Introduction

The transfer of Asyneuma comosiforme Hayek & Janch. to Campanula L. (Frajman & Schneeweiss 2009) has highlighted a well-known taxonomic problem that is particularly prevalent in the Campanulaceae. Many of the genera cannot be diagnosed explicitly by single characters, but instead by a combination of them, or by overall appearance (Gestalt), often in combination with geographic distribution. If we were to adhere to a narrow character-diagnosis, then several large genera such as Adenophora Fisch., Asyneuma Griseb. & Schenk and Campanula probably would have to be merged. From this perspective, Edraianthus A. DC. is only weakly separable from Campanula based on capsule dehiscence mechanisms, while similar “minor” criteria exist for other small genera. Yet, with a little experience of the Campanulaceae as a whole, these so-called “satellite” genera are readily separable. With the transfer of A. comosiforme, the already weak circumscription of Campanula (Eddie & al. 2003) is rendered even more impractical. Despite these difficulties, we present data to support the contention that a combination of morphological characters of this species differs significantly from that of the isophyllous species of Campanula (Park & al. 2006) and Asyneuma, and that, together with molecular sequence data, warrants the recognition of a new monotypic genus to accommodate A. comosiforme. We support this proposal by arguments that advocate the separation of monotypic genera in a Linnaean taxonomic classification, rather than an inclusive, “all-embracing” concept of the genus Campanula, and that such an arrangement better enhances our understanding of Campanulaceae evolution.

Material and methods

This study is based on the herbarium material deposited in the herbaria BEOU, BP, GB, JE, NHMR, TIR, W, WU and ZAGR (herbarium codes follow Thiers 2017+), personal field investigation (field. obs.) in all known sites, and living plants cultivated in the Botanical Garden Jevremovac, University of Belgrade. Names of taxa and taxonomic information and data are taken from the Euro+Med PlantBase (Euro+Med 2006–2017) and Lammers (2007). Abbreviations for the author(s) of scientific names follow the authors database of the International Plant Names Index ( http://www.ipni.org/ipni/authorsearchpage.do). Special emphasis was placed on the type materials and the descriptions in protologues. These materials are cited under Material studied. The locations for occurrences of the species in the field were recorded with a GPS (Garmin eTrex Legend HCx and Garmin eTrex Vista C).

Seeds and type of hairiness of leaves, corolla, pistil and fruits were examined under a LEICA stereo-binocular microscope. For the investigation of fine micromorphological structure of seeds, samples were coated with gold and studied with a scanning electron microscope (JEOL JSM-6460).

Root tips obtained from plants grown in pots were used for the analyses of mitotic chromosomes. They were pre-treated with 0.002M 8-hydroxyquinoline for 3 h at 4°C, fixed in cold 3/1 (v/v) absolute ethanol/glacial acetic acid for 48 h and stored in 70% ethanol at 4°C for further use. Hydrolysis was performed in 1N HCl for 12 min at 60°C and staining in Schiff reagent for at least 2 h (Feulgen & Rossenbeck 1924), followed by squash in acetic carmin. Chromosome plates were observed under a Leica DMLS light microscope (Leica Microsystems, Wetzlar, Germany) and photographed with Leica DCF 295 camera (Leica Microsystems). The chromosome number was determined from at least five individuals and from at least 10 cells per root tip.

Results

We provide new insights on the morphology of Asyneuma comosiforme based on detailed studies of living plants in all vegetative stages from different habitats and localities. Details are shown on Table 1, Fig. 1–3 and in the Expanded species description of the Taxonomy section.

The data in Table 1 should highlight the diagnostic characters for Asyneuma comosiforme and its separation from Asyneuma and the isophyllous species of Campanula, but, taken individually, most of these key features also occur in diverse genera of the Campanulaceae (Table 2), none of which can be diagnosed by a single character alone, but by a combination of them. We conclude that (1) linear corolla lobes, (2) corymbose inflorescence, (3) capsule dehiscing by lateral, medial valves and (4) rhombic leaves represent a combination of key features that clearly differentiate A. comosiforme from the other taxa.

The chromosome number 2n = 2x = 34, was determined here for the first time in Asyneuma comosiforme (Fig. 4). Given that this number is a common number in the bellflower family, not only in many Asyneuma and Campanula species, but also in other genera, such as Adenophora, Feeria Buser, Michauxia L'Hér., Physoplexis (Endl.) Schur, Phyteuma and Trachelium L. (Table 2), it cannot be used as an exclusive character for the segregation or conjugation of the genera within the Campanulaceae.

Table 1.

Comparison of morphological key features and chromosome numbers of Hayekia comosiformis, Campanula “isophyllous” group, Asyneuma Sect. Clinocarpium and Asyneuma s.str. (chromosome numbers are based on: Anchev 1975, 1993; Contandriopoulos 1980a, 1980b; Strid & Franzén 1981; Tzanoudakis & Kypriotakis 1987; Lammers & Hensold 1996; Probatova & al. 1998; Park & al. 2006; Probatova 2006; Rice & al. 2015 and references therein).

Discussion

A more rigorous, in-depth morphological examination of Asyneuma comosiforme reveals it to be enigmatic with respect to both Asyneuma and Campanula. Hayek (1921, 1924) thought the plant was closer to Asyneu ma, although perhaps not comfortably so, which was also our conclusion. In overall appearance, A. comosiforme resembles Asyneuma marginally more than the isophyllous bellflowers. It even bears a superficial resemblance to Diosphaera rumeliana (Hampe) Bornm. (≡ C. rumeliana (Hampe) Vatke), perhaps less so to Physoplexis comosa (L.) Schur. Despite the corroboration of the three molecular studies (Frajman & Schneeweiss 2009; Stefanović & Lakušić 2009; Bogdanović & al. 2014, 2015), the taxonomic status of A. comosiforme remains questionable. If this species is transferred to Campanula, it becomes even more difficult to find characters to clearly separate Asyneuma and Campanula on morphological grounds, but we disagree with the notion that most morphological characters are highly plastic and not suitable to delineate natural groups (Bogdanović & al. 2014: 1250, referring to Roquet & al. 2008, 2009). Bogdanović & al. (2014) provided a key to the species of C. ser. Garganicae Trinajstić, which included A. comosiforme (as C. comosiformis (Hayek & Janch.) Frajman & Schneew.), but most of the characters used in the key are equivocal. Within a group of species, discontinuously variable character states are not explicitly definable, and have tendencies towards one expression or another, but they can be used in combination to provide perfectly adequate diagnoses. Both the above-mentioned studies have influenced the respective authors to conclude that the deeply divided corolla is unreliable as the sole character to differentiate Asyneuma from Campanula.

Table 2.

Distribution of morphological key features and chromosome number of Hayekia comosiformis in comparison with other European genera of the Campanulaceae (chromosome numbers are based on: Rice & al. 2015, and references in: Anchev 1975, 1993; Contandriopoulos 1980a, 1980b; Strid & Franzén 1981; Tzanoudakis & Kypriotakis 1987; Lammers & Hensold 1996; Probatova & al. 1998; Park & al. 2006; Probatova 2006; Rice & al. 2015, and references therein).

Bogdanović & al. (2014: 1250) stated “There have been several attempts of classification (De Candolle 1830; Boissier 1875; Gadella 1966a, 1966b; Contandriopoulos 1984; Kolakovsky 1994), but none of them have been confirmed by any phylogenetic studies available to date…” However, Eddie & al. (2003: 565) concluded “Overall, there is a remarkable congruence between the ITS tree and traditional ideas on species relationships within the Campanulaceae (Eddie, 1999). The insights of early workers such as De Candolle and Boissier have proved to be remarkably clear, and their classification systems have, on the whole, been logically consistent with our findings on phylogeny.” Despite this, a cladogram is not synonymous with a classification and should not be used simply as a template for one. It is a useful means of data exploration and can provide phylogenetic insight, but should be interpreted with great care.

While the newly recovered “phylogenetic” position of Asyneuma comosiforme by Frajman & Schneeweiss (2009), Stefanović & Lakušić (2009) and Bogdanović & al. (2014) is strongly supported, the cladograms obtained by those authors invite explanation. Although the topology of a cladogram can highlight close phylogenetic relations, this is not always the case and it can vary greatly depending on the parameters selected for the analysis. Frajman & Schneeweiss's interpretation of the topology of their trees lacks the latitude necessary to obtain a broad vision of the inter-relationships within these lineages.

In fact, Asyneuma comosiforme is not as remote from Asyneuma as Frajman & Schneeweiss (2009) implied. It is not nested within the Garganica clade but is actually consistently sister to it. This may simply indicate that its molecular signature has diverged significantly from the bulk of species in Asyneuma, and therefore its position is perhaps not so surprising because both lineages are likely to have descended from a common ancestor anyway. There is no doubt that the genus Asyneuma and the isophyllous campanulas are closely related sister lineages within the Rapunculus clade of Campanula s.l. (Eddie & al. 2003; Park & al. 2006), or within the large clade CAM02-12 (Mansion & al. 2012; Jones & al. 2017). The apparent, closer relationship of A. comosiforme with the isophyllous campanulas on the molecular cladogram may be due to evolutionary stasis of its genome and/or differential rates of mutation. Asyneuma comosiforme may be ancestral to the isophyllous campanulas. How we classify A. comosiforme is another question because, from a morphological perspective, it is clearly “out of phase” with both Asyneuma and Campanula. This is the most important point at issue in this paper.

In the proposed taxonomic treatment of Asyneuma comosiforme, we have been guided by the recommendations or principles advocated by McVaugh (1945), for example: “The most important criterion of any supposed genus is not the width of the gap between it and another, but its own biological unity. Homogeneity of many characters, regardless of the degree of overlapping of these characters with those of other genera is the best indicator of this unity.” (Recommendation 3, in part).

Such segregation is hardly a new precedent. In early classifications, Phyteuma embraced all the segregate taxa that are now assigned to Asyneuma, Cryptocodon Fed., Cylindrocarpa Regel, Petromarula Vent. ex R. Hedw., Physoplexis and Sergia Fed. The western North American Campanula prenanthoides Durand has also been transferred to Asyneuma (McVaugh 1944, 1945; Gadella 1966b). The recognition of unique evolutionary grades of organization as mono/oligotypic genera should help us better understand the evolutionary history of the Campanulaceae, especially the rapunculoid bellflowers, because this assemblage is a major stumbling block in the reclassification of the family. Asyneuma comosiforme is isolated and unique within the Asyneuma/isophyllous lineages and therefore it is fitting for it to be given proper recognition in the genus Hayekia.

Taxonomy

Taking the above comparisons and discussion into account, we support the opinion of Radomir Lakušić (2001), who transferred Asyneuma comosiforme to the separate genus Hayekia. Based on additional morphological, karyological, chorological and ecological details provided here, which further highlight its taxonomic uniqueness, we propose the following.

Hayekia Lakušić ex D. Lakušić, Shuka & Eddie, gen. nov.

Type: Hayekia comosiformis (Hayek & Janch.) Lakušić ex D. Lakušić, Shuka & Eddie.

Generic diagnosis — Hayekia resembles Asyneuma, but differs as follows. Rhizomes subligneous and fragile, divergently branched, elongate and soboliferous, giving rise to many flexuous (often scorpioid), vertical herbaceous shoots; shoots glabrous with longitudinal papillate rows. Leaves ovate to rhomboid, with winged petiole. Flowers long pedicellate, clustered in a head-like, terminal raceme (corymb), with bracts of lower and middle flowers same as cauline leaves, much longer than pedicels. Capsule pendent, dehiscing by 3 clearly protruding, recurved axicorns, pushing fruit wall (pericarp), and opening medially by V-shaped, translucent, upward-curving, persistent valves.

Hayekia comosiformis (Hayek & Janch.) Lakušić ex D. Lakušić, Shuka & Eddie, comb. nov. (Fig. 1–3) ≡ Asyneuma comosiforme Hayek & Janch. in Oesterr. Bot. Z. 70: 20. 1921 ≡ Campanula comosiformis (Hayek. & Janch.) Frajman & Schneew. in Syst. Bot. 34: 599. 2009. – Lectotype (designated by Frajman & Schneeweiss 2009: 599): [Albania, Shija gorge, east of Bicaj] “Nordost-Albanien: Felsen am unteren Beginn der ŠijaSchlucht bei Bicaj, ca 400 m”, 15 Jun 1918, H. Zerny ( GB-004 7108 [photo!]; isotype: W [photo!]).

Expanded species description — Given that Hayek (1921) described this species on the basis of a few specimens collected in the flowering stage from a single locality, the following is an expanded description of living plants from different habitats and localities, and representing all vegetative stages.

Caespitose perennial. Rhizomes soboliferous, elongated and very branched, subligneous; stems many, erect, non-branched, 20(–30) cm long, subligneous, becoming herbaceous, lax and leafy, fragile, often with a flexuous (sometimes scorpioid) (Fig. 1B); shoots glabrous with longitudinal papillate rows. Leaves ovate to rhomboid, sometimes semi-succulent, glabrous; petiole long, narrowly winged; blade of basal leaves orbicular-ovate, small, withered at anthesis; blade of cauline leaves quadrangular, nearly rhombic, with obtuse lateral angles, rarely lanceolate, base triangular, cuneate, margin serrate with sharp teeth pointing forward (Fig. 1C, D), apex tapering to a point, ± acute. Inflorescence axis normally terminating in a loose, head-like, flat- or round-topped corymb, with lower pedicels longer than upper ones (Fig. 1C, D); occasionally inflorescence axis lacking a terminal corymb and then single flowers borne in leaf axils; bracts of lower and middle flowers same as cauline leaves, much longer than pedicels (Fig. 1C, D); bracts of upper flowers narrowly lanceolate, very small; pedicels slender, erect at anthesis (Fig. 1C, E, F), but elongated and recurved in fruit (Fig. 1D). Calyx green, ovoid, glabrous; calyx lobes narrowly triangular to almost linear-lanceolate, 2–3 × as long as ovary, with 1 (rarely 2) prominent, curved tooth on each margin (Fig. 1F). Corolla blue to dark lilac, rotate, divided nearly to base into 5 deflexed, long linear, strap-like lobes (Fig. 1A, C, E, F), each 12–20 mm long × 1.5–2.5 mm wide. Filaments blue, with narrowly deltoid base (Fig. 1E); anthers dirty green to dirty yellow, several times longer than distal, filiform part of filament. Ovary 5-ribbed, trilocular; style slender, as long as or longer than corolla lobes; collector hairs in upper , often encrusted with translucent, saffron-coloured pollen, usually curved backward in female phase (Fig. 1E). Capsule ovoid to deltoid, 2.7–3.8 mm long × 1.6–3.1 mm wide, 5-ribbed, glabrous; dehiscing by 3 clearly protruding, recurved axicorns pushing fruit wall (pericarp) and opening in medial part with V-shaped, translucent, persistent, upward-curving valves (Fig. 2A, B). Seeds shiny reddish to dark brown, 0.8–0.9 mm long × 0.5–0.6 mm wide, differentiated into flattened (Fig. 3A) and ellipsoid (Fig. 3C) types, both with similar striate testa with elongate, fibriform cells and slightly raised and anastomosed, radial and anticlinal walls forming a distinct, rugose, linear lumen (Fig. 3B, D).

Fig. 1.

Hayekia comosiformis – A: habit; B: soboliferous, subligneous rhizomes; C: inflorescence at flowering stage; D: inflorescence at fruiting stage; E: flower showing linear, strap-like lobes, filaments with narrowly deltoid filament base, anthers and style; F: flower showing calyx and calyx lobes with a tooth on each margin. – Photographs: A by B. Hallaçi (Shija gorge, 29 May 2012); B, C, E, F by L. Shuka (Shija gorge, 2 Jun 2017); D by B. Surina (Cape of Skanderbeg, 26 Jul 2012).

Chromosome number — 2n = 2x = 34, chromosomes 0.805 to 1.856 µm long (Fig. 4).

Phenology — Flowering from mid-May (at lower altitudes) until mid-July (at higher altitudes).

Fig. 2.

Hayekia comosiformis, capsules with recurved axicorns and V-shaped, upward-curving valves. – Scale bars: A, B = 1 mm. – Origin: Bicaj canyon (Shija gorge).

Fig. 3.

Hayekia comosiformis – A, B: flattened seed; C, D: ellipsoid seed. – Scale bars: A, C = 100 µm; B, D = 50 µm. – Origin: Bicaj canyon (Shija gorge)..

Distribution and ecology — Northeastern Albania, stenoendemic, monotypic (Fig. 5); 400–1985 m; crevices and screes of limestone cliffs and gorges.

Fig. 4.

Hayekia comosiformis – metaphase chromosome plate, 2n = 2x = 34. – Scale bar = 10 µm. – Origin: plants collected in Mali i Dejës and grown in Botanical Garden “Jevremovac” in Belgrade.

Generally, Hayekia comosiformis is restricted to steep, rocky slopes along semi-shaded canyons, providing relatively low temperatures and high air humidity. Here, it grows with other chasmophytic vegetation in cracks and crevices of small ledges or terraces. It is currently known from eight sites (four canyons and four localities in open screes or at the upper tree line) of the Korab–Gjalica and Deja–Pas Deja mountain groups. Two localities have a northwesterly aspect and are located in the tributaries of the right bank of the Black Drin, at 520 m (Bicaj canyon, site 1) and 630 m (Çaja canyon, site 2), respectively, whereas the third locality in the Seta canyon (Çidhna e Poshtme, site 3), occurs in the left bank of the Black Drin at 550 m and has an easterly aspect. The locality in the canyon of Kepi i Skënderbeut (site 8) is at the altitude of 1080–1150 m with a southeasterly aspect. Other localities of the Deja–Pas Deja mountains where H. comosiformis occur are not situated in canyons. They are found at altitudes of 1250–1985 m and 3.5–12.5 km from the canyon of Kepi i Skënderbeut, in semi-shady rock crevices and on open, stony and rocky pastures, with shallow soil, within the zone of forests of Fagus sylvatica L. and Pinus heldreichii Christ. Geologically, the rocks are upper Triassic to lower Cretaceous, and composed of homogeneous limestone in the Bicaj, Seta, Kepi i Skënderbeut and Guri i Vashës (site 6) localities, but of a mixture of limestone and thin, granitic tiles at the Lapaj bridge of the Çaja canyon. All the other localities of the Deja–Pas Deja mountains are on limestone substrates.

The tree flora of the upper parts of the rocky slopes in the three lowest localities comprises principally Carpinus orientalis Mill. and Juniperus deltoides R. P. Adams, rather than Fagus sylvatica and Pinus heldreichii, which occurs in the two localities above 1000 m, and Juniperus communis subsp. nana Syme in the Deja mounatins above 1500 m. Associated vegetation on the rocky slopes of all the canyons is generally sparse, but includes Allium flavum L., Asperula scutellaris Vis., Athamanta turbith subsp. haynaldii (Borbás & R. Uechtr.) Tutin, Campanula korabensis F. K. Mey., Clinopodium thymifolium (Scop.) Kuntze, Moehringia bavarica (L.) Gren., Moltkia petraea (Tratt.) Griseb., Ramonda serbica Pančić, Saxifraga paniculata Mill., Scabiosa crenata Cyr., Silene multicaulis Guss. and Teucrium arduini L. In addition, Heliosperma nikolicii (A. Seliger & Wraber) Niketić & Stevan., Hieracium waldsteinii Tausch, Saxifraga federiciaugusti subsp. grisebachii (Degen & Dörfl.) D. A. Webb and Sedum telephium subsp. maximum (L.) Krocker were noted in the Bicaj and Çaja canyons, while Crepis macedonica Kitan. was observed in the Seta and Kepi i Skënderbeut canyons. In the Kepi i Skënderbeut canyon, Saxifraga federiciaugusti subsp. grisebachii is replaced by S. sempervivum C. Koch. In the Deja–Pas Deja mountains localities, Hayekia comosiformis is associated with Alchemilla velebitica Borbás ex Janch., Amphoricarpos autariatus Blečić & E. Mayer, Asperula doerfleri Wettst., Carex kitaibeliana Degen ex Bech., Edraianthus graminifolius (L.) A. DC., E. serpyllifolius (Vis.) A. DC., Euphorbia myrsinites L., Hieracium villosum Jacq., Juniperus communis subsp. nana, Laserpitium siler L. subsp. zernyi (Hayek) Tutin, Leucanthemum vulgare Lam., Minuartia graminifolia (Ard.) Jáv., Pinus heldreichii, Potentilla apennina Ten., Saxifraga paniculata, Sedum dasyphyllum L., Sesleria interrupta Vis., Stachys recta L., Thalictrium minus L., etc.

Conservation status — The estimation of the potentially threatened status of Hayekia comosiformis was made according to the criteria and categories of the IUCN (IUCN Standards and Petitions Subcommittee 2017). It is included in the Red List of protected species of Albania and assessed as Endangered (EN A1b) (MoE 2013). It occurs in only eight localities, with an area of occupancy of 44 km² extending over 394.2 km². The population size was estimated between 1500 and 2000 mature individuals. Because the localities at higher altitudes and within canyons are subject to the same threats from climate change, and there is a reduction in available habitat in three other localities, mainly due to water course deviation for hydropower construction or quarrying activities, H. comosiformis is assessed as Endangered EN B1ab(iii)+2ab(ii,iv,v).

Etymology — The generic name Hayekia is feminine and therefore the specific epithet is comosiformis. The name Hayekia was chosen by Prof. Radomir Lakušić in honour of Prof. Dr August von Hayek (1871–1928), who was an Austrian botanist and associate professor at the University of Vienna, and who is particularly remembered for his phytogeographical investigations in the Balkans and the southeastern Alps. Prof. Lakušić (1933–2005) was a Yugoslavian botanist, a biogeographer and a vegetation biologist, whose field of work included, in particular, the subalpine and alpine high-mountain vegetation of the southeastern Dinarides in Bosnia and Herzegovina, and Montenegro. He was a professor at the Universities of Sarajevo and Banja Luka.

Fig. 5.

Distribution of Hayekia comosiformis – numbers on map correspond with numbers of sites listed under Material studied. – Country codes: Al = Albania; Ko = Kosovo; Ma = North Macedonia; Mn = Montenegro.

Material studied — This includes herbarium specimens examined or plants observed in the field. The site numbers correspond with the numbers on the distribution (Fig. 5). Because the species is listed as endangered, more precise locality data are omitted. This information could be supplied on request to any bona fide researcher.

District of Kukës (Rrethi i Kukësit):

Site 1 — Bicaj canyon (Shija gorge): BP524980!, BP589744!, BP768765!, GB-004 7108!, NHMR-1093!, TIR!, W 1958-0010312!, WU-4884!, ZAGR-32597!

Site 2 — Çaja canyon: TIR!, ZAGR-32598!

District of Peshkopi (Rrethi i Peshkopisë):

Site 3 — Seta canyon (Çidhna e Poshtme): TIR! District of Mirditë (Rreth i Mirditës):

Site 4 — Kurbnesh: BP- 22449! District of Mat (Rreth i Matit):

Site 5 — Mali i Dejës: BEOU-46886!, BP-21607!

Site 6 — Guri i Vashës: TIR!

Site 7 — Maja e Butrojes: Z. Barina & al., field. obs.

Site 8 — Kepi i Skënderbeut (Cape of Skanderbeg): JE5075!, NHMR-1982!, TIR!

Note: three literature records: “Gjallica–Koritniku” (Shuka & al. 2008), “ALBANIA. Gjalica, Mustafe” (Bogdanović & al. 2014) and “a 4 km territory in both sides of the road Kalimash–Morinë, Kukesi district” (Hallaçi 2008) refer to the same site: the Shija gorge E of Bicaj (site 1).