Printed in Excelsa 5:83 (1975).
(This article was subsequent to the revision of the Retusae by C.L.Scott).
The 162 named species in the genus Haworthia Duval have been split into 20 sections of which Retusae Haworth is one. The clearest and most obvious subdivision in Haworthia is into three subgenera which are based on both floral and vegetative characters. The Retusae fall into the subgenus Haworthia which contains the soft- leaved stemless species in which the three outer petals of the flowers completely enclose the inner petals. Division within this subgenus is the most difficult of the three and thus the composition of the section Retusae is understandably problematic. The type species of the section is H. retusa (L.) Haw., so named on account of the fat, bent, thumblike leaves. In theory this type of leaf should characterise all the other species in the section. However, in the field it soon becomes apparent that apart from the difficulty in recognising and identifying “species” related to H. retusa, some of these related elements do not have retused leaves. What then is the composition of the section Retusae and how can the individual species be recognised? So far the only really satisfactory criterion we have is the geographic distribution and relationship of a large series of rather localised populations. The individual plants in each of these populations differ to a greater or lesser degree from each other, and in similar manner, the populations differ from each other. Thus we have variability within and between populations. The composition of the Retusae is based here on visual assessment of the discontinuities of this variability, on geographic distribution, and on habitat.
Classification can be regarded as an attempt to reflect the natural relationships between groups of plants at a particular stage in time. In practice it may prove very difficult to achieve a clear-cut separation of a genus into its component species. The following 30 species were included in the section Retusae by H. Jacobsen (1965):-
H. asperula Haw. (H. retusa) (not recognized); *H. atrofusca G.G.Smith (H. magnifica var atrofusca); H. badia Poelln. (H. mirabilis);
*H. comptoniana G.G.Smith; H. correcta Poelln. (H. emelyae); H. cuspidata Haw. (insufficiently known) (horticultural hybrid); H. dekenahii G.G.Smith (H. retusa) (H. magnifica var dekenahii); *H. emelyae Poelln.; H. fouchei Poelln. (H. retusa); *H. heidelbergensis G. G. Smith; H. longebracteata G.G.Smith (H. retusa= turgida) (H. turgida var longibracteata); *H. magnifica Poelln.; H. maraisii Poelln. (H. schuldtiana); *H mirabilis Haw.; H. mundula G.G.Smith (H. mirabilis); H. nitidula Poelln. (H. mirabilis); *H notabilis Poelln. (H. maraisii var notabilis); H. otzenii G.G.Smith (H. mutica);
*H, paradoxa Poelln.(H. mirabilis var paradoxa); *H. parksiana Poelln.; H. picta Poelln. (H. emelyae); *H. pygmaea Poelln.; *H retusa (L.) Duv.; H. rossouwii Poelln. H. ryderiana Poelln. (insufficiently known) (horticultural hybrid).
*Mr. M. B. Bayer is Curator of the Karoo Botanic Garden in Worcester, Cape Province, which is one of the National Botanic Gardens of South Africa.
Excelsa No. 5, 1975. *H. schuldtiana Poelln. (H. maraisii); H. sublimpidula Poelln. (H. schuldtiana)(H. maraisii); H. triebneriana Poelln. (H. mirabilis); *H. turgida Haw.; H. willowmorensis Poelin. (H. mirabilis).
N.B. The names marked with an asterisk are accepted as good species by the author. Those without an asterisk are regarded as somewhat doubtful; these are each followed by a species in parenthesis with which they are probably conspecific. (Bold lettering are additionally as the names are recognised in my 1999 Revision).
There are several species from other sections which should be included here. These are:-
Section Muticae Berger
H. caespitosa Poelln. (H. turgida); *H. reticulata Haw.
Section Arachnoideae Haw.
H. guttata Uitewaal (H. schuldtiana) (H. maraisii); *H. herbacea (Mill.) Stearn.
Section Denticulatae Baker
H. laetivirens Haw. (H. turgida)
Finally there are several recently described species which must also be considered to belong to the Retusae:-
*H. springbokvlakensis C.L.Scott; H. geraldii C.L.Scott (H. retusa); *H. pubescens Bayer; *H. serrata Bayer.
There are also other species in the subgenus Haworthia which occur in the same geographic area as the species listed above and these include:-
H. mucronata Haw.; H. floribunda Poelln.; H. angustifolia Haw.; H. variegata L.Bolus; H. arachnoidea (L.) Duv.; H. decipiens Poelln.;
H. truncata Schonland; H. maughanii Poelln.; H. blackburniae Barker; H. graminifolia G.G.Smith; H. chlorocantha var. subglauca Poelln.
Except for H. floribunda, these latter species do not appear to be immediately involved in the section Retusae. Hutchison (1951) suggested that H. truncata and H. maughanii belonged in the Retusae. However, these two species have a different floral structure (a second paper by Hutchison intended to discuss differences in the flowers was apparently never published), and the leaves of the young seedlings are properly truncate as opposed to the bent leaf-ends in Retusae seedlings. Furthermore the two species do not fit into the general variability pattern at all.
Apart from geographic relationships, the members of the Retusae can be recognised by morphological characters. The first of these is thick bent-back (retused) leaf ends which is a character most pronounced in the eastern-most H. springbokvlakensis. This is not, however, characteristic of all the species and several may have retused leaves as well as erect or incurved leaves. The second main character is a floral one in which the buds are flattened at the tips and flared apart in “whale-tail” fashion. Again this is not true of all the species, but it does reach strongest expression in the west where the species may not have retused leaves.
Names are the basis of communication and the first problem in writing an account of these various elements is that of which name to use. It should be remembered that names can only be meaningful if they can be permanently and confidently associated with the herbarium specimen representing that species. None of Haworth’s species are represented by such types, neither are Baker’s, nor are a great many of von Poellnitz’s. Without adequate reference to geographic origin, it is virtually impossible to even guess at the relationship of such names to actual field populations. The illustrations in Salim Dyck’s “Monographia” (1836-43) are the only pictorial record of many of Haworth’s species, and it is unlikely that Baker or Berger had any other evidence on which to base their revisions in later years. The fact that many species are still missing is therefore more a reflection of the difficulties involved in applying old names to existing field populations, rather than of deficiencies in knowledge as to their whereabouts.
H. asperula Haw. is known only from the original short description and from a plate by Salm Dyck. Haworth likened this plant to H. retusa, whereas Berger (1908) compared it with H. mirabilis. It was described as having a roughened end-area with up to ten face lines as opposed to the smooth five-lined upper leaf surface of H. retusa. In some populations of H. retusa however the leaf faces may be roughened and the number of lines may vary from 5-12, even varying to a lesser degree in the leaves of one specimen. H. asperula thus could easily belong in synonomy with H. retusa. (I eventually decided to discard this name because of the associated confused historical record).
The name H. multifaria Haw. was set aside as a synonym under H. mirabilis by Baker (1896) and there is virtually no way of ascertaining the validity of this. Even had Haworth’s original collection been at Baker’s disposal there would still be some doubt that the association of names with plants was correct.
H. atrofusca G.G.Smith does not appear to be represented by a type specimen, but it was described from a known, albeit small, single locality near Riversdale. Characterised by its reddish-brown colour and rather obtuse roughened leaf-tips, it grows in very close proximity to H. retusa and can be recognised as distinct from that species. The probable relationship is with H. schuldtiana and H. magnifica.
H. badia Poelln. from near Napier in the SW Cape is an easily recognised element (subspecies?) in the H. mirabilis complex. It grows in pebbly sandstone amongst Macchia vegetation, near to H. mirabilis, but the latter however is found in Coastal Renosterbos vegetation on shale outcrops. The flower and flowering time are the same. Variability in Haworthia can often be seen to be associated with changes in habitat, though it is not always easy to recognise this.
H. comptoniana G.G.Smith occurs in the Georgida area between Willowmore and Uniondale. It is characterised by the highly reticulate and white-flecked, broad, smooth end-area to the leaves. Geographically it is allied to H. emelyae but separated by gaps in distribution which may exaggerate actual differences between the two species.
H. correcta Poelin. is one of the species of which the type no longer exists. It was very poorly illustrated in Kakteenkunde (1937) and von Poellnitz likened it to H. pygmaea. However, he was apparently confused in his naming of specimens from the two complexes and the fact that H. correcta was recorded confirmed in unpublished correspondence from the Calitzdorp area suggests that it is in fact a synonym for H. emelyae.
H. cuspidata Haw. is a source of considerable confusion. Haworth compared it with his H. mucronata, which is a name applied to plants from both an Eastern Cape complex and a Little Karoo complex, neither of which are associated with the Retusae. Von Poellnitz applied the name first to plants from George (which he later named H. turgida var. suberecta), and then to plants of unknown origin received from A. J. A. Uitewaal. Berger’s illustration (1908) of H. cuspidata appears to be of a hybrid involving a member of the H planifolia Haw./H. cymbiformis Haw. complex and similar plants are often seen in collections under the name H. cuspidata. It is possible to ally the name with variants of the Retusae complex but there can be little point in adding to the confusion and the name should be rejected on this basis.
H. dekenahii G.G.Smith is something of an anomaly – it is heavily flecked (with silvery-white flecks) and the leaf-ends are rather elongate. The single population is restricted to a few solitary surviving plants and this is near to a population of similarly flecked highly proliferous plants (H. turgida var. pallidifolia G.G.Smith). In the field the variety H. dekenahii var. argenteo-maculosa G. G. Smith can fairly certainly be associated with H. retusa: only a few of the plants at the type-locality are as heavily flecked as described in the variety. There seems to be little point in upholding the name H. dekenahii as a species for the few plants which do occur. These are not dramnatically different from other variants of H. retusa, although they may prove to be an interesting link with the series of plants occurring in the Gouritz River valley discussed later under H. turgida. The writer has seen a few isolated specimens from the Humor area east of the Gouritz River bridge in which the leaf-ends are very rugose and very rounded at the tips. However, it is also unlikely that even these occur in any consistent series which could require taxonomic recognition. (The problem has not gone away and is the subject of a manuscript in preparation Jan.2007).
H. emelyae Poelln. was not recorded from any precise locality when described and neither is a specimen preserved. However, collection data from both G. G. Smith (unpublished) and Mrs. Emely Ferguson (C.L.Scott, Aloe 11:4, p 44, 1973) indicate that this species came from the Vanwyksdorp area. Plants matching the description do indeed occur in that area and it seems fairly certain that H. picta and H. correcta are superfluous names for this same species. (Some authors seem to be intent on making this a personal issue and play with the names picta and correcta the latter for H. bayeri). It is usually characterized by pinkish reticulation of the upper leaf- face. Apart from a relationship with H. comptoniana, there is a population a few miles east of Muiskraal (Riversdale) which links H. emelyae with H. schuldtiana var. major G. G. Smith from Muiskraal itself. These latter populations are difficult to separate entirely from H. retusa as it occurs at Gouritzmond, and H. paradoxa at Vermaaklikheid.
H. fouchei Poefin. is simply a variant of H. retusa from near Riversdale which could perhaps be used to prove a continuity with H. longibracteata occurring in the lower reaches of the same river system.
H. heidelbergensis G. G. Smith is a small species from immediately east of Heidelberg. It has many leaves and grows in the same area as H. turgida, H. retusa, and H. schuldtiana. It appears to have affinities with a population from Springfontein north of the Langeberg mountain range as far as vegetative character is concerned. There is a similar population from Matjestoon to the southwest of Heidelberg, and another from Rooivlei near Bredasdorp.
H. longibracteata G.G.Smith cannot be upheld as a distinct species. It varies continuously with H. retusa along the Kafferkuils River east of Riversdale and perhaps also with the aberrant populations of H. turgida along the Duiwenhoks river south of Heidelberg. In the latter area there are populations which cannot positively be allied with either H. retusa or H. mutica.
H. magnifica Poelln. grows to the south east of Riversdale and at several other localities including west of Heidelberg. Were it not for a typical population of H. schuldtiana at Heidelberg itself, H. magnifica would be regarded simply as the eastern equivalent of this small dark green species. Nevertheless it may be an ecotype of H. schuldtiana associated with the Witteberg geological series. H. maraisii Poelln. is clearly a large form of H. schuldtiana from Stormsvlei and not a notable variant of that species. (The name H. maraisii is the earlier of the two names, and if the scheme of this paper was implemented, it would replace H. schuldtiana as the valid epithet for this species). South of the type locality there is an odd population which appears to be of hybrid origin between H. schuldtiana and H. mirabilis. As in H. herbacea and H. reticulata (Bayer, 1972) this would again indicate that related species may at some points of their distribution ranges be so discrete that hybrds are recognisable, whereas elsewhere it may appear that the two elements are still incompletely segregated.
H. mirabilis Haw. can easily be associated with a large complex in the Caledon, Bredasdorp, Greyton, Riviersonderend area. The relationship with H. schuldtiana is somewhat problematic although it appears to flower slightly earlier (Jan./Feb. as opposed to Mar./Apr.). The flower buds are longer and more slender and invariably the veins of the flower petals are brownish. The plants themselves are usually larger than in the case of H. schuldtiana and the generally accepted difference of a smooth end-area as opposed to roughened end-area may hold true. H. mirabilis is a fairly variable species but not to the extent that the many synonyms might suggest. H. triebneriana is almost certainly a synonym and the eleven described varieties of this species also nearly all belong under H. mirabilis along with H. nitidula and H. rossouwii. H. mundula, as is the case for H. badia, appears to be only an ecotypic segregate from southwest of Bredasdorp, having again the same flower and flowering time. In the upper Riviersonderend River valley, there are longer-leaved forms (H. triebneriana var. rubrodentata Poelln.) while at Skuitsberg a little further down the valley, very large broad-leaved forms occur (H. emelyae var. beukmannii Poelln.).
H. nitidula Poelln. is another inadequately located and typified “species” and it is only with the help of Smith’s collecting records that it is possible to pin the name to a plant from the Greyton area. The only species which occurs here, apart from a segregate of probably H. turgida (now H. mirabilis var consanguinea), very deep in the mountain behind the village, is H. mirabilis. Thus H. nitidula is clearly synonymous with this latter species.
H. notabilis Poelln. occurs in the low mountain foothills north of Robertson in several populations. However, these have a very close affinity with H. schuldtiana and is perhaps best regarded as a subspecies only.
H. otzenii G.G.Smith by some mischance is not represented by a type specimen. Contrary to statement in the original description, a specimen was never received at the Compton Herbarium. The species is allied by Smith to H. mutica, but at the locality recorded by Smith and following Otzen’s collecting numbers (obtained from an unpublished record by J. Luckhoff), only H. mirabilis occurs. It is most probable that Smith confused numbers with some other collection of Otzen’s which were in fact of H. mutica – which was also represented in Otzen’s collecting. Thus H. otzenii can be regarded as a synonym of H. mutica.
H. paradoxa Poelln. was described from Vermaaklikheid south of Riversdale and is only known to the writer from this locality. There is a suggestion in Dekenah’s collecting records that it may occur in the same calcareous ridges eastwards to Stilbaai. It is very similar to H. magnifica and likewise might also be an ecotype of H. schuldtiana. The nearest similar population known to the writer is several miles to the west at Brakfontein; this population is affiliated either to H. turgida or H. reticulata.
H. parksiana Poelln. is a very distinct small species from the Great Brak and Little Brak area between George and Mossel Bay. It derives its name from a strange confusion in which von Poellnitz took Major Long’s accession number Parks (for Port Elizabeth Parks Department) 636/32 to refer to a Mrs. Parks, and hence the species has a fictitious name. It was actually first collected by Mrs. Helm. The flower is not typical of the Retusae as the buds are long and thin and are rounded at the tips.
H. picta Poelin. was collected at Moeras River, cited by von Poellnitz as in the Little Brak River area. The locality is actually north of the Langeberg Mountains nearer Oudtshoorn. Populations in this area are continuous with H. emelyae and the name H. picta cannot be upheld. Some specimens in this complex can be very scabrous and von Poellnitz applied the name H. asperula to plants from the Oudtshoorn area.
H. pygmnaea Poelln. is from Great Brak River and it grows together with H. parksiana as well as near to H. turgida var. suberecta. It is not certain whether this species is distinct as it can often barely be separated from H. mutica from the Caledon/Bredasdorp area. The only differences readily apparent to the writer are that in H. pygrnaea the bract veins are purplish and the leaf faces tend to be papillate. There may be a relationship with the H. emelyae complex across the Langeberg mountains but this is not obvious.
H. retusa is a name which is associated with the Riversdale area where many populations occur varying to a greater or lesser degree from the earliest illustrations of the species. Several varieties were described by Smith, and ecotypes occur to add to the confusion. Essentially the species is comprised of fairly large plants up to 8 cm in diameter with broad (2 cm) strongly retused leaves. In the Riversdale area they may or may not be proliferous and are frequently bright green to yellowish and often brown in colour. In the south west the plants are less proliferous, darker green to purplish in colour (in cultivation too) and the leaves do not have an end awn. The writer regards these latter populations as comprising the species H. mutica (This decision is the subject of a series of articles with respect to H. mutica var nigra).
H. rossouwii Poelln. is another species of slightly doubtful origin but is does appear to have come from the mountains south of Bredasdorp – the same locality as H. triebneriana var. sublineata Poelln. This indicates that it is also simply a synonym of H. mirabilis.
H. ryderiana Poelln. is of unknown origin and in the absence of an illustration it is not possible to suggest an alliance with known field populations. The name should be rejected as insufficiently known.
H. schuldtiana was described from Robertson (as corrected by von Poellnitz in a subsequent note) but the collector, G. J. Payne’s unpublished records indicate that the original specimens came from MacGregor. The Worcester/Robertson area contains many populations referable to this small dark scabrous species. It is found at Barrydale and at Montagu in the Little Karoo as well as at Swellendam and Heidelberg to the east. It is very variable and it is unfortunate that the name H. maraisii was earlier used for the larger forms in the Stormsvlei area. H. schuldtiana is also found southwards to Bredasdorp and along the western side of the Breede River. At Bonnievale the populations are vegetatively very like H. herbacea, but can be distinguished by the comparatively short buds and small green-lined flowers.
H. sublimpidula Poelln. was described from Swellendam and the illustration clearly places the plant as one of the variants of H. schuldtiana. There are several populations from between Swellendam and Bonnievale from which the original specimens may have come. (More recent collections suggest the population now within the Bontebok National Park just east of the Breede River. My latest opinion is that considering my full collecting record, that it is not possible to distinguish H. mirabilis and H. maraisii and that there are other complicating considerations too).
H. triebneriana Poelln. has been the source of some considerable confusion in so far as applying the name to a field population is concerned. The species was, together with H. willowmorensis, said to have come from Strydomsvlei in the Willowmore area. No plants agreeing either with the original descriptions or illustrations are known in that area and it seems fairly certain that the locality was confused perhaps with Stormsvlei. Both species are considered by the writer to be synonomous with H. mirabilis.
H. turgida Haw. is another name which the writer feels has been wrongly applied. The species is illustrated by SaIm Dyck and can be allied with populations to the north of Heidelberg particularly, to which the names H. laetivirens (also illustrated by Salm Dyck) and H. caespitosa Poelln. can be applied. This is a very widespread species and in view of its apparent similarities to species such as H. herbacea and even H. notabilis, it is perhaps archetypal of the Retusae. It occurs in the wetter sandstone mountains from Swellendam eastwards to Riversdale. There are related populations in the mountains between MacGregor and Greyton. A specimen has been collected from the Potberg near Infanta, and there are notable similarities particularly to a form of H. herbacea high in the mountains north of Worcester. There are three populations in the lower-lying areas which can‘iot be clearly allied with this concept of H. turgida, and they may be intermediate between this species and H. reticulata. These populations occur at Brakfontein south of Riversdale, at Rooivlei to the north of Bredasdorp, and at the Breede River bridge near Swellendam. In addition, there are short, rounded-leaved plants from the Gouritz River area and nearer Albertinia which are also seemingly removed from this concept of H. turgida. The Gouritz plants are very obtusely round-leaved as well as being strongly flecked as in H. turgida var. pallidifolia. There is evidence to suggest that such plants extend up the Gourits river to as far as Die Hell north of Calitzdorp. It still needs to be shown that there is such a series of populations which could be taken to constitute a species. (This evidence was in the name rodiniiand it is certain that it came from lower down the Gouritz where it is formally named H. turgica var suberecta. I consider that the proper solution is that H. retusa and H. turgida are the same one species).
H. willowmorensis Poell. is a poorly recorded species as already noted. It appears to be synonymous with H. mirabilis on the basis of the poor illustration and doubts as to origin.
H. guttata Uitewaal was described from between Robertson and Bonnievale. The only species which occur there are the small form of H. reticulata (described as H. hurlingii Poelln.) and variants of H. schuldtiana. Thus H. guttata is easily allied with the latter species and is a superfluous name.
H. reticulata Haw. and H. herbacea (Lam.) Stearn have been the subject of two separate papers by the writer (Bayer, 1972). The association with the Retusae is through floral character and by indirect association in the field via H. pubescens Bayer and H. schuldtiana var. maculata Poelln. which the writer would uphold as a discrete species. H. pubescens shares characteristics of both H. herbacea and H. schuldtiana, while H. schuldtiana var. maculata is geographically and morphologically continuous with H. herbacea and some forms are vegetatively indistinguishable from H. notabilis. H. reticulata and H. herbacea have the very notable bi-arcuate and “whale-tail” buds found only in the Retusae. Both species are confined to the Worcester/Robertson Karoo. H. reticulata is the more variable of the two species and it is clear that populations at the Breede River bridge near Swellendam, have to be considered with either this species or H. turgida or both.
H. springbokvlakensis C.L.Scott is a new name for a species from the Steytlerville area previously taken by various persons to be Haworth’s H. mutica. It has very large strongly rounded thumb-like leaves and is a striking species. There is a related population in the Uniondale area which does not however have the convexity of leaf face found in H. springbokvlakensis. The affinity of either of these two elements with the geographically nearest species, H. comptoniana, needs clarification.
H. serrata Bayer is a distinctive element from southwest of Heidelberg which can perhaps be referred to those populations mentioned under H. heidelbergensis. It is a much larger species and some specimens have the same general form as H. herbacea. The floral structure is very like that of the population referred to from Springfontein.
The very nature of the problem makes the above account a very subjective one. The accompanying map is an attempt at illustrating the relationship of the various species in the field with the firm belief that Darwin’s words written to Hooker in 1845… “That grand subject, that almost keystone of the laws of creation, geographical distribution”, are true. If this is a fair account of the Retusae it is still extremely difficult to rationalise and justify without an intensive study of anatomy. Even was this available, the problem of variability would still be difficult to overcome as single specimens can seldom be taken to represent whole populations. Flowers in Haworthia, just like leaf form and structure, are very uniform throughout the subgenera. It was only in 1947 that Uitewaal drew attention to floral characters which could be used to sub-divide the genus but there do not appear to be other macroscopic floral characters which can effectively be used to base further subdivision on. A better understanding of the taxonomic situation in Haworthia may perhaps be derived from a closer study of a few selected species. It may be too much to expect, but it is hoped that the present paper will form a reliable guide to any future work that may be done.
The writer would like to thank Professor H. B. Rycroft, Director of the National Botanic Gardens of South Africa, for authorising preparation and delivery of this paper.
Jacobsen, H. in Kakteen und andere Sukkulenten, parts 3,4,6 and 7: 50 (1965).
Hutchison, P.C. in The Cactus and Succulent Journal of America, 23: 99 (1951).
Dyck, J.P., Monographia (1836-1849).
Berger, A. in Das Pflanzenreich, 4 (1908).
Baker, J.C. in Flora Capensis, 6 (1896).
Bayer, M.B. in National Cactus and Succulent Journal, 27: 10 (1972).
Bayer, M.B. ib., 27: 51(1972).
Uitewal, A. J. A. in Desert Plant Life, 19: 133 (1947).