Nuclear and Chloroplast DNA Variation in Cephalaria Squamiflora (Dipsacaceae), a Disjunct Mediterranean Species

TAXON • 12 October 2009: 12 pp. Rosselló & al. • DNA variation in Cephalaria squamiflora

Nuclear and chloroplast DNA variation in Cephalaria squamiflora (Dipsacaceae), a disjunct Mediterranean species

Josep A. Rosselló1,5, Raúl Cosín1, Gianluigi Bacchetta2, Salvatore Brullo3 & Maria Mayol4

1 Jardín Botánico, Icbibe, Universidad de Valencia, c/Quart 80, 46008 Valencia, Spain. [email protected] (author for correspondence) 2 Centro Conservazione Biodiversità, Dipartimento di Scienze Botaniche, Università degli Studi di Cagliari, S. Ignazio da Laconi 13, 09123 Cagliari, Italia 3 Dipartimento di Botanica, Università degli Studi di Catania, via A. Longo 19, 95125 Catania, Italia 4 CREAF (Center for Ecological Research and Forestry Applications), Autonomous University of Barcelona, 08193 Bellaterra, Spain 5 Marimurtra Botanical Garden, Carl Faust Fdn., PO Box 112, 17300 Blanes, Catalonia, Spain

Cephalaria squamiflora is a chamaephyte restricted to rupicolous habitats in islands of the Western (Balearic Islands, Sardinia) and Eastern Mediterranean (Crete and few Aegean islands). Four narrowly distributed races (subspp. squamiflora, mediterranea, ebusitana, balearica) have been described to encompass the morphological variation within the species. We have used nuclear ribosomal ITS and cpDNA sequences to assess how the patterns of molecular differentiation are related to taxonomic and geographic boundaries. Extensive intragenomic ITS variation was detected in samples from all territories, the average sequence divergence among cloned ribotypes was 1.339%. The parsimony network of cloned ITS sequences suggests a split between Eastern and Western Mediterranean accessions. Chloroplast DNA sequences showed five distinct haplotypes, only one of which was shared between islands (Majorca and Sardinia). Both nuclear and cpDNA markers supported the monophyly of the C. squamiflora complex and identified a highly structured pattern of molecular variation composed by sister monophyletic lineages that mirror major biogeographic units (Western and Eastern Mediterranean). The molecular evidence supports the hypotheses that vicariance events linked to the geological history of the region or dispersal across the Mediterranean may explain the distribution of the complex.

KEYWORDS: continental islands, Dipsacaceae, ITS paralogs, trnL-trnF spacer

(Sieber) Greuter (Dipsacaceae), have been reported to be INTRODUCTION restricted to both Western and Eastern Mediterranean Islands constitute major plant biodiversity centers in islands (Thompson, 2005; Galbany-Casals & al., 2006). the Mediterranean basin (Médail & Quezel, 1999). Medi- Cephalaria squamiflora, together with C. leucantha terranean islands and islets are not only singular for their (L.) Roem. & Schult., C. boetica Boiss. and C. lineari- species richness, but also for the narrow distribution of folia Lange, constitutes the subgenus Fimbriatocarpus most of their flora. Typically, high endemicity rates char- Szabó, a small assemblage of perennial species with a acterize the flora of many Mediterranean islands. Thus, narrow distribution centered in the Western Mediterra- the endemic element of the Balearic Islands (Médail & nean basin (Verlaque, 1985; Devesa, 2007). It was hy- Verlaque, 1997), Corsica (Gamisans & al., 1985), Sardinia pothesized that subgenus Fimbriatocarpus is primitive (Médail & Verlaque, 1997), Sicily (Médail & Verlaque, within Cephalaria and probably derived from ancestors 1997), Crete (Turland & al., 1993), and Cyprus (Alziar, belonging to the South African subgenus Lophocarpus 1995) ranges from 5.3% to 12% of their floras. Endemic (Verlaque, 1985). A cladistic study of the Dipsacaceae species may even characterize single small islands and using morphological and palynological characters showed islets of much reduced extension throughout the Mediter- a monophyletic Dipsacus clade, including Dipsacus and ranean basin. Cephalaria, characterized by a synapomorphic unilobate One of the salient features of the endemic flora of stigma (Caputo & Cozzolino, 1994). Dipsacus appeared the Mediterranean islands is that very few species are paraphyletic and basal to Cephalaria ; Cepahalaria subgg. shared between two or more distant archipelagos. In fact, Lophocarpus and Fimbriatocarpus formed the most basal only two single endemic taxa, Helichrysum microphyl- grade in Cephalaria, but were not sister taxa (Caputo & lum Willd. (Asteraceae) and Cephalaria squamiflora Cozzolino, 1994). Phylogenetic analysis of Dipsacaceae

1 Rosselló & al. • DNA variation in Cephalaria squamiflora TAXON • 12 October 2009: 12 pp.

using nuclear ribosomal and chloroplast DNA sequences fragmented distribution, including hotspots of endemism supported the Dipsacus clade, but further resolution and refuge centers for relict taxa (e.g., Simmons & al., within this clade was very limited due to the low sampling 1998; Bacchetta & al., 2008). Morphological variability (Caputo & al., 2004). within populations is usually low in rupicolous species, Cephalaria squamiflora is a long-lived rupicolous but discontinuities are usually present among allopatric chamaephyte that lives in calcareous rocky places (crev- populations. This geographical variability can lead to ices of vertical cliffs and overhangs), forming scattered unsatisfactory taxonomic treatments, since the periph- populations from sea level to about 1,500 m altitude. The eral populations of allopatric species may sometimes be species shows a distribution restricted to the Balearic Is- linked by a mosaic of morphologically intermediate, but lands (Majorca, Ibiza) and Sardinia in the Western Medi- geographically still discontinuous, non-interbreeding terranean, and Crete and several Aegean islands (Karpa- and long-isolated populations. A further question to be thos, Ikaria, Chios, Alonissos, Kyra Panaghia, Gioura, addressed in chasmophyte groups is to what extent the Skyros, Sifnos) in the Eastern Mediterranean (Greuter, groups that taxonomists have defined as cohesive con- 1967; Arrigoni, 1978; Mus & al., 1990; Fielding & Turland, specific populations may be the products of recurrent 2005; Bacchetta & al., 2008). evolution from separate lineages (Levin, 2001). This is a The populations of C. squamiflora show an overall possibility given the similar biological features associated morphological similarity. However, subtle variation in with saxicolous environments which have been favoured leaf shape and hairiness, shape of floral bracts, and epi- or retained by similar selective pressures in unrelated calyx ornamentation has been reported to be correlated taxonomic groups (Davis, 1951). with geography (Illario, 1938; Greuter, 1967; Pignatti, Due to its exclusively chasmophyte habitat, inter- 1977; Arrigoni, 1978), allowing the recognition of up population differentiation, and restricted and discontinu- to five very closely related races. With the exception of ous distribution, C. squamiflora is a suitable taxon (1) to Bacchetta & al. (2008), who adopt a very narrow species assess whether molecular discontinuities exist within the concept, all authors dealing with the C. squamiflora group species, and if these are linked to island or archipelago (Greuter, 1967; Pignatti, 1977; Arrigoni, 1978, Mus & al., boundaries, (2) to explore how many evolutionary lineages 1990; Devesa, 2007) have treated the major geographic may be detected within the species, (3) to assess the pat- variants as subspecies of C. squamiflora: subsp. squami- terns of molecular differentiation in narrowly distributed flora (Crete, Aegean islands), subsp. mediterranea (Viv.) insular endemics using molecular markers. In this paper Pignatti (Sardinia, Ibiza), and subsp. balearica (Coss. ex we report the patterns of molecular variation within C. Willk.) Greuter (Majorca). However, no agreement has squamiflora using sequences obtained from the nuclear been reached concerning the number and delimitation of and cpDNA genomes. The nuclear ribosomal ITS region taxonomic entities from the Western Mediterranean. Thus, (ITS1-5.8S-ITS2), the trnL (UAA) intron and the cpDNA Greuter (1967) pointed out that the recognition of subsp. intergenic spacer trnL (UAA) 3′ exon – trnF (GAA) have mediterranea from subsp. balearica on morphological been previously used to infer evolutionary relationships grounds was not justified and suggested that both subspe- at various taxonomic levels in Dipsacaceae (Zhang & al., cies should be merged within a single taxon. In addition, 2003; Caputo & al., 2004) and could constitute molecular authors have suggested that subsp. mediterranea should targets to assess patterns of molecular variation within include all populations from Sardinia and Ibiza (Mus & Cephalaria species. al., 1990; Devesa, 2007). But this view it is not followed by others claiming that the name subsp. mediterranea should be restricted to Sardinian individuals and the populations from Ibiza referred to a distinct entity either at the MATERIALS AND METHODS subspecific, C. squamiflora subsp. ebusitana (O. Bolòs & Plant material and DNA extraction. — Popula- Vigo) Romo (Bolòs, 1991; Romo, 1994) or specific level, tions of the C. squamiflora group were sampled through- C. ebusitana (O. Bolòs & Vigo) Bacch., Brullo & Giusso out their distribution range (Appendix; Fig. 1). The re- (Bacchetta & al., 2008). Recently, Bacchetta & al. (2008) port of the species in Corsica (Viviani, 1825) is likely have described a new species (C. bigazzii Bacch., Brullo due to a herbarium labeling error (Arrigoni, 1978) and & Giusso) from a single population of SW Sardinia previ- its presence has not been reconfirmed in the island in ously identified as C. squamiflora subsp. mediterranea recent times (Gamisans & Jeanmonod, 1993; Gamisans (Ballero & al., 2000; Maxia & Usai, 2004). & Marzocchi, 1996). For the purposes of this paper C. Cephalaria squamiflora belongs to the large group ebusitana and C. bigazzii have been included under C. sq- of endemic insular species growing exclusively in rupico- uamiflora subsp. mediterranea, and C. balearica has been lous habitats. These environments are usually character- treated at the subspecific level under C. squamiflora. Total ized by singular plant communities of very limited and DNA was extracted from fresh, silica gel–dried leaves or

2 TAXON • 12 October 2009: 12 pp. Rosselló & al. • DNA variation in Cephalaria squamiflora

herbarium specimens using the CTAB protocol of Doyle extension step at 72°C for 5 min was included. The PCR & Doyle (1987), scaled down to perform the process in products were separated on 1.0% agarose gels and puri- 1.5 mL microfuge tubes. fied using the High Pure PCR Product Purification Kit Nuclear ribosomal ITS sequences. — The re- (Roche Diagnostics). PCR products from eight accessions gion including ITS-1, 5.8S and ITS-2 was amplified us- encompassing all major biogeographic territories (West- ing the primer pair ITS-5/ITS-4 (White & al., 1990). In ern and Eastern Balearics, Sardinia, Aegean islands) were some instances, direct sequences were unreadable from gel-purified and ligated into the vector provided with the the electrophoretograms due to the presence of putative p-GEM-T (Promega) cloning kit. Plasmid DNA from in- paralogous products of different length. In these cases, dividual recombinant colonies was isolated according to a the primers ITS-2 and ITS-3 (White & al., 1990) were miniprep protocol (High Pure Plasmid Isolationkit, Roche also used to determine the bases of the remaining se- Diagnostics). For sequencing, purified PCR or cloned ITS quence. PCR reactions were performed in 50 μl, contain- products were reacted with BigDye Terminator Cycle Se- ing 75 mM Tris-HCl (pH 9.0 at 25°C), 5 mM KCl, 20 mM quencing Ready Reaction (Perkin-Elmer, Applied Biosys-

(NH4)2SO4, 0.0001% BSA, 1.5 mM MgCl2, 5% DMSO, tems) using the amplification primers. GenBank accession 200 μM of each dATP, dCTP, dGTP and dTTP, 0.2 μM of numbers are provided in the Appendix. each primer, approximately 50–100 ng of genomic DNA Integrity and secondary structure stability of and 0.75 units of Taq polymerase. Amplifications com- the ribosomal sequences. — The boundaries of the ITS prised an initial incubation step at 94°C for 3 min and 40 sequences and ribosomal coding regions were determined cycles of 15 s at 94°C, 15 s at 56°C, 30 s at 72°C. A final by comparison with published sequences of Dipsacaceae

Fig. 1. Distribution and sampling locations of Cephalaria squamiflora. Inset 1 (Balearic Islands), C. squamiflora subsp. balearica (squares) and C. squamiflora subsp. mediterranea (circles); inset 2 (Sardinia), C. squamiflora subsp. mediterranea (circles); inset 3 (Crete, Aegean islands), C. squamiflora subsp. squamiflora (polygons). Details of accession codes are indicated in the Appendix.

3 Rosselló & al. • DNA variation in Cephalaria squamiflora TAXON • 12 October 2009: 12 pp.

available in GenBank (Caputo & al., 2004). Sequences conventional tree building methodologies are used is were compared for length variation and guanine plus cy- likely due on the facts that ribosomal repeat units with tosine content (G+C%, hereafter). We explored the DNA multiple ancestry are related by a multifurcating rather sequences for the presence of several structural motifs. than dichotomous branching pattern and the true relation- Thus, at the ITS-1 region we searched for the presence and ship between groups of sequences is reticulate rather than length of the conserved angiosperm motif GGCRY-(4 to linear (Allaby & Brown, 2001). 7n)-GYGYCAAGGAA (Liu & Schardl, 1994). This core Parsimony network was implemented with the Net- sequence is most often associated with part of a hairpin work version 4.5 software (available at the website http:// structure, whereas the AAGGAA motif is predicted to www.fluxus-engineering.com) using the Reduced Median stand as a non base-paired sequence in an internal loop (RM) network method (Bandelt & al., 1995). In order to structure. It has been hypothesized that the whole or par- compare the parsimony network with other tree building tial conserved motif serves as a critical processing signal methods, the datasets were analyzed with conventional in the enzymatic processing of the ribosomal RNA (Liu parsimony and distance-based methods. & Schardl, 1994). Lastly, secondary structures of the Phylogenetic analyses using standard parsimony ITS-2 RNA transcripts were examined for the presence were conducted using Fitch parsimony with equal weight- of the tetranucleotide UGGU, which is preserved across ing of all characters and of transitions/transversions as the flowering plants. This motif is located within the C4 implemented in PAUP*4.0b10 (Swofford, 2002). Heu- domain and is contained near the apex of helix III on ristic searches were replicated 100 times with random the 5′ side (Mai & Coleman, 1997). Predicted secondary taxon-addition sequences, Tree Bisection-Reconnection structures of the ITS-1 and ITS-2 RNA transcripts and (TBR) branch swapping, and with the options Multrees associated free energy values were evaluated using the and Steepest Descent in effect. Support for monophyletic minimum-free energy (MFE) algorithm (Zuker, 1989) groups was assessed by both “fast” bootstrapping (10,000 and the latest free energy rules (Mathews & al., 1999; resamplings of the data) and “full” bootstrapping (100 Zuker & al., 1999). Fold predictions were made at the resamplings) using the heuristic search strategy as indi- M. Zuker web server (Zuker, 2003; http://www.bioinfo. cated above (see Mort & al., 2000 for discussion). In addi- rpi.edu/~zukerm/rna/) using the MFOLD program (ver- tion, phylogenetic reconstructions of ITS sequences were sion 3.1). Foldings were conducted at 37°C using a search also performed by using distance-based methods using within 5% of the thermodynamic optimality set. Individ- PAUP* 4.0b10, the Kimura 2-parameter distance model ual bases within foldings were annotated with the descrip- (Kimura, 1980), and the Neighbor-Joining method (Saitou tor P-num (Jaeger & al., 1989, 1990), which indicates the & Nei, 1987). For parsimony analysis, inferred indels from propensity of individual nucleotides to participate in base the aligned sequences (except a variable poly-A satellite pairs and consequently whether or not a predicted base region) were coded as an additional binary data and added pair is well determined. RNA structures were displayed to the molecular matrix. Available GenBank sequences with the RnaViz package (De Rijk & De Wachter, 1997). from Cephalaria (C. leucantha : ITS-1, AJ426523; ITS-2, cpDNA sequences. — The region including the AJ426523; trnL intron, AJ 427376; C. syriaca: ITS-1, trnL intron and the intergenic spacer between the trnL (3′) AJ426525; ITS-2, AJ426526; trnL intron, AJ427377) and and trnF coding sequences was amplified using the uni- Dipsacus species (that are sister to Cephalaria ; Caputo versal primers e and f described in Taberlet & al. (1991). & al., 2004) were downloaded for comparative purposes. PCR reactions, purifications and direct sequencing were performed as above. Phylogenetic analyses. — Cephalaria sequences were of similar length and were manually aligned and RESULTS edited with DAMBE (Xia & Xie, 2001). Ribosomal ITS sequences: direct sequenc- Network methods are designed for DNA haplotypes ing. — Direct ITS sequences from Cephalaria acces- and are believed to work better with infraspecific phy- sions showed polymorphic sites at 33 positions, of which logenies than standard parsimony algorithms (Posada & 14 were located in the ITS-1 and 19 in the ITS-2 region. Crandall, 2001) since it makes fewer assumptions about Unreadable electrophoretograms were found in direct the direction of evolution, the extent of sexual isolation, sequences at the ITS-1 region in all the Western Mediter- and the pattern of ancestry and descent (Allaby & Brown, ranean accessions analyzed. Particularly, the sequences 2001). The optimal use of networks when analyzing ri- could not be read after a poly-A tract located at the 5′ end. bosomal multigene families subjected to homogeniza- The three accessions from the Aegean islands showed tion processes, such as unequal crossing over and biased unreadable sequences near the middle of the ITS-2 re- gene conversion has been reported (Allaby & Brown, gion in forward and reverse sequencing electrophoreto- 2001). Poor resolution using multigene families when grams. The displacement observed between peaks in both

4 TAXON • 12 October 2009: 12 pp. Rosselló & al. • DNA variation in Cephalaria squamiflora

spacers suggests that intragenomic ITS variants differing across the whole dataset. Overall, 32 distinct clones were in length could be present within the PCR-amplified ribo- detected and the average sequence divergence among somal pool. In addition, most accessions showed double ribotypes (Kimura 2-parameter model) was 1.339% ± peaks in different sites at the ITS-1 and ITS-2 regions. 0.002. The most divergent ribotypes were from subsp. Overall, we inferred 8 intragenomic polymorphisms in squamiflora (Mt. Kali Limni, KA) and subsp. mediter- ITS-1 and 18 in ITS-2. Given such sequence ambiguities ranea (Lisandrus, BU), differing by 3.311% ± 0.007. Se- it was necessary to clone the PCR products in order to quence divergence values within and between ribotypes assess the number of intragenomic ribotypes present in retrieved from each subspecies, territories and accessions selected accessions. are depicted in Table 1. Within single accessions, intrage- Cloned sequences. — We cloned the ribosomal nomic divergence among ribotypes ranged between 0.000 products amplified from eight individuals coming from ± 0.000 (C. squamiflora subsp. mediterranea from Fumai, separate geographic regions for which direct ITS se- FU), but only three clones sequenced) to 1.328% ± 0.003 quences were previously generated: one from the Aegean (C. squamiflora subsp. mediterranea from Lisandrus, islands (Karpathos), four from Sardinia (Lisandrus, Mt. BU). Within territories, ribosomal sequences from Sar- Irveri, Mt. Fumai, Janna Nurai), two from Ibiza (Cap Jueu, dinian C. squamiflora subsp. mediterranea showed the Ses Roques Altes), and one from Majorca (Son Torrella). highest sequence divergence values (2.265% ± 0.006), Altogether, we sequenced 50 clones, 5 from subsp. squa- whereas subsp. squamiflora and Ibiza subsp. mediterra- miflora, 8 from subsp. balearica, and 37 from subsp. med- nea samples showed the lower values (1.221 ± 0.004 and iterranea (25 from Sardinia and 12 from Ibiza accessions). 1.227 ± 0.004), respectively. The number of analyzed clones per individual ranged from Comparison of direct and cloned ITS se- three to ten. The G+C% was similar in all clones, rang- quences. — Usually, the polymorphic sites found in ing from 63.8 to 65.2, and averaged 64.5%. The length of the cloned sequences matched the diversity found in the ITS-1 varied from 246 to 250 bp. Most of the length varia- electrophoretograms of the direct sequences. However, in tion was due to the presence of a poly-A stretch (showing some accessions cloning revealed the presence of bases between six and ten adenines) located at the 3′ end of the that were not apparent in the direct sequences. These ITS-1 spacer. In addition, clones from subsp. squamiflora intragenomic polymorphisms could be either related to showed slightly shorter ITS-1 sequences than those from cloning or sequencing artifacts. In addition, due to a non- subspp. mediterranea and balearica due to an indel 1 bp exhaustive cloning screening effort, not all the gene pool long. The number of variable sites (excluding length vari- of ribotypes that should be present within individual ge- ants) in this spacer was twelve. ITS-2 showed a uniform nomes, as inferred by direct sequencing, was retrieved. length of 248 bp, and 17 polymorphic sites were present A close inspection of both direct and cloned sequences

Table 1. Number of sequenced clones, number of ribotypes, and sequence divergence (Kimura 2-parameter) within accessions and among ribosomal ITS clones retrieved from C. squamiflora. Sequenced Sequence divergence [%] ± standard deviation Taxon Accession clones (ribotypes) Min. Mean Max. Cephalaria squamiflora subsp. squamiflora Karpathos KA 5 (4) 0.000 ± 0.000 0.711 ± 0.002 1.221 ± 0.004

subsp. mediterranea 37 (24) 0.000 ± 0.000 1.104 ± 0.002 2.265 ± 0.006 Ibiza 12 (9) 0.202 ± 0.002 0.598 ± 0.002 1.227 ± 0.004 CJ 6 (4) 0.202 ± 0.002 0.542 ± 0.001 1.018 ± 0.004 RA 6 (4) 0.000 ± 0.000 0.609 ± 0.002 1.227 ± 0.005 Sardinia 25 (15) 0.000 ± 0.000 1.152 ± 0.002 2.265 ± 0.006 BU 10 (8) 0.000 ± 0.000 1.328 ± 0.003 2.051 ± 0.006 IR 6 (3) 0.202 ± 0.001 0.814 ± 0.003 1.223 ± 0.004 FU 3 (2) 0.000 ± 0.000 0.000 ± 0.000 0.000 ± 0.000 JN 6 (4) 0.000 ± 0.000 0.612 ± 0.002 1.226 ± 0.004

subsp. balearica Majorca ST 8 (4) 0.000 ± 0.002 1.055 ± 0.003 1.634 ± 0.005

5 Rosselló & al. • DNA variation in Cephalaria squamiflora TAXON • 12 October 2009: 12 pp.

revealed (1) conspicuous fixed differences (seven mu- Structural integrity and RNA secondary struc- tations) between Western and Eastern Mediterranean ture. — Cloned sequences from C. squamiflora showed samples; (2) shared ribotypes between accessions from structural landmarks typical of angiosperm ribosomal different islands were extremely rare, and in fact only a ITS spacers (Mayol & Rosselló, 2001; Nieto Feliner & single ribotype was shared between subsp. mediterranea Rosselló, 2007): (1) the presence in the ITS-1 region of (Ibiza) and subsp. balearica ; (3) no diagnostic fixed ITS the universal motif GGCRY-(5n)-GYGYCAAGGAA markers differentiated subsp. balearica and subsp. medi- [GGCGCGATCTGCGCCAAGGAA] that is associated terranea accessions. with part of a hairpin structure and a non-base-paired ITS sequences in other Cephalaria and Dipsaca- sequence in an internal loop structure; (2) the universally ceae species. — Inspection of the aligned sequences conserved pyrimidine-pyrimidine mismatch in helix II in the C. squamiflora group with those available of the of the predicted structure of the ITS-2 RNA transcript Cephalaria-Dipsacus clade showed (1) that intrage- (C U, U U); (3) the predicted secondary models of the nomic polymorphisms (as inferred from base ambi- ITS-2 RNA transcripts conform well with a cruciform guities reported from direct readings) were apparently (four helix) structure; in addition, helix III is the long- rare (C. leucantha : three indeterminations, AJ426523, est stem-loop, as early reported for flowering plants; (4) AJ426524) or absent (C. syriaca: AJ426525, AJ426526; thermodynamically, the secondary structures of the RNA D. mitis: AY236187; D. pilosus : AY290016; D. sylves- transcripts of all sequences were similar and stable, as tris : AJ426527, AJ426528); (2) that the intragenomic inferred from the free energy values (near 98 kcal/mole) polymorphic sites present in the C. squamiflora group of the folded sequences; (5) we noted the presence of the were monomorphic in the other Cephalaria species; and tetranucleotide UGGU, contained near the apex of helix (3) intragenomic polymorphisms in C. leucantha were III of the ITS-2 RNA transcript on the 5′ side. monomorphic in the other congeneric taxa. Phylogenetic analysis. — The parsimony network of cloned ITS sequences is depicted in Fig. 2. Complex patterns were present in the network, including several unresolved loops that are the result of homoplasious sites. With the single exception of clones from subsp. squamiflora (that were grouped together and differ by the remaining Cephalaria accessions by a minimum of eight mutational events), there were no major groupings related to geographic or taxonomic boundaries. Clones from both Majorcan and Sardinian accessions were located on opposite terminals of the network, whereas Ibiza sequences were preferentially located between them. The direct ITS sequence of Cephalaria leucantha, the other species of subgenus Fimbriatocarpus, was linked closely to the Ibiza clones, differing by a minimum of only three mutations. By contrast, C. syriaca was clearly differentiated in the network and differed from other Cephalaria taxa by a minimum of 26 mutations. The joint analysis of Cephalaria ITS sequences using other phylogenetic methods (data not shown) always supported a monophyletic Eastern Mediterranean clade. However, they do not improve relationships between haplotypes from Western Mediterranean. cpDNA sequences. — The trnL intron and coding Fig. 2. Parsimony network of cloned ITS sequences from boundaries ranged from 498 to 505 bp across accessions C. squamiflora. Sequences from C. leucantha (leu) and C. syriaca (syr) have been incorporated for comparative whereas the trnL-trnF region showed a uniform length of purposes. Open symbols, inferred ribotypes not sampled; 438 bp. The concatenated cpDNA region yielded a 949-bp stars, C. squamiflora subsp. squamiflora; black circles, aligned sequence. Only six positions were polymorphic C. squamiflora subsp. balearica; grey circles, C. squa- within the ingroup (trnL intron: 5; trnL-trnF spacer: 1), miflora subsp. mediterranea (ribotypes from Ibiza show resulting in five distinct haplotypes (Table 2). Haplotype a black ring). Size of the symbols is proportional to the number of analyzed ribotypes. Numbers close to shaded 1 was exclusive of Aegean subsp. squamiflora accessions; ribotypes indicate the bootstrap values of the associated haplotype 2 was present in samples of subsp. mediter- clades assessed by maximum parsimony. ranea from Ibiza; haplotypes 3 and 5 were restricted

6 TAXON • 12 October 2009: 12 pp. Rosselló & al. • DNA variation in Cephalaria squamiflora

to subsp. mediterranea (Sardinia), and haplotype 4 was shared between samples of subsp. balearica and some accessions of subsp. mediterranea from Sardinia. The haplotypes of C. leucantha and C. syriaca were divergent from the ingroup. A parsimony network depicting haplotype relationships is shown in Fig. 3. Haplotype 4, the most frequent in the accessions studied, showed an internal position in the network, from which the haplotypes 1, 2, and 5 derived. Phylogenetic analyses using parsimony and distance methods of an expanded dataset including C. leucantha, C. syriaca, Dipsacus sylvestris, D. pilosus, and D. muticus strongly support (BS > 90%) a monophyletic C. squamiflora clade (data not shown), but further resolution of the group was not obtained.

Fig. 3. Parsimony network and geographical distribution (A, Balearic Islands; B, Sardinia; C, Aegean islands) of cp- DNA haplotypes (trnL-trnF region) from C. squamiflora. Se- quences from C. leucantha (leu) and C. syriaca (syr) have been incorporated for comparative purposes. Size of the symbols is proportional to the number of analyzed haplotypes. Numbers refer to haplotypes from Table 2.

Table 2. Polymorphic sites in the trnL-trnF region in C. squamiflora and congeneric taxa.

Haplo– trnL intron trnL-trnF Taxon Accession type 89 128 168 252 307 319 320 332 405 406 445 489 spacer Cephalaria squamiflora subsp. squamiflora KA1AC––ATTC–CC–C KA-2 1 A C – – A T T C – C C – C IK1AC––ATTC–CC–C Cephalaria squamiflora subsp. mediterranea CJ 2 A C – ATAAC G T T C – C C – A RA 2 A C – ATAA C G T T C – C C – A BU3AC––GTTCGTC–A TU4AC––GTTC–CC–A IR 5 A C TAACAAC G T T C – C C – A FU4AC––GTTC–CC–A JN4AC––GTTC–CC–A UR4AC––GTTC–CC–A Cephalaria squamiflora subsp. balearica PM4AC––GTTC–CC–A ST4AC––GTTC–CC–A CB4AC––GTTC–CC–A PA4AC––GTTC–CC–A FO4AC––GTTC–CC–A

C. leucantha AJ427376 A C – – G T T C – C – G ? C. syriaca AJ427377 T A – – G G A G – C – G ?

7 Rosselló & al. • DNA variation in Cephalaria squamiflora TAXON • 12 October 2009: 12 pp.

usually grow in distinct non-rupicolous environments, it DISCUSSION is unlikely that the ITS polymorphism could be gener- Ribosomal ITS paralogs in C. squamiflora. — The ated by hybridization events with other extant species. In ribosomal ITS region is the most frequently used nuclear fact, no other Cephalaria species is known to occur on region in phylogenetic reconstructions (cf. Baldwin & al., Majorca, Ibiza, Crete, Karpathos and Ikaria territories, 1995). Although the ITS region forms part of a multigene and only C. leucantha co-occurs with C. squamiflora in family with thousands of tandem copies clustering in ar- Sardinia. The inspection of ITS sequences reported for rays, and usually several ribosomal loci are present within C. leucantha does not support the view that this species plant genomes, only a single consensus ITS sequence is has contributed to the ITS heterogeneity found in C. sq- usually retrieved after sequencing PCR products from uamiflora. The haploid chromosome complement of C. single individuals. Some authors have cautioned about squamiflora shows secondary constrictions that could be the use of rDNA nucleotide sequences in phylogenetic sites of ribosomal loci in five out of nine chromosomes reconstructions after the finding that major ribosomal (Verlaque, 1985). This relative high number of secondary loci may move within and among chromosomes and that constrictions is not unique to C. squamiflora, but it has their movement may potentially occur via magnification also been reported in other species of Cephalaria and of minor loci consisting of a few rDNA copies (Dubcovsky in most genera of Dipsacaceae (Verlaque, 1985, 1986a, & Dvorak, 1995). However, this violation of the assumed b). Karyotype rearrangements resulting in gene duplica- orthology is usually neglected in standard phylogenetic tions have been hypothesized to have occurred before practices. In addition, if only direct, not cloned, sequences the divergence of several genera of the family (Sca- are routinely generated from an organism it is highly biosa, Cephalaria, Knautia, and Succisa ; Van Treuren likely that much of the intragenomic ribosomal diversity & Bijlsma, 1992), and could be also responsible for the present within organisms are overlooked. This is due to expansion of ribosomal loci in Dipsacaceae. Neverthe- the fact that molecular homogenizing mechanisms of the less, significant intragenomic ribosomal polymorphisms ribosomal multigene family could be relaxed, and from have not been apparently detected in other representatives two to several ribotypes can be present within organisms of Dipsacaceae so far analyzed (Bell, 2004; Caputo & if (1) the rate of mutation among copies is faster than that al., 2004). However, this does not necessarily suggest driving the concerted evolution of the array, (2) duplicate its absence since non-cloned, consensus sequences were ribosomal loci evolve without selective constraints and ac- generated in both studies. Independently of their origin, cumulate mutations, and (3) homeologous loci from other the persistence of ribosomal paralogs in C. squamiflora species are incorporated into a single genome through implies that the molecular forces driving concerted evolu- hybridization-mediated processes. In this study, cloning tion of this multigene family are not fully operating in this procedures have revealed that significant intragenomic narrowly-distributed species. If, as it is here speculated, variation in ribosomal sequences is present in a narrowly the divergent ribosomal families are on loci located in restricted species with a low population size. Levels of separate chromosomes, this could prevent concerted evo- intraindividual sequence divergence (up to 2.1%, Table lution since it has been suggested that the chromosomal 1) strongly suggest that the intragenomic ITS variability location of rDNA loci could have a more substantial im- found should not only be explained by PCR or sequenc- pact than the number of loci on the tempo of concerted ing artifacts. The presence of paralogous ribosomal loci evolution through either unequal crossing-over or gene within the nuclear genome or, alternatively, the occur- conversion (Zhang & Sang, 1999). rence of heterogeneous ribosomal copies within arrays Geographic patterns of molecular variation in should be postulated to occur in C. squamiflora. The C. squamiflora. — The distribution of C. squamiflora, length and G plus C content of the spacers and the cod- restricted to three Western and seven Eastern Mediterra- ing 5.8S region, the presence of all structural landmarks nean islands, is unique among the biogeographic patterns in the primary DNA sequence, the null rate of substitu- shown by the approximately 25.000 plant vascular species tions and deletions in the 5.8S region and in the highly native to the Mediterranean region (Contandriopoulos & conserved motifs, the high thermodynamic stability and Cardona, 1984; Thompson, 2005); a further entity, He- the conserved cruciform foldings of the secondary RNA lichrysum microphyllum, is restricted to a different com- structures suggest that the recovered ITS sequences they bination of Western and Eastern Mediterranean islands. are not pseudogenes. Four major hypotheses can be proposed to explain this Cephalaria squamiflora is diploid (2n = 18) and pattern of distribution. First, insular populations could shows two hypothesized plesiomorphic features, like a x have diverged from a common ancestor by the splitting of = 9 basic chromosome number and an asymmetric karyo- a continuous range through successive vicarious events. type (Verlaque, 1985). Since C. squamiflora is predomi- The scattered populations are thus relicts of a formerly nantly allopatric respect to other species of the genus that widespread species that went to extinction throughout

8 TAXON • 12 October 2009: 12 pp. Rosselló & al. • DNA variation in Cephalaria squamiflora

most of its area, excepting a few islands that acted as species (Altaba, 1998). In view of the similar ecological refuge. Second, during the late Miocene (Messinian pe- requirements of C. squamiflora throughout its range, it riod) closing of the Strait of Gibraltar and the concomi- is hard to envisage how a non-halophyte and strict rupi- tant negative hydric balance caused the desiccation of colous species could successfully spread in the unfavour- the Mediterranean. Connections across land corridors able saline environment. Further, it is difficult to explain could have allowed biotic interchanges between circum- how, if a Messinian dispersal is taken as a general expla- Mediterranean territories through dispersal. Third, nation for the shaping of its area, this has not resulted in some insular populations could have been originated by a more widespread, circum-Mediterranean distribution stochastic events involving long range dispersal (either of the species. by water, wind or birds) of fruits coming from a distant The co-occurrence of shared endemic species, or Eastern or Western gene pool. Lastly, the species could closely related species pairs, restricted to the Balearic not be a monophyletic lineage and the shared morpho- archipelago and Corsica and Sardinia islands is a relevant logical similarities between races could be obtained by biogeographic feature that has been linked to the paleo- convergence, since plants from all populations share the geographic history of the Western Mediterranean basin same rupicolous habitat. Under the first two hypotheses and has been repeatedly invoked as a paradigm for vicari- we would expect molecular markers to be geographi- ance and gradual plant speciation (Cardona & Contandrio- cally structured, and molecular footprints should show poulos, 1979; Contandriopoulos & Cardona, 1984). Usu- sister monophyletic lineages. In contrast, under the third ally, these floristic connections between Corsica, Sardinia hypothesis it would be expected to find an incomplete and the Balearic Islands involve endemic species occur- differentiation among major biogeographic units, and ring in the Eastern (Majorca or Minorca) rather than from paraphyletic lineages would be detected. The fourth hy- the Western (Ibiza and Formentera) Balearic Islands. The pothesis would be supported if polyphyletic evolutionary single puzzling exception to this pattern is C. squamiflora, units were depicted within a phylogenetic framework since subsp. mediterranea is shared between Ibiza and including other congeneric species. Sardinia whereas subsp. balearica is endemic to Majorca. In this work, both cpDNA and ribosomal sequences Post-Messinian terrestrial connections between the Ba- have shown a highly structured pattern of molecular learic Islands (including Ibiza) and the Corso-Sardinian variation composed by sister monophyletic lineages that microplate have not been documented. Accordingly, un- mirror major biogeographic units (Western and Eastern less a long–dispersal event should be invoked, the shared Mediterranean). This East-West vicariance is common in presence of C. squamiflora subsp. mediterranea in both the Mediterranean basin and several climatic and histori- islands could be a likely remnant of the Oligocenic land cal factors have been postulated to contribute to East- connections between Western Mediterranean territories West floristic and species differentiation (Thompson, (Thompson, 2005). 2005). Thus, available molecular evidence supports the The Mediterranean basin has a complex geological hypotheses that (1) vicariance events linked to the geo- history, and a different past configuration and situation logical formation of the present geography, or alterna- of the Western islands has been postulated (Thompson, tively, (2) dispersal across an almost dry Mediterranean 2005, and references therein). The Proto-Ligurian Massif could explain the distribution of the complex. Cephalaria was an Oligocene territory formed by emerged lands now squamiflora fruits, although having developed a short belonging to the Balearic Islands, the Corso-Sardinian fimbriated or scarious involucel-corona (Verlaque, 1984, archipelago, Southern France, the Kabylies, Calabria and 1985), are fully covered by the stiff receptacular bracts NE Spain (Alvarez, 1972, 1976; Alvarez & al., 1974; Van at ripeness (Caputo & Cozzolino, 1995). This favours der Voo, 1993; Westphal & al., 1973, 1976; Speranza & al., dispersal at short distances due to projecting stems and 2002). Successive splitting of the Proto-Ligurian Massif in a catapult mechanism (Verlaque, 1984). Since efficient the Late Oligocene and posterior tectonic movements of adaptations for dispersion over great distances by wind microplates led to the segregation of the Corso-Sardinian dispersal and epizoochory are lacking, it is unlikely microplate and the configuration of Balearic land masses. that long-range dispersal has played a significant role in The splitting of the Western and Eastern Balearic islands shaping its distribution and genetic structure across the took place after the Messinian transgression (about 5.3 Mediterranean. million years ago; Gautier & al., 1994) and allowed their The Messinian salinity crisis has been repeatedly in- isolation from other Western Mediterranean continental voked to explain the current distribution of many plant and insular lands until the present. and animal taxa in the Mediterranean basin (Bocquet & Parsimony network reconstruction of C. squamiflora al., 1978; Kiefer & Bocquet, 1979). However, dispersal genotypes from Western Mediterranean islands is to a across a dry Mediterranean basin that was likely a saline high extent concordant with the postulated paleobiogeo- hot desert would be a prowess beyond the abilities of most graphic relationships based on microplates movements.

9 Rosselló & al. • DNA variation in Cephalaria squamiflora TAXON • 12 October 2009: 12 pp.

The extensive ribosomal paralogy found in Western Medi- terranean samples of C. squamiflora complicates the use ACKNOWLEDGEMENTS of ITS sequences in postulating a sound biogeographic We thank M. Mus, N. Torres and M.A. Conesa who helped scenario for this area. However, some results are relevant in locating Balearic populations of C. squamiflora and effi- for the relationships of subsp. mediterranea accessions ciently assisted with field work. This work was partially funded from Ibiza and Sardinia. First, contrary to expectations, by the project CGL2004-00223/BOS. the unique shared ITS haplotype between islands was from Ibiza (subsp. mediterranea) and Majorcan (subsp. balearica) samples. Second, ITS sequences from subsp. LITERATURE CITED mediterranea from Ibiza are on average more closely related to Majorcan accessions of subsp. balearica (Kimura Allaby, R.G. & Brown, T.A. 2001. Network analysis provides insights into evolution of 5S rDNA arrays in Triticum and 2-parameter distances: 0.95% ± 0.44) than to Sardinian Aegilops. Genetics 157: 1331–1341. sequences of subsp. mediterranea (1.20% ± 0.53). In ad- Altaba, C.R. 1998. Testing vicariance: melanopsid snails and dition, results from the chloroplast genome indicated that Neogene tectonics in the Western Mediterranean. J. Bio- the haplotype present in Ibiza is derived and originated geogr. 25: 541–551. from the ancestral haplotype 4, which is present in both Alvarez, W. 1972. Rotation of Corsica-Sardinia microplate. Majorca and Sardinia islands. Thus, both kinds of mark- Nature 235: 103–105. Alvarez, W. 1976. A former continuation of the Alps. Bull. ers are consistent with the hypothesis that populations Geol. Soc. Amer. 87: 891–896. from Ibiza and Majorca derived from a recent ancestor Alvarez, W., Cocozza, T. & Wezel, F.C. 1974. Fragmentation not shared by Sardinian populations. of alpine orogenic belt by microplate dispersal. Nature Molecular markers agree with a biogeographic sce- 248: 309–314 nario for the Western Mediterranean involving two vicari- Alziar, G. 1995. Généralités sur la flore de l’île de Chypre. ous events. The splitting of the ancestral populations of Quelques données quantitatives. Ecol. Medit. 21: 47–52. C. squamiflora present in the Proto-Ligurian Massif, lead- Arrigoni, P.V. 1978 Le piante endemiche della Sardegna: 12– 18. Boll. Soc. Sarda Sci. Nat. 17: 177–214. ing to the Corso-Sardinian plate and its rafting eastward Bacchetta, G., Brullo, S. & Giusso del Caldo, S. 2008. Cepha- with a counterclockwise rotation (Cherchi & Montadert, laria bigazzii (Dipsacaceae), a new relic species of the 1982; Deino & al., 2001; Rosenbaum & al., 2002; Speranza Cephalaria. Edinburgh J. Bot. 65: 145–155. & al., 2002) is consistent with a first pattern of West-East Baldwin, B.G., Sanderson, M.J., Porter, J.M., Wojcie- vicariance (Sardinian and Balearic evolutionary units). The chowski, M.F., Campbell, C.S. & Donoghue, M.J. shared presence of the putative ancestral cpDNA haplotype 1995. The ITS region of nuclear ribosomal DNA: a valu- 4 in the populations of the Corso-Sardinian plate and the able source of evidence on angiosperm phylogeny. Ann. Missouri. Bot. Gard. 82: 247–277. remaining proto-Balearic Islands suggests a Proto-Ligu- Ballero, M., Cara, S., Marras, G. & Loi, M.C. 2000. La rian origin for the Western Mediterranean populations of flora del Fluminese (Sardegna sud-occidentale). Webbia C. squamiflora. This Balearic-Sardinian split would ex- 55: 65–105. plain why populations located in Majorca and Ibiza shared Bandelt, H.-J., Forster, P., Sykes, B.C. & Richards, M.B. ribosomal paralogs not found in Sardinia. The two private 1995. Mitochondrial portraits of human populations. Ge- cpDNA haplotypes found in Sardinia could have evolved netics 141: 743–753. Bell, C.D. 2004. Preliminary phylogeny of Valerianaceae after the fragmentation of the Proto-Ligurian microplate (Dipsacales) inferred from nuclear and chloroplast DNA under insularity environments. A later vicarious event in- sequence data. Molec. Phylog. Evol. 31: 340–350. volved the SW-NE fragmentation of the Proto-Balearic Bocquet, G., Widler, B. & Kiefer, H. 1978. The Messinian islands leading to Western and Eastern subarchipelagos, Model — a new outlook for the floristics and systematics leading to the persistence of the ancestral cpDNA hap- of the Mediterranean area. Candollea 33: 269–287. lotype in Majorca and the differentiation of the derived Bolòs, O. 1991. Notes florístiques, IV. Butl. Inst. Catalana Hist. haplotype present in Ibiza. If this scenario is correct, the Nat. 59: 145–146. Caputo, P. & Cozzolino, S. 1994. A cladistic analysis of Dip- taxonomic adscription of the populations from Ibiza to sacaceae (Dipsacales). Pl. Syst. Evol. 189: 41–61. subsp. mediterranea should be revisited, since the overall Caputo, P. & Cozzolino, S. 1995. Cladogeny in Dipsacaceae: morphology shared between them and Sardinia samples dispersal and vicariance models. Boll. Soc. Sarda Sci. Nat. is not due to recent ancestry but to convergence. Molecu- 30: 233–243. lar markers are not conclusive to support the splitting of Caputo, P., Cozzolino, S., & Moretti, A. 2004. Molecular C. squamiflora in five species, as recently suggested (Bac- phylogenetics of Dipsacaceae reveals parallel trends in seed dispersal syndromes. Pl. Syst. Evol. 246: 163–175. chetta & al., 2008). Other fast-evolving regions should be Cardona, M.A. & Contandriopoulos, J. 1979. Endemism and analyzed to assess in more detail the phylogeographic and evolution in the islands of the Western Mediterranean. Pp. taxonomic patterns of the Western Mediterranean popula- 133–169 in: Bramwell, D. (ed.), Plants and Islands. Aca- tions. demic Press, London.

10 TAXON • 12 October 2009: 12 pp. Rosselló & al. • DNA variation in Cephalaria squamiflora

Cherchi, A. & Montadert, L. 1982. Oligo-Miocene rift of Sar- internal transcribed spacer 1 of plant nuclear rRNA genes. dinia and the early history of the Western Mediterranean Pl. Molec. Biol. 26: 775–778. basin. Nature 298: 736–739. Mai, J.C. & Coleman, A.W. 1997. The Internal Transcribed Contandriopoulos, J. & Cardona, M.A. 1984. Caractère ori- Spacer 2 exhibits a common secondary structure in green ginal de la flore endémique des Baléares. Bot. Helvet. 94: algae and flowering plants. J. Molec. Evol. 44: 258–271. 101–131. Mathews, D.H., Sabina, J., Zuker, M. & Turner, D.H. 1999. Davis, P.H. 1951. Cliff vegetation in the Eastern Mediterranean. Expanded sequence dependence of thermodynamic pa- J. Ecol. 39: 63–93. rameters provides robust prediction of RNA secondary Deino, A., Gattacceca, J., Rizzo, R. & Montanari, A. 2001. structure. J. Molec. Biol. 288: 911–940. 40Ar/ 39Ar dating and paleomagnetism of the Miocene vol- Maxia, A. & Usai, C. 2004. Una nuova stazione di Cephalaria canic succession of Monte Furru (western Sardinia): im- mediterranea (Viv.) Szabó (Dipsacaceae) nella Sardegna plications for the rotation history of the Corsica-Sardinia sud-occidentale. Atti Soc. Tosc. Sci. Nat. Mem., Ser. B 110: microplate. Geophys. Res. Lett. 28: 3373–3376. 31–33. De Rijk, P. & De Wachter, R. 1997. RnaViz, a program for the Mayol, M. & Rosselló, J.A. 2001. Why nuclear ribosomal DNA visualisation of RNA secondary structure. Nucleic Acids spacers (ITS) tells different histories in Quercus. Molec. Res. 25: 4679–4684. Phylog. Evol. 19: 167–176. Devesa, J.A. 2007. Cephalaria Schrad. ex Roem. & Schult. Médail, F. & Quézel, P. 1999. Biodiversity hotspots in the Pp 276–286 in: Devesa, J.A., Gonzalo, R. & Herrero, A. Mediterranean Basin: setting global conservation priori- (eds.), Flora Iberica, vol. 15. Real Jardín Botánico, Madrid. ties. Cons. Biol. 13: 1510–1513. Doyle, J. & Doyle, J.J. 1987. A rapid DNA isolation procedure Médail, F. & Verlaque, R. 1997. Ecological characteristics for small quantities of fresh leaf tissue. Phytochem. Bull. and rarity of endemic plants from S-E France and Corsica. 19: 11–15. Implications for biodiversity conservation. Biol. Cons. 80: Dubcovsky, J. & Dvorak, J. 1995. Ribosomal RNA multi- 269–281. gene loci: nomads of the Triticeae genomes. Genetics 140: Mort, M.E., Soltis, P.S., Soltis D.E. & Mabry, M. 2000. Com- 1367–1377 parison of three methods for estimating internal support Fielding, J. & Turland, N. 2005. Flowers of Crete. Royal Bo- on phylogenetic trees. Syst. Biol. 49: 160–171. tanic Gardens, Kew. Mus, M., Rosselló, J.A. & Torres, N. 1990. De flora balearica Galbany-Casals, M., Sáez, L. & Benedí, C. 2006. A taxo- adnotationes (6–8). Candollea 45: 75–80. nomic revision of Helichrysum sect. Stoechadina (Astera- Nieto Feliner, G. & Rosselló, J.A. 2007. Better the devil you ceae, Gnaphalieae). Canad. J. Bot. 84: 1203–1232. know? Guidelines for insightful utilization of nrDNA ITS Gamisans, J., Aboucaya, A., Antoine, C. & Olivier, L. 1985. in species-level evolutionary studies in plants. Molec. Quelques données numeriques et chorologiques sur la flore Phylog. Evol. 44: 911–919. vasculaire de la Corse. Candollea 40: 571–582. Pignatti, S. 1977. Note critiche sulla Flora d’Italia. V. Nuovi Gamisans, J. & Jeanmonod, D. 1993. Catalogue des Plantes appunti miscellanei. Giorn. Bot. Ital. 111: 45–61. Vasculaires de la Corse (éd. 2) in Compléments au Pro- Posada, D. & Crandall, K.A. 2001. Intraspecific phylogenet- drome de la Flore Corse, Annexe nº3. Conservatoire et ics: trees grafting into networks. Trends Ecol. Evol. 16: Jardin Botaniques, Genève. 37–45. Gamisans, J. & Marzocchi, J.F. 1996. La Flore Endémique Romo, A. M. 1994. Flores Silvestres de Baleares. Editorial de la Corse. Edisud, Aix en Provence. Rueda, Madrid. Gautier, F., Caluzon, G., Suk, J.P. & Violanti, D. 1994. Age Rosenabum, G., Lister, G.S. & Duboz, C. 2002. Reconstruc- et durée de la crise de salinité Messinienne. Compt. Rend. tion of the tectonic evolution of the western Mediterranean Acad. Sci. Paris, Ser. 3, Sci. Vie. 318: 1103–1109. since the Oligocene. J. Virtual Explor. 8: 107–126. Greuter, W. 1967. Contributiones floristicae austro-aegeae Saitou, N. & Nei, M. 1987. The neighbor-joining method: a 10–12. Candollea 22: 233–253. new method for reconstructing phylogenetic trees. Molec. Illario, T. 1938. Revisione critica delle specie e varietá di piante Biol. Evol. 4: 406–425. vascolari stabilite da Domenico Viviani (1772–1840). Arch. Simmons, R.E., Griffin, M., Griffin, R.E., Marais, E. & Bot. (Forlí) 14: 144. Kolberg, H. 1998. Endemism in Namibia: patterns, proc- Jaeger, J.A., Turner, D.H. & Zuker, M. 1989. Improved esses and predictions. Biodiv. Cons. 7: 513–530. predictions of secondary structures for RNA. Proc. Natl. Speranza, F., Villa, I.M., Sagnotti, L., Florindo, F., Cosen- Acad. Sci. U.S.A. 86: 7706–7710. tino, C., Cipollari, P. & Mattei, M. 2002. Age of the Jaeger, J.A., Turner, D.H. & Zuker, M. 1990. Predicting op- Corsica-Sardinia rotation and Liguro-Provençal basin timal and suboptimal secondary structure for RNA. Meth. spreading: new paleomagnetic and Ar/Ar evidence. Tec- Enzymol. 183: 281–306. tonophysics 347: 231–251. Kiefer, H. & Bocquet, G. 1979. Silene velutina Pourret ex Swofford, D. 2002. PAUP*. Phylogenetic Analysis Using Par- Loiseleur (Caryophyllaceae) — example of a Messinian simony (*and Other Methods), version 4. Sinauer, Sun- destiny. Candollea 34: 459–472. derland. Kimura, M. 1980. A simple method for estimating evolutionary Taberlet, P., Gielly, L., Pautou, G. & Bouvet, J. 1991. Univer- rate of base substitutions through comparative studies of sal primers for amplification of three non-coding regions nucleotide sequences. J. Molec. Evol. 16: 111–120. of chloroplast DNA. Pl. Molec. Biol. 17: 1105–1109. Levin, D.A. 2001. The recurrent origin of plant races and spe- Thompson, J.D. 2005. Plant Evolution in the Mediterranean. cies. Syst. Bot. 26: 197–204. Oxford University Press, Oxford. Liu, J.-S. & Schardl, C.L. 1994. A conserved sequence in Turland, N.J., Chilton, L. & Press, J.R. 1993. Flora of the

11 Rosselló & al. • DNA variation in Cephalaria squamiflora TAXON • 12 October 2009: 12 pp.

Cretan Area: Annotated Checklist & Atlas. The Natural microcontinent, its initial position — paleomagnetic data History Museum, London. and geological fitting. Tectonophysics 30: 141–157 Van der Voo, R. 1993. Paleomagnetism of the Atlantic, Te- White, T.J., Bruns, T., Lee, S. & Taylor, J. 1990. Amplifica- thys, and Iapetus Oceans. Cambridge University Press, tion and direct sequencing of fungal ribosomal RNA genes Cambridge. for phylogenetics. Pp. 315–322 in: Innis, M.A., Gelfand, Van Treuren, R. & Bijlsma, R. 1992. Duplication of the struc- D.H., Sninsky, J.J. & White, T.J. (eds.), PCR Protocols: tural gene for glucosephosphate isomerase and phospho- A Guide to Methods and Applications. Academic Press, gluconate dehydrogenase in Scabiosa columbaria and their New York. phylogenetic implications in the Dipsacaceae. Biochem. Xia, X. & Xie, Z. 2001. DAMBE: Data analysis in molecular Genet. 30: 99–109. biology and evolution. J. Heredity 92: 371–373. Verlaque, R. 1984. A biosystematic and phylogenetic study of Zhang, D. & Sang, T. 1999. Physical mapping of ribosomal the Dipsacaceae. Pp. 307–320 in: Grant, W.F. (ed.), Plant RNA genes in peonies (Paeonia, Paeoniaceae) by fluores- Biosystematics. Academic Press, Toronto. cent in situ hybridization: implications for phylogeny and Verlaque, R. 1985. Étude biosystématique et phylogénétique concerted evolution. Amer. J. Bot. 86: 735–740. des Dipsacaceae. III. Tribus des Knautieae et des Dip- Zhang, W.-H., Chen, Z.-D., Li, J.-H., Chen, H.-B. & Tang, saceae. Rev. Cytol. Biol. Veg. Botaniste 8: 171–243. Y.-C. 2003. Phylogeny of the Dipsacales s.l. based on chlo- Verlaque, R. 1986a. Étude biosystématique et phylogénétique roplast trnL-F and ndhF sequences. Molec. Phylog. Evol. des Dipsacaceae. IV. Tribus des Scabioseae (phyllum nº 1, 26: 176–189. 2, 3). Rev. Cytol. Biol. Veg. Botaniste 9: 5–72. Zuker, M. 1989. On finding all suboptimal foldings of an RNA Verlaque, R. 1986b. Étude biosystématique et phylogénétique molecule. Science 244: 48–52. des Dipsacaceae. V. Tribus des Scabioseae (phyllum nº 4) Zuker, M. 2003. Mfold web server for nucleic acid folding and et conclusion. Rev. Cytol. Biol. Veg. Botaniste 9: 91–176. hybridization prediction. Nucleic Acids Res. 31: 3406–3415. Viviani, D. 1825. Appendix ad Florae Corsicae Prodronum. I. Zuker, M., Mathews, D.H. & Turner, D.H. 1999. Algorithms Typ. Gravier, Genuae. and thermodynamics for RNA secondary structure predic- Westphal, M., Bardon, C., Bossert, A. & Hamzeh, R. 1973. tion: a practical guide. Pp 11–43 in: Barciszewski, J. & Computer fit of Corsica and Sardinia against Southern Clark, B.F.C. (eds.), RNA Biochemistry and Biotechnology. France. Earth Planet. Sci. Lett. 18: 137–140. NATO Science Series 3. High Technology, vol. 70. Kluwer Westphal, M., Orsini, J. & Vellutini, P. 1976. Corsica-Sardinia Academic Publishers, Dordrecht, Boston & London.

Appendix. List of investigated accessions of Cephalaria squamiflora and related taxa, including GenBank codes. Taxon, location (abbreviation), date, collector (herbarium), GenBank accession numbers Cephalaria squamiflora subsp. squamiflora (Sieber) Greuter, Karpathos: Mt. Kali Limni (KA), 02.VII.2002, Brullo & Giusso s.n. (CAT), FJ379630, FJ379646, FJ379693–FJ379696, FJ379741–FJ379744; C. squamiflora subsp. squamiflora, Karpathos, Mt. Kali Limni (KA-2), 22.VIII.2003, Bacchetta & Brullo 449/03 (CAT), FJ379631, FJ379647; C. squamiflora subsp. squamiflora, Ikaria, Plagia, Agios Dimitrios (IK), 29.VIII.2003, Bacchetta & Brullo 594/03 (CAT), FJ379632, FJ379648; C. squamiflora subsp. mediterranea (Viv.) Pignatti, Sardinia, Urzulei, Costa Silana (UR), 15.IX.2003, Bacchetta, Casti & Demurtas 734/03 (CAG), FJ379626, FJ379642; C. squamiflora subsp. mediterranea, Sardinia, Lula, Janna Nurai (JN), 26.V.2002, Bacchetta s.n. (CAG), FJ379627, FJ379643, FJ379685–FJ379690, FJ379733–FJ379738; C. squamiflora subsp. mediterranea, Sardinia, Orgosolo, Mt. Fumai (FU), 25.V.2002, Bacchetta s.n. (CAG), FJ379628, FJ379644, FJ379691–FJ379692, FJ379739–FJ379740; C. squamiflora subsp. mediterranea, Sardinia, Dorgali, Monte Irveri (IR), 10.VIII.2002, Bacchetta & Manconi s.n. (CAG), FJ379619, FJ379635, FJ379661–FJ379666, FJ379709–FJ379714; C. squamiflora subsp. mediterranea, Sardinia, Galtellí, Mt. Tuttavista (TU), 10.VIII.2002, Bacchetta & Manconi s.n. (CAG), FJ379629, FJ379645; C. squamiflora subsp. mediterranea, Sardinia, Buggerru, Is Lisandrus (BU), 10.VII.2002, Bacchetta & Brullo s.n. (CAG, CAT), FJ379620, FJ379636, FJ379667–FJ379676, FJ379715–FJ379724; C. squamiflora subsp. mediterranea, Ibiza, Ses Roques Altes (RA), IX.2002, Torres & Rosselló s.n. (VAL), FJ379617, FJ379633, FJ379649–FJ379654, FJ379697–FJ379702; C. squamiflora subsp. mediterranea, Ibiza, Cap Jueu (CJ), IX.2002, Torres & Rosselló s.n. (VAL), FJ379618, FJ379634, FJ379655– FJ379660, FJ379703–FJ379708; C. squamiflora subsp. balearica (Coss. ex Willk.) Greuter, Mallorca, Puig Major (PM), VII.2002, Conesa & Mus s.n. (VAL), FJ379621, FJ379637; C. squamiflora subsp. balearica, Mallorca, Son Torrella (ST), VII.2002, Conesa & Mus s.n. (VAL), FJ379622, FJ379638, FJ379677–FJ379684, FJ379725–FJ379732; C. squamiflora subsp. balearica, Mallorca, Cavall Bernat (CB), V.2003, Conesa & Mus s.n. (VAL), FJ379623, FJ379639; C. squamiflora subsp. balearica, Mallorca, El Pal (PA), 18.XI.2003, Conesa, Mus & Rosselló s.n. (VAL), FJ379624, FJ379640; C. squamiflora subsp. balearica, Mallorca, Mirador Formentor (FO), 15.II.2004, Conesa, Mus & Rosselló s.n. (VAB), FJ379625, FJ379641; C. leucantha (L.) Roem. & Schult., Caputo & al. (2004), AJ426523, AJ426523, AJ427376; C. syriaca (L.) Roem. & Schult., Caputo & al. (2004), AJ426525, AJ426526, AJ427377; Dipsacus mitis D. Don, Bell & al. unpub., AY236187, AF446977; D. pilosus L., Bell & al., unpub., AY290016, AY290005; D. sativus (L.) Hon., Bell & al. unpub., AJ430982; D. sylvestris Mill., Caputo & al. (2004), AJ426527, AJ426527, AJ427378.