New ta \ a of Aneilema R . Br . ( Commelinaceae ) from southern and tropical East Africa

Four new species o f Aneilema are described: A. indehiscens Faden, with subsp. indehiscens (Kenya, Tanzania) and subsp. lilacinum Faden (Zimbabwe, Mozambique, South Africa); A. arenicola Faden (Mozambique, South Africa); A. brunneospermum Faden (Mozambique, Swaziland, South Africa); and A. tanaense Faden (Kenya). A new subspecies, Aneilema dregeanum Kunth subsp, mossambicense Faden (Mozambique), is also described, and A. johmtonii K. Schum. is lectotypified.


INTRODUCTION
A revision of the genus Aneilema for the Flora o f Southern Africa has included a detailed study of a group of taxa which were only superficially examined during a previous investigation (Faden, 1975).The completed research has revealed the presence in southern Africa and adjacent floral regions of three undescribed species and a new subspecies, which are described below: A .arenicola Faden, A .brunneospermum Faden, A .indehiscens Faden (with two subspecies), and A. dregeanum Kunth subsp.mossambicense Faden.Aneilema tanaense Faden, from tropical East Africa, is also described because of the need to validate the name for a forthcoming publication.The problem of typifying A .johnstonii K. Schum. is discussed, and a lectotype is selected.
Two related southern African species, Aneilema dregeanum Kunth and A .schlechteri K. Schum., were not clearly differentiated until Brenan (1961) studied them.They have continued to be confused with an unrelated but sympatric species that is described below as A .indehiscens Faden.In the discussion under A . schlechteri, Brenan (1961) noted that the Kew isotype (Schlechter 11748) differed from other specimens by having the cells of the outer capsule wall epidermis more or less isodiametric as opposed to longitudinally elongate.He considered that difference as probably due to the immature state of the capsules on the type, and he could find no reason to separate this collection taxonomically from the others.When I examined the specimens at Kew in 1974, I came to the same conclusion.
(a) Aneilema dregeanum Kunth subsp.dregea num Subsp.dregeanum is endemic to South Africa, occurring in Natal and eastern Cape Provinces.It grows in moist or mesic situations, most commonly in forest, but also in bush, along streams and (rarely) in grassland, from about sea level to 940 m altitude, usually in partial or dense shade.Flowering specimens have been seen in all months except October, June and July, although the main flowering period is December to April.The flowers open in the morning and fade by 13h30.
In addition to the type collection, the following specimens have been seen:  Plants of subsp.mossambicense are different in aspect from those of subsp.dregeanum because of their narrower, commonly more coriaceous, and usually more shortly petiolate leaves, as well as their laxer-appearing inflorescences due to the longer cincinnus peduncles and more widely spaced bracteoles.The differences in stamen filament hair length and colour are quite striking, even in the dried specimens, in which they seem to have been preserved with unexpected frequency.It is probable that further floral characters will be found when living material of subsp.mossambicense can be obtained.
The cells of the outer capsule wall in both subspecies are transversely elongate.In subsp.mossambicense these cells are arranged in regular files; in subsp.dregeanum they are less well-ordered.
The seeds of the two taxa are quite distinct.Although there is almost complete overlap in seed size, the seeds of subsp.dregeanum (1 ,9 5 -3 mm long) tend to be longer than those of subsp.mossambicense (1,9-2,3 mm long).They are also consistently darker in colour, and more deeply and finely pitted, with the hilum in a much narrow er groove (less than ê the width of the seed vs é -\ the width of the seed) than the seeds o f subsp.mossambicense.Furthermore, the seeds o f subsp.dregeanum typically have the farinaceous material confined to the testa depressions, with the edges of the separate patches granular, whereas the seeds of subsp.mossambicense are commonly completely covered by matted farinaceous material which is not at all granular.
In view of the suite of differences between these two taxa, and in particular the consistent and significant dissimilarities in seed and stamen filament hair characters, treating these taxa as distinct species was considered.Because our knowledge of both taxa, especially the Mozambique plant, is incomple te, it was deemed best to describe them as subspecies at this time.Brenan (1961) recorded the capsule locules o f A. dregeanum as 2-3-seeded.I have been unable to find more than two seeds per locule in any of the specimens of either subspecies that I have examined.
Aneilema schlechteri K. Schum., A. brunneospermum Faden and A. arenicola Faden comprise a group of closely related species.They may be distinguished by the above key.Further contrasting characters are given in Table 1.
This species is confined to south-eastern Zimbab we and eastern Transvaal.It is recorded from mopane woodland, Acacia nigrescens Community, and as a weed in irrigated sugarcane.It is noted as growing in basaltic soil or heavy, black turf.Flowering specimens have been seen in December and January, and a fruiting specimen in May.

Aneilema indehiscens Faden and A . tanaense
Faden belong to Aneilema section Lamprodithyros which is centred in tropical East Africa (Faden, 1975).One subspecies of A .indehiscens occurs within the Flora o f Southern Africa area, so that species is described below.Seeds of A. tanaense have been used in a series of germination experiments that are soon to be published.Therefore, although that species is endemic to Kenya, it is described herein.The accounts of both species have been largely adapted from Faden (1975), Because of the urgent need to validate the names, these descriptions are being published in advance of a monograph of section Lamprodithyros.

Key to the subspecies
In the field or with more complete specimens or detailed collector's notes, A .indehiscens is readily distinguishable from A .petersii on the basis of its vegetative shoots long-trailing, antherode lobes often falcate, connectives usually slightly elongate, lateral stamen filaments usually not crossing, capsules narrow, often indehiscent, and dorsal and ventral locule seed dimorphism only slight.Aneile ma indehiscens is consistently tetraploid and A .petersii regularly diploid (Faden, 1975(Faden, , 1983)).
Although the two subspecies are separated geographically by almost 1 500 km, few characters distinguish them unequivocally.These are given in the key above.However, other tendencies are shown by the subspecies which are sometimes useful diagnostically.Subsp.indehiscens usually has fewer uniseriate hairs on the bracteoles.It also tends to have shorter fruiting pedicels [4 ,5 -6 ,5 (-8 ) mm] than subsp.lilacinum [ (5 -)6 -1 0 mm].The shape of the antherodes apparently will also separate all or nearly all of the specimens, although further living material is required to determine the extent o f the variation in subsp.lilacinum.In subsp.lilacinum the anther sacs and/or sutures of all three anthers are blue-black, while in subsp.indehiscens those of the lateral anthers are entirely or partly yellow and those of the medial anther are wholly yellow or orange-yellow.In subsp.indehiscens mature cap sules are usually chestnut brown; in subsp.lilacinum they are grey-brown o r mottled light and dark brown.
In southern Africa A .indehiscens subsp.lilacinum has been overlooked or confused with the unrelated A. dregeanum or A. brunneospermum.The Swazi land collection of A. indehiscens listed in the above exsiccatae is cited by Compton (1976)   Aneilema calceolus Brenan, Kew Bull. 15: 223 (1961), pro Gregory s.n.
The taxonomic confusion of this species with A. clarkei is due to the publication by Rendle (1895) of six figures (his PI. 34, Figs 7 -1 2 ) which accompany the type description of A. clarkei. Fig. 8 clearly does not belong to that species.Through correspondence with Brenan -and subsequent examination of the specimens at the British Museum (Natural History) -the writer determined that Gregory had made two separate, unmixed collections of Aneilema with the same label data.One of them is the type of A. clarkei which, significantly, has on it ail of the drawings published by Rendle except Fig. 8.The second sheet is the one cited by Brenan (1961) as A. calceolus but is here treated as A .tanaense.
Aneilema tanaense is most closely related to A. benadirense Chiov. of Somalia and A .calceolus Brenan of Kenya and Tanzania.It may be distinguished from A .benadirense by its generally shorter leaves, smaller, often non-thyrsiform inflo rescences with fewer cincinni, shorter cincinnus peduncles, less widely spaced bracteoles, puberulous sepals, hook-hairs on the ovaries and capsules, and smaller capsules.It can be separated from A. calceolus by its less prostrate habit, more widely spaced bracteoles which have marginal glands, medial sepals generally with marginal glands, lateral anther pollen yellow, and smaller capsules.

LECTOTYPIFICATION O F ANEILEM A JOHNSTONII
Three collections from Tanzania were cited by Schumann (1895) when he described Aneilema johnstonii K. Schum.: Johnston s.n., Von Hóhnel 159 and Volkens 2146.Among the syntypes Johnston s.n. and Volkens 2146 belong to A. johnstonii, as it is usually interpreted, while Von Hóhnel 159 is A. hockii De Wild.Although most of the description applies equally well to all three collections, the colour of the petals is taken from the Volkens specimen, and the description of the capsule from the Volkens and Johnston specimens.Because these characters clearly separate A .johnstonii from A. hockii, there can be no confusion as to which element the name A .johnstonii should be applied.The Von Hóhnel collection cannot be the lectotype for A. johnstonii.
O f the remaining two collections, the Berlin specimen of Volkens 2146, which was presumably seen by Schumann, has survived, whereas that of Johnston s.n. has not.I am therefore designating Volkens 2146 (B!) as the lectotype of A. johnstonii.
Aneilema arenicola is restricted to southern Mozambique and extreme northern Natal.It is found on roadsides and hillsides with partially open woodland.It grows in sandy soil in full sun from 10-60 m altitude.Flowering specimens have been seen in November, December, and February to April.In the field the flowers were observed to fade at lOhOO.In cultivation they open about two hours after sunrise and remain open for approximately two-and-a-half hours.
as A .dregeanum, a species which probably does not occur in that country.(Compton has also included A .brunneospermum in his A .dregeanum.)Similarly, Ross (1972) omits any reference to A. petersii or a related species in Natal, although A .indehiscens is quite frequent in the northern part of that province.H e, too, may have included this species in A. dregeanum o r A .brunneospermum.