New synonyms and a new name in Asteraceae: Senecioneae from the southern African winter rainfall region

A review of the genera Othonna and Senecio undertaken for the forthcoming Greater Cape plants 2: Namaqualand-south- ern Narnib and western Karoo (Manning in prep.) led to a re-examination of the taxonomic status of several species. This was facilitated by the recent availability of high-resolution digital images on the Aluka website (www.aluka.org) of the Drege isotypes in the Paris Herbarium that formed the basis of many species described by De Candolle in his Prodromus systematis naturalis regni vegetabilis. These images made it possible to identify several names whose application had remained uncer­ tain until now. Each case is briefly discussed, with citation of additional relevant herbarium specimens. The following species are reduced to synonomy: O. incisa Harv. is included in O. rosea Harv.; O. spektakelensis Compton and O. zeyheri Sond. ex Harv. are included in O. retrorsa DC.; S. maydae Merxm. is included in S. albopunctatus Bolus, which is now considered to include forms w ith radiate and discoid capitula; S. cakilefolius DC. is included in O. arenarius Thunb.; S. pearsonii Hutch, is included in O. aspertdus DC.; S. parvifolius DC. is included in S. carroensis DC.; S. eriobasis DC. is included in S. erosus L.f.; and S. lobelioides DC. is included in S. flams (Decne.) Sch.Bip. The name S. panduratus (Thunb.) Less, is identified as a synonym of 5. erosus L.f. and plants that are currently know n under this name should be called S. robertiifolius DC. The


INTRODUCTION
The South African species o f Othonna L. and Senecio L. were last revised at a regional level by Harvey (1865). Since then, the sum m er rainfall species from KwaZulu-Natal have been studied intensively by Hilliard (1977) but many species in the southern African winter rainfall region remain poorly understood. The recent availabil ity o f high-resolution digital images o f type material o f many African taxa (ww w.aluka.org), especially the Paris isotype material o f the Drege collections that formed the basis o f a num ber o f De C andolle's (1838) species, has made it possible to identify several taxa that were never seen by Harvey (1865) and whose identity has remained uncertain. During the preparation o f a review o f the tribe Senecioneae (M anning in prep.), various nomenclatural and taxonomic issues were encountered that can now be addressed. The application o f some o f these names and other taxonomic and nomenclatural issues arising during the study are dealt with here.
The digital images, or virtual herbaria, available online through Aluka and other sites such as the Herbarium o f the Linnean Society o f London (www.linnean-online.org) are often adequate for providing confirmation o f the identity of names where the gross morphology o f the taxa in ques tion is diagnostic but they cannot replace microscopic examination o f actual material for critical features. We have, therefore, avoided making any taxonomic infer ences in instances where the identity o f the digital image depends on examination o f such micro-characters. In addi (Table 1). These vegetatively similar, deciduous geophytes from the Western Cape are tuberous perennials produc ing slender, often scandent, annual stems bearing perfoli ate or amplexicaul leaves and terminating in pedunculate capitula. Their remarkable vegetative similarity led Row ley (1994: 174) to treat both radiate and disciform plants in a single, variable species under the name O. filicaulis. He may be correct, but until the situation is fully investigated we continue to treat them as distinct.
Othonna perfoliata is distinguished from O. filicau lis by its generally fewer phyllaries (8-10 vs 10-12) and radiate marginal florets, although in some collections the rays are reduced and rather small. The marginal florets in O. filicaulis, in contrast, are filiform and rayless, w ith the truncate corolla less than half as long as the style. All fruiting material o f O. perfoliata that we have seen has the mature pappus in the marginal florets 5-8 mm long and ± as long as the cypselae or only slightly longer, whereas in O. filicaulis the pappus often elongates greatly, reaching 10-20 mm long, therefore 2-A times as long as the cypselae at maturity. O. filicaulis as currently under stood (e.g. Goldblatt & Manning 2000) is widespread in the southern African winter rainfall region, mainly on sandy, often coastal flats, from southern Namibia as far east as Uniondale, whereas O. perfoliata is essentially restricted to montane habitats in the southwestern Cape, where it occurs on rocky slopes between the Bokkeveld Mountains and Caledon (Goldblatt & Manning 2000).
The epithet perfoliata was first used in this context by Linnaeus f. (1782) in the combination Cineraria per foliata L .f, based on material collected at the Cape of Good Hope by Thunberg. There are no sheets under this name in the Linnaean Herbarium at Stockholm (S) but the Herbarium o f the Linnean Society (LINN) contains two sheets labelled as such, one o f them a radiate plant (LINN1000.32) and the other one evidently disciform (LINN1000.33). The protologue. unfortunately, makes no mention o f the condition o f the capitula but the descrip tion o f the leaves as ovate-cordate and o f the peduncles as elongated, with a solitary capitulum [foliis ovatis cordatis amplexicaulibus, pedunculis unifioris elongatis] matches the radiate-flowered specimen more closely than the disciform one. in which the leaves are lanceolateundulate and the inflorescence sparsely branched. It is in the sense o f a radiate taxon that Jacquin (1797)  At the time that he figured the radiate taxon under the name Othonna perfoliata. Jacquin (1797) described a second, similar taxon under the name O. filicaulis Jacq. This species was distinguished from O. perfoliata by the narrower, lanceolate-undulate leaves and by the smaller, linear rays. The ligulate condition o f the mar ginal florets is clearly described and is also illustrated in a detail o f a marginal floret but the rays are not evident in the draw ing o f the whole plant, which thus appears to be disciform. Both De Candolle (1838) and Harvey (1865) treated the species as circumscribed by Jacquin although neither knew it from actual material. The name later became associated with the disciform species, pre sumably because Jacquin's painting o f the whole plant appeared to represent a disciform individual, and it is this application that is current (e.g. Goldblatt & Man ning 2000). This transfer in the application of the name O. filicaulis to the disciform taxon appears to date to the Flora o f the Cape Peninsula (Adamson & Salter 1950), in which the name O. filicaulis is explicitly applied to the disciform species with filiform, truncate marginal corol las 'previously misidentified as O. perfoliata' (Adamson & Salter 1950: 820). In any event, the protologue o f O. filicaulis makes it quite clear that the name is properly applied to a radiate plant.
Examination of available herbarium specimens o f Oth onna perfoliata shows that the species exhibits a wide range of leaf shapes, ranging from suborbicular and plane (typical O. perfoliata) to lanceolate and undulate (typical O. filicaulis). The ray florets also vary from well-developed and oblong to smaller and narrow. We therefore conclude that O. filicaulis is conspecific with O. perfoliata and place the name in synonymy under it.  Lessing (1832: 89) reveals that he did not in fact publish this name. It first appears as D. lingua in De Candolle (1838: 471), where it is mis takenly attributed to Lessing as being based on the basio nym Othonna lingua Jacq. What Lessing (1832) actu ally published was the combination Doria digitata (L.) Less., based on Othonna digitata L., for a similar taxon with toothed leaves, followed by a list o f names that he regarded as conspecific with it. Among them was Othonna lingua sensu Jacquin (1797). Jacquin (1797) was in any event also not the author o f the name since he cited the Svstema vegetabilium o f 1784 (edition 14, Murray 1784), which in turn refers the name to Linnaeus fil. (1782). This is in fact the first appearance o f the name, which is thus correctly attributed to Linnaeus f. as O. lingua L.f.
There is no material under the name Othonna lingua in either LINN or S but there is a specimen under this name in the Thunberg herbarium (UPS-TIIUNB20882), which therefore represents the type. This specimen was cited by Harvey (1865: 342) and Harvey (1865) arose from an initial m isunderstanding by Jacquin (1797), whose illus tration and description apply to an erect-stemmed, tuber ous geophyte with petiolate radical leaves, lanceolate cauline leaves, and disciform capitula. Since all o f these are in fact radiate, this is strong cir cumstantial evidence that he had a radiate taxon in mind when describing O. lingua. There is also no indication as to whether the epithet refers to the shape o f the leaves or to the presence o f rays but the description o f the leaves as ovate-lanceolate and semi-amplexicaul suggests that it more probably alludes to the latter. In any event, the Thunberg collection, which is evidently the type, fixes the application o f the name.

4.
Othonna lingua is a species with radiate capitula and is, according to our interpretation o f the type, conspe cific with O. bulbosa (see discussion above). It is, there fore, necessary to consider the identity o f the disciform species to which this name has been applied in many her baria. The collections under this name represent a tuber ous geophyte with disciform capitula and ± erect, annual stems bearing variously oblanceolate to ovate leaves, the lowermost leaves ± tapering to the base and ± truncate or sessile, and the upper leaves mostly auriculate.
We have examined herbarium material filed under the name O. lingua and are able to distinguish two sets o f populations. One represents a western, coastal spe cies occurring from Namaqualand to the Olifants River, mostly with fleshy, oblanceolate leaves with conspicu ously revolute margins, capitula in which the phyllaries are connate in the basal third, and disc cypselae with short, caducous pappus bristles 1-3 mm long. A sec ond series o f plants with ± plane leaves occurs inland, from the Cedarberg eastwards to Port Elizabeth, and is distinguished by the phyllaries connate for ± half their length or more, and most strikingly by the disc cypselae entirely lacking a pappus (very rarely with one or two short bristles on the outer disc florets in some collections from the Little Karoo).
Only two species o f Othonna are known in which the disc florets lack a pappus and both were described by De Candolle (1838). O. gymnodiscus (DC.) Sch.Bip. was based on a plant collected by Ecklon near Port Elizabeth, and O. semicalva (DC.) Sch.Bip. on a collection made by Drege in the O lifant's River Valley. The two taxa were distinguished primarily by differences in leaf shape (respectively oblong-ovate and auriculate vs linear-lan ceolate and sessile). Both taxa were known to De Can dolle (1838) from the type specimens only. Collections made since then have filled in the distribution between the type localities and also suggest that the purported dif ferences between them in leaf shape and size o f capitula are not significant. Leaves in herbarium material range from narrrowly lanceolate to obovate, with the leaf base narrowed or ± auriculate, and the phyllaries vary from 6 -10 mm long, sometimes even on the same plant depend ing on their stage o f development. Our examination of type material o f both names leads us to conclude that they represent forms o f a single species, for which we choose the name O. gymnodiscus as being most appropriate. The name O. semicalva is accordingly reduced to synonymy.
The second series o f populations from the west coast with pappus bristles present in the disc florets appears to represent an undescribed species but further study is required to assess this.

5.
Only three species of Othonna are known with pur ple or magenta rays, all of them tuberous geophytes from Namaqualand. They are distinguished essentially by the shape o f the leaves, specifically the degree of lobing. At the time that Harvey (1865)

6.
Othonna retrorsa DC. is an easily recognizable, cushion-form ing perennial with a many-headed caudex producing rosettes o f very distinctive, leathery, reticu late-veined leaves. These are oblanceolate with carti laginous margins hearing few to many pale, patent or retrorse denticles. The leaves are persistent at the base and their dried remains accumulate around the short stems, gradually decaying into a fibrous mass. One or more, sparsely branched flowering stems are produced from each rosette. The species was described by De Can dolle (1838)  . These plants were grow ing in cracks in exposed granite sheets, which is con sistent with their more compact and dw arf habit. This habitat struck Compton (1953) as significantly differ ent from that o f O. retrorsa, which lie knew from rocky slopes in somewhat deeper soils, but subsequent collec tions o f typical O. retrorsa have been made from bare rock cracks and sandy depressions on granite (O liver 5965 NBG) and the variation in growth form is evidently purely ecological. The salient differences between the three varieties are given in Rowley (1994

7.
Cacalia rigida Thunb. is a thorny shrublet w ith obovate, sparsely denticulate leaves and shortly pedunculate, disciform capitula (Thunberg 1823 Candolle's epithet was preoccupied in the genus. Thun berg's Cacalia rigida is actually the earliest name for the taxon but this epithet is also preoccupied in Othonna. and Schultz's new name is thus the correct name for the spe cies in Othonna. The species, which is rarely collected, has been overlooked in southern African checklists for the family under any o f its available names (Welman 2006). We provide the complete nomenclature here.

8.
Senecio alhopunctatus Bolus (1887) was described from a single gathering o f several plants collected in Namaqualand at Klipfontein, west o f Steinkopf on the old Steinkopf-Port Nolloth railway. S. alhopunctatus is a subshrub with terminal tufts of pinnatifid leaves, the lobes tipped with characteristic pale thickenings or calli that give the species its name, and solitary, radiate capitula on long peduncles. The plant was described as glabrous but careful examination of the type material shows the leaves and peduncles to be scantily clad in minute, sessile glands. No similar plants with radiate capitula have been collected since then but several collections have been made of plants that are vegetatively identical to S. albopunctatus but differ from it in having discoid capitula. The first o f these collec tions was made in 1935 by R.H. Compton at Klipfontein hill, west o f Steinkopf (Compton 5442 NBG), at the type locality o f S. albopunctatus. It was identified tentatively by Compton as S. albopunctatus following careful comparison with the type, with the comment that apart from the slightly less indurated leaflet tips, he could find no significant dif ferences between the two species and that he, therefore, interpreted his material as a rayless variant o f S. albopunc tatus. Since then several additional collections o f this rayless form have been made in the Richtersveld, all of them agreeing exactly with Bolus's and Compton's original gath erings and also proving very good matches with the type of S. albopunctatus, apart from the clear absence o f rays.
The occurrence o f both discoid and radiate forms in a single species is rare in Senecio but not unknown. Among species from the Cape Floristic Region (CFR) it has been recorded in S. agapetes C.Jeffrey and S. crispus Thunb. (Goldblatt & Manning 2000), and in several species from KwaZulu-Natal, including S. conrathii N.E.Br., S. hypochoerideus DC., S. polvodon DC. and S. poseideonis Hilliard & B.L.Burtt (Hilliard 1977). Such species are scattered throughout the genus. In most instances one o f the forms is dominant, with the other rare or occasional, but in a few species both forms are common. A strikingly similar example is provided by S. erosus L.f., in which both radiate and discoid plants have been collected near Mooreesburg (Helrne 213/ [dis coid] and Helme 2339 [radiate] NBG) in this otherwise entirely radiate species. We have no hesitation, therefore, in following Compton in treating both discoid and radi ate forms as a single species.
However, it is now clear that the discoid material of Senecio albopunctatus from the Richtersveld is indistin guishable from S. maydae Merxm. (1960), which was based on several collections from the Huib Hoch Plateau and adjacent hills in southern Namibia (Dinter 1932

9.
Among the 20 annual species o f Senecio recog nized by Harvey (1865), are five with mauve or pur ple ray florets. Two o f these species are well known: S. elegans L. is essentially a species o f sand dunes along the western and southern Cape coast, from Saldanha in Western Cape to Port Alfred in Eastern Cape, whereas 5. arenarius Thunb. is widely distributed on sandy and gravelly flats and in washes along the west coast and interior, from central Namibia through much o f the west ern half o f South Africa as far south as Agulhas in West ern Cape (Goldblatt & Manning 2000). Both are ± glandular-pubescent herbs, extremely variable in leaf form [polymorphous is how they are described by Harvey (1865)] but readily separable by their different involu cres: ± cylindrical and with a few subulate bracteoles in S. arenarius; ± campanulate and closely enveloped at the base by several imbricating, lanceolate, black-tipped bracteoles in S. elegans. The remaining three species in the group were known to Harvey from the type speci men or description alone and they remain poorly known today.
One o f them. Senecio cakilefolius DC., was based on a collection made by Drege at Silwerfontein, southeast of Springbok. This material was not seen by Harvey (1865), who relied entirely on De Candolle's (1838) description. The species was distinguished from S. arenarius by its glabrous stem and leaves, and supposedly larger capitula but examination o f the type collection confirms that the plants are actually sparsely but quite evidently glandularpubescent and the capitula are no larger than commonly encountered in S. arenarius. The name has subsequently been applied rather indiscriminately to any arenariuslike plants with less than the usual pubescence. Subglabrous or thinly pubescent plants o f the cakilefolius type are common between Springbok and Kamiesberg but also occur further south near Clanwilliam and through the Tanqua River Basin to Whitehill, whereas more densely pubescent plants o f the arenarius type are wide spread. With a full range o f material from Namaqua land and the West Coast now available, it is clear that there are any number o f intermediate conditions from almost glabrous plants to those with sparsely glandularpubescent stems and leaves to densely glandular-pubes cent plants. The size o f the capitula also varies greatly and independently o f the vestiture. Populations from Namaqualand and the Bokkeveld Plateau have slightly larger capitula with phyllaries 5-7 mm long compared with those from the Cold Bokkeveld and Little Karoo, in which the phyllaries are 4 -6 mm long, but this small size difference is not correlated with other differences, and is not uncommonly encountered in other species in the genus. With the ample collections now at our disposal, it appears to us that the concepts o f 5. arenarius and S. cakilefolius represent the extremes o f a continuous range o f variation and we thus treat them as a single species.
Senecio a re n a riu s Thunb., Prodromus plantarum capensium: 158 (1800b). Type: South Africa, without precise locality or date, Thunberg UPS-TH U NB19545 (UPS-THUNB, holo.-m icrofiche!). 10. Senecio pearsonii Hutch. (1917) was based on a collection from the Kamiesberg and diagnosed against S. hypochoerideus DC., from which it was distinguished by its more finely serrate leaves and striate-papillate, as opposed to uniformly pubescent, achenes. Actually, the achenes o f 5. hypochoerideus, like those of most species o f Senecio, are striate-hispid (Hilliard 1977). Examina tion o f the type o f S. pearsonii confirms H utchinson's opinion o f the Kamiesberg material in respect o f S. hypo choerideus but shows his species to be a perfect match for the closely allied S. asperulus DC. This species dif fers primarily from S. hypochoerideus in its narrower leaves 2-10 mm wide, with more finely serrate margins with simple teeth vs broader, doubly-serrate leaves 10-40 mm wide in 5. hypochoerideus (Hilliard 1977). At the time o f Hilliard's (1977)

Senecio carroensis DC. (1838) was based on a collection made by Drege along the southern margin o f the Great Karoo at Kendo [Kendouslaagte] between
Klaarstroom and Willowmore. The species is a slen der. laxly branched shrublet with deeply incised, almost bipinnatisect leaves, the lobes narrow and deeply toothed. The sparsely woolly or subglabrous branches are characteristically yellowish striate when young, often flushed purple when older. The radiate, yellow capitula are in lax corymbs, with glabrous phyllaries.
Ample material o f the taxon has now been collected from the drier mountains o f the Little Karoo westwards to Karoopoort and thence northwards through the Cold Bokkeveld and Swartruggens as far as the Bokkeveld Mountains. This material displays significant variation in the shape and size o f the leaf lobes, a feature that was already evident to De Candolle. Although mostly nar rowly oblong or linear and 2 -7 mm long, the lobes in some collections are much reduced, almost quadrate, and 1-2 mm long. These plants are an exact match tor Sene cio parvifolius DC., another o f D rege's collections, from the Kamiesberg in Namaqualand, and which was distin guished from S. carroensis essentially by its smaller leaf lobes. Further collections from the Kamiesberg confirm the general constancy o f this leaf character among the Namaqualand plants but also include plants in which the leaf lobes are longer and narrower and thus indistin guishable from more xeromorphic forms o f S. carroen sis. With this larger range o f material now available, we conclude that these two species represent extreme leaf forms in a single species.

12.
Among the taxa with yellow, radiate capitula that were included in Harvey's (1865) Sinuosi are two spe cies characterized by a short, vertical rhizome closely covered by imbricating leaves, with the base o f the peti oles expanded and encircling the rhizome. The invest ing leaf bases form an almost bulb-like structure, with their inner faces densely covered with woolly hairs that also run up along the margin o f the sheath and onto the petiole for a short distance. These woolly rhizomes set these two species apart from others in the group but the difference between them is not clear. Senecio eriobasis DC. (1838), based on a collection from Worcester, was distinguished from S. erosus L.f. (1782) by its glabrous vs scabro-pubescent leaves and involucre. Like many other species o f Senecio, S. erosus is extremely variable in the degree o f developm ent o f the vestiture, varying from sparsely to densely pubescent, with no clear dis tinction between the two conditions. More significantly, however, exam ination o f the type o f S. eriobasis shows the leaves and peduncles to be distinctly pubescent and quite indistinguishable from those o f S. erosus, and there is no doubt that Goldblatt & Manning (2000) were cor rect in their conclusion that the two could not be sepa rated. We accordingly formally include S. eriobasis in the synonomy o f S. erosus. The species is distributed from Namaqiialand to the southern Cape.
Examination o f the type o f Doria incisa Thunb.. treated as a synonym o f Senecio erosus by Harvey (1865), confirms that it has radiate capitula, despite its initial placement in Doria, and that it cannot be distin guished from S. erosus. Harvey (1865) also included in S. erosus the species described by Thunberg (1823)  There are two specimens labelled Cineraria pandu rata in UPS-THUNB. One o f them, a complete plant w ith the diagnostic erect, villous rhizome o f Senecio ero sus, is also labelled Doria incisa and constitutes the type o f that name. The second specimen, securely identified by the locality data as the type o f Cineraria pandurata, comprises just an inflorescence and a single, unattached basal leaf. This incomplete specimen, in our opinion, might equally be assigned either to S. erosus or to the taxon that is currently known under the name S. pan duratus but Thunberg's (1823)

13.
The name Senecio panduratus (Thunb.) Less, wa misapplied by De Candolle (1838) to plants that resem ble 5. hastatus L. in general appearance but which are distinct from it in their sparsely branched corymbs with much larger capitula. Both S. hastatus and S. panduratus sensu DC. have a short. ± horizontal rhizome, and radi cal leaves with long petioles and inciso-pinnatifid blades, and the stems and leaves are thinly or densely pubescent w ith a mix o f short, glandular hairs and longer eglandular hairs. True S. hastatus is characterized by few-to wellbranched corymbs of up to 20, relatively small, cylindri cal capitula, 9-12 x 5-7 mm, with 12-14 phyllaries. It is widely distributed in moister situations from Ceres in the southwestern Cape eastwards to Lesotho and the Free State (Hilliard 1977;Goldblatt & Manning 2000). The taxon currently known under the name S. panduratus, in contrast, has sparsely branched corymbs o f (1-)3-10, larger, campanulate capitula, 1 0 -1 2 x 8 -1 2 mm. with 20-24 phyllaries. It has a more restricted distribution in the interior southwestern Cape, in more arid environments. The name S. hastatus is in fact a synonym of S. erosus (see above) and the plant currently known under that name thus requires a new name. Examination o f the type o f S. robertiifolius DC. (1838), a poorly known taxon based on a collection made in the Kamiesberg, confirms that it precisely matches the plants currently identified as S. hastatus and this name is therefore available for use.

14.
Senecio lobelioides DC. (1838) was based on material collected by Drege on the Farm Silwerfontein, midway between Springbok and Kamieskroon. It was not seen by Harvey (1865), who merely repeated De C andolle's description. Examination o f isotype material at Paris confirms that it is readily recognized among the other annual species by being completely glabrous, and with characteristic leaves, the lower conspicuously petiolate with ovate-reniform blades, and the upper sessile and cordate-amplexicaul. Although becoming progressively sm aller towards the end o f the branches, the upper leaves retain their distinctive shape, giv ing the inflorescences a rather leafy character. Another distinctive feature o f the species is the narrowly cylindrical capitula, which are obscurely radiate with very short rays barely exceeding the involucre. Drege's material o f 5. lobelioides, how ever, is indistinguishable from collections at SAM that have been identified as Senecioflavus (Decne.) Sch.Bip. subsp. flavus by both Merxmiiller (1967) and later, in 1988. by Aaron Lister (now Department o f Botany and Plant Pathology. Oregon State University). Although we have not been able to examine type material o f S. flavus, the protologue (Decaisne 1834), as well as the descrip tion and accompanying illustration in Boulos & Hind (2002), give us no reason to doubt this opinion and we accordingly treat S. lobelioides as a synonym o f S. fla vus.