Nomenclature and typification of southern African species of Euphorbia

Types have been located for most of the 185 species of Euphorbia L. that are known to occur naturally in southern Africa and also for most of their synonyms. Lectotypes or neotypes are selected where possible for those names for which a holotype cannot be found. The synonymy largely follows previous accounts and reasons are given where new synonymy is proposed. Euphorbia huttonae N.E.Br. is reinstated at the level of species and E. franksiae var. zuluensis A.C.White et al . is raised to the level of species as E. gerstneriana Bruyns, nom. nov. A new name, E. radyeri Bruyns, is provided for the rhizomatous plants previously referred to as E. caerulescens Haw., which is synonymous with E. ledienii A.Berger.


InTRODuCTIOn
The large, cosmopolitan genus Euphorbia L. is represented in southern Africa (taken here to include the countries Botswana, Lesotho, namibia, South Africa, and Swaziland) by 185 species.For these there are 368 validly described names at the level of species.In this listing only naturally occurring species are included.These are distributed among the four subgenera established by Bruyns et al. (2006) and based on molecular and morphological data as shown in Table 1, where additional information on growth habit is also supplied.Many of the succulent species such as E. globosa Sims, E. meloformis Aiton and E. obesa Hook.f. are popular subjects in specialist collections around the world, but the other species have little economical use, though their medicinal use is probably underestimated.The last revisions of the genus for this region were those of n.E. Brown (1911Brown ( -1912Brown ( , 1915)), with the succulent species receiving further attention by White et al. (1941).The taxonomy of Euphorbia in southern Africa remains disorganised, with many names applied in different herbaria in South Africa to quite different species.In an attempt to bring order to the situation, types have been located and, where these are missing, lectotypes or neotypes are selected as applicable.

MATERIALS AnD METHODS
For each validly published name for a southern African species of Euphorbia, the protologue was consulted in the relevant literature.Type specimens have been searched for among material in the herbaria B, BM, BOL, g, gRA, K, KMg, M, nBg, nH, nu, nY, OXF, P, PRE, S, SAM, SBT, W, WInD, WRSL, Wu, Z (herbarium acronyms according to Holmgren et al. (1990)).A specimen is taken as the holotype if it was indicated as such by the author, or if it is clear from where it is located relative to where the author worked that it must be the holotype.In many cases it has proved to be impossible to be sure which specimen is the holo-type.In such instances, a lectotype is generally selected from among the duplicates of the 'type number' or from among the 'syntypes' mentioned by the author in the protologue.A particular specimen is chosen over others according to its quality or, in the case of syntypes, according also to how widely duplicates (if any) are represented.In cases where no appropriate material for use as a lectotype was located, a neotype was selected.All material cited has been seen unless it is expressly stated otherwise.The JSTOR Plant Science website (http:// plants.jstor.org/)has been consulted in all applicable cases and the Kew Herbarium Catalogue (http://apps.kew.org/herbcat) was consulted for many species and names as well.Data on localities is given as on the specimens, with the present-day country where the specimen was collected added.

RESuLTS
The species are arranged alphabetically within the four subgenera of Euphorbia that were established in Bruyns et al. (2006).The synonymy is as in Bruyns et al. (2006), except where otherwise mentioned and discussed.
Jan. 1820, Mund & Maire 15 (K 000253186, lecto., designated here).[There are two specimens under this number at K and it is not certain that either was seen by Klotzsch & garcke.This is the larger specimen, the other is a 'branch from the type', according to n.E.Brown.] A. setigerum E.Mey.ex Klotzsch & garcke: 29 (1860).Type: South Africa, Cape, Drège (missing).

E. siliciicola
E. aequoris n.E.Br.: 279 (1915).Type: South Africa, Cape, Middelburg div., Schoombie, Feb. 1897, Trollip (sub SAM 20091) (SAM, lecto., designated here; K, isolecto.).[Brown (1915) cited also: Rosmead Junction, 4 000 ', 22 Mar. 1900, Sim sub Galpin 5626 (PRE); between Colesburg & Hanover, 1871, Bolus 2201 (K).]Leach (1988a) discussed E. juttae in detail but considered that E. aequoris, although closely related, was 'sufficiently distinct in vegetative characters and habit alone for it to be disregarded' in those discussions.He did not say in what way it was so distinct.It appears that this distinctiveness lay in the much more robust plants formed by E. aequoris, with longer and more slender stems and branches, with more widely spaced and less prominent tubercles and a longer rootstock, as well as the lack of the peculiar habit that the branches have in E. juttae of bending over to the north or west.nevertheless, among the material that he cited under E. juttae were two specimens from near Olifantshoek and near Kenhardt respectively that are rather more typical of E. aequoris than of E. juttae.While many specimens of E. aequoris are unmistakeable (especially those from the great Karoo and drier parts of the Eastern Cape), those from the northern Cape and calcareous pans on the southern edge of the Kalahari are not clearly referable to either species.Some of these (especially plants from exposed spots) may even exhibit a similar, almost prostrate habit to E. juttae, and have shorter stems and branches with more prominent tubercles while more protected plants are erect, slender, and more typical of E. aequoris.I have found no clear distinctions between the two species and have placed E. aequoris in synonymy.
In Bruyns et al. (2006), E. tenax was treated as a synonym of E. arceuthobioides.The respective types make it clear that they are the same species.However, E. tenax was published first and so this treatment was wrong.
Bergius only cited 'Herm.Afr.23', which refers to page 23 in J. Burman's Catalogi duo plantarum africanorum of 1736 that was in turn part of his Thesaurus zeylanicus.no illustration or specimen is listed, only a phrase which corresponds to the same phrase on page 23 in Burman's Catalogi.However, there is a specimen at SBT annotated by Bergius as 'Euphorbia mihi genistoides' and 'e.Cap.b. sp.grubb'.It is known that a consignment of specimens collected at the Cape by J.A. Auge was bought from Auge by Michael grubb during a brief visit to the Cape in 1764 and presented to Bergius, and that these formed the basis of Bergius' 'Descriptiones' (gunn & Codd 1981).Consequently, this specimen is taken as the type.Haworth (1812) did not refer to Bergius' publication directly, but to 'Willd.,Sp. Pl. 2: 908' where references 'Mant. 564' and 'Berg. cap. 146' were given, the latter clearly the same as above.Klotzsch & garcke: 99 (1860).Type: South Africa.Cape of good Hope, Ecklon & Zeyher 2 (missing).

E. hottentota
The name E. hottentota was maintained as distinct from E. avasmontana in Bruyns et al. (2006).Marloth (1930: 335) separated E. avasmontana and E. hottentota by the number of angles on the branches (7-angled in E. avasmontana; 5-6-angled in E. hottentota) but White et al. (1941: 824) pointed out that 'some of Marloth's herbarium specimens do not agree entirely with the typical form' so that the identity of this 'species' is less clear than Marloth thought.Over the large area where it occurs branches are frequently 4-angled and may have up to eight angles and no clear separation into 5-6-angled and 7-angled plants is possible.no differences in the floral structures have been detected on which they could be separated.E. barnardii A.C. White et al., The Succulent Euphorbieae 2: 965 (1941).Type: South Africa, Transvaal, Sekukuniland, farm Driekop, east of Lulu Mountain, 3 000 ', 6 Jan. 1937, Barnard 449 (PRE, holo.;MO, iso.).
Possible types for E. caerulescens include (1) a specimen 'Cape of good Hope, Bowie (K)', which was made by n.E. Brown in november 1876 from 'the type plant (still in cultivation at Kew) dried by myself' (Brown 1915: 365) and (2) a drawing by Bond (423/292) of the apex of a branch and annotated 'drawn from the plant from which Haworth described' and 'Received in 1823 from the Cape of good Hope by Mr Bowie'.I propose that we accept that the plant in cultivation was among those (if there were more than one) from which Haworth drew up his description so that I have designated the specimen made by n.E. Brown as the lectotype.
E. ledienii var.drègei n.E.Br.: 366 (1915).Type: South Africa, near Port Elizabeth, received 9 Sept 1912, I.L.Drège (K, lecto., designated here).[For E. ledienii var. drègei, Brown (1915) cited two collections: Humansdorp div., near Zeekoe River, Thunberg; near Port Elizabeth, received 9 Sept 1912, I.L.Drège (K).He annotated both the specimen uPS-THunB 11416 and that of Drège as 'var.dregei' so one is designated as lectotype.]Brown (1915) mentioned that he had not seen any flowers of E. caerulescens, nor any dried specimens that he could definitely refer to it, other than the 'type'.He distinguished E. caerulescens and E. ledienii by the glaucous or bluish-green stems, with spines 6-12 mm long in the former; green, not glaucous stems, with spines 2-6 mm long in the latter (Brown 1915: 244).Dyer (1931) and White et al. (1941) found that these distinctions were not useful and they maintained that the only difference between E. caerulescens and E. ledienii was the rhizomatous habit of the former.This character was neither mentioned by Haworth nor is it visible in either the type specimen or the drawing by Bond.It was also not mentioned by n.E.Brown, who knew the type specimen in cultivation.Therefore the association by Dyer (1931) and White et al.(1941) of a rhizomatous habit with E. caerulescens and a non-rhizomatous habit with E. ledienii is erroneous and the name E. caerulescens must refer to the same non-rhizomatous plants as E. ledienii.Consequently, E. ledienii is a synonym of E. caerulescens.This confusion was not recognised in Bruyns et al. (2006), where E. ledienii was treated as a separate species from E. caerulescens.The rhizomatous plants are here treated as a separate species, E. radyeri Bruyns and the differences between them are discussed under that species.
The authorship of this species is usually given as 'goebel' (e.g. Brown (1915); White et al. (1941)) or 'goebel ex n.E.Br.' (e.g.Carter (2002)).However, while n.E. Brown (1897) published the first detailed description of E. grandicornis, the name was in use for a long time before this and there are several earlier brief descriptions that validated the name.The first known published appearance of the name E. grandicornis is Oudemans ( 1865), but the name was not validly described there.The earliest validation of the name is that by A. Blanc (1888), in which it is said that 'Euphorbia grandicornis is still more remarkable on account of its tremendous spines and queer, contorted form'.According to White et al. (1941), a figure of E. grandicornis appeared in an earlier catalogue of A. Blanc of 1887, but I have not been able to trace this.The next one that has been detected is that of goebel (1889), in which the diagnosis is similarly rudimentary but still constitutes valid publication.In J.E. Weiss' account of 1893 a more detailed diagnosis of E. grandicornis appeared.Since both Weiss' and goebel's names are illegitimate, lectotypes are not selected for either of them.
E. grandidens Haw., Philosophical magazine and journal 66: 33 (1825).Type: Illustration number 807/323 by T. Duncanson at K of specimen received 1822 from Cape of good Hope collected by Bowie (lecto., designated here).Pax: 86 (1909).Type: South Africa, Transvaal, Lowveld, near Barberton, Pole Evans (missing).[Carter (2002) cited the type specimen at PRE, but this does not exist, nor is there any material known elsewhere that could have been seen by Pax.] Euphorbia evansii was said to differ (White et al. 1941) from E. grandidens in being shorter (reaching 10 m as opposed to 16 m), with 3-to 4-angled secondary branches with gently sinuate margins (as opposed to 3-angled or rarely 2-to 4-angled in E. grandidens with more prominently toothed margins), spines lacking the pairs of prickles at their bases, these often present in E. grandidens.none of these differences are clear-cut and I have found it impossible to separate the known collections into two distinct species.Consequently, the name E. evansii is placed in synonymy, although it was kept separate in Bruyns et al. (2006).Systematik 34: 375 (1904).Type: Botswana, Lobatsi, Marloth 3413 (missing).neotype (Leach 1967) 1862).Type: South Africa, natal, in woods near Durban, Drège 4614 (S, holo.; K, iso.).[Boissier (1862) cited a specimen at 'h. Bunge' and that at S was annotated by him, so is taken as the holotype.That at K is a 'fragment from type'.]E. similis A. Berger, Sukk. Euph.: 69 (1906a).Type: South Africa, natal ?(missing).

E. griseola Pax, Botanische Jahrbücher für
n.E. Brown pressed two specimens from plants in cultivation at Kew that were reputed to be E. similis and mentioned that he had sent a branch to Berger who had confirmed that this was what he named E. similis.However, many of the pressed branches on the two specimens at K bear foliage-leaves 15-80 mm long and consequently they cannot represent either E. ingens or E. similis in which the leaves were 'minute' according to Berger and where such foliage-leaves are only present on the young stem.P.R.O.Bally determined one of these specimens at K as E. obovalifolia A.Rich.(= E. ampliphylla Pax) and this is more likely to be the correct identity of this plant, which Brown (1915) used for his description of E. similis, but which is not the same as that which Berger (1906a) Carter (2002) cited the type from PRE, the specimen is not present there.It is assumed that this was sent to K on this occasion.This specimen was collected from the same plant from which the figure was painted.]E. knuthii Pax, Botanische Jahrbücher für Systematik 34: 83 (1904).Type: Moçambique, Ressano garcia, 1 000', 27 Dec. 1897, Schlechter 11949 (K, lecto., designated here; BM, BOL, BR, g-2 sheets, gRA, HBg, PRE, WAg, isolecto.).[The sheet at K was annotated by Pax ('Knuthii Pax !') and here he also scratched out Schlechter's proposed name for the plant.nevertheless, n.E. Brown annotated it as 'part of type'.This sheet is then taken as the lectotype.Carter & Leach (2001) informally selected the specimen at K as lectotype, but this is invalid and so it is formally designated here.]E. limpopoana L.C. Leach ex S.Carter, Kew Bulletin 54: 960 (2000).Type: Zimbabwe, Fulton's Drift, 25.5 km nnW of Beitbridge, Sept. 1963, Leach 11582a (SRgH, holo.).
Although E. caerulescens and E. radyeri are similar, they are easily separated.Branches around the perimeter of most plants of E. radyeri are usually rhizomatous and this phenomenon is unknown in E. caerulescens.The branches tend to have a more bluish green colour in E. radyeri than in E. caerulescens, though the colour varies greatly in the latter, with greener branches on plants from more sheltered habitats.The branches of E. radyeri are thicker, deeply articulated into almost spherical segments, while those of E. caerulescens are generally more slender and only indistinctly articulated into considerably longer, cylindrical segments.In E. radyeri the tubercles are often much longer and broader and the leaf-rudiments are somewhat larger than in E. caerulescens.Florally E. caerulescens and E. radyeri are very similar.In E. caerulescens the cyathia are often slightly narrower, becoming more abruptly narrow beneath the glands, while the female florets are borne on a slightly longer pedicel and are without the elongated calyx of E. radyeri.

E. waterbergensis
In Bruyns et al. (2006) E. brakdamensis was included under E. filiflora.Careful examination of Schlechter's many pressings of the type collection of E. brakdamensis, shows, however, that this is not correct.In E. filiflora the stem and branches are very similar in shape and thickness, with the stem usually slightly longer than the branches, if it can be detected at all.In E. brakdamensis, on the other hand, the branches are very much more slender than the stem, which is largely buried in the ground and is greatly exceeded in height by the branches.E. filiflora has unusually long cyathia (often around 8 mm long), with especially long styles (7-9 mm long) and long male pedicels.The cyathia in E. brakdamensis do not exceed 5 mm long and the styles are not longer than 6 mm.The marginal processes on the glands in E. brakdamensis are much more brightly coloured than those of E. filiflora where, however, they are longer, more slender and considerably more numerous.

E. rangeana
[Euphorbia rangeana was very similar to E. marientalii and was distinguished by 'E.rangeana ist grünbraun und graubraun' (Dinter 1914: 31), which does make it validly published.However, no specimens were cited here.]Euphorbia rudis was maintained as a distinct 'species' in Bruyns et al. (2006).However, for E. rudis and E. braunsii White et al. (1941: 474) mentioned that 'there is really no sharp line of distinction between the two plants, but rather a gradation.The typical forms of the two are fairly clearly distinguishable, while many of the intermediate forms are very confusing indeed and difficult to classify satisfactorily.'The distinctions between the two included: the smaller 'average size of the main stem', the 'more slender' branches with the tubercles 'somewhat more recurved at the apex' and 'rather smaller' cyathia and 'more completely united styles' in E. rudis.These are all subject to considerable variation so that the name E. rudis has been abandoned here.

E. crassipes Marloth,
Transactions of the Royal Society of South Africa 1: 318 (1909).Type: South Africa, Cape, Biesiespoort, Marloth (4399)4397 (PRE, holo.; K, iso.).Marloth: 38 (1910b).Type: South Africa, Cape, Britstown, Sept. 1909, Marloth 4682 (PRE, holo.; K, iso.).[In the cases of both E. crassipes and E. fusca, Brown annotated the specimens at K as parts from Marloth's type specimens and so the holotype is the specimen at PRE in each case, with isotypes at K.] E. baliola n.E.Br.: 327 (1915).Type: namibia, great Karas Mountains, between 1st & 2nd outspan between Kraikluft and narudas Süd, 5400', 26 Dec. 1912, Pearson 8095 (K, holo.; BOL-2 sheets, gRA, SAM, iso.).[Brown annotated the specimen at K himself as 'Type' but those at BOL and SAM were not annotated by him.Therefore that at K is taken as the holotype.]E. inornata n.E.Br.: 586 (1925).E. inelegans n.E.Br.: 322 (1915), nom. illegit., non n.E.Br. (1911).Type: South Africa, Cape, near Kimberley, Sept. 1912, Moran (sub Schonland 1718) (K 000253322, holo.; gRA, K, iso.).[Brown (1915) mentioned that E. inornata was described from a living plant sent by Schonland in 1912, grown at Kew and pressed by Brown himself in June 1913.This is the specimen 'near Kimberley, Moran, living plant sent to Kew by Schonland' (K000253322).Mounted on the same sheet is another specimen, namely 'Moran sub Schönland 1718' (K000253323) and both had 'Type' written on them by Brown.The former is taken as the holotype.]E. eendoornensis Dinter: 196 (1932).Type: namibia, between Wittsand and Eendorn, 26 Mar. 1924, Dinter (missing).neotype (designated here): namibia, Vrede, Bruyns 11362 (nBg). . hopetownensis nel: 192 (1933b).Type: South Africa, Cape, Hopetown, 1930, E.Markoetter sub SUG 5529 (missing).Type: Illustration in Kakteenkunde: 192 (1933) (lecto., designated here).[Although Carter (2002) cited a specimen at STE (now incorportated into nBg), this does not exist.]Marloth (1910b) said that Euphorbia fusca differed from E. crassipes by the non-persistent peduncles (some peduncles being persistent in E. crassipes).In most populations of E. crassipes one finds plants with persistent peduncles and others without them so this character cannot be used to distinguish between them and the type of E. crassipes at PRE is a typical specimen of what is usually referred to as 'E.fusca'.White et al. (1941) maintained that the main differences between E. crassipes and E. fusca were the slightly more cylindrical stem, thicker branches, the deeper involucres and the green glands.However, in the description Marloth did not mention the glands at all and they were only represented in a small black and white drawing so that their colour was unknown.none of these other differences are significant in this widely distributed and quite variable species.

E
Although the glands of E. inornata were given as olive-green on their upper surface, which is unusually pale for E. crassipes, the shape of the plant, the relative thickness of the branches and the shape of the cyathia and glands all fit E. crassipes, under which it is included here.
Euphorbia hopetownensis was described from a small plant (only 5 cm broad) with ascending, relatively stout branches which bore unusually short peduncles at 5-7 mm long and 'pink-purple' glands with five teeth.The small figure in the text and these few details are strongly suggestive of E. crassipes, under which this name is subsumed here.

E. suppressa
According to n.E. Brown (1915), Marloth considered his number 4397 from between Prince Albert and Prince Albert Road ('the railway') to belong to E. crassipes.However, Brown believed that the absence of a 'flat top to the stem' was significant and that it represented a distinct species, which he named E. albertensis.The plants pressed (K, PRE) have a relatively slender stem (far too slender to belong to E. crassipes) with numerous short branches towards their apex (which are also much more slender than in E. crassipes) with many long, slender, spine-like persistent, sterile peduncles.Vegetatively these plants are extremely similar to E. decepta and, although Brown (1915) was unable to supply much detail about the floral parts of E. albertensis, this name is included here under E. decepta.
Both E. gamkensis and E. suppressa were treated as distinct species in Bruyns et al. (2006).but are here relegated to synonymy.
Euphorbia suppressa was compared extensively with E. albertensis and E. arida (Marx 1999a).The basis for comparison with E. albertensis was mainly Figure 445 of White et al. (1941).However, it is uncertain whether this figure is of the 'species' described by n.E. Brown as E. albertensis.Apart from the fact that Dyer (gRA records) had tentatively attributed two specimens from the area between Prince Albert and Klaarstroom to E. arida, it remains unclear what this new species has to do with E. arida (a species of the north-eastern great Karoo and southern Free State) and why it was not compared with E. decepta, which is fairly well-known on the southern portion of the great Karoo between Beaufort West and Willowmore.Florally E. arida and E. decepta are not easily separated except by the somewhat shallower cyathium (and slightly shorter styles) with fewer, often obsolete teeth on the outer margins of the cyathial glands in E. decepta (deeper cyathium, longer style and more prominent and more numerous marginal teeth in E. arida).However, although the plant appears to be very similar in both species, beneath the soil plants of E. arida have a system of swollen tuberous roots which develop from and extend the tap-root.These structures are entirely absent in E. decepta.In all these respects E. suppressa is identical to E. decepta and so this name is included here under E. decepta.
Euphorbia gamkensis was compared extensively with E. crassipes (and its synonym E. fusca).However, it differs from E. crassipes by its much smaller stature (main stem at most 90 mm thick) by the considerably deeper cyathium whose glands are more-or-less without marginal processes (these are particularly prominent in E. crassipes and are usually strongly deflexed).Again, it ought to have been considered how it differs from E. decepta.Vegetatively the two are difficult to separate and I have been unable to find any reliable differences.The cyathia differ in that the styles are shorter and more deeply divided in E. gamkensis, but no other significant differences have been detected.As I consider this to be insufficient on which to base a separate and otherwise so similar species, I have included E. gamkensis under E. decepta.
E. franksiae n.E.Br.: 315 (1915).Type: South Africa, natal, Camperdown, 2 000 ', 19 Oct. 1910, Franks sub Medley-Wood 11727 (K, holo.;nH, PRE, iso.).[The specimens at nH and PRE were not seen by Brown, though that at nH is annotated as 'part of Type Spec.'.The sheet at K contains two specimens, one collected by Franks on 19 October 1910, pressed by Wood and sent to K (this being the other 'part of Type Spec.') and another made from two plants sent in Apr.1913 to, and cultivated at, Kew.Only the former specimen is annotated by Brown as 'type' and is taken as the holotype.]E. woodii n.E.Br.: 315 (1915).Type: South Africa, natal, Clairmont Flats, Wood 4090 (K, lecto., designated here; nH, isolecto).[Brown (1915) cited also: Clairmont Flats, Wood 11803 (K) and Wood 12612 (K).]E. passa n.E.Br.: 313 (1915).Type: South Africa, natal, Cooper, cult.J. Corduroy, 6 July 1905 (K 000253311, lecto., designated here, K, isolecto.).[Brown (1915) cited also: Scottsburg, Pole Evans (missing); umzumbi, Wood (K).]E. discreta n.E.Br.: 316 (1915).Type: South Africa, natal, banks of umzimkulu River near shore, 25 Feb. 1837, Bachmann 757 (K, holo.).Brown (1915) recognised a host of 'species ' here, including E. discreta, E. ernestii, E. flanaganii, E. franksiae, E. gatbergensis, E. passa, and E. woodii. White et al. (1941) reduced the number slightly by placing E. discreta and E. passa in synonymy under E. woodii and recognising E. ernestii, E. flanaganii, E. franksiae, E. gatbergensis, and E. woodii as distinct species.Brown (1915: 314) commented on the remarkable extent to which these plants can vary in size; in particular, how one of them increased in size in cultivation from 30-40 branches at 3-8 inches long to 140 branches that were 9-14.5 inches long and this underlines the vegetative variability that one may observe here.nevertheless, he distinguished E. flanaganii from E. woodii by the 'much shorter branches' (Brown (1915): 314) and E. discreta from E. woodii by the fact that the 'body of the plant is much smaller' (Brown (1915): 316).As commented on extensively by White et al. (1941), this makes no sense in view of such strong variation in the size of individuals.Plants producing more than one rosette of branches are not unusual and are found in many populations.Although this feature was not mentioned by Brown (1915) in his descriptions, this was supposed to separate E. gatbergensis from E. ernestii (White et al., 1941: 75), but they recognised that plants of both 'species' could produce several rosettes.E. flanaganii and E. woodii were separated by 'Ovary puberulous = E. flanaganii'; Ovary glabrous to thinly pubescent with long hairs = E. woodii' (White et al. (1941): 75, adapted from Brown (1915): 239).In practise some populations have plants with pubescent ovaries and others with glabrous ovaries and to distinguish two species on the basis of the length and density of this pubescence is untenable.Consequently all these names are reduced here to synonymy under a single species.E. fortuita A.C. White et al., The Succulent Euphorbieae 2: 962 (1941).Type: South Africa, Cape, 27 miles from Ladismith towards Barrydale, Aug. 1939, Dyer 4074 (PRE, Sheet I, holo.; K, PRE-2 sheets, iso.).
Euphorbia fortuita was included under E. esculenta in Bruyns et al. (2006).However, although in both species the cyathial glands are mostly dark and the centre of the cyathium is densely filled with white hairs, there are significant differences between them that warrant their recognition as distinct species.In E. fortuita the glands are much broader and the cyathium is more conical, having a rather rounded, almost spherical shape in E. esculenta.Furthermore, the pedicels of the male florets in E. esculenta are glabrous (densely pubescent in E. fortuita) but in E. fortuita the bracteoles are uniformly pubescent in their upper half, while in E. esculenta they are densely pubescent only at their apices.The ovary is entirely glabrous in E. esculenta and densely pubescent above in E. fortuita.
E. huttonae N.E.Br., Flora capensis 5(2): 316 (1915).E. inermis var.huttonae (n.E.Br.)A.C. White et al.: 395 (1941).Type: South Africa, Cape, Carlisle Bridge, on the Fish River, fl. nov. 1903, H. Hutton (K, holo.;gRA, iso.).[n.E. Brown based his description on a small dried specimen sent to Kew by Schonland in June 1913.Brown kept two branches and one 'flower' at Kew and sent one branch back to gRA.Thus, although he annotated each as 'Half of the type specimen', that at Kew is actually two thirds of the specimen and is taken as the holotype.]E. superans nel ex Herre: 15 (1950).Type: South Africa, Eastern Cape, July 1948, Rosenbrock sub SUG 7215 (missing).neotype (designated here): South Africa, Carlisle Bridge, nov. 1903, H. Hutton (gRA, holo.; duplicate at K). [The specimen cited here by Herre is missing (though Carter (2002) cited it as being at STE, now incorporated into nBg).The name is usually cited as 'E.superans nel' but the article in which it was published was written by H. Herre.no photograph was included with the protologue and so a neotype is selected.]Euphorbia huttonae is re-instated at the level of species for various reasons.Vegetatively it differs from E. inermis in that the rootstock does not develop a series of swollen, fusiform roots below the stem, but tapers rapidly off into fine roots.There are several clear differences in the cyathia.In E. huttonae the whole of the upper surface of the gland is bright yellow.Each gland may be divided deeply down the middle into two broad, convex, yellow structures which remain pressed together towards their bases or it may be an entire, solid wedgeshaped structure that is convex above.The outer edges of the glands are irregularly toothed and notched and may be slightly paler in some populations.In E. inermis each gland possesses a dark green part towards the base above which it is divided deeply and finely into antlerlike, white processes.Other floral differences are the spreading, white cyathial lobes in E. inermis (rather than the pale yellowish green inwardly pressed lobes of E. huttonae) and the longer styles in E. inermis which are only divided near their apex (divided much more deeply to near their middle in E. huttonae).Some confusion exists over the identity of Euphorbia superans, which was maintained as a distinct species in Bruyns et al. (2006).A figure appeared in the Euphorbia Journal (Vol.2: 138, as 'supernans') which was cited by Carter (2002) as E. superans, but the slender, bright green branches and finely toothed, broad cyathial glands make it clear that this figure is of E. flanaganii.Herre (1950) compared E. superans with E. inermis and mentioned that the glands were 'yellow...shortly bifid with two processes denticulate at the apex, divided [to] about a third with two diverging processes...slightly revolute'.This is very similar to the structure of the glands in E. huttonae but is not similar at all to that in E. flanaganii.The length of the styles and the length to which they are divided also correspond closely to E. huttonae under which E. superans is now included.
E. eustacei n.E.Br.: 122 (1913).Type: South Africa, Cape, near Matjiesfontein, Oct. 1912, C.E. Pillans (K 000253356, holo.;K, PRE, iso.).[From the material sent by Pillans and cultivated at Kew, n.E. Brown made and annotated three specimens on two sheets at K and also sent 'part of the type' to PRE. Brown annotated only one of them (K 000253356) as 'Type Specimen' (others as 'Type, branches from type plant' and 'Type Plant') and so this is taken as the holotype and the others as isotypes.] In Bruyns et al. (2006) E. eustacei was maintained as distinct from E. loricata.This does not reflect the position correctly.Dense, low-growing and mound-forming plants with slightly broader, more obovate leaves and spines drying out white have always been taken as typical of E. eustacei and were assumed to be restricted to the Matjiesfontein area (White et al. 1941), while the more diffuse, taller plants with narrower leaves and spines drying out brown that are characteristic of the valley of the Olifants River between Citrusdal and Clanwiliam are typical of E. loricata.nevertheless, White et al. (1941) hinted at a wider distribution for E. loricata and included some more densely branched plants (e.g.figure 264) in their concept of this species.now that the respective distributions have become better known it has been found that there is a gradation from the one into the other as one progresses eastwards from the valley of the Olifants River to the great Escarpment (rather than two disjunct and distinct species each confined to particular areas) so that E. eustacei and E. loricata are ecotypes of one considerably more widespread species.
[The specimen in the Herbarium of Alwyn Berger at nY consists of several small stems but has no information apart from the name on it.It was undoubtedly seen by Berger and may well be the Marloth specimen, but is designated here as lectotype.]E. multifolia A.C. White et al., The Succulent Euphorbieae 2: 962 (1941).Type: South Africa, Cape, 30 miles from Laingsburg towards Ladismith, Aug. 1939, Herre (PRE, lecto., designated here).[White et al. (1941) listed two specimens from the same locality, collected by Smith and Herre respectively and designated that by Smith the type.This being missing, the specimen of Herre is designated as lectotype.]E. namaquensis N.E.Br., Flora capensis 5(2): 325 (1915).Type: South Africa, between Aggeneys and Pella, Pearson 2992 (BOL, lecto., designated by Williamson 2007;K, SAM, isolecto.).E. multiramosa nel: 29 (1935).Type: South Africa, Cape, Little Bushmanland, flats between Jakkalswater and Vioolsdrift, Oct. 1930, Herre sub SUG 5890 (missing).neotype (designated here): South Africa, Cape, between Jakkalswater and Vioolsdrift, 600 m, Sept. 2006, Williamson 6048 (BOL, duplicate at E). [The specimen SUG 5890 cited by Carter (2002) at STE, now incorportated into nBg, does not exist.In designating a neotype, Williamson (2007) cited 'Williamson 6048 (BOL, E)'.Since two specimens are cited, this neotypification was invalid and this is rectified here.] Euphorbia namaquensis was included under E. friedrichiae in Bruyns et al. (2006).These two species are very similar and (among various 'medusoid' species occurring in the arid south of namibia and north-western South Africa), they share the feature of particularly slender branches which become thicker towards their bases.The two differ in that the cyathia-bearing pedun-cles arise in E. friedrichiae at or near the tips of the branches (the tip of the branch elongating into a peduncle in some cases) around the apex of the plant while in E. namaquensis the cyathia-bearing peduncles are shorter (and more densely tuberculate) and arise lower on the branches mainly in the lower half of the plant.The cyathia in both are of a similar size but the glands have longer and more slender processes in E. friedrichiae, while the ovary is densely pubescent with short styles (glabrous to pubescent with often much longer styles in E. namaquensis).In E. friedrichiae in namibia, the capsules often have an unusual array of warts and slightly raised wing-like ridges along the three angles while they are quite without these in E. namaquensis.However, these excrescences are usually (though not always) absent in plants in South Africa of E. friedrichiae (from east of Onseepkans), where the capsules are also often larger than in namibian plants.White et al. (1941) expressed doubt as to whether the two names E. multiramosa and E. namaquensis represented distinct species.E. multiramosa was also included under E. friedrichiae in Bruyns et al. (2006).Williamson (2007) made extensive notes on E. multiramosa and E. namaquensis.He concluded that they represented distinct species, since 'the general appearance of both plants is quite different....Cymes in E. multiramosa are only produced on the leeward aspect mostly from half to the lower third of the plant....the cymes are solitary with very short peduncles and the involucral glands smaller, sessile, horizontally curving outwards with 4-8 marginal processes.The capsules are glabrous and ± 8 mm in diameter.Euphorbia namaquensis has a single or up to two pairs of cyathia with elongated peduncles at branch apices and with involucral glands larger, shortly stipitate, suberect to erect and with 3-6 marginal processes and capsules densely pubescent ± 10-12 mm in diameter.'In practice, the 'general appearance' of plants from north of Steinkopf (taken to be typical of E. multiramosa) and those from west of gamoep (taken to represent E. namaquensis) is identical; all the other features mentioned are actually very variable within populations.Consequently E. multiramosa and E. namaquensis differ only in the glabrous vs. pubescent capsules, though even this feature has been found to be variable in E. multiramosa.The name Euphorbia patula Mill.has been a source of considerable confusion.n.E. Brown (1915: 293) suggested that it was a weak form of E. mauritanica and this was taken up by White et al. (1941: 120), while Carter (2002) referred it to E. tridentata.Both Brown (1915) and White et al. (1941) considered, wrongly, that Dactylanthes patula was published Haworth (1812), while it was merely a new combination for Miller's name E. patula.White et al. (1941) also believed that Robert Sweet (1818) described a new species 'Euphorbia patula'.However, there he referred to 'H.S.', which meant 'Haworth on Succulent Plants', i.e. Haworth (1812).Since this provided a clear reference to Haworth's book and hence back to Miller (1768), it did not constitute publication of a new, and then illegitimate name Euphorbia patula Sweet, as was assumed in White et al. (1941) and Carter (2002) but merely referred to Miller's E. patula.White et al.(1941) mation 1913: 122 (1913).Type: South Africa, Cape, near Doornkloof River, between Muiskraal and Ladismith, Aug. 1907, N.S.Pillans sub BOL 12543 (BOL, holo.; K, iso.).[n.E. Brown wrote '1 piece kept for Kew' on the specimen at BOL, which is a much larger specimen, so this is taken as the holotype.]E. pillansii var.albovirens A.C. White et al.: 965 (1941).Type: South Africa, Cape, Paardekop near Spes Bona, 650 m, 3 Oct. 1925, Marloth 12543 (PRE 0258928-1, lecto., designated here; PRE, isolecto.).[There are two specimens at PRE, neither was selected by the authors as type so a lectotype is selected here.]E. pillansii var.ramosissima A.C. White et al.: 965 (1941).Type: South Africa, Cape, between Montagu and Touws River, Aug. 1939, Dyer 4100 (missing).Marloth, South African gardening & Country Life 21: 133 (1931).Type: South Africa, Cape, near Mortimer, Aug. 1913, Shoesmith sub Marloth 5295 (PRE, lecto., designated here).[Marloth (1931) cited also Marloth 12644 but this is missing.]E. polygona Haw., Miscellanea naturalia, sive dissertationes variae ad historiam naturalem spectantes: 184 (1803).neotype (designated here): South Africa, Cape, Witpoortsberg, 2 000-3 000', Aug., Drège 8212 (S 2583; duplicates at BM, HBg-2 sheets, K, MO, P, S, W-3 sheets).[Haworth (1803) mentioned that E. polygona was described from material introduced before 1790, but nothing was preserved.A neotype has therefore been selected.]E. horrida Boiss.: 27 (1860).Type: South Africa, Cape, Witpoortsberg, 2 000-3 000', Aug., Drège 8212 (S 2583, lecto., designated here; BM, HBg-2 sheets, K, MO, P, S, W-3 sheets, isolecto.).[Boissier (1860) did not cite a herbarium and so a lectotype is chosen.]E. horrida var.striata A.C. White et al.: 964 (1941) Berger: 230 (1910).Type: South Africa, Cape, neither collector nor locality (missing).[Carter (2002) cited a specimen of Burtt-Davy at PRE as the type, but this does not exist, nor is there any evidence that it could possibly be the type of Berger's name.In his discussion of E. gorgonis, Berger (1910) mentioned having obtained plants of the recently described E. davyi from Burtt-Davy but not that Burtt-Davy had supplied him with E. gorgonis.This appears to have been mis-interpreted by Carter (2002).]

E. polycephala
The name E. procumbens was not used in Bruyns et al. (2006).This followed White et al. (1941), who did not adopt E. procumbens Mill.as the name for these plants, even though it antedated E. pugniformis (based on the same figure) by nearly 100 years, apparently because Miller's 'description is too incomplete to permit of any certainty' (p.337) in its identity and 'that name cannot be maintained at all' (p.338).However, its identity is clear from Haworth's references which lead to the present neotypification and the replacement of E. pug-niformis by this name.White et al. (1941) assumed that Sweet's (1818) use of 'Euphorbia procumbens' was a new name but, since Sweet (1818) referred to Haworth (1812) and thus indirectly to Miller, they were not correct.Marloth, South African gardening & Country Life 19: 191 (1929).Type: South Africa, Cape, near Krombeks River, Riversdale distr., Sept. 1933, Muir 4089 (PRE, holo.).
[Both White et al. (1941) and Boissier (1862) cite this name.Actually no description or diagnosis was given by Burman and he merely listed Euphorbia viminalis of Linneaus, which is the basionym of Sarcostemma viminale (L.) R.Br.(Apocynaceae).Here Burman (1768) cited 'Alp. aegypt. t. 190. Dill elth. t. 368' and he appears to have copied these references directly from among the five given by Linneaus ( 1753) for E. viminalis L. (= Sarcostemma viminale).In fact these references are wrong.In Alpini (1735) there is no t.190, but the figure referred to is t.53 on page 190.This figure is the lectotype of S. viminale, selected by Liede & Meve (1993), though it is wrongly cited there too.Dillen's Hortus Elthamensis (Dillen 1732) had only 324 plates in it and here page 386 was meant, where there is no plate.This was again cited incorrectly in Liede & Meve (1993).]E. viperina A. Berger, Monatsschr. Kakteenk. 12: 39 (1902a).Type: South Africa, Cape of good Hope?, collector unknown (missing).[White et al. (1941) placed E. viperina under E. inermis.However, the description of Berger does not correspond closely to what we know today as E. inermis.no type has been located for E. viperina.Berger (1902a) compared the inflorescences of E. viperina to those of E. caput-medusae and E. parvimamma, but in fact the inflorescences of the latter were never described and it is not clear that what he called E. parvimamma (Berger 1899) corresponds to Boissier's concept of it.]

ACKnOWLEDgEMEnTS
The curators of the herbaria B, BM, BOL, g, gRA, K, KMg, M, nBg, nY, OXF, P, PRE, S, SBT, SAM, W, WInD, Wu and Z are thanked for access to the material in their care.Christiane Anderson, university of Michigan is thanked for copies of many pieces of little-known literature and for much assistance with, and helpful discussion of, nomenclatural matters concerning several of the names in Euphorbia.Paul E. Berry, university of Michigan, with funds from the u.S. national Science Foundation PBI program (award # DEB-0616533), assisted with the costs of a trip to examine specimens in some European herbaria during which many types were located.He also provided invaluable advice on the status of many other types, among extensive comments on cited 'inter Zuurebergen et Klein Bruintjeshoogte et inter Vischrivier et Fort Beaufort (Drège nº 8206)'.The collection from 'between Fish R. & Fort Beaufort' is now labelled Drège 8206c (K) and the other as Drège 8206a (K, MO, S).]

Table 1 .
-numbers of species, available (i.e.validly published) names and showing the numbers of species exhibiting different growth forms (annuals, herbs, succulents, and geophytes) in the subgenera of Euphorbia.
E. caput-medusae L., Therefore Miller's information makes it clear that his name cannot be applied to E. loricata, even though some of the references he gave refer to that species.The reference to 'very narrow leaves' makes it more likely that this name refers to E. caput-medusae than E. inermis, among the species with a 'thick short stalk' and 'trailing branches'.At present no preserved material of Miller's Euphorbia no 10 is known and so a neotype is selected.
E. nesemannii R.A.Dyer, Klotzsch & garcke (1859)zsch & garcke (1859)published a new name Medusea patula but this, too, is wrong and this was also a new combination for E. patula Mill.Consequently, they missed the fact that Miller's name E. patula was the earliest valid name for E. ornithopus.E. pedemontana L.C.Leach, South African Journal of Botany 54: 501 (1988).Type: South Africa, Cape, foot of Matsikamma, Vanrhynsdorp distr., Lavranos & Bleck 20828 (nBg, holo.).Flanagan 2344 (BOL; duplicates at gRA, PRE).[The painting number 296/926 at K by g.Bond represents a very weak and imperfectly developed branch without spines (as noted by n.E. Brown on the painting) and it is doubtful whether this is a reasonable lectotype as it could belong to one of several species.Brown (1915) also doubted whether it was made from the plant from which Haworth described the species, since Haworth (1828) mentioned spines and these are absent from this painting.Rather than use this painting of somewhat doubtful identity as lectotype, a neotype has been selected.]E. pentops Marloth ex A.C.White et al., The Succulent Euphorbieae 2: 963 (1941).Type: South Africa, Cape, near Komaggas, 10 June 1930, Herre 5562 (PRE, holo.).