The Cheilanthes hirta complex and allied species ( Adiantaceae / Pteri daceae ) in southern Africa

The very variable complex of plants until now ascribed to Cheilanthes hirta Swartz, together with some derived or allied species, is investigated. The type of the species as established by N. C. Anthony (1984), is accepted. Three new varieties are distinguished: Cheilanthes hirta Swartz var. brevipilosa W. & N. Jacobsen, var. inferacampestris W. & N. Jacobsen and var. nemorosa W. & N. Jacobsen. One new form of Cheilanthes hirta var. brevipilosa W. & N. Jacobsen is recognized: forma waterbergensis W. & N. Jacobsen. Var. laxa Kunze (1836) is given the new status o f forma: Chei­ lanthes hirta var. brevipilosa W. & N. Jacobsen forma laxa (Kunze) W. & N. Jacobsen. Three allied species are discussed and included in the key to all taxa mentioned. Taxa are described and information given includes notes on distribution and ecology. Special attention is paid to spore morphology. Most taxa are illustrated by a habit photo, and all by line drawings and scanning electron micrographs (SEM) of the spores. Possible evolutionary conclusions, particularly on the nature of the spores and on frond dimensions in relation to geographical distribution and climate are discussed.


INTRODUCTION AND AIMS
Amongs the 24 currently accepted species of Chei lanthes Swartz (Anthony 1984) in southern Africa, the widespread C. hirta Swartz contains in its broader con cept a number of rather different looking plants.In this respect it resembles the equally widespread C. viridis (Forssk.)Swartz complex which has been split into a number of varieties or separate species (Anthony 1984).This has so far not been done in the case of C. hirta.Since the first naming of the species by Swartz as Adiantum caffrorum in 1801 and its description by him in 1806 as Cheilanthes hirta, it has been referred to under a number of different specific epithets in different genera and with varying breadth of concept until as recently as 1983 (Jacobsen) and 1984 (Anthony).
The present paper deals with the varying forms, varie ties and allied species which were at one time or another included in the species concept.The authors realize that, similar to the C. viridis complex, the various forms of C. hirta will propagate partly sexually and partly apogamously with the possibility of the more xerophytic ones being apogamous.Cytological studies might clarify in what way polyploidy and hybridization affect their evo lution.This is at present being done with a number of cheilanthoid ferns of Europe, the Mediterranean and Macaronesia and it has led to the creation of a great number of hybrids, each with its own name.The authors are not equipped to do this, but they have made a thorough examination of morphological features, includ ing those of the spores, and they conclude that their findings warrant separation of the C. hirta complex into a number of taxa.

HISTORY AND SYNONYMY
The typification of the species presents difficulties.It could either be based on a collection by Dr Groendal in Herb.Gastroemi of a plant from Mauritius in the Riksmuseet, Stockholm, or on a collection by Thunberg from the Cape of Good Hope in Herb.Thunberg, Uppsala.Certain entries by Wikstroem on the former sheet seem to indicate that this is indeed the specimen on which Swartz (1801: 85) based his Adiantum caffrorum and on which he himself changed the name to Chei lanthes hirta (Swartz 1806: 128, 329).Anthony (1984) accepts the sheet from Mauritius as the type of the spe cies.Figure 1.
Professor N. Lundqvist of the Cryptogamic Section of the Riksmuseet (personal correspondence) is of the opi nion that Swartz typified both his Adiantum caffrorum and his Cheilanthes hirta with material from South Africa and that the sheet from Mauritius is not the type.He also mentioned that it was not marked thus by Prof. Schelpe, who quotes it as the type (1969: 72).The sheet in the Thunberg Herbarium is, according to Lundqvist, not the type, as Swartz had not seen it.He did not com ment on the possible locality of the type sheet.Thus the question of the whereabouts of the type sheet is not solved.It appears best, however, to follow Anthony (1984) and to accept the sheet from Mauritius at least as the lectotype.The specific epithet hirtum or hirta based on Swartz' material was combined by various authors with various genera, which are now included in Chei lanthes.
Kunze recognized the variability of the species in 1836 and distinguished four forms: a contracta, (3 inter media, Y laxa and 5 parx iloba.The types of these names were housed in his herbarium in Leipzig and were destroyed during the last war in air attacks on the city.In later works, beginning with Pappe & Rawson (1858), and including those of Sim (1915), Jacobsen (1983) and  Anthony (1984), the forms of Kunze are referred to as varieties, a procedure in agreement with Article 35.3 of the Code.Anthony (1984) and Roux (1986) have investi gated the typification of Kunze's forms.They have established the following: 1, var.contractu is synonymous with C. contractu (Kunze) M ett., a species on its own; 2, for var.interm ediu, Anthony (1984) selects as lectotype a collection by Drêge from the W itberg in the southern Cape, now in B under No. 000414 and deter mined by Kunze.A duplicate of this collection is in SAM.Roux (1986) also studied material from B but he states that he could not find any material listed by Kunze in his text and he concludes that the variety is untypified; 3, for var.lu.xu, Anthony (1984) selects a collection by Drege from near Bokpoort and Nieuweveld as lectotype in B under No. 000413, with a duplicate of this collection in SAM.Roux notes that the specimen Drêge s.n.(BM) collected near Enon and Bontjes River, mounted on one sheet with some plants collected by Cooper can be identified with var.lu.xu, but they are not cited by Kunze and cannot be considered as syntypes; 4, var.purvilobu is cited by Roux (1986) as synony mous with C heilunthespurvilobu Swartz.
The authors of this paper have investigated the dupli cates of the lectotypes of var.intermediu and var.lu.xu in SAM as well as photostats of a probable syntype of var.intermediu and of an isosyntype of var.The lectotype chosen by Anthony (1984) for C. hirtu var.la.xa consists of a rather damaged plant with thin stipes and rhachises and rather crumpled pinnae as well as part of a stipe and lamina of similar poor preservation.When Kunze described the forms of the C. hirtu com plex he had only material from the southern and eastern Cape at his disposal.The authors of this paper worked with much more material, including specimens from the Transvaal, the Orange Free State, Lesotho, Natal and the northern Cape Province.On account of shape, character of stipe and lamina and vestiture, they consider the lecto type to represent a form of their var.brevipilosu, so that the correct name of the plant is now C. hirtu Swartz var.brevipilosu W. & N. Jacobsen forma lu.xu (Kunze) W. & N. Jacobsen.The photostat of an isosyntype of C. hirtu Swartz var.luxu Kunze shows six very damaged fronds, collected by Drêge and now housed at the Museum in Leiden under Morton 751.The locality is not known.The habit of these plants is similar to the lectotype above and the plants are considered to be a good match of it.Pappe & Rawson (1858: 252) described as Chei lunthes glundulosu a plant collected by the missionary R. Moffat, possibly from a locality near Kuruman.Accord ing to their text this appears to be a lax form of C. hirtu.The whereabouts of the type are not known.Anthony (1984) incorporated C. nielsii Jacobsen (1983) in her concept of C. hirtu as a shade form.

MATERIALS AND METHODS
The authors have studied the large collection in the National Herbarium at Pretoria as well as their own col lections.The following features were used to distinguish taxa: rhizome and basal stipe scales, lamina configura tion, length to width ratio of lamina and largest pinna, vestiture of stipe, rhachis and lamina, angle setting of pinnae to rhachis, and sculpturing of the spores by means of electron microscope scanning.The latter was carried out by the Botanical Research Institute, Pretoria.

RESULTS
The authors have come to the conclusion that the com plex of forms should be subdivided, that the obviously allied and consistently deviating forms without interme diates should be considered separate species and that the remaining forms of C. hirtu sensu luto should be split into a number of forms or varieties.
As an allied species which might have evolved from a common ancestor, Cheilunthes purvilobu (Swartz) Swartz was considered a separate species by its author, who had originally named it Adiuntum purvilobum in 1800, but transferred it to Cheilunthes in 1806.Though later included as a variety of C. hirtu by Kunze (1836), Pappe & Rawson (1858) and others up to Christensen's Index filicum (1906( ), Sim (1915) ) accepted it again at species level and this view is supported by the present authors.Due to its glabrous and viscid lamina surface it is immediately distinct from all other more or less hirsute members of the complex.It is therefore not discussed any further in this study.
The general essential features of all forms of the com plex are set out by Jacobsen (1983: 2 6 1 -2 6 3 ) and by Anthony (1984: 6 8 -6 9 ).An important diagnostic fea ture is the presence of patent or adpressed multicellular, often gland-tipped hairs on stipe, rhachis and lamina surfaces.As gradations occur between a number of the taxa, separation at species level has been possible only in a few instances.Three of the hirsute members are sepa rated at specific level, and four at varietal level.One of the varieties is split into two formae.The main differ ences are shown in the following key.The species (Figures 2b, 3) has recently been described by Anthony (1984) so that only a few pertinent morphological features need be stressed here, which dis tinguish the species from all other taxa of the complex: the long creeping rhizome; the rhizome scales which are darker than those of other members of the complex; the dark green coloration of the ventral surface of the lamina (only common with C. hirta var.hirta) and especially the twisted 3 -4 (-6)-cellular hairs on stipe, rhachis and lamina surfaces, adpressed downwards along the rhachis (Figure 3c).The pinnae are mostly 1 2 -1 5 mm apart, 9 -2 8 x 2 ,5 -1 0 mm, with up to 10 pinnules, spreading at an angle of 3 8 °-5 2 ° (average 46°) from the rhachis, bu curving upwards to run parallel or even towards it and touching the pinnae above (Figure 2 b).The pinnules may be very close together or more often well separated and alternate on the costae, about 4 x 2 mm where measurable, but often too contracted, especially when mature or drying out; the sori are dark brown, about 0,5 mm in diameter.The spores are globular, about 50 fim in diameter, rugulose with anastomosing rugae and a prominent trilete ridge (Figure 4 a -c ) .

Distribution
The species is essentially southern, but has recently been found to occur in the eastern Transvaal, where it overlaps in its distribution with that of C. hirta var.hirta and of var.brevipilosa.In the Cape Province it extends from the Kamiesberg in Namaqualand along the Cape mountain ranges to the south-western, southern and south-eastern parts of the province, approximately as far as the Grahamstown area.Northwards it is found as far as the Swartberg.The species is quite frequent in the Cape, but rare in the Transvaal (Figure 5).

Ecology
C. contracta usually grows in shallow, sandy lithosols amongst rocks or even on sheetrock mats.Quartzite soils seem to be preferred.Aspect is of no importance, but plants at the foot of large boulders in half or part shade tend to be larger.In arid fynbos the stipe may be quite long, frequently as long as the lamina.In spite of the creeping rhizome the fronds are mostly only ± 3 mm apart, the stipes irregularly sprouting, young ones indis criminately mixed with mature ones, so that large clus ters of fronds develop.
The altitudinal range is large, according to Anthony (/.c.) between 60 and 1 830 m in the Cape, but apparently not below 1 000 m in the Transvaal.An herbarium sheet

2.
C heilanthes hirta Swartz var.h irta .Chei lanthes hirta Swartz: 128, 329 (1806).Type: Mauritius, Groendal (S, lecto .-Herb .Jacobsen, photo.!).Figures 2a, 4 d -f The lectotype is a rather contracted plant, of which only the upper part is preserved, not dissimilar to certain specimens from South Africa (Figure la, b).The conclu sion that the South African populations agree with the lectotype is based on investigation of a photo of the type sheet and on the following observations: what is here considered to be var.hirta is, apart from var. inferacam pestris, the only eastern representative of the complex, reaching the coast of South Africa in Transkei (Figure 7).Tardieu-Blot (1958) describes under Cheilanthes hirta Swartz var.contracta Kunze a plant with con- trad ed , elongate fronds, with long, white, patent hairs and with globular trilete spores with convex faces and cristate-reticulate sporoderm.The senior author saw a contracted, elongate plant with long white hairs on the surfaces which he identified as a form of Cheilanthes hirta on Réunion in 1978.No other forms of the variable species are recorded from these localities.It is accepted therefore that the variety described here agrees with the Mauritian plant.

Distribution
Apart from the type locality in M auritius, the presence of the variety has been confirmed in Reunion, in Mada gascar, Zimbabwe and southern Africa.Here it is known from the Transvaal Highveld, northern and western Transvaal, Swaziland, the Orange Free State, Lesotho, Natal, Transkei and the eastern and southern Cape, with one occurrence as far west as the Somerset West area (Figure 7).The variety is frequent in the Transvaal, Natal, Transkei and the eastern Cape, rare in the OFS, Upper Karoo and southern Cape.

Ecology
The variety grows mostly in shallow sandy to loamy lithosols in rocky ground, amongst boulders and out crops, often at the foot of boulders or cliffs, usually amongst some herbs and grass, but also in open bush country in sunlight or preferably in part shade.The alti tude ranges from almost sea level in Natal, Transkei and the Cape to more than 2 000 m (almost 3 000 m in Leso tho in a deviating, dwarf, highly contracted form) at 400 to 1 000 mm annual rainfall, however, in the wetter areas not in forests, but always in open well drained positions.In the winter rainfall area of the Cape the plants dry and shrivel during the hot summer months, while on higher ground in the Free State, Lesotho and Transvaal they fade during the winter to develop new shoots at the onset of the summer rains.They may be evergreen, but are not soriferous during the winter in the coastal areas of higher rainfall.

Notes
It had initially been thought, that the fronds of C. hirta var.hirta were always highly contracted, corresponding to Kunze's (1836) var.contracta, but detailed investiga tions have shown that the essential differences between the taxa of the complex lie in the type of vestiture of rhachis and lamina and the spore characters.Contracted habit as well as more spread out fronds may occur in C. contracta, C. hirta var.hirta and var.brevipilosa.All three are well separated by the nature of the rhachis and surface hairs as well as by differences in the nature and size of the spores.A varietatibus et formis ceteris speciei pilis brevissimis glandulosis in superficiebus ambabus laminae differt.

C heilanthes h irta
Rhizome creeping with broadly based lanceolate, acu minate, light brown to reddish scales, 5 -9 x 0 ,5 -1 ,2 mm, basal ones shorter with black median stripe.Fronds erect, tufted, closely spaced.Stipe terete, dark brown, dull or shiny, 1,2-2,0 mm in diameter, stout, basal scales as those on rhizome, acuminate to linear, 5 -1 0 x 0 ,4 -0 ,8 mm, becoming less and smaller upwards and replaced about half way up by patent, twisted or shaggy light brown or reddish hairs up to 1 mm long protruding from a shaggy coat of shorter hairs.Lamina linear-lan ceolate to lanceolate, often broadest above the middle (Figure 9a), 3-pinnatifid to 3-pinnate, slightly contracted to densely crumpled (Figure 8) with the pinnae flat or inrolled and twisted out of the plane of the frond; apex rounded or subacute with a tiny pinnatifid terminal seg ment 2 -5 mm long.Pinnae mostly spreading at 6 3 °-7 9 ° from the rhachis, but often at more acute angles, es pecially in contracted forms, otherwise pinnae and pin nules as in var.hirta, except for very short, glassy, occa- sionally glandular hairs, arranged flabelliform along the costules of the pinnules on the upper side (Figure 9 0 , and very sparse, short, gland-tipped hairs on the under side which may be almost glabrous.Rhachis densely covered by short and adpressed, 1-3-celled glandtipped hairs 0 ,2 -0 ,5 mm long, with occasional patent longer hairs of this type, up to 5-celled and 1,8 mm long (Figure 9d, e).Spores averaging 48 /xm in diameter (Std Dev. ± 1,34), strictly triangular, trilete ridge prominent, essentially strongly and coarsely rugulose-cristate, especially on distal face, but with isolated short rugae and rare verrucae on proximal face (Figure 4g, h).

Distribution
Known so far from South Africa and SWA/Namibia.In the former mainly in a strip trending north-south from the northern, eastern and western Transvaal through the Orange Free State and western Lesotho to the eastern Cape, with an isolated outlier in the northern Cape Pro vince south of Vryburg and three north-south orientated occurrences in southern SWA/Namibia (Figure 10).A find allotted to this variety in the Natal Drakensberg is doubtful and may be a hybrid of var.hirta and var.nemorosa (Jacobsen 4628).The variety overlaps in its distribution to a large extent with that of var.hirta, but seems to be absent from Natal and the Transkei.

/K V Ecology
Ecologically its requirements are similar to those of var.hirta.The altitudinal range also tends to be alike, probably from about 1 0 0 -2 000 m. Highly alpine forms are missing.On the whole the variety prefers more xerophytic localities, yet has a broad moisture tolerance range, and occurs in both summer and winter rainfall areas with an annual precipitation ranging from as low as 100 mm to about 1 000 mm.It is common in the Transvaal and frequent in the OFS, but becomes rarer in the eastern Cape Province except for the Hogsback-Keiskamma area.

Notes
The variety is distinguished by its smaller, triangular spores from var. hirta, which has large, rounded spores.The original concept of a taxonomic separation on account o f the generally more open and spreading lamina in var.brevipilosa could not be upheld, as some fairly contracted forms with triangular spores (N.Jacobsen 4522, Figure 8) were seen.A subsequent study disclosed the different type of vestiture, which forms a most important additional criterion for differentiation and which was used in the key to the taxa of the complex.Rhizome erect, small, with closely tufted fronds with scales lanceolate, acute, 4 -5 X 0 ,6 -0,8 mm, lower ones light brown with dark median stripe, upper ones light brown to foxy red, broader and larger.Stipe terete, reddish brown, more rarely dark brown to almost black, thin, reaching only rarely 1 mm in diameter, basal scales lanceolate-acuminate, 4 x 0,6 mm, pale brown to brown, replaced higher up by patent, occasionally glandtipped, light brown hairs up to 2 mm long.Lamina lan ceolate to oblong (Figures 11a, 12a), broadest in the middle, 3-pinnatifid to 3-pinnate, 6 0 -2 5 0 x 1 5 -5 0 mm, averaging 140 X 32 mm, about 4% times longer than broad, apex rounded, pinnatifid, terminal segment 7 -8 mm long.Pinnae triangular, mostly with 3 -4 pairs of pinnules, spreading at 7 0 °-9 0 ° from the rhachis, 0 -3 pairs slightly reduced towards base, lowest 8 -1 5 x 6-10 mm, 13-27 mm distant from next above; centre pinnae 8-32 x 5 -1 4 mm, 7 -15 mm apart, apex rounded, slightly basiscopically developed at the base (Figure 12b); top pinnae crowded or up to 5 mm distant from terminal segment.Pinnules oblong, lobed to pinna tifid, apex obtuse, terminal segment unequally pinnati fid, decurrent, some simple or gland-tipped hairs on the

Distribution
The form extends from the south-western Transvaal into the northern Cape Province and south-eastern SWA/ Namibia and southwards in a broad strip through the central Karoo, the OFS and part of Lesotho to the south eastern Cape.A few finds of the form have been made in central SWA/Namibia (Figure 14).

Ecology
It is a xerophytic representative of the C. hirta com plex, growing mostly on stony slopes or amongst boul ders or in rock crevices (very often of dolerite) under fairly arid conditions, never on north aspects, rarely on west-facing slopes, but frequently on east and south aspects.It shelters usually under boulders or overhang ing krantzes or, if sufficient scrubby vegetation is avail able, as for instance in the Karoo Nature Reserve at Graaff-Reinet or in the south-eastern Cape, under shrubs, sometimes belonging to the genus Euphorbia.The habit of the plants is therefore not xeromorphic, but mesomorphic, with a small rhizome and frail stipes and rhachis.
The altitudinal range extends from practically sea level in the south-eastern Cape to about 2 000 m in west ern Lesotho in areas with the rainfall ranging from 200 to ± 1 000 mm annually.Phenologically it follows the pattern of C. hirta var.hirta and var.brevipilosa.It is occasional to rare and mostly isolated.

Notes
C. hirta var.brevipilosa forma laxa probably corre sponds with C. glandulosa of Pappe & Rawson (1858), the type of which was collected in Griqualand in 1857 by Moffat, but is now lost.According to the description it had purple stipes and rhachises with glandular hairs.Pappe & Rawson (1858: 36) remark that this species was 'very like C. hirta var.laxa of Kunze (1836)', which is obviously a lax variety of the species.It appears there fore that Baker, cited in Anthony: 70 (1984), erred in comparing a contracted form with the type specimen of C. glandulosa in Herb.Rawson.
C. hirta var.brevipilosa forma laxa differs from all other taxa of the complex, except C. hirta var.brevipi losa forma waterbergensis and C. nielsii, in the smaller size and fragile stipe.It differs from C. hirta var.brevi pilosa forma waterbergensis in the 1-2 mm long hairs on the rhachis which are without long basal pyramidal cells, and from C. nielsii in the tripinnate fronds.All morphological features coincide with those of C. hirta var.brevipilosa forma laxa, but pinnae mostly with 5 pairs of pinnules.Pinnules triangular to oblong, deeply lobed, terminal segment lobed, apex rounded, margins not hairy (Figure 15b), but hirsute on both surfaces with short 1 -3-celled, gland-tipped hairs (Figures 15f, g).Rhachis well covered with mostly thin and fine hairs, often gland-tipped, mostly 2-or 3-celled, but also with some longer ones with up to 6 cells, with a characteristic pyramidal long basal cell (Figures 15d, e).Spores about 51 /xm in diameter (Std.Dev. ± 0,57), triangular with faintly prominent trilete ridge; spore surface forming a network of irregular rugulose cristae, sparse on proximal face and on the trilete ridge.

Distribution
This form occurs to the north of the area of forma laxa in the western and north-western Transvaal with the northern limit north of the Waterberg.There are isolated finds from the Loskop Dam and from near Pretoriuskop in the Kruger National Park (see Figure 14).

Ecology
Forma waterbergensis is a xerophytic form of the drier areas of the Transvaal, where the annual rainfall does not exceed 600 mm.Scarcity of material prevents an assess ment of the altitudinal range.It may be between 700 and 1 100 m.Similar to forma laxa, the plants prefer south aspects, growing in shallow lithosols of sandstones, felsite or granite amongst and in shelter of boulders or in rock crevices.The form is rare and has been collected only a few times.

Notes
The form was separated from the similar and wide spread forma laxa on account of the different nature of the hairs of the rhachis.It differs from C. hirta var.brevipilosa in the pyramidal basal cells of the rhachis hairs; in its small size and in the frail stipe and rhachis, which are very different from the robust stipe of the variety.The pinnae spread from the rhachis at almost 90° while those of var.brevipilosa are always ascending.

Distribution
The variety is confined to the high rainfall areas of South Africa, following essentially the eastern escarp ment and the mountains of the southern Cape.The distri bution is somewhat patchy with isolated areas along the Soutpansberg-Woodbush escarpment, a long strip from The Downs in eastern Transvaal through Swaziland to the Louwsburg area and some isolated occurrences in the Witbank-Belfast and Ermelo areas.After a gap in the Natal Midlands it is found again all along the OFS/Lesotho border, along the Drakensberg and into southern Natal and Transkei, ending in the Hogsback area of the eastern Cape.The southern Cape area extends from the Suurberg to the Grootvadersbos (Figure 19).

Ecology
The variety grows as a hemicryptophyte in shallow lateritic, humiferous soils of fersiallitic to ferallitic nature, mostly amongst boulders on mountain slopes or in gullies and ravines, as long as there is a certain amount of shade.It does not enter the wet forest of its environm ent, but prefers forest edges and scrub forest in areas o f 800 to more than 1 200 mm annual rainfall.It was thought not to occur at greater altitudes than about 1 800 m, but a dwarf, typical montane form occupies elevations from 1 800 to about 2 300 m in the Witzieshoek-Sentinel area, either on scree slopes in the shelter of ericoid scrub or sometimes in company with Polystichum alticola Schelpe & N. C. Anthony or chasmophytic on basalt cliffs, again with P. alticola or Cheilanthes eckloniana (Kunze) Mett.These plants are usually be tween 150 and 200 mm, rarely up to 300 mm tall and about 50 mm broad.

Note
C. hirta var.nemorosa is morphologically an easily recognized type and its spores are rather different from the other members of the C. hirta complex.This might have justified specific rank, but some intermediates be tween it and var.brevipilosa or var.inferacampestris respectively were seen, so that varietal status is prefer able.Jacobsen,var. nov. Type: Transvaal,Kruger National Park,V. d. Schijff 3990 (PRE,holo.).Figures 18d, e; 20 & 21.

Distribution
Var. inferacampestris is found from the eastern Transvaal (southern Kruger National Park) through eastern Swaziland, Zululand and Natal, to the Port Shepstone area.One isolated specimen was seen from the Devuli Ranch in the Bikita District of Zimbabwe (Figure 22).

Ecology
The variety, although confined to low-lying hot areas is by no means typical of xerophytic habitats, but is usually found in sheltered shady localities, on southfacing rocky slopes with humiferous loamy soil.It may be found in scrub forest or in kloofs and ravines, in the shade of large boulders or amongst undergrowth.The annual rainfall varies between 600 mm (where the variety grows only in moist localities) and ± 1 000 mm.The altitudinal range is between 50 and 600 m.  (1985).Type: Transvaal, farm Wanhoop 78 JT, N. Jacobsen 5255 (PRE, holo.).Figures 13g, h; 23.The species has been described recently, but it was felt that some supporting drawings and spore micrographs might be informative.As the spores have not yet been described a few words are necessary.Spores average 60 /xm in diameter, are roughly triangular with prominent cristate trilete ridge, the surface is loosely cristate with broad anastomosing cristae, up to 12 /xm high on a smooth base.

Distribution and ecology
The species is known only from a few localities above 1 800 m altitude in the eastern Transvaal escarpment (Figure 22), where they grow in shallow lithosols over quartzite, along rocky shelves and amongst boulders on east and north-facing slopes in full sunlight.The fronds wither away during the cold season.This species was fully described by Jacobsen (1983) and merely a few line drawings of the species as well as micrographs of the spores are presented here.As the latter have not yet been described this is done in this paper.Spores average 36 /xm in diameter (Std Dev. ± 0,52), are globular, the trilete ridge is not visible due to densely anastom osing, well developed cristae about 3 /urn high (Figure 1 8 f-h ).

Distribution
The species is so far known with certainty from the northern and north-western Transvaal (because of poor preservation the few records from near Pretoria are dubious) and from a few isolated occurrences in the northern Cape and from one place in Botswana.A record from SW A/Namibia is doubtful, again due to poor preservation of the specimen (see Figure 22).

Ecology
The species inhabits very dry situations, growing on rocky hillsides or hilltops mostly in full sunlight amongst rock outcrops, on ledges and at boulder bases on north and west aspects, very much rarer on SSW -facing slopes and then in partial shade.Rainfall requirements are in the 4 0 0 -6 0 0 mm range annually and the altitude lies between 700 and 1 100 m.

Notes
C. nielsii is included in this study merely because its indumentum has a certain similarity to that of the mem bers o f the C. hirta complex.Anthony (1984) declares that it is indeed a shade form of C. hirta.In view of the data given above, this view cannot be upheld.Anthony also states that the spores are indistinguishable from those o f C. hirta.While it is true that some members of the C. hirta complex have similarly strongly cristate spores, as for instance C. hirta var.hirta, others within the complex have deviating spores.The spores of C. nielsii are in fact not like any of the other members of the complex as they are very small, having an average size of 36 /xm with very little variation.The smallest spores found within the complex are those of C. hirta var.infe racampestris with an average of 41 /xm, while those of var.hirta average 64 /xm.Morphologically C. nielsii is entirely different from var. inferacampestris, a form always growing in shade under fairly dry climatic con ditions.As pointed out by Jacobsen (1983) the plants have morphologically a greater affinity to C. inaequalis (Kunze) Mett.var.buchananii (Bak.)Schelpe and to C. leachii (Schelpe) Schelpe as well as to C. capensis (Thunb.)Swartz than to C. hirta.Although the possibil ity of a derivation from the latter cannot be ruled out, it is felt that the specific status should be maintained because of the consistent morphological differences and lack of intermediates.

Spores and possible evolution
The separation of a complex of species on the basis of spores alone is debatable and the authors quote C. Wood (1973), that 'great caution must obviously be exercised when we assume that a character such as spore mor phology is of prime importance in either taxonomy or phylogeny, when so little is known concerning the methods by which such sculpturing is produced' (this refers to the sculpturing of the sporoderm).He found that there was a great variation in the spores of the Thelypteridaceae.Bearing this in mind, as well as the tendency of xerophytic ferns to be apogamous and to produce hybrids, the authors have put emphasis on morphological characters, which are of immediate use to the botanist in the field, and used the nature of the spores only in a supporting function.Study of the spores has, however, revealed the existence of a distinct and separate taxon within the group of rather contracted members of the complex.Later morphological studies confirmed that the taxon differed also substantially in other respects.This variety was named Cheilanthes hirta var.brevipilosa.
The size of the spores of members of the complex varies greatly.The type variety, var.hirta, shows an average of 64 /xm, whereas spores of other members are as small as 41 /xm (var.inferacampestris) and those of C. nielsii are consistently around the 36 /xm mark.This agrees with R. M. & A. F. Tryon's (1973) findings for the entire cheilantho'id group.The large size of the spores of C. hirta var.hirta seems to point to an ancient type and its occurrence on M adagascar, Réunion and Mauritius may indicate that the variety existed already during the Miocene, if not the Oligocene periods.The rather large standard deviation (Std Dev.) of ± 2,71 confirms the noticeable irregularity in size and ornamen tation, which would indicate apogamy and tendency to form hybrids.
Var. brevipilosa likewise seems to have separated from the ancestral form at a fairly early age as it is pro bably diploid (Prof.T. Reichstein, pers.com m .),but apparently originated later than the Miocene period.The spores are of average size (48 /xm) with relatively little variation (Std Dev. ± 1,34).Var.nemorosa with its relatively large spores (average 53 /xm) and greater variability-the standard deviation here is ± 2,29-is probably also an older segregate.The sporoderm is generally consistently finely reticulate.With its lax growth and its spreading, finely divided pinnae, var.nemorosa is easily distinguished from all other members of the complex, even when unusually small, as in the montane forms.
The spores of var.inferacampestris are much smaller, averaging 41 /xm.The variety has probably evolved from var. nemorosa in more recent times.This assump tion is merely based on morphological similarity, as the sporoderm of the varieties is entirely different, being loosely and strongly cristate in inferacampestris and finely reticulate to reticulate-rugulose in var.nemorosa.
The spores of the two formae of var.brevipilosa with their distinctly triangular shape, prominent trilete ridge and rugulose-cristate sporoderm, are similar to the typi cal forma, which was the main reason for considering them as derivatives of it-even though their small and delicate, widely spreading fronds have little morpho logical affinity with it.The spores vary very little, hav ing a standard deviation of only ± 1,10 for forma laxa and ± 0 ,5 7 for forma waterbergensis.
C. contracta has entirely different spores and is there fore thought to have originated from a non-related tribe, which developed in the southern Cape, and in a later period extended its range to the Transvaal.Spore varia tion is minimal.C. hyaloglandulosa has large, strongly cristate spores (± 60 /xm), similar to var.hirta and has probably evolved from it, but is considered to be a sepa rate species because of its dense cover of unicellular glands on several-celled stalks.The tiny spores of C. nielsii are indeed superficially similar in their cristate, round sporoderm to those of C. hirta var.hirta, but they are almost only half as large.This fact and the consistent differences in habit exclude the plants from the C. hirta group.C. nielsii is therefore considered a distinct spec ies, even though it may have derived from the C. hirta group.Cristate sporoderms are, according to Tryon & Tryon (1973), predominant in the cheilanthoid ferns, even if this is .lesspronounced amongst the South Afri can species.Yet the spores of C. marlothii (Hieron.)Schelpe ( = Notholaena marlothii Hiem), a plant of very different habit, are very similar to those of certain types of the C. hirta complex [compare Anthony (1984), Plate 2, C -D with Plate 13, C -E],

Morphology, phytogeography and climate
The length and width of the lamina and of the largest pinnae were measured as well as the angle of the pinnae to the rhachis.These parameters show some correlation with the climatic conditions under which the plants grow.The results are summarized on Table 1 and  It becomes evident that most of the members of the complex are typically associated with a certain type of climate.This is discussed in the sections on the ecology of the various taxa.The succession from east to west, i.e. from mesophytic to meso-xerophytic to xerophytic is reflected in the following sequence of the taxa: The size and division of fronds decreases generally from wet to dry conditions.
The relatively tall mesomorphic habit with finely divided fronds which allows for maximum light expo sure in C. hirta var.inferacampestris and C. hirta var.nemorosa gives way to the elongate meso-xeromorphic forms of the C. hirta var.hirta/C.hirta var.brevipilosa/C.contracta group with upwards directed, often contracted pinnae suited to cope with water loss.Towards the xerophytic end, however, no further con traction takes place and the formae laxa and waterber gensis escape desiccation by growing only in the shade of rocks, herbs or scrub on south-facing slopes, or wherever they are sufficiently protected from desicca tion.They are small and delicate and not at all xeromorphic.C. nielsii, the most brittle of all, grows on fully exposed sites, in sunlight and frequently under very hot conditions.It is, however, entirely seasonal, being green only when the rainy season has fully set in and shrivel ling rapidly when it gets too dry.In this respect and with its small rhizome and generally small size it resembles the small therophytes of arid areas.Its deviating position from the other members of the complex is well demon strated in Figure 27, where its lamina ratios plot outside the general trend and where the ratios of the largest pin nae, although beginning within the trend, leave it at a steeper angle.
lu.xu .The lectotype of C. hirtu var.intermediu is a frond without a rhizome, with a thick stipe and a contracted, narrowly linear lamina.Stipe, lamina and vestiture coincide with what the authors consider to be C. hirtu Swartz var.hirtu and they sink var.intermediu under it.The photostat of a probable syntype of var.intermediu shows a small whole plant and a large stipe and lamina on the left and right sides of the sheet respectively, col lected by Drêge, Ecklon & Zeyher, s. n., s. d.The sheet is housed at the Museum at Leiden under Morton 750.The plant on the left is from the Paarl Mountain and the frond on the right from the Sneeuwbergen.They are obviously different species, the latter is Cheilunthes contructu (Kunze) Mett.ex Kuhn, the former in habit and size similar to what the authors describe as C. hirtu Swartz var.brevipilosu W. & N. Jacobsen forma lu.xu (Kunze) W. & N. Jacobsen (see below).