A taxonomic revision of the genus Hydnora (Hydnoraceae)

Summary

A systematic monograph is presented for Hydnora (Hydnoraceae), a poorly known genus of holoparasitic plants distributed across Africa, Madagascar and southern Arabia. Species of Hydnora are characterised by their underground habit, unusual fleshy flowers and complete absence of leaves or photosynthetic tissue. This is the first detailed monograph of the genus Hydnora since 1935 and is informed by a comprehensive survey of herbarium specimens and literature. Detailed descriptions, full synonymy, distribution maps and discussion concerning confusable taxa are provided for each species, along with notes on ethnobotany, ecology and conservation. We place particular emphasis on the taxonomic value of osmophore geometry and positioning in living and dried material, which are highly consistent within species. We also provide the first detailed assessment of host range across the genus. Hydnora hanningtonii Rendle and H. solmsiana Dinter are reinstated from synonymy, and H. bolinii S.Hatt is newly described here. The infrageneric classification is reviewed and a key is provided for both living and dried material. Species are accompanied by both illustrations and photographs of living and dried material where possible.

Introduction

Hydnora Thunb. is a genus of parasitic plants distributed across much of southern and eastern Africa, Madagascar, and southern Arabia (Map 1). Together with the American sister genus Prosopanche de Bary, Hydnora is holoparasitic, meaning that it is entirely dependent on a host plant for all of its carbon and nutrient intake. The two genera are placed in the family Hydnoraceae, which is nestled within the Piperales (Jost et al. 2021). Following Cassytha L. (Lauraceae), Hydnoraceae are the second earliest diverging lineage of extant parasitic plants (Naumann et al. 2013; Nickrent 2020).

Map 1.

Distribution of all known species of Hydnora, compiled from herbarium specimens, iNaturalist records, literature references and personal observations.

The genus is unusual amongst flowering plant diversity in that all plants are completely devoid of photosynthetic vegetative parts, and because of their extreme morphological divergence, the plants are sometimes mistaken for a fungus at first glance (Bolin et al. 2011). Primarily a large subterranean rhizome (Tennakoon et al. 2007), Hydnora only breaches the surface of the soil to flower, in some cases for just a few weeks following the rains (Bolin et al. 2018). Their bizarre, fleshy flowers attract pollinating beetles with a potent, foetid odour, trapping them within their hollow reproductive chamber (Seymour et al. 2009). The beetles can escape when the flowers wither a few days later, and pollination follows (Bolin et al. 2009a). The ovary then develops into a large swollen fruit containing thousands of tiny black seeds. The fruit is sought after by jackals and other smaller mammals, hence its common name in South Africa: jakkalskos (Bolin et al. 2011).

Currently represented by eight species, the genus has suffered prolonged neglect and taxonomic confusion since it was described in 1775 by Carl Thunberg (Musselman & Visser 1989; Bolin et al. 2018). A combination of a largely hypogeous habit and remote distribution, often in politically unstable regions, has meant that Hydnora has remained elusive to plant collectors and taxonomists (Bolin 2009). Furthermore, a considerable number of the specimens stored in herbaria are in relatively poor condition; many are shattered, deformed fragments of little or no taxonomic value. These factors partly explain the vague and incomplete descriptions and consequent synonymy that is characteristic of much of the research into Hydnora in the 19^th and early 20^th centuries. The work of Lytton Musselman and Jay Bolin over the last few decades has significantly helped to clarify some of this confusion, although considerable uncertainty remains (Musselman & Visser 1989; Bolin et al. 2018). The lack of a published widely sampled phylogeny has also greatly hindered our understanding of the evolutionary relationships among species.

Host preference varies considerably across the genus (Musselman & Visser 1989). Subgenus Hydnora (syn. subgenus Euhydnora Decne., nom. inval. [Art. 21.3 of the ICN]) in southern Africa exclusively parasitise succulent Euphorbia L. Within this group, H. visseri Bolin, E.Maass & Musselman, H. longicollis Welw. and H. triceps Drège & E.Mey. are restricted to one, two or a few mutually exclusive host species, while H. africana Thunb. has at least 12 recorded hosts (see Host Specificity and Speciation). The remaining species of Hydnora parasitise Fabaceae (primarily Vachellia Wight & Arn. and Senegalia Raf. species although other genera have been recorded), with the exception of H. sinandevu Beentje & Q.Luke which parasitises Commiphora Jacq. (Burseraceae) hosts. What little information exists about host preference is scattered across herbarium specimen label data, observational notes, and floristic accounts. However, these data are often unreliable as collectors rarely excavate Hydnora to confirm host-parasite contact. The complex pattern of host preference exhibited across the genus indicates potential co-evolution between host and parasite; like many aspects of the ecology and evolution of Hydnora, this remains unexamined (Bolin et al. 2011).

Distributional data are limited to the relatively small number of herbarium specimens (Teixeira-Costa et al. 2022), a few scattered observations in floristic accounts and more recently the small, but steadily growing, collection of observations recorded by digital means such as iNaturalist (https://www.inaturalist.org). Consequently, conservation assessments are scarce and often unreliable given the paucity of data to inform them (Dougherty et al. 2015). Sparse collecting, a lack of records, and poor preservation of specimens together point to significant gaps in our knowledge of the distribution of Hydnora. For example, a population of H. abyssinica A.Braun was described from Nigeria just five years ago, more than 1,500 miles from the nearest other collection of the same species (Agyeno et al. 2018).

This monograph is the first comprehensive revision of the genus Hydnora since 1935 (Vaccaneo 1934; Harms 1935). In 1989, Musselman & Visser estimated that there were fewer than 100 meaningful specimens of Hydnora in the world’s herbaria. Now, at least 443 specimens are known to the authors, most of which were examined for this monograph. This study includes a detailed list of recorded host species, together with detailed descriptions, accompanying illustrations, distribution maps and all available ecological and ethnobotanical data for each species. These are complemented by a key for living and dried specimens. Hydnora solmsiana Dinter and H. hanningtonii Rendle are reinstated from synonymy, and H. bolinii S.Hatt is newly described here.

Taxonomic History

The taxonomic history of Hydnora begins with the collection of H. africana Thunb. in South Africa by the Swedish botanist Carl Thunberg (1775), who published it as an extraordinary new species of fungus (see illustration in Fig. 1). A year later, Erik Acharius, student of Carl Linnaeus, asserted the plant genus, Aphyteia, with a single species Aphyteia hydnora [Hydnora africana] (Acharius & von Linné 1776). In the decades that followed, Aphyteia and Hydnora were used somewhat interchangeably, until the early 1800s, by which time Hydnora had become the dominant name in use. According to International Code of Nomenclature for algae, fungi, and plants, the initial name, Hydnora, is valid even though it was described as a fungus, as the Code encompasses both plants and fungal nomenclature (Preamble 8, Turland et al. 2018) Indeed, it appears Thunberg was unsure whether it should be described as a plant or a fungus, as evidenced in his report to the Royal Academy of Science in Stockholm in 1775: ‘but of all that I have so far had the opportunity to see and discover, nothing has seemed to me more strange than the fungus whose description I have the honour of submitting. So strange is its composition that many would certainly doubt the existence of such a plant on the face of the earth’ (Svedelius 1944). In the early 1800s, a small number of collections were made and deposited in European herbaria by ‘entrepreneurial collectors’ Christian Ecklon, Karl Zeyher and Johann Franz Drège (Arnold 2021). One of these collections by Drège precipitated the description of a second species, H. triceps Drège & E.Mey, by Ernesto Meyer (1833), who published a small but relatively detailed report on the two accepted species.

Fig. 1.

Early illustrations of Hydnora species. A, B Hydnora africana from Thunberg (1775); C H. longicollis from Welwitsch (1869); D H. hanningtonii from Rendle (1896).

Debate ensued over the affinities of Hydnora, with British botanist Robert Brown believing it to be closely related to Rafflesia R.Br. ex Gray (Rafflesiaceae) due to its shared trait of parasitism, while others disagreed, such as Brown’s student William Griffith (Brown 1844; Arnold 2021). Over the remainder of the 19^th century, several new species of Hydnora were published by European botanists including Welwitsch (1869; Fig. 1), Beccari (1871), Decaisne (1873) and Rendle (1896, Fig. 1). Decaisne made the first attempt to divide Hydnora into subgenera (Decaisne 1873). He grouped those with four tepals into subgenus Dorhyna Decne. and those with three into subgenus Euhydnora (now subgenus Hydnora). Many of the species during this period were described in isolation and authors rarely observed other specimens in person (Arnold 2021). As a result, many of the descriptions share insignificant differences and the species have since been synonymised (Musselman & Visser 1989).

The turn of the century saw an increase in floristic work, with floras by Dinter (1909), Thistleton-Dyer (1912) and Marloth (1913) consolidating Hydnora species in southern Africa. New species continued to be described until Vaccaneo’s exhaustive monograph (1934). In addition to describing a new species, H. cornii Vaccaneo, he presented detailed descriptions and ecological data. He also suggested that many species described thus far did not hold up to scrutiny. He retained Decaisne’s two subgenera, although altered the criteria such that they were based on rhizome characters rather than the variable and unreliable trait of floral merosity. Hermann Harms published another monograph a year later (1935), largely based on the foundations laid by Vaccaneo. Most notably, he extended the subgeneric classification to include subgenus Neohydnora Harms (H. esculenta Jum. & H.Perrier), and subgenus Tricephalohydnum Harms (H. triceps).

Only a few publications on Hydnora emerged in the decades that followed. This hiatus was lifted by a series of publications by Lytton Musselman and Johann Visser, partly triggered by the instigation of various floristic projects across Africa in the 1980s (Visser 1981; Musselman & Visser 1987). This culminated in a report in 1989 on the natural history and updated taxonomy of the genus (Musselman & Visser 1989). Accounts that followed this paper, mostly published by Musselman, synonymised 11 species described from East Africa under H. johannis Becc. (Musselman 1993, 1997, 2001). Beentje & Luke (2002) determined that H. abyssinica A.Braun was in fact the valid name for this species, although many herbarium specimens are still labelled as H. johannis. In the years since, Musselman’s student Jay Bolin has described two new species, H. visseri and H. arabica Bolin & Musselman (Bolin et al. 2011, 2018). In his PhD thesis, he produced a revision of subgenus Euhydnora (now subgenus Hydnora) in southern Africa, and the first molecular phylogeny for the genus, including seven species of Hydnora using rpoB and ITS regions (Bolin 2009). As it stands, there are eight accepted species of Hydnora, grouped into four subgenera, and a comprehensive monograph of the entire genus has not been updated for almost a century (Harms 1935; Bolin et al. 2018).

Materials & Methods

This monograph has been compiled through the examination of herbarium specimens, literature, recorded observations and fieldwork. Specimens of Hydnora were observed by the authors at BM, BOL, EA, GHPG, GRA, K, NBG, OBG, ODU, SAM, TAN and WIND. All specimens from other herbaria (AAU, B, BR, E, FT, G, HARG, HBG, HUH, JUHN, L, LD, LISU, M, MO, P, PRE, PSUB, RO, S, SRGH, UPS, US) were examined digitally. Species descriptions were updated with new and relevant data from herbarium specimens and the literature.

Species Concepts. The species concept used follows the ‘morphological cluster’ concept outlined in Vorontsova & Knapp (2016): “assemblages of individuals with morphological features in common and separate from other such assemblages by correlated morphological discontinuities in a number of features” (Davis & Heywood 1963). Although the taxonomic decisions made here are based on clear and consistent morphological characters, DNA sequence data would provide a higher level of resolution. As such, the taxonomic decisions made here will be corroborated with molecular data in the future.

Host Specificity. A list of potential host species for each Hydnora species was compiled using herbarium specimen label data and literature (Table 1). Each entry is accompanied by a reference and a confidence rating, with a brief justification. These are ranked from ‘Low’, ‘Medium’ or ‘High’ based on the number and quality of the records in the context of all known records for the species. Host species are marked ‘High’ if they have been reported numerous times, or observed in situ by the authors; ‘Medium’ if there is a single or two reports, with the species in the expected family/genus for that Hydnora species (e.g., Euphorbia for subgenus Hydnora), or ‘Low’ if the host is not from the expected family/genus for that Hydnora species.

Table 1. List of recorded host species for Hydnora. Each entry is accompanied by a reference and a confidence measure with a brief justification. These are ranked either ‘Low’, ‘Medium’ or ‘High’ (see Materials and Methods). They are subjective measures and should be taken as such, but will hopefully provide some indication on how valid the host record is likely to be.

Distribution Maps. Distribution maps were assembled by combining all available data from herbarium specimens, literature and iNaturalist observations (iNaturalist 2022). Most herbarium specimens lacked co-ordinates, and were therefore georeferenced where possible using Google Earth and collecting locality indices such as Polhill 1988. Point data were input into ArcGIS Pro 2.8, from which distributional maps were produced (ESRI 2021). Collecting localities written in the ‘specimens examined’ sections appear as written on herbarium labels. Any spelling corrections of historic localities are put in square brackets. For specimens from Kenya, Tanzania and Uganda, the geographical system used by Flora of Tropical East Africa (Polhill 1988) is used (e.g., K1, T1, U1).

Figures. Illustrations and line drawings were prepared by the first author for all species, with the exception of an illustration for Hydnora bolinii, which lacked sufficient photographic material to produce a reliable illustration in this way and therefore specimen photographs and line drawings are supplied instead. The illustrations were prepared from collated reference material of photographs of living material and herbarium specimens.

Morphology

Hydnora possess unique morphology, such that certain characters are impossible to describe succinctly using conventional botanical terminology. This confusion is reflected in former descriptions of Hydnora, where the same words have been employed to describe different features. For example, the term ‘bait body’ is sometimes used to refer to the apical osmophore, and at other times the long setae that line the tepal margin in H. africana (Musselman & Visser 1989; Bolin 2009). To overcome this obstacle, existing terms (Beentje 2020) have been adapted here and outlined in an illustrated glossary (Fig. 2), with efforts stay as coherent with previous descriptions of Hydnora as possible. To ensure clarity, the terms defined here are deployed consistently throughout the monograph.

Fig. 2.

Illustrated glossary of terms used in this monograph to apply to characters on Hydnora tepals. Each word is accompanied by an illustrated example of how the term is applied. The bottom-right box illustrates the two floral axes referred to in the text.

Hydnora have fewer taxonomically valuable morphological characters than most autotrophic plants, thus putting added value on ecological traits and molecular data for identification. This is further complicated by the loss of certain key characters only available in the field, such as colour, smell, and gross morphology. Even within a single population, flowers can vary in morphology and colour; partly due to differing growing conditions or the differing stages of transformation that the flower undergoes during anthesis. However, there are a set of key characters outlined below that can be reliably used to distinguish dried specimens of almost any age, assuming both the flower and rhizome have been collected.

Rhizome. Gross morphology and surface texture of the rhizome is inherently variable depending on environmental conditions (Fig. 3A – E). However, the positioning of the tubercles along the rhizome surface has significant taxonomic value (Fig. 3H, J). Either the tubercles are randomly distributed across the surface, with the rhizome therefore generally terete or subterete in cross-section, or they are arranged in distinct parallel lines and thus the rhizome is angular in cross-section (Fig. 3F, G). This trait reliably separates East African subgenera Dorhyna and Sineseta from the southern African subgenus Hydnora and the Madagascan subgenus Neohydnora. Beyond this trait, the rhizome has not been found to be of taxonomic value. The colour of the internal rhizome flesh can vary from pink to red to black depending on age at the time of cross-section.

Fig. 3.

Rhizome morphology of Hydnora. A Hydnora africana: rhizome and associated host root; B H. africana: flesh of rhizome detached from outer skin; C, E H. solmsiana: fragments of dried rhizomes; D H. solmsiana: surface of dried rhizome; F H. africana: rhizome cross-section; G H. africana: rhizome side view; H angular rhizome type (subgenus Hydnora and Neohydnora); J terete rhizome type (subgenus Dorhyna and Sineseta). H – J drawn by sebastian hatt.

Flower. Gross floral anatomy is conserved across the genus (Fig. 4A – G). Floral merosity has been used in the past as a subgenus-defining character. Although it is no longer considered a reliable character due to natural phenotypic fluctuation in this trait, most individuals in subgenus Hydnora have three tepals, most in subgenera Dorhyna and Sineseta have four and most in subgenus Neohydnora have three to five. While it can be a valuable character for identification, it cannot be used in isolation. Fruit and seed morphology have not been found to vary considerably between species (Fig. 4J – N).

Fig. 4.

Flower morphology of Hydnora. A Hydnora visseri flower cross-section; B H. triceps flower cross-section; C H. esculenta flower cross-section; D H. solmsiana flower cross-section; E diagram of general floral structure (H. hanningtonii in this example); F androecial chamber cross-section (H. visseri); G gynoecial chamber cross-section (H. visseri); H four stages of decay of H. visseri flowers, fresh (left) to dry (right); J fruit (H. triceps); K, M fruit cross-section (H. triceps); L germinating seeds (H. triceps); N diagram of general fruit structure. E & N drawn by sebastian hatt.

Osmophore. ‘Osmophore’ is the term used to describe the spongy, pale coloured tissue that generates the foetid odour emitted by the flowers (Fig. 5A – F). In dried specimens, they usually appear as a distinctly paler region (e.g., Fig. 5D), though are sometimes distinctly blackened by poor drying conditions. There are two types of osmophore: the apical osmophore, positioned apical to the tepal margin (Fig. 5B), and the recessed osmophore, positioned within the tepal cavity (Fig. 5A). We found that the shape and positioning of these osmophores on the tepals is highly consistent within species and therefore has taxonomic value (Fig. 5G – P). Species in subgenus Hydnora possess just one, large recessed osmophore (Fig. 5G), while those in subgenus Dorhyna have both an apical and recessed osmophore (Fig. 5K – N), though the former is most conspicuous. Subgenus Neohydnora (H. esculenta) only has an apical osmophore (Fig. 5J), while subgenus Sineseta (H. sinandevu) only has a recessed osmophore (Fig. 5P).

Fig. 5.

Tepal morphology of Hydnora. A Hydnora africana tepal showing recessed osmophore; B H. esculenta tepal showing apical osmophore; C H. abyssinica tepal showing apical osmophore; D H. hanningtonii dried tepal showing apical osmophore (Thesiger s.n. [BM]); E H. solmsiana dried tepal showing apical osmophore (Bolin 09-06 [WIND]); F H. bolinii dried tepal showing apical and recessed osmophore (Gillett 12781 [K]); G – P osmophore positioning for all species; Q H. visseri tepal setae; R H. esculenta tepal setae; S H. solmsiana tepal setae; T H. hanningtonii tepal setae. G – P drawn by sebastian hatt.

Setae. Setae coverage on the tepal margin does have some taxonomic value despite also being a highly variable trait with size, density and colour changing considerably over the lifespan of the flower (Fig. 5Q – T). H. sinandevu is unusual in being entirely glabrous. H. hanningtonii has dense, strigose setae that cover the entire tepal margin (Fig. 5T) while H. abyssinica and H. solmsiana have setae of variable length and density that rarely reach the ventral edge (Fig. 5S). Similarly, H. visseri has distinctly shorter and more sparsely distributed setae (Fig. 5Q) than H. africana. Setae can be dimorphic in some species, H. africana, H. visseri and H. longicollis, with an additional type of thicker, longer seta found only on the tepal edge. H. esculenta is unique in having fleshy ribs in place of setae (Fig. 5R).

Live material. Tepal colour is not preserved in dried specimens, and is quite variable in living material (Fig. 4H). However, H. abyssinica is consistently white inside the floral tube, while H. hanningtonii is consistently bright red. Different species are reported to vary in the smell emitted by the flower and the taste of the fruit. While these characters may be useful for field identification and should be recorded in label data, they are variable and subjective and cannot be used in isolation.

Beyond these characters, little else has been found to be of taxonomic value in dried or living specimens. The dimensions of the flower appear to be highly variable, as well as the relative distance between the components of the reproductive chamber and the tepals.

Distribution

Hydnora is widespread across southern and eastern Africa, southern Madagascar and the southern Arabian Peninsula (Map 1). The genus has a preference for arid to semi-arid climates, and has been reported from Sudan, Eritrea, Somalia, Ethiopia, Rwanda, Uganda, Kenya, Tanzania, the Democratic Republic of the Congo, Mozambique, Zimbabwe, Angola, Namibia, Botswana, Eswatini and South Africa. Recently, a population resembling H. abyssinica was reported from the semi-arid Jos plateau in Nigeria, ~1,500 km from the nearest known population recorded hitherto (Agyeno et al. 2018). The broad distribution of Hydnora, particularly that of the most widespread species H. abyssinica and H. solmsiana, indicate the tolerance of a range of climatic conditions. For example, H. esculenta has been recorded growing in both sub-arid scrubland and transitional rainforest areas (Bolin & Musselman 2013). Distribution may be more restricted by host range. For example, H. triceps has a relatively narrow range in South Africa that follows the narrow range of its host, Euphorbia dregeana E.Mey. ex Boiss. In contrast, H. abyssinica has a vast range and it can parasitise a number of different Fabaceae hosts with broad ranges across Africa. In addition, it has also been shown to be able to parasitise introduced species such as Pithecellobium dulce (Roxb.) Benth. and Tamarindus indica L. that may, in theory, have triggered a range expansion. However, very little is known about the climatic requirements of Hydnora and, certainly in the case of H. abyssinica, many of its hosts can grow in regions that Hydnora clearly cannot. The recalcitrance of Hydnora to cultivation ex-situ may also suggest that it does have specific requirements, perhaps edaphic as well as climatic, that are not yet appreciated (Thorogood et al. 2022).

The range of Hydnora may be underestimated. This is plausible given the dearth in collection in many of these regions, hindered also by a hypogeous habit and infrequent flowering (Sosef et al. 2017). For example, the distribution of H. sinandevu was previously thought to be restricted to Kenya and Tanzania. However, our research has revealed collections of H. sinandevu in Saudi Arabia, Somalia and Ethiopia.

Host Specificity

Our understanding of host specificity is poor for holoparasitic plants, because collectors rarely excavate to confirm the point of host attachment. For this reason, many host records are unreliable. For example, Hydnora solmsiana was observed by the author growing many metres from its verified host; despite flowering next to a different species of potential host.

Host range across the genus is summarised in Table 1. Species in the southern African subgenus Hydnora appear to be restricted to succulent Euphorbia (Euphorbiaceae) hosts. Interestingly, the four Hydnora species in this subgenus have mutually exclusive hosts, even in sympatry (Bolin et al. 2011). While H. visseri, H. longicollis and H. triceps appear very restricted to either one, two or three host species, H. africana has been recorded to parasitise at least 12 species. These 12 known hosts of H. africana are found in different subgenera and sections of Euphorbia, rather than being closely related as might be expected given the level of specificity seen in H. visseri, H. longicollis and H. triceps. This raises the question of why these specific hosts are preferred, but not other more closely related Euphorbia species native to the same locality. Understanding this pattern of host preference in subgenus Hydnora requires knowledge of the mechanisms of host preference and rejection in these species that we currently lack. Notably, in the only germination study of Hydnora, H. triceps only germinated in response to cut root exudates of its host and not for other Euphorbia species (Bolin et al. 2009b).

Species of subgenus Dorhyna parasitise trees in the Fabaceae. The widespread Hydnora abyssinica has 16 recorded hosts, H. solmsiana has 14 recorded hosts and H. hanningtonii has three recorded hosts, with most of these belonging to Vachellia and Senegalia (both ex Acacia). In subgenus Neohydnora, H. esculenta parasites at least five species of Fabaceae from four different genera. H. sinandevu is unusual in parasitising exclusively species of Commiphora (Burseraceae).

According to the most recent phylogeny for the genus (Jost et al. 2022), a shift has occurred from the parasitism of CAM-photosynthesising Euphorbia (Malpighiales) to C3-photosynthesising Fabaceae (Fabales). It is unknown whether there is any significance for the parasite with regards to the difference in photosynthetic metabolism between these two groups. The identity of the ancestral host family remains uncertain, although the phylogeny suggests that the ancestral Hydnora parasitised Fabaceae (Bolin 2009; Jost et al. 2022). Therefore, a transition must presumably also have been made from Fabaceae to Commiphora (Burseraceae) in H. sinandevu. The fact that these family-level transitions appear to correlate with genetically defined lineages of Hydnora may suggest that host specificity has contributed to genetic divergence in the genus, as has been evidenced in the endoparasitic Apodanthaceae (Bellot & Renner 2014).

Interestingly, some Hydnora appear to be able to parasitise certain non-native hosts that have only been introduced to the native region in the last few centuries. There are at least two incidences of H. abyssinica parasitising introduced Fabaceae hosts, namely Albizia lebbeck (L.) Benth., Delonix regia (Bojer ex Hook.) Raf. and Tamarindus indica L. (Bosser 1994; Agyeno et al. 2018). Furthermore, both H. hanningtonii and H. esculenta have been recorded on the non-native Pithecellobium dulce (Roxb.) Benth. Indeed, Bolin & Musselman (2013) report that this introduced species was the most common host encountered for H. esculenta in Madagascar. Interestingly, increased susceptibility of introduced species to parasitic plants has also been reported in mistletoes (Kołodziejek et al. 2013; Díaz-Limón et al. 2016; Ma et al. 2020).

Infrageneric Classification

We follow the subgenera of Decaisne (1873), added to by Harms (1935), but we make modifications to accommodate new taxa, in line with the existing phylogenetic data for the genus (Bolin et al. 2009b; Jost et al. 2022). The revised subgeneric classification is presented summarised in Table 2. Subgenus Euhydnora is corrected here to subgenus Hydnora; an autonym generated in line with nomenclatural code requirements that the subdivision of a genus that includes the type of the generic name must repeat the generic name unaltered as its subdivisional epithet (Art. 21.3, 22.1, 22.2, Turland et al. 2018). Hydnora triceps was previously positioned in its own subgenus, subg. Tricephalohydnum by Harms (1935) due to its uniquely fused tepals. Subgenus Tricephalohydnum is included here in subgenus Hydnora due to its shared characters of recessed osmophores and angular rhizomes with tubercles in distinct parallel lines, as previously suggested by Musselman & Visser (1989).

Table 2. Revised subgeneric classification for Hydnora. A summary of characters is provided for each subgenus.

Subgenus Dorhyna previously contained Hydnora abyssinica and H. hanningtonii (as H. arabica). Here H. solmsiana and H. bolinii are assigned to subgenus Dorhyna based on their shared rhizome morphology, Fabaceae hosts and positioning of the osmophores. Subgenus Neohydnora remains in place and contains only H. esculenta. On the basis that H. sinandevu lacks both setae and an apical osmophore, and parasitises Commiphora (Burseraceae) rather than Fabaceae, thus conflicting considerably with the criteria for subgenus Dorhyna, it is moved from subgenus Dorhyna to its own new subgenus Sineseta (translating to ‘without setae’ in Latin). A widely sampled molecular phylogeny will be required to confirm these taxonomic decisions, although they already comply with existing phylogenetic trees for the genus (Bolin et al. 2009b; Jost et al. 2022).

Systematic Treatment

Hydnora Thun. (Thunberg 1775: 69). Type: Hydnora africana Thunberg (1775: 69).

Aphyteia Ach. & L. (Acharius & von Linné 1776: 7). Type: Aphyteia hydnora Ach. (Acharius & von Linné 1776: 10).

Herbaceous subterranean perennial root holoparasite, without leaves or stem. Rhizome terete, subterete or angular; rhizome surface coriaceous, dark brown, lighter coloured (when fresh) near growth tip; rhizome spreads laterally and often bi- or trifurcating or branching irregularly to form smaller terminal branches; rhizome ornamented with numerous lateral tubercles of variable size and shape, randomly distributed or arranged in parallel lines on the surface; tubercles remain the same or develop into a flower or develop into a haustorium; numerous flowers and flower buds on single rhizomes; rhizome fleshy, brick red to reddish-pink internal tissue when broken. Flower 3 – 5-merous; flower emerges only partially from soil or not at all; two floral chambers, androecial chamber subtended by gynoecial chamber, inner surfaces of chambers glabrous, pedicel absent or present up to 2.5 cm long. Perianth external tissues coriaceous, dark brown to reddish-brown; perianth tissues fleshy; perianth tube red, pink or white inside, darkening to red-brown or black when dried; tepal margins present with setae present or absent; tepal margins red to white to orange, lanceolate to elliptic, fused or not fused. Osmophore positioned either on tepal apices or recessed in tepal cavity or both; spongy, cream to white (in life), darkening to tan or black when dried, generating foetid odour. Androecium antheral ring formed by connate inverted V-shaped anther lobes, forming a central orifice, transversely striate and divided into numerous horizontal pollen sacs. Gynoecium ovary inferior, unilocular with numerous ovules produced from apical placenta; lobed and cushion-like stigma on the floor of the gynoecial chamber. Fruit subterranean, globose, leathery brown pericarp. Seed spherical, black-brown, thousands of seeds embedded within mealy white pulp.

distribution. Ten species distributed across the arid and semi-arid regions of southern and eastern Africa, southern Madagascar, Nigeria and the southern Arabian Peninsula. Map 1.

Key to species of Hydnora

• 1. Rhizome angular, usually 4 – 7-angled, with tubercles in distinct parallel lines …………….2

• 1. Rhizome terete, subterete or laterally compressed, with tubercles randomly distributed……6

• 2. Osmophores only apical, not recessed; tepal margins white, ribbed, without setae; flowers often 5-merous, (Madagascar and Mascarene Islands)...…………....…5. H. esculenta

• 2. Osmophores not apical, only recessed; tepal margins red, orange or pink, not ribbed, with setae; flowers usually 3-merous, (mainland Southern Africa)…………………………3

• 3. Tepals entirely fused for apical half of length thus forming 3 isolated funnel-shaped apertures; flower entirely hypogeous; parasitising Euphorbia dregeana…4. H. triceps

• 3. Tepals not fused at apex or only transiently; flower not or only partially hypogeous, parasitising Euphorbia spp. other than E. dregeana……………………………………4

• 4. Tepals 0.9 – 2.4 cm long; tepal margins often orange; less than ${}^{1}\!\left/ {}_{5}\right.$ flower above ground; parasitising Euphorbia damarana (Angola, northern Namibia)………3. H. longicollis

• 4. Tepals 2.5 – 9 cm long; tepal margins often pink to red; more than ${}^{1}\!\left/ {}_{5}\right.$ flower above ground; parasitising Euphorbia spp. other than E. damarana (Namibia, South Africa)……….5

• 5. Tepals 2.5 – 6 cm long; tepal margins red to orange; tepals usually lightly fused at the apex at anthesis; parasitising Euphorbia spp. other than E. gregaria or E. gummifera………………………………………………………………..1. H. africana

• 5. Tepals 5.5 – 9 cm long; tepal margins red to pink; tepals usually free at the apex at anthesis; parasitising Euphorbia gregaria or E. gummifera………………………….2. H. visseri

• 6. Tepal margins entirely glabrous; apical osmophore absent, recessed osmophore inconspicuous; parasitising Commiphora spp.………………………..10. H. sinandevu

• 6. Tepal margins with at least some setae; apical osmophore present, recessed osmophore present but may be inconspicuous; parasitising Fabaceae spp…………………………7

• 7. Tepal margins and perianth tube entirely red; apical osmophore occupying ½ or more of the total tepal length; setae strigose……………………………………7. H. hanningtonii

• 7. Tepal margins red to pink and perianth tube white internally; apical osmophore occupying less than ½ of the total tepal length; setae villosa………………………………………8

• 8. Apical osmophore reduced to narrow inverted V-shaped strip above setae margins, outside this a glabrous outer cucullus; reduced tepal margins with setae not reaching dorsal edge of tepal; recessed osmophore prominent, extending down almost to basal end of tepal margin………………………………………………………………..9. H. bolinii

• 8. Apical osmophore clearly present, not as above; tepal margins with setae reaching dorsal edge of tepal; recessed osmophore present or inconspicuous…………………………..9

• 9. Recessed osmophore small or inconspicuous; tepal margins usually reddish-pink; border between apical osmophore and tepal margins very distinct, descending sharply from dorsal to ventral edge………………6. H. abyssinica

• 9. Recessed osmophore prominent, extending down almost to basal end of tepal margin; tepal margins usually pink to whitish-cream; border between apical osmophore and tepal margins very indistinct, not descending sharply from dorsal to ventral edge, often horizontal or ascending from dorsal to ventral edge…………………..8. H. solmsiana

I) Hydnora subgenus Hydnora. Type: Hydnora africana Thunb.

Hydnora subgenus Euhydnora Decne. (Decaisne 1873: 75)

Rhizome angular (4 – 8-sided) in cross-section, although growth tip and terminal several cm of rhizome may be terete with thinner periderm; rhizome ornamented with numerous lateral tubercles arranged in distinct parallel lines, with these lines defining the number of angles of the rhizome in cross section, on older specimens many of the tubercles may be swollen and distorted such that it is less evident that they are arranged in clear parallel lines. Flower usually 3-merous (rarely 4). Perianth tube red to pink to orange (in life), then darkening to dark red to brown or black over several days; tepals apically fused or not fused, if not fused then elliptic-lanceolate and curving inwards towards apex; tepal margins with setae though often conspicuous, tepal edges with or without large setae. Osmophore recessed in tepal cavity, spongy and white (at anthesis), darkening to grey/brown over several days, generating strong odours of rotting meat. Gynoecium with fleshy pedicel absent or rarely present.

1. Hydnora africana Thunb. (Thunberg 1775: 69); Meyer (1833: 98); Pappe (1862: 51); Thunberg (1823: 499); Solms-Laubach (1901: 6); Dinter (1909: 57); Thistleton-Dyer (1912: 486); Marloth (1913: 177); Dinter (1923: 424); Vaccaneo (1934: 436); Harms (1935: 288); Schreiber in Merxmüller (1968: 41); Musselman & Visser (1989: 323); Bolin et al. (2009b: 259); Bolin et al. (2018: 107). Type: South Africa, Caput Bonae Spei [Cape of Good Hope] [Western Cape], 1772 – 1775, Thunberg s.n. (lectotype LD! [LD1239903], selected here; isolectotype UPS [UPSV126487).

Aphyteia hydnora Ach. (Acharius & von Linné 1776: 10). Type: as for Hydnora africana.

Aphyteia africana (Thunb.) Oken (1841: 801); Merrill (1950: 270). Type: as for Hydnora africana.

Herbaceous subterranean perennial root holoparasite, without leaves or stem. Rhizome angular (4 – 8-sided) in cross-section, 1 – 3 cm in diam., growth tip and first several cm of rhizome terete with thinner periderm; rhizome surface coriaceous, brown to dark brown, lighter coloured near growth tip (in life); rhizome spreads laterally, may occasionally bi- or tri-furcate; rhizome ornamented with numerous lateral tubercles arranged in distinct parallel lines, with these lines defining the number of angles of the rhizome in cross section, tubercles remain the same or develop into a flower or develop into a haustorium, on older specimens many of the tubercles may be swollen and distorted such that it is less evident that they are arranged in clear parallel lines; internally fleshy, coloured bright red to pink. Flower usually 3-merous (rarely 4); flower emerges only partially from the soil; two floral chambers, androecial chamber subtended by gynoecial chamber, inner surfaces of chambers glabrous. Perianth external tissues coriaceous, brown, total length 8.2 – 19.5 cm; perianth tissues fleshy; perianth tube pink (in life), then darkening to orange and red over several days, 2 – 3.5 cm wide; tepal margins red to orange, tepals (measured from apex to point of connation with adjacent tepal) 2.5 – 6 × 2 – 6.5 cm (measured at midpoint); tepals elliptic-lanceolate, curving inwards towards apex, not fused at apex; tepal margin with short setae up to 2 mm, with almost complete coverage across surface, largest on dorsal side but still visible even on ventral side; dorsal tepal edge lined with a single row of considerably larger, conspicuous, broad, pale setae up to 6 mm. Osmophore recessed in tepal cavity, spongy and white (at anthesis), darkening to grey/brown over several days, generating strong odours of rotting meat. Androecium antheral ring formed by connate w-shaped anther lobes, 1.0 – 2.1 cm wide, forming a central orifice, transversely striate and divided into numerous horizontal pollen sacs; pollen bisulcate. Gynoecium ovary inferior, unilocular with numerous ovules, ovary 2.2 – 4.1 cm wide; sessile, cushion-like stigma on the floor of the gynoecial chamber, stigma 1.9 – 2.4 cm wide, fleshy pedicel sometimes present, 0 – 5 cm. Fruit subterranean, turbinate berry, leathery pericarp, 7 – 18 cm in diam. Seed spherical, black-brown, 0.7 – 1.2 mm, thousands of seeds within fruit, embedded within white pulp. Fig. 6.

Fig. 6.

Hydnora africana. A illustration of flower and rhizome; B example of herbarium specimen flower and rhizome (both Bolin 09-11 [WIND]); C – E live flowers. A illustrated by sebastian hatt.

distribution. Hydnora africana follows the distribution of its succulent Euphorbia hosts across southern Africa. It occurs from the Karas Region of southern Namibia (with a single collection in the Erongo Region), down into South Africa through the Northern, Western and Eastern Cape provinces. There are a few records in KwaZulu-Natal, and single records from the North West province and Free State province. It has generally been collected within 200 miles of the coast, with almost no specimens having been found in the interior provinces of South Africa. Within Namibia, it tends to be restricted to inselbergs or areas with winter rainfall (Bolin 2009). Map 2.

Map 2.

Distribution range of Hydnora africana.

specimens examined. namibia. Erongo Region: Omaruru, Brandberg, 15 April 1987, Craven 2765 (WIND!); Karas Region: Luderitz Distr., Namib-Naukluft Park [26°40'12.3"S, 15°22'48.8"E], 10 Nov. 2008, Bolin 09-24 (WIND!); Luderitz Distr., Spitzkoppe Farm [27°51'46"S, 16°43'04"E], 4 Aug. 2016, Chase & Rugheimer FC1071 (WIND!); Tschaunaup [2617DA], 25 Feb. 1947, Gerstner 6309 (PRE!); Warmbad, Farm Sperlingspütz [2181CA], 27 May 1972, Giess & Muller 12241 (WIND!); Rosh Pinah, Farm Namuskluft [S27 52.127, E16 51.727], 18 Oct. 2005, Bolin 08-12 (WIND!); Rosh Pinah, Farm Namuskluft [S27 52.127, E16 51.727], 3 Dec. 2008, Bolin 05-3 (WIND!). south africa. Northern Cape: Port Nolloth, Nick Kotze Farm [S29 16.225 E17 6.477], 17 Feb. 2009, Bolin 09-11 (WIND!); Farm Gemsbokvlei, ~30 km E of Port Nolloth [-29.264963, 17.103708], 30 Sept. 2023, Hatt 33 (K!, NBG!); Port Nolloth [2916BD], 6 June 1951, Hughes s.n. (NBG!); Richtersveld, E foothills of Numeesberg [2816BD], 3 Oct. 1972, Wisura 2523 (NBG!); Richtersveld, Cornel’s Kop [2816BD], 1 July 1978, Williamson 2818 (NBG!); Cornellskop, N face, [-28.4165241, 16.8850214], 3 Oct. 2023, Hatt 41 (K!, NBG!); Namaqualand [c2917], May 1916, Marloth 7425 (PRE!); Richtersveld, Khubusi Distr., SW side below Kornelskop on Bloeddrif Road [28°25.237'S, 16°35.047'E], 11 Sept. 2002, Venter & Venter 9804 (MO!, NY!, S); Richtersveld 11 [28°22'29"S, 16°54'52"E], 10 Sept. 2010, Bester 10105 (PRE!); Lekkersing, Karuchabpoort [2917AA], 26 Aug. 1977, Oliver et al. 144 (PRE!); Pofadder Distr., Great Bushmanland, 15 Nov. 1961, Horn 5 (NBG!); Pofadder Distr., Naroep [2818DC], 10 July 2008, Bruyns 11226 (BOL!, NBG!); Little Namaqualand, Anenous, 1931, Orpen s.n. (BOL!); Namaqualand, Veldevreden?, Oct. 1939, Hanekam? s.n. (SAM!); Kleinsee, 1931, Orpen s.n. (BOL!); Holgat, 1926, Pillans s.n. (BOL!); Between Zilverfontein, Kooperberg and Kaus, no date, Drege s.n. (HBG!, K!, L); Western Cape: Klaver, Varyhnsdorp [3118DC], 28 Sept. 1959, Drumsden s.n. (NBG!); Belville Distr., N of Cape Town [3318CB], 3 Jan. 2001, Burgoyne & Lewis 3 (PRE!); Malmesbury, Langebaan, 10 May 1944, Chaplin 4 (NBG!); Haakgat, off Otto du Plessis Drive, Blauwwberg Nature Reserve, Melkbosstrand [-33.752298, 18.441915], 27 Sept. 2023, Hatt 26 (K!, NBG!); Calvinia [3119BD], July 1937, Schmidt 602 (PRE!, K!); Malmesbury, Hopefield [3318AB], no date, Marloth 10559 (PRE!); Saldana, no date, Osbeck s.n. (S); Rooifontein Farm, 6 km N of Karoopoort, [3319BA], 4 Oct. 1986, Hilton-Taylor 1666 (NBG!); Near Ceres Karoo and Witzinberg, 20 Oct. 1973, Carlquist 4840 (HUH!, US!); Calvinia, Touws-Rivier, 1 Dec. 1986, Breckle 10179 (M!); Calvinia, Lokenburg [3119CA], 12 Dec. 1953, Acocks 17363 (PRE!); Worcester, Brewelskloof [3319CD], 15 Oct. 1978, Bayer 1564 (NBG!); Worcester, in veld opposite entrance to “Kanetvlei” [3319DA], 16 Sept. 1980, Walters 2448 (NBG!); Worcester [3319CB], 19 Nov. 1925, Pole-Evans 3023 (PRE!); Worcester [3319CB], Oct. no year, Marloth 5701 (PRE!); Worcester, Karoo Botanic Gardens [3319CB], 4 Oct. 1991, Steiner 2418 (NBG!); Worcester, Karoo Botanic Gardens, 1946, Leighton 2835 (BOL!); Worcester, Karoo Botanic Gardens, 6 April 1985, Musselman 7041 (K!); Worcester, Karoo Botanic Gardens, 21 Dec. 1946, Compton 18661 (NBG!); Karoo Desert Botanic Garden, along one of the eastern paths [-33.61469, 9.452395], 26 Sept. 2023, Hatt 20 (K!, NBG!); Swellendam, 20 miles from Heidelberg Cape, 7 Oct. 1949, Kramer s.n. (NBG!); Swellendam, 1930, Hurling s.n. (BOL!); Prince Albert, Swartzkraal, 17 July 1939, Oosthuyzen s.n. (NBG!); Ladismith, on farm of Mrs R. W. von Moltke, 16 Oct. 1959, Von Moltke s.n. (NBG!); Riversdale, near Riversdale, 22 Feb. 1926, Muir s.n. (K!); Riversdale, near Riversdale, 16 Jan. 1928, Muir s.n. (K!); Cape Peninsula, Llandudno, 26 June 1940, Bond 403 (NBG!); Cape Peninsula, Llandudno, 9 Jan. 1938, Hafström & Acocks 409 (S); Cape Peninsula, Houdeklip Bay, between Llandudno and Hout Bay [3418AB], 4 Oct. 1925, Dumsday 6653 (PRE!); Cape Town, Llandudno Beach, 1941, Esterhuysen 7621 (BOL!); Llandudno, no date, Acocks s.n. (S). Eastern Cape: East London [3327BB], Oct. 1964, Buys 34 (PRE!); East London, Buffalo R., 2^nd Creek [3327BB], 2 March 1898, Heaton & Galpin 3189 (PRE!); East London, 2^nd Creek [3327BB], Jan. 1927, Smith 3852 (PRE!); East London, bank of Nahoon R., 6 Oct. 1900, Medley-Wood 53471 (PRE!); East London, 1933, Everitt s.n. (BOL!); East London [3327BB], Nov. 1917, Doidge s.n. (PRE!); East London, Nov. 1962, Bokelmann 3 (NBG!); Kidd’s Beach, 30 miles W of East London [3327BA], 9 Feb. 1967, Venter 3273 (NBG!, PRE!); Queenstown Distr., banks of White Kei R., Gwatyu [c3127CC], Jan. 1911, Spence 8005 (PRE!); King William’s Town, Convent of the Sacred Heart [3227CD], no date, de Victoria s.n. (PRE!); Port Alfred, Kowie River Preserve [S33 34.055, E26 51.751], 25 Nov. 2008, Bolin 0-82 (WIND!); Kowie River West [3326DB], Aug. 1917, Tyson TRV19227 (PRE!); Bathurst Distr., Port Alfred, 23 Aug. 1948, Symons s.n. (NBG!); Bathurst Distr., Slankpe Farm, 1959, collector unknown s.n. (GRA!); Alexandria, 1946, Holland s.n. (BOL!); 15 km E of Alexandria, 1978, Kopke s.n. (GRA!); Jansenville [3224DC], 4 Jan. 1934, Long 1322 (K!, PRE!); Jansenville, Wolverfontein [3224DC], Nov. 1885, MacOwan 1724 (BM!, HUH!, K!, PRE!); Steyterville, 1915, Hepburn s.n. (GRA!). KwaZulu-Natal: Tugela Ferry and Keats Drift, Aug. 1937, Smith 64032 (PRE!); Tugela Valley [2830D], no date, Smith s.n. (PRE!). Free State: Close to Rietrivier, before 1887, MacOwan 1488 (BM!). North West Prov.: Zwartruggens Distr., Somerset, ad Loot’s Kloof, no date, Leonard s.n. (BM!). Unlocalised: no date, collector unknown s.n. (MO [MO2740130]); no date, collector unknown s.n. (MO! [MO2861368]); early 1800s, Zeyher 1511 (G!, K!); 1877, Barkley s.n. (K!); no date, Brown s.n. (BM!); Kap, early 1800s, Ecklon & Zeyher s.n. (B!); Cape of Good Hope, no date, Masson s.n. (BM!); no date, McGibbon s.n. (K!); Cap. B. Spei., 1772 – 1775, Thunberg s.n. (lectotype LD!, isolectotype UPS); ex Africae aristidis desertis, 1772 – 1775, Thunberg s.n. (UPS); Habitat ad Cap. B. Spei, no date, collector unknown s.n. (B!); Kaap Prov, Sept. 1887, collector unknown s.n. (L!); no date, collector unknown s.n. (BM! [BM000939563]); no date, collector unknown s.n. (BM! [BM000939564]); no date, Atherstone s.n. (K!); Karroo Desert, before 1867, Kochmann s.n. (K!); Cap. Ban. Op.?, no date, collector unknown s.n. (M!); no date, Thunberg s.n. (S); Goda Hoppsudden, no date, Thunberg s.n. (S); Habitat ad Cap. B. Spei, no date, Thunberg s.n. (S); no date, collector unknown s.n. (S [S11-11398]); no date, collector unknown s.n. (S [S-PLE-E10324]); before 1867, Baron von Ludwig s.n. (K!); no date, collector unknown s.n. (BM! [BM000939567]).

habitat. Rocky or sandy soils amongst succulent Karoo vegetation and coastal forest grasslands, in close proximity to its Euphorbia hosts.

conservation status. Hydnora africana is relatively widespread across South Africa and grows on a number of different Euphorbia hosts. In 2008 it was listed as ‘Least Concern, Not Threatened’ in the Red List of South African Plants (Williams et al. 2008). The same ranking is true for all of the recorded hosts, suggesting that H. africana is not currently threatened. However, populations may be locally threatened where harvesting for food/medicinal utility is common.

phenology. Flowering time seems to be dependent on rainfall, which varies considerably across its distribution in South Africa. For example, flowering occurs more around November to January in the summer rainfall areas of the Eastern Cape, while around August to October in the winter rainfall areas of the Northern and Western Cape and around February to April in southern Namibia (Bolin 2009). Fruit maturation time is thought to be at least several months to a year.

etymology. The specific epithet refers to its distribution in South Africa.

vernacular names. In Afrikaans, Hydnora is commonly known as jakkalskos, meaning ‘jackal food’, referring to the tendency of jackals to dig up and eat the fruits. It is also less commonly known as bobbejaankos, meaning ‘baboon food’, Baviaanskos, Baviaanskost, Jakhalskost, Kannip, Kannikan and Kaw-imp (Williams et al. 2008; De Beer & Van Wyk 2011). It is known as uMavumbaka in Zulu, meaning ‘the one that pops up’, a name that is shared with Sarcophyte sanguinea (Williams et al. 2011a). It is also known as Idolo-lenkonyane, Umafumbuka and Ubuklunga in Xhosa (Williams et al. 2008; Bisi-Johnson et al. 2010).

ethnobotany. Xhosa men and women are known to use the roots and fruiting body in skin care (Mwinga et al. 2019). They are crushed and macerated with a coarse stone into a thin reddish paste, which is applied to the skin for preventing acne and treating pimples, with daily application until symptoms disappear (Dold & Cocks 2005). The paste is also used for sun protection and is recorded to be used on a daily basis by many outdoor workers. The apparent effectiveness has been suggested to be linked to the high tannin content in the fruit. It has been recorded being sold at traditional medicine markets in South Africa, such as Grahamstown Market, Eastern Cape. The fruit is reported to be delectable (Visser 1981) and apparently eaten by Nama people (Gerstner 6309 [PRE0369198]). In 1934, Vaccaneo recorded that the Khoekhoe people eat the fruit raw or roasted. The rhizomes have been used in the treatment of diarrhoea, swollen glands or inflamed throat (Bisi-Johnson et al. 2010), oral thrush (De Beer & Van Wyk 2011), dysentery (Olajuyigbe & Afolayan 2012), cardiovascular diseases (Omoruyi et al. 2012) and fever, asthma, constipation, hypertension and oesophageal cancer (Otang et al. 2012). Baboons and jackals have been recorded eating the fruits (Williams et al. 2008; De Beer & Van Wyk 2011).

host specificity. Hydnora africana appears to be restricted to succulent Euphorbia hosts (Table 1). The following hosts have been recorded, grouped here into subgenera following Bruyns et al. 2006: Euphorbia subgenus Chamaesyce: E. burmannii (Klotzsch & Garcke) E.Mey. ex Boiss., E. chersina N.E.Br., E. karroensis (Boiss.) N.E.Br., E. rhombifolia Boiss.; E. subgenus Rhizanthium: E. caput-medusae L.; E. lignosa Marloth; E. subgenus Euphorbia: E. caerulescens Haw., E. ingens E.Mey. ex Boiss., E. grandidens Haw., Euphorbia tetragona Haw., Euphorbia triangularis Desf. ex A.Berger.; E. subgenus Esula: E. mauritanica L. Within these species, there are representatives from at least four subgenera of Euphorbia. Outside Euphorbiaceae, Cotyledon orbiulatum L. was reported as a host by Thunberg in the protologue for H. africana (1775). However, the validity of this plant as a host is highly unlikely given no evidence has been found of non-Euphorbia hosts since then, despite a considerable number of collections and observations having been made.

notes. Hydnora africana is the type for the genus and perhaps the most well-known Hydnora species. Although H. visseri, H. solmsiana and H. abyssinica have been recorded from South Africa, H. africana is by far the dominant Hydnora species in South Africa. For much of its range, particularly in the Western and Eastern Cape provinces, it is the only recorded Hydnora species. In a few 19^th century journals, there are scattered mentions of a species called Hydnora capensis from South Africa (McGibbon 1868) or Hydnora capensi Thunb. by Welwitsch (1869). These names are not validly published and are almost certainly synonymous with H. africana, although the origin of this confusion is unclear.

Hydnora africana is only confusable with H. visseri and H. longicollis and, morphologically, it falls in between the two. H. longicollis predominantly grows in Angola and northern Namibia rather than South Africa, and is distinctly smaller, with tepals < 2.4 cm long and more sharply curved. H. visseri also has a more northerly range, although there is some overlap with H. africana around the South African-Namibian border region. They are distinguishable from herbarium specimens, with H. visseri being distinctly larger with flowers up to 24 cm long and tepals 5.5 – 9 cm long that are more erect than H. africana. H. africana parasitises a range of species but has not been recorded on the hosts of H. visseri or H. longicollis, although this should not be treated as a reliable character on its own.

2. Hydnora visseri Bolin, E.Maass & Musselman (Bolin et al. 2011: 255, 2018: 107). Type: Namibia, Karas Region: Rosh Pinah, Farm Namuskluft, parasitising Euphorbia gummifera, 554 m, 27°54.283S, 16°50.308E, 12 Dec. 2005, Bolin 05-4 (holotype WIND!; isotype US!).

Herbaceous subterranean perennial root holoparasite, without leaves or stem. Rhizome angular (4 – 6-sided) in cross-section, 1 – 7 cm in diam., growth tip and first several cm of rhizome terete and with thinner periderm; rhizome surface coriaceous, brown to brown-orange, lighter coloured near growth tip (in life); rhizome spreads laterally, usually at shallow depths less than 30 cm below the soil surface, may occasionally bi- or trifurcate; rhizome ornamented with numerous lateral tubercles arranged in distinct parallel lines, with these lines defining the number of angles of the rhizome in cross-section, tubercles remain the same or develop into a flower or develop into a haustorium; rhizome fleshy, internally coloured dark red (thicker portions) to pale pink (near growth tip). Flower usually 3-merous (rarely 2, 4 or 5); flower emerges only partially from the soil; two floral chambers, androecial chamber subtended by gynoecial chamber, inner surfaces of chambers glabrous. Perianth external tissues coriaceous, brown, often resembling the soil, total length 10.5 – 24 cm; perianth tissues fleshy; perianth tube pink (in life), then darkening to orange and red over several days, 4.8 – 11.2 cm wide; tepals reddish-pink to red, tepal (measured from apex to point of connation with adjacent tepal) 5.5 – 9 × 2.0 – 5.7 cm (measured at midpoint); tepals elongate-lanceolate, curving inwards towards apex, not fused at apex; tepal margin with small setae up to 4 mm, but generally smaller than this and inconspicuous; dorsal tepal edge lined with single row of considerably larger, conspicuous, broad, pale setae up to 6 mm, but generally smaller than this. Osmophore recessed in tepal cavity, spongy and white (at anthesis), darkening to grey/brown over several days, generating odours of rotting meat. Androecium antheral ring formed by connate w-shaped anther lobes, forming a central orifice, transversely striate and divided into numerous horizontal pollen sacs; pollen bisulcate, 22 – 27 × 15 – 20 µm. Gynoecium ovary inferior, unilocular with numerous ovules, ovary 2.2 – 4.5 cm wide; sessile, cushion-like stigma on the floor of the gynoecial chamber, stigma 1 – 3.1 cm wide. Fruit subterranean, turbinate to globose berry, leathery pericarp, 5 – 20 cm in diam. Seed spherical, black-brown, 0.7 – 1.2 mm, thousands of seeds within fruit, embedded within white pulp, seed with hard testa and undifferentiated spherical embryo embedded in endosperm. Fig. 7.

Fig. 7.

Hydnora visseri. A illustration of flower and rhizome; B example of herbarium specimen flower (Bolin 09-24 [WIND]) and rhizome (Mannheimer 250 [WIND]); C – E live flowers. A illustrated by sebastian hatt.

distribution. Known primarily from the Hardap and Karas regions of southern Namibia, but also areas of the Northern Cape of South Africa close to the border with Namibia, in proximity to the Orange River. These are all areas of mostly summer rainfall. The distribution appears to match closely with those of its two hosts, Euphorbia gregaria and E. gummifera. Although not recorded from the eastern portion of the Northern Cape, Hydnora visseri is thought likely to grow there given the abundance of E. gregaria present (Bolin et al. 2011). Collected from an elevation of 0 – 1600 m. Map 3.

Map 3.

Distribution range of Hydnora visseri.

specimens examined. namibia. Hardap Region: 50 km S of Maltahöhe, 1981, Lavranos & Pehlemann 20205 (WIND!); Karas Region: Bethanie Distr., near Kuibis, May 1913, Range s.n. (SAM!); South Luderitz, Aug. 1971, Logan & Jensen 4966 (WIND!, only rhizomes but likely H. visseri due to locality); Farm Altdorn 3 [27°54'S, 17°41'E], 19 Jan. 1987, Ward 10117 (K!, PRE!, WIND!); Sandveld, on koppie next to road [27°1'1"S, 18°14'9"E], 27 March 1998, Strohbach & Dauth 3755 (WIND!); Sperrgebiet [26°59'47"S, 15°21'58"E], 11 July 1995, Strohbach 766 (WIND!); Sperrgebiet, Roter Kamm. [27°37'30"S, 16°22'30"E], 29 Sept. 1996, Mannheimer & Mannheimer 425 (WIND!); Eastern Sperrgebiet [27°0'27"S, 15°47'23"E], 25 Sept. 1996, Mannheimer & Mannheimer 303 (WIND!); Eastern Sperrgebiet [27°1'13"S, 15°42'58"E], 25 Sept. 1996, Mannheimer & Mannheimer 299 (WIND!); Sperrgebiet, approach to Klinghardt Mts [27°21'2"S, 15°41'11"E], 21 Sept. 1996, Mannheimer 250 (WIND!); Karasburg, Farm Karios 8 [27°37'30"S, 17°52'30"E], 15 April 2000, Strassen 114 (WIND!); Seeheim, 1927, Pellans 5850 (BOL!); Seeheim, Farm Kanas, 20 Oct. 2005, Bolin 0-51 (WIND!); Karas, Gondwana Canyon Park [S27 32.850, E17 52.972], 14 Feb. 2009, Bolin 908 (WIND!); Farm Jerusalem 73 [28°23'46"S, 19°36'51"E], 7 March 2008, Klaassen & Rugheimer EK1823 (US, WIND!); Farm Sandfontein 131 [28°45'36"S, 18°25'18"E], 11 March 2008, Klaassen & Rugheimer EK1891 (WIND!); Rosh Pinah, Farm Namuskluft [S27 54.283, E16 50.308], 3 Dec. 2008, Bolin 0-85 (WIND!); Rosh Pinah, Farm Namuskluft [S27 54.283, E16 50.308], 12 Dec. 2005, Bolin 0-54 (holotype WIND!, isotype US!); Unlocalised: 28 Jan. 1929, Dinter 6011 (B!). south africa. Northern Cape: Sendlingsdrift [S28 20.699, E16 55.522], 16 Feb. 2009, Bolin 9-10 (WIND!); 1 km E along road to Sendelingsdrif from junction with Bloeddrif, just N of Annisrivier [-28.40688, 16.901673], 2 Oct. 2023, Hatt 39 (K!, NBG!); Aggeneys Black Mountain Mine [2918BA], 2 Feb. 2000, Burgoyne 7981 (PRE!).

habitat. Flat, sandy soils amongst Nama-Karoo and succulent Karoo vegetation types, in areas where Euphorbia gregaria and E. gummifera are prevalent. Occasionally seen in rocky soil types.

conservation status. The authors of the type description (Bolin et al. 2011) state that Hydnora visseri should be considered Least Concern (LC) according to IUCN criteria. It is relatively widespread and common across its range. Furthermore, neither of its hosts are listed as threatened on either the Namibian or South African red lists (Loots 2005; Williams et al. 2008). Both host species receive some degree of protection as succulent Euphorbia under CITES Appendix 2. Although intensive land use for grazing occurs in southern Namibia, the host plants are not palatable food for grazers, and thus their abundance appears to be largely unaffected (Bolin et al. 2011). Indeed, as they have been observed persisting on farms in Namibia it is unlikely that H. visseri is currently threatened under the existing regime of land use. More local threats such as mining activity near Rosh Pinah have been noted but are unlikely to be impactful unless there is a significant increase in activity (Bolin et al. 2011).

phenology. Flowers primarily from October to January, although sporadic flowering has been observed all year round, likely due to flowering being dependent on adequate rainfall. Fruit maturation time is thought to be greater than eight months, with ripe fruits often existing alongside flowers from the next year (Bolin et al. 2011).

etymology. The specific epithet refers to Professor Johann H. Visser (1931 – 1990), a South African botanist and expert on South African parasitic plants.

vernacular names. All Hydnora species in South Africa are known as jakkalskos, bobbejaankos or stinkblom in Afrikaans. In Nama-Damara/Khoekhoe it is known as kani or kanip (Bolin et al. 2011).

ethnobotany. It is unknown whether local people in Namibia discriminate between Hydnora africana and H. visseri (see ethnobotany entry for H. africana). Mature fruits are reported to be enjoyed as a raw food source by local peoples and bitter immature fruits are made palatable by preparing them by roasting in coals, with or without animal milk (Ward 10117 [PRE0796363], Bolin et al. 2011).

host specificity. Euphorbia gregaria Marloth and E. gummifera Boiss. are confirmed hosts, both in E. subgenus Euphorbia sect. Tirucalli, and there is a single report of E. gariepina Boiss., in E. subgenus Rhizanthium (Table 1). Even in areas where other known Euphorbia hosts of H. africana co-occur with these species, H. visseri seems to exclusively favour E. gregaria and E. gummifera (Bolin et al. 2011).

notes. The largest of the species in subgenus Hydnora, H. visseri is distinguished from the similar H. africana and H. longicollis by its larger, more elongate, erect tepals, and by its unique hosts, Euphorbia gregaria and E. gummifera. It was only recently separated from H. africana as a distinct species (Bolin et al. 2011).

3. Hydnora longicollis Welw. (Welwitsch 1869: Tab. XXI). Baker & Wright in Thistleton-Dyer (1909: 132); Dinter (1923: 424); Vaccaneo (1934: 439); Harms (1935: 290); Bolin (2009: 85); Bolin et al. (2018: 107). Type: Angola, Mossamedes Distr. As far as Cabo Negro, maritime sandy areas, 1859, Welwitsch 530 (syntypes BM!, G!, K!, LISU!).

Hydnora africana var. longicollis Welw. (Welwitsch 1869: 66); Hiern (1900: 910); Musselman & Visser (1989: 323). Type: as for Hydnora longicollis Welw.

Hydnora longicollis Welw. sub var. africanae (Welw.) Solms-Laubach (1901: 6). Type: as for Hydnora longicollis Welw.

Herbaceous subterranean perennial root holoparasite, without leaves or stem. Rhizome angular (4 – 6-sided) in cross-section, 1 – 3 cm in diam.; rhizome surface coriaceous, brown to dark brown; rhizome spreads laterally, usually at shallow depths, may occasionally bi- or trifurcate; rhizome ornamented with numerous lateral tubercles arranged in distinct parallel lines, with these lines defining the number of angles of the rhizome in cross-section, tubercles remain the same or develop into a flower or develop into a haustorium; rhizome fleshy, internally coloured rich red. Flower usually 3-merous (rarely 4); flower emerges only partially from the soil; two floral chambers, androecial chamber subtended by gynoecial chamber, inner surfaces of chambers glabrous. Perianth external tissues coriaceous, brown, often resembling the soil, total length 5.1 – 14.5 cm long; perianth tissues fleshy; perianth tube pink (in life), then darkening to orange and red over several days, 1.2 – 2.7 cm wide; tepals pinkish, tepal (measured from apex to point of connation with adjacent tepal) 0.9 – 2.4 × 1.2 – 2.7 cm (measured at midpoint); tepals elliptic-lanceolate, curving inwards towards apex, not fused at apex; tepal margin with setae up to 2 mm, with almost complete coverage across surface, largest on dorsal side but still visible even on ventral side; dorsal tepal edge lined with a single row of considerably larger, conspicuous, broad, pale setae up to 4 mm. Osmophore recessed in tepal cavity, spongy and white (at anthesis), darkening to grey/brown over several days, generating foul odour. Androecium antheral ring formed by connate anther lobes, 1.0 – 2.1 cm wide, forming a central orifice, transversely striate and divided into numerous horizontal pollen sacs; pollen bisulcate. Gynoecium ovary inferior, unilocular with numerous ovules, ovary 1.7 – 3.3 cm wide, sessile, cushion-like stigma on the floor of the gynoecial chamber, stigma 1.0 – 1.8 cm wide. Fruit subterranean, turbinate to globose berry, leathery pericarp, 3 – 5 cm in diam. Seed spherical, black-brown, 0.7 – 1.2 mm, thousands of seeds within fruit, embedded within white pulp. Fig. 8.

Fig. 8.

Hydnora longicollis. A illustration of flower and rhizome; B example of herbarium specimen flower (Carisso & Sousa s.n. [BM]) and rhizome (Humbert s.n. [P]); C – E live flowers. A illustrated by sebastian hatt.

distribution. Known from southwestern Angola and the Erongo region of Namibia. In Namibia, it largely follows the distribution of its host, Euphorbia damarana. There are only a few collections in Angola, most of which are in the sandy, coastal Namibe Province. However, a recent collection was made considerably further inland in the Huíla Province. The full extent of its distribution in Angola is likely undiscovered, especially given much of the country has been considerably understudied in the last century due to civil unrest (Bolin 2009). Its distribution is largely disjunct from and adjacent to Hydnora visseri and H. africana. Map 4.

Map 4.

Distribution range of Hydnora longicollis.

specimens examined. angola. Namibe Prov.: approx 60 km E of Namibe, 2009, Voigt 67 (GHPG!); South of Namibe, 2006, Bird s.n. (ODU!); Mossamedes Distr. as far as Cabo Negro, 1859, Welwitsch 530 (syntypes BM!, G!, K!, LISU!); Mossamedes, desert SE of Mossamedes, 11 June 1937, Carrisso & de Sousa s.n. (BM!); desert SE of Mossamedes, 1937, Humbert s.n. (P!). namibia. Erongo Region: ~ 65 km E of Hentiesbaai [-21.905083, 14.85075], 6 June 2022, Hatt 17 (K!, WIND!); 10 km SE of Bloedkoppie [-22.908526, 15.34823], 6 Nov. 2023, Hatt 46 [GBB]; 25 km SE of Uis [S21 26.346, E15 03.115], 3 Feb. 2009, Bolin 0-92 (WIND!); Omaruru, approx. 15 km S of Omaruru R. [21°30'18"S, 15°09'41"E], 10 April 2009, Bolin 09-30 (WIND!); Swakopmund, no date, Weiss s.n. (M!); Henties Baai, Messum Crater, 2003, Maass s.n. (ODU!).

habitat. Sandy, rocky or gravel plains, sometimes in relative proximity to the coast or in dry riverbeds.

conservation status. There is very limited data on the conservation status of Hydnora longicollis, largely because the full extent of its distribution remains unclear. Much of Namibia’s west coast falls within designated areas with at least some degree of protection. Alternatively, conservation in Angola has largely been neglected in the past, with much of its natural landscape devastated by extensive conflict or unregulated development (Huntley et al. 2019). However, one of its primary hosts, Euphorbia damarana, has been listed as ‘not threatened’ according to the national red list, which is encouraging for H. longicollis (Ministério do Ambiente 2018). Considerable further research is clearly required on this matter to make a confident decision on the threat status.

phenology. In Angola, flowering has mainly been observed in January, while in northern Namibia it occurs more in February to April. However, sporadic flowering has been observed all year round, likely due to flowering being dependent on adequate rainfall. Fruit maturation time is thought to be at least several months (Bolin 2009).

etymology. The specific epithet translates to ‘long neck’ from latin. This likely refers to the proportionately long perianth tube seen in some specimens, although this a variable character.

vernacular names. In Nama-Damara/Khoekhoe, ‘Hydnora’ is known as kani or kanip (Bolin et al. 2011). It is unclear whether this is specific to H. visseri or whether it also applies to H. longicollis as well. Vernacular names for Hydnora in Angola are currently unrecorded.

ethnobotany. Data for the uses of Hydnora longicollis are limited to the type description in 1869. Welwitsch (1869) writes that all parts of the plant, particularly the rhizomes, are astringent. At the time, they were sometimes used in fishing nets and fabric dyes by the Mossamedes locals. He also notes a potential value in healing sores as a mucous membrane. It is unknown whether local people discriminate between H. longicollis, H. visseri and H. africana in terms of usage, particularly in Namibia where there is overlap in their distributions. One specimen from Munich reports that it belongs to a collection of ‘plants that provide tannins’ (Weiss s.n. [M]). Tannins have been reported from other Hydnora species (see H. solmsiana: Ethnobotany).

host specificity. Euphorbia damarana L.C.Leach (E. subgenus Euphorbia sect. Tirucalli) is a confirmed host (Table 1). There is a single report on E. gariepina and two reports on E. virosa Willd. The type description notes Roepera orbiculata (as Zygophyllum orbiculata) (Zygophyllaceae), although this was most likely a mistake as it belongs to an entirely unrelated family to Euphorbia and no reports have corroborated this since (Welwitsch 1869). The shortage of herbarium specimens and observational data has considerably limited our understanding of host preference for this species.

notes. Species description adapted from (Bolin 2009). Hydnora longicollis has suffered much neglect since its discovery in 1869. This is largely due to the political instability in Angola that has made collecting there a challenge, hence the relatively small number of herbarium specimens (Bolin 2009). There is some taxonomic confusion surrounding its name, as the protologue cites it as H. africana var. longicollis Welw., but the illustration in the same report is labelled as H. longicollis Welw. (Welwitsch 1869). Therefore, both names are technically valid according to the International Code of Nomenclature (Turland et al. 2018). Since then, both names seem to have been used interchangeably (Vaccaneo 1934; Harms 1935). Bolin et al. (2011) consider H. longicollis Welw. to be a species in its own right due to its morphological distinction from H. africana, as well as its non-overlapping distribution and distinct host species. Furthermore, an unpublished phylogeny of Hydnora from Jay Bolin’s PhD thesis supported this species concept (Bolin 2009). Therefore, this species is accepted here.

4. Hydnora triceps Drège & E.Mey. (Meyer 1833: 779); Harvey (1863: 187); Solms-Laubach (1901: 6); Thistleton-Dyer (1912: 487); Marloth (1913: 178); Vaccaneo (1934: 446); Harms (1935: 293); Musselman & Visser (1989: 323); Bolin et al. (2018: 107). Type: South Africa. Zwischen Zilverfontein, Kooperberg und Kaus, Sept./Oct. before 1833, Drège s.n. (syntypes HBG!, K!, L!).

Aphyteia triceps Steud. (Steudel 1840: 111). Type: as for Hydnora triceps Drège & E.Mey.

Aphyteia multiceps Burch. (Burchell 1822: 213). Type: as for Hydnora triceps Drège & E.Mey.

Herbaceous subterranean perennial root holoparasite, without leaves or stem. Rhizome angular, 4 – 7 (– 9)-sided, in cross-section, 1 – 3 cm in diam., rhizome surface coriaceous, dark-brown to almost black; rhizome spreads laterally at various depths between 5 – 15 cm below ground; rhizome ornamented with numerous lateral tubercles arranged in distinct parallel lines, with these lines defining the number of angles of the rhizome in cross-section, tubercles remain the same or develop into a flower or develop into a haustorium; rhizome fleshy, internally coloured deep scarlet (thicker portions) to pinkish (near growth tip). Flower usually 3-merous; flower almost entirely underground, usually mature flowers are up to 5 cm underground although sometimes the top 2 cm of the perianth represented by the 3 opening between tepals will be exposed; two floral chambers, androecial chamber subtended by gynoecial chamber, inner surfaces of chambers glabrous. Perianth external tissues coriaceous, pinkish-brown, total length 10 – 15 cm long; perianth tissues fleshy; perianth tube pinkish-yellow, 2 – 10 cm long and 2.5 – 3.5 cm wide; tepals broadly clavate but incurved and apically fused to form 3 distinct lateral apertures, each with a pronounced cucullate flesh-coloured mouth; well-developed horizontal flange fused into a prominent circular ridge on the inside below the apertures; tepal interior glabrous. Osmophore recessed below each horizontal flange, below the apertures, spongy and white (at anthesis), darkening to dark brown over several days, generating strong foetid odour shortly after opening. Androecium 3 erect connate anther lobes, inserted near apex of perianth tube, forming central orifice, transversely striate and divided into numerous vertical pollen sacs with abundant, yellow, slightly sticky pollen. Gynoecium ovary inferior, unilocular with numerous pendulous ovules; sessile, cushion-like stigma on floor of the gynoecial chamber, stigma distinctly trilobed. Fruit subterranean, turbinate to globose berry, leathery external pericarp, internally bright pink, astringent taste, faint coconut odour, slightly dehiscent at apex when ripe, 3 – 10 cm in diam. Seed spherical, black-brown, embedded within white pulp. Fig. 9.

Fig. 9.

Hydnora triceps. A illustration of flower and rhizome; B example of herbarium specimen flower and rhizome (both Bolin 08-7 [WIND]); C – E live flowers. A illustrated by sebastian hatt.

distribution. Known from the Northern Cape of South Africa and the Karas Region of Namibia, following the narrow distribution of Euphorbia dregeana. Map 5.

Map 5.

Distribution range of Hydnora triceps.

specimens examined. namibia. Karas Region: Rosh Pinah, road verge next to Farm Namuskluft, 14 Sept. 2001, Maass & Musselman 1 (WIND!); Rosh Pinah, Farm Namuskluft [S27 56.427, E16 48.141], 3 Dec. 2008, Bolin 0-87 (WIND!); Rosh Pinah, Farm Namuskluft [S27 56.427, E16 48.141], 11 July 2009, Bolin 09-41 (WIND!); Rosh Pinah, Farm Namuskluft 88 [27°53'27"S, 16°51'21"E], 3 Aug. 2002, Maass EM18 (WIND!); Namian’s Saddle [27°53'27"S, 16°51'21"E], 20 Sept. 2003, Mannheimer CM2438 (WIND!). south africa. Northern Cape: Port Nolloth, Farm Gemsbokvlei [29°23'47"S, 17°12'40"E], 21 Dec. 2002, Maass & Musselman 2 (WIND!); Port Nolloth, Nick Kotze Farm, 17 Feb. 2009, Bolin 09-14 (WIND!); 20 km N along Lekkersing Rd from the R382 to Port Nolloth [-29.19116, 17.130254], 1 Oct. 2023, Hatt 37 (K!, NBG!); ~ 7 km N along Wolfberg Rd from R355 [-29.4761, 17.364589], Hatt 43 (K!, NBG!); Modderfontein, Whitehead, Namaqualand, near O’Kiep, c. 1887, Hofmeyer s.n. (K!, SAM!); Zwischen Zilverfontein, Kooperberg und Kaus, Sept./Oct. before 1833, Drège s.n. (syntype HBG!, K!, L!); Farm Droëskraal, in the vicinity of Naroegas se Berge, 1988, Visser STE31789 (PRE!). unlocalised. 1887, Marloth 1863 (PRE!); no date, collector unknown s.n. (K!, mixed specimen with H. africana).

habitat. Sandy plains and stony hills, following the preferred habitats of Euphorbia dregeana. Visser (1988) noted a preference in Hydnora triceps for sandy habitats.

conservation status. Both Hydnora triceps and its host Euphorbia dregeana appear as ‘Least Concern’ on the Red List of South African Plants (Williams et al. 2008). While there are some potential local threats such as mining activities near Rosh Pinah, the host population seems to be largely stable. While there is intensive land use in the area, populations of both host and parasite appear to be able to survive regardless. This is perhaps because the host plant is not eaten by grazing animals. Indeed, a significant population was found growing on a farm in Namibia in 2004 (Maass & Musselman 2004).

phenology. Full details of phenology are unknown, but Hydnora triceps has been collected in flower during September on two occasions and collected in fruit in December on one occasion (Visser 1988; Maass & Musselman 2004).

etymology. The specific epithet translates to ‘three-headed’, referring to the three distinct apertures formed by the incurved tepals.

vernacular names. None recorded.

ethnobotany. No uses recorded. It is possible that it is used indiscriminately from Hydnora africana or H. visseri, given the overlap in range. However, given that most specimens don’t even breach the surface of the soil, it is also possible that they go largely unnoticed by local people.

host specificity. The only known host is Euphorbia dregeana E.Mey. ex Boiss. (in E. subgenus Rhizanthium) (Table 1).

notes. Species description adapted from (Maass & Musselman 2004). Hydnora triceps is very distinctive and unlikely to be confused with any other species. No other species has fused tepals and an entirely hypogeous habit. It has only been collected a handful of times since its discovery in 1833. It was thought extinct until it was rediscovered in South Africa in 1988 by Johann Visser and later in Namibia in 2001 by Erika Maass (Visser 1988; Maass & Musselman 2001). The rarity of this species is unclear. Despite its occurrence in South Africa, a relatively thoroughly surveyed country, the small number of collections may simply reflect that the fact that it is notoriously difficult to see. In most cases, the only evidence of its presence of small cracks in the soil, over a small bulge in the soil, leading to the openings of the flower.

II) Hydnora subgenus Neohydnora Harms (1935: 293). Type: Hydnora esculenta Jum. & H.Perrier

Rhizome angular (4 – 7-sided) in cross-section; tubercles arranged in distinct parallel lines, with these lines defining the number of angles of the rhizome in cross-section. Flower (3 – (4 – 5) – 6)-merous. Perianth external tissues coriaceous, brown to purplish-brown; perianth tube whitish inside, fading to brown-black at the tepals; tepal margins whitish-cream, darkening to brown with age; tepal often strongly recurved towards the apex, not fused at apex; tepal margin with numerous setae, c. 3 mm in diam., transitioning into a distinctly ribbed surface. Osmophore apical, not recessed, often split down the middle such that it is divided into two triangles barely fused at the apex, spongy and white to tan (in life), darkening to yellow when dried, generating musty, foetid odour, but not one of rotting flesh. Gynoecium fleshy pedicel sometimes present, 0 – 5 cm long.

5. Hydnora esculenta Jum. & H.Perrier (Jumelle & Perrier 1912: 327); Vaccaneo (1934: 446); Harms (1935: 293); Musselman & Visser (1989: 323); Bolin & Musselman (2013: 1); Bolin et al. (2018: 107). Type: Madagascar, SW Region. Linta and Menandra watersheds, 7 Dec. 1910, Perrier 8911 (holotype P!; isotype K!). Epitype (designated by Bolin & Musselman 2013): Madagascar, Tulear, grounds of hospital, 6 Jan. 1947, Humbert 19805 (epitype P!, isoepitypes G!, K!, MO!).

Herbaceous subterranean perennial root holoparasite, without leaves or stem. Rhizome angular (4 – 7-sided) in cross-section, 0.7 – 5 cm in diam.; rhizome surface coriaceous, dark brown to orange-brown, lighter coloured near growth tip (in life); rhizome spreads laterally, may occasionally bifurcate; rhizome ornamented with numerous lateral tubercles arranged in distinct parallel lines, with these lines defining the number of angles of the rhizome in cross-section, tubercles remain the same or develop into a flower or develop into a haustorium; rhizome fleshy, internally coloured light pink to red (in life). Flower (3 – (4 – 5) – 6)-merous; flower emerges only partially from the soil; two floral chambers, androecial chamber subtended by gynoecial chamber, inner surfaces of chambers glabrous. Perianth external tissues coriaceous, brown to purplish-brown, total length 11.6 – 26.9 cm; perianth tissues fleshy; perianth tube whitish inside, fading to brown-black at the tepals; tepals whitish-cream, darkening to brown with age, tepal (measured from apex to point of connation with adjacent tepal) 4.3 – 7.4 × 1.2 – 3.9 cm (measured at base); tepals elongate-lanceolate, often strongly recurved towards the apex, not fused at apex; tepal margin with numerous setae, c. 3 mm in diam., transitioning into a distinctly ribbed surface. Osmophore apical, not recessed, often split down the middle such that it is divided into two triangles barely fused at the apex, spongy and white to tan (in life), darkening to yellow when dried, generating musty, foetid odour, but not one of rotting flesh. Androecium antheral ring formed by connate w-shaped anther lobes, forming a central orifice, transversely striate and divided into numerous horizontal pollen sacs, pollen bisulcate. Gynoecium ovary inferior, unilocular with numerous ovules, ovary 2.1 – 4.6 cm wide; lobed and cushion-like stigma on the floor of the gynoecial chamber, stigma 1 – 3 cm wide, fleshy pedicel sometimes present, 0 – 5 cm long. Fruit partially subterranean turbinate berry, 7 – 15 cm in diam. Seed spherical, black-brown, 0.7 – 1.2 mm, thousands of seeds within fruit, embedded within white pulp. Fig. 10.

Fig. 10.

Hydnora esculenta. A illustration of flower and rhizome; B example of herbarium specimen flower and rhizome (both Humbert 19805 [P]); C – E live flowers. A illustrated by sebastian hatt.

distribution. Endemic to southern Madagascar, specifically the Atsimo-Andrefana, Anosy and Androy regions. There exists a single report of its occurrence on Réunion Island, although this was later thought to be a short-lived population attached to an introduced host species, Pithecellobium dulce (Roxb.) Benth. (Fabaceae) (Bosser 1994). While there have been no reports of Hydnora from the island since, despite the efforts of Baillon to search for it, it is possible it still grows there and has not been formally recorded (Bosser 1994). The initial description indicates it was recorded growing in abundance in two separate towns on the island. Although the description provided makes it impossible to confirm whether or not this population was H. esculenta, it is perhaps likely given the geographic position of the island relative to Madagascar, and that it was reported on a known host of H. esculenta. It is possible that the range of H. esculenta is increasing given its ability to parasitise the now widely distributed and planted non-native host Pithecellobium dulce, although this has not been studied and is mere speculation at present (Bolin & Musselman 2013). Map 6.

Map 6.

Distribution range of Hydnora esculenta.

specimens examined. madagascar. Toliara: NW of Amboasary, Berenty Reserve [25°01'S, 46°19'E], 23 Dec. 1987, Phillipson 2716 (MO!, P!, TAN!); Berenty Reserve, Ankoba Forest [25°00.053"S, 46°17.834"E], 7 Dec. 2007, Bolin & Razafindraibe 71 (TAN!); Anosy Region, Parc Andohahela-Tsimelahy [24°56.186"S, 46°38.546"E], 10 Dec. 2007, Bolin & Razafindraibe 72 (TAN!); Stream bank in Ampanihy [24°41.808"S, 44°44.441"E], 15 Dec. 2007, Bolin & Razafindraibe 73 (TAN!); Tuléar, no date, Richey 2 (P!); Tuléar, Sudwestliches kustengebiet, 26 Dec. 1959, Schlieben 8260 (B!, BM!, BR!, G!, K!, M!); Tuléar, cours de l’Hopital, 1 Jan. 1947, Humbert 19805 (epitype P!, isoepitypes G!, K!, MO!); near hospital, 1956, Abbays s.n. (TAN!); Sud de Madagascar, 7 Dec. 1910, Perrier 8911 (holotype P!, isotype K!).

habitat. Details of habitat preference for Hydnora esculenta are scarce. The region of Madagascar in which it grows is sub-arid, dominated by spiny thicket and occasional woodland cover. Its distribution even edges into the transitional-rainforest areas in the south-east of Madagascar (Bolin & Musselman 2013).

conservation status. Hydnora esculenta has been described in 2013 as uncommon, but locally abundant in the areas in which it occurred (Bolin & Musselman 2013). The conservation status of this species remains unclear and largely unstudied. However, there are notable potential threats such as widespread deforestation, with a 45% loss in woodland from 1973 to 2013, particularly in remote regions such as the Tulear province due to ‘clandestine pioneer-agriculture’ (Brinkmann et al. 2014). While there are a few protected areas in southern Madagascar known to be home to H. esculenta such as the Berenty Reserve, much of the spiny forest habitat in the region remains unprotected and at risk of destruction over the coming years. Furthermore, increased browsing by small livestock has already been noticed by local Tsimelahy villagers as having depleted the Hydnora populations (Bolin & Musselman 2013). As such, the conservation status of H. esculenta is perhaps the most concerning within the genus and warrants more attention.

phenology. Flowers primarily from November to January, with fruiting from April to May (Bolin & Musselman 2013).

etymology. The specific epithet ‘esculenta’ is latin for ‘edible’. This has roots in the Malagasy name for Hydnora, which is voantany. This can be translated and divided into voa, meaning ‘fruit’, and tany, meaning ‘earth’.

vernacular names. Hydnora is known as voantany in Malagasy, translating into ‘fruit of the earth’ (Bolin & Musselman 2013).

ethnobotany. The fruit is known to be something of a delicacy in the southern regions, apparently delectable with a scent reminiscent of a rennet apple (Jumelle & Perrier 1912). The rhizomes are sold at local drug markets and have also been recorded as being used by the local semi-nomadic ‘Vezo’ people (fishermen, or those that live from sea fishing) of southern Madagascar in the composition of the ‘mohara’ or ‘mandemilahy’ (Jumelle & Perrier 1912). This appears to be a fetish primarily made from cow’s horn that is used, likely among other things, to help cure rheumatism (Boiteau 1997).

host specificity. The only confirmed hosts belong to the Fabaceae (Table 1). Confirmed hosts include the native Albizia tulearensis R.Vig. and Alantsilodendron decaryanum (R.Vig.) Villiers, whose distributions are a close match for H. esculenta, and also on the non-native introduced Pithecellobium dulce (Roxb.) Benth., whose range is now almost global. Curiously, Bolin & Musselman (2013) reported that P. dulce was the most commonly encountered host in the region. There are historic reports of Casuarina sp. (Casuarinaceae) as a host (Schlieben 8260 [K!]), although this was likely a mistake as this is not in the Fabaceae. Tamarindus indica has also been a recorded host in the past, although this was in 1947 [Humbert 19805 (P!)] and has not been observed since. However, T. indica is within the Fabaceae, and is a recorded host on H. abyssinica, so this association is entirely plausible. There is a single report of H. esculenta growing on P. dulce and Albizia lebbeck on Ile de Réunion. Both these trees are non-native, so it is suspected that Hydnora was transferred to the island accidentally on the roots of these introduced trees (Bosser 1994).

notes. Species description adapted from Bolin & Musselman (2013). Hydnora esculenta is the only species found on Madagascar, and the only species belonging to subgenus Neohydnora. It is distinguished from all other species by its unique combination of an angular rhizome with tubercles in parallel lines, together with apical, but not recessed osmophores. This species was described in 1912, although the type specimen is little more than a shattered fragment of fruit. Therefore, it was epitypified and re-described by Bolin & Musselman in 2013.

III) Hydnora subgenus Dorhyna Decne. (Decaisne 1873: 75) Harms (1935: 291). Type: Hydnora abyssinica A.Braun.

Rhizome irregular, often both terete, subterete and compressed at different points, rhizome ornamented with numerous lateral tubercles of variable size and shape, randomly and densely distributed on the surface, may occasionally form what appear to be lines. Flower usually 4-merous (3 and 5 also observed several times); pedicel absent or present up to 2.5 cm long. Perianth external tissues coriaceous, brown to reddish-brown; tepal margins with setae; sometimes glabrous outer cucullus present dorso-apical to the apical osmophore and tepal margin. Osmophore apical always present but may be reduced; recessed osmophore usually present to at least some extent.

6. Hydnora abyssinica A.Braun (Braun 1867: 217); Decaisne (1873: 76); Martelli (1886: 70); Engler (1895: 169); Rendle (1896: 55); Engler (1900: 386); Baker & Wright in Thistleton-Dyer (1909: 133); Musselman (1997: 18); Beentje & Luke (2002: 1); Bolin et al. (2018: 106). Type: Ethiopia, mountains near Dehli-Dikeno, 1853, Schimper 963 (syntypes B [not seen, may be destroyed?], P!).

Hydnora abyssinica var. quinquefida Engl. (Engler 1900: 386); Baker & Wright in Thistleton-Dyer (1909: 134). Type: Tanzania, Uhehe, Lukosse river, before 1900, Goetze 487 (holotype B [not seen, may be destroyed?]).

Hydnora aethiopica Decne. (Decaisne 1873: 77); Solms-Laubach (1901: 6); Baker & Wright in Thistleton-Dyer (1909: 132); (Vaccaneo 1934: 441). Type: Sudan, voyage aux sources du Nil Blanc, before 1873, Sabatier s.n. (holotype P).

Hydnora angolensis Decne. (Decaisne 1873: 76); Solms-Laubach (1901: 7); Baker & Wright in Thistleton-Dyer (1909: 134); (Vaccaneo 1934: 444). Type: Angola, ad oram angolensem, before 1873, collector unknown s.n. (P [not seen, may be missing?]).

Hydnora bogosensis Becc. (Beccari 1871: 6); Martelli (1886: 70); Rendle (1896: 55); Solms-Laubach (1901: 7); Baker & Wright in Thistleton-Dyer (1909: 134); Chiovenda (1916: 156); Vaccaneo (1934: 432); Harms (1935: 291); Cufodontis (1972: 36). Type: Eritrea, Keren, July – Aug. 1870, Beccari s.n. (holotype FT).

Hydnora gigantea Chiov. (Chiovenda 1916: 156); Harms (1935: 291). Type: Somalia, Berdale to El Ualac, c. 1911, Paoli 980, 1035 (both syntypes FT [not seen, may be missing?]).

Hydnora gigantea var. trimera Chiov. (Chiovenda 1916: 157); Harms (1935: 291); Cufodontis (1972: 37). Type: as for Hydnora gigantea Chiov.

Hydnora johannis Becc. (Beccari 1871: 5); Solms-Laubach (1901: 7); Chiovenda (1916: 156); Vaccaneo (1934: 427); Harms (1935: 291); Peter (1932: 185); Lebrun (1948: 393); Cufodontis (1972: 37); Troupin (1978: 289); Malaisse (1982: 115); Musselman (1984: 23); Musselman & Visser (1987: 81); Musselman & Visser (1989: 323); Musselman (1993: 31). Type: Eritrea, Keren-Bogos, 1870, Beccari 170 (holotype FT!).

Hydnora johannis var. gigantea (Chiov.) Vaccaneo (1934: 448); Cufodontis (1972: 37). Type: as for Hydnora gigantea Chiov.

Hydnora johannis var. quinquefida (Engl.) Solms (Solms-Laubach 1901: 7); Vaccaneo (1934: 430); Harms (1935: 291); Peter (1932: 185). Type: as for Hydnora abyssinica var. quinquefida Engl.

Hydnora michaelis Peter (1932: 185). Type: Tanzania, Usagara, Kidete, Pori on Romuma, before 1932, Peter 32690 (holotype B!).

Hydnora ruspolii Chiov. (Chiovenda 1917: 57); Vaccaneo (1934: 447); Harms (1935: 291); Cufodontis (1972: 37). Type: Somalia, Marro Umberto I ville, April 1893, Ruspoli & Riva 1091 (syntypes F!, G!, RO!).

Herbaceous subterranean perennial root holoparasite, without leaves or stem. Rhizome irregular, often both terete, subterete and compressed at different points, 5 – 8 (– 10) cm in diam.; rhizome surface coriaceous, dark brown, lighter-coloured (when fresh) near growth tip; rhizome spreads laterally and often bi- or trifurcating or branching irregularly to form smaller terminal branches; rhizome ornamented with numerous lateral tubercles of variable size and shape, randomly and densely distributed on the surface, not in clear consistent lines (although may occasionally form what appear to be lines), tubercles remain the same or develop into a flower or develop into a haustorium; numerous flowers and flower buds on single rhizomes; rhizome fleshy, brick red to reddish-pink internal tissue when broken, producing sticky, astringent exudate, lighter colours at growing tip (in life). Flower usually 4-merous (3 and 5 also observed several times); flower emerges only partially from soil; two floral chambers, androecial chamber subtended by gynoecial chamber, inner surfaces of chambers glabrous, pedicel absent or present up to 2.5 cm long. Perianth external tissues coriaceous, brown to reddish-brown, often with thin fleshy white streaks where the outer layer has cracked, total length (5 –) 11 – 25 (– 33) cm; perianth tissues fleshy; perianth tube and tepal cavity pinkish to ivory white inside (in life), darkening to brick red-brown (dried), 3 – 5.5 cm wide; tepal margin reddish-pink to pink-orange inside (in life), darkening to brick-red brown, tepal (measured from apex to point of connation with adjacent tepal) (4 –) 6 – 9 (– 14) × 1.7 – 3.8 cm (measured at midpoint); tepals lanceolate to spathulate, initially erect, gently curved and connivent, later spread out flat on ground, this change hastened by moist weather but usually inevitable with age; tepal margins covered in diffuse setae, longest on dorsal side, often entirely absent on ventral side, from 1 – 7 mm, generally quite variable between individuals. Osmophore apical osmophore present, border between osmophore and hairy tepal margin sloping down basally in a sharp diagonal from the dorsal side to the ventral side; recessed osmophore present but barely so, usually extending no more than 1 cm down the length of the tepal margin; spongy, cream to white (in life), 2.6 – 5 × 1.6 – 2 cm, darkening to tan when dried, generating very foetid odour. Androecium antheral ring formed by connate w-shaped anther lobes, forming a central orifice, transversely striate and divided into numerous horizontal pollen sacs. Gynoecium ovary inferior, unilocular with numerous ovules produced from apical placenta, ovary 2.2 – 2.8 (– 4.5) cm wide; lobed and cushion-like stigma on the floor of the gynoecial chamber. Fruit subterranean, globose, 10 – 15 cm in diam., leathery brown pericarp. Seed spherical, black-brown 1 – 1.8 mm, thousands of seeds embedded within mealy white pulp. Fig. 11.

Fig. 11.

Hydnora abyssinica. A illustration of flower and rhizome; B – E examples of herbarium specimens: B flower (Luke 3016 [K]); C osmophore (Greenway & Kirrika 11132 [K]); D osmophore and diffuse tepal setae (Gillett 14151 [K]); E rhizome (Jameson & Michelmore 134 [K]); F – H live flowers. A illustrated by sebastian hatt. photos: f & h mathew rees; g william burger, slide no. 4174 [K].

distribution. Hydnora abyssinica has a considerably larger distribution than all the other species. It has been recorded in northern South Africa, Eswatini, Mozambique, Tanzania, Kenya, Uganda, the DRC, Rwanda, Ethiopia, Somalia, Eritrea and Nigeria. There is a single report from 1873 of a section of rhizome collected from Gabon, although this material has since been lost (Decaisne 1873). However, given the recent discovery of H. abyssinica in Nigeria, far outside the previously accepted range, it is possible that Hydnora may be found in Gabon or surrounding countries such as the Central African Republic or Chad (Agyeno et al. 2018). Map 7.

Map 7.

Distribution range of Hydnora abyssinica.

specimens examined. nigeria. LGA Plateau State: Nekong Kanke, c. 2018, Gosomji s.n. (JUHN!). eritrea. Keren, July 1870, Beccari s.n. (holotype FT); Abita, Keren, July 1870, Beccari 170 (holotype FT!); Barka Valley, Keren-Bogos, Sciotel?, c. 1870, Beccari 417, 423, 424, 3685, 3687, 3690 (FT [not seen, may be missing?]); Sciotel, c. 1870, Martelli s.n. (FT!); Beni-Amer, Monte Damba, 25 Sept. 1903, Pappi 6110 (FT!); Habab, Melchet-Tzaroba, 18 April 1909, Pappi 8222 (FT!). somalia. Somaliland: Meid, April 1875, Hildebrandt s.n. (BM!, BM!, K!, L!); W of Gacan Libaax camp on road, 17 May 2017, Awale AIA20 (HARG!); Hargeisa, 26 Sept. 1932, Gillett 4071 (K!, FT!); Northern rangelands [9°58'N, 46°12'E], 11 March 1981, Beckett 929 (EA!); Hahi, June 1885, James & Thrapp s.n. (K!); Harrar, da Gildessa a Zeila, 1889, Robecchi-Bricchetti 184 (FT!); Northeast Region: Sultanato di Obbia, 26 April 1924, Puccioni & Stefanini 443 (FT!); Central Region: Marro Umberto I ville, April 1893, Ruspoli & Riva 1091 (syntypes FT!, G!, RO!); Harli Bomani, April 1893, Ruspoli & Riva s.n. (G!); South Region: Berdale to El Ualac, c. 1911, Paoli 980 (syntype FT [not seen, may be missing?]); Berdale to El Ualac, c. 1911, Paoli 1035 (syntype FT [not seen, may be missing?]). ethiopia. Oromia Region: Harar, slopes W of Midaga above Gobelli River Valley [42°4'E, 8°46'N], 9 May 1964, Burger 3496 (K!, US!); Pozzi di el Banno, 2 May 1939, Corradi 2866 (FT!); Road Ginir, 119 km E of Ginir, 17 Nov. 2015, Friis, Abebe & Getachew 15708 (K!); Sidamo, 20 km S of Neghelle, 18 May 1982, Friis, Tadesse & Vollesen 3101 (K!); Sidamo, El Siro water holes, 22 May 1983, Gilbert & Ensermu 7727 (K!); Kilil, Lake Margerhita, 5 April 1958, Eriksson 645 (S [not seen, identity not confirmed]); Amhara Region: Gondaraba nei dintorni del fortino, 28 May 1939, Corradi 2862 (FT!); mountains near Dehli-Dikeno, 1853, Schimper 963 (syntypes B [not seen, may be destroyed], P!); Dehli-Dikeno, near Dschadscha, Agau, 19 July 1853, Schimper 1189 (P!); Tigray Region: Amhara-Uolcait, 5 Feb. 1913, Pappi 9007 (FT!); SNNP Region: Balkakà-Neldà, 1893, Ruspoli & Riva 1623 (FT!); Dire Dawa Region: 20 km from Haresa, 26 April 1969, Ebba 734 (K!); 27 km NE of Dire Dawa on road to Djibouti, 15 April 1972, Gilbert 2390 (EA!, K [appears to be missing]); Somali Region: 22 km from Qarsonney, 14 May 2006, Thulin et al. 11140 (UPS); Unlocalised: likely 1854, Schimper s.n. (P! [P05591526]); 4 April 1854, Schimper s.n. (P! [P05591525]). kenya. K1: Balambala, no date, Adamson 450 or Bally 6151 (G!, K!, S); a few miles S of Bura, 3 Jan. 1943, Bally 2050 (K!, EA!); Mandera, 30 km on Ramu-Malka Mari road [4°4'N, 40°59'E], 6 May 1978, Gilbert & Thulin 1534 (EA!, K!, UPS); Moyale, 15 km out on Wajir road, 6 Nov. 1952, Gillett 14151 (EA!, K!); Kilifi, Sala E [3914E, 0306N], 28 Dec. 1991, Luke 3016 (EA!, K!); Mandéra, Oct. 1889, Sacleux 965 (P); K2: Munyen, Turkana, March 1947, Bally 5042 (K!); Wei Wei, Katuw, 23 July 1978, Meyerhoff 71M (K!); K3: Baringo Distr., Lake Hannington E shore, 28 June 1972, Bally 15177 (EA!); Baringo Distr., Chemolingot borehole area [0°58'N, 35°57'E], 6 Sept. 1976, Timberlake 712 (EA!); Rumarute, 1 Nov. 1937, Searle s.n. (K!); K4: Nairobi, Feb. 1964, Beecher s.n. (EA!, K!); Nairobi National Park, just outside Cheetah Gate [1°26'S, 36°58'E], 26 April 1975, Friis & Hansen 2605 (K!); Machatos Distr., Kanza, 14 Nov. 1982, Howard 16717 (EA!); K6: Kampi ya Bibi [1°33'S, 36°33'E], 25 April 1965, Archer 13150 (EA!); Ngong Hills, Nov. 1938, Bally 12514 (K!); Nairobi-Magadi Rd, prehistoric site, 13 Dec. 1968, Early 14080 (EA!); Chyulu Hills, 18 Nov. 1972, East African Herbarium 15231 (EA!); Kajaido Distr., Chyulu Hills [2°31'S, 37°45'E], 18 Oct. 1969, Gillett & Kariuki 18797 (EA!, K!); Chyulu Hills, Mbirikani-Nbi [02320S, 3731E], 12 Dec. 1991, Luke 2963 (EA!, K!); Chyulu Hills, below Bonham House [0232S, 3744E], 22 Dec. 2000, Luke & Luke 7182 (EA!, K!, MO, US!); Olorgasailie Prehistoric Site, 28 Dec. 1958, Napper 1220 (EA!, K!); Loita Hills [0140S, 3555E], 26 April 1948, Vesey Fitzgerald 6359 (K!); K7: no locality, Feb. 1977, Adamson 16079 (EA!); Voi, Dec. 1962, de Jong 362/CL (EA!); Mackinnon road, 14 Nov. 1950, Lt. Col. Venour 114/50 (EA!); Mt Kisigau, Nov. 1938, Joanna 8635 (EA!, K! [both just rhizomes, identity impossible to confirm]). uganda. U1: N. Karamoja, near Kaabong, June 1945, Dale 416 (EA!); Lira, Lango, Dec. 1929, Davies 1038 (K!); Amudat, Karamoja, May 1948, Eggeling 5812 (K!); 30 – 50 miles N of Kacheliba–Karamoja, 9 May 1953, Padwa 106 (EA!, K!); Lochoi, Karamoja, 24 May 1940, Thomas 3539 (K!); U2: Kijumbura, Victoria Nile, 31 Jan. 1907, Bagshawe 1559 (BM!); Kasumba, Ankole, 9 April 1953, Jameson & Michelmore 134 (K!); Ruizi R., 24 Feb. 1951, Jarrett 444 (EA!, K!); U4: Sekamuli, near Bamunanika [0°41'N, 32°36'E], 13 June 1967, Ferreira 133 (EA!, K!, MO); Unlocalised: 28 April 1997, Badaza 64808 & 64809 (K!). rwanda. Gabiro, no date, Hoier s.n. (BR); Kibugabuga [Bugesera], 17 Nov. 1953, Liben 953 (BR); Akagera National Park [1.5088°S, 30.6814°E], 9 June 2018, Luke 18431 (EA!). tanzania. T1: Mwanza Distr., Ukiriguru, 7 Dec. 1970, Ebbela 71/1 (EA!); Musoma, below Sid Downies Dam, 30 Dec. 1962, Greenway & Harvey 10917 (EA!, K!); Mwanga, Ukiriguru, 15 Jan. 1948, Prentice 1298 (EA!); Benagi Hill, Kikima, 24 March 1961, Pullman 12381 (EA!); Bukoba Distr., Nov. 1983, Kanywa 84 (K! [just rhizome, impossible to confirm identity]); T2: Kilimanjaro, N of Loloari Hill, 26 Nov. 1967, Bigger 1442 (EA!); Lake Manyara National Park, near Mto wa Mbu R., 6 Jan. 1962, Dingle HD1002 (EA!); Arusha, Ngorongoro Conservation Area [3°20'S, 35°2'E], 15 Oct. 1993, Ellemann 654 (AAU, MO!); Mbuka Distr., Tamamgwe R., 3 Jan. 1959, Mahinde 415 (EA!); T3: Msasa, Lake Manyara National Park, 4 Dec. 1963, Greenway & Kirrika 11132 (EA!, K!); Usambara Mts, May 1998, Schlage CS265 (B!); Muhesa, 4 April 1924, Shantz 198 (K!); T5: Kongogo, 8 Jan. 1966, Newman 88 (EA!); Landschaft Ussagara, Kidete, Dec. 1925, Peter 32690 (holotype B!); Kondoa Distr., Kondoa, 11 Jan. 1962, Polhill & Paulo 1139 (B!, BR!, K!); T6: between Uhehe and Khutu, Jan. 1899, Goetze 395 (B [not seen, may be destroyed]); T7: Uhehe, Lukosse R., no date, Goetze 487 (holotype B [not seen, may be destroyed]); Unlocalised: Serengeti Park, near a river, 2004, Vandervinne s.n. (photograph, BR!). malawi. Mchiru Mts, Nov. 1989, Jenkins 87 (K!). angola. Ad oram Angolensem, before 1873, collector unknown s.n. (holotype P [not seen, may be missing and may not be H. abyssinica). namibia. Hardap Region: Karibib, Farm Nuiskasaken, 24 July 1963, Häblich & Giess 7652 (M!). south africa. KwaZulu-Natal Prov.: Tembe Park [27°2'40"S, 32°25'6"E], 12 Dec. 2001, Vahmeijer HV00385 (PRE!). unlocalised. Voyage aux sources du Nil Blanc, before 1873, Sabatier s.n. (holotype P); no date, Bouxin 1257BIS (BR); no date, Musselman 198 (E).

habitat. Wide range of arid to sub-arid habitat types following the distribution of its generally Acacia host species. Recorded from dry woodland, wooded grassland, bushland, scrubland, often on black cotton, clay or sandy alluvial soils or rocky slopes (Musselman 1997; Beentje & Luke 2002).

conservation status. Hydnora abyssinica has a vast range across Africa, although the population density within that range is unclear. The species as a whole is unlikely to be threatened given the size of the distribution, although certain localities will be more at risk than others. The population in Nigeria is reported to be under considerable threat from over-harvesting and land development in the area (Otuwose Agyeno, pers. comm.). Similarly, in Uganda there are concerns about increasing awareness of its medicinal potential leading to over-harvesting (Nyafuono et al. 2000). The South African Red List records H. abyssinica as ‘Not Threatened’, although nearby populations in Mozambique are known to be harvested to a considerable degree (Williams et al. 2008; 2011b). It is difficult to comment with confidence on the threat harvesting causes, as the full extent of the populations are largely unknown. Furthermore, H. abyssinica may later prove to be composed of distinct species or subspecies, some of which may be under more threat than others.

phenology. Flowering time varies considerably across its vast range but tends to occur following the rains and lasts for a few days to a month (Beentje & Luke 2002). Fruits tend to mature around five months after flowering (Musselman 1997).

etymology. The specific epithet refers to the locality of the type specimen in Ethiopia (formerly known as Abyssinia).

vernacular names. In Somalia, it is known as likke, lécche, lipti or leka (Vaccaneo 1934) or laka (Gilbert & Thulin 1534 [EA!, K!, UPS]) or liki, like or dingah (Mandu et al. 1999) or likaha (Ebba 734 [K!]). In Gobato, Galla and Malakati, spoken in Ethiopia, it is known as tuka (Ebba 734 [K!]; Adamson 450 [G!, K!, S]). In East Africa, it has names in many languages. In Maasai or Maa it is known as orukunyi (Luke & Luke 7182 [EA!, K!, MO, US!]; Beentje & Luke 2002) or erukunyi or erkyunyi (Mandu et al. 1999). In Borana it is known as toga (Gilbert & Thulin 1534 [EA!, K!, UPS]; Mandu et al. 1999). In Burji it is known as guli (Mandu et al. 1999). In Kikuyu it is known as muthigira (Mandu et al. 1999). In Pokot it is known as aurieng’o or kaworiongo (Mandu et al. 1999). In Turkana it is known as auriong’o (Mandu et al. 1999). In Swahili it is known as kioga (Engler 1900) or nyambo or mnyambo (in Tanzania according to Goetze 395 [B]; Mandu et al. 1999). In Uganda, in Luganda, it is known as omutimagwensi or omutimagwettaka (Nyafuono et al. 2000) or mutima gwenai/gwensi which translates to ‘the heart of the earth/world’ (Bagshawe 1559 [BM!]; Kanywa 84 [K!]) or awuriongo (Thomas 3539 [K!]) or in Ankole, spoken in Uganda, it is known as otumimawasi (Jarrett 444 [EA!, K!]). In Tanzania, it is known also known as mulumbi (Goetze 487 [B]; Vaccaneo 1934) or amamaso (Newman 88 [EA!]). In Wasukuma, spoken in Tanzania, it is known as wija wa hasi, which translate to ‘half-blind person of the earth’, in reference to its ‘screwed-up peering look’ when it first breaches the ground surface (Prentice 1298 [EA!]). In Nigeria, the Ngas and Hausa speaking community know it as kurshim or kaushekasa (Agyeno et al. 2018). In southern Africa, in Xitsonga/Xichangana, it is known as mavumbule (Williams et al. 2011b). It is known as uMavumbaka in Zulu, meaning ‘the one that pops up’, a name that is shared with Sarcophyte sanguinea (Balanophoraceae) (Williams et al. 2011a).

ethnobotany. Hydnora abyssinica has a wide range of uses across its range. In Somalia, locals cut it into thin slices and heat it in water for a few hours, obtaining a strongly coloured liquid use to dye mats (Vaccaneo 1934). The fleshy rhizome and fruits are known to be eaten in Somalia (Ruspoli & Riva 1091 [FT!, G!, RO!]; Vaccaneo 1934). The fruit are reported to be cooked like a baked potato, then cut open with the seeds eaten with sugar and milk (Gillett 4017 [K!]). In Ethiopia, the fruit is reported to be eaten by local people (Burger 3496 [K!, US!]). In Kenya, the flower buds are reported to be eaten by the Pokot and Maasai (Gillett 14151 [EA!, K!]; Timberlake 712 [EA!]), with additional minor medicinal use during childbirth (Beentje & Luke 2002) or to treat post-partum haemorrhage (Searle 2575 [K!]). In northern Uganda it is used as an antidiarrheal agent, while in eastern and central Uganda, the root is used to treat cardiovascular disorders and diabetes (Nyafuono et al. 2000). One report indicates that in Uganda, if it is used to treat heart infections, as reported by Nyafuono et al. (2000), the sufferer who used this treatment must for-go goat hearts for the rest of their life (Bagshawe 1559 [BM!]). In Tanzania, it is used in the treatment of sore throat, swollen tonsils and throat inflammations (Schlage CS265 [B!]; Nyafuono et al. 2000). Furthermore, in Tanzania it is eaten with milk by herd boys, who also use the flower buds as a small cup to milk the cows in the daytime, although apparently this is strictly prohibited (Elleman 654 [AAU, MO!]). The usage of H. abyssinica in local medicine in Nigeria far pre-dates its discovery there in 2018. The plant is harvested at considerable rates for use by doctors in the treatment of diarrhoea, dysentery, stomach-ache, piles and erectile disfunction, with some locals reported using it to ‘rejuvenate the body from the physical and mental effects of the sun’ (Agyeno et al. 2018). It is sold at traditional medicine markets in South Africa, although harvested from Mozambique, in the treatment of bleeding, diarrhoea, acne, piles, menstrual problems, stomach cramps (Williams et al. 2011a; 2011b). H. abyssinica (it is unclear which part of the plant) has been reported being ‘dug out’ and consumed by wild pigs and warthogs in East Africa (Bally 12514 [K!]; Bally 5042 [K!]; Beentje & Luke 2002). Beyond uses, H. abyssinica has been reported causing damage to local streets, driveways and patios in Ethiopia, where its rhizomes have burst through the road or tile surface (Maass & Musselman 2001).

host specificity. Hydnora abyssinica appears to be restricted to Fabaceae hosts, mostly Senegalia and Vachellia (Acacia), but also a few other trees from different genera in the same family (Table 1). The following hosts have been reported: Albizia lebbeck (L.) Benth., Delonix regia (Bojer ex Hook.) Raf., Piliostigma thonningii (Schumach.) Milne-Redh., Senegalia asak (Forssk.) Kyal. & Boatwr. [Acacia glaucophylla], Senegalia mellifera (Benth.) Seigler & Ebinger [Acacia mellifera], Senegalia nigrescens (Oliv.) P.J.H.Hurter [Acacia nigrescens], Tamarindus indica L., Vachellia amythethophylla (Steud. ex A.Rich.) Kyal. & Boatwr. [Acacia macrothyrsa], Vachellia drepanolobium (Harms ex Y.Sjöstedt) P.J.H.Hurter [Acacia drepanolobium], Vachellia elatior (Brenan) Kyal. & Boatwr. [Acacia elatior], Vachellia gerrardii (Benth.) P.J.H.Hurter [Acacia gerrardii], Vachellia grandicornuta (Gerstner) Seigler & Ebinger [Acacia grandicornuta], Vachellia hockii (De Wild.) Seigler & Ebinger [Acacia hockii], Vachellia nilotica subsp. tomentosa (Benth.) Kyal. & Boatwr. [Acacia arabica], Vachellia tortilis (Forssk.) Galasso & Banfi [Acacia tortilis], Vachellia xanthophloea (Benth.) Banfi & Galasso [Acacia xanthophloea]. An entry from iNaturalist cited a Prosopsis sp. (Fabaceae) as the host [bryanadkins 13223800]. Given the rhizome appeared excavated in the image, this could indeed be the host. There are a handful of other host species that have been recorded in the past on old herbarium specimens, although the validity of these is doubtful given the ease with which the host species can be mistaken. As they belong to completely different families and have only been reported once, they are not considered potential hosts here. These erroneous hosts are Adansonia digitata (Malvaceae), Dobera glabra (Salvadoraceae), Hildebrandtia sp. (Convolvulaceae), Kigelia africana (Bignoniaceae) and Terminalia sericea Burch. ex DC. (Combretaceae).

notes. Species description adapted from Beentje & Luke (2002). Hydnora abyssinica is the most widespread and variable species of Hydnora, hence the numerous synonyms described since its initial publication in 1867. H. abyssinica was previously known as H. johannis Becc., until Beentje & Luke (2002) noticed that an earlier description for H. abyssinica was indeed valid. As such, many specimens around the world are still labelled as H. johannis. The various synonyms have been studied to as much detail as possible in this monograph, despite the loss of many type specimens and obscurity of many of the protologues. None of them appeared to be sufficiently distinct from H. abyssinica besides H. solmsiana, which has been reinstated here. Further field collections in under-studied areas may well provide new specimens that may lead to H. abyssinica being divided further.

Hydnora abyssinica is also similar to H. hanningtonii, but is distinguished by having a proportionately smaller apical osmophore that only occupies considerably less than ½ the tepal. H. hanningtonii has dense strigose setae that cover the entire tepal margin and a red perianth tube and tepal cavity, while H. abyssinica has longer setae, that rarely reach the ventral side of the tepal margin, and a pinkish-white perianth tube and tepal cavity.

7. Hydnora hanningtonii Rendle (1896: 55); Solms-Laubach (1901: 7); Baker & Wright in Thistleton-Dyer (1909: 133); Chiovenda (1916: 156); Vaccaneo (1934: 448); Harms (1935: 291); Cufodontis (1972: 37). Type: Tanzania, Jordan’s Nullah, Lake Victoria, E arm of Mwanza Gulf, Dec. 1882, Hannington s.n. (syntype BM!) & Ethiopia, Ginia, Galla highlands, 1895, Donaldson Smith s.n. (syntype BM [not seen, likely missing]).

Hydnora arabica Bolin & Musselman (in Bolin et al. 2018: 1). Type: Oman, Dhofar Region, Ayn Ayuoon, 0.66 km SE of the spring [17.241294, 53.891644], 16 Dec. 2014, Bolin, Rahbi & Musselman JFB2014OM3 (holotype OBG!, isotype US!).

Hydnora cornii Vaccaneo (1934: 414); Harms (1935: 291); Cufodontis (1972: 37). Type: Somalia, Mogadishu, Villa Governationale di Afgoi, Corni s.n. (holotype TOM).

Herbaceous subterranean perennial root holoparasite, without leaves or stem. Rhizome terete, subterete or compressed, 1 – 5 cm diam., may be much larger on mature specimens (none found); rhizome surface coriaceous, dark brown to light tan, lighter coloured (when fresh) near growth tip; rhizome spreads laterally and may bifurcate or branch irregularly; rhizome ornamented with numerous lateral tubercles, randomly and densely distributed on the surface, not in clear consistent lines, tubercles remain the same or develop into a flower or develop into an haustorium; numerous flowers and flower buds on single rhizomes; rhizome swollen and irregular at haustorial interface with host root; rhizome fleshy, pink to red internal tissue when broken, lighter colours at growing tip (in life). Flower usually 4-merous, (though 3 and 5 observed); flower emerges only partially from soil; two floral chambers, androecial chamber subtended by gynoecial chamber, inner surfaces of chambers glabrous. Perianth external tissues coriaceous, brown to reddish-brown, total length 10 – 28 cm; perianth tissues fleshy; perianth tube red inside (in life), darkening to brick red-brown, 1.9 – 3.6 cm wide; tepal margins red inside (in life), darkening to brick red-brown, tepal (measured from apex to point of connation with adjacent tepal) 4.0 – 8.5 × 1.9 – 3.5 cm (measured at midpoint) (this on 4-merous flowers, wider on 3-merous flowers and narrower on 5-merous flowers); tepals clavate to elongate-linear, typically curved, not fused at apex; tepal margins always with dense strigose setae covering most or all of the tepal margin from ventral to dorsal edge, max. 3.3 mm. Osmophore apical and very prominent, greater than or equal to ½ length of tepal, spongy and white (in life), darkening to tan when dried, generating foetid odour, described as a strong ammoniac smell. Androecium antheral ring formed by connate w-shaped anther lobes, forming a central orifice, transversely striate and divided into numerous horizontal pollen sacs. Gynoecium ovary inferior, unilocular with numerous ovules, ovary 2.5 – 4.5 cm wide; lobed and cushion-like stigma on the floor of the gynoecial chamber, stigma 1.9 – 2.2 cm wide. Fruit only one observed, subterranean, globose, 7 cm in diam. Seed spherical, black-brown, 0.7 – 1.2 mm, thousands of seeds within fruit. Fig. 12.

Fig. 12.

Hydnora hanningtonii. A illustration of flower and rhizome; B – E examples of herbarium specimens (all Thesiger s.n. [BM]): B flower; C, D strigose tepal setae; E rhizome; F – H live flowers. A illustrated by sebastian hatt.

distribution. Known from the southern Arabian Peninsula (recorded from southwestern Saudi Arabia, southern Oman [Dhofar region], and Yemen). Also recorded from East Africa (Somalia, Tanzania and Kenya). Collected from an elevation of 200 – 1900 m. Map 8.

Map 8.

Distribution range of Hydnora hanningtonii.

specimens examined. saudi arabia. Najran Region: ~42 km NW of Najran, 30 April 1979, Collenette 1485 (K!). oman. Dhofar Region: 50 km W of Mudhai, 13 Sept. 1985, Miller 7619 (E!, ODU!); Wadi Haluf, SE of Rub el Khali, 8 Nov. 1945, Thesiger s.n. (BM!); Umshadid, 31 Oct. 1946, Thesiger s.n. (BM!); Ayn Ayuoon, 0.66 km SE of the spring [17°14'28.66"N, 53°53'29.92"E], 16 Dec. 2014, Bolin, Rahbi & Musselman JFB2014OM3 (holotype OBG!, isotype US!); ~ 2 km S of Haluf, Wadi Haluf [17°20'3.56"N 53°57'35.82"E], 16 Dec. 2014, Bolin, Rahbi & Musselman JFB2014OM2 (OBG!); ~17 km NE of Mirbat, Wadi Ayn [17°7'2.64"N 54°47'53.21"E], 15 Dec. 2014, Bolin, Lupton, Musselman & Rahbi JFB2014OM1 (OBG!); ~16 km NE of Mirbat, 17 Dec. 2014, Bolin, Lupton, Musselman & Rahbi JFB2014OM4 (OBG!). yemen. Hadramawt, Wadi Masila, 14 April 1947, Thesiger s.n. (BM!). somalia. Central Region: Mogadishu, Villa Governationale di Afgoi, before 1932, Corni s.n. (holotype TOM); Afgoi, 6 km from Mogadishu, 13 Aug. 1959, Moggi & Bavazzano s.n. (FT!). kenya. K7: Taita Taveta Distr., Ndara Ranch to Kajeri [0330S, 3937E], 26 Dec. 1991, Robertson 6537 (EA!, K!). tanzania. T1: Jordan’s Nullah, Lake Victoria, E arm of Mwanza Gulf, Dec. 1882, Hannington s.n. (syntype BM!); T3: near Gonja, Jan. 1950, Bally 7694 (K!).

habitat. In the southern Arabian Peninsula, it is typically found in stony and sandy wadi beds. It has been observed growing in sand. The type locality of Hydnora arabica is mountainous with strongly dissected rocky plateaus and loamy-skeletal to sandy-skeletal shallow soils underlain by Tertiary, Hadhramaut group formations: beige bioclastic, calcarenitic and micritic limestone with abundant fossilised molluscs (Bolin et al. 2018). In East Africa, it also appears to have a degree of preference for sandy soils, as in (Robertson 6537 [EA!, K!]). In Somalia, it has been reported from a sand dune (Robecchi-Bricchetti 184 [FT!]). H. cornii was described growing in an Albizia grove in the garden of the Consulate’s villa, though it is unclear how sandy the soil is there. The paucity of collections and infrequent sightings of this species mean that our understanding of its habitat or ecological requirements are limited.

conservation status. Fieldwork in Dhofar, Oman by Bolin et al. (2018) has indicated that populations of both Hydnora hanningtonii and its host Acacia tortilis (Fabaceae) are abundant in the wadi beds, despite being difficult to locate due its hypogeous habit. They suggest that the conservation status of H. hanningtonii in southern Oman is secure. Details of its abundance and distribution in Yemen are unclear. The conservation status in East Africa is unknown, although the ability of the plant to grow on artificially planted, introduced species, as seen in Afgoi where it grows on an Albizia grove in the Consulate’s villa garden (Vaccaneo 1934), is encouraging as this suggests it can survive disruption to the natural habitat so long as a suitable host is present.

phenology. Observed flowering in southern Oman in December, but local villagers report that flowering also occurs in other months (Bolin et al. 2018). At some sites, flowering was reported in July and at others in May. In East Africa, flowering has also been reported around December and January, although in Somalia it has been reported flowering in April. The apparently sporadic flowering may be dependent on adequate rainfall.

etymology. The specific epithet refers to the collector of the type, Bishop Hannington, whom Rendle apparently named this species after.

vernacular names. Known in Oman as dhanuna, xamleg (Jibbali), khamlayyeh, khumla’ah (Dhofari) (Bolin et al. 2018). Known in Yemen as nabeekh, fateekh and tarateef (Lawdar and Dathina) (Al-Fatimi et al. 2016). Vernacular names in East Africa for this species are likely the same as Hydnora abyssinica (see Vernacular Names), as there is currently no evidence that local people distinguish between these two species.

ethnobotany. In Oman, Jibbali settlers in Dhofar use all parts of the plant, and typically collect it following a rainstorm (Miller & Morris 1988). The flower heads are eaten, which smell of strong ammoniac and taste like very green cheese. However, the fruit is the more desired part for consumption. The matured fruit is apparently very sweet and is enjoyed by both humans and foxes. The dried remnants of the plant are used by locals as an indicator of where to dig for it the next time the rains fall. The rhizomes are crushed to a rough paste, which has been used to treat odorous or crusted leather. It has also been used to strengthen cotton material in fishing nets and ropes. Both the paste and dried out remnants have been used as a black ‘antimony’ powder which can serve as a dye or as eye shadow. In South Yemen, in the Lawdar and Dathina districts of Abyan, local people collect the flowers after the rains (Al-Fatimi et al. 2016). They eat the flowers, fresh or grilled, and use them in medicinal therapy to cure various stomach diseases, gastric ulcers and cancers. The dried flowers are ground down and applied with milk or water to the afflicted area. The rhizomes and fruits are not known to be used in Yemen. In Kenya it has been recorded being dug up by rhinoceroses (Bally 7694 [K!]). Uses in East Africa for this species are likely the same as Hydnora abyssinica (see Ethnobotany entry), as there is currently no evidence that local people distinguish between these two species.

host specificity. All known hosts are within the Fabaceae (Table 1). In Oman and Yemen, the primary host is Vachellia tortilis (Forssk.) Galasso & Banfi (Acacia tortilis) (Bolin et al. 2018). However, Bolin et al. (2018) found a robust population of Hydnora arabica growing on Pithecellobium dulce (Roxb.) Benth. in a goat yard. Goats feeding in the nearby wadi where H. hanningtonii is abundant are thought to be the likely vector for the Hydnora seeds into the goat yard. This tree is native to the Americas and has been introduced into much of Africa and Asia. Similarly, the type of H. cornii (now H. hanningtonii) was found on Albizia lebbeck, another introduced tree.

notes. Species description adapted from Vaccaneo (1934) and Bolin et al. (2018). Hydnora arabica was described in 2018 and was thought to be restricted to the southern Arabian Peninsula. However, examination of the protologues and type specimens of H. hanningtoni Rendle and H. cornii Vaccaneo revealed that both species share the same morphological characters as H. arabica; notably an apical osmophore that is equal to or greater than ½ the length of the tepal, dense strigose setae that cover most or all of the tepal margin, and an entirely red perianth tube. As H. hanningtoni Rendle is the oldest synonym described, it is selected here as the true name for this species. According to Art. 60.12 of the International Code of Nomenclature (Turland et al. 2018), the specific epithet is changed to hanningtonii, as is appropriate for an epithet named in honour of a male person. Examination of herbarium material revealed that several specimens from Tanzania, Kenya and Somalia previously labelled as H. abyssinica were in fact H. hanningtonii. Furthermore, a set of photographs online by Jabruson of (supposedly) H. abyssinica growing on Acacia roots in Kenya, Tsavo National Park, were identified as H. hanningtonii (https://www.mindenpictures.com/stock-photo/fly-on-flower-of-(hydnora-abyssinica)-parasite-growing-on-acacia-roots-tsavo/search/detailmodal-0_90179754.html). Morphologically, this species is quite similar to H. abyssinica, which can make identification from dried specimens relatively difficult, although not impossible as there are clear characters that reliably separate the two species.

8. Hydnora solmsiana Dinter (1909: 57); Marloth (1913: 178); Dinter (1923: 424); Harms (1935: 291); Schreiber in Merxmüller (1968: 41). Type: Namibia, Windhoek, Klein–Windhoek, before 1909, Dinter 356 (syntype B [not seen, likely destroyed]) & Namibia, am auß bei Keetmanshoop, before 1909, Dinter s.n. (syntype B [not seen, likely destroyed]). Neotype: Namibia, Horn River at Warmbad Township, no date, Bayer s.n. (neotype selected here PRE!, isoneotypes HUH!, K!, LD, US!).

Herbaceous subterranean perennial root holoparasite, without leaves or stem. Rhizome irregular, terete, subterete and often compressed, 1.5 – 6.5 cm in diam., rhizome surface coriaceous, dark brown; rhizome spreads laterally and often bi- or trifurcating or branching irregularly to form smaller terminal branches; rhizome ornamented with numerous lateral tubercles of variable size and shape, randomly and densely distributed on the surface, not in clear consistent lines (although may occasionally form what appear to be lines), tubercles remain the same or develop into a flower or develop into an haustorium; numerous flowers and flower buds on single rhizomes; rhizome fleshy, pinkish internal tissue when broken. Flower usually 4-merous (rarely 3- or 5-merous); flower emerges only partially from soil; two floral chambers, androecial chamber subtended by gynoecial chamber, inner surfaces of chambers glabrous. Perianth external tissues coriaceous, rust-brown, total length 10 – 18 cm; perianth tissues fleshy; perianth tube usually creamish-white inside, darkening to brick red brown (dried), 2 – 4.5 cm wide; tepal lobes creamish-white (in life), darkening to brick-red brown (dried), tepal (measured from apex to point of connation with adjacent tepal) 3.5 – 10 × 1.2 – 3 cm (measured at midpoint); tepals lanceolate to spathulate, gently curved and connivent; tepal margins covered in diffuse setae, longest on dorsal side, less dense on ventral side, c. 2 mm long, generally quite variable between individuals. Osmophore apical osmophore present, with an indistinct border between osmophore and setose tepal margin, roughly horizontal or sloping up apically from dorsal side to the ventral side; recessed osmophore present, reaching down to basal end of setose tepal margins and often slightly beyond, elliptic or spathulate, often fleshy and pitted in life; both osmophores visible in dried specimens as distinctly paler regions; no unpleasant smell reported. Androecium antheral ring formed by connate w-shaped anther lobes, forming a central orifice, transversely striate and divided into numerous horizontal pollen sacs. Gynoecium ovary inferior, unilocular with numerous ovules produced from apical placenta, ovary c. 1.5 – 2 cm wide; lobed but nearly flat and cushion-like stigma on the floor of the gynoecial chamber. Fruit subterranean, globose, c. 4.5 cm in diam., leathery brown pericarp. Seed spherical, black-brown, c. 1 mm, thousands of seeds embedded within mealy pink-white pulp. Fig. 13.

Fig. 13.

Hydnora solmsiana. A illustration of flower and rhizome; B – F examples of herbarium specimens: B tepals (Kotze & Giess 10892 [WIND]); C tepal showing apical and recessed osmophore (Bayer s.n. [US]); D indistinct border between osmophore and tepal setae region (Bolin 09-06 [WIND]); E flower (Bolin 09-06 [WIND]); F rhizome (Musselman 6279a [BM]); G open flower (Malaisse 11290 [BR]); H flower cross-section; J freshly-opened flower. A illustrated by sebastian hatt.

distribution. Hydnora solmsiana has been recorded from Namibia, Botswana, South Africa, Eswatini, Zimbabwe, the DRC, Sudan, Rwanda and western Tanzania. Given how patchy this known distribution is, it is likely that H. solmsiana also grows in Angola and Zambia but has not been reported yet. H. solmsiana appears to be distributed more across southern and central Africa, while H. abyssinica, H. hanningtonii and H. sinandevu tend to be found more across East Africa. Map 9.

Map 9.

Distribution range of Hydnora solmsiana.

specimens examined. sudan. Central Sudan: Gezira, Um Barona, 27 Aug. 1982, Musselman 6129 (E, K); Blue Nile Prov., Abu Naama, no date, Musselman 6228 (E!); Khartoum, S of Khartoum, Dom Island, 4 Jan. 1989, Musselman 6279A (BM!, BR!, E!, K!, LD!). tanzania. T4: Katisunga, 19 Jan. 1950, Bullock 2273 (K!). democratic republic of the congo. Katanga: Dikulushi, 20 km N of Kiankalamu, likely 1980, Malaisse 11290 (BR!); Dikulushi, 15 Dec. 1980, Malaisse 11348 (BR!). zimbabwe. Midlands: Gwelo, Gwelo Teacher’s College, 21 Feb. 1967, Biegel 1929 (K!, SRGH!); Selukwe, Uoncima Farm, 2 Jan. 1965, Guy 159891 (K!, SRGH!); Matabeleland South: Filabusi, 26 Dec. 1962, Guy 139469 (BR!, K!, SRGH!); Matabeleland North: Bulawayo, no date, Musselman & Obilana s.n. (M!, ODU!). botswana. Ngamiland: Okavango, no date, Smith s.n. (ODU!); Okavango, Qhusai Island, 25 Nov. 1979, Smith 2904 (PSUB!); Ngamiland, Aug. 1946, Wilmot PRE52465 (PRE!); Unlocalised: July 1994, Bolnick 22 (MO). namibia. Karas Region: Horn R. at Warmbad, no date, Bayer s.n. (neotype K!, isoneotypes HUH!, LD, PRE!, US!); Gondwana Canyon Park [S27 24.157, E17 56.768], 11 Feb. 2009, Bolin 09-06 (WIND!); Am Auß bei Keetmanshoop, no date, Dinter s.n. (B [not seen, likely destroyed?]); Bethanie, Plaas Umub, 13 April 1970, Kotze & Giess 10892 (WIND!); Gobas, 27 Aug. 1927, collector unknown s.n. (NBG!); Hardap Region: Maltahöhe, Bergzebrapark Naukluft, 24 Feb. 1974, Giess & Robinson 13256 (WIND!); Maltahöhe, Bergzebrapark Naukluft, 3 Sept. 1972, Merxmüller & Giess 28200a (M!); Khomas Region: Windhoek, Klein Windhoek, no date, Dinter 356 (syntype B [not seen, likely destroyed]); Hatsamas, 6 March 1911, Dinter 1951 (B [not seen, likely destroyed], SAM!); Hereroland, near Windhoek, Jan. 1912, Bohr 5093 (PRE!); Erongo Region: 10 km E of Uis [S21 13.454, E15 03.295], 2 Feb. 2009, Bolin 08-1 (WIND!); Uis [S21 13.454, E15 03.295], 31 Oct. 2008, Bolin 08-1-2 (WIND!); Otjozondjupa: Otavi, Grootfontein, no date, Freyer & Giess 7469, 7650 (WIND [not found, likely missing]); Okahandja, no date, Gaerdes & Giess 7651 (WIND [not found, likely missing]); Tsumkwe, no date, Kat 807 (WIND [not found, likely missing]); Tsumkwe, edge of Nama Pan, no date, Story 5162 (WIND!); Grootfontein, no date, Bauer s.n. (WIND [not found, likely missing]); Kavango West: Okavango Mile 46 [18°18'22"S, 19°15'3"E], 19 Feb. 2003, Strohbach BS5635 (WIND!); Kavango East: Okavango, Hukapama [17°53'116"S, 20°15'26"E], 7 Nov. 1996, Burke 95373 (WIND!); Rundu, along Kavango R. [S17 53.164, E20 15.461], 3 May 2009, Bolin 93-9 (WIND!); Oshikoto: Halali, no date, Musselman & Visser s.n. (ODU!); Namutoni, no date, Musselman & Visser s.n. (ODU!); Etosha Pan, Tsumeb, Namutoni, 29 Dec. 1986, Dujardin 10491 (M!); Oshana: Etosha National Park, Okaukuejo Rest Camp, 20 Nov. 2005, Bolin 05-2 (WIND!). eswatini. Nsoko, lowveld of Swaziland, Feb. 1977, von Wissel s.n. (K!, PRE!). south africa. Limpopo Prov.: Transvaal, Naboomspruit, 12 Jan. 1939, Galpin s.n. (PRE!); Transvaaal, Potgietersrust, Jan. 1940, McDonald 213 (K!); Potgietersus Distr., Tugela Hotel, 3 Jan. 1978, Muller s.n. (PRE!); Besjeskuil, before 1931, van der Byl 8829 (K!); Northern Cape Prov.: Namaqualand, Richtersveld, Sendelingsdrift, 10 Jan. 1996, Williamson 5882 (PRE!); Little Namaqualand, banks of Orange R., Arris Drift [Beesbank], 31 Aug. 1925, Marloth 12526 (PRE!).

habitat. Appears to have some degree of preference for riverside habitats, although not exclusively. In Namibia, it has often been collected and seen by the authors near rivers (Bolin 939 [WIND!]; Bayer s.n. [HUH!, K!, LD, PRE!, US!]; Williamson 5582 [PRE!]), with Dinter describing several observations along rivers in Namibia (Dinter 1909). Similarly in Sudan it has been collected along the Blue Nile (Musselman 1984) and in Botswana in riverine woodland in the Okavango delta (Smith 2904 [PSUB!]). However, it is not necessarily restricted to this habitat, and has been collected from several non-riparian habitats across its range.

conservation status. Due to the poor understanding of this species and the lack of recognition of it as distinct species over the last century, its conservation status is currently unknown. It has several reported uses across its range, although the extent of harvesting and the damage this is causing to population levels has not been ascertained. Given it appears to have a number of possible hosts and a distribution that spreads across several countries, it is hoped that the plant is unlikely to be critically threatened, although it may face local threats at a population level.

phenology. In southern Africa, it is often recorded flowering in December or January following the rains (Marloth 1913; pers. obs.), with many specimens being collected around December and January. Therefore, across its vast range, flowering time varies considerably.

etymology. Dinter does not explain his choice of specific epithet, although it is likely that he named it in honour of the German botanist Hermann zu Solms-Laubach, who wrote about the seeds of the Hydnoraceae and Rafflesiaceae (Solms-Laubach 1874) and produced a short monograph on the Hydnoraceae in 1901 for Adolf Engler’s Pflanzenreich (Solms-Laubach 1901).

vernacular names. Vernacular names for Hydnora solmsiana may apply to other species of Hydnora, particularly H. abyssinica to which it bears close resemblance. In southern Africa, in Xitsonga/Xichangana, it is known as mavumbule (Williams et al. 2011b). It is known as uMavumbaka in Zulu, meaning ‘the one that pops up’, a name that is shared with Sarcophyte sanguinea Sparrm. (Williams et al. 2011a). In Sudan, a common name for the plant is tartous, tartousch or dambu, deriving from the Arabic word utartis, meaning ‘to hold fast’ (Musselman 1984).

ethnobotany. In Sudan, the dried rhizomes are used as a charcoal/fuel alternative. It is reported to be far superior to normal charcoal, producing a heat of unrivalled intensity and evenness (Musselman 1984). Also in Sudan, the plants are powdered and boiled to produce a tannin-rich astringent paste used in the treatment of diarrhoea and assorted intestinal ailments (Musselman 1984). The white pulp of the fruit is also eaten as a delectable food source in their own right and is reported to resemble custard apple (Annona squamata) (Annonaceae) in taste and an apple in texture (Musselman 1984). Curiously, a large shipment of Hydnora solmsiana [then H. abyssinica] was seized from the Germans during the First World War, although the purpose of this shipment remains unknown (Dinter 1909; Musselman 1984). Perhaps it is linked to its high tannin content and subsequent use in tanneries before the war (Musselman & Visser 1987). In the Democratic Republic of the Congo, the rhizomes are crushed into a decoction and used in the treatment of swollen legs and feet (Malaisse 1982). This decoction is also apparently poured on the soil of Cassava fields in order to ensure good harvest and restore soil fertility.

Hydnora solmsiana is consumed as food by a wide range of animals, including monkeys and goats along the Blue Nile in Sudan (Musselman 1984; Musselman & Visser 1989) and elephants at Etosha Pan National Park (Musselman & Visser 1989). Beyond uses, H. solmsiana has been reported causing damage to local streets, driveways and patios in both Namibia and Sudan, where its rhizomes have burst through the road or tile surface, even causing cracks in the walls of a nearby house (Maass & Musselman 2001).

host specificity. Hydnora solmsiana is restricted to Fabaceae and has several overlapping host species with H. abyssinica (Table 1). So far, it has only been recorded on the following hosts: Acacia saligna (Labill.) H.L.Wendl. [Acacia cyanophylla], Faidherbia albida (Delile) A.Chev. [Acacia albida], Senegalia nigrescens (Oliv.) P.J.H.Hurter [Acacia nigrescens], Senegalia polyacantha subsp. Ampylacantha (Hochst. ex A.Rich.) Kyal. & Boatwr. [Acacia polyacantha subsp. Campylacantha], Vachellia karroo (Hayne) Banfi & Galasso [Acacia horrida, Acacia karroo], Vachellia luederitzii (Engl.) Kyal. & Boatwr. [Acacia luederitzii], Vachellia nebrownii (Burtt Davy) Seigler & Ebinger [Acacia nebrownii], Vachellia nilotica (L.) P.J.H.Hurter & Mabb. [Acacia nilotica], Vachellia reficiens (Wawra & Peyr.) Kyal. & Boatwr. [Acacia reficiens], Vachellia rehmanniana (Schinz) Kyal. & Boatwr. [Acacia rehmanniana], Vachellia seyal (Delile) P.J.H.Hurter [Acacia seyal], Vachellia sieberiana (DC.) Kyal. & Boatwr. [Acacia sieberiana], Vachellia tortilis (Forssk.) Galasso & Banfi [Acacia tortilis], Vachellia tortilis subsp. heteracantha (Burch.) Kyal. & Boatwr. [Acacia heteracantha, Acacia litakunensis]. It is likely not limited to the hosts above, particularly because one of the hosts, Acacia saligna, is an introduced species from Australia. Dinter records Acacia horrida as the host in the protologue, although this is likely a misidentification of Acacia karroo; a mistake made by many collectors in Namibia at that time (Namibia Biodiversity Database 2021).

notes. Species description adapted from (Dinter 1909; Marloth 1913). Hydnora solmsiana was described in 1909 and was accepted and considered a valid species over the 25 years that followed. The monographs of Vaccaneo (1934) and Harms (1935) then expressed doubt about the strength of the distinction between H. solmsiana and H. abyssinica (then H. johannis). It was scarcely mentioned again until 1989, when Musselman officially synonymised it with H. abyssinica (then H. johannis) (Musselman & Visser 1989). The uncertainty stems from the brevity of Dinter’s original description, which omits several key details about the defining morphological characters, and lacks a clarifying illustration (Dinter 1909).

Although the description is brief, it does mention that Hydnora solmsiana is parasitic on Acacia karroo (A. horrida), and that the flower is 4-merous, with a whitish-pink interior and a subterete rhizome. This is barely sufficient as a distinguishing description and makes no reference to the important character of the apical osmophore. Unfortunately, the type specimen was almost certainly destroyed in Berlin during the Second World War as it is no longer found in the Berlin herbarium, making it difficult to be without any doubt that the name: ‘H. solmsiana Dinter’ indeed applies to the specimens described here. However, while Dinter’s original 1909 description is brief, its contents do agree with the characters described here as belonging to the species, and H. solmsiana is by far the most dominant species of Hydnora with terete rhizomes in Namibia. More thorough descriptions and illustrations with associated herbarium specimens are later provided by Marloth in 1913, and in Flowering Plants of South Africa by Pole-Evans in 1931, accompanied by a colour illustration by Cynthia Letty. Both of these concur with, and expand on, the original description by Dinter. The specimens of H. solmsiana cited in this paper, along with several iNaturalist observations, have consistent morphological differences from H. abyssinica and these characters match those in the combined descriptions of Dinter, Marloth and Pole-Evans. As a result, the species name H. solmsiana Dinter is reinstated here. Molecular data would provide further confidence to this taxonomic decision.

As a result of the destruction of the type in Berlin, a neotype was designated here. Bayer s.n. ([HUH!, K!, LD, PRE!, US!) was selected as it has multiple duplicates and the specimens are in good condition, clearly illustrating the morphological characters of the species.

Hydnora solmsiana is most similar to H. abyssinica, from which it is distinguished by having a conspicuous recessed osmophore that extends at least as far as the basal end of the tepal margins, and an indistinct border between the apical osmophore and the tepal margins that often slopes up apically from the dorsal side to the ventral side. While colour is not a reliable trait, the tepal margins are generally whitish-cream when fresh, particularly in southern Africa.

Examination of specimens of Hydnora abyssinica across Africa revealed that many of these are in fact H. solmsiana and clearly display traits associated with that species. These specimens were mostly in southern or central African countries and not East African countries. It appears that the distribution of H. solmsiana and H. abyssinica are largely distinct, with H. solmsiana generally west of the Great Rift Valley, and H. abyssinica to the east. The two populations co-exist in the Kruger National Park in South Africa, although seem to maintain morphological distinction. Malaisse (1982) wrote that the specimens of H. abyssinica [then H. johannis] he observed in the Democratic Republic of the Congo, Dikulushi, may in fact be H. solmsiana, and indeed he was correct.

Curiously, there is a single specimen of what appears to be Hydnora abyssinica in central Namibia, which is unexpected as it is far from the distribution of other H. abyssinica (Häblich & Giess 7652 [M!]) in an area apparently dominated by H. solmsiana. It is also possible, and perhaps likely given the distribution, that the population of H. abyssinica reported in Nigeria is indeed H. solmsiana, given its proximity. However, the images published in Agyeno et al. (2018) are not sufficient to determine this.

9. Hydnora bolinii S.Hatt sp. nov. Type: Ethiopia, Harar Province, Schevelli Valley, flat area above river, 10 km E of the Fafan Valley [42°40'E, 9°16'N], 1432 m, 14 April 1963, Burger 2695 (holotype K!; isotype US!).

http://www.ipni.org/urn:lsid:ipni.org:names:77340707-1

Herbaceous subterranean perennial root holoparasite, without leaves or stem. Rhizome irregular, often both terete, subterete and compressed at different points, up to 3 cm in diam., although likely has the potential to be larger; rhizome surface coriaceous, dark grey-brown; rhizome spreads laterally; rhizome ornamented with numerous lateral tubercles of variable size and shape, randomly and densely distributed on the surface, not in clear consistent lines (although may occasionally form what appear to be lines), tubercles remain the same or develop into a flower or develop into a haustorium; numerous flowers and flower buds on single rhizomes; rhizome fleshy, internal colour not seen. Flower usually 4-merous (3 and 5 also observed several times); flower emerges only partially from soil; two floral chambers, androecial chamber subtended by gynoecial chamber, inner surfaces of chambers glabrous. Perianth external tissues coriaceous, rust-brown, total length up to 16 cm, may grow even larger; perianth tissues fleshy; perianth tube salmon and white inside, up to 3 cm wide; tepal salmon and white inside, tepal (measured from apex to point of connation with adjacent tepal) 5 – 10 × 2 – 3 cm (measured at midpoint); tepals lanceolate, gently curved and connivent; tepal apex tissue glabrous outer cucullus and apparently not an osmophore; tepal margins setose, setae c. 2 mm long, tepal margin only reaching half way from ventral to dorsal side of tepal. Osmophore positioned in two places; apical osmophore a narrow upside-down V-shaped strip, apical to the tepal margin and basal to the outer cucullus; recessed osmophore in tepal cavity; unknown if smell is produced. Androecium antheral ring formed by connate w-shaped anther lobes, forming a central orifice, transversely striate and divided into numerous horizontal pollen sacs. Gynoecium ovary inferior, unilocular with numerous ovules produced from apical placenta, ovary c. 1.5 – 2.5 cm wide; lobed and cushion-like stigma on the floor of the gynoecial chamber. Fruit and Seed not seen. Fig. 14.

Fig. 14.

Hydnora bolinii. A – D examples of herbarium specimen material; A tepal (Gillett 12781 [K]); B tepal, side view (Burger 2695 [US]); C very desiccated tepal apex (Gillett & Watson 2358 [K]); D rhizome (Gillett 12781 [K]); E tepal illustration, side view (left), front view (right); F rhizome illustration; G, H type specimen (Burger 2695 [K]): G herbarium sheet; H spirit specimen; J close up of apical osmophore, reduced tepal margins and recessed osmophore (Gillett 12781 [K]). E, F drawn by sebastian hatt.

recognition. Hydnora bolinii is similar to members of subgenus Dorhyna, but is distinguished by having an apical osmophore reduced to a narrow strip (vs apical osmophore as the actual apex of the tepal), together with a recessed osmophore that extends to the basal end of the tepal margins (as seen in H. solmsiana but not H. abyssinica). The setose tepal margin never touches the dorsal edge of the tepal (vs H. abyssinica and H. solmsiana where it touches the dorsal edge for the entire length of the tepal below the osmophore). There is a glabrous outer tissue that forms a large glabrous cucullus at the tepal apex (this tissue is not present in H. abyssinica or H. solmsiana).

distribution. Africa: Ethiopia, northern Kenya, Somalia. The true extent of Hydnora bolinii is likely greater than we currently understand given the limited number of collections made. Map 10.

Map 10.

Distribution range of Hydnora bolinii.

specimens examined. somalia. Somaliland: N Somalia site A/9 [9°50'N, 45°33'E], 23 June 1981, Gillett & Watson 2358 (EA!, K!). ethiopia. Harar Prov., Schevelli Valley [42°40'E, 9°16'N], 14 April 1963, Burger 2695 (holotype K!, isotype US!). kenya. K1: Dandu [3°26'N, 39°54'E], 14 April 1952, Gillett 12781 (K!).

habitat. Collected in rich Commiphora-Acacia scrub scattered with larger trees such as Delonix, Terminalia and Gyrocarpus (Gillett 12781 [K!]) and in degraded open Acacia woodland on eroded yellow-red silty sand (Gillett & Watson 2358 [EA!, K!]). The limited number of collections likely do not represent the full range of this species’ preferred habitat.

conservation status. Due to the limited number of collections of this species and the lack of recognition of Hydnora bolinii as distinct species over the last century, its conservation status is currently unknown. The very small number of collections and observations made of this species may be due to its rarity, in which case it is of likely conservation concern. However, this may simply be an artefact of it growing in fairly remote and inaccessible locations. As such it is difficult to comment on how threatened this species is, although perhaps much of what has been written about H. abyssinica (see species entry) applies here.

phenology. Specimens collected in flower in April and June. (Gillett 12781 [K!]) reports that it rained on 27^th March 1952, and it was collected on 14^th April that year. This may suggest that like other Hydnora species, it often flowers following rains.

etymology. This species is named in honour of Dr Jay Bolin, in recognition of his vast contribution to the study of Hydnora over almost two decades of intrepid field work and innovative research.

vernacular names. Known as tukha in Boran and liki in Somali (Gillett 12781 [K!]). It is likely that many of the vernacular names for Hydnora abyssinica also apply to this species, as it is currently unknown whether local people differentiate between the two species.

ethnobotany. The fruit is reported to be eaten by people (Gillett 12781 [K!]). (Burger 2695 [K!]) writes that it is ‘dug up by Somalis’. It is likely that many of the uses of Hydnora abyssinica also apply to this species, as it is currently unknown whether local people differentiate between the two species.

host specificty. No confirmed host known (Table 1). Reported to grow amongst Acacia and Delonix (Fabaceae), Commiphora (Burseraceae), Terminalia (Combretaceae), Gyrocarpus (Hernandiaceae), Boscia (Capparidaceae), Ziziphus (Rhamnaceae), Zygophyllum and Balanites (Zygophyllaceae), and Salsola (Amaranthaceae) (Gillett & Watson 2358 [K!]; Gillett 12781 [K!]), so the host is likely one of these. (Burger 2695 [K!]) describes it as ‘under Acacia’, although this single report is not sufficient to be confident that this is the true host. Further collections are required. Given its apparent relatedness to Hydnora abyssinica based on morphology, it is perhaps likely that it parasitises Acacia (Vachellia or Senegalia), though this is not possible to confirm with the data available.

notes. Three specimens were united in showing the same set of striking morphological differences in the tepals from any other Hydnora species. They are distinguished by the large glabrous outer cucullus, in combination with a reduced setose tepal margin that does not touch the dorsal edge of the tepal, a reduced apical osmophore to a small upside-down v-shape above the tepal margin, and an enlarged recessed osmophore. It is described here as a new species, H. bolinii, and placed within subgenus Dorhyna due its terete rhizome with randomly distributed tubercles, presence of an apical and recessed osmophore, and its probable Acacia host. Due to the very small number of known specimens, molecular data and further collections/observations in the field would provide considerable confidence to this taxonomic decision.

Certain characters, such as the glabrous outer cucullus, are shared with Hydnora sinandevu, while others, such as the setose tepal margin, are shared with H. abyssinica and H. solmsiana. No other species has the apical osmophore reduced to a narrow strip above the tepal margin, and no other species has a setose tepal margin that does not touch the dorsal edge of the tepal. Both osmophores appear as distinctly paler regions in herbarium specimens, and are thus immediately recognisable. Unfortunately, there are currently no known photographs of a live specimen.

IV) Hydnora subgenus Sineseta S.Hatt, subg. nov. Type: Hydnora sinandevu Beentje & Q.Luke.

http://www.ipni.org/urn:lsid:ipni.org:names:77340708-1

Rhizome irregular, brown or caramel-coloured, often both terete, subterete and compressed at different points, rhizome ornamented with numerous lateral tubercles of variable size and shape, randomly and densely distributed on the surface, may occasionally form what appear to be lines. Flower usually 4-merous. Perianth external tissues coriaceous, brown to distinctly caramel-coloured; perianth tissues somewhat fleshy; tepal whitish to yellow-orange inside (in life and dried), tepals narrowly lanceolate, typically often strongly curved, not fused at apex; tepal margins or outer cucullus entirely glabrous. Osmophore recessed in tepal cavity, fresh flower not seen by authors, reported to generate faint musty smell or no smell at all.

recognition. Hydnora subgenus Sineseta is distinguished by having a combination of terete rhizomes with randomly distributed tubercles, entirely glabrous tepals, a recessed osmophore and a lack of an apical osmophore. Only known to parasitise Commiphora (Burseraceae).

10. Hydnora sinandevu Beentje & Q.Luke (2002: 5); Bolin et al. (2018: 107). Type: Kenya, Kilifi Distr., Arabuko-Sokoke, Mida, between mangroves and forest, root parasite on Commiphora africana [39°59'E, 03°19'S], 2 m, 1 Jan. 1992, Luke 3033a (holotype EA!, isotypes K!, MO!, US!).

Herbaceous subterranean perennial root holoparasite, without leaves or stem. Rhizome terete, 0.7 – 2 cm diam., may be larger on mature specimens (none found); rhizome surface coriaceous, brown, caramel-coloured; rhizome ornamented with numerous lateral tubercles usually distributed in roughly parallel lines on the surface, although sometimes tubercles only loosely follow these lines or not at all, tubercles remain the same or develop into a flower or develop into a haustorium; numerous flowers and flower buds on single rhizomes; rhizome flesh (live) not seen. Flower usually 4-merous; flower emerges only partially from the soil; two floral chambers, androecial chamber subtended by gynoecial chamber, inner surfaces of chambers glabrous. Perianth external tissues coriaceous, brown to distinctly caramel-coloured, total length 8 – 16 cm; perianth tissues somewhat fleshy; perianth tube reddish to pale pink inside, 5.3 – 9 cm × 1.9 – 2.8 cm; tepals whitish to yellow-orange inside (in life and dried), tepal (measured from apex to point of connation with adjacent tepal) 2.7 – 6.5 × 0.9 – 2.9 cm (measured at midpoint); tepals narrowly lanceolate, typically often strongly curved, not fused at apex; tepal margins or outer cucullus glabrous and not setose. Osmophore recessed in tepal cavity, fresh flower not seen by author, reported to generate faint musty smell or no smell at all. Androecium antheral ring formed by connate w-shaped anther lobes, forming a central orifice 15 – 20 mm high, with the top 46 – 73 mm above the flower base, transversely striate and divided into numerous horizontal pollen sacs. Gynoecium ovary inferior, unilocular, subglobose, 15 – 30 mm (when dry) in diam., lobed and cushion-like stigma on floor of gynoecial chamber. Fruit only one observed, subterranean, globose, 9 cm in diam. Seed spherical, black-brown, 0.5 – 2 mm, thousands of seeds within fruit. Fig. 15.

Fig. 15.

Hydnora sinandevu. A illustration of flower and rhizome; B – D examples of herbarium specimens: B flower cross-section (Luke 3033A [EA]); C young flower cross-section (Luke 4304 [EA]); D rhizome (Luke 3033A [EA]); E live bud (left) and open flower (right); F cut rhizomes and associated Commiphora sp. host (right). A illustrated by sebastian hatt. photos: e william burger, slide no. 3490 [K].

distribution. Known primarily from Kenya and Tanzania. A small number of specimens have been collected from Ethiopia, Somalia and Saudi Arabia. This suggests that the distribution of Hydnora sinandevu is likely greater than it seems. Collected from an elevation of 0 – 1500 m. Map 11.

Map 11.

Distribution range of Hydnora sinandevu.

representative specimens. saudi arabia. Jizan Prov.: between Sabiya and Idabi, 7 March 1985, Collenette 5119 (E!, K!). somalia. Northeast Region: Hobyo Distr., 20 km N of Hobyo [5°55.2'N, 48°53.8'E], 30 May 1987, Wieland 4422 (MO!); South Region: Lower Shabeelle, Cara Cadde, 1 Aug. 1988, Kilian & Lobin 2124 6976 (B!). ethiopia. Oromiya Region: Filtù, 18 Oct. 1937, Vàtova 814 (FT!); Slopes W of Midaga above Gobelli R. Valley, 9 May 1964, Burger 3490 (K!, US!); SNNPR: Turkana, a road going to the Ileret Pump [4°29.71'N, 36°22.40'E], 10 June 2012, Kimeu & Brown 52 (EA!). kenya. K1: Tana R., S of Garissa, Jan. 1943, Bally 2051 (EA!, K!, MO); South Turkana, Kailongoi Mts [1°50'N, 35°50'E], 15 June 1970, Matthew 6850 (K!); K7: Kwale, Taru Desert, Oct. 1956, Bally 10434 (EA!, MO); Taita in Ost-Afrika, July 1877, Hildebrandt s.n. (G!); Kilifi, Arabuko Sokoke, 1 Jan. 1992, Luke 3033A (EA!, K!, MO, US!); Tana River Distr., Tana River National Primate Reserve [01°54'S, 040°04'E], 19 March 1990, Luke et al. TRP677 (K!, EA!, MO); Tsavo National Park, Voi Hill, 14 May 1963, Sheldrick 177/63 (EA!). tanzania. T2: Oldurai Gorge, April 1964, Leakey 97/64 (EA!); T3: Lushoto Distr., Tanga, Muheza, 22 Aug. 1935, Davies 1004 (EA!, K!, MO); Pangani Distr., Msubugwe, 17 March 1950, Verdcourt AH9924 (EA!, MO); T4: Katavi National Park [0706S, 3103E], 18 Dec. 2006, Luke 11673 (K [not seen, may be missing], EA!); T6: Pwani, Bagamoyo, 1884, Kirk 150 (K [not seen, may be missing], MO!); Selous Game Reserve, Sand Rivers Lodge [07°46'S, 38°08'E], 8 Feb. 1995, Luke & Luke 4303 (EA!, K!, MO).

habitat. In Kenya/Tanzania, found in scattered tree grassland, Acacia-Commiphora bushland, thicket or the forest margin between mangrove and forest. In Ethiopia it has been found on limestone slopes ~1500 m under Terminalia (Combretaceae) shrubs/woodland, although most records are from lower altitudes, < 500 m. In Saudi Arabia it has been found in a shallow sandy wadi over basalt slabs, in shade under Commiphora (Burseraceae), Grewia (Malvaceae), Maytenus (Celestraceae) and Cissus (Vitaceae) shrubs.

conservation status. An IUCN red list assessment from 2008 listed Hydnora sinandevu as Least Concern (Gereau et al. 2020). It is difficult to speculate on the conservation status of the species, particularly as recent discoveries of specimens in Ethiopia and Saudi Arabia indicate that the range is significantly greater than previously thought.

phenology. Collected in flower in January and August in Kenya/Tanzania, in March in Saudi Arabia and in May in Ethiopia and Somalia. These seemingly random flowering times may indicate that flowering is determined by recent rainfall rather than seasonal changes.

etymology. The specific epithet translates to ‘without beard’ in Kiswahili. This is a reference to its glabrous tepal margins (Quentin Luke, pers. comm.).

vernacular names. In northern Kenya it is known by Rendille herdsman as lekiyomoy (pers. obs.). Vernacular names for this plant may be the same as for Hydnora abyssinica, and it is unknown whether locals differentiate between these two species.

ethnobotany. The relatively recent description of this plant means that ethnobotanical information for this species is scant. It is possible that much of the ethnobotanical information for Hydnora abyssinica also applies to this species, as the two were conflated in the past. Beentje & Luke (2002) note in the protologue that it has minor medicinal use against throat complaints. Discussions with Rendille herdsman in northern Kenya revealed that they make a drink from the rhizomes that has medicinal value related to birthing (pers. obs.).

hosts. Restricted to Commiphora (Burseraceae) hosts, making it the only species of Hydnora known to do so (Table 1). The only confirmed hosts are Commiphora campestris Engl. and C. africana (A.Rich.) Engl. in Kenya and Tanzania. However, H. sinandevu in Saudi Arabia was recorded growing amongst several shrubs including Commiphora quadricincta Schweinf (Collenette 1985). Although no direct host-parasite contact was recorded, it seems likely that this was the host plant for this specimen. It may well be the case that populations of H. sinandevu discovered in the future are found to parasitise other Commiphora species not mentioned here.

notes. Species description adapted from Beentje & Luke (2002). Hydnora sinandevu was described in 2002 by Beentje & Luke as part of the Flora of Tropical East Africa project. It is easily distinguished by being the only Hydnora species to be entirely glabrous and the only species to parasitise Commiphora (Burseraceae). It was previously thought to be restricted to Kenya and Tanzania, although collections from Somalia, Ethiopia and Saudi Arabia have been found that were previously misidentified as H. abyssinica. This considerably expands the potential distribution of this species. Due to its unique set of characters (see Infrageneric Classification), it has been assigned to a new subgenus, Sineseta.

Conclusions and recommendations for future work

This monograph is the first comprehensive taxonomic revision of Hydnora for nearly a century, and should serve as a bedrock for future evolutionary and ecological studies into the genus. Following an exhaustive review of herbarium specimens and literature, the number of accepted species is raised from eight to ten, following the reinstatement of H. solmsiana, and the description of H. bolinii S.Hatt sp. nov.

While much can be gleaned from existing collections, further field work is critical to better understand all aspects of the biology of Hydnora, particularly distribution and host ecology. An important next step will be the production of a widely sampled molecular phylogeny for the genus. This will not only illuminate evolutionary relationships between the species, but also serve as a foundation for future studies. The distribution and host data recorded here has revealed interesting patterns in host specificity. Research into the mechanisms in operation at the host-parasite interface across a range of species will greatly assist our understanding of host specificity and evolution in the genus, and in parasitic plants more broadly.