Systematics of the southern African genus Ixia (Iridaceae: Crocoideae): 4. Revision of sect. Dichone

The southern African genus Ixia L. comprises ± 90 species from the winter-rainfall zone of the subcontinent. Ixia sect. Dichone (Salisb. ex Baker) goldblatt & J.C.Manning, one of four sections in the genus and currently including 10 species and three varieties, is distinguished by the following floral characters: lower part of the perianth tube filiform and tightly clasping the style; filaments not decurrent; upper part of the perianth tube short to vestigial; style branches involute-tubular and stigmatic only at the tips; and so-called subdidymous anthers. We review the taxonomy of the section, providing complete descriptions and distribution maps, and a key to the species. I. amethystina Manning & goldblatt is recognized to be a later synonym of I. brevituba g.J.Lewis. Most collections currently included under that name represent another species, here described as I. rigida. We recognize five additional species in the section: early summer-blooming I. altissima from the Cedarberg; I. bifolia from the Caledon District; I. flagellaris, a stoloniferous species from the Cedarberg; I. simulans from the western Langeberg; and I. tenuis from the Piketberg. We also raise to species rank I. micrandra var. confusa and var. minor, as I. confusa and I. minor respectively. Foliar and associated floral variation in the widespread I. scillaris has led us to recognize two new subspecies among its northern populations, broad leaved subsp. latifolia and the dwarfed, smaller flowered subsp. toximontana; subsp. scillaris is restricted to the immediate southwestern Cape, from Darling to Somerset West. Sect. Dichone now has 17 species and two subspecies.


InTRODuCTIOn
Restricted to the winter-rainfall zone of southern Africa, the genus Ixia L. comprises some 90 species divided among the four sections Dichone (Salisb.ex Baker) goldblatt & J.C.Manning, Ixia, Hyalis (Baker) Diels and Morphixia (Ker gawl.)Pax (goldblatt & Manning 2011).Sect.Dichone, the smallest of these, comprised eight species and three varieties when last revised by Lewis (1962).In the later account of Ixia for Flora of southern Africa, De Vos (1999) followed Lewis's taxonomy closely but included the new species, I. collina, described by goldblatt & Snijman (1985), and altered the arrangement of species to place those with specialized, unilateral stamens at the end of the account.The most recent addition to the section, I. amethystina, was described by Manning & goldblatt (2006), but we have now discovered it to be conspecific with the type of I. brevituba.Most plants currently referred to I. brevituba in herbaria represent a separate species, described here as I. rigida.
Field work conducted since 1999 has resulted in the discovery of three sets of populations belonging to sect.Dichone that do not accord with any of the species currently recognized.We describe one of them from the Langeberg and allied to I. scillaris, as I. simulans, another from the Piketberg as I. tenuis and the third, evidently allied to the I. micrandra complex, as I. bifolia.
Examination of living plants and herbarium specimens matching I. micrandra var.confusa and I. micrandra var.minor provide evidence that both are better regarded as separate species, which we treat as I. confusa and I. minor respectively.We identified two further undescribed species from the Cedarberg in herbaria, namely I. altissima, an early summer blooming species of uncertain affinities and I. flagellaris, allied to I. scillaris but the only stoloniferous species in the section.
We provide a complete review of sect.Dichone, expanding it to include 17 species, with I. scillaris subdivided into three subspecies.

TAXOnOMIC HISTORY
As is the case for many Cape plants, the taxonomic history of sect.Dichone is complex.The genus Dichone (the meaning of the greek translation, two tubes, might refer to the incompletely dehiscent anther thecae) was proposed without description by R.A. Salisbury in 1812 as 'Dichone crispa Laws.Cat.' for a plant known at the time as Ixia crispa L.f. ( 1782) but now treated as I. erubescens goldblatt.The cryptic reference, 'Laws.Cat.', refers to an unpublished catalogue, perhaps merely a handwritten list, of plants evidently grown in the garden of the Scottish nurseryman, Peter Lawson, about which little appears to be known.C.F Ecklon (1827) likewise treated Ixia crispa and its immediate allies in a separate genus but for which he used the name Agretta, also without description.His species and combinations in Agretta are currently invalid.Ecklon intended to admit five species to his genus, one of which (A. stricta) is now I. stricta (Eckl.ex Klatt) g.J. Lewis, and another (A. pallideflavens), which is now I. odorata Ker gawl.(sect.Hyalis according to goldblatt & Manning 2011).Agretta stricta was referred by Klatt (1882) to his genus Tritonixia and the species was only transferred to Ixia (sect.Bothalia 42,2 (2012) Dichone) by Lewis in 1962.We have not identified Ecklon's A. grandiflora.
The name Dichone was validated by J.g. Baker (1877) as sect.Dichone Salisb.ex Baker of Tritonia, at which time it included T. scillaris (L.) Baker, T. trinervata Baker and T. undulata (Burm.f.) Baker (which he erroneously believed was an earlier synonym of Ixia crispa).T. undulata is correctly a species of Tritonia and is currently known by that name (goldblatt & Manning 2006).Ixia micrandra Baker (sect.Dichone) remained in Ixia on account of its actinomorphic flowers, although it has the filiform perianth tube, short anthers and involute style branches of Dichone.To add to the series of misunderstandings, Ixia retusa Salisb., described in 1796 by R.A. Salisbury Klatt (1882) under his combination Watsonia retusa, which like its basionym, I. retusa, is without a type.Both Lewis (1962) and De Vos (1999) excluded I. retusa from Ixia on the basis that the protologue is inadequate to identify the species.We concur.
The four species of sect.Dichone that occur in the western Karoo remained unknown until the early twentieth century, when collections of the three species Ixia curvata g.J.Lewis, I. rigida goldblatt & J.C.Manning, and I. trifolia g.J.Lewis were made by Rudolf Marloth in 1920.The first of the western Karoo species to be recognized, I. trifolia, was described by Lewis (1934), based on her own collection grown at Kirstenbosch Botanical garden.In assigning the species to Ixia, Lewis evidently did not then realize that I. trifolia was closely allied to species of sect.Dichone, which at the time was included in Tritonia.Although she noted the unusual anther dehiscence via very narrow slits in I. trifolia, she regarded the species as belonging in sect.Euixia (i.e.sect.Ixia) and close to I. ovata (Andrews) Sweet, now I. abbreviata Houtt.
By 1954 Lewis regarded sect.Dichone as more closely allied to Ixia than to Tritonia and recommended its recognition as a separate genus (Lewis 1954: 100) because of its unique anther dehiscence.Later (Lewis 1962) she formally transferred sect.Dichone to Ixia, noting that the section was also unusual in the genus in its involute-tubular style branches and its distinctive anthers.She described I. brevituba g.J. Lewis and I. curvata in 1962, assigning Marloth's collection of what we here describe as I. rigida to her I. brevituba (Lewis 1962), which is typified by a 1929 collection made by grant & Theiler.Time has provided new evidence for Lewis's conclusion that Dichone is more closely related to Ixia than to Tritonia.Species examined cytologically have the same basic chromosome number, x = 10, and karyotype as other species of Ixia whereas Tritonia has x = 11 (goldblatt 1971a;De Vos 1982;goldblatt & Manning 2011).Ixia also has derived pollen grains with a single banded operculum (goldblatt et al. 1991 and unpublished) whereas Tritonia has the plesiomorphic pollen grains of subfamily Crocoideae with a two-banded operculum.The two genera are clearly closely allied and available molecular studies (goldblatt et al. 2006, 2008) indicate a sister relationship (with moderate bootstrap support, 80%) between the one species each of Ixia (I.latifolia D.Delaroche) and Tritonia (T.disticha (Klatt) Baker) that were sequenced.Although goldblatt & Manning (1999) initially treated sect.Dichone as one of two sections of subg.Ixia they later reverted to Lewis's foursection classification of Ixia.There have been no molecular studies that shed light on the relationship of Ixia and Dichone but the unstated assumption implicit in the current taxonomy is that Dichone is nested within Ixia, hence its status as a section of that genus.

MATERIALS AnD METHODS
using standard methods of taxonomic investigation, we examined the holdings of Ixia in herbaria with significant southern African collections, BOL, K, MO, nBg, PRE, and SAM (acronyms following Holmgren et al. 1990).Our herbarium studies were accompanied by field investigation to determine variation within populations for some species, and their ecology, especially soil, aspect, and altitude.

MORPHOLOgY
Morphology: the vegetative morphology of sec.Dichone corresponds closely to that of other species of Ixia, with few exceptions.Leaves are plane and form a basal fan in most species but I. erubescens has crisped and undulate leaf margins, and leaves of I. scillaris are sometimes undulate, or rarely crisped in a few populations from the Piketberg, Pakhuis-Biedouw Mtns, and near Tulbagh.Leaves of I. minor (= I. micrandra var.minor) are terete to subterete or plane whereas those of I. micrandra are filiform, and both species usually have two or sometimes three leaves, the uppermost ± entirely sheathing the lower part of the stem.I. bifolia, I. trinervata, and usually I. simulans also have two leaves, a basal foliage leaf and a second, largely to entirely sheathing, upper leaf.In I. trinervata the lanceolate foliage leaf is distinctive in having three ± equally prominent veins.The lowermost leaf of I. simulans has one prominent vein and thickened margins, and the upper sheathing leaf is inserted above the stem base.
Flowers are produced in spikes typical of Crocoideae.The floral bracts are typical of Ixia, being membranous and ± translucent, becoming dry with age.The outer of each bract pair usually has 3 prominent veins and 3 teeth but is occasionally 1-toothed, sometimes varying within a species but consistently so in I. rigida.Flowers of most species are predominantly pale to deep pink, sometimes mauve pink, but with a white to pale yellow throat often edged darker pink to mauve.The perianth is occasionally white in some individuals or populations of I. scillaris, notably in the Olifants River Valley populations of subsp.latifolia, but is purple in I. brevituba, the flowers of which have a dark centre and stamens.Flower orientation ranges from fully upright to half to fully nodding, thus facing to the side.Flowers of several species are scented.I. bifolia, I. confusa, I. stricta, and at least some populations of I. scillaris produce a roselike scent, and I. collina and I. rigida are also sweetly scented (goldblatt & Snijman 1985).
The perianth tube is usually described as filiform (or slender throughout) in sect.Dichone but the tepals are in fact fused for some distance beyond the lower filiform part of the tube.The length of the upper part of the tube is often so short as to be conveniently ignored but in Ixia altissima and I. rigida fully half the length of the tube is flared for ± 2 and 1 mm respectively and the tube is thus clearly funnel-shaped.In I. trifolia the flared upper part of the tube is also developed to a significant degree.The walls of the slender, lower part of the tube always tightly clasp the style and the tube contains no nectar.
Although radial floral symmetry is the common, and by inference ancestral, condition in the group, flowers of Ixia collina, I. erubescens, I. simulans, I. scillaris, I. tenuis, and probably also I. flagellaris are weakly zygomorphic.The flowers in these species face sideways and are thus nodding, the tepals held more-or-less vertically, with the stamens unilateral and the anthers horizontal or drooping.At dehiscence the stamens in the nodding flowers of I. curvata and I. stricta are symmetrically arranged but become unilateral with age, evidently through gravity (De Vos 1999).
Filaments are usually free and filiform, and inserted at the mouth of the narrow part of the perianth tube.In some populations of Ixia trifolia the filaments are united basally for up to 0.7 mm, or up to one fifth their length.Anthers are oblong to subrotund and slightly to markedly recurved above the base.Species from the Western Karoo typically have oblong anthers, 4-5 mm long, with the thecae ± twice or more as long as wide, whereas those from the southwestern Cape below the Escarpment are characterized by their derived, short, suborbicular to broadly oblong anthers, 2-4 mm long and < twice as long as wide.Several species within this group have unilateral stamens (I.collina, I. erubescens, I. simulans, and I. scillaris), with the thecae sharply acute at their bases.The filaments in these species are characteristically sharply recurved apically such that the thecae are held almost horizontally.Anthers in sect.Dichone dehisce from the base via narrow slits that do not open fully, a unique feature in the genus and family.Dehiscence is either ± complete to the locule apex or incomplete and restricted to the lower third or half of the anther, as in I. scillaris.Anthers have been described as subdidymous (Lewis 1962), a term meant to imply the vestigial nature of the anther connective, which in sect.Dichone is a narrow, adaxial strip of tissue that continues from the apex of the filament to shortly below the anther apex.The two thecae are closely appressed without the usual welldeveloped sterile tissue of the connective on their adaxial surfaces.
The filiform style divides at the mouth of the narrow part of the perianth tube, thus at or shortly beyond the filament bases, and the style branches spread outwards, either ± straight (e.g.I. bifolia, I. micrandra, I. trifolia) or falcate to recurved.The style branches are involute, with the margins rolled upward to form an enclosed, tubular channel open only at the apex, with the stigmatic surfaces restricted to the apices, which are often expanded or slightly bifurcate (e.g. in I. collina and I. micrandra).
The capsules are, as far as known, typical of Ixia (goldblatt & Manning 2011), as are the seeds, which have a smooth surface and an excluded vascular trace (goldblatt et al. 2006;goldblatt & Manning 2011).Capsules are known for only a few species and we do not routinely include capsule features in the species descriptions below.notably, however, the few available capsules of I. scillaris are ± obovoid, possibly a distinguishing taxonomic feature.
InFRASECTIOnAL CLASSIFICATIOn no classification below sectional rank has until now been proposed for sect.Dichone.However, with 17 species, it seems useful to us to recognize two species groups, or series.We include species with radially symmetric flowers and either complete or incomplete anther dehiscence in ser.Euanthera, thus including the Cedarberg I. altissima, all western Karoo species, and several from the southwestern Cape.The anther thecae in ser.Euanthera dehisce conventionally along narrow slits for their entire length or almost so and are either ± rounded or acute and recurved at the base.Species with mainly zygomorphic flowers and ± unilateral stamens, I. stricta excepted, are included in ser.Dichone.The anthers in this group of species are mostly horizontal or pendent, acute and recurved at the base, and dehisce incompletely from below.no vegetative specializations consistently accompany these floral differences.Ser.Dichone, which is evidently monophyletic, is centred in the southwestern Cape below the Escarpment, with populations of I. scillaris ranging into northern namaqualand and onto the gifberg-Matsikamma Mtn complex.Ser.Euanthera is comparatively widespread.
Distribution and habitat: Ixia altissima is known from two sites in the central and southern Cedarberg, from the Matjies River nature Reserve and from gonnafontein to the south (Figure 1).Plants grow along stream banks or in marshy seeps on sandstone derived soil, at the gonnafontein site among Kniphofia plants.The habitat remains moist at least until December and possibly later into the summer.We suggest a conservation status for I. altissima of Rare (R), in light of its evident rarity, but we see no current threat to the species.
Diagnosis and relationships: Ixia altissima is a surprising plant, sometimes standing over 1 m in height and flowering late in the season, well into December.The anthers with vestigial connective and the involute style branches, stigmatic only at the tips, indicate its placement in sect.Dichone but the funnel-shaped perianth tube with the filaments inserted in the middle of the upper part of the tube is unusual for the section, most species of which have the tube virtually filiform throughout with the filaments inserted at the apex of the tube.The perianth tube, only ± 4 mm long, and the pale floral bracts are reminiscent of two Roggeveld Escarpment species, I. trifolia and I. rigida, the latter also flowering relatively late in the season, often in november.These species share relatively unspecialized, longitudinally dehiscent anthers.I. altissima seems most like I. rigida, also a tall plant, up to 600 mm high and sometimes flowering as late as november.The latter has flowers with a similarly funnel-shaped perianth tube 2-3 mm long and silvery translucent floral bracts, contrasting with the slightly longer perianth tube, ± 4 mm long, and shorter anthers, just 3 mm long, in I. altissima, which is also unusual in the section in having the style dividing opposite the middle of the anthers.Plants (150-)300-600 mm high, stem usually with 2-3 ± erect branches held close to axis, sheathed at base by firm, chestnut brown cataphylls, combining with old corm tunics and leaf bases to form collar around base.Corm 12-18 mm diam., tunics of firm, relatively hard fibres.Leaves (3-)5-7, lower (2-)3-6 with expanded blades, upper 1 or 2 leaves sheathing lower 1/3 to middle of stem, blades sword-shaped to sub-linear, (3-)5-8 mm wide, main vein, margins, and often secondary vein pairs moderately thickened, hyaline when dry. Spike weakly flexuose, nodding in bud, mostly 8-15-flowered, lateral spikes with fewer flowers; bracts membranous, silver-translucent, ± 5.5-7.0 mm long, outer with prominent brown or purple central vein and few secondary veins not reaching upper margin, acute or with 2 obscure lateral lobes, inner 2-veined and acutely 2-toothed.Flowers rotate, pale pink, darker at tepal bases and ± white in throat, sweetly rose-scented; perianth tube funnel-shaped, ± 2.5 mm long, filiform and clasping style in basal 1.5 mm or less, widely flared distally for up to 1 mm; tepals spreading, ovate-oblong, ± 12(-15) × 5(-7) mm, obtuse.Stamens symmetrically arranged, erect; filaments inserted at mouth of narrow part of tube, filiform, ± 1.5(2.5)mm long, ± white; anthers oblong to linear, curved back at base, 3-4(5) mm long, slightly recurved basally, thecae subacute at base, dehiscing fully by longitudinal slits, yellow.Style dividing at base of filaments, branches ± straight, 2.0-3.0(3.5)mm long, ± filiform-tubular, stigmatic apically, white.Flowering time: Sept. to mid-nov. Figure 2. Distribution and habitat: Ixia rigida is restricted to the Roggeveld, where it is known from the Escarpment edge on the Farms Fransplaas, Quaggasfontein, uitkyk, Houdenbek and a short distance inland northwest of Sutherland on the Farm geelhoek (Figure 1).Plants are restricted to dolerite outcrops and grow in rocky ground in the heavy red clay characteristic of this habitat.The first record of the species, made by Rudolf Marloth in 1920 and said to be from the 'Komsberg flats'[some distance south of Sutherland] may be erroneous, for I. rigida is otherwise unknown south of the town.The relatively few records show that the species is rare but plants often flower late, usually after mid-October on the escarpment edge, when relatively little collecting is done in the Roggeveld, so the present records may not fully reflect its range.There appears to be no imminent threat to its survival but at sites where we have seen the species, plants were heavily overgrazed and often do not set seed as a result.We tentatively suggest a conservation status of near-threatened, (nT).
Diagnosis and relationships: Ixia rigida is a distinctive species easily recognized by the moderately tall stem up to 600 mm high with 2-3 upright branches held close to the axis, and the silvery, translucent floral bracts subtending very short-tubed flowers, the tube up to 2.5 mm long with the narrow portion up to ±1.5 mm long.Plants have up to seven leathery leaves with prominently thickened margins and main and secondary veins.I. rigida was known to Lewis (1962) from only a single specimen, which she associated with I. brevifolia and De Vos (1999) likewise misunderstood the species, citing specimens of I. rigida under the name I. brevifolia.
Although fairly uniform across its range the populations of Ixia rigida inland of the Roggeveld escarpment and flowering three to four weeks earlier than those along the edge of the escarpment, thus in September, have slightly larger flowers and less fibrotic leaves.Too little is known about I. rigida to assess the significance of this variation.We note, however, that the escarpment populations mostly have 6 or 7 leaves, typically 5-7 mm wide, tepals ± 12 × 5 mm, anthers ± 3 mm long, and style branches ± 2 mm long.The inland populations have 4 or 5 leaves, mostly 7-8 mm wide, tepals ± 14-16 × 6-8 mm, slightly longer filaments, anthers 4(5) mm long, and style branches ± 3.5 mm long.The bracts of the inland plants also stand out in their darker purple veins.3. Ixia trifolia G.J.Lewis in Flowering Plants of South Africa 14: t. 543 (1934); Lewis: 171 (1962);De Vos: 69 (1999).Type: South Africa, [Western Cape], Tweedside, cultivated in Cape Town, Sept. 1932, Lewis s.n. (nat. Bot. gard. 2706/32 in BOL, lecto., here designated; BOL!, K, PRE!, SAM!, isolecto.)[Lewis: 171 (1962) cited this number at BOL as the holotype but there are two sheets, hence our designation of a lectotype].
Distribution and habitat: Ixia trifolia has a relatively wide range, extending along the Roggeveld Escarpment from the Farm uitkyk southward through the Klein Roggeveld to the Laingsburg Karoo as far west as Tweedside (Figure 3).Plants grow on stony clay ground in low, karroid scrub or mountain renosterveld.The flowers offer no nectar and are pollinated by pollen-collecting female bees.The anthophorine Amegilla spilostoma (Apidae) has been captured visiting flowers near Sutherland (goldblatt & Manning 2007) and we have also noted small hopliine beetles consistently visiting flowers at a population on the Farm Blesberg and clearly actively transferring pollen from flowers of one plant to those of another.
Diagnosis and relationships: Ixia trifolia is recognized by the combination of bright pink, rarely mauve flowers, white to yellow edged with dark pink in the centre, always held upright on inclined spikes, and two or three, relatively broad foliage leaves plus one (or two) more sheathing the stem.The specific epithet trifolia refers to the three foliage leaves, thus ignoring the presence of the upper, entirely sheathing leaf.Like other species of sect.Dichone from the Western Karoo, the style branches are slender and nearly straight and the anthers dehisce completely.unique to this species in the section is that the filaments may be united basally for up to one quarter of their length, and the style is exserted up to 1.5 mm from the filiform part of the perianth tube.Ixia trifolia is perhaps most closely allied to I. brevituba, which has similarly oriented, upright flowers on an inclined spike but with narrower leaves, up to 1.5 mm wide and violet flowers darker in the centre.Both species stand out in the section in having the spikes drooping in bud.
Ixia trifolia is relatively common across its range but is poorly collected.The earliest collection of the species that we have found was made by Rudolf Marloth in 1920 at the top of Verlatekloof south of Sutherland although the type population was discovered at Tweedside near Touws River sometime before 1932.The type specimens comprise plants cultivated at Kirstenbosch Botanic gardens under g.J. Lewis's name.Plants 150-300 mm high, stem either unbranched and then with scale-like leaves at aerial nodes, or with up to 2 short branches, erect below, inclined above, sheathed below by submembranous, red-brown cataphylls.Corm 12-20 mm diam., tunics of medium-textured, wiry, reticulate fibres, extending upward in papery collar 30-70 mm long.Leaves 3 or 4, basal, uppermost completely sheathing lower 2/3 of stem, blades suberect or lowermost slightly falcate, 1.5-2.0mm wide, reaching to near top of stem, firm-textured, margins and midrib hyaline, lightly thickened.Spike inclined, nodding in bud, crowded, 5-7-flowered, branches fewer-flowered; bracts dry-membranous, translucent or flushed purple above, outer 5-7 mm long, acute or obscurely 3-dentate, inner ± as long as outer or slightly shorter, 2-veined FIguRE 3.-Distribution of Ixia bifolia, ▲; I. brevituba, ∆; I. trifolia, •; and I. trinervata, ○.  08:00, even in cool, overcast conditions and remain fully open until late afternoon.Flowers of the closely related I. trifolia, like many other species of sect.Dichone, will not open fully or at all under cool conditions.The only pollinators recorded are small hopliine beetles, Kabousia axillaris.The form of the flowers is consistent with this pollination system.In view of its narrow range we concur with Raimondo et al.'s (2009) conservation status of vulnerable (Vu), but we note that the entire range of I. brevituba falls within the Tanqua national Park and future disturbance seems unlikely.
Diagnosis and relationships: distinguished by its pale purple flowers with small, dark eye, Ixia brevituba has relatively broad, blackish anthers that dehisce laterally so that the light-coloured pollen forms a contrasting pale margin to each anther, and short style branches that are ± as long as the filaments, thus scarcely projecting from between the filaments.The lovely amethyst-coloured flowers are borne on inclined spikes so that they face directly upward in an elegant, arching spray.I. brevituba falls among a small group of species that are endemic or near-endemic to the Roggeveld Escarpment and Klein Roggeveld, including I. curvata, I. rigida, and I. trifolia, all of which share relatively unspecialized, longitudinally dehiscent anthers 3-5 mm long.Among these species I. brevituba appears to be most closely allied to I. trifolia on the basis of their unusual, inclined spikes, and lateral branches that are decurved and nodding when young.All other species have spikes that are erect from bud to fruit, a feature not explicitly mentioned by Lewis (1962) in the protologue but noted and illustrated by De Vos (1999).The two species also share a perianth tube 2-3 mm long.I. brevituba differs from I. trifolia in its blackish purple anthers and purple perianth tube, which gives the flowers a small, dark central eye, consistently narrower leaves 1.5-2.0mm wide, and medium-textured corm tunics drawn into a well-developed neck.I. trifolia, like all other species in sect.Dichone, has yellow anthers, although the filaments may be pale mauve, and a pale perianth tube, giving the flowers a well-defined whitish central eye.The leaves of I. trifolia are broader than in I. brevituba, (2.5-)5.0-12.0mm wide, and the tunics are more coarsely fibrous, with the lower fibres developed into woody claws but not drawn into a neck above.In addition, the style branches of I. trifolia are longer than the filaments, 4-5 mm long, and project conspicuously between them.
The ranges of Ixia brevituba and I. trifolia complement one another-only a short distance separates the two species and we infer that they share a common ancestor.The differences in their flowers appear to be related to their pollination biology.I. brevituba has flowers that conform to the hopliine beetle pollination syndrome (goldblatt et al. 1998) and the beetle Kabousia axillaris has been captured while visiting the flowers.All individuals of this insect carried pure loads of Ixiatype pollen on their dorsal thorax and frons (Manning & goldblatt 2006).In contrast, Ixia trifolia is pollinated by solitary female anthophorine bees (goldblatt & Manning 2007) and possibly also by hopliine beetles.
When described by Lewis (1962), Ixia brevifolia included a collection of a second species, now I. rigida, and the protologue is a composite of the two species.
Distribution and habitat: Ixia curvata has a relatively modest range for sect.Dichone, extending in the western Karoo from the eastern end of the Hantamsberg at Moordenaarspoort northeast of Calvinia southward to Middelpos on the Roggeveld Plateau (Figure 1).The range remains poorly documented and there are no collections from Middelpos itself although we have seen the species on the commonage there.Our description is based largely on the Middelpos population and the illustration here is from this population.Plants favour the red clay soils derived from dolerite but are also found in pale clay ground derived from Karoo shales.Pollination of I. curvata has not been studied.Raimondo et al. (2009) regarded the conservation status as of Least Concern (LC) and we concur, noting the relatively modest range of the species.Plants 250-450 mm high, stem usually unbranched or with a single short branch, with membranous cataphylls enclosing base.Corm 9-18 mm diam., tunics of firm, medium-textured, netted fibres.Leaves 2, lower 1 ± basal, upper 1 sheathing stem in lower 1/2, blades lanceolate, 4-10 mm wide, conspicuously 3-veined, main vein moderately thickened, secondary veins slightly less so, margins not or barely thickened.Spike flexuose, mostly 7-122-flowered; bracts membranous, translucent, sometimes turning rusty brown in upper 1/3, outer 5-6 mm long, with prominent central vein and up to six secondary veins, 1-toothed or ± fringed, inner ± as long as outer, 2-veined and 2-toothed.Flowers rotate, deep rose pink, unscented; perianth tube filiform, 4-5 mm long, clasping style; tepals spreading, ovate-elliptic, obtuse to retuse, 12-15 × 6-8 mm.Stamens symmetrically arranged, erect; filaments inserted at mouth of tube, filiform, 2-4 mm long, ± white; anthers erect, suborbicular, ± 2 mm long, yellow, thecae acute and recurved at base, dehiscing by narrow, longitudinal slits almost to apex.Style dividing at base of filaments, branches falcate, 2.5-3.0 mm long, tubular-filiform, stigmatic apically, pale pink.ure 3).Plants favour clay soils and grow in renosterveld, often in stony ground.There is evidence that flowering is stimulated after fire and we have seen plants blooming well on the Farm Vleitjies after a summer burn.We agree with Raimondo et al.'s (2009) estimate of near threatened status (nT) for I. trinervata.
Diagnosis and relationships: first described by J.g. Baker (1892) as Tritonia trinervata and transferred to Ixia by g.J. Lewis (1962), I. trinervata is readily recognized by the presence of just one foliage leaf, the blade narrowly sword-shaped and with three ± equally conspicuous veins and unthickened margins.A second leaf is largely to entirely sheathing, although relatively broad.The flowers are an intense, bright mauve pink or red-magenta.Lewis (1962) considered the species most closely related to usually pale pink-flowered I. stricta, which blooms later in the season, at which time its 3-5 leaves are usually dry.Both species favour clay or clay loam soils.The new I. bifolia, described here, recalls I. trinervata in its single foliage leaf, but the leaf blade is much narrower than in I. trinervata, with a single main vein embedded in the leathery-succulent leaf tissue.When dry the leaf margins appear particularly thickened, unlike those of I. trinervata.The flowers of I. bifolia are slightly smaller than those of I. trinervata and have unusual, almost straight, deep pink style branches in marked contrast to the pale pink, falcate style branches of I. trinervata.Plants mostly 160-300 mm high, stem unbranched, with short scale ± 1 mm long, inserted 12-30 mm below base of spike, sheathed basally by dry, brown cataphylls.Corm 10-12 mm diam., tunics of fine, soft to firm fibres extending upward as collar, bearing cormlets at base.Leaves 2, lower 1with long sheath, ± lanceolate, blade 60-80 × (3-)5-8 mm, reaching up to middle of stem, leathery-succulent, main vein not raised when alive (visible as a pale line when held to light), slightly raised when dry, margins thickened, hyaline, transparent when alive, when dry raised above mesophyll and yellow, upper leaf largely sheathing, sometimes with short free blade similar to basal but 2-3 mm wide.Spike mostly 4-7-flowered, ± straight; bracts translucent pink with purple veins, 5-6 mm long, outer with 3 main veins terminating in acute teeth and 1 or 2 secondary veins along margins, inner 2-veined and acutely 2-toothed.Flowers rotate, upright, deep pink, darker toward tepal bases, rose-scented; perianth tube filiform, 5-6 mm long; tepals obovate, 12-14 × 5-7 mm.Stamens symmetri-cally arranged; filaments filiform, inserted at mouth of narrow part of tube, ± 1.5 mm long, deep pink (drying purple); anthers suborbicular, 2.0-2.5 mm long, yellow, thecae obtuse basally, dehiscing completely or almost so by narrow slits.Style dividing at base of filaments, style branches 3.8-4.0mm long, deep pink (drying purple), ± straight, extending outward below anthers, expanded and bilobed at tips.Capsules globose, ± 4 mm diam.Flowering time: mostly late Aug. to mid-nov.but Dec. and early Jan. in the Caledon Swartberg.Figure 6.

Representative specimens
Distribution and habitat: restricted to the Caledon District, Ixia bifolia is known from the northern slopes of Shaw's Mtn south of Caledon and from the lower slopes of the Swartberg east of Caledon (Figure 3).The soil at the Shaw's Mtn site is hard, stony clay and the habitat there is shared by the rare Tritoniopsis flexuosa (L.f.) g.J.Lewis (known from just two other localities).With its narrow range and just two known populations, one small and near a major road, we suggest a conservation status of (Vu), Vulnerable.
Diagnosis and relationships: Ixia bifolia seems at first to belong in the I. micrandra-confusa complex, which also has very slender stems and small flowers but it stands out in the very short, leathery-succulent foliage leaf blade, up to 80 mm long and 5-7 mm wide.The leaf has a single, prominent vein and moderately thickened margins.The flowers seem most like those of I. confusa but the short, deep pink filaments ± 1.5 mm long and similarly deep pink, straight style branches ± 4 mm long (both drying purple) are quite different from I. confusa, which has white filaments 1.3-2.5 mm long and white, strongly falcate style branches 1-2 mm long.We suspect that I. bifolia may be more closely allied to I. trinervata, which has slightly larger flowers on longer spikes but also a single foliage leaf, this relatively broad and always with three prominent veins and margins not thickened.I. bifolia shares with other members of the I. micrandra complex the finely fibrous corm tunics that extend upward around the stem base.Stems are almost always unbranched and have an odd feature, a small, nearly microscopic, translucent scale and prophyll, ± 1 mm long, inserted 12-30 mm below the base of the spike.Stems of I. confusa may be branched and are normally taller than in I. bifolia but also have a translucent stem scale in populations from the Swartberg and sometimes elsewhere across its range.I. confusa has 3 or 4 linear leaves, usually ± 2 mm, but occasionally to 4 mm wide.Plants (150-)250-500 mm high, stem simple or 1-branched, with short scale ± 1 mm long, inserted 25-40 mm below base of spike, sheathed basally by green, leathery to firm, dry cataphylls.Corm 8-10 mm diam., tunics of fine, soft fibres sometimes extending upward in a weakly developed collar.Leaves (2)3(4), lower (1)2 or 3 with linear to narrowly sword-shaped blades, ± 1/3 as long to reaching top of stem, 1.5-4.0mm wide, upper leaf largely sheathing, with short free blade.Spike slightly flexuose, (2-)5-9(-12)-flowered; bracts membranous, becoming dry from tips, 4-5 mm long, outer 1-or 3-veined, irregularly 2-3(-5)-lobed or shortly toothed, inner often slightly longer than outer, 2-veined and 2-toothed.Flowers rotate, upright, pink to light red-purple, rose-scented; perianth tube filiform, 3-4 mm long; tepals spreading, ± ovate, mostly 12-15 × 6-7 mm.Stamens symmetrically arranged, erect; filaments inserted at mouth of tube, filiform, 1.0-2.8mm long; anthers oblong, 2.0-2.3 mm long, yellow, thecae acute and recurved at base, usually dehiscing longitudinally from base almost to apex.Style dividing at or up to middle of filaments, style branches 1.5-2.0mm long, falcate, tubular with expanded ciliate, stigmatic tips, sometimes shortly forked apically.Flowering time: mostly Jul.-early Dec.

Representative specimens
Distribution and habitat: Ixia confusa occurs in the mountains of southern Western Cape, from near Elim and the Riviersonderend Mtns in the west through the Langeberg to the Paardevlei Mtns near Farm Bonnievale at the western end of the Outeniqua Mtns (Vlok 709 nBg, PRE), and inland in the great Swartberg (Figure 7) where it has been recorded from Swartberg Pass, Meiringspoort, and Blesberg slightly further east.An outlying collection (Marloth 2429 PRE) from Montagu Pass near george extends the range slightly south.We have encountered plants (Goldblatt & Manning 12946) on moist southwest-facing slopes in loamy clay in renosterveld-thicket transition vegetation.The flowers of our collection were strongly rose-scented.We suspect that plants flower particularly well after fire but our collection was made in veld that had not burned for over 10 years.Occurring mostly in montane habitats, there seems no current threat to the species and we agree with Raimondo et al.'s (2009) (LT, as I. micrandra var.confusa), Least Threat conservation status.
Diagnosis and relationships: known to g.J. Lewis (1962) as Ixia micrandra var.confusa, we raise the taxon to species rank after examining living plants and comparing them with I. micrandra.Both I. micrandra and closely related I. minor (= I. micrandra var.minor g.J.Lewis) have corms with finely fibrous tunics that extend upward in a collar around the base of the stem, terete to subterete or linear leaves mostly 1.0-1.5 mm at widest, and spikes of up to 6 flowers.I. confusa has corms with similar finely fibrous tunics, sometimes without a collar of fibres around the base, but plane leaves 1.5-4.0mm wide, and spikes with as many as 12 flowers.I. micrandra has anthers 3-4 mm long, and slender, ± straight, ascending style branches 3-4 mm long, thus quite different from the shorter anthers, ± 1.0-2.5 mm long, and falcate style branches 1.5-2.5 mm long of I. confusa.The style branches of I. confusa match those of I. minor but the terete or subterete leaf blades, flattened filaments and consistently few-flowered spikes of the latter immediately separate them.
Ixia confusa may also be confused with the new I. bifolia but this species has a single, relatively broad, leathery-succulent foliage leaf and flowers with shorter filaments ± 1 mm and longer, straight style branches ± 4 mm long.Both the filaments and style branches of I. bifolia are dark pink and dry dark purple, unlike the pale filaments and style branches of I. confusa, which dry to a pale, almost colourless state.
Annotations by g.J. Lewis on collections of Ixia confusa from the Swartberg indicate that at one time she considered these populations a separate species, which she intended to call I. aestivalis, in part because of their late flowering, from late October to early December.We find that collections from the Swartberg also differ in having slightly shorter filaments, 1-2 mm long, than typical I. confusa (1.5-2.8 mm long), the style sometimes dividing above the mouth of the perianth tube (opposite the middle of the filaments in Goldblatt 7962), and anther dehiscence sometimes not reaching the middle of the thecae.The plants have a small, translucent scale and prophyll on the upper part of the stem, a feature occasionally present in southern populations of I. confusa.Additional study may show that the later-flowering Swartberg populations merit separate taxonomic status but we find no consistent character differences between them.
Distribution and habitat: extending from Houw Hoek, Bot River and the Caledon Swartberg to Bredasdorp and De Hoop (Figure 7), Ixia micrandra typically occurs on sandstone slopes and flowers well only in seasons immediately after veld fires.A collection from De Hoop is from pockets of sandy soil in limestone pavement, an unusual habitat for the species.Largely montane, we see no current threat to the species and endorse Raimondo et al.'s (2009) (LT), least threat, status.
Diagnosis and relationships: treated by g.J. Lewis (1962) as including the two additional varieties, var.confusa and var.minor, we have no doubt that I. micrandra is allied to these two taxa, which are here raised to species rank.I. micrandra stands out in the group in its long, narrow, almost linear filaments and its style branches, which are unusually long at 2.0-3.5 mm, and straight or slightly arched, thus extending between the bases of the anthers.I. confusa has similarly slender but somewhat shorter filaments, slightly shorter anthers, and shorter, strongly arched style branches, 1.5-2.0mm long.The leaves of I. confusa have plane, usually wider blades up to 4 mm wide (vs.0.5-1.5(-2.0mm) and without the prominently thickened main vein of M. micrandra.The style branches are also strongly arched in I. minor but that species has unusual flattened, lan-ceolate filaments, narrowed and kinked at the tips.The ranges of I. micrandra and I. minor are almost complementary, that of I. micrandra extending east of I. minor, but the two have been collected growing and flowering on the same day in the Houw Hoek Mtns, where their distributions overlap.
Distribution and habitat: a narrow endemic of the southwestern Cape, Ixia minor occurs from Franschhoek Pass and the Wemmershoek Mtns south though the Hottentots Holland and Kogelberg Mtns to Houw Hoek (Figure 7).The habitat, stony sandstone slopes, is the same as that for I. micrandra and as in that species we see no current threat and suggest a least threat (LT) conservation status.
Diagnosis and relationships: Ixia minor, treated as var.minor of I. micrandra by Lewis (1962) but raised here to species rank, is evidently closely allied to the more widespread I. micrandra, with which it is easily confused unless floral details are examined carefully.Both share a slender, usually unbranched stem, small corms with fine, netted tunics that extend upward in a collar around the base, and a dry, often brown, upper cataphyll.The critical differences between the two species lie in their flowers: I. minor has flattened, ± wedgeshaped to oblanceolate filaments abruptly narrowed at the tips, shorter and broader anthers, often not dehiscing to the apex and kinked at the base with the bases of the thecae turned inwards and touching, and shorter, wider, strongly arched style branches slightly compressed and narrowed at the tips.To these microscopic details can be added a deep pink perianth.Flowers of I. micrandra have pale pink to purple-pink tepals, linear filaments, longer, oblong anthers dehiscing to the apex and longer, more slender style branches, ± straight to slightly arched.
The leaves of the two species also differ: those of Ixia minor are subterete to oval in section, thus when alive without margins or evident central vein, whereas I. micrandra has plane, linear leaf blades with somewhat thickened margins and, when dry, a raised main vein.Floral differences between I. minor and I. micrandra are illustrated in detail in the accounts by both Lewis (1962) andDe Vos (1999).A color photograph of I. minor (Manning et al. 2002: 242) under the name I. micrandra clearly shows the short, recurved style branches of the species.Compelling evidence for their separation is provided by collections of both species made by the late J. Beyers in the Houw Hoek Mtns flowering on the same date and sufficiently different in appearance that Beyers gave them different collection numbers.Habitat differences, if any, were not recorded.73 (1999).Tritonixia stricta Eckl.ex Klatt: 357 (1882).Tritonia scillaris var.stricta (Eckl.ex Klatt) Baker: 191 (1892).Type: Western Cape, Caledon, Swartberg and vicinity of the Baths, December 1830, Zeyher 4008 (SAM, lecto.!, designated by Lewis: 162 (1962); K, PRE!, isolecto.).
Distribution and habitat: restricted to the Caledon-Bredasdorp area of Western Cape, Ixia stricta extends from the Houw Hoek Mtns in the west to Bredasdorp in the southeast (Figure 1).Typically flowering late in the year, in november and December, I. stricta is poorly collected.not surprisingly the foliage is often more-orless dry at flowering.Plants occur mostly in montane habitats, but usually at fairly low elevations, in sandstone and loamy clay soils.Raimondo et al. (2009) suggest a conservation status of (LC), least concern, but we suggest that (nT), near threatened, is more appropriate.
Diagnosis and relationships: while not particularly distinctive in its radially symmetric flowers with short, erect stamens, Ixia stricta is identified in ser.Dichone by a combination of the rotate flowers and a basal fan of (3)4 or 5 sword-shaped leaves, the blades with a prominent main vein and thickened margins.The sweetly scented flowers have a filiform tube about half as long as the tepals and a particularly short style ± 1 mm long.Blooming late in the season, in november and December, the leaves are often dry or partly so, a feature unusual in the genus.
History: the plant illustrated by Jacquin (1792) and misidentified as Ixia polystachya L. is the earli-est record of Ixia stricta and this illustration is also one of two sources cited by Ker gawler (in Sims 1801) for his I. scillaris var.angustifolia.Both Lewis (1962) and De Vos (1999) associated Jacquin's illustration with I. stricta and the illustration shows clearly the relatively broad, prominently veined basal leaves and the sheathing upper leaves subtending straight, several-flowered branches.Ker gawler's varietal name angustifolia is nomenclaturally illegitimate and superfluous because he included the earlier name I. scillaris var.incarnata, this now most likely I. scillaris subsp.latifolia.
Watsonia retusa (Salisb.)Klatt ( 1882), a combination based on indirect citation of the basionym Ixia retusa Salisb., was regarded by Lewis as a synonym of I. micrandra var.confusa.De Vos (1999), however, associated the species with I. stricta and we concur.Klatt evidently intended to provide a name for the plant called I. polystachia [sic] by Ker gawler (1803) in Curtis's botanical magazine but as Ker gawler specifically mentioned I. retusa Salisbury (1796) in his discussion of the plate in question, Klatt's name must be treated as a combination, but with a basionym that has no known type.Salisbury's brief protologue matches the plant depicted in Curtis's botanical magazine published seven years later in all respects, notably the style branches emerging between the filaments and the tepals twice as long as the perianth tube.But there is no connection between Salisbury's species and the painting.12. Ixia erubescens Goldblatt in Journal of South African Botany 37: 233 (1971b), as a new name for I. crispa L.f.: 91 (1782), a superfluous name for I. undulata Burm.f.Agretta crispa Eckl.: 24 (1827), as a new name for I. crispa L.f.Tritonia thunbergii n.E.Br.: 137 (1929), as a new name for Ixia crispa in Tritonia, non T. crispa (L.f.) Ker gawl.Type: South Africa without precise locality, Thunberg s.n.[LInn 58.25, lecto., designated by Lewis: 173 (1962)].
Distribution and habitat: predominantly a species of near southwestern Western Cape, Ixia erubescens extends from Tulbagh southwards to Stellenbosch and the Caledon District (Figure 8) where its southeasternmost station is at Franskraal, but is absent from the Cape Peninsula.Outlying populations occur on the lower slopes of the Piketberg.The species is most often encountered on heavy clay or granitic alluvium in seasonally damp places, where plants typically form tight clusters through the germination of daughter cormlets.The Piketberg plants are restricted to cooler south-facing slopes where they grow on well-drained clay slopes in the shade of shrubs and small trees, and grow singly rather than in clusters.We concur with the conservation status of I. erubescens, listed by Raimondo et al.(2009) as (LC), least concern, although we note that its range has been historically much reduced through agriculture, especially in the Swartland and Boland.
Diagnosis and relationships: a relatively small species, Ixia erubescens is immediately recognized by the fine corm tunics, the 4-6 narrow, deeply undulate and crisped leaves, usually unbranched stem, and the vertically oriented flower with the stamens unilateral and slightly pendent when fully open.Both Lewis (1962) andDe Vos (1999) regarded the flowers as fundamentally actinomorphic with symmetrically arranged stamens and noted that only late in anthesis do the anthers become unilateral and pendulous, simply due to gravity.Our own observations counter this: the stamens are unilateral immediately the flower buds open.We assume that I. erubescens is most closely allied to I. scillaris, which typically has medium to coarse corm tunic fibres, broader leaves, usually with plane margins, and somewhat to substantially larger flowers, also with unilateral stamens, but with a perianth tube up to 4.5 mm long and tepals up to 14 mm long (vs.tube 2-3 mm long and tepals 11-12 mm long).Occasional populations of I. scillaris with undulate or crisped leaves are readily distinguished from I. erubescens by their broader leaves, larger flowers, and coarser corm tunic fibres.
The taxonomic history of Ixia erubescens is somewhat confused.Linnaeus fil.(1782) described I. crispa based on C.P. Thunberg's collection but cited the earlier name I. undulata Burm.f.(Burman 1768) (now Tritonia undulata) in synonymy.According to current rules of nomenclature this rendered I. crispa superfluous and therefore illegitimate.The two were often confused by later authors and Baker (1877) used the name Tritonia undulata for the species, not realizing that its type is a different plant.Brown (1929) was the first to realize that I. crispa and I. undulata were different species and transferred I. crispa L.f. to Tritonia under a new epithet, T. thunbergii, because he believed that the name T. crispa (L.f.) Ker gawl.(based on Gladiolus crispus L.f.) barred a combination based on Linnaeus fil's I. crispa.On transferring sect.Dichone to Ixia, Lewis (1962) used the name I. crispa for the species.Then goldblatt (1971b), realizing that I. crispa was superfluous and thus illegitimate, established the new name I. erubescens.Transfer to Ixia of the legitimate T. thunbergii n.E.Br. is prevented by use of that epithet in Ixia by Roemer & Schultes (1817), their species being a synonym of T. securigera Ker gawl.13.Ixia scillaris L., Species plantarum ed. 2, 1: 52 (1762); Lewis: 173 (1962);De Vos: 78 (1999).Tritonia scillaris (L.) Baker: 163 (1877).Tritonixia scillaris (L.) Klatt: 357 (1882).Type: South Africa, without precise locality or collector [LInn 58.25-digital image!, lecto., designated by Lewis: 174 (1962)].
Distribution and habitat: predominantly a species of the west coast and near interior of Western Cape, Ixia scillaris extends from the Cape Peninsula north to the gifberg and Kobee Mtns, along the coastal plain and mountain ranges parallel to the coast as far inland as the Cedarberg and the Pakhuis-nardouw Mtns.Isolated populations occur in the Spektakel Mtns in northern namaqualand, northern Cape (Figure 10).Plants grow on a variety of soils but most commonly on heavy clays derived from shale or granite.With so wide a range its conservation status (LC), Least Concern, seems appropriate.The range of subsp.scillaris is, however, much reduced and may be Vulnerable (Vu).
Studies of the floral biology of Ixia scillaris (goldblatt et al. 2000) show the flowers are adapted for buzz pollination by anthophorine bees.Pollen remains in the anthers until vibrated at high frequency by the wings of visiting bees.unusual for flowers with this pollination strategy, the pollen falls back into the centre of the flower through narrow slits at the base of the anthers.Like all species of sect.Dichone, the flowers do not produce nectar.
Diagnosis and relationships: the most common and widespread species of sect.Dichone, Ixia scillaris is recognized primarily by the nodding flower with vertically oriented tepals with unilateral, slightly pendent and often relatively long anthers, combined with a fan of 3-7 basal leaves, (3)6-20 mm wide.Related I. erubescens has narrow, ± falcate leaves, 2-4(-6) mm wide, with closely crisped margins, and is a smaller plant with smaller flowers, the tepals 8-11 × 4-6 mm and anthers 2.0-2.5 mm long vs. tepals mostly 10-15 × 5-9 mm and anthers 3-4 mm long in I. scillaris.Apparently closely allied Ixia simulans, with half to fully nodding flowers and unilateral stamens with pendent anthers, has a single basal, ± linear foliage leaf and a second leaf sheathing the stem for most of its length.Ixia tenuis from the Piketberg is a particularly slender species with small corms with finely fibrous tunics, just three, narrow, softtextured leaves and relatively small flowers with dark magenta tepals.I. tenuis grows in a habitat unique for species allied to I. scillaris, seeps on sandstone rocks, growing in moss in soils less than 10 mm deep.
Variation: Typical Ixia scillaris from Darling and Tulbagh to the Cape Peninsula has (4)5-7 relatively narrow leaves, mostly 5-8 mm wide, with 3-5 secondary veins conspicuous only when dry, mostly branched stems (rarely with up to 7 branches), and moderately dense spikes with flowers spaced 8-12 mm apart.The flowers are relatively small, with tepals 8-10 × 4-5 mm and ± symmetrically arranged, although the lowermost tepal is held slightly apart and recurved, and anthers

C
Bothalia 42,2 (2012) 2.0-2.8mm long.In contrast, plants from the Olifants River Valley and low elevations in the Cedarberg ranging into the Bokkeveld and Spektakel Mtns are always taller, with 2-4, broader leaves mostly 15-20 mm wide, usually with at least two conspicuous secondary veins in addition to the main vein, stems simple or with 1(2) branches, and lax spikes with flowers distantly spaced, 15-25 mm apart.The flowers are larger, with tepals 14-16 × 5-9 mm, the upper tepals overlapping and separated from the three lower tepals, and anthers 3.2-4.0mm long (Table 1).Although Lewis (1962) noted the variation in leaf number and venation between the southern and northern populations she preferred not to recognize the two forms taxonomically as she believed them to intergrade in leaf form and number to some extent.She did not, however, note that the distinction was accompanied by differences in size of the flowers, which we consider as significant as the leaf differences.
A third set of divergent populations occurs on the gifberg-Matsikamma Mtn complex.These consist of unusually small plants, seldom exceeding 150 mm in height, with small flowers (tepals ± 8 × ± 4 mm, anthers ± 2 mm long), thus like those of the southern populations of Ixia scillaris but with just three leaves (rarely two in depauperate individuals), recalling the northern populations, at least in leaf number, although the leaves are shorter and narrower.
Our taxonomic solution to the variation in Ixia scillaris is to recognize the three sets of populations as separate subspecies, those of the larger-flowered, northern populations as subsp.latifolia and the gifberg-Matsikamma populations as subsp.toximontana.We choose subspecies rank because of the small overlap between some populations of each subspecies and the comparatively small degree of difference between them relative to that between species of sect.Dichone.
In addition to these major variants, Lewis recognized populations with undulate leaves as var.subundulata.The type of the variety is from east of Pakhuis Pass (Barker 4727) and there are several more collections from this area.Plants with undulate leaves also occur locally near Tulbagh.Flowers of the Pakhuis-northern Cedarberg have the relatively large flowers that correspond with subsp.latifolia whereas those from the south have smaller flowers of matching subsp.scillaris.We see no value in recognizing undulate-leaved plants taxonomically.undulate leaf margins have evidently evolved independently in different parts of the range of Ixia scillaris.
History: the taxonomic history of Ixia scillaris begins with a brief polynomial entry in A. van Royen's (1740) Florae leydensis prodromus, an account of the plants grown in the botanical garden of Leiden university.Linnaeus (1753), in the first edition of the Species plantarum, adopted van Royen's species and, using his actual phrase name, provided the specific epithet ramosus to the description Gladiolus caule ramoso, foliis linearibus.The specimen is still at the Leiden Herbarium.Then, in the second edition of his Species plantarum, Linnaeus (1762) described Ixia scillaris, based on a specimen in his own herbarium.Evidently Linnaeus simply did not relate this specimen to the earlier name, although he may well have seen Van Royen's specimen or the living plants when he was in The netherlands some 15 years earlier.Gladiolus ramosus was forgotten for many years but was listed by Ker gawler (1827) as a synonym of Melasphaerula graminea (L.f.) Ker gawl.and by Lewis et al. (1962) as a synonym of the same species under the name M. ramosa (Burm.f.) n.E.Br.
Distribution and habitat: restricted to the southwestern Western Cape, subsp.scillaris extends from Tulbagh and Darling to the Cape Peninsula and Somerset West (Figure 10).The only record outside this area, a collection from Hawston near Bot River, Wilman s.n.nBg, is the Clanwilliam form of Ixia scillaris, subsp.latifolia, and we assume the label is incorrect.Subsp.scillaris is common and fairly conspicuous where it occurs, mainly on granite-derived sandy gravels.
Distribution and habitat: subsp.latifolia is centred in the Olifants River valley and lower elevations of the surrounding Cedarberg, Pakhuis-Bidouw Mts, extending from near Citrusdal to Bulshoek and locally in the Bokkeveld Mtns.Isolated populations occur on the Spektakel Mtns of northern namaqualand, a notable disjunction (Figure 10).It is replaced on the gifberg/ Matsikamma Mtns by subsp.toximontana.Plants favour rocky situations, mostly in sandy ground.Subsp.latifolia is common throughout its range, but it flowers well only after fire or clearing of the veld.The spring following a fire plants are often seen in masses where before the fire none were evident.
Distribution: restricted to the gifberg and Matsikamma Mtn complex of northern Western Cape, subsp.toximontana grows in shallow, sandy ground overlying sandstone pavement (Figure 10).The habitat is permanently wet in the rainy winter and early spring months but becomes dry by the end of September and remains so though the summer and autumn.Most of its range is virtually intact and pristine at present.Diagnosis: subsp.toximontana is distinguished in Ixia scillaris by its low stature, seldom exceeding 150 mm.combined with relatively small flowers and only 2 or 3 leaves, these short and relatively narrow, 3-8 mm wide.The flowers have tepals ± 8 × 4 mm and anthers ± 2.3 mm long, smaller than in subsp.latifolia or subsp.scillaris (Table 1).In their small size they recall subsp.scillaris rather than subsp.latifolia, in whose range it is nested.The particularly small size of all parts may be an adaptation to the habitat, nutrient-poor, shallow, sandy ground.Plants slender, 180-350 mm high, stem simple or rarely 1-branched.Corms globose, 7-10 mm diam., tunics of fine fibres.Leaves mostly 3(2), occasionally 4 but then this very reduced in size, ± linear to narrowly sword-shaped, 1/4-1/2 as long as stem, mostly 1.5-3.0mm wide, relatively soft-textured, pale grey-green, when alive main vein slightly raised (appearing thickened when dry), margins not raised (rarely undulate, then blade up to 5 mm wide-Barker 4671, 5764, see comments below).Spike mostly 4-10-flowered, flowers mostly 10-15 mm apart; bracts 2-3 mm long.Flowers deep magenta with tepal bases yellow edged with darker magenta (or pale pink or white-Barker 4671, 5764), throat pale yellow, perianth tube 3.0-3.5 mm long, expanded in upper 1 mm; tepals 8-9 × 4-5 mm, oriented vertically but with lowermost tepals recurved and ± horizontal.Stamens unilateral; filaments ± 2 mm long; anthers 2.0-2.5 mm long.Style branches ± 1 mm long, reaching to ± middle of filaments.Flowering time: mid-Oct.-mid-nov.

Representative specimens
Distribution and habitat: known from just a handful of sites, Ixia tenuis is restricted to rocky, sandstone habitats in the Piketberg Mtns, mostly at elevations of 650-800 m (Figure 10).We collected the type population on the Farm noupoort on south-trending slopes of a rocky hill growing in wet moss on shallow soil on sandstone outcrops.The remaining collections lack detailed habitat notes.We note that the undulate-leaved variant (see below) is from the foot of the Piketberg Range and blooms earlier, in September.The species is rare, and at present merits a conservation status of Vulnerable (Vu), but we see no imminent threat.
Diagnosis: Ixia tenuis is distinguished from the widespread I. scillaris by its very slender habit, the stems rarely branched, usually three, narrow leaves, mostly 1.5-3.0mm wide, and relatively small flowers with tepals 8-9 × 4-5 mm and anthers 2.0-2.5 mm long (Table 1).The corms are also relatively small, mostly less than 10 mm in diameter and with relatively fine tunic fibres.Corms of I. scillaris are larger, typically 12-18 (rarely only 10) mm in diameter, with relatively coarse tunic fibres, and the species has broader leaves, mostly 8-25 mm wide.We provisionally include collections by W.F. Barker (Barker 4671 and 5764) from De Hoek and flowering in September here.These specimens have slightly broader but shorter leaves than typical I. tenuis and have undulate margins.This variant, which we have not seen alive, was included by Lewis (1962) as I. scillaris var.subundulata, the type of which we include in I. scillaris subsp.latifolia.Plants 150-330 mm high, stem simple, rarely with 1 branch.Corm small, up to 10 mm diam., bearing stolons up to 50 mm long from base, with a large terminal cormlet up to 5 mm diam., tunics finely fibrous to moderately coarse.Leaves 4, uppermost sheathing for most of its length, foliage leaves with blades narrowly sword-shaped to falcate, ± 1/3 as long as stem, mostly 2-4 mm wide, main vein prominent, margins slightly thickened.Spike mostly 5-9-flowered; bracts translucent with purple veins, 3-4 mm long, outer 3-veined, acutely 3-toothed, inner slightly shorter than outer, 2-veined and 2-toothed.Flowers evidently zygomorphic, nodding with tepals vertical, pink, throat and tepal bases yellow edged with deep pink; perianth tube filiform, ± 3 mm long; tepals ovate, ± 9.0 × 4.5 mm.Stamens evidently unilateral; filaments slightly flattened, ± 2 mm long; anthers oblong-orbicular, ± 3 mm long, yellow, thecae recurved and acute at base, splitting incompletely from base.Style dividing at base of filaments, branches ± 1.8 mm long, falcate, slender.Flowering time: Oct. and nov.
Distribution and habitat: restricted to middle and upper elevations in the Cedarberg, Ixia flagellaris is currently known from just two collections from the Cedarberg, one from Heuning Vlei in the north and the other from Sneeuberg in the south (Figure 8).The type collection was made the season after a fire.The species evidently grows on damp flats, most likely on peaty sandstone ground, but this remains to be established.A collection from Kobee Pass (see below) is provisionally included here.
Diagnosis and relationships: the outstanding feature of Ixia flagellaris is the corm, which is remarkable in the I. scillaris complex in the finely fibrous corm tunics and in the production of slender stolons (Latin: flagellae) up to 50 mm long, each terminating in a relatively large cormlet.no other member of sect.Dichone produces stolons although they are known in a few species of sect.Ixia and Morphixia (goldblatt & Manning 2011).I. flagellaris otherwise accords closely with I. scillaris subsp.toximontana, notably in its relatively low stature, 150-330 mm high, and small flowers with tepals ± 9.0 × 4.5 mm, but differs in lacking prominent, thickened leaf margins..The late flowering and moist habitat, seeps and marshy sites, are unusual for the complex, although known in the Piketberg species, I. tenuis.Plants from Kobee Pass (Hall 4517) cited below may belong here as they have fine corm tunics and unthickened leaf margins but they evidently do not produce stolons.Their exact status remains uncertain.Plants 40-100 cm high, stems usually simple, rarely with a single branch, sheathed basally by membranous cataphylls.Corm ± 10 mm diam., tunics of fine, netted fibres.Leaves 2(3), lower 1(2) with narrowly swordshaped to linear blade with single prominent vein, ± 1/2 as long as stem, 1.5-6.5 mm wide, margins thickened and slightly raised, hyaline when dry; upper leaf sheathing lower 1/2 of stem, free and unifacial distally.Spike erect, fairly lax, weakly flexuose or straight, 4-8-flowered; bracts membranous, translucent, purple at tips, ± 4 mm long, outer usually with 3 prominent veins and 3-toothed, inner 2-veined and forked at apex.Flowers ± nodding with tepals held vertically, zygomorphic with stamens unilateral, bright pink (rarely pale pink), white at tepal bases, unscented; perianth tube filiform, ± 4.5 mm long, tightly clasping style; tepals spreading, ovateelliptic, (10-)12-14 × 5-6 mm, inner whorl slightly larger than outer.Stamens unilateral, filaments inserted at mouth of tube, filiform, ± 2.5 mm long, white, anthers oblong, 2.0-2.5 mm long, yellow, horizontal or slightly pendent, thecae acute and slightly recurved at base, dehiscing incompletely from base by narrow slits.Style dividing opposite lower 1/3 of filaments, branches ± 1.3 mm long, white, tubular, falcate, stigmatic apically.Flowering time: mid-Aug.-mid-Oct.Figure 11.

Representative specimens
Distribution and habitat: Ixia simulans is currently known from just four collections, three from the western Langeberg, between Dassieshoek near Robertson and Swellendam, and one from Lemoenpoort in the northern foothills of the Riviersonderend Mtns (Figure 10).Records are all from stony ground, and at the Keisie River gorge near Montagu hot springs plants grew on steep, rocky, south-facing slopes.Plants are rooted in shallow sandy soil, often growing in clumps of Restionaceae.All known localities for I. simulans are in virtually undisturbed montane vegetation and we see no threat to the species: a conservation status of Least Concern (LC) seems appropriate.
Diagnosis and relationships: Ixia simulans is a surprising novelty given that the western Langeberg has been botanized for over two centuries and intensely so in the past 50 years.The flowers closely resemble those of I. scillaris in general appearance and have unilateral, horizontal to slightly pendent anthers that dehisce incompletely by narrow slits near the base.I. simulans differs from I. scillaris most conspicuously in its single, or rarely two, long, basal leaves with ± linear blades 1.0-4.5 mm wide, spikes of only 4-8 flowers, relatively short floral bracts ± 4 mm long and the spreading, subequal tepals, those of the inner whorl being slightly shorter than the outer.The tepals are inclined and all held in the same plane whereas those of I. scillaris are always held vertically and the lowermost tepal is weakly to strongly recurved and held apart from the other five tepals.
The range of Ixia simulans and I. scillaris complement one another: I. scillaris extends from the Cape Peninsula and Somerset West in the south through the Olifants River Valley, Cedarberg-Pakhuis Mtn complex to the Bokkeveld Mtns, and locally well to the north in the Spektakel Mtns west of Springbok whereas I. simulans occurs well to the east of this line in the western Langeberg.globose, 7-12 mm diam., tunics of medium-textured fibres, often with cormlets at base.Leaves 3 or 4, lower 2 or 3 with sublinear to narrowly sword-shaped blades, ± 1/2 as long as stem, 4-8 mm wide, main vein moderately thickened, uppermost leaf sheathing lower half of stem.Spike flexuose, 7-16-flowered, lateral spikes with fewer flowers; bracts membranous, translucent, 4.5-7.0mm long, outer with 3 prominent, purple veins, 3-or 5-toothed, inner 2-veined and 2-toothed.Flowers zygomorphic, half nodding, pink, white to yellow edged deep pink at tepal bases, sweetly scented, perianth tube filiform, ± 5 mm long, expanded in upper 0.5 mm, tepals subequal, ovate, ± obtuse, 9-12 × ± 5 mm.Stamens unilateral, ± horizontal, inserted at mouth of filiform part of tube, filaments ± 3 mm long, filiform below, curved at right angles in upper 1 mm and becoming flattened and as wide as anthers, anthers ± 3 mm long, oblong, yellow, thecae acute and recurved near base, dehiscing incompletely from base.Style dividing at base of filaments, style branches ± 2.5 mm long, tubular, falcate, bifid at apex, stigmatic apically.Flowering time: late-Aug.-midSept.

Representative specimens
Distribution and habitat: Ixia collina is a narrow endemic of the mid Breede River Valley, its only known locality the Farm Alfalfa near Aan de Doorns south of Worcester (Figure 8).Plants grow on the southern slopes of low clay hills above the Breede River flood plain and adjacent to cultivated land.With so narrow a range Raimondo et al. (2009) regard the species as endangered (En) and we agree.We recommend efforts be made to locate additional populations but there is little doubt that I. collina is extremely rare.
Diagnosis and relationships: Ixia collina is easily distinguished from other members of ser.Dichone by the unusual branching pattern: the relatively short branches extend outward horizontally and curve upward distally as well as by an unusual floral feature.The filaments are unique in the section in being filiform below, curved at right angles in the distal 1 mm and broadened so that they are as wide as the anthers.

EXCLuDED SPECIES
Ixia retusa Salisb., Prodromus stirpium in horto ad Chapel Allerton vigentium: 35 (1796).The identity of this species has remained a mystery as the description is inadequate to identify the plant (De Vos 1999) and no type is known.Lewis (1962) suggested I. retusa might be I. stricta but with a query.Ker gawler (in Sims 1801; 1803) associated the name with what he called Ixia polystachya (actually I. scillaris, thus not the same as Linnaeus's species) [see extensive discussion under Taxonomic History].Although we suspect that I. retusa is a member of sect.Dichone, it is impossible to tell which species and the name must be rejected.Likewise, Klatt's combination, Watsonia retusa (Salisb.)Klatt is rejected (see additional discussion under Ixia stricta).
toothed, margins connate in lower 1.5 mm around ovary.Flowers rotate, light purple with small dark purple stain in centre, lightly scented; perianth tube ± funnel-shaped, ± 2.3 mm long, filiform below and clasping style, flaring in distal 1 mm; tepals spreading and slightly concave distally, obovate, ± narrowed below into a short claw, 12-13 × 7-8 mm.Stamens symmetrically arranged, erect, filaments inserted at apex of tube, diverging above, 2.5 mm long, blackish purple, anthers ± oblong, 4.5-5.0mm long, blackish purple, dehiscing laterally by narrow slits along entire length.Style dividing at or slightly above base of filaments, ± 2 mm long, branches filiform, involute, stigmatic at apex, purple, falcate, 2.0-2.5 mm long.Flowering time: late Sept. to early Oct. Figure4A-D.Distribution and habitat: Ixia brevituba is known from a small area on the edge of the Roggeveld Escarpment near ganagga Pass southwest of Middelpos (Figure 3).The few known collections were made within a few kilometres of each other on Farms Zoekop and Agterkop (the type is not localized exactly but is very likely from the same area).Plants grow in stony, light clay in renosterveld.The flowers open widely around
Diagnosis and relationships: relatively low growing, Ixia curvata is a small plant, seldom exceeding 200 mm in height (although taller plants are known) and has nodding, pink flowers with tepals held vertically with stamens borne horizontally.Like other members of ser.Euanthera, the anthers split completely.The four or five leaves are typically falcate and leathery rather than fibrotic.Its immediate relationships remain uncertain.Other species of the series have upright flowers with tepals spreading horizontally.The plants are notable in the thick collar of fibres around the base and chestnutbrown cataphylls.The species was first collected by Rudolf Marloth in October 1920, but the label data localizing the plants from 'plains near Tulbagh' is an obvious mistake.Marloth collected in the Roggeveld and nearby in October 1920 when he also made the first collections of Ixia rigida (initially assigned to I. brevituba) and I. trifolia, thereby establishing the presence of sect.Dichone in the Western Karoo.Acocks 18504 (PRE).3120(Williston):35 km SE of Calvinia to Middelpos on Blomfontein road, (-CA), 30 Sept. 1976, Goldblatt 4281 (MO).uncertainlocalities:'Plains near Tulbagh,' Oct. 1920, Marloth 9936 (PRE);Van Rhyns Pass, 15 Sept. 1953, Taylor  3967 (nBg).
Stamens symmetrically arranged (Ixia stricta) or unilateral and horizontal to pendent; anthers oblong, recurved and acute at base, dehiscing incompletely from base.Style branches always falcate.7 spp., centred in the western half of Western Cape, but with isolated populations in northern namaqualand and the Bokkeveld Mtns of northern Cape.11.Ixia stricta (Eckl.ex Klatt) G.J.Lewis in Journal of South African Botany 28: 167 (1962); De Vos: Ser. DichoneFlowers upright or nodding, radially symmetric (Ixia stricta) or zygomorphic.
Variation in leaf morphology in subsp.latifolia is notable.Plants from Clanwilliam and nearby have a conspicuous submarginal vein, but those from elsewhere have a prominent secondary vein pair distant from the margins, and plants from the Spektakel Mtns of namaqualand show no prominent secondary veins and appear thinner in texture.no other feature seems to set the Spektakel population apart.Populations east of the Pakhuis-nardouw Mtns have leaf margins slightly to markedly undulate and without conspicuous veins apart from the main vein.