Notes on the systematics and nomenclature of Tritonia ( Iridaceae : Crocoideae )

Study o f some early types o f species now known to belong to the genus Tritonia Ker Gaw l.. a member o f Iridaceae. sub­ family Crocoideae, comprising some 28 species o f southern and south tropical Africa, has shown the need for some nomenclatural adjustments. Ixia undulata Burm.f. (1768) is an earlier name for T. crispa (L.f.) Ker Gawl. based on Gladiolus crispus L.f. (1782) and the combination T. undulata (Burm.f.) Baker must be used for the species, which is native to the western half o f Western Cape, South Africa. The variety T. crispa var. parx iflora is also reduced to synonymy. The type specimen o f Ixia gladiolaris Lam. (1789), currently considered a synonym o f Tritonia securigera (Aiton) Ker Gawl.. has flowers that lack the characteristic tooth-like ridges on the lower tepals o f the latter, and corresponds closely to the eastern southern African T. lineata (Salisb.) Ker Gawl., based on Gladiolus lineatus Salisb. (1796). The new combination T. gladiolaris (Lam.) Goldblatt & J.C. Manning is made and T. lineata is reduced to synonymy. Montbretia lacerata and Tritonia lacerata. erroneously regarded as synonyms o f T. crispa, are combinations based on Gladiolus laceratus Burm.f., a species that remains unidenti­ fied because no type is known and the description is too vague to permit its identification. Lastly, field studies have shown that the cnsped-leaved T. watermeveri is connected by a series o f morphological intermediates to typical T. securigera. which has straight leaves and identical flowers. The new combination T. securigera subsp. watermeveri (L.Bolus) J.C.Manning & Goldblatt is proposed for this taxon.


INTRODUCTION
Our continuing studies o f the systematics and biology o f the African Iridaceae led us to examine type material o f several species thought to belong to the genus Tritonia Ker Gawl.but excluded by De Vos (1982, 1983) from her revision o f the genus as insufficiently known because o f the difficulty in relating type material to any known species.Among the names in question are Gladiolus fla vus [Sol.in] Aiton (1789), G. laceratus Burm.f.(1768), G. undulatus Burm.f.(1768), and Ixia gladiolaris Lam.(1789).The type o f Ixia gladiolaris is what is currently called Tritonia lineata (Salisb.)Ker Gawl., which thus becomes T. gladiolaris.Gladiolus flavus was based on a specimen o f T. flabellifolia (D.Delaroche) Ker Gawl., to which it is now assigned as a later synonym.The type o f G. undulatus matches T. crispa (L.f.) Ker Gawl.The latter, based on G. crispus L.f. (1782), closely resembles the shorter-tubed T. crispa var.parvijflora Baker, and the species must therefore be known by the earlier name T. undulata, a combination made by Baker in 1877.We have, however, failed to locate authentic material that can be associated with G. laceratus and this species must continue to be excluded from Tritonia.Taxonomic adjustments required as a result o f our investigation are made below.Lastly, new collections o f T. securigera (Aiton) Ker Gawl.have led us to re-examine the distinc tion between this species and the western Little Karoo plants referred to as T. watermeyeri.The latter species, defined by its crisped and undulate leaves, is now known to be connected by a series o f intermediates to the wide spread T. securigera, which has plane leaves but flowers identical to those o f T. watermeyeri.We conclude that it The type o f Gladiolus flavus is readily identified as a specimen in the British Museum (Natural History) col lected in early bloom (Figure 1).The long-tubed flowers are quite evident, and the characteristic dry.brown, longattenuate and acuminate bracts immediately identify the plant as the Western Cape species, Tritonia flabellifolia.
This plant has white to cream-coloured flowers and it is puzzling that Solander, who drew up the description, should have called it G. flavus (actually describing the flower as intensely yellow).Possibly the flower buds were sufficiently cream-coloured for him to have thought they deserved the epithet he chose.Despite the apparent inconsistency in flower colour we are confident that G.
flavus is a synonym o f T. flabellifolia.

2.
T rito n ia g la d io laris (Lam.)Goldblatt & J.C.Manning, comb. nov.The association by De Vos (1983, 1999)   The identity and generic disposition o f lxia undulata.described by N.L. Burman in 1768, has proved enduringly controversial.It was referred to Tritonia.as T. undulata by Baker (1877).and remained so treated in Flora capen sis (Baker 1896).Klatt (1882), however, referred I. undu lata to his new genus Tritonixia, a segregate o f Tritonia erected for the shorter-tubed members o f the genus (but including the type o f Tritonia, T. crocata).Brown (1929) actually made a new combination, T. undulata, because although Baker had already made the same combination, he had included plants o f lxia crispa L.f. (now I. erubescens Goldblatt) under the same.

Ixia gladiolaris
Later, Lewis (1962) in her monograph o f lxia excluded lxia undulata from that genus but did not explicitly say to what genus she thought the plant belonged.Then, De Vos (1983Vos ( , 1999) ) excluded the species from Tritonia in her accounts o f the genus, commenting that although it superfically resembled a Tritonia, the short, stout anthers recalled lxia subgenus Dichone, although the funnelshaped tube did not.She therefore suggested that the type might represent a hybrid between I. crispa and I. vanzijliae M.P.de Vos.
We have examined the type specimen o f lxia undulata.
a single plant without a conn, in the Burman Herbarium at Geneva (Figure 2), and are amazed that its identity could have caused such confusion.It is an exact match for plants from the Tulbagh District o f Western Cape, referred by both Baker (1892, 1896) and De Vos (1983, 1999) to Tritonia crispa var.parviflora.The leaf margins are tightly crisped and the entire blade is loosely undulate and twisted, while the salver-shaped flower has a narrow, almost cylindric (not funnel-shaped as De Vos stated) Perianth tube.± 18 mm long with [apparently] spreading tepals, stamens with filaments exserted ± 5 mm from the tube, and anthers ± 3 mm long.
We have no hesitation in regarding lxia undulata as nonspecific with Tritonia crispa (basionym Gladiolus crispus L .f ( 1782))./. undulata is the earlier name, and T. crispa now falls into the synonymy o f T. undulata.
We also question the merit o f recognizing infraspe cific taxa in this species.Perianth tube length is variable, ranging from 18-85 mm across its range.1882/36, NBG).The pattern is not entirely clear but there is a trend for shorter tubes in populations in the south and interior o f its range and any division into subspecific taxa based on tube length would be arbitrary.Tube length is closely associated with the local pollinators, longproboscid flies (Manning & Goldblatt 1997).Where the long-proboscid species Moegistorhynchus longirostris occurs within the range o f T. undulata, the perianth tube is longest but inland and on the Cape Peninsula where M. longirostris does not occur, shorter-proboscid flies, presumably either Philoliche gulosa or P. rostrata, are the inferred pollinators.Predictably, tube length tracks the proboscis length o f the pollinator, a situation that has been documented in Disa draconis and Geissorhiza confusa (Johnson & Steiner 1997).We therefore include var.parviflora in T. undulata.
No type for this species, described by Burman fil. in 1768, has ever been identified, and the description is too brief to permit its association with any known species.Baker (1877), who referred the species to Montbretia (now included in Tritonia), believed it to be an earlier name for T. crispa, now T. undulata.Both Baker and Klatt, who made the combination in Tritonia, are silent on their reasons for identifying Burman's species as a Tritonia.In her most recent account o f Tritonia, De Vos (1999) placed Montbretia lacerata Baker (sic) and T. lacerata (Baker) Klatt (sic) as synonyms o f T. crispa, and designated the type o f T. crispa as the lectotype o f M. lacerata.She thus ignored the fact that Baker's name is a combination based on Burm an's Gladiolus laceratus (and thus has no type o f its own but is nomenclaturally identical to G. laceratus).De Vos did not specifically deal with the identity o f the latter.In our search for type material we have confirmed that no type is located at G (Herb.Burman) or L (the Leiden Herbarium), the only places where it is likely to be preserved.

microfiche!).
There is no question about the identity o f the type of this predominantly Little Karoo species, to our knowl edge first collected by Masson and Thunberg in the early 1770s.M asson's collection was later grown at the Royal Botanic Gardens, Kew, where plants in flower were described as Gladiolus securiger a decade later by Solander for A iton's Hortus kewensis (Aiton 1789).Plants o f the same collection were illustrated in Curtis's Botanical Magazine in 1797 under that name (Ker Gawler 1797) but the species was later transferred to Tritonia (Ker Gawler 1804).
Distinctive features o f the species, in the narrow sense, are a basal fan o f firm, plane leaves (4-)6-12 mm wide with acute tips, a typically unbranched stem (robust plants may have one or more lateral branches), and an erect spike o f (3)4-8 orange, rarely yellow, and evidently unscented flowers.The dorsal tepal is largest, 16-18 x 13-15 mm, and the lower tepals each bear a pronounced bright yellow median tooth, ± 4 x 2.5 mm.shaped like an axehead (securiger is Latin for axe-bearer).These teeth, called calli in the early literature and calluses by De Vos (1982, 1983, 1999) in her accounts o f Tritonia.are not unique to the species but are frequent in Tritonia.
They are particularly well developed in most species of section Montbretia, but in the remaining sections usually form low, thickened ridges, and are not developed at all in some species, such as T. tugw'elliae L.Bolus.The teeth are especially well developed in T. securigera and its western Little Karoo relative, T. watermeyeri L.Bolus.In fact there is little difference between the flowers o f the two taxa, apart from the slightly smaller tepal teeth (± 3.0 x 1.3 mm) and faint woody scent in T. watermeyeri.
The major difference between the two are in their leaves, and the crisped and undulate leaf blades o f T. watermey eri contrast strikingly with the plane leaves of typical T. securigera.Scent in flowers o f Iridaceae is notoriously variable, even within a species (Goldblatt & Manning 1998) and little reliance can be placed on it alone as a specific character.helped us identify Ixia gladiolaris in the Herbier Lamarck at P; and J. Veldkamp, Leiden Herbarium, for searching for a possible type o f Gladiolus laceratus.We are also grateful to Mary Stiffler who cheerfully helped solve bibliographic questions and provided copies o f numerous articles not readily available to us.
of Ixia gladio laris with T. securigera is refuted by two features: the presence of prominent submarginal veins on the leaves, and the low ridges on the lower tepals.The leaves o f T. securigera have no obvious submarginal veins and the lower tepals o f the flower bear large, tooth-like ridges, mm high, that resemble elaborate axeheads.We regard the type and associated description o f Ixia gladiolaris as representing one o f two largely eastern southern African species, either Tritonia disticha (Klatt) Baker or T. lineata.Both these species have leaves with prominent submarginal veins, unusual in Tritonia, and flowers with low median ridges on the lower tepals.The two species are distinguished largely by flowering time, flower size and perianth colour (De Vos 1999): T. disticha blooms in summer, mainly January to March, and has red to pink (or rarely yellow) flowers 20-30 mm long, whereas T lineata flowers earlier, mainly September to November, and has yellow to apricot (pale orange) flow ers (25-)30-40 mm long, lxia gladiolaris most closely matches the latter morphologically and is evidently also early flowering.Plants grown in Paris bloomed in spring, March and April in the northern hemisphere, whereas we would expect T. disticha to flower in June or July, even under glass.I. gladiolaris is an earlier name for T. lineata, which thus falls into synonymy under the new combina tion T. gladiolaris.
Flowers with the longest tubes are recorded from the Olifants River Mountains (85 mm in Goldblatt & Manning 10345, NBG), south to Joostenberg (60 mm in Lewis 5901.NBG).Plants with shorter-tubed flowers.25-40 mm long, occur inland and in the south, as at Sir Lowry's Pass (35-40 mm in Barker 3378, NBG), and Somerset West (25-30 mm in Lewis 5675.NBG).Some plants from the Tulbagh Valley fall at the lower extreme o f the range with a tube 22-25 mm (Hansford 4 from Welgelegen House.NBG).In con trast, plants from the slopes o f Roodezandberg.not far from Tulbagh, have a tube ± 50 mm long (Compton et al. FIGURE 2.-Holotype o f lxia undulata Burm.f.(G-BU).

ACKNOWLEDGEMENTS
The distinction between Tritonia securigera and T. watermeyeri becomes even less clear when plants with flowers matching the two species from sandy habitats in the Anysberg and at the foot o f the Little Swartberg in the western Little Karoo are taken into considera tion.These plants are often dwarfed, up to 0.6 m high, and have either fairly soft-textured, undulate leaves(Goldblatt & Porter 12060, NBG, from  the foothills o f the Little Swartberg), or narrow, twisted leaves with straight to slightly crisped margins (Martin 60, NBG, from Anysberg Nature Reserve) while the spikes may have 1-3 flowers in the most dwarfed individuals, or up to 4 flowers in taller plants.Both T. securigera and typical T. watermeyeri are usually taller plants, at least 150 mm high, and grow on shale slopes.The plants with undulate leaves seem perfectly intermediate between typical examples o f T. securigera and T. watermeyeri and maintenance o f the latter as a separate species seems arbitrary.Existence o f plants with leaves o f intermediate form, either undulate or loosely twisted, make it more appropriate to treat T. watermeyeri as a subspecies o f T. securigera.Typical T. securigera ranges widely through the southern Cape, from the Touwsberg in the western Little Karoo to Bedford in the Eastern Cape, whereas subsp.watermeyeri is restricted to the extreme west, between Montagu and Anysberg (Figure 3).subsp.w aterm ey eri (L.Bolus) J.C.Manning &  Goldblatt, comb, et stat.nov.Tritonia watermeyeri L.Bolus in Annals o f the Bolus Herbarium 4Support for this study from the National Geographic Society (grants 6704-00, 7103-01, 7316-02, and 7799-05) is gratefully acknowledged.We thank the Curators o f the following herbaria for loan o f type specimens: BM, G, and L; Pete Phillipson, Missouri Botanical Garden, who