Synopsis of the genus Salix (Salicaceae) in southern Africa

One species of Salix, S. mucronata Thunb. (= 5 . subserrata Willd.), with five subspecies, is recognized as indigenous to the southern African region. Problems of delimitation in the genus in southern Africa are discussed, and a key to the indigenous and exotic taxa is presented. The synonymy of the subspecies is presented, with leaf silhouettes and a distribution map of each. The following new combinations are made: S. mucronata subsp. hirsuta (Thunb.) Immelman, S. mucronata subsp. capensis (Thunb.) Immelman, 5. mucronata subsp. woodii (Seemen) Immelman and S. mucronata subsp. wilmsii (Seemen) Immelman.

Identification of taxa within the genus Salix in southern A frica is difficult for the following reasons: 1, all species are dioecious; 2, spring leaves, which are present when the flowers first appear, differ con siderably from sum m er leaves, and are similar in all the southern African m aterial; 3, flowers and fruits are similar in all the taxa; 4, leaf and pubescence characters are highly variable within the taxa, so that it is difficult to discern a pattern.

M ETHODS
Specimens were borrow ed from all m ajor South African herbaria, as well as from W indhoek and H a rare.The following m acrom orphological characters were exam ined: leaf size, shape and margins; petiole length; pubescence of leaves and twigs; and struc ture of flowers and fruits.The glands in male and female flowers were exam ined.The num ber of sta mens in the male flowers was counted, because this character has been often used to distinguish species of Salix in oth er parts of the world.Adaxial and abaxial leaf surfaces, as well as pollen and seeds, were viewed with the SEM.

RESULTS
No taxonom ically useful results were obtained from either the SEM exam inations or the stam en counts.The num ber of stam ens varied from 3-12 per flower, and varied even within the same inflores cence.The glands of both male and female flowers varied considerably, with 2 to num erous glands in the male flower and a glandular ring in the female.This ring varied from entire to deeply lobed, the num ber and size of the lobes being irregular, but it * Botanical Research Institute, Private Bag X101, Pretoria 0001.
did not present any pattern.Fruits and flowers of all southern African specimens were found to be essen tially similar.
Leaf length, breadth, proportions and m argin, pu bescence of both leaves and twigs, and petiole length, however, did differ between taxa, and these differences were correlated with distribution.In northern SW A/Namibia, the south-western Cape and the northern Transvaal, the sum m er leaves are large, relatively broad, glabrous or som etim es with grey canescence, and have either entire or toothed margins.A round the Olifants River (south-western C ape) the leaves are similar in shape to the above but have a dense covering of long silvery trichom es.Specimens from the rest of the Cape Province, in cluding the O range River system, and extending along the Vaal into the Transvaal and O range Free State, and into Lesotho, have short, relatively broad leaves, always toothed and glabrous, with short pe tioles and glabrous twigs.Plants from Natal and most of the Transvaal (excluding the Vaal and its tributaries), have long narrow leaves, always toothed, with longer petioles and (usually) grey-canescent twigs.Plants with larger, broader leaves, nearly always with entire margins, occur in the Transvaal Lowveld and escarpm ent from Swaziland to the northern Transvaal border and into Zim babwe.

DISCUSSION
It was difficult to distinguish clearly any of the taxa m orphologically, except the Olifants River taxon which has densely silvery-pubescent leaves.
Most characters showed at least some overlap and, in the case of the Transvaal Lowveld taxon.the characters were especially variable.H ow ever, it is possible to recognize most specimens as belonging to one or other of several taxa and, in addition, the variation is geographically correlated.For these reasons it was decided not to place all indigenous plants in a single highly polymorphic taxon.Burtt Davy (1922) in his study of the genus in the whole of A frica, was the first to suggest that each species or variety of Salix was restricted to a certain drainage basin(s).In southern Africa at least, this conclusion can be accepted, though distributions of the taxa do overlap to some degree in the south-western Cape and to a larger extent in the Transvaal.Each taxon shows a well defined range and for this reason, al though the morphological differences are small, the taxa are recognized at subspecific rather than var ietal level.
It is not certain w hether the variability and inter grading of characters is due to inter-taxon hybridiza tion since no attem pt has been made to hybridize the indigenous taxa under controlled conditions.It is possible that S. mucronata subsp.wilmsii is a hybrid between subsp.woodii and subsp.mucronata.It is in many ways interm ediate between these two subspe cies, it occurs within the range of both, and it ex hibits a great range of variation within the subspe cies.Its variation, however, may also be ascribed to inter-taxon diversity.Therefore, until experim ental studies have been done, it appears best to m aintain it as a separate taxon.The taxa are difficult to identify because the differences between them are both small and variable.A further difficulty results from the fact that there is a great difference on each plant between the spring (im m ature) and sum m er (m a ture) leaves.It is preferred to call them spring and sum m er leaves respectively, as 'im m ature' implies further developm ent, whereas the leaves m aintain their differences even when growth is com plete.The spring leaves are smaller, broader in proportion to their length, often obovate, the margins are entire and the apices may be rounded rather than acute.The transition to summer leaves is gradual, but is usually com plete by November.In the following key, only specimens collected from November to May have been considered.For specimens collected at other times of the year the locality may present the only clue to identity.U nfortunately, as stated earlier, no flower or fruit characters were found to distinguish taxa.This made the identification of type specimens difficult, as a num ber were collected in spring and in some cases no locality was given.
It is often stated that the southern African taxa do not have stipules.Examination of fresh m aterial of subsp.woodii, however, showed that this was not strictly true, as vestigial stipules are present on young twigs.These are ± 1 mm long, thick in tex ture, and are later deciduous; in dry herbarium m a terial they are not noticeable.S. mucronata subsp.mucronata does have stipules further north in its range, but not within southern Africa.
A few exotic species appear to have become natu ralized in southern Africa, and are included in the key.S. babylonica (W eeping Willow) (Figure 1.13) is w idespread and well known, while the other species, from the few specimens available, appear to be more restricted.No specimens have been seen which appear to represent hybrids between the in digenous and exotic taxa.More collecting of exotic willows is needed, together with inform ation about whether they are cultivated, naturalized or spread ing.Only female plants of the exotic species are present in this country, and none of the specimens seen were male.The unusual distribution of subsp.mucronata, which occurs in northern SW A/Namibia and in the south-w estern C ape, suggests that it may once have occurred in the intervening area, and that it has since died out there due to the drying up of perennial rivers in much of SW A/Namibia and Nam aqualand.There is no noticeable difference between the SWA-/Nam ibian and Cape populations that might justify describing a new subspecies.Subsp.mucronata is widely distributed in Africa, entering the area under consideration also in the northern Transvaal, with one record from the eastern Transvaal.Figures W ithin the subspecies as a whole there is a great range of variation, from specimens with glabrous branches and entire glabrous lanceolate leaves, to (in various com binations) broadly elliptic leaves with regularly to irregularly serrate margins, and slender or stout stem s with dense canescence.None of the specimens from our area have stipules, though they do occur in the subspecies elsewhere in Africa.
B urtt Davy (1922) seems to have had a mixed con cept of S. mucronata.His S. mucronata var.m ucro nata com prises elem ents of both subsp.mucronata (as delim ited here) and of subsp.capensis.The range of distribution given on B urtt Davy's map in cludes that of both subspecies, while the synonyms cited com prise S. aegyptica Thunb.(S. mucronata subsp.mucronata), 'S.capensis auct.non T hunb.' , and S. mucronata var.integra (placed here under 'taxa insufficiently know n').His illustrations of the leaves are clearly those of S. mucronata subsp.m u cronata.B urtt Davy therefore probably had speci mens of both subsp.mucronata and subsp.capensis in mind when delimiting his S. mucronata var.m u cronata.B urtt Davy's concept of S. capensis is narrow er than that adopted here for subsp.capensis, and I include taxa he accepts as separate.These are S. crateradenia from the northern Cape and S. capensis var.caffra from the eastern Cape.S. capensis var.integra, possibly from the eastern C ape, may also belong here but is m entioned under 'taxa insuffi ciently know n' below.5. mucronata var.mucronata sensu B urtt Davy pro parte, excluding those speci mens from the southern and eastern C ape, must also be included.Figures 1.5, 1.6; 2C.
The type of S. crateradenia (from the northern Cape) has not been located but, judging from Seem en's description, it is almost certainly S. mucronata subsp.capensis.He com m ented that it is close to S. capensis, but distinguished it by its well defined style For a discussion of specimens interm ediate be tween this subspecies and subsp.capensis, see under that subspecies.It also appears to grade into subsp.wilmsii.
A lthough the type has not been found, the de tailed description allows it to be confidently identi fied as this subspecies.The leaves are said by See men to be 90 x 11 mm, with the margin having small sharp teeth, and the shape narrowly lanceolate to linear, which can only be subsp.woodii.The fact that the type comes from Natal, from the Tugela River near Colenso, confirms this identification.(e) subsp.wilmsii (Seemen) Immelman, comb.nov.Specimens of this subspecies have been seen which are very close to subsp.mucronata, e.g.Hardy 401 and H em m 452.In its typical form, subsp.wilmsii is easily distinguishable, but it may approach subsp.woodii and subsp.mucronata in appearance.Only further research can establish whether the taxon is simply very variable or whether it has under gone introgressive hybridization with these other subspecies.As m entioned in the Discussion, subsp.wilmsii itself may also be a hybrid between subsp.woodii and subsp.mucronata.In its 'pure' form, subsp.wilmsii has stout, densely canescent twigs and large, broadly lanceolate, entire leaves, which are densely grey-canescent when young.Figures 1.10-1.12& 2E.capensis but the one from Kew is subsp.hirsuta, with large entire leaves and dense silvery pubescence on the young leaves and twigs.No other specimen re sembling this has been seen from the area, and it is possible that the Kew specimen has been misla belled.Drêge did collect at the Olifants River, w here subsp.hirsuta occurs.

summer leaf, Transvaal, Krugersdorp, near Skeerpoort River, Codd 10096. 10-12, Salix mucronata subsp. wilmsii, summer leaves. 10, Transvaal, Lydenburg, Lowveld Botanic Garden, Buitendag 997; 11, Transvaal, Kruger National Park, Sigaas, Van der Schijff 357; 12, Transvaal, Kruger National Park, near Punda Maria, C odd 5558. 13, S. babylonica, summer leaf. Cape, near Cape Town, Ecklon 713. All in PRE, x 1. auEjr FIG URE 2.-A , Salix mucronata subsp. mucronata-, B, S. mucronata subsp. hirsuta', C, S. mucronata subsp. capensis; D , S. mucronata subsp. woodii; E, S. mucronata subsp. wilmsii. 1, Orange River drainage basin; 2, Olifants River (Cape) drainage basin; 3, Limpopo River drainage basin; 4, Olifants River (Transvaal) drainage basin; 5, Maputo River drainage basin; 6, Natal rivers; 7, Cape rivers. and
the entire glands in the female flower.Styles have not been found to vary significantly in this study, and the variation in glands was found to be considerable, without any discrete ranges of varia tion.Burtt Davy states that the species comes from the headw aters of the Kuruman River, which is in the northern C ape, and is a tributary of the Orange River.A population of Salix has been found at this approxim ate locality by Mr A. G ubb, of the M cG re gor M useum , Kimberley, who states that he does not consider it to differ from 5. mucronata subsp.capen sis (pers.com m .).Some specimens which might be interm ediate be tween subsp.capensis and subsp.woodii have been seen.These are Green 88, Leendertz 3752, Muller 1053 and Sutton 884 (all at PRE).These occur near the boundaries of the two subspecies, both between the Orange and Limpopo drainage basins and the Orange River drainage basin and the Natal rivers. Vouchers: