Introduction

It is well known that the genus Centaurea L. presents taxonomic difficulties. In Central Europe, the examples of intricate groups in which the definition of taxa has been controversial for a long time and which have been rearranged recently based on modern morphometric, karyological and molecular approaches include the C. stoebe group (Španiel et al. 2008; Mráz et al. 2011) and the C. triumfetti group (Olšavská et al. 2009, 2011). However, C. sect. Jacea (= C. subgen. Jacea as delimited in Flora Europaea; Dostál 1976) accounts for the majority of Central European taxa. The section includes several species complexes (aggregates) that can be relatively easily delimited based on the shape of leaves and the size and appearance of capitula (e.g. C. jacea agg., C. nigrescens agg., C. nigra agg., and C. phrygia agg.). The base chromosome number is uniform within the section (x = 11) and all of the aggregates include both diploid (2n = 2x = 22) and tetraploid (2n = 4x = 44) taxa (e.g. Dostál 1976; Koutecký 2007; Koutecký et al. 2012). An autotetraploid origin causing poor morphological differentiation from diploids has been reported for C. jacea (Hardy et al. 2001).

The major causes of taxonomic and determination problems in the section Jacea are hybridization and introgression. Both crossing experiments and field studies (Gardou 1972; Hardy et al. 2000, 2001; Koutecký et al. 2011, 2012) have consistently shown that taxa of the same ploidy level (regardless of whether diploid or tetraploid) hybridize freely to produce fertile hybrids, but hybridization between plants of different ploidy levels is rare. The interploidy hybrids are more often tetraploid rather than triploid (intermediate). The tetraploid hybrids are fertile and capable of backcrossing with tetraploid parents (Koutecký et al. 2011). The morphological variability of hybrids is enormous, especially with respect to the key determination character, the shape of appendages of involucral bracts (e.g. Marsden-Jones and Turrill 1954; Saarisalo-Taubert 1966; Vanderhoeven et al. 2002; Koutecký et al. 2011). Some hybrid morphotypes have even been treated as autonomous (nonhybrid) taxa (Koutecký 2009). Hybridization and introgression result in extensive hybrid zones where variable intermediate populations prevail over parental taxa (Koutecký 2007).

The C. jacea aggregate (henceforth referred as “C. jacea agg.”, while “C. jacea” refers only to the species C. jacea L.) ranks among the most intricate groups within C. sect. Jacea. The aggregate is easily recognized by two morphological characters: (1) appendages of all involucral bracts (i.e. scarious, brown or black structures on the top of the green parts of the bracts) are rounded and entire or only finely denticulate on margin, while they are toothed or fimbriate in other taxa of the section; (2) achenes lack any remains of a pappus. In this circumscription, which corresponds to C. subgen. Jacea sect. Jacea in Flora Europaea (Dostál 1976), the C. jacea agg. involves more than ten microspecies throughout Europe (Dostál 1976). Due to a lack of obvious determination characters and generally unresolved taxonomy, these taxa are often treated as subspecies of a broadly defined C. jacea (more recent, e.g. Wagenitz 1987; Ochsmann 1998; Kubát et al. 2002; Štěpánek and Koutecký 2004; Jäger and Werner 2005; Fischer et al. 2008). In some of these works even two taxa with fimbriate appendages are included in C. jacea due to frequent hybridization and intermediate morphotypes [C. jacea subsp. macroptilon (Borbás) Hayek, C. jacea subsp. oxylepis (Wimm. et Grab.) Hayek]. However, for consistency with the generally accepted concept of other aggregates within the section Jacea (e.g. C. phrygia agg.), and to avoid impractical use of a nothosubspecific rank for “intermediate morphotypes”, I use a narrow species delimitation in this paper, following e.g. Flora Europaea (Dostál 1976).

The C. jacea agg. includes both diploids (2n = 22) and tetraploids (2n = 44). The former seem to be confined to southern Europe, while tetraploids occur throughout the range of the group. Diploid counts for C. jacea were reported by Gardou (1972) for France (Eastern Pyrenees, Alpes-Maritimes), Italy (Liguria, Southern Alps, Padova, Northern Apennines) and Croatia (the island of Krk), and by Kuzmanov et al. (1986) for Bulgaria. The other taxa of the C. jacea agg. with reported diploid chromosome counts include an unclear taxon C. dracunculifolia Dufour (Dostál 1976), C. rocheliana (Heuff.) Dostál (Bogdan et al. in Löve 1979), C. weldeniana (one record from the island of Krk, Croatia; Lovrić in Löve 1982), and C. pannonica (Heuff.) Simonk. [= C. jacea subsp. angustifolia (DC.) Gremli]. In the last species a diploid count was reported in Flora Europaea (Dostál 1976; most probably based on records for C. jacea subsp. angustifolia from Gardou 1972, see below) and for Bavaria (Lippert and Heubl 1988); the latter is the only record of diploids in Central Europe. However, all these counts from the literature should be understood as C. jacea agg. due to determination problems and due to imprecise or incomplete citations from the primary sources. For instance, in the most important work of Gardou (1972), the broad taxonomic concept was used and only one species, C. jacea, was recognized, with two subspecies, subsp. jacea and subsp. angustifolia. Each subspecies comprised several varieties, including “C. j. subsp. angustifolia var. bracteata” [= C. bracteata Scop.; C. jacea subsp. gaudinii (Boiss. & Reut.) Gremli] and “C. j. subsp. angustifolia var. weldeniana” [= C. weldeniana Rchb.; C. jacea subsp. weldeniana (Rchb.) Greuter]. Unfortunately, the exact determination of subspecies/varieties is lacking for the majority of localities. It is probable that (1) most of the records of Gardou (1972) for France and Italy belong to C. bracteata, which is common there and comprises at least some diploids (Koutecký et al., unpublished data), (2) both the records of Gardou (1972) for Croatia belong to C. weldeniana (one of the localities is marked as “C. jacea subsp. angustifolia var. weldeniana” on p. 318), and (3) all these data are cited by later authors simply as C. jacea subsp. angustifolia, or even C. jacea, without examining the lower ranks used by Gardou (1972).

Numerous tetraploid counts have been reported for C. jacea and C. pannonica for Central Europe, e.g. Albers (1998), Dobeš and Vitek (2000), Marhold et al. (2007), Koutecký et al. (2011). Tetraploid counts have also been reported for C. bracteata for France (Gardou 1972, p. 355 as “C. jacea ssp. angustifolia var. bracteata”) and Croatia (Lovrić in Löve 1982) and for Croatian C. weldeniana var. balcanica (Lovrić in Löve 1982), an unclear taxon intermediate between C. weldeniana and C. pannonica (Dostál 1976). Single individuals of higher ploidy levels (pentaploid, hexaploid) are exceptionally found within tetraploid populations as a result of rare polyploidization events, and triploids rarely occur as a result of hybridization between diploids and tetraploids (Koutecký et al. 2011).

In 2005, I discovered a population near the village of Gießhübl near Wien, Austria (see Table 1 for the exact location), that is markedly different from other Central European populations of the C. jacea agg. It was preliminarily determined as C. weldeniana Rchb., which has not previously been known to occur in Austria. Interestingly, there is one taxon that has been generally overlooked and was described in the same locality at the beginning of twentieth century, namely C. argyrolepis Hayek. Based on the original description (Hayek 1901), it belongs to the C. jacea agg. and may be identical to the recently found population. The aim of the work reported here was therefore (a) to inspect the morphological variation of the studied population and of C. weldeniana to confirm the determination and demonstrate differences from other Central European taxa of the C. jacea agg., (b) to examine potential hybridization with C. jacea, which also occurs at the locality, (c) to provide karyological and genome size data for C. weldeniana, and (d) to revise the original material of C. argyrolepis.

Table 1 Studied populations of C. weldeniana, C. jacea and C. pannonica
Full size table

Materials and methods

Field sampling

All available (i.e. 15) individuals of putative C. weldeniana and three individuals with a morphology intermediate between C. weldeniana and C. jacea (putative hybrids) were sampled for morphometric analysis during September 2010 at the locality near Gießhübl. Only undamaged stems with a fully developed terminal capitulum were collected and only one stem was sampled from each tussock (genet). An additional 11 individuals were sampled for flow cytometry only (damaged or sterile individuals). An equivalent number of putative C. jacea individuals were collected at the same locality for morphometric analysis. For comparison, material of C. weldeniana from another three sites in Croatia and Bosnia was included (11 individuals altogether); no other karyologically/cytometrically confirmed material of C. weldeniana was available. The morphometric analysis was supplemented with a subset of plants collected for morphometric study of C. jacea and C. pannonica from the Czech Republic, Slovakia and Ukraine (Koutecký 2008, and unpublished data). To avoid strongly unbalanced numbers of samples, only five individuals were chosen at random from each of 34 additional populations of the C. jacea agg. All localities are listed in Table 1. Voucher specimens are stored in the herbarium CBFS.

Flow cytometry

DNA ploidy level and genome size were determined using flow cytometry. The method was as described by Koutecký et al. (2012) in detail. Briefly, fresh leaves were chopped in Otto I buffer with an internal standard Glycine max ‘Polanka’ (2C = 2.50 pg; Doležel and Greilhuber 2010), and stained after about 1 min with Otto II buffer containing 2-mercaptoethanol (2 μl/ml) and a fluorescent dye. For ploidy level estimation, DAPI fluorochrome (4 μg/ml) was used and samples were analysed using a Partec PA II flow cytometer equipped with a mercury arc lamp. It was possible to use bulked samples of three to ten plants due to high-resolution histograms and absence of endopolyploidy. From two populations of C. weldeniana from Bosnia and Croatia, silica gel-dried leaves were used and only one individual per sample was analysed. For genome size measurements, the same internal standard and the same method of sample preparation from fresh leaves was used, only replacing DAPI by propidium iodide (50 μg/ml) and RNase IIa (50 μg/ml). The genome size was determined using a Partec CyFlow SL flow cytometer equipped with a 532-nm (green) diode-pumped solid-state laser (100 mW output). One individual per sample was measured and the fluorescence intensity of 5,000 particles was recorded. Each individual was analysed three times on three different days and the average value used as the genome size; the repeated measurements of a sample were always consistent and the range did not exceed 2 % of the average value. The fluorescence histograms were processed using FloMax 2.6 software (Partec, Germany).

Chromosome counting

Chromosomes of the putative C. weldeniana from the Gießhübl population were counted to calibrate the results of flow cytometry. Chromosomes in the apical root meristems of seedlings that were germinated from field-collected achenes were counted. The protocol of Španiel et al. (2008) was used, with minor modifications. Seedlings were pretreated with a saturated water solution containing 0.002 M 8-hydroxyquinoline for 10 h at 4 °C, fixed in a mixture of ethanol and acetic acid (3:1) for 24 h at 4 °C and stored in 70 % ethanol at 4 °C. Maceration lasted about 5 min in a mixture of ethanol and hydrochloric acid (1:1). The apical part of a root was then squashed using a cellophane square and stained for 45 min in 10 % Giemsa solution in 0.2 M sodium phosphate buffer, pH 7.2. Three samples were analysed and at least two mitoses per plant were studied.

Morphometric analysis

Morphometric analyses were performed (a) to demonstrate differentiation of the newly recognized population of C. weldeniana from other Central European material of the C. jacea agg. and (b) to examine hybridization between C. weldeniana and C. jacea. For this analysis, 29 characters were used, including 15 quantitative characters, seven ratios and seven binary characters (Table 2). Characters such as branching of the stem, shape of leaves and dimensions of the involucre were measured on fresh plants. Plants were then dried and used to measure the other characters.

Table 2 List of morphological characters and transformations used
Full size table

Basic statistical measures (mean, median, maximum and minimum values, quartiles, five and 95 percentiles and standard deviation) were computed for each population. The normality of the distribution of each character was examined and transformations were applied to the characters which markedly deviated from a normal distribution (Table 2). Pearson and Spearman correlation coefficients were calculated for pairs of characters for each taxon and for the whole data set to study relationships between the characters. For principal component analyses (PCA), the data were standardized to have zero mean and unit standard deviation (PCA based on correlation matrix).

Two PCA analyses were performed. In the first (PCA 1), differences between C. weldeniana and other Central European material were evaluated using individuals as operational taxonomic units. In the second (PCA 2), the Gießhübl population was analysed in detail to evaluate differences between C. weldeniana and the local population of C. jacea, and especially to identify potential hybrids with intermediate morphology.

The statistical analyses were performed using Statistica 9 (StatSoft 2010) and Canoco for Windows 4.5 software (ter Braak and Šmilauer 2002).

Herbarium material

Centaurea material in the herbaria W and WU was revised. Centaurea argyrolepis was described by Hayek (1901) based mainly on collections of Müllner, which are currently stored in these herbaria.

Results

Ploidy levels and genome size

Flow cytometry revealed that all studied individuals of C. weldeniana were DNA diploid. There was negligible variation in the relative fluorescence among populations for which fresh leaves were available and some variation among silica gel-dried samples. The genome size of C. weldeniana was 2C = 2.09 pg. All studied individuals of C. jacea were DNA tetraploid. The putative hybrids C. jacea × C. weldeniana were all DNA tetraploid. The genome sizes are shown in Table 3, and typical fluorescence histograms are presented in Fig. 1.

Table 3 Genome sizes of C. weldeniana and C. jacea
Full size table
Fig. 1
figure 1

Representative fluorescence histograms of Centaurea weldeniana and its hybrids. Glycine max ‘Polanka’ was used as the internal standard. a Diploid C. weldeniana (DAPI stain). b Diploid C. weldeniana (propidium iodide stain). c Tetraploid hybrid C. weldeniana × C. jacea (DAPI stain). d Simultaneous analysis of C. weldeniana and C. phrygia L. corroborating a difference of about 6 % in the genome size (DAPI stain)

Full size image

Chromosome counting confirmed the results of flow cytometry. All three samples of C. weldeniana from the Gießhübl population were diploid, 2n = 22 (Fig. 2).

Fig. 2
figure 2

Somatic metaphase of Centaurea weldeniana, 2n = 22. Scale bar 5 μm

Full size image

Morphometric analysis

There were only slight differences between the Spearman and Pearson correlation coefficients. No highly correlated characters (r > |0.95|) were found and all characters were used in the multivariate analyses except for AC1 that was almost invariable.

PCA 1 revealed differentiation of C. weldeniana from Central European populations of the C. jacea agg. Centaurea weldeniana forms a compact cloud on the periphery of the ordination space, which does not overlap with other taxa, although it is not separated by any obvious gap. Individuals from the Gießhübl population are intermingled with C. weldeniana from the Balkans (Fig. 3). The cloud of C. weldeniana is also slightly separated along the third ordination axis (not shown). Centaurea jacea and C. pannonica markedly overlap (however, separation of these two taxa is beyond the scope of the present paper, and will be discussed in a separate contribution). The characters that most contribute to the separation of C. weldeniana are shape of leaves and bracts on branches (smaller width and higher length/width ratio in C. weldeniana; characters LW, LBW, LLW and LBLW), presence of greyish hairs (more frequent in C. weldeniana; characters HM and HU), size of capitula (smaller in C. weldeniana; characters IW and IL) and branching angle (higher in C. weldeniana; character BA) (Table 4).

Fig. 3
figure 3

PCA of Central European populations of the Centaurea jacea agg. The first and the second ordination axes explain 21.6 % and 13.9 % of the variation, respectively. a Position of individuals. Black diamonds C. weldeniana, Gießhübl; black triangles other populations of C. weldeniana; empty circles C. jacea; empty triangles C. pannonica; grey diamonds putative hybrids C. jacea × C. weldeniana. b Correlation of characters with ordination axes

Full size image
Table 4 Variation of the characters of Centaurea weldeniana and Central European populations of C. jacea and C. pannonica
Full size table

Within the Gießhübl population, diploids (C. weldeniana) and tetraploids (C. jacea) were clearly separated by PCA 2 (Fig. 4). In addition to characters indicated by PCA 1, the two species were also separated by size of appendages (AL, AW), total stem height (ST), branching of lateral branches (BB), length of internodes on branches (BLLF) and length of leaves and bracts (LL, LBL). Concerning the tetraploid individuals tentatively marked as hybrids, the hybrid origin of at least two of them was confirmed—one was very close to C. weldeniana, one was intermediate and one was on the margin of the cloud of C. jacea (Fig. 4).

Fig. 4
figure 4

PCA of the Gießhübl population. The first and the second ordination axes explain 38.2 % and 13.1 % of variation, respectively. a Position of individuals. Black diamonds Centaurea weldeniana (diploid); empty circles C. jacea (tetraploid); crosses putative hybrids C. jacea × C. weldeniana (tetraploid). b Correlation of characters with ordination axes

Full size image

Centaurea argyrolepis

The original material of C. argyrolepis comprises at least eight herbarium sheets that conform with the protologue (Hayek 1901): accession numbers W 1912-10588, W 1912-10589, W 1912-15090, W 1912-15123, W 1912-21094, W 1912-21101, WU 43222, and one sheet in W without an accession number (“Centaurea gaudinii Boiss. et Reut. Tirolerhof b. Gießhübl” 28.8.1900 leg. A. Teyber). All the material is homogeneous and identical to the recently found population, and can be classified as C. weldeniana.

Discussion

Determination

Morphometric analysis and karyological data confirmed differences between the Gießhübl population from other Central European representatives of the C. jacea agg. The key features were conspicuous greyish indumentum, linear leaves and bracts, relatively long, straight (virgate) and patent branches and relatively large and pale appendages of involucral bracts. The population was determined as C. weldeniana and was not different from Croatian and Bosnian material of C. weldeniana in multivariate morphometric analysis. The determination was also confirmed by karyological data. All studied populations assigned to C. weldeniana were diploid. In contrast, all Central European populations of C. jacea and all but one population of C. pannonica studied so far were tetraploid (see Introduction for references and Koutecký et al., unpublished data). Centaurea weldeniana is a Balkan taxon whose native distribution includes Italy, Croatia, Serbia, Bosnia, Montenegro, Albania and Greece (Dostál 1976; Greuter 2006–2009).

There is one other putatively diploid taxon (see Introduction) from the C. jacea agg. native to Austria: C. bracteata (=C. jacea subsp. gaudinii). Its distribution includes France, Switzerland, Austria, Italy, Slovenia and Croatia, and north-west Africa (Morocco, Libya and Tunisia) (Dostál 1976, Greuter 2006–2009). In Austria, it occurs only in the south-west in western Kärnten and Osttirol (Fischer et al. 2008), about as far from the Gießhübl locality as are the nearest populations of C. weldeniana in Italy and Croatia. Interestingly, the Gießhübl population was determined as C. bracteata (under the name C. gaudinii) when found for the first time by M. F. Müllner (1888) (see below). Also Hayek (1901), who described it as a new taxon C. argyrolepis, regarded it as close to, but different from, C. bracteata because C. bracteata has somewhat larger appendages of involucral brats than is usual in C. weldeniana (Table 4). However, the width of appendages 4.6–6.4 mm is still considerably smaller than reported for C. bracteata (6–8 mm), is fully within the range (4–)5–7 mm reported for C. weldeniana and is also within the variation range of “small-appendage” taxa such as C. jacea and C. pannonica (Table 4). Moreover, the Gießhübl population differs from C. bracteata also in other characters, such as markedly smaller capitula (0.9–1.1 cm wide; while 1.2–2.0 cm is reported for C. bracteata) and branching about in the middle of the stem with long branches (while C. bracteata is short-branched in the upper part of the stem; for measurements cited here see Dostál 1976 and Pignatti 1982).

Centaurea weldeniana is not known to occur in Austria. The (re)discovery of the present population implies addition of the new, probably adventive species to the flora of Austria.

History of the population

Centaurea weldeniana is most probably not native in Austria. The population is isolated. There are no records of C. weldeniana from Austria and of C. bracteata from the eastern half of Austria nor any specimens of these taxa in the studied herbaria (especially W and WU and also BP, which contains material from the formerly Hungarian part of contemporary Austria). The locality lies in an intensive agricultural landscape, where survival of a “relic” population is improbable. Interestingly, the adventive origin of this population was suggested by Hayek (1901), who mentions three large dairy farms that used imported hay.

The population of C. weldeniana was discovered “auf einer Wiese zwischen Perchtoldsdorf and Giesshübel” by Müllner in August 1886 and determined as C. gaudinii [= C. bracteata] (Müllner 1888). The same population is also cited by Beck (1893) as C. gaudinii, by Hayek (1901) as C. argyrolepis and by Hayek (1917) as C. jacea subvar. argyrolepis. The locality (both in the literature and on herbarium labels) was referred to as between Perchtoldsdorf and Gießhübl or near the former, or is narrowed to the village of Tirolerhof between them. Since than, there are no records until 2005 when I discovered the population at the same site or apparently very close to the original locality. Hence, it is most probable that the species has survived there for at least 125 years and was merely not recorded for a long period. The other possibility (repeated dispersal from far localities to the same place and nowhere else) is rather improbable.

Genome size

The genome size is reported for the first time for any diploid taxon of the C. jacea agg.. The value of 2.09 pg is fully within the range for diploid members of C. sect. Jacea found by Bancheva and Greilhuber (2006), Dydak et al. (2009) and Koutecký et al. (2012). Interestingly, the value for the other diploid member of the C. jacea agg., C. bracteata, is the same (Koutecký et al., unpublished data) and is different from diploids of the C. phrygia agg. (Fig. 1d), which suggests that the genome size may be a useful marker for internal classification of C. sect. Jacea. Compared to published values for related tetraploids (Bancheva and Greilhuber 2006, Dydak et al. 2009), the monoploid genome size (Cx-value) of tetraploid C. jacea is about 0.95 of that of diploid C. weldeniana, which is similar to the mean difference between diploid and tetraploid taxa of the sect. Jacea found by Bancheva and Greilhuber (2006). However, such comparisons should be viewed with caution, since there may be considerable differences among individual studies using different methodologies (Feulgen densitometry vs. flow cytometry) and different internal standards. Indeed, the values of relative fluorescence of tetraploid C. jacea in the present study were 2.01 times higher than those of the diploid C. weldeniana (Table 3).

Hybridization

The persistence of an isolated diploid population among abundant tetraploids for at least 125 years is a unique natural experiment. Although there is a certain phenological shift (C. weldeniana starting to flower about 3 weeks after C. jacea), the flowering times of the two species overlap widely. Theoretically, the population of C. weldeniana should suffer from a severe minority cytotype disadvantage (Levin 1975) due to crossing with more frequent tetraploids. The hybrids, irrespective of whether they are sterile triploids or fertile tetraploids (Koutecký et al. 2011), cannot contribute to (sexual) reproduction of the diploids. Although C. sect. Jacea plants are perennials and can survive up to 10 years under cultivation (Koutecký, personal observation), there must have been several generations through sexual reproduction since 1886. Survival of this population is thus a good example of the interploidy reproductive barrier and confirms the results of previous studies of heteroploid hybridization in Centaurea sect. Jacea (Gardou 1972; Hardy et al. 2000, 2001; Koutecký et al. 2011, 2012). Specifically, it confirms the results of pollinations with a mixture of pollen, in which the vast majority of progeny are of the same ploidy level as the maternal plant and intercytotype hybrids are almost absent (Koutecký et al. 2011). Although in a minority, the frequency of diploid–diploid crosses has obviously been sufficient to prevent massive hybridization and sustain the population of C. weldeniana.

However, due to massive influx of tetraploid pollen to the diploids, the barrier is broken occasionally. Morphometric analysis revealed at least two tetraploid individuals that were morphologically intermediate or even close to diploids and were most probably hybrids (Fig. 4). As in previous field studies (Hardy et al. 2000; Koutecký et al. 2011, 2012), these natural hybrids were tetraploid instead of triploid and must have originated from unreduced gametes (probably ovules) of diploids (Koutecký et al. 2011). They were fully fertile (bore filled seeds), which indicates back-crossing with tetraploid C. jacea and possible introgression of genetic material of diploids to the genome of tetraploids.