New species of Sparaxis and Ixia (Iridaceae: Ixioideae) from Western Cape, South Africa, and taxonomic notes on Ixia and Gladiolus

Sparaxis auriculata is a new species of this western Cape and western Karoo genus It resembles S. villosa (Burm.f.) Goldblatt in its (lowers but is probably allied to S. #aleata (Jacq.) Ker Gawl. Sparaxis metelerkampiae, currently a sub­ species of S. variegata Sweet is raised to species rank. We propose changes to the infrageneric classification of the southern African winter rainfall genus Ixia. and describe a new species, /. aurea It is related to I odorata Ker Gawl. but differs in the larger deep yellow or orange flowers which are unscented In Gladiolus, we propose the new name. G saxatilis. for the Mpumalanga endemic, originally described as G litliicola Goldblatt & J.C.Manning, a homonym for an Ethiopian species.

Distribution and habitat: the species is so far known only from the lower slopes of the Gifberg Mountains in the northernmost part of the Western Cape ( Figure 2). Plants were found to be very local but fairly common among rocks and in cleared places on bush-covered, sandstone slopes.
Apparently first recorded in 1979 by I.S. Walters on the Farm Vleikraal (or Veekraal) on the slopes of the Gifberg east of Klawer, Sparaxis auriculata was initially associated with the superficially similar S. villosa (Burm.f.) Goldblatt. then classified in the genus Synnotia Sweet. A second gathering was made in 1997 on Gifberg Pass east of Vleikraal and we re-collected the species there on the middle slopes of the Pass on the Farm Koorlandshoek in early August. 1998. Examination of living plants, in particular the underground organs, con firmed our expectation that the two prior collections rep resented an undescribed species. S. auriculata can be dis tinguished mainly on vegetative features. The conns are narrowly conical and have tunics of medium-textured fibres, whereas the leaves have the sheaths liberally spot ted with dark purple, blades with acute apices and. espe cially unusual lor Sparaxis. a prominent submarginal vein that makes the margins appear thickened.
ifr/A m both northwestern Cape species (Lewis 1956;Goldblatt 1992). The underground part of the stem is enclosed by a thick neck of fine fibres and leaves with speckled sheaths, but both species differ in floral details from 5. auriculata. Sparaxis galeata has short-tubed, extremely sweetly fragrant flowers with the dorsal tepal reflexed, the leaves are prostrate, oblong and obtuse, and the species favours dry, stony, clay flats. The rare S. roxburghii has long-tubed flowers (the tube 20-25 mm long), otherwise resembling those of S. galeata, and is known from a single population in the mid Olifants River valley on stony slopes. Neither species has the submar ginal vein of S. auriculata.
The flowers of Sparaxis auriculata are much like those of the common southwestern Cape S. villosa in general appearance, but they are substantially larger in all respects. In S. auriculata the dorsal tepal is ± 25 x 18-20 mm and the anthers are 7-8 mm long, compared with 14-19 x 10-12 mm for the dorsal tepal and 3-4 mm for the anthers in S. vil losa. Furthermore, the style in the latter species divides near the anther bases, the style branches are ± 3 mm long, rarely reaching beyond the middle of the anthers, and the lower tepals are inclined slightly below the horizontal and without auriculate basal lobes. In S. auriculata the style branches are ± 4.5 mm long and exceed the anther apices, and the lower tepals are directed downwards and are prominently lobed near the base. The corms and leaves of the two species dif fer substantially and in S. villosa the corms are globose with coarse, netted fibres, the vertical elements of which are thickened and claw-like below, while the leaves lack the unusual spotting so marked in S. auriculata.  Goldblatt (1992) recognised two subspecies of Sparaxis variegata with subsp. metelerkampiae (L.Bolus) Goldblatt being distinguished from the typical subspecies by its small er. longer-tubed flowers which arc largely purple (including the anthers), a style which usually divides below the muddle of the anthers, and short style branches only 2-4 mm long. The two were maintained as subspecies because of the pres ence of some apparently intermediate populations.
Field observations show that populations of Sparaxis variegata and S. m etelerkam piae consistently differ and do not intergrade. This is particularly significant as their geographic ranges overlap. Apparent intermediates in herbaria may either be depauperate individuals or unusu al forms of either taxon. The morphological differences between the two reflect differences in their pollination biology. S. m etelerkam piae is pollinated by the long-proboscid Hies P rosoeca peringueyi and P sp. (I)iptera: Nemestrinidae) (Manning & Goldblatt 1996). whereas S. variegata is pollinated by the large anthophorine bee, Anthophora diversipes (Hymenoptera: Apidae) (unpub lished observations). These differences in morphology and reproductive biology suggest that it is more appro priate to recognise S. m etelerkam piae as a species.
The two species can be distinguished as follows. Sparaxis variegata has flowers with a perianth tube 37-44 mm long with the lower part 25-32 mm long and the dor sal tepal 25-30 mm long. The style divides at or near the anther apices, the style branches are 4-5 mm long and the cream-coloured anthers are 6-7 mm long. The perianth is usually predominantly yellow with the dorsal tepal and the tips of the others coloured violet whereas the lower part of the throat is streaked with dark violet lines. In S. m etel erkampiae the perianth tube is 45-50 mm long with the lower part 35-40 mm long and the dorsal tepal is 14-18 mm long. The style usually divides below the middle of the anthers, the style branches are 2-4 mm long and the purple anthers are 2^4 mm long. The perianth is uniform ly purple in colour except for the lower tepals which are marked with narrow white longitudinal streaks. Greenhouse-grown plants of S. metelerkampiae are self compatible and autogamous, whereas S. variegata is selfincompatible (Goldblatt 1992). 3. Ixia aurea J.C.Manning <£ Goldblatt. sp. nov.
Distribution and habitat: Ixia aurea is known only from the granitic slopes above the Darling-Yzerfontein road ( Figure 2) and is conspicuous when in flower, the brilliant yellow-orange spikes making a bright splash of colour among the late spring flora.
Despite occurring alongside the road in such a wellvisited region as Darling, the species has rarely been col lected. Plants from the type locality were originally mis takenly identified as a form of I. dubia Vent, of section Ixia by Goldblatt et al. (1998) because of the general resemblance to that species, despite the lack of a dark central mark to the flowers. More critical examination shows that I. aurea does not belong in section Ixia as cur rently circumscribed, for it has flowers with a narrowly funnel-shaped tube, tightly enclosing the style in the cylindrical lower half but flaring above and containing nectar in the upper third. In addition the filaments diverge from the base and do not occlude the mouth of the tube. In section Ixia, to which I. dubia belongs, the perianth tube is filiform-cylindrical almost to the apex, closely envelops the style and does not contain more than minute traces of nectar. The filaments are also coherent or connate from the base and thus close off the mouth of the tube. We suspect that records of this species were never made because even from a short distance the plants may be confused with two other common species from the Darling area, I. maculata L. and /. dubia. Apart from the floral differences, I. aurea may be distinguished from I. dubia by the leaves which lack thickened margins and midrib and are lightly coiled in the upper half. The blades of /. dubia and its allies have thickened margins and midribs and are plane. We suspect that the relationships of I. aurea lie with the two other yellow-flowered, nar row-tubed species, I. tenuifolia Vent. (= I. framesii L.Bolus) and I. odorata Ker Gawl. O f these, I. odorata may be its closest ally, based on general appearance, the broad, twisted leaf blades, and the many-flowered spikes. The flowers of I. odorata are in general smaller than those of I. aurea, with tepals up to 15 mm long, are white to lemon-yellow with a darker tube and never deep golden yellow to orange and are always sweetly scented of magnolia. The flowers are visited by a variety of bees for nectar and pollen.
The flowers of /. aurea are pollinated by a range of insects, including the hopliine beetles Heterochelus arthriticus, Lepithrix om atella and Pachycnema crassipes (Coleoptera: Scarabidae) and the horsefly Mesomyia edentula (Diptera: Tabanidae). The beetles use the flowers as sites for assembly and mating while the flies are attracted to the flowers as a source of nectar. Both the beetle species and the horsefly become covered in pollen after visiting several flowers. Ixia aurea thus joins the long list of species of this genus, mostly in section Ixia, which are pollinated in part at least by hopliine beetles (Goldblatt e ta l. 1998).

INFRAGENERIC CLASSIFICATION OF IXIA
The following currently accepted infrageneric classi fication of Ixia, was proposed by Lewis (1962) (subgenus Ixia) share the synapomorphy of a filiform perianth tube which tightly clasps the style for its entire length. The tube also contains little or no nectar and the filaments are contigu ous or connate from the base, effectively closing the mouth of the tube. We assume that this character arose once, for there seems no evidence to the contrary. Observations on the pollination biology of the genus are also consistent with this assumption and suggest that the species with filiform tubes have adopted derived pollina tion strategies which offer pollinator rewards other than nectar. The two taxa have flowers which are modified in different ways for different pollination strategies. In Dichone the flowers are uniformly pink and the anthers are short, somewhat inflated and dehisce tardily or only from the base by short slits. The flowers are buzz-polli nated by anthophorine bees (Apidae: Anthophorinae) which vibrate the flowers to extract the pollen (unpub lished observations). In contrast, flowers of section Ixia are variously coloured, often with dark centres, and the anthers are large and longitudinally dehiscent. These flow ers are adapted for pollination by hopliine beetles (Scarabidae) or by bees foraging for pollen. In the remain ing sections H yalis (Salisb. ex Baker) Diels and Morphixia (Ker Gawl.) Pax, the flowers have hollow peri anth tubes which normally contain significant quantities of nectar; they are largely pollinated by nectar-foraging insects, particularly bees and long-tongued flies. Because of the shared synapomorphy we believe that Dichone, w ith section rank, and section Ixia should together consti tute the subgenus Ixia. This decision is reinforced by the close resem blance between I. m icrandra (subgenus Dichone) and / flexuosa (section Ixia). Sections Hyalis and Morphixia are placed in a second subgenus for which we propose the name subgenus Morphixia.
We, therefore, put forward the following infrageneric classification for Ixia: Subgenus Ixia. Type: Ixia polystachya L.