The Eriosema cordatum Complex . I . The Eriosema populifolium Group

The relationships among three species of Eriosema are discussed. E. distinctum N.E.Br. and E. populifolium Harv. are maintained as distinct from E. cordatum E. Mey. The relationships between these three species and extra-South African species of the E. cordatum complex are discussed in the light of new evidence.


INTRODUCTION
Because of their close affinity, Verdcourt (1971) suggested that a field study should be carried out to elucidate the relationships among E. cordatum, E. populifolium and E. distinctum. This paper reports the results of such a study. It does not deal, however, with the identity of E. cordatum, nor with the relationships of this species to E. nutans and E. buchananii. These aspects will be dealt with in subsequent papers in this series.

• FIELD STUDY
A field study (Stirton, 1975) has revealed that E. cordatum, E. distinctum and E. populifolium are morphologically different and that they may be distinguished on three main characters, namely the type of underground system, the vesture of the plant and the shape and size of the flowers. Of these, the last is probably the most significant but all three should be considered collectively.

Underground system
Plants of all three species develop perennial under ground organs. E. cordatum has a simple, vertical, daucate rootstock ( Fig. 1), whereas in E. distinctum and E. populifolium the underground systems are more complex. There is some difficulty in accurately interpreting these organs since without anatomical and developmental studies, it is sometimes unclear where root gives place to stem.
In E. distinctum there is usually an erect rootstock ( Fig. 2.1) similar to that of E. cordatum, but in addition a number of rhizomes arise from the stylopodial region (Fig. 2.2). These grew horizontally ( Fig. 2.3), branching periodically at acute angles, (Fig. 2.4) and ultimately extend over considerable areas. Leaves are produced at intervals (Fig. 2 .5 ) and rootlets develop from the main rootstock and the lateral rhizomes. Near Balgowan in the Natal Mid- lands, plants of this species are densely spread over areas as extensive as one hectare. Each colony consists of probably no more than ten plants. If more plants are represented this seems to result from fragmentation of the rhizomes. This feature has also been noted in Argyrolobium speciosum Eckl. & Zeyh. in the south eastern Transvaal.
No erect rootstocks have been found in E. populi folium. Here the underground system appears to consist of a thin, horizontal rootstock ( Fig. 3 .1) from which arise stumpy right-angled branches ( Fig. 3.2) that terminate in rootlets (Fig. 3.3). The horizontal rootstock also produces short rhizomes that branch repeatedly, often reaching the surface where they develop leaves (Fig. 3.4). These horizontal rootstocks cover extensive areas. One plant in the Kununata area of southern Natal had a rhizome system that was unearthed intact for over forty metres. (This was distressing since the first author had spent the previous day sampling transects in this large Kununata colony. The interplant data collected was suddenly trans formed to intraplant data by this finding). In all subsequent population studies cognisance was taken of clonal propensity before sampling was carried out.
All three species produce root nodules. Grobbelaar, Van Beyma andTodd (1967) andCorby (1974) both reported root nodule formation in E. cordatum. This was verified by Stirton (1975), who also reported for the first time nodule formation in E. distinctum and E. populifolium.

Vesture
The vesture of young unexpanded leaflets is a useful distinguishing character. In E. cordatum the young leaflets are densely glandular with red, foxy or white hairs. E. distinctum is also densely gland encrusted and has a short yellow pubescence between the raised veins that are covered with long silver appressed hairs. Glands are not obvious on the young leaflets of E. populifolium which are silky and yellow or creamy white. The young freshly unfolded leaflets of this species have a characteristic shiny, silvery-grey pubescence.
The vesture of old leaves is characteristic of each species. In E. cordatum there is a scattered patent pubescence on the adaxial surface. E. distinctum is finely pubescent with appressed hairs on both surfaces. The leaves of E. populifolium, however, are densely covered with appressed silvery-grey hairs, with or without pale yellow hairs along the margins or mixed with the grey above.
Some colonies in E. distinctum are characterized by the presence of very distinct, short, yellow bulbous-

Flowers
The three species are separated by flower colour, flower size and by the different shapes of the corolla parts. E. distinctum and E. populifolium have the largest flowers known within the genus in South Africa.
The exsertion of the stigma from the anthers is a useful field character which separates E. distinctum and E. populifolium from all other South African species of Eriosema. It has been seldom considered by taxo nomists. But during routine naming the first author found that, where it occurred, it was a useful diagnostic character in nearly all tribes of South African papilionates.

GEOGRAPHICAL DISTRIBUTION
The three species under consideration are all eastern southern African in distribution. E. cordatum is the most widespread. E. distinctum is sympatric with E. cordatum in Natal, but has two outliers, one i i the Transvaal and one in the eastern Cape. Closer study of these distributions shows that E. cordatum is frequent along the Natal coast whereas E. distinctum occurs mainly in the midlands and uplands, extending along high ground to northern and southern Natal. Where coast grades into midlands the species are sympatric but occupy different ecological niches.   E. populifolium is much more restricted. The typical subspecies with unifoliolate leaves occurs in southern Natal, where it is sympatric with E. cordatum, but allopatric with E. distinctum which occurs further inland. Subsp. capensis, with trifoliate leaves, is known from further south than E. cordatum, but is more or less sympatric with the most southerly record of E. distinctum. TAXONOMY While E. cordatum sensu lato is easily separated from the E. populifolium group comprising E. popu lifolium and E. distinctum, it is nevertheless a complex assemblage of plants with four distinct nodes of variation linked by numerous intermediates (Stirton 1975). As this complex is still under investigation only the E. populifolium group is treated taxonomically in this paper.  Ross. FI. N atal: 208 (1972). Lectotype: South Africa, Natal, without precise locality, Wood 6357 (K).
Eriosema distinctum is distributed mainly in the midlands and uplands of Natal, extending along the lower Drakensberg to southern Natal and the eastern Cape Province. A few specimens have been collected along the Natal-Orange Free State border, but only one has been recorded from the Transvaal (Fig. 6).  Eriosema distinctum has been commonly confused with E. cordatum E. Mey. It is readily distinguished from this species by its rhizomatous underground system, the presence of persistent acrorachial stipels, the general shape of the flower parts, the length of the fruits and the exerted stigma.
Three broad nodes of variation were found in E. distinctum. These ncdes occurred in the north-east, the centre and the south of the range. Roughly in a north to south direction there was, throughout the range, a decrease in plant size, flower size and general pubescence.
Plants from semi-isolated north-eastern populations near Ntumeni and Eshowe differed from those consti tuting the bulk of the species in their larger leaves, drnser patent indumentum, larger stipules and in the white pubescence of their unfolding leaflets (Acocks 11785, Haygarth s.n. Oct. 1915, Hilliard 1937, Lawn 931, Stirton 1024. Although very distinctive in the field they were difficult to assess taxonomically as a number of other specimens (Edwards 2220, Gerstner 2607, Hilliard 3001) possessed features of both the Ntumeni-Eshowe populations and the most typical midlands taxon. Acocks 11785 (PRE) is a mixed sheet which carries two specimens that grade into typical E. distinctum.
The central populations of the species comprise true E. distinctum. Clones are, nevertheless, very variable in the field, differing markedly in size and habit. The method of producing a vast, underground network of shallow acutely-angled branching rhizomes has probably accounted for the successful spread of this entity along roads after roadbuilding operations. Grading into these populations is a minor segregate, variable in habit but which is characterised by the presence of very distinct, short yellow bulbous-based hairs on the calyx and less often on the stem and petioles (Story 635, Oliver 46, 496, Goossens 276,  Mogg 2116, 2336, Trauseld 97). The poorly collected southern and northern populations might prove worthy of ranking after further gatherings have been made. The presence of a pure yellow-flowered taxon in the southern popula tions needs verification.
Attention must be drawn to a number of atypical specimens. Johnstone 487 from Hlobane (2730-DB) is definitely linked to the north-eastern populations and, with these populations, gives an impression of a close relationship with E. populifolium. This specimen is very distinctive and has the longest stipules recorded so far. It differs from typical E. distinctum in the long, grey pilose calyx lobes. Van der Zeyde 131/73 (2829-AD) has the largest flowers recorded in the species. It is a very robust plant and shows some resemblance to Hutchinson 1986 (2429-AA), the only known record of the species in the Transvaal. These three collections are retained in E. distinctum somewhat tentatively, as on present information it is inopportune to give them any formal recognition.
The clonal nature of E. distinctum, its extremely poor seed set, and its great morphological variability has led us to suspect that the species may have had a hybrid origin. We are therefore loathe to designate infraspecific rankings when the entities recognized might prove to be nothing more than clones. Likewise we consider it an unsatisfactory practice to give formal recognition to single collections when these belong to the periphery of a large variable group.
Flowering in the species occurs from late August to December and reaches a peak during November, followed by a later flush between February and April. Northern populations tend to flower earlier and for a shorter duration than southern populations. This bimodal flowering behaviour coupled with a N-S direction time cline has also been found in E. psoraleoides (Lam.) G.Don and apparently occurs in a number of papilionaceous genera in South Africa.
Apart from the two collections in the eastern Cape and a single uncertain collection from the Transvaal, the bulk of the specimens has been collected in southern Natal. Eriosema populifolium, was originally based on a single collection by Sanderson from the Transvaal. There appears to be some doubt as to the veracity of this locality. In referring to the type housed in the Kew Herbarium, Burtt Davy (1932) noted that the addendum "N atal" given in the Flora Capensis by Harvey (1862) did not appear on the sheet in the Hooker Herbarium. The inscription on the type is "Transvaal, S. Africa, Sanderson" (Fig. 9). Since no further collections have been made in the Transvaal and because the ecological preferences and the distri bution of extant populations argue against its occurence in the Transvaal, it could be concluded that Sanderson's specimen was mislabelled.
Of interest, therefore, are two nameless specimens from the M. E. Barber Herbarium (at GRA). One of these sheets has the legend "Vaal Heights" (Vaal Hoogte). We wonder whether these closely matched specimens have any connection? According to Gunn (1974)  It does not answer the question as to whether both collections were made in the Trans vaal. Secondly, " Vaalheights" is a farm or other locality name in Natal and was mistaken for the Vaal Heights area of the Transvaal. No record of such a name in Natal has yet been traced. Finally, Sanderson collected the plant in Natal but inadvertently mixed his labels. This last explanation seems the most plau sible at present.
Field studies support Verdcourt's (1971) retention of specific rank for this taxon. However, apart from the superficial resemblance given by its large cordate leaves, this species does not have much in common with E. cordatum. It is clearly very close to E. distinc tum, however, particularly to plants of the north eastern populations near Eshowe, Ntonjaneni, Ntumeni and Melmoth. Eriosema populifolium has been much confused with the names Rhynchosia villosa (M eisn.) Druce, R. sigmodes Benth., Eriosema distinctum N.E.Br., and E. cordatum E. Mey. It can be separated from all these taxa by its large incurved connate stipules ( Fig. 10.1), together with the shiny, silvery grey pubescence on the surfaces of the leaves (Fig. 10.2) and the persistent deep boat-shaped flower bract. ALBASV MUSEUM. This species has been found to include both unifoliolate ( Fig. 9) and trifoliolate (Fig. 11) leaved plants and so its concept has had to be widened. The Cape speci mens Flanagan 2369 and Pegler 683 are predominantly trifoliolate-leaved specimens with the basal leaves unifoliolate and the general dimensions overall being remarkably smaller than those of any Natal plants. These two collections, named above, are noteworthy for their smaller flowers, shorter racemes and trifolio late leaves. They seem worthy of separate rank and are described as subspecies capensis. The leaf blade of the terminal leaflet is larger (11-16 cm long, 11-16 cm wide) and the flowers longer (15-17 cm) than in subsp. capensis (Fig. 12). x 3 ; 5, calyx opened out, x 3 £ ; 6a, standard opened out, x 3 ; 6b, standard closed, x 3 ; 7, wing, x 3 ; 8 , keel, x 3; 9, vexillar stamen, x 3 $ ; 10, staminal sheath, x 3 | ; 11, discoid floral nectary,