Two additions to the Jacea-Lepteranthus complex : parallel adaptation in the enigmatic species Centaurea subtilis and C . exarata

Centaurea subtilis from south east Italy and C. exarata from south west Iberia were classifi ed in the AcrolophusPhalolepis group and therein in section Maculosae. A molecular survey based on ITS sequence data indicates that both species should rather be placed in the Jacea-Lepteranthus group instead. This placement is consistent with the chromosome number of the two species, which is x = 11 like the rest of species of the Jacea-Lepteranthus group, and differs from the x = 9 of the other taxa included in sect. Maculosae. These results confi rm previous suggestions on the unnaturality of sect. Maculosae. Centaurea exarata and C. subtilis are quite different from the other species of Jacea-Lepteranthus in some striking morphological characters, which we hypothesize to be the result of parallel adaptation to dryer climates. The lack of competitors for pollination might be a good explanation for the partial or even total loss of showy fl owers in these two species.


INTRODUCTION
Centaurea L. is one of the largest genera of the Asteraceae with ca.250 species (Susanna & Garcia-Jacas, 2007).Only thanks to recent studies (Garcia-Jacas et al., 2000;Garcia-Jacas et al., 2006), which widely modifi ed former classifi cations (Dostál, 1976;Wagenitz, 1955), a natural delineation of the genus and an outline of the classifi cation were established.Several species and groups formerly considered as Centaurea are now classifi ed as independent genera, and Centaurea is reduced to three groups.The fi rst two are Acrocentron and Cyanus, and conform well delimited subgenera (Susanna & Garcia-Jacas, 2007).The third group, which encompasses the remaining of the genus, is a wide group of species of manifold morphological characters, the Centaurea jacea group.A comprehensive molecular survey of the Jacea group revealed two large complexes: a fi rst group of taxa mostly with spiny involucral appendages (Eastern and Western Mediterranean clades, cf.Garcia-Jacas et al., 2006), and a second extense group comprising sections Acrolophus (Cass.)DC., Phalolepis (Cass.)DC., Willkommia G. Blanca, Jacea (Mill.)DC., and Lepteranthus (Neck.)DC.The fi rst three sections build up the monophyletic Acrolophus-Phalolepis-Willkommia group clearly separated from the Jacea-Lepteranthus group.These two diverse groups await further classifi cation, although some approaches have been already done, a diffi cult task because of intense hybridization (Suárez-Santiago et al., 2007a, 2007b).
Centaurea subtilis was described by Bertoloni (1853) and it has been traditionally placed in sect.Acrolophus, more precisely into the direct proximity of C. stoebe L. or C. paniculata L. Fiori (1927) included it in sect.Acrolophus as part of his broad concept of C. paniculata together with many other taxa, which are nowadays considered either independent species or subspecies of C. aplolepa Moretti.Dostál (1976)  Centaurea subtilis is a dwarf shrub 10-30 cm high with little branched or simple stems.The whole plant is white tomentose.Lower leaves are 3-5 cm long, 1-pinnatisect into linear laciniae (1×10 mm) ending in a small mucro; middle leaves are smaller, often trifi d, while the uppermost are often simple.The ovoid capitula are solitary and distant from the leaves, and the involucre is 7-11 mm wide.The bracts are green, with outstanding yellowish nerves, and pubescent on the margin.The brownish appendages have an appressed apical spine of 0.5-1 mm, and lateral fi mbriae of 0.5 mm.The fl orets are of about 17 mm length, purple to wine red, the outer ones are sterile and showy (Fig. 1A).The achenes are ca.3 mm long, with a pappus 1-2 mm long, 1/3-1/2 as long as the achene, purple in our collection.The species is a narrow endemic from southeastern Italy and grows in the Puglia and Basilicata regions (Fig. 2; Conti et al., 2005).It occurs only in the southern part of the Gargano peninsula, near Monte San Angelo and near Matera (Pignatti, 1982).The species gives name to two phytosociological societies, Centaureetum subtilis, which settles in dry, sunny limestone rocks at low altitudes (Bianco et al., 1988) and Centaureo subtilis-Thymetum capitati, which grows in garrigues on shallow, calcareous soils (Terzi & D'Amico, 2006).
Centaurea exarata was described by Cosson (1851) and it was placed by Dostál (1976), as mentioned above, in subg.Acrolophus sect.Maculosae, mainly because of the similarity of the involucral appendages to those of other species from this group.
Centaurea exarata is a 30-60 cm high erect perennial.The stem is simple or sparingly branched.The whole plant is arachnoid-pubescent.Leaves are undivided, the lower ones oblong-lanceolate and the upper ones linear-lanceolate and semi-amplexicaul or auriculate.The capitula are solitary with a bract-like leaf on their base.The involucre is ca.14-18 mm long, ovoid.The bracts are adpressed with outstanding nerves and narrowly triangular, reddish-brown, erect, long fi mbriate appendages, without apical spine.The fl orets are purple, all fertile and non-radiant (Fig. 1B).The achenes are 3-4 mm long, with a pappus 1.2-2 mm long, half as long as the achene.Like C. subtilis, it is also a narrow endemic and occurs along the Atlantic coast of occidental Andalusia and central Portugal (Fig. 2; Coutinho, 1939;Franco, 1984;Talavera, 1987).It gives the name to the association Centaureo exaratae-Armerietum gaditanae (Allier & Bresset, 1977).Its ecology is quite different from the preferences of C. subtilis: Centaurea exarata inhabits there sandy beaches, dunes and sandy depressions which get frequently inundated and dry out later.
First doubts on the systematic position of these two species came from their base chromosome number, x = 11 in both cases (Damboldt & Matthäs, 1975;Tornadore & Marcucci, 1988;Valdés-Bermejo, 1980), instead of x = 9 found in members of the Acrolophus-Phalolepis group.Damboldt & Matthäs (1975) already pointed out that C. subtilis had no close relationship with the C. stoebe complex due to the difference in the chromosome number and also due to morphological differences.The doubts were confi rmed for C. exarata by Garcia-Jacas et al. (2006) through molecular data, where it was shown that this species is placed within the clade Jacea-Lepteranthus.The phylogenetic position of C. subtilis was still unknown and further studies were still wanting, which led to the present work.
With the aim of exploring and confi rming the phylogenetic and systematic relationships of C. exarata and C. subtilis, we carried out new chromosome counts, a phylogenetic analysis of molecular data (ITS sequences) and a morphological survey.

Plant material
During summer 2008 vouchers, seeds and leaves of one population of C. subtilis were collected and stored in silica gel.Vouchers are deposited in Herbaria BC and BOZ.

Chromosome counts
To count the chromosome number we used the squash technique on somatic metaphases of root meristems from germinating seeds collected in the wild.
After a pretreatment with 0.002 M 8-hydroxyquinoline at 4ºC for 8 h, the material was fi xed with Carnoy at low temperatures for 24 h.Then it was hydrolysed with 5N HCl at room temperature for A B 1 h.The staining was done with 1% acetic orcein at room temperature for 2 to 12 hours, and the root tips were mounted in 45% acetic acid.Five metaphase plates from different individuals were examined.Preparations were made permanent by freezing with CO 2 , dehydrating in ethanol and mounting in Canada balsam.Digital photographs were taken with an Olympus 3030 camera, mounted on an Olympus microscope U-TV1 X.

Molecular phylogeny study
For the molecular phylogeny, we selected several species from the western and widely-distributed clades showed in Garcia-Jacas et al

DNA extraction, amplification and sequencing strategies
Total genomic DNA was extracted following the CTAB method of Doyle & Doyle (1987) as modifi ed by Cullings (1992) from silica-gel-dried leaves collected in the fi eld.Double-stranded DNA of the ITS region was amplifi ed using the 17SE, forward, and the 26SE, reverse, primers (Sun et al., 1994).The PCR was executed with the following conditions: 2 min denaturing at 94ºC, followed by 30 cycles of 94ºC denaturing for 1 min 30 s, 57ºC annealing for 2 min and 72ºC extension for 3 min, with an additional 15 min at 72ºC.Double-stranded PCR products were purifi ed with QIAquick® Purifi cation Kit (Qiagen Inc., Valencia, CA, USA) and sequenced with the primers 17SE as forward primer, and 26SE as reverse.Direct sequencing of the amplifi ed DNA segments was performed with a "Big Dye® Terminator v3.1 kit" (Applied Biosystems, Foster City, CA, USA), following the protocol recommended by the manufacturer.Nucleotide sequencing was carried out at the "Serveis Científi co-Tècnics" of the University of Barcelona on an ABI PRISM 3700 DNA analyzer (Applied Biosystems).

Phylogenetic analyses
Nucleotide sequences were edited using Chromas 2.0 (Technelysium Pty. Ltd., Tewantin, Australia) and aligned visually by sequential pairwise comparison (Swofford & Olsen, 1990).Data matrices are available on request from the corresponding author.Parsimony analysis of the ITS dataset involved heuristic searches conducted with PAUP 4.0b10 (Swofford, 2002) using TBR branch swapping with character states specifi ed as unordered and unweighted.The indels were coded as missing data.All most-parsimonious trees (MPTs) were saved.To locate other potential islands of MPTs (Maddison, 1991), we performed 1000 replications with random taxon addition, also with TBR branch swapping.Bootstrap analyses (BS) (Felsenstein, 1985) were performed using 1000 replicates of heuristic search with the default options.For the strict consensus tree, consistency index (CI) and retention index (RI) are given, excluding uninformative characters.Bayesian inference estimation of the ITS dataset was calculated using MrBayes 3.1.2(Huelsenbeck & Ronquist, 2001;Ronquist & Huelsenbeck, 2003).The best available model of molecular evolution required for Bayesian estimations of phylogeny was selected using hierarchical likelihood ratio tests (hLRT) and Akaike information criteria (AIC) as implemented in the software MrModeltest 2.2 (Nylander, 2004), which considers only nucleotide substitution models that are currently implemented in PAUP and MrBayes.
The symmetrical model, with equal base frequencies with variable sites assumed to follow a discrete gamma distribution SYM+G (Zharkikh, 1994), was selected as the best-fi t model of nucleotide substitution.The Bayesian inference analyses were initiated with random starting trees and were run for 1 × 10 6 generations.Four Markov chains were run using the Markov Chain Monte Carlo (MCMC) principles to sample trees.We saved one out of every 100 generations, resulting in 10000 sample trees.Data from the fi rst 1000 generations were discarded as the "burn-in" period, after confi rming that likelihood values had stabilized prior to the 1000 th generation.Internodes with posterior probabilities ≥ 95% were considered statistically signifi cant.

Morphological survey
To prepare Fig. 3, the middle involucral bracts of fi ve different species, all taken from herbarium specimens (Table 1), were glued onto a black paper and scanned using a scanner Epson Perfection 4990 Photo.The background of the resulting image was underlain in black using Adobe-Photoshop 7.0.

Chromosome counts
Centaurea subtilis is a diploid with 2n = 2x = 22 chromosomes (Fig. 4).The basic chromosome number is x = 11.This result is coincident with previous reports (see introduction and references therein).

Molecular phylogeny
The results of both analyses are commented together because the topology of the trees produced by parsimony and Bayesian approaches was coincident.The parsimony analysis resulted in 30 equally parsimonious trees of 76 steps.Descriptive statistics: the consistency index (CI) was 0.750 and the retention index (RI) was 0.932, excluding uninformative characters.
Figure 5 shows the Bayesian phylogram with the addition of Bayesian posterior probabilities (PP) above branches, and parsimony bootstrap values (BS) below branches.The results are coincident with the ones in Garcia-Jacas et al. (2006) for the clade that was named "widely distributed clade".Within this group, our analyses showed three subclades: one which comprised species of the Acrolophus-Phalolepis sections (BS = 91%, PP = 1.00), a second one which comprised species of sect.Willkommia (BS = 97%, PP = 1.00), and a third subclade which comprised the taxa from the Jacea-Lepteranthus group plus Centaurea subtilis with high support values (BS = 93%, PP = 1.00).
By showing C. subtilis within the Jacea-Lepteranthus group, our results confi rm that sect.Maculosae is an artifi cial assemblage.After C. exarata (moved to the Jacea-Lepteranthus group by Garcia-Jacas et al., 2006), this is the second species of sect.Maculosae which belongs to a different clade and is placed within the Jacea-Lepteranthus group.Most of the remaining species of sect.Maculosae are part of the Acrolophus-Phalolepis group as verified by previous molecular studies [Centaurea corymbosa (Garcia-Jacas et al., 2006), C. fi liformis (Mameli, 2008), C. stoebe and C. triniifolia (Ochsmann, 2000)], and they do not form a natural group in any of these studies.The delimitation of Maculosae from some other sections, namely sections Arenariae Dostál, Paniculatae Dostál and Dissectae Dostál, is doubtful and the whole taxonomic treatment by Dostál (1976) for the Jacea group needs a deep revision.

Morphology
Centaurea subtilis shows intriguing morphological paralellisms with C. exarata.Their appendages of the involucral bracts are very similar (Fig. 3).When compared with other members of the Jacea-Lepteranthus group, the outer showy fl owers are much smaller in C. subtilis, while they are totally reduced in C. exarata (Fig. 1).Most members of the Jacea-Lepteranthus group have broad, undivided, mesophyllic, mostly glabrous or sparsely hirsute leaves, whereas C. subtilis and C. exarata have narrow, tomentose, more Two additions to the Jacea-Lepteranthus complex: parallel adaptation in the enigmatic species Centaurea subtilis and C. exarata xerothermic leaves.Centaurea subtilis is hitherto the only species of the group with divided leaves.All these morphological characters, which in part led to their wrong taxonomic classifi cation in former times, can be considered as adaptations to dryer micro-and macroclimatic environments.The distributional centre of the Jacea-Lepteranthus group are the mountain ranges in the northern Mediterranean region, especially the southern Alps, the northern Balkanic mountains, the Pyrenees, the Apennines and the mountains of central Iberia.Its members usually grow in mountain meadows with deep soil and good water supply or in some cases in well water-provided crevices.In contrast, C. subtilis and C. exarata grow in drier habitats, like calcareous rocks, garrigues or dunes, which share the tendency to suffer water defi ciencies, induced by their local ecology on one hand and by the mediterranean climate on the other.Strong droughts can occur in late spring and early summer, at a time when this two species have still not concluded their reproduction cycle.Thus, it was evolutionary essential to develop adaptations to avoid water loss.Ecological differences of these two species relative to other taxa of the Jacea-Lepteranthus group might be a good explanation for their morphological divergence.
This adaptation to a drier climate would also contribute to explain the secondary loss of the showy peripheral sterile fl orets in both species, much reduced in C. subtilis and totally absent in C. exarata.According to Lack (1976), the presence of radiant peripheral fl orets in the related C. nigra L. is directly related to competition for pollinators (mainly Bombus sp., cf.Hegland & Totland, 2004) with a larger species of Centaurea with similar ecological preferences, C. scabiosa L. Radiate heads in C. nigra predominate in populations which grow in presence of C. scabiosa, while non-radiated heads predominate in stands of C. nigra alone.According to Font et al. (2009), C. scabiosa sensu latissimo is an Euro-Siberian species and thereafter totally absent from the areas where C. exarata and C. subtilis grow.There is a species of Centaurea with very large radiate heads which grows in the vicinity of C. exarata, C. polyacantha Willd., but its fl owering period is much earlier (Talavera, 1987).

Concluding remarks
This study emphasizes once again the importance of chromosome numbers as a predictor of the systematic relationships in the genus Centaurea, especially in the Jacea group.In the case of a species where the base chromosome number does not correspond with the one of its systematic group, caution is required: it may be a sign of a mistake in its taxonomic classifi cation, as demonstrated by molecular phylogeny analysis in the case of C. exarata and C. subtilis.

Figure 5 .
Figure 5. Phylogram obtained from the Bayesian analysis of ITS sequences.Numbers above branches indicate Bayesiancredibility values (PP); numbers below branches indicate bootstrap values (BS) from parsimony analysis.

Table 1 .
Two additions to the Jacea-Lepteranthus complex: parallel adaptation in the enigmatic species Centaurea subtilis and C. exarata Origin of materials, herbaria where vouchers are deposited and GenBank accession numbers (the new sequence is boldfaced).