The classification of Haworthia and particularly the sub-genus Haworthia is fraught with difficulty, because it is so difficult to circumscribe species either in writing or pictorially. The solution I have followed is to recognise species as geographic systems, knowing that there are continuities which can only be resolved where the systems co-exist as apparently independent entities. Even this is difficult because it is actually seldom that true co-existence occurs. Usually when two species (even of different sub-genera) occur in close proximity, they occupy different habitats. In some such cases it is possible to reasonably conclude that the differences in appearance are only at the level induced by that difference in habitat, or at a higher level at which they can be taken to be discrete systems. With the small qualitative differences that actually determine these subjective judgements, there seems to be little that can be done to quantify them.
In this article I will just discuss a few examples which I can support with photographic evidence, and inform readers that there is far more extensive published and unpublished manuscript and photographic evidence available.
Starting with H. decipiens var. pringlei. This element has an independent history in which it can be argued that it can be upheld as discrete if one so wishes, placed with bolusii var. blackbeardiana with good reason, and combined either with H. aristata sensu Bayer, or with H. cooperi. The picture I use is of JDV93/48-1 of co-type material of Scott’s pringlei from north-east of Middleton. Then I take MBB6872-2 of H. gracilis var. isabellae from the Krom River estuary south-west of Humansdorp. I also use JDV90/80-3 from Engelandsekloof in the Baviaanskloof which is blue-green in colour and hence H. cooperi var. gordoniana. From there I leap to H. mucronata var. mucronata MBB6872-4 from Pietersfontein north-west of Montagu .
It is then not a substantial leap qualitatively, although it is geographically, to go to JDV97/8-5 Ouplaas at the base of the Cockscomb, north of the Baviaanskloof. This element is H. gracilis var. isabellae, and an observable connection between that element and H. decipiens var. minor. But I have illustrated two clones – one which is more characteristic in that it has few spines, and one which is a lot spinier. There is not much difficulty in finding reciprocity with JDV97/63-6 H. cooperi var. cooperi at Glen Avon, Somerset East, or with MBB6556-7 at the inland Ripon Station near Middleton (cooperi or pringlei!). Nor is it difficult to go from there to MBB6799-8 north-west of Patensie, again cooperi var. gordoniana. There is a problem that there are again both spined and unspined forms MBB6799-9.
Going back to arachnoidea and mucronata. There are a few localities where these two elements can be said to co-exist, but within the limits which I described above. Mucronata is not always clearly separable from arachnoidea. The difference lies only in the degree of translucens of the leaf margins, with arachnoidea having no marginal translucence, and mucronata with. MBB6867-10 is two clones of arachnoidea from north-east of Montagu, near Ouberg. The plants do not have translucence, but they do exhibit a block-pattern on the leaves. This block-pattern can apparently become translucent but I have not observed it closely enough to really know what the significance is. When spines with translucent bases occur on the leaf surfaces, there is similarity with H. marumiana var. dimorpha. But before getting excited about these ‘characters’, lets look at a collection MBB6883-11 from between Montagu and Barrydale. This population is not unique in terms if the variation it exhibits. I show three clones. one is probably comparable with -10 from Ouberg, while the other two are somewhere on the way to mucronata. I would refer them to arachnoidea var. nigricans only because of the absence of translucence.
How significant this tranlucence is, is hard to say. JDV98/83-12 is a plant from Sapkamma, which I class as H. decipiens var. minor. It is unusual in that it has the block pattern and no translucence in contrast to all the other specimens in the collection which are highly translucent. Thus the distinction we use to separate arachnoidea and mucronata is somewhat diluted, and it certainly means that xiphiophylla is more probably linked to decipiens than it is to arachnoidea, where I have placed it.
A further demonstration of the value of translucence and patterning of the leaves is obtainable from MBB6850-13. This is a collection of H. cymbiformis var. obtusa frm Swartwaterpoort. The collection is probably unusually variable. One plant has no tranlucence and only the block patterning, another has marginal tranlucence, and a third has a very reticulate patterning of the leaf-end, with marked translucence. The degree of ‘windowing’ of the leaves is very variable in other populations too. H. cymbiformis var. setulifera in MBB6573-14, Highclere, Cathcart (and the classification here is shaky, as the plants are uncharacteristic for either setulifera or reddii and certainly also for the only other geographic candidate which is blackbeardiana). My picture shows a plants with a highly reticulate windowing of the leaf.
In MBB6771-15, which is a collection of H. gracilis var. picturata from Moordenaarskloof in the Longkloof, I try to show that the translucent reticulation can vary. The illustrations do not show this very well, but the windows can be large and rounded, or narrow and elongate. The degree of spination is also variable. It has to be noted too that this element falls into a no-mansland between cooperi as gordoniana, and cymbiformis as transiens – what to say of its classification as a gracilis variant. Lastly I illustrate for interest mainly, a collection of H. cymbiformis var. reddii frm Inverbolo in MBB6843-16. This was actually collected first by P.V.Bruyns and the illustrations must convince anyone of the affinity with Waterdown Dam – reddii. Bruyns has also recorded this element from Inversomo, although the clones in cultivation have a small (very small) element of doubt attached to it because of the Inverbolo collection. The plants are interesting because the translucence is fairly obscure. If viewed with the light source behind the plants, the translucence of the leaves is very much more obvious; showing up as block-patterning with the light behind the viewer.
It is a fascinating subject and I say repeatedly that the problems of classification in Haworthia are very much more common in other plant families and genera than plant taxonomists want to admit.