Herbertia 56: 61-66 (2001)
SOUTH AFRICA'S JEWELS OF THE DESERT:
A REAPPRAISAL OF THE GENUS DAUBENYA (HYACINTHACEAE)

John Manning
Compton Herbarium, Private Bag X7, 7733 Claremont, South Africa

Fig. 1. Daubenya aurea

Fig. 2. Daubenya capensis

Fig. 3. Daubenya namaquensis

Fig. 4. Daubenya stylosa

Fig. 5. Massonia jasminiflora.

Fig. 6. Whiteheadia bifolia.

There must he few lovers of bulbs who have not heard of, if not actually seen, that extraordinary Cape member of the Hyacinth family, Daubenya aurea. Looking for all the world like the discarded peelings of some brilliantly colored citrus fruit, the plants have an exotic, almost bizarre appearance that is in stark contrast to the austerity of the high plateau landscape from which they spring. Rare in nature, the species is restricted to a handful of colonies scattered in a narrow strip along the base of a stony ridge that meanders along the escarpment at the southeast margin of South Africa's interior plateau. From north to South this ridge is no longer than 10 km in length. Nowadays, the location of some of these colonies is well known, and, as they are adjacent to good quality secondary roads, it is not a difficult matter to visit the plants. This was not always the case, however, and the early history of the species is one of mystery and excitement. But no matter how familiar the species might have become, it never fails to excite admiration in all who see it (see "Magnificent Daubenya" by John Lavranos & Charles Craib in Herbertia 32: 22-25 for a detailed discussion of the species). It certainly succeeded in amazing the British botanist, John Lindley, when he described it in 1835, and he did not hesitate erecting a separate genus just for it.

The genera of Hyacinthaceae are notoriously difficult to define adequately. In practice, this has resulted in the segregation of various groups of closely related species as separate genera by taxonomists. The species of each segregate genus share one or more morphological peculiarities that set them apart. The remaining, less distinct species, have then been relegated to more or less ill‑defined genera. Known more picturesquely as portmanteaux genera, these botanical catch-aIls include Ornithogalum and Scilla. An alternative, and arguably even less meritorious practice, favoured by some botanists has been to recognise every small group of species within these "catch-all" genera as a separate genus in itself. A proliferation of small or even monotypic genera has been the result, each of which is defined by various minutiae that serve only to obscure more general relationships between these groups of species. All uneasy mix of these two philosophies currently holds sway within the Hyacinthaceae. The morphological tangles that are strangling the emergence of a meaningful circumscription of genera within the family are only now beginning to give way. The sword that is cleaving this Gordian knot is DNA sequencing. Employed correctly, this technology constitutes a powerful new source of data for analysing the relationships between species without the confusing façade that morphology often throws over the underlying structure. One of the groups within the Hyacinthaceae that is yielding particularly well to this methodology is the alliance of largely South African genera recognised by some workers as the tribe Massonieae. This includes the somewhat confusing assemblage of stemless Cape genera characterised by a pair of prostrate leaves: Androsiphon, Amphisiphon, Daubenya, Massonia, Neobakeria and Whiteheadia.

Epitomising the traditional taxonomic practices that have been employed in the family, the genera Androsiphon, Amphisiphon, Daubenya and Whiteheadia were all erected to accommodate single species that differed from their relatives in various attributes of floral morphology. Floral variation within genera of the Hyacinthaceae is on the whole relatively modest, thus where relatively major differences occurred, they were construed to represent sufficient grounds for recognising separate genera. The intellectual climate of the time in botany stressed differences over similarities. For example, in Androsiphon, the stamens are fused into a slender tube. This peculiarity also characterises Amphisiphon but is accompanied here by a fusion of the tepals. Daubenya. was characterised by the highly exaggerated outer tepals in the lower flowers, while Whiteheadia was recognised by its elongate spike of sessile flowers sheltered beneath a conspicuous crown of sterile bracts. Predictably, the two remaining genera in this alliance, Massonia and Neobakeria, were less easy to recognise and, in fact, of late are usually combined as one. This is in contrast to earlier workers who reserved the genus Massonia for those species with large bracts and very condensed, rounded inflorescences. Neobakeria accommodated the remaining few species with smaller bracts and more elongate inflorescences.

Most of these small genera are as rare or even rarer than Daubenya aurea itself. Both Androsiphon and Amphisiphon are known from only two or three populations around the village of Nieuwoudtville in Northern Cape. This small farming settlement lies a little way from the western edge of the Bokkeveld Escarpment, whose sandstone ramparts form the northernmost extension of the Cape Floral Region. Nieuwoudtville, with good reason known locally as the Bulb Capital of the World, lies across three very dissimilar soils and their associated vegetation types. This geological complexity coincides with a sharp rainfall gradient and relatively open vegetation to provide a diversity of habitats highly suitable for geophytes. Excavation of soil samples in the vicinity has yielded the almost unbelievable total of 25,000 bulbs in a square meter!

Most of the four accepted species of Neobakeria are likewise known from small areas. The two most localised are N. namaquensis, from the sandy flats east of Springbok in Namaqualand, and the species recognised as N. burchellii, which is known from two populations on calcareous coastal sands between Saldanha Bay and Paternoster, just over 100 km north of Cape Town. The two remaining species, N. comata and N. marginata are more widespread across parts of the southern African interior plateau. In sharp contrast to the highly localised nature of these species, most species of Massonia and Whiteheadia are relatively widespread across suitable parts of the country. For all six genera this means the winter‑rainfall region of southern Africa, particularly the western parts of the interior plateau.

Although the species in all of these genera, with the possible exception of Massonia itself, are fairly well understood, the same could not be said until recently about the relationships between them. It seemed likely that Amphisiphon and Androsiphon were related to one another by virtue of their unusual, fused stamens. In turn, both were considered to he near Daubenya, if only because they were not white or pink in color. Any more exacting relationships between the genera remained unresolved. Massonia was usually combined with Neobakeria and, in an extension of this, Lavranos & Craib (l.c.) were led to the conclusion that Daubenya was 'closely related to Massonia, from which it seems to differ only in the dimorphism of its florets...'. Actually, irregular lower flowers are also found in the inflorescences of Neobakeria namaquensis and Massonia comata, although of a different form than that evident in Daubenya. This floral dimorphism is quite marked in N. namaquensis and led the German workers, U. & U. Muller‑Doblies to restrict the genus to that one species alone, removing the remaining species of Neobakeria to the genus Massonia. Matters rested uneasily here until the investigation of DNA sequences within the family began to shed some extraordinary light onto the problem.

The least surprising result of these studies was the close relationship that they indicated between Whiteheadia, Massonia and the genus Lachenalia. The large, predominantly winter-rainfall genus Lachenalia had long been considered to form part of this alliance, largely on the basis of its smooth, rounded seeds arid more or less tubular flowers. The results from DNA sequence analysis were thus merely a vindication of the morphological data. What was quite unexpected, however, were the relationships suggested for the remaining genera in the group. Although there was no doubt that the genera Amphisipihon, Androsiphon, Daubenya and Neobakeria were extremely closely related to one another, there was no evidence at all for an anticipated relationship between these genera and the Massonia-group. What DNA sequence analysis in fact suggested was that the similarity in the leaves and inflorescence structure of these two groups was an instance of convergence or parallel evolution. Moreover, as often occurs when molecular evidence of phylogenetic relationship is incongruent with morphological data, a closer examination of these characters supported this conclusion. The earlier distinction between the species of Massonia with large bracts and the species of Neobakeria with small bracts was thus spectacularly vindicated. However, another almost more startling revelation was to emerge. The genetic differences between the species of the Daubenya-alliance were so small as not to justify the recognition of more than one genus for the lot! All of the species previously recognised as comprising the four genera Amphisiphon, Androsiphon, Daubenya and Neobakeria are most accurately treated as a single genus. Fortunately for us, the oldest name for this group is Daubenya. The result of all this molecular genealogy is that the genus Daubenya has been enlarged to accommodate another six species. With one stroke the genus has thus become the morphologically most diverse in the family in terms of flower form, but retains a striking underlying similarity in the vegetative morphology of these apparently disparate species. All of them share unusual glossy leaves which have the surface marked with longitudinal depressed lines and which fragment at the base into narrow papery strips when they dry. The cause of the startling differences in the flowers of these species has become evident with further study of the species in the wild. They relate to the wide variety of different pollinators to which each of the species is adapted, including honeybees, noctuid moths, monkey beetles, sunbirds and possibly even rodents. Such a diverse spectrum of pollinators is not known in other genera of Hyacinthaceae and is clearly the reason for the unprecedented floral diversity among these species. Botanists, like the pollinating agents that fuelled the process of natural selection, discriminated between the species. While this was a matter of survival for the animals concerned, it merely resulted in a great deal of intellectual confusion for the humans!