Synopsis of the genus Brachylaena (Asteraceae) in southern Africa

A synopsis of the southern African representatives of the genus Brachylaena R. Br. (Asteraceae) is presented, in which seven species and two subspecies are recognized. B. uniflora Harv. is placed in synonymy with B. discolor subsp. transvaalensis (Phill. & Schweick.) J. Paiva and the division of subsp. discolor in two varieties is not accepted. Descriptions, synonymy, voucher specimens, distribution maps, line drawings and keys to the species and subspecies are given.


INTRODUCTION
The latest taxonomic revision of all southern African species of Brachylaena was published by Phillips & Schweickerdt (1937). They recognized nine species in South Africa, including a new species. B. transvaalensis Phill. & Schweick., as well as B. discolor DC. and B. uniflora Harv.
According to Hilliard & Burtt (1971) it is not always possible to distinguish between B. discolor. B. uniflora and B. transvaalensis on the basis of the characters de scribed by Phillips & Schweickerdt (1937). These three species became known as the B. discolor-uniflora com plex, which was regarded as the only remaining taxo nomic problem in the genus.
In an effort to solve that problem. Paiva (1972) reduced B. transvaalensis to subspecific rank, under B. discolor. He also distinguished two varieties (var. mossambicensis and var. discolor) in B. discolor subsp. discolor on the basis of reproductive characteristics. Unfortunately he did not study B. uniflora because it does not occur in his study area, namely that of the Flora zambesiaca.
Hilliard (1977) did not agree with Paiva (1972) and speculated that B. discolor and B. uniflora represented merely the extremes in a wide range of head sizes within the B. discolor-uniflora complex. According to Hilliard (1977) the variation of the characteristics used by Paiva (1972) as diagnostic, is continuous. Hilliard also men tioned that the variation can be interpreted in two ways: a single species showing clinal variation in the number of flowers in a head loosely linked to an ecological cline. or two species which are hybridizing. Gibbs Russell et al. (1987) followed the delimitation proposed by Paiva (1972), and recognized eight species and four infraspecific taxa in southern Africa, including B. discolor DC. subsp. discolor with the varieties discolor and mossambicensis J. Paiva; subsp. transxaalensis (Phill. & Schweick.) J. Paiva; as well as B. uniflora Harv.
The views presented here are supported by a detailed study of the vegetative and reproductive morphology and the anatomy of leaves and stems of the southern African representatives of Brachylaena (Cilliers 1990). Material was gathered during extensive field studies and loaned from all major South African and some overseas herbaria.
A brief review is given of the research on the mor phology of the heads of the B. discolor-uniflora complex. The variation of certain quantitative characters was pre sented in the form of dice diagrams (Radford et al. 1974) as shown in Figure 1. 2. the two varieties of B. discolor subsp. discolor cannot be separated on the basis of any of the eight parameters; 3. on the basis of six of the parameters shown. B. discolor subsp. transvaalensis and B. uniflora cannot be separated: the number of flowers both on male and female heads, however, indicate a discontinuity between the two entities; 4. the parameters of the male heads shown are taxonomically more useful than those of the female heads, which tend to show continuous variation: the length of the female involucre, however, clearly supports the recognition of the two groups mentioned under 1. On this evidence, backed by other findings by Cilliers (1990) Trees or shrubs, dioecious. Leaves alternate, simple, petiolate or subsessile, entire, toothed or crenate, some times 3-lobed at apex, mostly coriaceous, often white-or rusty-tomentose abaxially. Heads in open or dense axillary or terminal racemes or panicles, rarely solitary; unisexual, rarely with fertile hermaphrodite flowers; 1 -30(-50) flow ers per head; involucre oblong, ovoid or campanulate, bracts in 3-7 rows, free, dry, inner ones progressively longer. Receptacle epaleate, honeycombed. Male flowers: corolla tubular. 3-5-lobed; anthers tailed at base, exserted; style filiform, sometimes thickened above, simple or bifid, branches very short, flat, acute or obtuse; ovary usually abortive, pubescent; pappus poorly developed, of scabrid bristles, uniseriate. Female flowers: corolla tubular, 5lobed; staminodes occasionally present; style filiform, bifid, branches very short, flat, acute or obtuse. Achenes 4-5-angled, pubescent or subglabrous, glandular.
The leaves of B. elliptica are very variable in their shape and often conspicuously 3-lobed or 3-toothed at the apex; they sometimes resemble those of B. glabra, but are shorter, not as wide and white-tomentose (never gla brous) abaxially.
Distribution and habitat: B. elliptica is widespread in the eastern Cape from Uitenhage eastwards through Transkei to Natal and Zululand (Figure 11). It can be found in coastal and river bush, valley bushveld, in scrub or on grassy slopes, often on rocky ridges.
Distribution and habitat: within southern Africa B. rotundata occurs in the northeastern, southern and western parts of Transvaal and occasionally on the Free State side of the Vaal River ( Figure 11). It is common on rocky ridges and hills and the edges of dry kloof bush. Multistemmed shrubs or small to large trees (2-)6-10 (-20) m high; bark light grey to brown, fissured; young branches with white down, glabrescent. Leaves petiolate, oblanceolate to narrowly obovate, sometimes elliptic to narrowly elliptic; (50-)70-100 (-190) x (15-) 25-45(-60) mm, glabrous adaxially, white-tomentose abaxially, glan dular, apex acute, sometimes obtuse, rarely rounded, mar gin entire, sinuated or serrated over whole margin or in upper half and in coppice shoots, base decurrent into a petiole; petiole (3-)8-10 ( parts of Natal and Transkei up to the mouth of the Bush mans River in the eastern Cape. Its habitat varies from inland forests on valley slopes and plateaux to coastal for ests, scrub and dune bush.
Distribution and habitat: within southern Africa B. dis color subsp. transvaalensis occurs in the northern and eastern Transvaal and in Swaziland, as well as in the northeastern and southern parts of Natal and the Transkei ( Figure 16). It can be found in inland and coastal forests, but not in dune bush or forests directly facing the sea.

INTRODUCTION
The genus Ophioglossum L. has received no taxonomic attention in Africa since Clausen's (1938) confusing monograph of the genus. The simple morphological struc ture of the plants together with apparently considerable intraspecific variation has caused compilers of African re gional floras to either follow Clausen's nomenclature or assume a conservative stance by maintaining broadly cir cumscribed species which, in the light of recent studies, are often composed of two or more clearly distinct taxa.
The most important development since Clausen's work is the publication by Wagner & Wagner (1983) of their 'genus communities' concept, based upon work on the genus Botrychium Swartz (Ophioglossaceae) in North America. This concept made use of the fact that Botrych ium (like Ophioglossum) frequently grows together in multiple species communities and, when this occurs, one is able to make rational comparisons between taxa occur ring within a community, particularly if these differences are maintained in communities found elsewhere under dif ferent edaphic and climatic conditions. This concept has provided taxonomists w ith a valuable additional tool with which to distinguish between environmentally-induced variation and differences that are genetically controlled.
During recent studies, the senior author has discovered an unique area in northwestern Zimbabwe which supports 11 species of Ophioglossum. Additional genus communi ties have since been located in both Zimbabwe and South Africa which have made a better delimitation and defini tion of the southern African species of Ophioglossum pos sible. However, these studies have resulted in the need to revise the nomenclature of several of the taxa occurring on the African subcontinent.

DISCUSSION
In 1983, the senior author collected nine species of Ophioglossum growing within an area of about six square kilometres in the Sengwa Wildlife Research Area of north western Zimbabwe. Subsequent collecting at Sengwa has yielded a further two species, bringing the total to eleven, a situation not recorded to date elsewhere in the world for the genus Ophioglossum. Additional genus communities have been discovered by the senior author on Ngomakurira. northeastern Zimbabwe (6 species), Elim Hospital, eastern Zimbabwe (4 species), Treur River, eastern Trans vaal (5 species) and several communities composed of two or three species each. These localities occur on var ious substrates (sandstone, granite, quartzite and dolomite) and under varying climatic conditions, and the consistent differences displayed between members of these genus communities have allowed for clearer definitions to be established between taxa.
However, the specific concepts within the African members of Ophioglossum entrenched by Tardieu-Blot (1953, 1964 and Schelpe (1970Schelpe ( , 1977 do not adequately cover the number of taxa that are evident from genus com munity studies. This was noted by Dr Mary' Pocock of Grahamstown who spent much time studying the South African species of Ophioglossum. She recorded her con cepts in a manuscript which unfortunately was never pub