Aspidonepsis ( Asclepiadaceae ) , a new southern African genus

Aspidonepsis, an endemic southern African genus, is described and compared to the closely allied genus Aspidoglossum. This newly described genus is composed of two subgenera, Aspidonepsis and Unguilobium. consisting of three and two species respectively. Asclepias diploglossa, A. flava, A. cognata and A. reneensis are transferred to Aspidonepsis. and A. shebae is newly described. All species are discussed, illustrated and a key is given to aid in their identification.


INTRODUCTION
A. A. Bullock's work on the family Asclepiadaceae (1952 to 1967) has received wide acceptance in Africa north of the Limpopo River. In southern Africa, however, his generic concepts and names have seldom been applied. This is explained partly by the fact that his research seldom included southern African plants and partly by the rejection of his work by Dyer (1975).
Unfortunately, three elements detract from Bullock's work: 1, he admitted that his delimitation of genera was only tentative (1952); 2, when resurrecting or expanding existing genera he seldom gave new descriptions for these taxa. As a result, the generic circumscriptions and exact application of some of these names is still unclear; 3, his species concepts were often very broad and there is now growing consensus that some species will need to be re-split.
Most southern African herbaria therefore still follow N.E. Brown's treatment of the Asclepiadaceae as outlined in the Flora capensis (1907)(1908)(1909). However, workers like N.E. Brown had followed the tradition of their time and separated genera using floral differences only. They even separated some genera on the basis of a single character. Phenomena like convergent evolution were seldom taken into account, and workers were unaware that the evolu tion of analogous floral morphologies had taken place within the family. Bullock (1952) was the first to realize that such convergent evolution had taken place and that many genera in the family not only contained a number of unrelated entities, but that these entities could only be identified in terms of consistently produced correlated character combinations. He was the first taxonomist to attempt a phylogenetically based classification for the African members of the tribe Asclepiadeae. What Bullock has done at the generic level, N.E. Brown has accomplished at the specific level. Consequently the work of N.E. Brown (species delimitation) and Bullock (generic delimitation) should be seen as complimentary rather than antagonistic.
Recent investigations concerning the southern African members of the genus Asclepias sensu N.E. Brown have shown that Bullock's generic concepts should be redefined and extended to embrace the taxa of this subcontinent (Nicholas 1981). Bearing in mind that the type species of the genus Asclepias L. is A. syriaca L., the authors agree with Bullock in the exclusion of Asclepias from Africa except as an adventive. The process of moving the southern African taxa of Asclepias sensu N.E. Brown to their correct generic position has already begun (Nicholas & Goyder 1990). The authors understand the desirability of giving a brief generic synopsis of the subtribe Asclepiadineae in Africa at this early stage of their work. However, as a number of genera still need to be: 1, resurrected from synonomy; 2, newly described; 3, extensively redefined; they feel that it is at present unwise to publish information that may change as their research progresses. 1, a globose, fusiform or napiform tuber just below the soil surface; 2, a single erect stem (rarely up to 3 in A. flava); 3, spreading to ascending linear to narrowly elliptic leaves which are ranked up the stem; 4, inflorescences gathered together at the top of the flower ing stem, even if nodally produced; 5, persistent inflorescence bracts, often grading with the leaf system; 6, cucullate corona lobes which are produced 0.5 to 1.8 mm above the insertion of the corolla; 7, a saccate corona lobe cavity; 8, wishbone-shaped pollinaria, with semicircular to hemiovoid pollinia.
Aspidonepsis is confined to high altitude grasslands of the Drakensberg and its foothills, although outlying popu lations of some species may be found in mountainous situations as far south as Grahamstown and on montane 'islands' nearer the Transkei-Natal coast. The northern limit of distribution of this endemic southern African genus is the eastern Transvaal. Species are usually, but not always, found in situations subject to annual burning. Popu lations are intermittent in the wild and usually consist of few widely dispersed individuals. Occasionally up to three tubers are produced in a connected series, possibly representing subsequent year's growths.
Aspidonepsis bears a number of similarities to the genus Aspidoglossum (Table 1), and it is the authors' opinion that the two genera may have originated from the same distant, ancestral stock. However, if this is the case, then the two taxa have since evolved along very different lines, for a number of major disjunctions in morphology now exist, such as the aggregation of inflorescences near the stem apex and the central cavities in the corona lobes of Aspidonepsis. In contrast Aspidoglossum bears inflores cences that are produced along the length of the stem and there is no corona lobe cavity.
The affinity of these two genera can be clearly seen in the corona lobe and pollinarium morphology of Aspidoglossum delagoense (Schltr.) Kupicha, which is very similar to Aspidonepsis (Figure 1). However, all other features of this species place it clearly within Aspidoglos sum, of which A. biflorum E. Mey. is not only the type species but also typical of the genus as a whole (Kupicha 1984). Aspidoglossum has more species and is morpho logically more diverse than Aspidonepsis.
The five species recognized in Aspidonepsis fall into two well-defined groups that require recognition at subgeneric level. The first group is characterised by spreading or ascending corolla lobes and cup-or dish-shaped corona lobes with a tooth-like appendage projecting from the floor of the corona lobe cavity. The second group has reflexed corolla lobes and corona lobes with a more angled outer margin and no tooth-like structure projecting from the floor of the corona lobe cavity.
A total of 187 pressed specimens were examined during the course of this study from the following herbaria: BOL, CPF, GRA, J, K, NBG, NH, NU, PRE, SAM and TCD*. Additional data were obtained from spirit collections and supplemented by observations in the field. * Herbarium abbreviations are taken from Holmgren el a l. (1990).
This subgenus is composed of three species: Aspidonep sis diploglossa, A. flava and A. cognata. For a number of years these three species were considered conspecific, and lumped together under the oldest name, viz. A. diploglossa. However, although all three species are vegetatively similar, close examination shows that they are distinct entities with very different floral morphologies (Nicholas 1987). They are usually found in annually burnt or grazed, high to medium altitude, montane grasslands. They are found along the Natal-Transkei Drakensberg, and on scattered island mountain ranges in the eastern Cape and Natal midlands. The flowers of this subgenus are predominantly yellow or yellow-green, although occa sionally flowers with brown or purple markings can be found.
Aspidonepsis diploglossa is found in annually burnt montane grasslands, normally on south-or east-facing hill side slopes or mountain plateaux. Usually, but not always, occurring in wettish areas. Collectors often report it as rare, although a great many collections exist. It is usually found growing at altitudes ranging from 1 500 to 2 400 m, but occasionally also at lower altitudes. Plants flower from October to January. The tubers of this plant lie just below the soil surface, and when sectioned reveal white, woody flesh that oozes sticky, milky latex.
A. diploglossa, a mountain-loving species, exhibits a rather strange distribution. It may be found at high altitudes around Grahamstown and Hogsback in the eastern Cape, then there is a gap in the Transkei Drakensberg (which may be an artifact caused by poor collection in this area) and then it occurs abundantly along the Natal Drakens berg and its foothills as far as Van Reenen's Pass. After yet another gap it is found again in the Wakkerstroom area. A. diploglossa may also inhabit mountain islands in the Natal midlands at places such as Inanda. Greytown and Weenen. However, it occurs in the most unlikely place near the southern Natal coast at the Umtamvuna Nature Reserve, where it grows at an altitude of only 350 m. This nature reserve is well known scientifically because it lies within the narrow belt of Natal Group sandstone in the coastal region between Port Shepstone and Port St Johns. Its rich flora includes a number of rare plants and en demic species. However, the occurrence of Aspidonepsis diploglossa at such a low altitude and so near the sea, is surprising and inexplicable ( Figure 3).
Unfortunately, when R. Schlechter described Asclepias schizoglossoides in 1894 he not only failed to cite the specimens he examined, but was also unaware that he was dealing with an already described taxon. Turczaninow had named this species Gomphocarpus diploglossus in 1848. citing Ecklon 23 as the type. N.E. Brown picked up these two errors when preparing the Asclepiadaceae for Flora capensis, and in correspondence with Schlechter discovered that the latter taxonomist had based the name Asclepias schizoglossoides on a Barber specimen 'probably collected in British Kaffraria'. As a result, N.E. Brown (1908) suspected that the specimen may be part of Mrs Barber's gathering numbered 35. N.E. Brown's selection of Barber 35 as the type of the name Asclepias schizoglossoides for Flora capensis was probably correct. However, due to the destruction of Schlechter's asclepiadaceous collections housed at Berlin herbarium during the Second World War, we cannot confirm this. In this paper we have, therefore, chosen Barber 35 (K) as the neotype of the name Asclepias schizoglossoides.
W.H. Harvey has written (in pencil) on two Trinity College Dublin herbarium (TCD) sheets of this species, the name Gomphocarpus luteus (var.) 0 heterophyllus. This name was never validly published, and must be con sidered nothing more than a manuscript name.
Aspidonepsis diploglossa differs from A. flava and A. cognata in possessing longer (occasionally narrower) leaves, a deeply cleft anther appendage, yellow to yellowbrown flowers and a simple cup-shaped corona lobe, the upper proximal ends of which are no more than blunt rounded shoulders level with the style apex. See Table 3.

# Discontinuities between A. diploglossa and the other two taxa; • discontinuities between A. fla va and the other two taxa; + discontinuities between A. cognata and the other two taxa; $ discontinuities between A. fla va and A. cogn ata; * discontinuities between all three taxa.
Rootstock a globose tuber, occasionally several connec ted in series, 6-10 X 4 -9 mm. Stems usually 1, rarely as many as 3, slender, erect, 180-475 mm long. Leaves spreading to erect, lanceolate, linear to narrow-elliptic, 7-83 x 0.5-6.0(-7.0) mm; apex acuminate or rarely acute; base petiolate to cuneate. Inflorescences 1-3(-6) per plant, 1-3 per stem, 4-18(-24)-flowered; bracts present at anthesis, 2.6-5.3(-7.5) X 0.15-0.5 mm; pedun cles (4->10-175 mm long. Flowers 3-5 (-6 ) x 5 -8 mm; pedicel 5-11 mm long. Calyx: lobes lanceolate, 2.0-3.6(-4.0) X 0.7-1.2 mm. Corolla', lobes ovate, oc casionally elliptic, free to the base, 3.5-5.0 x 2.0-3.2 mm, inside greenish yellow or yellow, outside yellowish green, pale yellow or yellow with a purple apex, margins occasionally slightly revolute. Corona lobes produced 0.5-0.8 mm above corolla, cucullate-crateriform, in side view boxing glove-shaped, 1.0-1.6 mm long, upper proximal ends extending into 2 short (0.25 -0.70 mm), subulate or arm-like appendages that meet and are then reflexed back to point to distal end of lobe, lower than style apex, distal end dilated and bowl-shaped with upper margin overtopping the style apex (even if only slightly); cavity crateriform, 0.4-0.7 mm deep with a central sausage-shaped appendage projecting 0.4-0.7 mm above cavity margin; orange-yellow, golden yellow, yellow-green or yellow. Aspidonepsis flava is usually found growing in annually burnt montane grasslands. Colonies are usually scattered and occur at altitudes between 600 and 2 000 m, rarely at altitudes as low as 450 m. Distributed from Grahams town in the eastern Cape through Transkei to Natal. This species is commonly found in the Drakensberg or its foothills, although it can be found in the midland and coastal belts if mountainous areas provide it with a suitable refuge (Figure 4). A. flava flowers in the midsummer months between November and January, although there is one record of a plant flowering in October. This taxon was first described by N.E. Brown in Flora capensis (1908), and is abundant in southern Natal and the Transkei interior. The limits of its southern distribu tion is near Grahamstown where a few specimens have been collected. Plants grow in small colonies in annually burnt grasslands, and usually occur on hillside slopes amongst scattered rocks where they receive some protec tion from grazing animals and fire. Plants may have up to three tubers connected in series, each probably representing a previous year's growth. Like A. diploglossa these globose tubers are found just below the soil surface, and have white, latex-filled flesh.
Aspidonepsis cognata may be found scattered in annually burnt (but not always) montane grassland, usually occur ring in river valleys or near streams where the soil is quite damp. This graceful species flowers between November and December (although there is one record for October), and occurs at altitudes between 1 200 and 2 100 m, rarely lower. A. cognata is confined to a small area in the south ern Natal and northern Transkei Drakensberg ( Figure 6). It is unfortunate that N.E. Brown (1908) chose Schlech ter 6469 as the type of Asclepias cognata, because this collection is not typical of the species as a whole. However, all specimens of Schlechter 6469 examined, although not typical, clearly belong to this species. In appearance Hilliard & Burtt 7855 is more representative of the species.
Aspidonepsis cognata can be distinguished from the other species in subgenus Aspidonepsis by its larger flowers, larger corona lobes which are broadly helmet shaped, wider anther appendages and its longer translator arms and pollinia. (See Table 3). The corona lobe shape is highly diagnostic, in particular the subulate or arm-like proximal appendages which may project over the style apex, and the raised distal end which is usually truncated along its upper margin and overtops the style head.  and broader, margins noticeably revolute. Inflorescences 4-11-flowered, bracts present at anthesis and grading in size and shape with leaf system. Flowers purple, brown, lilac and yellow. Corolla reflexed, lobe apices ascending, abaxial surface pubescent. Corona produced high on staminal column, ± 1 mm above corolla; lobes with proximal appendages deltoid-falcate with obtuse apex level with or projecting over style apex, distal end of corona with arm-like appendage reflexed into corona lobe cavity (A. reenensis) or without appendage (A. shebae). Staminal column: anther wings ear-like in outline; anther append ages pescapriform, deeply cleft at apex (A. shebae), or ovate to rectangular and occasionally cleft at apex 04. reenensis). Style apex with slightly thickened, undulate margins. Pollinia: distal end noticeably dilated and narrowed towards proximal end. Etymology: from the Latin words ungu(is) (claw) and lob(us) lobe, in reference to the claw-shaped corona lobes of this subgenus (Table 2).
There are two species in subgenus Unguilobium, viz. A. reenensis (the type species) and A. shebae. Both are confined to mountainous areas of the southern African Drakensberg. A. reenensis is found in the southern regions of this mountain system (namely Natal), whereas A. shebae is found in the northeastern region (the eastern Transvaal). As such, these species are quite widely separated geo graphically ( Figure 6). Although probably related (even if somewhat distantly), they can be easily told apart using corona lobe and anther appendage shape. A Nicholas & D.J. Goyder,comb. nov. Type: South Africa,Natal,Van Reenen,Wood8635 (K! holo.;GRA!,NH!,PRE!,SAM!,iso.).
A. shebae probably occurs in montane grasslands, and is restricted to high altitude areas (1 400 to 2 100 m) of the Pilgrim's Rest region of the eastern Transvaal ( Figure  6). Plants flower in December-January, and according to one set of collections is said to be frequent.
Vegetatively A. shebae is very similar to A. reenensis, and it is probably closely related to this species (Table  4). In floral morphology, however, these two species differ greatly, especially in corona lobe structure ( Figure   10).