A synopsis of Tephrosia subgenus Barbistyla (Fabaceae) in southern Africa

A synopsis o f the southern African species o f Tephrosia Pers. subgenus Barbistyla Brummitt is given. Twenty-one species are recognized. A new species, T. brummittii B. D. Schrire, is described and six new combinations are presented. The key is partly descriptive to provide diagnoses of the taxa and includes information on distributions. Vouchers are noted from at least three different herbaria to aid identification. Detailed notes are provided on atypical collections and taxa that require further investigation. Typification has been done as far as necessary to ensure that the names used are correct.


INTRODUCTION
Tephrosia is a pantropical genus in the tribe Millet tieae (Papilionoideae) comprising over 400 species mostly from Africa.An overview of the nomenclatural history, systematic position, taxonomy and morphology of the genus is given by Bosman & De Haas (1983) and the most recent reviews of the tribe are by Geesink (1981Geesink ( , 1984)).
In the Flora of southern Africa region it is estimated that there are 52 species eight subspecies and 10 varie ties.There are at least three interspecific hybrids with the likelihood of more being found in the future.
My treatment of the genera closely allied to Tephrosia follows that by Brummitt (1980), i.e.Requienia and Ptycholobium are retained as separate genera whereas Lupinophyllum is sunk in Tephrosia.Geesink (1984) there is good reason for retaining the two subgeneric divisions in Tephrosia (Brummitt 1980).Although precise characters such as arrangement and position of hairs either along the style or just below the stigma, or the degree of style twisting, are variable, the two subgenera are character ized by flowers with different pollination strategies.

Despite recent criticism by
The salient features of each subgenus are summarized in the key below.
In subgenus Barbistyla, 21 species are recognized; a number of H.M.L. Forbes's (1948) species are reduced to synonomy or to subspecific status.The subgenus has a high degree of endemism in southern Africa with the T. natalensis, T. macropoda, T. albissima and T. grandiflora complexes almost all radiating exclusively within the Flora of southern Africa area.The three latter groups appear to be closely related to one another.It is however only these species in the subgenus that are endemic to us.The T. longipes complex and some of the other isolated species are more tropical in origin and they represent southerly extensions into the FSA area.A number of other areas of diversity for the subgenus occur in tropical Africa.
Selection pressures for evolutionary divergence at the generic level appear to have been centred on the breeding system.The subgeneric categories have different flower structures and this is reflected in their different polli nation strategies.Within the subgenus Barbistyla, inflo rescence characters (bracts) and fruit characters (pod size and seed position) separate the T. grandiflora allies from those of T. longipes.Inflorescence specialization has been particularly marked in the T. grandiflora alliance.An interpretation of the above might be that in the early stages of the evolution of Tephrosia the floral char acters were fairly labile, and only after the build up of various constraints, did two major alternative strategies become dominant.The multifunctional-structural flower had now become constrained by the accumulated adapt ions and interactions (burden) that made it successful.This became a fixed feature around which the other parts of the breeding system could interact.Further selection could be channelled in either of two directions: speciali zation in the ovules and ovary and ultimately fruit and seed dispersal, or specialization in inflorescences acting on floral development, presentation and protection.
Further examination of these hypotheses could add to our knowledge of the radiation and interrelationships within the genus Tephrosia as a whole.This synopsis is a preliminary account of the subgenus Barbistyla aimed at providing names and a means of identification for a difficult and variable group of plants.Hybridization is so far only known from isolated cases in disturbed areas, between closely related species.
The genus appears to comprise a number of actively evolving complexes in which peripheral variation is dif ficult to assess, and much of the underlying genetic va riation is not readily quantifiable in morphological terms.A full revision of the genus for the Flora o f southern Africa is in preparation.4. Tephrosia longipes Meisn. in Hooker's London Journal of Botany 2: 87 (1843).Type: Natal, in graminosis ad latera montium Tafelberge, Port Natal, Krauss 20 (K, holo.!).4a.subsp.longipes.Brummitt in Boletim da Sociedade Broteriana, s6r. 2, 41:309(1968).
(2) A great deal of confusion has arisen in the past concerning the T. longipes-T.lurida complex.Forbes (1948) and Brummitt (1968) treated the two as separate species while Gillett (1958) placed T. lurida as a variety of T. longipes.Present research has shown that tradi tional characters used to separate them are based on eco logical which cannot be reliably upheld taxonomically.
(4) This complex together with the following species, i.e.T. reptans Bak., T. caerulea Bak.f., T. euprepes Brummitt and T. euchroa Verdoom forms a clearly recognizable assemblage that is widespread in tropical Africa, and with the exception of T. longipes Meisn.only occurs in the northern extremities of the Flora area.

Notes
(1) H. M. L. Forbes examined material in B before the Second World War for her revision (1948).She desig nated a specimen, Drege (Herb.Meyer 5471) as a type thus choosing one of the syntypes in Meyer's protologue as the lectotype.Since this material was destroyed and it is not clear which syntype she chose, a new lectotype can be designated from any of the original syntypes.
(2) The T. macropoda complex is a widespread group of trailing to prostrate open grassland herbs and over the major part of its range, i.e.Transvaal and much of Natal, Transkei and the eastern Cape, it is easily recognizable as one species, T. macropoda.On the periphery of this range, however, there is considerable divergence from the typical form: (a) Along the eastern coastal strip plants are particularly robust but are clearly recognizable as T. macropoda.
(b) In the following regions plants may be much reduced in size: (i) the higher altitude foothills and slopes of the Dra kensberg in Natal to Transkei; (ii) the southerly end of the species range in the E Cape; (iii) the coastal hinterland of Transkei and Natal; (iv) parts of the E Transvaal and Swaziland.
(3) One distinct form (referred to here as var.diffusa) occurs in these areas, and is associated with edaphically dry granite or sandstone outcrops.Plants are procum bent, diffusely branching and often form dense mats.Typical T. macropoda may be found in the surrounding grassland and the two may appear to be quite separate taxa.
(4) T. macropoda is, however, a variable species and smaller-leaved forms do occur and will intergrade with the typical form, particularly at the extremes of its range and with increasing altitude.These are sometimes diffi cult to distinguish from the diffuse plant in the herbar ium.The latter is clearly an ecotype and the name is retained as a variety of T. macropoda but doubtful plants should be retained in T. macropoda sensu lato. 12.
T. brummitti occurs sympatrically with T. macropoda and, although a few problem plants occur in overlapping areas, e.g.Huntley 194 (PRE) from Mtunzini, Natal, they are not readily identifiable as intermediates.
Two specimens from southern Mozambique, cited by Brummitt: 372 (1968)  This species is named in honour of Dr R. K. Brummitt of the Royal Botanic Gardens, Kew, who has contrib uted much to the knowledge of the genus Tephrosia from the Flora Zambesiaca area.

Notes
(1 )7 .albissima is another variable complex including a number of names published simultaneously by H. M. L. Forbes (1948).A geographically distinct segre gate, T. zuluensis is given subspecific status and this includes those forms from western Zululand and nor thern Natal which have almost sessile leaves and very reduced densely pilose leaf rhachides.This creates prob lems in the choice of key characters because the typical subspecies becomes very much reduced in size at the opposite end of its range in the higher altitude areas of Kokstad and Harding in eastern Griqualand (following a similar trend to T. macropoda sensu lato.).
(2) Another entity, T. unifolia H. M. L. Forbes is placed in synonomy under subsp.zuluensis but further field study is needed to be certain of its position.The best example is J. Turner 1 (NH) from the Babanango Distr. in western Zululand.The leaves are distinctly petiolate with leaflets narrowly oblong-elliptic to oblanceo late, both longer and narrower than typical subsp.zu luensis.The collections Brown & Shapiro 73 (K) from Itala Nature Reserve and Ross 1741 (PRE) from near Goedgegun in the Louwsberg, need further investigation as they appear to be somewhat intermediate between T. albissima and T. macropoda sensu lato.
Distribution: E Swaziland.

Notes
This species is very variable, with three distinct forms:

Notes
The largest and most difficult complex, which is allied to both T. albissima and T. macropoda sensu lato, com prises those erect or bushy subshrubs related to T. gran diflora (species 14 to 21).T. grandiflora has tradition ally been confused with many of the species in this group but presents no problems itself.The isolated T. gobensis from the Lebombo region seems most closely allied to it as does the endemic T. bachmannii from the sandstone outcrop areas of northern Pondoland/southem Natal.
as T. incarnata, also belong to this species.These are: 1, Sabie, Lebombo Mountains, Meponduine, Pedro & Pedrogao 735 (LMJ); 2, Namaacha, near the Canada Dry factory, Barbosa & Lemos 7534 (COI, LISC, LMJ).Both localities are near the border with Swaziland in the Lebombo Mountains.The plants grow at altitudes from 0-450 m in the coastal and lowveld vegetation types of N Zululand, S Mozambique and E Swaziland.These prostrate herbs occur in open grassland from coastal dunes to streambanks and forest margins along the Lebombo Mountains.