Printed in Cactus and Succulent Journal of Great Britain, 1972.

M. B. Bayer, Karoo Botanic Garden, Worcester, South Africa

Confusion in nomenclature has robbed an attractive and interesting genus of much of its charm and enticement to the collector.  For the past few years considerable effort has been made at the Karoo Botanic Garden at Worcester, South Africa, into the investigation of Haworthia.  It has been disturbing to find the extent to which current nomenclature is at variance with the natural groups as they are found in the field.  Perhaps more disturbing is the difficulty which obviously lies in reducing the complex field relationships to an understandable and usable system of classification, Haworthia can clearly be divided into three subgenera on the basis of floral characters; these are the subgenus Haworthia in which the base of the floret is rounded-triangular and in which the outer three segments completely enclose the three inner ones, and the subgenera Hexangulares Uitewaal and Robustipedunculares  (Uitcw.) Bayer in which the floret base,is rounded-hexangular (the florets stipitate in the former) with the outer floral segments separated by the midribs of the inner segments.

The Robustipedunculares comprises the sections Albicantes (H. marginata (Lam.) Stearn) and Margaritferae (H. margaritifera (L.) Duval, H. minima (L.) Duv. and H. kinqiana V. Poelln.).  The level at which confusion exists can be demonstrated by species such as H. attenuate Haw., H. fasciata (Willd.) Haw. and H. longiana V. Poelln., which once placed with the Margaritiferae, clearly belong with the Coarctatae (Hexangulares).  The four species constituting the Robustipedunculares can be defined in terms of geographic distribution alone, and the recognition of varieties is quite purposeless from the botanical point of view. This is not only because the degree of variability within and between populations does not appear to warrant such a step.  Natural hybridisation occurs between the species (as it does too between H. margaritifera and Astroloba aspera (Willd.) Uitew.), producing a large number of intermediate forms.  Also the variability in terms of the distribution ranges of each species needs a very careful study to see if there is discontinuity of variability associated with differences in locality (the best basis for recognition or sub-species).  Horticulturally the various forms are of course of interest, but to confound the naming of plants for this purpose with the naming for botanical reasons is to court chaos.  Species such as H. semiglabrata Haw., H. subfasciata (S-D) Baker, H. papillosa (S-D) Raw., H. mutabilis V. Poelln., H. poel/nitziana Uitew, and H. uitewaaliana V. Poelln. are some names clearly derived from the naming of individual plants out of their proper natural context.

The problem of redundant and invalid species is far greater in the larger subgenera Hexangulares and Haworthia.  In the latter, H. arachnoidea (L.) Duv. (including H. setata Haw.) is very widely distributed and occurs under a wide range of conditions.  Variability is enormous and this species alone may merit the same attention that the remainder of the genus does.  Species such as H. bolusii Bak., H. blackbeardiana V. Poelln. and H. decipiens V. Poelln. may prove to be a few of the lesser parts of one large composite species.  H. helmiae V. Poelln. may also be derived directly from this same complex as plants from other localities tend to indicate.  Here it can be pointed out that von Poellnitz described H. unicolor from the same population as H. he/miae and later dropped it under H. aristata Raw.  A similar situation could easily have arisen in H. venteri V. Poelln. where glabrous and setate forms also occur.

H. reticulata Haw. and H. herbacea (Miller) Steam form a rather distinctive species alliance in the S.W. Cape Province and it is interesting to note that the variety H. pallida var paynei V. Poelln (if not the species too) and the species H. hurlingii V. Poelln., H, intermedia V. Poelln., H. haageana V. Poelln., H. submaculata V. Poelln., H. luteorosea Uitew. and H. aegrota V. Poelln. were all named from within the ranks of these two species.  Field investigation shows no support for the recognition of subspecies (unless that H. herbacea and H. reticulata themselves be merged), and very little even for the rank of variety.  Confusion also reigns in the section Retusae, and here it seems improbable that a system will ever be devised to satisfy both botanist and collector.  As one example shows, all the varieties of H. triebneriana V. PoelIn. can be shown to be forms within H. mirabilis Haw., while H. badia V. Poelln. and H. mundula Smith are undoubtedly elements in this species too.  Intermediate populations between H. mirabilis and H. schuldtiana V. Poelln. also are known.  Von Poellnitz identified both H. asperula Haw, and H. pygmaea V. Poelln. from the same two widely separated localities, also indicative of a reluctance to identify new finds with the early species described by Haworth and others.  Until recently species such as H. reticulata, H. mirabilis and H. herbacea had never been associated with field populations.  Problems of nomenclature strike the collector most forcibly when dealing with the Hexangulares simply because this is the easiest group in cultivation and vegetative differences easily discernible.  H. reinwardtii must be the despair of most collectors as there are some 24 named varieties.  All these varieties certainly have very little foundation in fact, and this is also true of several ‘species’ with as little claim to recognition.  In G. G. Smith’s fieldbook, a number of plants from one locality at Bathurst are named H. greenii Baker, H. greenii var. silvicola Smith, H. reinwardtii (S-D) Haw., H. fulva Smith, H. coarctata Haw.,  H. chalwinii Marl. et Berger and H. musculina Smith!  Collections made by Mr. F. Stayner at Peddie and by the writer at Howiesonspoort have produced a similar range of ‘species’.  In several other cases more than one variety or species is named from one locality, e.g. H. reinwardtii vars brevicula Smith and diminuta Smith from Frazers Camp and the varieties kaffirdriftensis Smith, olivacea Smith, zebrina Smith, grandicula Smith and pulchra V. PoelIn. from Kaffirdrift.  This kind of taxonomy suggests that as soon as a plant is collected for naming and cultivation, the variety or species is effectively destroyed in nature!  Of course there certainly are populations such as that of the variety bellula Smith which may be distinctive as a population.  Smith was very critical of von Poellnitz and Resende for their approaches to the naming of Haworthias and yet he also made errors.  In the, case of H. baccata Smith, the species was named from a plant reputed to have come from Stutterheim, a locality quite at variance with the known distribution of the Coarctatae.  In actual fact the plant was collected at Frazers Camp, already the source of two varieties of H. reinwardtii.   The Coarctatae also include a number of species such as H. coarctatoides Resende et Viveiros, H. henriquesii Res., H. kewensis V. Poelln., H. lisbonensis Res., H. resendeana V. Poelln., H. revendettii Uitew., H. rubrobrunea V. Poelln. and H. sampiaiana Res. which in all probability are garden hybrids and as such unlikely to figure in any definitive monograph on the Haworthia.

Some further examples demonstrating the state of taxonomy in Haworthia include the description of H. planifolia var incrassata V. Poelln.  The same collection number, Triebner 861, is cited in Fedde’s Reperatorium of the previous year under H. cymbiformis.  The collection number Triebner 1059 is cited in the same way in the descriptions of H. rossouwii V. Poelln. and H. altilinea var bicarinata V. Poelln. Von Poellnitz described H. planifolia var transiens and H. cymbiformis var translucens from plants from Prince Alfred’s (not Albert’s) Pass near Knysna, and the two varieties H. cymbiformis var brevifolia and multifolia from Hell’s Gate, Uitenhage.  Von Poellnitz confessed that he was not sure of the distinction between H. cymbformis and H. plani folia (both Haworth’s species), and it does seem as though he and Smith had transposed concepts of these two species.  H. planifolia var setulifera V. Poelln. is certainly a plant from the complex around East London which Smith refers to as H. cymbiformis, and von Poellnitz also named H.cymbiformis var. obesa from the Transkei, which is nothing but an eastern form of the same East London complex.  Where H. leightonii Smith, H. lepida Smith, H. ramosa Smith and H. perplexa V. Poelln. now fit in, within the overlapping distribution ranges of H. planjfolia and H. cymbiformis will only be resolved by field investigation.  It is interesting to note that Berger wrote of H. planifolia…’Its characteristic and shortly pedicelled flowers also entirely remove it from H. cymbjformis, nevertheless it is close, but for all that must be considered a distinct species and must be placed in this special section’!  This sentence must surely have altered in its translation from the original German!  H. perplexa V. Poelln. is certainly a hybrid as suggested by von Poellnitz but probably involving H. angustfolia, rather than H. altilinea as he surmised, and H. cymbformis.

The sections as presently constituted are not very meaningful as has already been indicated. Species such as H. tuberculata V. Poelln. and H. smitii V. Poellu. have affinities with the Scabrae both ecological and morphological.  H. morrisiae is a known hybrid between H. tuberculata and H. starkiana V. Poelln..   H. caespitosa V. Poelln. is certainly a synonym of H. laetivirens Haw. and yet is placed in the section Muticae instead of Denticulatae.  H. mclarenii V. Poelin. is actually from Barrydale (not Tulbagh) and very likely is synonomous with H. aristata, yet it is placed in the Loratae and not Denticulatae.  H. nigra Haw. and H. schmidtiana V. Poelln. were also placed in different sections until Uitewaal showed them to be synonomous.  Species from the H. reticulata/herbacea complex were distributed over three sections and one of these (H. maculata (V. Poelln.) Bayer) was rated as a variety of H. schuldtiana in the Retusae.

All this confusion seems to stem from the fact that species and varieties have more often than not been described from single specimens with insufficient regard for variability within local let alone different populations which make up the species.  The choice of habitat, coupled with insect population, produces a situation in Haworthia in which populations are highly localised and inbred.  To know all the species in all their variability and association with other species is possibly a superhuman task.  One soon learns where to look for Haworthias in stable rocky situations with well-drained shallow soils, and in the absence of competitive vegetation.  Some striking examples of habitat preference can frequently be found.  At one locality near Worcester, four species are found growing on a rocky buttress some 200 metres by 50 metres, each species retained in its own localised community.  H. herbacea occurs on a lower southwestern aspect, H. reticulata on a higher more exposed western slope, H. marqaritifera on the highest level areas, and H. setata on the cool moister southern slopes.  There are many other instances where two or three species grow in very close proximity without hybridising or transgressing sometimes indefinable habitat requirements.  At Howiesonspoort near Grahamstown, H. cymbformis is found under shade on the upper rock ledges of a steep precipice, while within yards H. reinwardtii can be found in the open or among grass on the eastern slopes.  H. angustifoIia Haw. is also present and hybrids have been reported in the past.  At Thomas River near Cathcart, H. nigra is found along the immediate river bank while H. vittata Baker is abundant in the grassy slopes a short distance away.  At Oudtshoorn H. tuberculata, H. graminfo1ia Smith and H. setata are found within close distance of one another with occasional hybrids of the latter two species.  In most of these cases, species belonging to different subgenera are involved but there is nevertheless a paucity of hybrids in the field when one considers the ease with which Haworthias may hybridise in cultivation.  Differences in flowering time, apart from the genetic sterility which may occur between members of different subgenera, may account for the few hybrids.  Seedling survival may also be responsible either due to chance where the proportion of hybridised seed to pure seed is low, or where the hybrid seedling cannot cope with habitat conditions.

From the writer’s experience it appears fairly certain that geographic factors provide the most practical key to understanding Haworthia.  River drainage systems, mountain barriers, geological systems, aspect, rainfall, vegetation and altitude all need to be considered in determining the distribution ranges of the various species.  If a satisfactory two-dimensional system of classification will ever be devised is however, still open to question.  Species in Haworthia appear to have evolved as a result of selection and adaptation along ecological gradients producing an endless array of species and their intermediates.  Vegetative characters are frequently difficult to define even where differences can clearly be discerned in the overall facies of the plants. Floral characters will be of value in future classification attempts, but here too differences are small and also not absolute.

For any sense to be made in the genus it remains essential that a return be made to commonsense acceptance of basic principles of plant classification.  For names to mean anything they must surely portray the relationships of plants to one another, and to have stability and continuity, have their origin in natural populations. A species concept based on an understanding of distribution and isolationary mechanisms will help to produce a better classification in Haworthia, but it may be impossible for one man to obtain enough experience of plants in the field for this purpose.  H. reticulata and H. herbacea occupy a relatively small area and this is some 800 square miles in extent.  H. altilinea occurs in an area probably ten times this size, and H. arachnoidea is found throughout the greater part of the whole Cape Province.  It has been said that ‘many the ship of a prospective taxonomist has been shipwrecked on the rocks of the Liliaceae’.

Haworthia x submaculata, a hybrid between H. herbacea ““ind H. maculata (photo: M. B. Bayer)

Fig. 1. Haworthia margaretifera, a sturdy specimen (photo: 114. B. Bayer)

Fig. 2. Haworthia fasciata var. ovato-lanceolata from Vanstaadens Pass (photo: M. B. Bayer)

Fig. 3. Haworthia X submaculata, a hybrid between H. herbacea and H. maculata (photo: M.B.Bayer)