius Cheirolophus arbutifolius 1 1 Plectocephalus tweediei 1 Centaurodendron dracaenoides 1 Plectocephalus cachinalensis 1 Rhaponticoides alpina Rhaponticoides centaurium 0.91 1 Psephellus simplicicaulis Psephellus mucroniferus 1 1 Crupina vulgaris 1 Crupina crupinastrum Phalacrachena inuloides 1 1 Carthamus tinctorius 1 Phonus arborescens 0.97 Carduncellus dianius 1 Centaurea lanigera 0.87 Centaurea triumfettii 1 Centaurea tauromenitana 1 Centaurea clementei 0.75 Centaurea aspera 0.89 Centaurea babylonica 2.0 Centaurea benedicta