Warionia ( Asteraceae ) : a relict genus of Cichorieae ? by

Katinas, L., Tellería, M.C., Susanna, A. & Ortiz, S. 2008. Warionia (Asteraceae): a relict genus of Cichorieae? Anales Jard. Bot. Madrid 65(2): 367-381. The genus Warionia, with its only species W. saharae, is endemic to the northwestern edge of the African Sahara desert. This is a somewhat thistle-like aromatic plant, with white latex, and fleshy, pinnately-partite leaves. Warionia is in many respects so different from any other genus of Asteraceae, that it has been tentatively placed in the tribes Cardueae, Cichorieae, Gundelieae, and Mutisieae. Until now, a comprehensive study of Warionia to have a complete context for discussing its taxonomic position is lacking. The general morphology, anatomy, palynology and chromosome number of W. saharae are investigated here, and the species is described and illustrated. Laticifers in leaves and stems indicate a relationship with Cichorieae, and are associated with the phloem, in contact with it or with the surrounding sclerenchyma sheath. The pollen features indicate a strong relation with Cardueae, namely the structure with Anthemoid pattern where the columellae are joined to the foot layer, the ectosexine with thin columellae, the endosexine with stout and ramified columellae, the conspicuous spines with globose bases and conspicuous apical channels, and the tectum surface very perforate. Chromosomal counts resulted in 2n = 34. The morphological and palynological evidence positions Warionia between the tribes Cardueae and Cichorieae suggesting that it could be a remnant of the ancestral stock that gave rise to both tribes.


Introduction
Warionia Benth.& Coss. is a monotypic genus of Asteraceae, endemic to the northwestern edge of the African Sahara desert.The species Warionia saharae Benth.& Coss., known by the vernacular name of "afessas" or "abessas", may be found in several localities on dry shale in northwestern Africa, in Morocco and Algeria (Lewalle, 1986;Audissou, 1999) (Fig. 1).This is a thistle-like aromatic plant, with white latex, and pinnately-partite, somewhat fleshy leaves.The capitula are homogamous with tubular corollas, the anthers are caudate, and the style branches are dorsally covered by acute collecting hairs extending somewhat below the branches bifurcation.
One peculiarity of this plant is the penetrating odor that some people find unpleasant, given by the glandular hairs in the epidermis of the leaves at the slightest contact.The plant is considered to have medicinal properties mainly by its essential oils (Watillon & al., 1987).Decoction of dried leaves is used as antirrheumatic and against epileptic crisis (Bellakhdar & al., 1986).Crude extracts of the plants showed antibacterial and cytotoxic activities against a cancer cell line (KB cells) (Hilmi, 2002).In addition, the local women anoint themselves with the perfume of the leaves, and believe that the supernatural powers attributed to the plant make them more seductive (Audissou, 1999).
Finally, the pollen of Warionia has another particular feature such as the exinal dimorphism detected by Dimon (1971), i.e., the specimens have pollen grains with different exine thickness; although this feature is not mentioned in other works (e.g., Wortley & al., 2007).
Warionia is in many respects so different from any other genus of Asteraceae, that it has been described and tentatively placed in several tribes.In the original description of the genus and its species, Bentham and Cosson (1872) included the new taxa in the Cardueae, but further authors (Dimon, 1971;Cabrera, 1977;Dittrich, 1977) considered Warionia a member of the tribe Mutisieae sensu lato.Hansen (1991a), in his morphology-based cladistic analysis of Mutisieae sensu lato, suggested again a possible relationship of Warionia with the Cardueae by the acute style-hairs and the bell- shaped corolla.The relationship with Cardueae was also supported in a scanning electron microscopy palynological study of Mutisieae (Hansen, 1991b) due to the spiny grains of W. saharae.In the morphologybased cladogram of Karis & al. (1992), Warionia appears as an independent branch between the Cardueae and the Vernonioid complex (tribes Arctoteae, Cichorieae, Liabeae, and Vernonieae).Bremer (1994) placed Warionia among the genera unassigned to a tribe in the subfamily Cichorioideae.In a monographic study of the Mutisieae sensu lato (Katinas & al., in press), the exclusion of Warionia was confirmed.Some molecular phylogenetic studies place Warionia as sister to the tribe Cichorieae.Funk & al. (2004) placed Warionia in the tribe Gundelieae, together with the monotypic Gundelia.This branch had a low support, but the sister relationship of this tribe with Cichorieae resulted in 100 % support.Gundelia had been previously included in the Cichorieae by Karis & al. (2001).In the ITS molecular phylogenetic analysis of Goertzen & al. (2003) Warionia appears as the only member of a branch sister to the tribe Cichorieae, although in this case not well supported.This result was repeated in the supertree of Funk & al. (2005).Finally, the combined analysis of 10 chloroplast DNA loci of Panero & Funk (2008) yielded a tribe Gundelieae, constituted by Warionia and Gundelia, sister to Cichorieae with stronger statistical support.
Until now, a comprehensive study of Warionia to have a complete context for discussing its taxonomic position is lacking.The main references to Warionia are phytochemical studies and floristic treatments (e.g., Ozenda, 1991, Gómez, 2001), although there are references to pollen (e.g., Wortley & al., 2007), laticifers (e.g., Augier & Mérac, 1951) and chromosomes (e.g., Oberprieler & Vogt, 1993).A critical investigation of Warionia that gathers all the relevant information on the genus, i.e., morphological, anatomical, palynological, and chromosomal, is needed and have led us to the present study.

Materials and methods
Vegetative and reproductive organs were obtained from herbarium (BC, MA, MO, US; Holmgren & al., 1990) and living specimens.Plants for anatomical and part of the palynological studies were cultivated from seeds of the specimen Romo & al. 10543 (BC) in the Greenhouse of the Botany Department, University of Wisconsin-Madison.Samples from herbarium specimens were reconstituted in boiling water.Whole foliar segments and phyllaries were cleared according to Strittmatter (1973).The midregion of plant organs were isolated and free hand cut transversely.Some sections were stained with Safranin while the rest were left unstained.Observations and drawings of morphological and anatomical features were carried out on a Wild M5 stereomicroscope and on a Leitz SM Lux light microscope equipped with a camera lucida; the drawings of the plant were performed by the authors (LK).Photographs were taken with a Nikon Coolpix S10.For scanning electron microscopy (SEM), portions of styles of FAA-fixed samples were taken, critical-point dried in CO2, and examined in a JEOL/EO JSM-6360.
Pollen grains of Warionia saharae were obtained from herbarium specimens deposited at BC, MO, and US.Pollen of Gundelia tournefortii L. was also analyzed (Barbey 522, BC) to compare with the pollen of Warionia saharae.For light microscopy (LM), pollen grains were acetolyzed and mounted in glycerol jelly.Measurements of equatorial and polar diameters exclude the spines.For scanning electron microscopy (SEM), acetolyzed pollen grains were suspended in 90 % ethanol, mounted on stubs, sputter-coated with gold palladium and examined with a JEOL JSM T-100 microscope.Pollen terminology follows Punt & al. (1994).

Warionia
Etymology: The specific epithet refers to the distribution area, in the African Sahara desert.
Biology: Flowering is abundant and lasts several months.The flowering season has been recorded from April to June, while it may extend to July or August if the spring rains are abundant and well spaced (Audissou, 1999;Gómez, 2001).In cultivation (USA) the plants flowered in November.The florets have perfume.In habitat W. saharae loses it leaves in the dry season (Audissou, 1999).
Observation: There are two specimens at herbarium K, labelled as types in the Royal Botanic Gardens Kew database (accessed: September 2008), collected by Merdochée in southwestern, in 1875.They have a label with the inscription: "Par les soins de M. Beaumier.Herb.E. Cosson, Recd.8/77".Despite the fact that they might belong to Cosson's collections, these specimens do not correspond to type specimens because they have a different place of collection and collector than those indicated in the protologue.

Anatomy
Leaf transection (Fig. 4 a, b) shows an adaxial and abaxial epidermis with 2-seriate glandular hairs and slightly exerted stomata of the anomocitic type (seen in paradermal section).There is a layer of hypodermis on the adaxial surface.Collenchyma at the midvein level forms extensions from the hypodermis and the abaxial epidermis to the three vascular bundles.The mesophyll consists of a central layer of colorless parenchyma, an upper, and a lower layer of palisade.A sheath of sclerenchyma fibers surrounds the veins.Laticifers are closely associated with the phloem, in contact with it or with the surrounding sclerenchyma sheath.Paradermal sections of leaves show the laticifers associated with the vascular tissue.Parenchyma cells separate tangentially anastomosed strands of laticifers.The primary stem in transection (Fig. 4 c, d) is characterized by the epidermis covered with scarce 2-seriate glandular hairs, followed by a layer of collenchyma.Vascular bundles are embedded in the parenchyma, and surrounded by sclerenchyma sheath.Lignified parenchyma constitutes the pith.Laticifers with dense content are somewhat larger than the neighboring cells, and are present in the cortex surrounding the sclerenchyma sheath of the phloem.

Pollen
Pollen is radially symmetrical and isopolar; spheroidal to suboblate; more or less circular in equatorial and polar view (Figs. 5 a-d Exine is tectate, echinate, ca. 12 µm thick at the equator slightly slender at the poles; spines are ca.7 µm long, with globose bases, acute or rounded tips and conspicuous apical channel (Fig. 6 a-d); tectum surface is very perforate excepting the spine tip.Ectosexine Pollen size is variable among specimens, and strongly variable in the specimens Gómez s.n (BC) and Gattefosée 1340 (MO).Pollen grains from Gattefosée 1340 (MO) are unusually spiny (and therefore, it was not possible to take measures of the diameters) having two types of spines: (1) acute tips, and (2) rounded tips or dome-shaped apex (Fig. 6 d).Both spine types have many apical channels.Rounded spines are rare in Asteraceae, generally spines have an acute tip.Similar spine morphology was observed in apparently aborted pollen of some Eupatorium species (Skvarla & al., 2001), and it was attributed to stress conditions during pollen development.

Chromosomes
Our results agree with three previous counts (Reese, 1957;Humphries & al., 1978;Oberprieler & Vogt, 1993;recorded in Watanabe, 2007), all of them with 2n = 34 (Fig. 7).The chromosome number n = x = 17, frequent in Asteraceae (Jeffrey, 2007a), is very common in Cardueae (e.g., Cirsium, Cynara, Jurinea, Notobasis, Onopordum) but very high for the Cichorieae, in which x = 9 (and lower numbers) predominates, with a few genera having x = 10 (Lack, 2007).Perhaps the most striking feature is the extreme asymmetry of the karyotype.The combination of a very high chromosome   number and extreme asymmetry suggests processes of dysploidy in Warionia.Descending dysploidy was first detected and classically studied in another Cichorieae genus, Crepis L. (Stebbins, 1971).We suggest two events of polyploidy followed by dysploidy to explain both asymmetry and high number.A first duplication of an ancestor with x = 5 would have led to x = 10, which by descending dysploidy would have resulted in x = 9.A second polyploidization event would result in x = 18 followed by the loss of a chromosome by dysploidy.

Discussion
This is the first global study of Warionia.Although previous chromosome counts in Warionia were reported (Reese, 1957;Humphries & al., 1978;Oberprieler & Vogt, 1993) and some pollen studies developed (Dimon, 1971;Wortley & al., 2007), we reanalyzed these features.We were particularly interested in comparing the pollen of Warionia with that of Gundelia in order to add another criterion for analyzing the taxonomic relationship between both genera.Considering this information together with the existing molecular data (e.g., Goertzen & al., 2003;Funk & al., 2004) we intended to provide a general overview for discussing the systematics of the genus.In this sense, it is crucial that we discuss some points here: the sister relationship between Warionia and Gundelia, the tribal position of Warionia, and some insights about the probable evolution of the genus.
The Warionia-Gundelia affinity Gundelia with its unique species G. tournefortii was considered by Robinson (1994) the unique genus of the tribe Gundelieae, and he considered it a connecting link between Arctoteae and Cardueae.Later, Funk & al. (2004) added Warionia to Gundelieae, and showed the tribe sister to Cichorieae.The thistle-like habit of Gundelia, the presence of latex, the spurred anther thecae, and the vernonioid style (Karis & al., 2001) would suggest some resemblances with Warionia.However, Gundelia also shows important morphological differences regarding Warionia such as the secondary heads (vs.solitary or 2-3-headed in Warionia), the functionally male peripheral florets (vs.all bisexual florets), the annuliform pappus (vs.capillary pappus) and, most important, the pollen features.Pollen of G. tournefortii has Helianthoid pattern, distinguished mainly by the presence of cavea (Fig. 6 f) (Skvarla & al., 1977;Robinson, 1994;Wortley & al., 2007), where the layer of columellae is separated from the foot layer from aperture to aperture.Warionia saharae, on the other hand, has an Anthemoid or ecaveate pattern (Fig. 6 c).Although the pollen sculpture is echinate in both genera, they differ in the shape of the spines.In Gundelia, the spines are conical with subglobose-shaped bases (Fig. 6 e), whereas in Warionia the spines are mameliform with globose-shaped bases (Fig. 6 a, b).Another remarkable feature of the spines of Gundelia is their tendency to unite at their bases (Fig. 5 f) as happens in species with subechinolophate pollen of Cichorieae, Vernonieae, and Arctoteae (Blackmore, 1984;Skvarla & al., 1977).We agree with Wortley & al. (2007) in that it is strongly incongruent to link Warionia and Gundelia into the tribe Gundelieae.Our observations suggest that the pollen morphology of Warionia shows a close relationship with the Serratula type of Cardueae (Wagenitz, 1955), whereas pollen of Gundelia shows that it is related to some Cichorieae, Vernonieae and Arctoteae.
Molecular phylogenetic analyses are not congruent regarding the degree of support of the branch Warionia-Gundelia.Combined molecular data (trnL-F, ndhF, ITS) display weak support (less than 50%) for the sister relationship between both genera (Funk & al., 2004), whereas the strict consensus tree resulting from maximum parsimony of combined data of 10 chloroplast loci (Panero & Funk, 2008) shows higher support (86%).Based on the overall evidence, we consider Warionia to be a separate lineage of Gundelia and thus Warionia should be excluded from Gundelieae.

Tribal position of Warionia
Warionia, as commented before, was included by some authors (Dimon, 1971;Cabrera, 1977;Dittrich, 1977) within the Mutisieae sensu lato.Some characters coincide with those of the African Mutisieae, principally with the members of the tribe Dicomeae (Ortiz, 2000(Ortiz, , 2001)), whereas many characters of Warionia show a departure from that group.A comparison be-tween Warionia and the African Dicomeae shows that: (1) Leaves of Warionia are sinuate to pinnately-partite whereas all the African Dicomeae have entire leaves; (2) Warionia's involucre is rather similar to that of the core Dicomeae (Dicoma Cass., Macledium Cass., Cloiselia S. Moore, Pasaccardoa Kuntze) but its phyllaries are not coriaceous and have not an acuminatepungent apex as in those genera; (3) Warionia's corolla tube dilates abruptly into the limb (being very similar to that of the genera Pleiotaxis Steetz and Erythrocephalum Benth.and Pasaccardoa) whereas in Dicoma and Macledium that separation is gradual; (4) the corolla epidermal cell cuticle ornamentation is conspicuously longitudinally striate and transversely ondulate ("intestine-like", see Karis & al., 1992) as in almost all the genera of the Dicomeae.The corolla lobes of Warionia, however, do not have thick-bundled apical veins, whereas these are present in the African Dicomeae (with the exception of some species of Dicoma and Pleiotaxis); (5) the anther apical appendage of Warionia is acute and thin, whereas in Dicomeae it is apiculate to acuminate and enlarged; (6) style of Warionia is covered by acute sweeping hairs as in almost all the African Dicomeae, but extending somewhat below the branches bifurcation point, whereas in the African Dicomeae the sweeping hairs are subapical, apical or lacking.Katinas & al. (in press), finally excluded Warionia from the Mutisieae s. l. mainly on the basis of style and pollen features.
The morphological and palynological evidence would place Warionia in a middle position between the tribes Cardueae and Cichorieae, since the genus has features of both tribes (Table 1).
The characters that link Warionia with Cardueae are mainly the type of corolla, the morphology of the cypsela apex, and the pollen features.The thistle-like habit of these plants, with its coarsely lobed leaves and large capitula of campanulate corollas probably led Bentham and Cosson (1872) to include Warionia in the Cardueae and relate it with the European genus Berardia.The thistle habit, however, may also be found in genera of Cichorieae.The discoid, homogamous capitula, with all campanulate florets are another feature typical of the Cardueae.Also, S-shaped corollas, as described by Dittrich (1977) for some genera of Cardueae, are found in Warionia.The apical plate at the end of the cypsela pericarp in Warionia is also characteristic of the Cardueae, as described by Dittrich (1977).Apically, the cypselae of Warionia have a crown-like structure or apical plate upon which are inserted the pappus and the nectary disc.More importantly, the pollen features indicate a strong relation with Cardueae, namely the structure with Anthemoid pattern where the columellae are joined to the foot layer, the ectosexine with thin columellae, the endosexine with stout and ramified columellae, the conspicuous spines with globose bases and conspicuous apical channels, and the tectum surface very perforate.Its highly perforate exine surface led Zao & al. (2006) to relate the pollen of Warionia with some Gochnatiinae (tribe Mutisieae s. l.).Pollen of Warionia can be placed among the echinate types of Cardueae (Tormo & Ubera, 1995) and, to some degree, it is related to the genus Macroclinidium Maxim.(tribe Mutisieae s. l.), which was regarded near the Cardueae by Katinas & al. (in press).The pollen of Warionia is very different from that of Cichorieae whose members have caveate exine (helianthoid pattern) with either echinate, echinolophate (with large window-like spaces lacking a tectum, and echinate ridges), or subechinolophate sculpture.When echinate, the spines do not have globose bases, as occurs in Warionia.Another feature that would link Warionia with Cardueae is the radially arranged style branches (Fig. 3 a, c), different for example from the tangentially arranged branches in other tribes such as Cichorieae (Robinson, 1984).The radial condition of the style occurs when the branches diverge along a radius of the head, and the tangential condition when the branches spread laterally and transversely to a radius of the head.
On the other hand, the characters that link Warionia with Cichorieae are the presence of laticifers, the type of style, and the molecular evidence.Although latex has been reported in Cardueae, it is stored in latex sacs rather than in elaborate laticifers as in the Cichorieae (Metcalfe & Chalk, 1965;Bremer, 1994).Leaves and stems of Warionia have conspicuous laticifers.As hap-pens with the palynological evidence, the style features are another important criterion when reevaluating the systematic position of taxa.Warionia has the typical vernonioid style of Cichorieae, Liabeae and Vernonieae, very different from the carduoid style characterized by a ring of hairs on the shaft below the branches.The style of Warionia has slender, long, filiform style branches and acute collecting hairs, distributed all along the branches extending down on the shaft of the style (Fig. 8).Finally, the molecular data represents another line of evidence regarding the position of Warionia close to Cichorieae.In the ITS analysis of Goertzen & al. (2003), Warionia appears at the base of Cichorieae, sister to the genera of the tribe although not well supported.In the phylogenetic analyses of Funk & al. (2004) and Panero & Funk (2008), the position of Warionia is also basal to Cichorieae, with higher support; in the later study, the Gundelieae branch, which includes Warionia and Gundelia, has the lowest support value when compared with the other tribes of the tree.
In summary, the morphological and palynological evidence positions Warionia between the tribes Cardueae and Cichorieae, whereas the phylogenetic molecular analyses place the genus at the base of the tribe Cichorieae.These two views are not contradicting but complementary, suggesting the following possibilities: (1) to regard Warionia as a primitive, basal genus of Cichorieae, or (2) to regard Warionia as the only member of a new, distinct tribe, sister to Cichorieae.The second option would need fewer assumptions regarding evolutionary changes of key features within Cichorieae, such as type of corollas and pollen grains.However, we cannot disregard the molecular data and

Evolution of Warionia
Direct dispersal easterly from South America to Africa was postulated by Panero & Funk (2008) as a plausible sweepstakes route across the Atlantic where early lineages of Asteraceae reached Africa.This may have given origin to the earliest African branching clades outside South America.African clades are represented by Tarchonantheae, Dicomeae, and Warionia; mostly African/Eurasian clades are Cichorieae, Gymnarrheneae; and the Cardueae are a Mediterranean/Central Asian clade.According to Panero & Funk (2008) the African dispersal gave rise to explosive radiation of Asteraceae across the world.The basal lineages of Carduoideae and Cichorioideae appear to be African or Mediterranean.Understanding the means by which Warionia saharae evolutionary arose is a challenge.Some chromosomal rearrangements may trigger speciation (Levin, 2002), particularly when geographical barriers to gene flow are absent.In these situations, the rearrangements may decrease gene flow sufficiently among diverging populations to allow selected differences or hybrid incompatibilities to accumulate (Riesberg & al., 1999).Hybrid speciation by means of allopolyploidy in which reproductive isolation arises as an instantaneous by-product of hybrid genome duplication could be one of such mechanisms.It is not possible to formulate any founded hypothesis about hybrid speciation in Warionia between parental species of Cardueae and Cichorieae since it would require, for example, crossing experiments.It is also possible that from an ancestral stock some genera remained in isolated pockets as relicts, conserving primitive set of features.The Sahara endemic Warionia could be a remnant of the ancestral stock that gave rise to Cardueae and Cichorieae, among other groups, and conserved morphological features of both tribes.Other genera of Asteraceae appear to have undergone similar processes.In North America, for example, Panero & Funk (2008) postulated that the monotypic genus Hecastocleis A. Gray endemic to western United States, linked to Mutisieae, could be the only relict of an Asteraceae lineage that reached North America from the Old World probably via the Bering land bridge.Jeffrey (2007b) placed the Asian Dipterocome under the carduoid genera of uncertain placement.He postulated that the carduoid pollen and cypsela features are probably plesiomorphic in Dipterocome, and that this genus, like the north African and Middle East Gymnarrhena, would lie at or below the base of Cichorioideae as a relict of a distinct line.Thus, monotypic genera such as Dipterocome, Gymnarrhena, Hecastocleis, and Warionia may illustrate that not all sweepstakes dispersing organisms may be equally successful in terms of speciation after colonization, depending on where and when they arrive and on their capacity to compete with other organisms.

Fig. 8 .
Fig. 8. SEM photographs of the style of Warionia saharae: a, upper part of the style showing the vernonioid type morphology; b, detail of one branch showing the acute collecting hairs.[a, b, Gattefossé 1340 (MO)].

Table 1 .
Comparison of the main morphological and palynological characters between Warionia saharae, and the members of the tribes Cardueae and Cichorieae.
prefer to maintain Warionia as a member of Cichorieae until new molecular studies determine other position.