Numerical taxonomic studies in the subtribe Ruschiinae ( Mesem-bryanthemaceae ) — Astridia , Acrodon and

A numerical taxonomic study of three genera of the Ruschiinae (Mesembryanthemaceae) is presented. Seven species are recognized in Astridia. Four species are recognized in Acrodon and two new combinations are made: A. leptophyllus (L. Bol.) Glen and A. duplessiae (L. Bol.) Glen. Five species are recognized in Ebracteola and two new combinations are made: E. wilmaniae (L. Bol.) Glen and E. fulleri (L. Bol.) Glen. This study is largely based on the cited herbarium material, and the characters used are mainly the following: dimensions of plants, leaves and internodes, number and dimensions o f parts of flower and fruit, colour of petals, and colour, dimensions and surface structure o f seeds.


INTRODUCTION
The subtribe Ruschiinae is one of the largest groups in the family Mesembryanthemaceae. with some 540 described species and infraspecific taxa in 17 described genera.Most of these genera have not been revised since they were first described, except for partial revisions of the genera occurring in SWA/Namibia (Friedrich 1970).The only exception is the genus Astridia, of which a brief account was published by Bolus (1961b, key revised in Bolus 1966a).
The subtribe is defined by the fruits, which are 5locular, with covering membranes and placental tu bercles, and (usually) without valve wings.In habit, members of this subtribe range from dwarf succu lents which are not as reduced as, for example, Lithops or Conophytum, to some of the largest shrubs in the family Mesembryanthemaceae.The leaves are always opposite, and are usually triquetrous to semiterete.Flowers are solitary or in larger or smaller cymose inflorescences, and individual flowers vary from among the smallest to among the largest in the family.Petal colours found in this subtribe include the full range of yellows, reds, pinks, magentas and whites found in the family: in some species each pe tal has a central dark-coloured longitudinal stripe, but petals in which the base and apex are of different colours are very rare.
Taxonomic work on this group was started with the intention of producing a comprehensive account for the Flora of southern Africa.Rising costs and in creasingly stringent limits on what was financially vi able rendered this goal unattainable in the foresee able future, and so it was determined to complete treatments of the minor genera where this could be done with minimum extra input.In the case of both Ebracteola and Acrodon, about half of the species have, until this treatment, been included in the genus Ruschia.This illustrates one of the major diffi culties in the taxonomic treatment of this subtribe, namely the generally poor delimitation of genera.The apparently ambiguous limits of genera in the Ruschiinae is, no doubt, due to the fact that newlvdiscovered species have been described in great numbers over a long period of time, and assigned to genera which have not been revised since they were first described.In particular, it appears that a re vision of the over 300 species at present assigned to the genus Ruschia would, if it were to be of benefit to the users of taxonomic treatments, distribute these species among about three major genera and some (extant) minor ones.This process is started in this paper.

METHODS AND MATERIALS
Both dried and living specimens, the latter mainly cultivated, were examined in the course of this study.For numerical processing, both scored and measured characters were recorded, and an Opera tional Taxonomic Unit (OTU) was defined as being equivalent to one micro-taxon as recognized by L. Bolus.As her stated aim (Bolus 1936(Bolus -1958: iii) : iii) was to describe as much as possible of the variation in the family as new taxa, so that the types would remain in South Africa 'for the convenience of future work ers', it was found that the danger of overlooking any variation in the specimens examined by her was neg ligible.Very few.if any, specimens not referable to any of her micro-taxa have been collected since her Bothalia 16,2 (1986) death.Scored characters were taken as being for all obtainable specimens that could be regarded as be longing to the OTU in question.Measured charac ters were taken from as many individuals as possible with a maximum of 25, subject to the same con straint.
Eighty-six characters were recorded; some of these were noted as minimum, mean and maximum and others as minimum and maximum without a mean, yielding a total of 121 character strings (Hall 1973), as listed in Table 1.A preliminary examina tion of a very incomplete portion of this matrix was made using the BOLAID package of programs for numerical taxonomy (Hall 1973); this was the basis for the taxonomic observations of Glen (1984).It was found to be financially unviable to continue pro cessing data for the whole of the subtribe Ruschiinae (538 OTU's) using BOLAID.so a cheaper and faster system was sought and found in NT-SYS (Rohlf et al. 1977).BOLAID was used on the Bur roughs B7700 of the Department of Agriculture and Water Supply, while NT-SYS was made available by the CSIR on their CDC computer.The present ac count is derived from an examination of 103 OTU's drawn from all genera recognized as belonging to the subtribe Ruschiinae.The first step in processing the data with NT-SYS was to standardize the raw matrix so that each character had a mean of zero and a standard devia tion of one.The standardized data matrix was then used for further processing, thus ensuring that all characters would be strictly equally weighted in later computations.A matrix of correlations was calcu lated according to an algorithm described by Rohlf et al. in the NT-SYS manual.An 'average taxonomic distance' matrix, giving a measure of the dissimilar ity between pairs of OTU's, was calculated accord ing to the algorithm of Sokal (1961).This gives, in effect, a normalized matrix of Euclidean distances between pairs of OTU's.Finally, NT-SYS calcu lated cophenetic matrices and dendrograms for both distance and correlation matrices using UPGMA (unweighted pair-group method with arithmetic av erages; Sneath & Sokal 1973).The dendrograms are referred to below as distance and correlation phenograms, respectively.
The classification and circumscriptions of taxa presented here were derived from both distance and correlation phenograms, with constant reference back to the original specimens.
In the specimen citations below, the following ab breviations, which are not found in Index Herbario rum (Holmgren et al. 1981), indicate garden acces sion numbers:-KG -Karoo Garden, Worcester NBG -National Botanical Garden.Kirstenbosch.Used where the number is of the form <accession num ber>/<date>.e.g.1234/56 is no.1234 of 1956

SUG -Stellenbosch University Garden
All other abbreviations in the specimen citations in dicate herbaria and are according to Index Herbario rum.

ASTRIDIA
The genus is a very natural and cohesive group of species, not only very similar in appearance but also confined to a small area in the lower reaches of the Orange River.
In habit, Astridia Dinter is very similar to Ruschianthemum and Stoeberia, all three of which occur in the same area.Ruschianthemum and Stoeberia differ markedly from Astridia in having small (± 1 0 mm in diameter) white flowers rather than large showy ones, in having capsules showing varying degrees of schizocarpy rather than the hygrochastic capsules of Astridia, and non-baculate seeds with well differen tiated embryo and micropylar regions (see Figure 1).Astridia can be separated from species of Ruschia of similar habit by its flower size, bracts, stamens, seed surfaces, and by the leaf surface, which is glabrous in Ruschia but velutinous in Astridia.In addition, the flowers of Astridia are solitary, whereas those of Ruschia are typically borne in large, repeatedly branched cymes.The overall appearance of plants of the genus Astridia is so distinctive that they are un likely to be confused with any other group of Mesembryanthemaceae. Dinter (1926) published the new combination As tridia velutina Dinter for the plant which he had named Mesembryanthemum velutinum Dinter (non L. Bol.), but he gave only a very brief description of his new genus.Schwantes (1927) supplied an amended generic description in the following year, and recognized a second species.A. maxima (Haw.)Schwant., based on M. maximum Haw.The distin guishing characters of the new genus were stated to be the seeds, which appeared to be covered with hol low spines, and the overall form of the plants.
N. E. Brown (1928) noted that although no distin guishing characters could readily be found for the genus, plants were so distinctive in overall appear ance that the genus was possibly a good one.He also noted that the seed character mentioned by Schwantes was not unique in Mesembryanthema ceae.Bolus (1961b) noted a number of relatively subtle characters by which Astridia can be distin guished from Ruschia and other genera.These in clude: 1, the manner of attachment of the leaves to the stem (in Ruschia old leaves may be broken off com plete, but in Astridia a portion of leaf base always adheres to the stem); 2, the form of the bracts (cymbiform in Astridia but semi terete to triquetrous in Ruschia); 3, the ciliate inner stamens of Astridia, as opposed to the glabrous stamens of Ruschia.
Bolus admitted the possibility that any or all of these characters may be found in the genus Ruschia, but the combination of all three, together with the growth habit, appears to define a natural group.Friedrich (1970) distinguished between Astridia and species of Ruschia of similar habit bv 1, the large flowers of Astridia (± 50 mm in di ameter when open in Astridia; ± 20 mm in Ruschia); 2, the form of the bracts (cymbiform in Astridia; semiterete to triquetrous in Ruschia); 3, the stigmas, which are always six in Astridia and usually five in Ruschia (this difference is not as sig nificant as it may seem at first sight: carpel numbers are variable throughout the Mesembryanthemaceae, and this variability increases in proportion to the usual number of carpels).
Friedrich was the first to suggest that too many taxa had been described in the genus, and he recog nized six species and no infraspecific taxa in SWA/ Namibia.
The seed character, used by Schwantes in the orig inal delimitation of the genus, is best seen in the type species, A. velutina, where long, spine-like baculae cover the entire surface of the seed (Figure 1A).These characteristic baculae are not generally as well developed in the other species of the genus, and may be restricted to the micropvlar region of the seed, the surface of the cells of the embryo region of such seeds being in the form of low, roughly conical bacu lae (Figure 4A-D).As Brown (1928) has pointed out, this character is not unique to Astridia, being present, and in fact better developed, in Braunsia (see Figure 1), Antegibbaeum and possibly other genera.
On the south bank of the Orange River, the genus is found in the Richtersveld, in Acocks's (1975) Western Mountain Karoo, Succulent Karoo and Namaqualand Broken Veld, in an area no more than 100 km x 100 km; on the north bank, it is restricted to a hardly larger area of Giess's (1971) 'Desert and succulent steppe (winter rainfall area)'.
The phenograms used to generate the classifica tions are shown in Figures 2 & 3.The overall corre lation coefficient of the distance phenogram was slightly less than that of the correlation phenogram (0,75 as against 0,80).Nevertheless, it was found by checking back against the original specimens that the former yielded groups that were less likely to lead to the misidentification of new specimens than the lat ter.In cases where the two phenograms differ, the distance rather than the correlation dendrogram was followed in the construction of groups and recogni tion of taxa.trous, usually grey, blue-grey or glaucous green, Stigmas 6, subulate to filiform.Capsules 6-locular; usually firmly attached and not falling readily.Flow-covering membranes well developed; valve wings ers solitary, on relatively short pedicels or subsessile, absent or awn-like; placental tubercles present; exthe pedicels each with a pair of relatively small, dis-panding keels diverging, unadorned.Seeds various tinctively boat-shaped bracts which tend to remain shades of maroon to black, with long, spine-like baattached to the plant on drying.Sepals 6, in two se-culae at least on the micropylar region.Robust shrubs 200-300 mm tall.Stems pale buff to dark brown when young; internodes ± 15 x 4 mm.Leaves minutely velutinous, 15-34 mm long, obscu rely keeled, 7-14 mm wide and slightly less thick.Pedicel ± 7 x 2,5 mm.Bracts up to 12 mm long and 6 mm thick.Flowers ± 30 mm in diameter when open.Sepals 6, outer pair up to 7 x 6 mm, inner 4 up to 7 x 5 mm.Petals 40-70 in 1-2 series, white to pink, 12-23 x up to 3 mm, but often narrower.Sta minodes present, 2-10 mm long, distinct from petals.Stamens many; filaments 6-9 (-11) mm long.Stig mas subulate to filiform, shorter than longest sta mens, -1 -8 mm long.Capsule dark grey, broadly obconical, ± 11,5 mm in diameter when closed and 8,5 mm long; valve wings absent; placental tubercles large, radial diameter 1,2 mm.Seeds pale yellow to dark brown, echinate, 1,1-1,5 (-1,85) x 0,7-1,3 x 0,5-1,2 mm, micropylar region 0,3-0,55 mm long; baculae very prominent; microbaculae long but vari able.Dinter (1923) is a later homonym of M. velutinum L. Bol. (1922), and is therefore invalid in Mesembryanthe mum.However, there is no earlier homonym of the combination Astridia velutina Dinter & Schwant., with Mesembryanthemum velutinum Dinter non L. Bol. as basionvm, nor is there any earlier name for this species.According to Art. 72 note 1 of the Rules of Botanical Nomenclature, this combination must be accepted.For this reason, the name Astridia dinteri L. Bol. was superfluous when published and is to be rejected as illegitimate (Art.63.1).

The name Mesembryanthemum velutinum
It will be seen from the scatter diagram in Figure 5 that the various entities included in this species can not be distinguished on the basis of their leaves.The same is true for all other characters examined, and so several previously-accepted names, as listed in the synonymy above, must become synonyms of A. velu tina.
Voucher specimens: The relatively short internodes sheathed by the leaves for half their length, and the long narrow leaves distinguish this species from others of the genus.The distribution of this species is shown in Figure 9. 1.4 Astridia citrina (L.Bol.) L. Bol. in Journal of South African Botany 32: 230 (1966b); Jacobsen: 413 (1974).Glen: t. 1917Glen: t. (1985)).Type: SWA/Nami bia, without precise locality, cultivated in Windhoek Government Garden, July-August 1937, Rusch c& Erni sub Holloway 58 (BOL!).Robust shrubs 300-400 mm tall.Stems pale buff to dark brown when young; internodes ± 20 x 5,5 mm.Leaves minutely velutinous, 30-55 (-70) mm long, strongly keeled, 11-19 mm thick.Pedicel ± 13 x 3 mm.Bracts up to 19 mm long and 10 mm thick.Flower ± 50 mm in diameter when open.Sepals 6, outer pair up to 12 x 8 mm, inner 4 up to 11 x 7 mm.Petals ± 50 in 1-2 series, white to yellow, 18-23 x up to 3 mm.Staminodes present, 6-9 mm long, distinct from petals.Stamens many; filaments 4-9 mm long.Stigmas subulate to filiform, shorter or longer than longest stamens, 8,5-11 mm long.Cap sule dark grey, broadly obconical, ±11,5 mm in di ameter when closed and 10 mm long; valve wings absent; placental tubercles small, radial diameter 0.5 mm.Seeds deep maroon, echinate, 1,1-1,4 x 0,75-1,05 x 0,5-0,9 mm, micropylar region 0,3-0,55 mm long; baculae prominent, more so on micropylar region than embryo region; microbaculae large, el liptical-conical.The yellow colour of the flowers in this species is unique in the genus.Other distinguishing characters are the combination of wide and deep leaves relative to their length, relatively long pedicels and long cap sules.The distribution of this species is shown in Fig ure 10.The flowers of this species are the largest and showiest in the genus.Like A. vanheerdei.to which it is very similar, it may be distinguished from A. longifolia by its subsessile flowers (and.presumably, capsules).It is similar to A. longifolia and different from A. vanheerdei in the narrowing of the leaf bases and the presence of staminodes.These three species are very close, and still more detailed studies may show that they are all one species.The distribution of this species is shown in Figure 10.Robust shrubs 200-300 mm tall.Stems pale buff when young; internodes ± 30 mm long and 8 mm in diameter.Leaves glaucous, triquetrous, keeled, 42-78 (-118) mm long.(4-) 14-20 mm wide and thick, minutely velutinous, sheathing stem for ± 7 mm.Pedicels ± 1 3 x 4 mm.Bracts up to 24 mm long and 11 mm thick.Flower ± 60 mm in diameter when open.Sepals 6, outer pair up to 13 x 12 mm, inner 4 up to 10 x 6 mm.Petals ± 70, white or rarely pale pink, 22-36.5 x up to 3 mm.Staminodes absent.Sta mens many, 3,5-11 mm long.Stigmas filiform.4-11 mm long.Capsule broadly obconical.± 13,5 mm in diameter when closed and 9.5 mm long; covering membranes present, covering most of interior; valve wings present or absent, if present then long and awn-like; placental tubercles large, radial diameter ± 1,4 mm: expanding keels slightly to widely diverg ing.Seeds medium brown to dark maroon, 0.95-1.5 x 0,7-1.0x 0.6-0,9 mm.micropylar region 0.3-0,55 mm long; baculae prominent, more so on embryo region than on micropyle; microbaculae conspicu ous, long, cylindrical to elliptical.As will be seen from the scatter diagram in Figure 11, Astridia ruschii is quite indistinguishable from A. hallii in leaf characters.The same is true for all other characters examined.These two names must there fore be regarded as referring to the same species.The relatively wide leaves, long pedicels and broad, shallow capsules distinguish this species from all others.The distribution of this species is shown in Figure 12.This species differs from A. longifolia in the rela tively broad leaf-bases, the subsessile flowers and the absence of staminodes.The first of these charac ters distinguishes it from A. speciosa, as do the lat erally compressed leaves and the absence of stami nodes.The distribution of this species is shown in Figure 13.This species is known only from the type speci men, and so must be regarded as the rarest and leastknown member of this genus.

ACRODON
Acrodon N.E.Br. is a genus of four rather similarlooking species of dwarf habit.Two of these species have, until now.been included in Ruschia, and the arrangement of specimens in BOL indicates that L. Bolus and her co-workers were in some doubt as to whether these two genera should be retained or merged into one.Ruschia is regarded here as a genus of typically shrubby plants, most of which have flow ers in cymose inflorescences.Only in the section Uncinata does one find plants in which the leaf keel is toothed; these plants are shrubs with leaves decur rent on the internodes, and flowers with petals of uniform colour, often arranged in five 'fascicles'.Plants of this small genus were among the first highly succulent members of the family Mesem bryanthemaceae to become known in Europe.Acro don bellidiflorus was well known in England (Brad ley 1717; Dillenius 1731) and the Netherlands (Lin naeus 1738; Van Royen 1740) early in the eighteenth century.It appears that this species was introduced into cultivation, at least in England and Germany, as early as the end of the seventeenth century, as the first descriptions of it date from 1700 (Plukenet 1700; Volckamer 1700).Linnaeus evidently knew the plant from his period of work at Clifford's gar den near Leyden in the Netherlands.His descrip tions in the Hortus Cliffortianus and the Species Plantarum (1753) make use of the sharply trique trous leaves and the small teeth on the leaf keel near the apex; these characters are still used to distinguish Acrodon from others of the family.Haworth (1821) distinguished two varieties of A. bellidiflorus.The genus Acrodon was separated from Mesembryanthe mum L. by N. E. Brown (1927), who at that time included only A. bellidiflorus in his new genus.
The classification presented here is based on the phenograms shown in Figures 14 & 15.The great similarity between species of Acrodon is shown by the scatter diagrams in Figures 17-19.From these it can be seen that the species are difficult to separate reliably on the basis of individual characters taken pairwise and plotted on scatter diagrams, but that multivariate methods using as many characters as possible together, not just two at a time, distinguish more certainly between very close species.Dwarf succulents with internodes completely hid den by leaf bases.Leaves bright green, sharply tri quetrous, often with a few small teeth on keel near apex.Flowers solitary; pedicels with leaf-like bracts.Sepals 5, in two series, outer pair fleshy, inner 3 less so, with membranous margins and generally signifi cantly smaller.Petals 30 or more.1-3-seriate.white to pink, usually with conspicuous darker pink to ma genta longitudinal stripes.Staminodes present or ab sent.Stamens numerous, erect; filaments papillate at least at base.Stigmas 5, subulate.Capsules relatively large, turbiniform, woody, 5-locular; covering mem branes well developed; valve wings awn-like, rarely absent; placental tubercles relatively large.Seeds various shades of maroon to black, with conspicu ous, well spaced baculae on the embryo region.Ficoides africana humilis, folio triangulari breviori nonnihil spinoso seu denticulato, Volckamer: 166 (1700).Ficoides seu fi cus aizoides africana folio triangulari crasso brevi glauco ad tres margines aculeata, Boerhaave 1: 290 n.21 (1720).
Voucher specimens: A. bellidiflorus differs from A. parvifolius in that the former species grows in small tufts of long leaves and has large flowers and relatively larger fruits.A. parvifolius grows in large mats with trailing stems bearing short, wide leaves, small flowers and rela tively small fruits.The leaves of A. duplessiae have marginal teeth, are glaucous green and are dorsally compressed, while those of A. bellidiflorus have teeth only on the keel, and are bright green and sym-  No individual character or group of characters can be used to distinguish between specimens hitherto assigned to A. bellidiflorus, A. subulatus and Rus chia longifolia (cf. Figure 20).It is therefore neces sary to regard these three names and their nomenclatural synonyms as referring to the same species, the correct name being A. bellidiflorus.The distribu tion of this species is shown in Figure 6.Creeping, succulent-leafed herbs forming mats up to 1 m in diameter.Stems with internodes not com pletely hidden by leaf bases; young internodes, when visible, pale brown and shiny.Leaves bright green to olive green, reddish brown at apices, laterally com pressed.sharply triquetrous, 13-17,5 mm long.2-3 (-4) mm wide and 2,5-5 mm thick, with a few small teeth on keel at apex.Pedicels ± 14 mm long and slightly over 1 mm in diameter, with a pair of leaf like bracts at base.Flowers opening about midday, small for genus, 17-20 mm in diameter when open.Sepals 5, in tw'o series, outer pair up to 5 x 4 mm, inner 3 up to 4 x 3 mm.Petals ± 35, white to pink, with a central longitudinal pink to magenta stripe and magenta margins.6-8 x 1 mm.Staminodes few, sharply distinct from petals, white with pink apices, 3,5-4,5 mm long.Stamens in several series; fila ments papillate at base or in lower half, pink.2-3 mm long; anthers deep purple; pollen pink.Stigmas subulate, pink.Capsule 6,5 x ± 8.5 mm; covering membranes covering most of the interior; valve wings aw'n-like; placental tubercles small, radial di ameter ± 0.8 mm; expanding keels almost parallel, unadorned.Seeds deep brown.0.7-0,9 (-1,0) x 0,6-0,7 x 0.45-0,65 mm.micropylar region 0,22-0.38mm long; baculae large, hemisphericalcylindrical and well spaced on embryo region, less regular on micropylar region; microbaculae rod shaped, conspicuous.
The name Ruschia leptophylla' was published in a fascicle of Notes on Mesembryanthemum and allied genera dated 29th January 1932, while the fascicle in which the name R. macrophylla' was published is dated 24th June 1932.The reason for choosing the epithet leptophylla rather than macrophylla for this species, therefore, rests on consideration of a prior ity of about six months.
The only difference between plants identified as Ruschia macrophylla and those previously called R. leptophylla and R. compressa is in the ratio of leaf thickness to leaf length, and even in this character there is a small degree of overlap (Figure 21).In all other measured characters the overlap is complete (see, for example.Figures 18 & 19), and the states of 'multi-state' characters are the same for all the taxa included in the present species.For this reason the taxa united here are not retained as separate species.

EBRACTEOLA
This genus comprises five remarkably similarlooking species.They are readily distinguished from most genera in the subtribe Ruschiinae by their dwarf habit, leaves without teeth and petals without longitudinal stripes, and from most Mesembryanthemaceae of dwarf habit by their thickened, woody rootstocks.The whitish colour of the seeds and the absence of bracts are useful accessory characters but do not appear in all species of Ebracteola.
The first species of Ebracteola Dinter & Schwant.to become known was described under the name Mesembryanthemum wilmaniae from near Kimber ley (Bolus 1916).This was followed by M. montismoltkei from the other extremity of the range of this genus, near Windhoek (Dinter 1922), M. derenber gianum in the following year (Dinter 1923), Ruschia fulleri a few years later (Bolus 1929) and finally, after a thirty-year interval, Ebracteola Candida (Bo lus 1961a).The genus was first described to accom modate M. montis-moltkei and M. derenbergianum (Dinter & Schwantes 1927).
Apart from Friedrich's (1970) study of the SWA/Namibian material, no critical study of the genus has been published until now.Three species of the genus were examined in the course of a study of the subtribe Lampranthinae (Glen 1978), but the only conclusion to be drawn at that time was that Ebracteola was not a member of that subtribe, and was probably better placed in the subtribe Ruschii nae.
The generic name refers to the absence of bracts in most specimens of these two species, with a dimin utive ending to indicate the dwarf habit of plants of this genus.
The phenograms used to generate the classifica tion presented here are shown in Figures 22 & 23.The great similarity between the different species of Ebracteola is shown graphically by the scatter dia grams in Figures 24-26.From these it can be seen that the species are difficult to separate reliably on the basis of character pairs plotted on scatter dia grams, but that multivariate methods using as many characters as possible together, not just two at a time, distinguish more certainly between very close species.
The protologue of the genus does not indicate which of the two first-accepted species was to be taken as the type of the genus.The choice of E. montis-moltkei (Dinter) Dinter & Schwant. was made by Von Poellnitz (1933: 40), without any reason being given.This choice must necessarily be followed, but it may be pointed out that the other species placed in their new genus by Dinter & Schwantes, E. derenbergiana (Dinter) Dinter & Schwant., was alone transferred from Ebracteola to Ruschia by Bolus (in Jacobsen 1955) and Weber (1968).
The species here transferred to Ebracteola from Ruschia, namely E. wilmaniae and E. fulleri, are dwarf cushion-forming succulents with essentially semiterete leaves, although forms of E. wilmaniae with sharply triquetrous leaves are known.They also have solitary flowers which are very similar to each other and to other species previously included in Ebracteola.These characters are rare in the Ruschii nae, and so these species were found to be closer to other species classified under Ebracteola than to any species of Ruschia, regardless of which measure of similarity was used.For this reason they are transfer red from Ruschia to Ebracteola.Ebracteola Dinter & Schwant. in Zeitschrift fur Sukkulentenkunde 3: 24 (1927);Friedrich: 44 (1970);Herre: 146 (1971);Dyer: 110 (1975).Lectotype species: E. montis-moltkei (Dinter) Dinter & Schwant. (cited by Von Poellnitz 1933: 40).
Voucher specimens: E. wilmaniae is the easternmost member of the genus, and is hardly known west of Prieska.Its near est neighbour is E. fulleri, which is not found east of Keimoes, about 150 km west of the westernmost re cord of the present species.The range of E. wilma niae extends eastwards as far as Makwassie in the Transvaal, northwards as far as Kuruman and south wards as far as Niekerkshoop.near Prieska (see Fig ure 10).
The absence of bracts in this species and widely separated distribution ranges separate E. montismoltkei on the one hand from E. fulleri and E. wil maniae on the other.The sharply triquetrous leaves provide another useful character distinguishing this species from E. fulleri.while the staminodes merg ing into petals serve to distinguish E. montis-moltkei from E. wilmaniae.Differences between this species and others of the genus not mentioned above are dis cussed under those species.The distribution of this species is shown in Figure 13.
. alba s. s t r .0 -R.d u lc ls s. s t r .® -fl. 1 at 1sepa 1 a & ■ R. 1ong i f o 1 la f -R.rubra s. s t r .
FIGURE 11. -Leaf measure ments of plants included in L e a f l e n g t h , mm Astridia hallii.
. R. du Plessis s.n. in NBG I048,32 (BOL): Robinson Pass (-CC).I. de Jager s.n. in BOL 24718 (BOL).3421 (Riversdale): Melkhoutfontein (-A D ).L. Bolus s.n.(BOL).Differences between A. duplessiae on the one hand and A. bellidiflorus and A. parvifolius on the other are discussed under those species.Leaves of A. duplessiae are generally much broader and slightly shorter than those of A. leptophyllus; they also have marginal teeth, which are lacking in the latter species.In A. duplessiae the petals are some what longer than in A. leptophyllus, and the capsules are shallower in relation to their diameter.Distribu tion of this species is shown in Figure 9. 2.4 Acrodon leptophyllus (L.Bol.) Glen, comb.nov.

FIG
FIG URE 22. -Phenogram of Ebracteola calculated from a distance matrix using UPGMA.Irrelevant O T U 's are omitted.
FIGURE 24.-Leaf measure ments of all species of Ebracteola.

Phenogram of Acrodon calculated from a correlation matrix using UPGMA. Irrelevant OTU's are omitted.
Acocks's (1975)crodon are longer and wider than those of Ruschia section Uncinata, and dark green rather than greyish.The flowers of Acrodon are soli tary, and the petals are evenly spaced.They are characterized by a central longitudinal stripe of a darker colour than the rest of the petal.Acrodon is therefore among the more distinctive genera in the family Mesembryanthemaceae.It may be distinguished from Ruschia section Un cinata not only on morphological but also on geo graphical grounds.Ruschia section Uncinata is characteristically found in the upper Karoo and SWA/Namibia, with outlying species in the Orange Free State and western Transvaal, while Acrodon is restricted to the southern Cape and Little Karoo, inAcocks's (1975)Fynbos, Coastal Rhenosterbosveld, Coastal Macchia, Succulent Mountain Scrub and Karroid Broken Veld types.

Stigma and sta men measurements of Ac- F i 1 ament length, mm rodon spp.
metrical in transverse section.Capsules of A. duplessiae are generally slightly broader in relation to their depth than those of A. bellidiflorus, and seeds are smaller and darker in colour.A. leptophyllus also has glaucous leaves, but these tend to be longer than those of A. bellidiflorus.The capsules of A. lepto phyllus are significantly longer relative to their di ameter than those of A. bellidiflorus.

C a n d id a L. Bol. in
E.