Codonographia Gorgonum, or the description of a pleiad of bellflowers (Campanula, Campanulaceae) from the Cabo Verde archipelago

Abstract Gardère, M.L., J. Florence, S. Muller, Y. Savriama & J.-Y. Dubuisson (2021). Codonographia Gorgonum, or the description of a pleiad of bellflowers (Campanula, Campanulaceae) from the Cabo Verde archipelago. Candollea 76: 13–40. In English, English and Portuguese abstracts. The previous taxonomic study of Cabo Verdean bellflowers (Campanula L., Campanulaceae) recognized four endemic species. In light of a recently published geometric morphometric study, the reassessment of the Campanula jacobaea complex allows to restrict Campanula jacobaea C. Sm. ex Webb to the island of Santiago. A new lectotype is proposed for this name that conforms to the protologue. Three further new species are described: Campanula monteverdensis Gardère from São Vicente, Campanula fransinea Gardère and Campanula cochleromena from São Nicolau, and Campanula vicinituba Gardère from Santo Antão. Campanula hortelensis Gardère, also from Santo Antão is placed in synonym of Campanula feijoana Gardère. Seven species are recognized in the Cabo Verde archipelago including Campanula bravensis (Bolle) A. Chev., the most widespread bellflower occurring in all southern mountainous islands. A determination key, descriptions, photographs and a distribution map are provided.


Introduction
The genus Campanula L. (Campanulaceae), commonly known as bellflowers, comprises approximately 420 species which are distributed across Arctic and temperate regions of the Northern Hemisphere (Lammers, 2007), with some in tropical continental regions, especially in Africa where it occasionally extends to the Southern Hemisphere (Thulin, 1976(Thulin, , 1977; others are found in insular environments, particularly those of the Cabo Verde archipelago (Gardère, 2020), the focus of the present study. Situated in the mid-Atlantic Ocean c. 570 km off the West African coast and c. 1360 km south from the Canary Islands, the islands of Cabo Verde are the southernmost of Macaronesia (Fig. 1). Confined to the mountainous islands, Cabo Verdean bellflowers [CVB] occur mainly in humid, rupicolous environments (Gardère et al., 2017) and more rarely in shrublands. CVB form very floriferous and lignified subshrubs of remarkable size, which have always fascinated collectors and are of significant horticultural value (Hooker, 1883;Lobin et al., 1993). However, the restricted access to material due to the isolation of the archipelago, and the difficulty of preserving the exact shape of the corolla in herbarium specimens, has long impeded morpho-taxonomic research.

Feijó or the first collections
The first collections of bellflowers in the Cabo Verde archipelago date back to the end of the 18 th century and were made by the Portuguese naturalist João da Silva Feijó (1760 -1824) during the famous "Philosophical Journeys" (Gardère et al., 2019a). Feijó was assigned by the Portuguese Crown to study the natural history of the archipelago and made numerous collections of plants between 1783 and 1789, including bellflower specimens. In 1788, Feijó wrote a short unpublished manuscript entitled Plantae Insulanae, where he described some monospecific genera which he considered as new for science. Among these was the genus "Fransinea", including a single species "Fransinea insulana", since identified as belonging to Campanula (Gardère et al., 2019a). At present the only two existing bellflower specimens collected by Feijó are conserved at P. The precise localities where these plant collections were made, unknown until now, have now been determined thanks to examination of the flowers, as are discussed below.

Collections from the first half of the 19th century
During the first half of the 19th century, scientific expeditions bound for as-yet unexplored regions of the globe made frequent stopovers, used by botanists to collect plant specimens. In 1816 the Norwegian botanist C. Smith, a member of the expedition of Captain Tuckey, made the second known collection of bellflowers from Santiago during a short visit (Sunding, 1980). Smith later died on the banks of the Congo River during the expedition; his herbarium was brought back by his assistant D. Lockard and his notes were published in the accounts of the expedition (Tuckey, 1818). In his list of specimens, Smith considered the bellflower collected around Pico da Antónia to be new and named it "Campanula jacobaea" (Tuckey, 1818: 251), but lacking a description, this name remained a nomen nudum. The next collections were made in 1822 by J. Forbes during stopovers in São Nicolau, Santo Antão and Santiago (Owen, 1833). Then in 1832, on the second voyage of the Beagle, C. Darwin made a short visit to Santiago (Vala, 2009), where he collected material around the valley of São Domingo. A few years later, in 1839, J.D. Hooker, heading south with the Antarctica Expedition of Captain Ross, also collected material in the same locality and gave a brief description of it in a letter to his father W.J. Hooker: "A most beautiful blue flowered Campanula (nº 125) next appeared with corollas 1 ½ inch long quite European in appearance" (Rico et al., 2017: Appendix S1); in addition, he wrote in his journal: "a lovely fern with beautiful soft green foliage growing like our Cystopteris [Hypodematium crenatum (Forssk.) Kuhn, Dryopteridaceae] out of the crevices of the rocks; it grew with lots of the Campanula and Umbellifer (found on the way up) which so put me in mind of old Scottish forms of plant" (Huxley, 1918: 94). In 1841, the collections made by T. Vogel joined the list: during a stopover of the Niger Expedition in São Vicente, he collected some specimens from the summit of Monte Verde (Vogel, 1849).

Taxonomic genesis and successive concepts
In 1848, Webb gathered the collections of Forbes, Darwin, Hooker and Vogel, and, as a prelude to his Spicilegia Gorgonea (Webb, 1849), made the first description of a species based on these specimens (Webb, 1848). At first, the name "Campanula daltonii Webb" was indicated on the specimen labels and included in a list of plants of the Cabo Verde islands prepared by Decaisne (1848) as part of a study by C. Sainte-Claire Deville on the geology of the archipelago. However, Webb ended up adopting Smith's epithet when he formally described the CVB in Icones Plantarum, hence validly publishing the name C. jacobaea C. Sm. ex Webb (Webb, 1848: tab. 762). Later, Bolle (1861) undertook the first study in which infraspecific taxa were proposed, subdividing C. jacobaea into four entities: var. "genuina", an invalid name for the type variety (Santo Antão and São Nicolau), var. humilis (São Vicente and São Nicolau), var. hispida (Santiago), and var. bravensis (Brava). This first taxonomic division was not be entirely upheld, as it was founded on a complex combination of many inconsistent morphological features: ratios of the lengths of the calyx/ corolla and different types of indument. Besides this, the supposed geographical ranges of the different varieties were not all confirmed. Chevalier (1935), in his Flore de l'archipel, raised the var. bravensis to the species level, as C. bravensis (Bolle) A. Chev. (Chevalier, 1935: 889), but most botanists retained a monospecific concept accepting only C. jacobaea (Nogueira, 1976;Sunding, 1982;Lobin, 1982Lobin, , 1986Hansen & Sunding, 1985;Rustan & Brochmann, 1993;Figueiredo, 1995). It is only after the taxonomic revision of Leyens & Lobin (1995) that C. bravensis has been re-established as a distinct species. For the first time, a description was provided of its remarkable tubular corolla and the conical pubescent roof of the ovary, characters used to distinguish it from C. jacobaea, a highly polymorphic species with a campanulate corolla in addition to a flat and glabrous roof of the ovary. However, faced with the high level of variability in the floral characters, this taxonomic concept was challenged once again. Alarcón et al. (2013), unable to find reliable identifying characters nor geographic or genetic evidence to support the recognition of two species, retained a single, highly variable species, C. jacobaea, for the entire archipelago. More recently, the bellflowers from Santo Antão, hitherto identified as C. jacobaea, were described as two new species under C. feijoana Gardère and C. hortelensis Gardère (Gardère, 2015). These two species are clearly distinguishable from other CVB species by their infundibuliform corolla, constricted in its median part but distinguish from one another by several floral and foliar characters as the size, shape and position of the lobes and appendages of the calyx, and the shape and texture of the lamina of the rosette leaves.

Campanula jacobaea complex: one or several species?
Although the geographic range of C. jacobaea was reduced in 2015 to the islands of Santiago, São Nicolau, and São Vicente (Gardère, 2015), the species still was considered widespread in the archipelago with a high level of polymorphism. This is particularly pronounced in its floral characters. Field work conducted during numerous expeditions undertaken between 2013 and 2017 made it possible to untangle this enigmatic polymorphism. Campanula jacobaea in the wide sense indeed shows great diversity in the shape of its corolla: from campanulate to infundibuliform, including all the potential intermediate states. Nonetheless, based only on these distinctive shapes, four morphologically and geographically distinct entities can be identified: (1) a first, corresponding to the bellflowers from São Vicente with flared infundibuliform corollas, (2) a second, corresponding to the bellflowers from the western part of São Nicolau (henceforth W São Nicolau) with narrow infundibuliform corollas; and two other different entities with campanulate corollas: (3) one, corresponding to the bellflowers from the eastern part of São Nicolau (henceforth E São Nicolau), and (4) another, corresponding to the bellflowers from Santiago, with a longer corolla that is slightly constricted at the throat. Moreover, on an expedition in November and December 2015, a new entity with a flared, non-constricted infundibuliform corolla was discovered on the far eastern end of Santo Antão, in an isolated mountainous region.

The contribution of geometric morphometrics to distinguishing the species
To distinguish between these previously unsuspected morphogeographic entities, we recently analysed the corolla shape of all of the CVB using geometric morphometrics (Gardère et al., 2019b). The corolla shapes of 221 photographed flowers were characterized in high detail by placing 2D landmarks and semi-landmarks, and were then subjected to a Generalized Procrustes Analysis to extract shape by removing information on size, location and orientation (Savriama et al., 2012;Gunz & Mitteroecker, 2013;Dryden & Mardia, 2016;Savriama, 2018). The results of this study support the division of C. jacobaea into four morpho-geographic entities and suggest that the concept of C. jacobaea should be restricted to the island of Santiago, i.e. the entity that best fits the original description of the species made by Webb (1848). Moreover, the results confirm the taxonomic re-establishment of C. bravensis proposed by Leyens & Lobin (1995) in which we adhere, as well as the close floral morphological affinities between C. feijoana and C. hortelensis, and the presence of new taxon on Santo Antão. Based on the results of the geometric morphometric analysis and the observation of numerous herbarium specimens, a taxonomic revision of Cabo Verdean species of the genus Campanula is presented here, along with an identification key. The discriminating characteristics are mainly floral (and foliar to a lesser extent), enabling the recognition of seven species (Fig. 1), all endemic to the archipelago, four of which are newly described and illustrated. In this way, C. jacobaea is now characterised as a species restricted to Santiago, and three entities are newly described for populations occurring on the other islands that were once included in its range: C. cochleromena Gardère and C. fransinea Gardère in São Nicolau, and C. monteverdensis Gardère in São Vicente. A new species is described for bellflowers from Santo Antão, C. vicinituba Gardère, whereas C. hortelensis is placed in synonymy under C. feijoana. Finally, given that the lectotype of C. jacobaea designated by Porter (1986) is in serious conflict with the protologue, a new lectotype is designated here.

Material and methods
The taxonomic revision presented below is essentially based on field observations made by the first author (MLG) during several trips to the Cabo Verde archipelago between 2009 and 2017 and specimen collections mainly deposited in the Herbaria of P, LISC and in the Herbarium of Cabo Verde -as yet unreferenced in the Index Herbariorum (Thiers, Lobin, pers. comm.). A large collection of flowers prepared in the field is conserved in alcohol and stored in P. Corolla are described following the terminology proposed by Gardère et al. (2019b: Table 5). For Scanning Electron Microscopy (SEM), leaves were sampled from herbarium specimens and observed using a Hitachi SU-3500 SEM at the Electron Microscopy technical platform of the MNHN. Vernacular names. -Thirteen vernacular names in Cabo Verdean Creole language have been noted from the literature and herbarium labels; new names have been recorded on the field. Historically, the most ancient vernacular name has been reported by Feijó who noted "Campainhas", a Portuguese vernacular name to designate the bellflowers which is no longer used in the archipelago (Gardère et al., 2019a). Short lists of vernacular names are published by Basto (1988) and Feijão (1960) but without any information about localities. More recently, Gomes et al. (1995b) and Szpera (2015) have grouped all the CVB under "Contra-Bruxas" but this vernacular name is, in fact, peculiar to the bellflowers from Santo Antão (Leyens & Lobin, 1995;Figueiredo, 1995).

Key to the Cabo Verdean species of Campanula
Distribution. -In the archipelago, the genus is found on all islands with elevations above 700 m, i.e. Santo Antão, São Vicente, São Nicolau, Santiago, Fogo, and Brava ( Fig. 1).
Etymology. -The specific epithet "jacobaea", from the Latin Jacobus, refers to Santiago or "Saint James"; the island was given the name because it was discovered on Saint James Day. Jacobaea was initially chosen by Smith in 1816 to name the bellflowers from Santiago, which he wanted to describe as new to the Cabo Verde flora.
Notes. -Webb (1848: tab. 762) based the description of C. jacobaea on collections from different geographical origin and different collectors; only the names of the collectors were indicated. The following year in his Spicilegia Gorgonea Webb (1849: 148) added information such as localities, dates and sometimes collection numbers. Eight syntypes have been located. Original material is extant in the Hooker Herbarium (now K) and mounted on two sheets (Fig. 4,5) with duplicates in CGE and L and a further syntype has been located in G.
On the first sheet of original material in K (Fig. 4), four specimens are mounted of which only those on the bottom half display the names of the collectors ( Fig. 4A -B). On the bottom right, the Darwin's specimen [K000865902] collected in Santiago represents C. bravensis (Fig. 4A). A small footnote cross leads to a label ( Fig. 4A') with a printed reference "from J.S. Henslow" and "Campanula" in Hooker f.'s handwriting and "C. Darwin 's Ms. No. 279 (γ.)" and "166 " in an unidentified handwriting -probably numbering by Hooker f. or Henslow (Porter, 1986). Just above Darwin's specimen on the left is the Forbes' specimen ( Fig. 4B) from São Nicolau ("n. 35 " according Webb, 1849;[K001134406], identified here as C. fransinea) and with the locality and collector handwritten: "Isle San Nicol. Forbes ". The other two specimens on the upper half [K000865901] (Fig. 4C) and [K001134405] (Fig. 4D), are erroneously annotated "Teneriffe " in the Canary Islands by Hooker f. Leyens & Lobin (1995) were misled by the placement of the label (Fig. 4A') and wrongly attributed them to Darwin. These two specimens represent C. fransinea and most likely collected in São Nicolau. Among the collectors cited in the protologue, Forbes was the only one who visited São Nicolau from March 27 th to 31st, 1822 (Owen, 1833); Darwin only visited Santiago (Vala, 2009). Therefore, the two specimens [K000865901] ( The original material located in G is a specimen collected by Forbes [G00426961]. The locality is not mentioned but should be from Santiago where Forbes made a stopover in early April 1822 (Owen, 1833). On the original determination label on the bottom left is written "Campanula Daltonii Webb" and "sp.
[retained] to be figured, Cap Verd, Forbes" both in Webb's handwriting. This specimen has served for the preparation of the illustrations of C. jacobaea (Webb, 1848: tab. 762) and is here identified as C. jacobaea.
Among these syntypes, Porter (1986) chose Darwin 279 from K [K000865902] (Fig. 4A) as the lectotype of C. jacobaea with a duplicate CGE [CGE03087] but Darwin 279 is in serious conflict with the protologue. Webb (1848) described the species as having a campanulate corolla, three times longer than the calyx-lobes, illustrating the throat as slightly constricted and the roof of the ovary as flat (Webb, 1848: tab. 762). However, the flower of Darwin 279 has a tubular corolla that is barely longer than twice the length of the calyx-lobes, and the roof of the ovary appears to be conical. Leyens & Lobin (1995) correctly identified Darwin 279 as C. bravensis. They chose Forbes s.n.

Campanula bravensis
Etymology. -The epithet bravensis refers to the type locality, the island of Brava; brava meaning "wild" in Portuguese.
Distribution and habitat. -Campanula bravensis has the broadest geographical range: it occurs on the three mountainous southern islands, i.e. Santiago, Fogo and Brava (Hansen & Sunding, 1993;Leyens & Lobin, 1995;Brochmann et al., 1997;Sánchez-Pinto et al., 2005), and is found in the widest range of elevations but in different plant communities according to habitats and islands. In Brava, C. bravensis is found from around 500 m elevation to the highest summits, on rocks regularly submitted to fog with Launaea thalassica N.  (Gardère 1405). In this way, it occupies a diverse range of habitats such as the depths of wet lowland valleys (ribeiras), grassy slopes around Euphorbia tuckeyana Webb (Euphorbiaceae) shrubland, isolated on volcanic ash slopes, on wet rocks, near springs (chupadeiros), or at the entrance of caves with Adiantum capillus-veneris (Pteridaceae), Pteris vittata L. (Pteridaceae) and sometimes with Asplenium adiantum-nigrum L. (Aspleniaceae) for the highest elevation locations. Campanula bravensis is very common across these two islands. On the other hand, in Santiago, it is quite rare and is only known from some field stations in the two main mountain ranges, Serra do Pico da Antónia and Serra da Malagueta, where it generally occurs around wet rocks with Pteris vittata (Pteridaceae) or more rarely along riverbeds, and often in sympatry with C. jacobaea.
Notes. -For Bolle (1861), the concept of his variety bravensis is limited to its type locality, i.e. Brava, because in the protologue he only cited his own collections from this island. Indeed, Bolle made two expeditions to the archipelago in 1851 and 1853 (Salinger & Strehlow, 1991), and collected only in Santo Antão, São Vicente, São Nicolau and Brava (Barbosa, 1962). Later, Andrade (1908) extended the concept of the variety bravensis to Fogo. Then, Chevalier (1935) raised the variety to the rank of species, adopting a broader concept than currently accepted and which included the islands of Brava, Fogo and Santiago. However, some authors have also extended the distribution of C. bravensis (Pettersson, 1960;Sunding, 1973;Eriksson et al., 1974Eriksson et al., , 1979 to W São Nicolau probably owing to the presence in this island of plants with white narrow infundibuliform corollas (C. fransinea), which slightly resemble C. bravensis.
For C. bravensis, the tubular corolla shape is the most noteworthy and dependable diagnostic feature to distinguish it from the other species. However, this feature is absent from the original description of the basionym. Bolle (1861) described the variety bravensis using features of the calyx-lobes, a calyx/ corolla length ratio, and on the indument. That said, he did accurately describe the colour of the corolla: yellowish-white with green veins and with the edge of lamina (i.e. edges of corolla-lobes) slightly purplish. Chevalier (1935) added depth to the description of Bolle (1861) by describing the leaf shape which he used, along with the colour of the flower as a diagnostic feature. However, he did not make any descriptions of the shape of the corolla, even if considered unique in the genus, the character being described much later by Leyens & Lobin (1995).  Sub-frutex 20 -80 cm tall, highly woody in lower part; floriferous stems branched, decumbent to pendulous arising from the base of one or several sterile basal rosettes, sometimes understated, glabrous to glabrescent in the woody basal parts with indument hispidulous toward the extremity, consisting of trichomes 0.15 -0.4 mm long. Leaves: pseudorosette leaves or rosette leaves obovate to spatulate, sometimes falciform, (1.4 -) 2 -6(-9) × (0.4 -)0.9 -2(-2.7) cm, base cuneiform to gradually attenuate concave sometimes asymmetric, apex acute to ± obtuse; cauline leaves narrowly obovate to narrowly elliptic rarely elliptic, (0.7 -)1. 5 -5(-7.5) × (0.5 -)0.8 -1.8(-2.5) cm, base attenuate sometimes asymmetric, apex acute to ± obtuse; margin weakly revolute and sometimes obscurely ondulate, crenelate to serrulate, rarely entire; adaxial side pure green to dark green in vivo, weakly or densely covered with hispidulous to hispid indument, consisting of trichomes 0.2 -0.6(-0.8) mm long, sometimes glabrescent; abaxial side greenish in vivo, venation whitish, hispidulous to hispid indument on primary and secondary veins consisting of trichomes (0.2 -)0.3 -0.6(-0.8) mm long and hispidulous indument on tertiary and ultimate veins consisting of trichomes 0.1 -0.3 mm long, lamina glabrescent. Inflorescences in monochasial pauciflorous cyme or rarely in pluriflorous thyrse. Flowers erect, pedicel 0.9 -4.5 cm long, with the same indument as the stem; axillate by one or two bracts subopposite, ovato-triangular or ovate to narrowly ovate, base semi-amplexicaul, apex acute, with the same indument as the leaves. Calyx: calyx-lobes narrowly triangular, 12 -17 × 3 -9.5 mm, pressed up against the corolla to recurvate, margin weakly revolute; appendages ovate, reflexed, 1 -3 mm long; lobe edges, appendages and median main vein covered with an indument hispidulous to hispid, consisting of trichomes 0.15 -0.7 mm long. Corolla infundibuliform, purple-blue to pure white passing through all intermediate tones; base wide round; tube sub-cylindrical c. 20 mm long, constricted in the middle, 9 -12 mm large in the larger basal part and 7 -10 mm in the constricted part; throat flared, mouth 22 -32 mm large; lobes spreading to recurvate, 6 -10 × 11 -18 mm, apex apiculate; primary external veins micro-hispidulous to hispidous, 0.1 -0.12 mm long. Stamens with glabrous filaments; anthers, 2 -4 mm long. Ovary with glabrous to glabrescent roof, flat, topped by a yellowish-with nectary disk. Style thick, fleshy, (8 -)9 -11 mm long, included in the corolla, stigma trifid and papillose.
Etymology. -The species is dedicated to the Portuguese naturalist João da Silva Feijó (Gardère, 2015) who undertook the first scientific expedition entirely dedicated to the study of the natural history of the Cabo Verde islands between 1783 and 1796 during the "Philosophical Journeys" (Gardère et al., 2019a) and who made the first collections of CVB between 1783 and 1789.
Vernacular names and uses. -"Guinchino" (Cardoso Júnior, 1905;Chevalier, 1935), "Mataquim" (in Água das Caldeiras, Hiemstra H236; Cardoso Júnior, 1905), "Contra-Bruxas" and "Dedal" (Chevalier, 1935;Barbosa, 1961;Leyens & Lobin, 1995;Figueiredo, 1995); the colour adjectives "branco" (white) or "azul" (blue) are sometimes added to the name; "Hortelãoda-Rocha" in Moroços areas (Gardère 1560). However, in the archipelago according Chevalier (1935) "Mataquim" can also designate Corchorus trilocularis L. (Malvaceae) and on the island of Santo Antão, Barbosa (1961) records this name for Antirrhinum orontium L. (Scrophulariaceae). Campanula feijoana was once used for its magical and medical properties. Healers [curandeiros] used the leaves and flowers to make a tea to treat flu (Leyens & Lobin, 1995) and used it as a fetish plant against curses: "Contra-Bruxas" meaning "anti-witch"; the practice was to wash the body with the infused water and to put a few drops on the tongue. This tradition was still very much alive in the beginning of the last century in the Alto Mira region and in the Ribeira Corvo (Gardère, unpubl. data). Cardoso Júnior (1905) also mentioned, without going into detail, the use of this plant for medical purposes by the islanders.
Distribution and habitat. -Campanula feijoana, endemic to Santo Antão, is a rupicolous species confined to steep, moist rocks. It is found from 150 m to 1700 m, in deep and shadowy valleys [ribeiras] (Fig. 7A), surrounding waterfalls, on seeping rock faces or near to springs [chupadeiros], or rarely on river banks; and up to the highest mountainous areas, on rock faces (Fig. 7F) regularly submitted to dense fog. Campanula feijoana grows with ferns, notably Adiantum capillus-veneris (Pteridaceae) and Pteris vittata (Pteridaceae) and with other species that are characteristic components of these rupicolous environments, like Kickxia elegans (G. Forst.) D.A. Sutton (Scrophulariaceae) and Blumea axillaris (Lam.) DC. (Asteraceae).
Etymology. -The epithet vicinituba means the "neighboring trumpet" from the Latin vicinus, "close neighbor" and tuba, "trumpet"; the flowers closely resemble those of the bellflowers from São Nicolau which is easily visible from the rocks of Borda Perdia.
Distribution and habitat. -Campanula vicinituba is only known from its type locality: Borda Perdia, an isolated mountainous area on the far eastern part of Santo Antão at around 1200 m elevation and regularly covered by fog. This microendemic species is strictly chasmophyte and colonizes dry cliffs exposed ESE above Ribeira Brava along with Aeonium gorgoneum J.A. Schmidt (Crassulaceae), Phagnalon melanoleucum Webb (Asteraceae) and Polycarpaea gayi Webb (Caryophyllaceae).
Three Forbes' specimens collected in 1822 in W São Nicolau ([K000865901, K001134405, K001134390]) were chosen by Webb (1848: tab. 762) to be part of the syntypes of C. jacobaea (see under C. jacobaea). Later, in his taxonomic treatment, Bolle (1861) grouped the bellflowers of high-elevation humid areas from W São Nicolau with those from Santo Antão in the type variety "genuina" (see under C. feijoana), and the bellflowers of the more "xeric" areas from W São Nicolau and those from São Vicente in the variety humilis (see under C. monteverdensis).
All the populations from W São Nicolau are described here as new under C. fransinea. This species differs from other CVB species by its narrow infundibuliform corolla of 20 -34 mm long (Fig. 10C). However, it remains quite close to C. vicinituba which has a corolla also narrow infundibuliform but shorter (20 mm) and more flared (Fig. 8D); and differs from the characters indicated in the key. According to its habitats, C. fransinea shows different forms with upright forms in high-elevation shrubland (Fig. 10A), and tufted forms in drier rupicolous areas (Fig. 10F). specimens, date and locality of which are unknown. Despite the poor state of conservation of the Feijó's specimens, some crumpled flowers have been rehydrated, thus restoring the roof of the ovary and corolla shapes. These restored characters revealed that Feijó collected at least two different species: (1) C. bravensis (Feijó V-V-2), most probably from Fogo where he identified a collection under "Fransinea" in his plant list of Fogo between April 1786 and April 1787 (Gardère et al., 2019a); and (2) C. fransinea (Feijó V-V-1), probably from São Nicolau where he collected 24 specimens around 1784 but no detailed list was able to be found (Gardère et al., 2019a). Only one taxon, C. jacobaea var. bravensis, remains from the first revision of CVB by Bolle (1861), of which we retain the combination and the enlarged concept proposed by Chevalier (1935). On the other hand, no morphological, or even geographical group could be considered for the three other Bollean varieties ("genuina", humilis and hispida). Among the diagnostic features used to separate these varieties, Bolle (1861) attached great importance to the indument of stems and leaves but in Bolle's work the interpretation of this feature seems to be strongly related to the microclimatic environmental conditions. Indeed, for the northern taxa, Bolle described the prostrate xero-mesophytic forms from São Vicente and São Nicolau under the variety humilis whereas the erect hygrophytic forms from Santo Antão and São Nicolau are described under the variety "genuina". The same bold true for the southern taxa where the bellflowers from Brava are described under the variety bravensis and the prostrate and strongly hispidulous ones from Santiago under the variety hispida (Bolle, 1861).
As emphasized by Leyens & Lobin (1995), the variations of the indument density observed in the CVB are correlated to the microclimatic conditions of the habitat. The bellflowers that grow on cliffs or ridges exposed to the wind and clouds are often prostrate with a dense indument. On the contrary, those found in deep humid valleys or close to springs and shielded from the wind have an erect habitus and a thinly scattered indument.