Taxonomy and Natural History of Hydnora ( Hydnoraceae )

Hydnora is a genus of subterranean holoparasitic herbs found in arid and semiarid regions of Africa, Madagascar, and the southwestern part of the Arabian peninsula. Results from field and herbarium studies suggest the genus consists of four or five species, although more than 12 have been described. The recent rediscovery of H. triceps, a plant that had remained uncollected for a century, supports the need for additional field work. Taxonomic research has been impeded by a paucity of collections, which are often fragmentary in nature and poorly preserved. Supraspecific classification, species complexes, floral biology, and uses are discussed.


INTRODUCTION
The two genera of the Hydnoraceae, Hydnora and Prosopanche, include some of the strangest plants in the world.The vegetative plant body is highly reduced and the plants are entirely subterranean except when flowering.Prosopanche is found in South America (Kuijt 1969); a new species has recently been described from Central America (Gomez 1984).Hydnora is essentially an African genus, although two species are found in the southern Arabian peninsula.Little information on Hydnora is available due to its furtive nature and often seasonal appearance.
This paper reviews the infrageneric taxonomy of the group and provides notes on the natural history of H. johannis Becc.and H. africana Thunb., two species we have studied intensively during the past several years.We have drawn heavily upon earlier papers (Musselman 1984;Musselman and Visser 1987) in preparing this work.
The most recent monograph of Hydnora is that ofHarms (1935), a work based largely on the studies ofVaccaneo (1934), who worked with material collected in East Africa.Kuijt ( 1969) reviews the biology of the genus.Visser discusses the biology of H. africana (Visser 1981) and H. triceps (Visser 1988) while Musselman (1984) and Musselman and Visser (1987) treat H. johannis.

MATERIALS AND METHODS
Hydnora was studied in the field in Sudan, Zimbabwe, Botswana, South Africa, and Namibia.Specimens from the following herbaria were examined: BM, BR, C, E, K, L, USC, M, MO, P, PRE, RO, SRGH, UPS, and WIND.Work towards a monograph of this genus has been severely hindered by insufficient and poorly preserved herbarium material.Very few of the type specimens could be located; several were at Berlin and destroyed during the Second World War.Herbarium specimens are usually of poor quality and fragmentary, since the plants are very fleshy and often decay while drying.After drying, the material is very brittle.We estimate that there are fewer than 100 usable Hydnora specimens in the world's major herbaria.

Distribution
Hydnora is largely an African genus.Two species, H. africana and H. johannis, are widespread.Distribution data, however, are very spotty (Fig. 1).

Description
Hydnoraceae are the only angiosperm family entirely devoid of leaflike structures, lacking even scales (Kuijt 1969).The following description of the genus follows Musselman (1984) except that "root" is used for the underground portion of the plant in the sense of Visser (1981) rather then "rhizome" (Musselman 1984).
Subterranean holoparasitic herbs with often massive root systems spreading laterally from the host.Roots up to 1 dm wide, 4-5 angled, terete or sometimes flattened.Periderm well developed, brick-red except at the tip of the root.Fresh roots pink to flesh-red with sticky exudate, extremely bitter and astringent.Entire root (except for the tip) covered with warty outgrowths ofhaustoreogenic capacity which may be regularly distributed and up to 0.8 em long or irregularly distributed and less than 0.5 em long.Latent and active buds scattered along the roots, usually in clusters of 2-4.Buds tubular with valvate lobes.Flowers 3-, 4-, or 5-merous.Perianth lobes patent and resting on the soil, or lobes not reflexed and flower opening by a separation of the lobes.Flowers variable in size from 5 to 25 em, the length depending on the depth of the root, pedicel 4-9 em.Ovary inferior, unilocular, with numerous infolded, pendant placentae.Stigmas sessile and with distinct grooves on the surface.Stamens basifixed, anthers 2.5-3 x 2-2.5 em.Pollen very sticky, adhering to anthers.Perianth lobes 6-8 em; in some species, "bait bodies" are present between the inner margins ofthe lobes; in other species a well-developed osmorphoric region ("cucullus") is present.Fruit entirely subterranean, fleshy, globose, 10-15 em wide, many-seeded, often splitting irregularly after maturation.Fruiting pedicel very short and easily separated from the root.Pericarp with scaly outer periderm; the inner layer mealy, white, very sweet.Fruiting placentae similar to inner layer of pericarp in taste and texture.Seeds brown, very hard, irregularly shaped, oblong to globose, 1-1.8 mm.Seedlings unknown.
Representatives of the genus are illustrated in Figure 2, 3, and 4.

Hosts
Hydnora species have a narrow host range.Hydnora africana and H. triceps are restricted entirely to Euphorbia species, with one report of Zygophyllum as a host (Table 1), whereas other species of Hydnora parasitize only species of Acacia and related trees.A report of H. johannis on Kigelia africana (Lam.)Benth.has been shown to be erroneous.The hosts of Hydnora are given in Table 1.
Host-parasite relationships have received very little attention.Since the parasite has no obvious transpiration, how are materials moved from the host to the parasite?In times of water stress, can water move from the parasite to the host?The answer to these and other questions requires additional research.

Taxonomy
Decaisne (1873) first suggested that subgeneric classification be based on the number of floral parts.In his scheme, subgenus Dorhyna contained plants with Fig. 4. Hydnora triceps.Rower and root; host root at left.(From Visser [1988]).(Three-fourths life size.)four-parted perianth parts, stigma lobes, and stamens.The subgenus Hydnora, on the other hand, contained species with trimerous perianth parts, stigma lobes, and stamens.We have found, however, that individuals in any population of H. africana can have four stigma lobes, although this species usually has three stigma lobes.Later, Vaccaneo (1934) amended Decaisne's subgeneric classification to include features of the roots.In this classification, Vaccaneo's subgenus Hydnora included species with angled roots and large protuberances and subgenus Dorhyna those plants with more or less terete roots and smaller protuberances.Harms (1935) established four sections.One of these, sect.Tricephalohydnum Harms, was monotypic and included the poorly known H. triceps, which we tentatively place in the now sectionless subg.Hydnora due to its 3-merous flowers, the presence of"bait bodies," and its angled warty roots.Our preliminary taxonomic treatment of the group follows.Listed species follow Harms (1935)  Hydnora angolensis, known only from a fruit presumably collected in Gabon, was placed by Decaisne (1873) in subgenus Hydnora.We know of no other collections of Hydnora from Gabon, so the status of this species is unclear.Decaisne ( 1873) described H. aethiopica based on a specimen from Sabatier's expedition to find the source of the White Nile.This name also appears to be based on an incomplete specimen.If it represented subg.Hydnora, it would provide the basis for a significant range extension for this essentially southern African group.
Lastly, there is a noteworthy collection from the Dhofar region of Oman, Miller 7619 (E), that has been determined as H. africana, a species known only from the southernmost part of Africa; a collection from Arabia would represent a considerable disjunction.The specimen resembles H. africana but the label data indicates that it was growing beneath Acacia.Although no reference to parasitism is made, this Hydnora may have been parasitic on Acacia.

Uses
Human uses.-Thefruits of H. johannis are edible, even delectable (Vaccaneo 1934;Musselman 1984;Musselman and Visser, 1987).Hydnora africana fruits are also sought after as food (Visser 1981).Less well known is the medicinal use of the roots (Vaccaneo 1934;Musselman 1984).These have been used as an antidiarrheal agent, probably due to the high concentration of tannin, imparting a strong astringency to the dried roots.The high tannin content also renders the roots useful for tanning leather (Musselman and Visser 1987).In Sudan, the dried roots are used like charcoal (Musselman 1984).Parker (1988) reports that in the Dire Dawa region of Ethiopia, H. johannis damages asphalt streets by breaking through the surface to flower.
Animal uses.-Thefruits of Hydnora johannis are eagerly sought and eaten by monkeys (Musselman 1984), whereas those of H. africana are eaten by various other mammals (Visser 1981 ).According to one herbarium label, Gilbert & Thulin 1534 (UPS), flower buds of H. johannis are eaten.We assume that animals are implied here, for goats are known to eat the buds of H. johannis in Sudan.Because the roots contain large quantities of water, they are often excavated and eaten by animals during the dry season.In the Namib Desert, evidence of foraging for H. africana by small mammals is often found.We have observed extensive excavations of H. johannis by elephants in Etosha Pan National Park (Musselman and Visser 1987).Rhinoceroses are reported to forage for H. johannis, according to the herbarium label for Bally 7694 (K).
Culture Kuijt (1969) wrote, regarding the Hydnoraceae: "A fascinating story awaits the botanist who is fortunate enough to have access to viable seeds."Seeds are abundantly produced by Hydnora but we know of only one person who has grown these plants in cultivation.We are indebted to this "fortunate botanist," Sherwin Carlquist, for sharing his "fascinating story." Seeds of H. africana were collected in 1973 from plants growing on Euphorbia mauritanica L. at Worcester, South Africa.Carlquist planted the seeds in a pot of Euphorbia caputmedusae L. in Claremont, California.For four years the Euphorbia was repotted as it grew, the Hydnora parasite not yet evident.The plant was then transplanted to Carlquist's garden where, three years later, a flower of Hydnora appeared!Other Euphorbia species were added and they, too, were parasitized.No fruits have been produced, however.

Floral Biology
The structure of the Hydnora flower is simple.The perianth lobes, and sometimes part of the tube, are the only portions ofthe plant emerging above the soil.References to the fetid smell of the flower are numerous, but little information on the floral biology and no information on the breeding system of the parasite are available.
The flowers of H. africana and H. triceps are apparently unique in possessing "bait bodies" ("Koderkorpen"), first described by Harms (1935) for H. africana.These are strong-smelling particles on the perianth lobe margins (Fig. 2).These bait bodies attract beetles, often dermestids, which enter the flower and harvest the pollen.Carrion flies have also been collected in the flowers.Since the stigma is below the anthers, pollen falls onto it.The beetles sometimes crawl onto the stigmatic surface, which is receptive a few days before anthesis.Because beetles bearing pollen have been trapped before anthesis, H. africana may be an outcrossing species.This hypothesis is also supported by the fact that flies are sometimes found inside the flower buds.These observations on floral visitors are our own.
The floral biology of H. johannis is different.Bait bodies are not present but there is a well-developed osmophore ("cucullus," sensu Vaccaneo 1934) at the tip of each perianth lobe (Fig. 1 ).Hydnora johannis is pollinated by scarab beetles, rather than dermestid beetles.
Blow flies, as well as dermestid beetles, have been found in the flowers of H. triceps.The flowers of this species seldom emerge from the soil; insects enter them through cracks in the soil, cracks that form as the flowers grow (Visser 1988).

Fig. I .
Fig. I. Distribution of Hydnora species based on herbarium specimens, literature, and field work.Ranges are approximate.

Table I .
Recorded hosts for Hydnora.