Justicia L. (Acanthaceae: Ruellieae: Justiciinae) (Scotland & Vollesen 2000) is the largest genus of Acanthaceae, containing at a conservative estimate some 600 spp. (Graham 1988). As noted by Graham (1988) in her landmark review of the delimitation of Justicia, the circumscription of the genus has historically followed one of two divergent trends – either the adoption of a broad view of the genus, or towards the recognition of greater or lesser numbers of small segregate genera. The worldwide distribution of Justicia and the high degree of morphological variation amongst the species has hampered the general acceptance of one or the other of these two options, with workers generally focused on regional floristic accounts and thus missing the taxonomic wood for the trees. The result has been a confusing situation, whereby some regional floras recognise one or more of the segregate genera whilst others do not. The need for a stable, or at least comprehensive, classification of the genus stimulated Graham (1988) to undertake the first worldwide morphological survey of Justicia since Lindau's (1895) revision for Die natuürlichen Pflanzenfamilien. The result of her extensive study of vegetative and floral morphology and pollen ultrastruture was the conviction that the genus is most appropriately and satisfactorily circumscribed in the broad sense favoured by Lindau (1895), thus including such well-established segregates in an African context as Adhatoda Miller, Aulojusticia Lindau, Duvernoia E.Mey. ex Nees and Siphonoglossa Oersted.

The resulting infrageneric classification for Justicia proposed by Graham (1988) recognised sixteen sections, seven of them in southern Africa, and enumerated 295 species, representing an estimated half of the species in the genus. Amongst the species not listed were several from southern Africa. Although clearly not complete, Graham's classification forms the basis for future taxonomic studies in the genus and a framework in which to incorporate additional species.

Monechma Hochst. was retained by Graham (1988) as distinct from Justicia. The genus is traditionally diagnosed by its two-seeded capsules containing smooth, compressed seeds; whereas, Justicia is characterised by having mainly four-seeded (rarely two-seeded) capsules containing at most weakly compressed, mostly rugose (rarely smooth) seeds. The validity of this distinction in the light of the occurrence of intermediate forms was questioned by Hedrén (1990), who concluded that the two taxa could not be upheld at generic level. Accordingly, he reduced Monechma to a section within Justicia, as had been done earlier by Lindau (1895), thus increasing the number of sections in southern Africa to eight. In the same year, J. subsect. Ansellia (C.B.Clarke) V.A.W.Graham was raised by Ensermu (1990) back to sectional level as sect. Ansellia C.B.Clarke.

Graham's (1988) classification of Justicia was not implemented for the floristic treatment of the tribe for southern Africa (Baden et al. 1995) but has been adopted in later regional treatments for sub-Saharan Africa (Scotland & Vollesen 2000). As a result, the generic circumscriptions currently accepted for the southern African subregion are now out of line with those for the Flora of tropical East Africa (Darbyshire, Vollesen & Kelbessa 2010) and the Flora of Somalia (Hedrén & Thulin 2006) to name just two, both of which adopt the broad circumscription of Justicia accepted by Graham (1988) and Hedrén (1990), as will the planned volume for the Flora Zambesiaca region.

A start at integrating the treatment of the southern African flora with recent accounts for the rest of the continent was made by the formal transfer to Justicia of the remaining species of Aulojusticia (= Siphonolossa spp.) by Manning and Goldblatt (2012) and Duvernoia by Cubey (2006). What remains to be effected is the transfer to Justicia of the two southern African species of Adhatoda and all of the southern African species of Monechma, several of which require new combinations or replacement names, as well as the placement within the appropriate sections in Graham's (1988) classification of the southern African species that were left unplaced by her.

We do this here and provide a complete enumeration of the southern African species within the infrageneric classification developed by Graham (1988) and modified and expanded by Ensermu (1990) and Hedrén (1990). This will facilitate the uniform curation of the southern African material in a system compatible with that adopted elsewhere in sub-Saharan Africa.

The primary characters used by Graham (1988) to circumscribe her sections are the structure of the inflorescence and the pollen, notably the number of apertures and the condition of the margocolpi. Placing the currently unplaced species in an appropriate section in the system was mostly straightforward through direct association with allied species already in the system. Notable exceptions to this are the two southern African species of Adhatoda, A. andromeda (Lindau) C.B.Clarke and A. densiflora (Hochst.) J.C.Manning. Their association with the genus Adhatoda dates from the treatment of the family for Flora capensis by Clarke (1901), who included them in Adahatoda rather than in Justicia on account of their pollen having entire rather than areolate margocolpi, the latter then considered to be diagnostic for Justicia. Their placement in Adhatoda was later questioned by Manning and Getliffe-Norris (1995), who considered that they merited segregation from the tropical African and Indian members of the genus. These species, including the type of Adhatoda, were assigned by Graham (1988) to sect. Vasica, diagnosed by a shrubby habit and 2-aperturate pollen.

The South African Adhatoda species, both perennial herbs with 3-aperturate pollen, would certainly be misplaced in sect. Vasica and conform best to the small sect. Rhaphidospora (Nees) T.Anderson, based on the combination of their contracted, cymose infloresences and 3-aperturate pollen with entire margocolpi. This pollen type is otherwise found only in sect. Betonica and in the New World sect. Drejerella, both of which have strictly simple, spike-like inflorescences. The pollen type in Adhatoda andromeda and Adhatoda densiflora is evidently ancestral in Justicia (Graham 1988) and thus not phylogenetically informative, whilst the highly condensed inflorescences are difficult to interpret. It is possible that the two species are better segregated in a separate section but a trend towards a similar contraction of the inflorescence is evident in Justicia bolusii C.B.Clarke, which seems correctly placed in this section, and we include them here pending evidence to the contrary.

We place Justicia minima A.Meeuse, unusual amongst African species in its 4-fid calyx, in sect. Tyloglossa, largely on its similarity in inflorescence and pollen morphology to Justicia linearispica, placed here by Graham (1988).

Our placement in sect. Harniaria of the species previously treated as the genus Aulojustcia, as recommended by Hedrén (1990), is entirely consistent with the derived, sessile inflorescences and 2-aperturate pollen with areolate margocolpi that define this section. Similarly, placement of Justicia cuneata Vahl. in the same section accords with its sessile, 1-flowered axillary cymes and areolate margocolpi.

The southern African species of Monechma as treated by Immelman (1995) were placed en bloc in sect. Monechma, following Hedrén (1990). Inflorescence morphology in the section is very variable (Hedrén 1990), raising the possibility that the section is not monophyletic, although all species examined share derived 2-aperturate pollen with areolate margocolpi.

Justicia L., Species plantarum: 15 (1753).

Type: J. hyssopifolia L., lecto., designated by Hitchcock & Green (1929), fide Graham (1988).

We are most grateful to the referees for their valuable comments.

Babiana Ker Gawl. (93 spp.) (Goldblatt & Manning 2007, 2012) is relatively widespread in southern Africa and centred in the winter rainfall zone in the west of the subcontinent. One of the better known species in Western Cape Province, Babiana stricta (Aiton) Ker Gawl. (1803), is a combination based on Gladiolus strictus Aiton (1789). The location of a type, if it exists, is unknown and Goldblatt and Manning (2004) designated the illustration in Curtis's Botanical Magazine accompanying the combination as a neotype. Lewis (1959), in her monograph of Babiana, cited Gladiolus nervosus Lam. (1788) as a synonym of B. stricta and used the later epithet stricta for the species. Her apparently inexplicable action is perhaps due to the presence of the name Babiana nervosa Ker Gawl. in Index Kewensis, which we assume she would have seen. Checking the reference therein, we find neither G. nervosus nor B. nervosa mentioned. In our accounts of Babiana we overlooked Lewis's error and maintained B. stricta, although in somewhat revised circumscription (Goldblatt & Manning 2004, 2007). The type of G. nervosus, cultivated at the Jardin Botanique in Paris, was grown from material collected by Pierre Sonnerat at the Cape, which he visited briefly en route to and from Mauritius in 1773 and 1781 (Gunn & Codd 1981). The specimen is well preserved and is without doubt the species known as B. stricta, as evinced by the stiff, acute, upright leaves of that species, flowers with a narrow, straight perianth tube ± as long as the tepals, inner floral bracts divided to the base, a pubescent ovary and most significant, arrow-shaped anthers wider at the base.

We note that Lewis's circumscription of B. stricta included four varieties: of these var. erectifolia (G.J.Lewis) G.J.Lewis is included in B. nervosa, var. grandiflora G.J.Lewis is now Babiana tubaeformis Goldblatt & J.C.Manning, var. regia G.J.Lewis is Babiana regia (G.J.Lewis) Goldblatt & J.C.Manning and var. sulphurea (Jacq.) Baker has been excluded as the type cannot be associated with any known species (Goldblatt & Manning 2007; Goldblatt, Manning & Gereau 2008).

Lamarck's epithet nervosus is valid in Gladiolus and is available in Babiana [indexes of plant names show the name B. nervosa Ker Gawl. (1804) but it does not appear in the purported place of publication nor in Ker Gawler's earlier article dealing with B. stricta]. We provide the new combination here and reduce B. stricta to synonymy. For completeness we include the full synonymy of the species.

Babiana nervosa (Lam.) Goldblatt & J.C.Manning, comb. nov. Gladiolus nervosus Lam., Encyclopédie méthodique 2: 724 (1788). Type: South Africa, without precise locality, ex hort. Paris, Sonnerat s.n. (P: Herb. Lamarck, holo.—digital image!).

Gladiolus strictus [Sol. in] Aiton: 63 (1789) syn. nov. Babiana stricta (Aiton) Ker Gawl.: t. 621 (1803). Type: South Africa, without precise locality or collector, illustration in Ker Gawl.: t. 621 (1803) [neotype designated by Goldblatt & Manning: 94 (2004)].

Babiana erectifolia G.J.Lewis: 3 (1938). Babiana stricta var. erectifolia G.J.Lewis: 43 (1959). Type: South Africa, [Western Cape], Brand Vlei, near Worcester, Sept. 1932, G.J. Lewis s.n. as Nat. Bot. Gard 2686/32 (BOL, holo.!).

Gladiolus ringens Thunb.: 186 (1800), hom. illegit. non Andrews (1798). Type: South Africa, without precise locality, Thunberg s.n. (UPS: Herb. Thunberg, syn.).

?Babiana caesia Eckl.: 32 (1827). Type: South Africa, without precise locality, ?Ecklon s.n. (location unknown, not at S).

Babiana flavocaesia Eckl.: 32 (1827). Type: South Africa, vicinity of Stellenbosch, ?Ecklon s.n. (S, holo.!).

In Geissorhiza Ker Gawl. (100 spp.) (Goldblatt & Manning 2009, 2013), the nomenclature of G. ornithogaloides and its two subspecies need adjustment. The name Geissorhiza ornithogaloides is currently treated as a new species described by F.W. Klatt in 1866 (Goldblatt 1985). We note, however, that Klatt cited as a syntype, Lichtenstein s.n. from the ‘Koue Bokkeveld’. This specimen is the type of Ixia ornithogaloides Lichst. ex Roem. & Schult. (1817a). Although Klatt did not cite a basionym for his species, it should be treated as a combination according to the International Code of Botanical Nomenclature Art. 41.4 (especially examples 7, 8 and 10) as it is based in part on the same type (Greuter et al. 2000).

The identity and type locality of the Lichtenstein collection matches what is currently G. ornithogaloides subsp. marlothii, based on Geissorhiza marlothii R.C.Foster (1941). As a result, the name subsp. marlothii must be replaced by subsp. ornithogaloides and this taxon, in turn, requires a new name in place of its current circumscription as subsp. ornithogaloides. We accordingly replace it with the name subsp. flava based on Geissorhiza flava Klatt (1882). The choice of Ecklon & Zeyher Irid. 225 (51.8) from Caledon as lectotype of G. ornithogaloides Klatt by Goldblatt (1985) falls away. We also cite an epitype for G. ornithogaloides because the type specimens lack corms, important in distinguishing the two subspecies. The revised nomenclature is as follows.

Geissorhiza ornithogaloides (Licht. ex Roem. & Schult.) Klatt in Linnaea 34: 656 (1866). Ixia ornithogaloides Licht. ex Roem. & Schult.: 376 (1817b). Trichonema ornithogaloides (Licht. ex Roem. & Schult.) A.Dietr.: 583 (1833). Type: South Africa, [Western Cape], Koue Bokkeveld [Cold Bokkeveld], possibly Nov. 1803, Lichtenstein s.n. (B, holo.!).

subsp. ornithogaloides

Geissorhiza marlothii R.C.Foster: 66 (1941), syn. nov. G. ornithogaloides subsp. marlothii (R.C.Foster) Goldblatt: 325 (1985). Type: South Africa, [Western Cape], Cold Bokkeveld, Houdenbeck, 850 m, Marloth 10612 (B, holo.!; NBG!, PRE!, iso.); [Western Cape], Schoongesig, Ceres, Hanekom 1222 (NBG, epi., designated here).

subsp. flava (Klatt) Goldblatt & J.C.Manning, comb. & stat. nov. Geissorhiza flava Klatt: 392 (1882). G. ornithogaloides var. flava (Klatt) R.C.Foster: 68 (1941). Type: South Africa, [Western Cape], without precise locality, Breutel s.n. (B, holo.!).

[Waitzia flava Reichb. in Klatt: 392 (1882), cited in synonymy]

[Geissorhiza romuleoides Eckl: 27 (1827), nom. nud.]

P.B. Phillipson (Missouri Botanical Garden) kindly confirmed details of the type of Gladiolus nervosus in the Paris Herbarium [P] and R.E. Gereau (Missouri Botanical Garden) provided nomenclatural advice.

The genus Reyemia Hilliard (1992) (Scrophulariaceae: Limoselleae) was established for two species of annual herbs from the Hantam and Roggeveld Plateau in Northern Cape Province, South Africa, neither of them well known. Reyemia nemesioides (Diels) Hilliard (previously Zalusianskya nemesioides Diels) has been collected several times along the length of the Hantam-Roggeveld escarpment between Calvinia and Sutherland, but Reyemia chasmanthiflora Hilliard is still known only from the type locality near Williston. The relationships of Reyemia were presumed by Hilliard (1994) to lie with Zalusianskya F.W.Schmidt, with which it shares a derived, shortly toothed and plicate calyx and similar, centrally depressed, cushion-shaped seeds. It was distinguished from Zaluzianskya by what was described as a ‘loosely paniculate’ inflorescence, a resupinate corolla with a patch of clavate hairs inside the throat on the posticous side, and with two stamens and two staminodes. Examination of the specimens confirms that the inflorescence comprises essentially spicate units, sometimes with the lower flowers shortly pedicellate. Species of Zalusiaskya have simple or sparsely branched, spicate or racemose inflorescences, non-resupinate flowers, sometimes with a ring of clavate hairs in the mouth, and either four or two stamens, usually without staminodes (Hilliard 1992). The differences between the two genera are thus centred on the strongly branched inflorescence and more zygomorphic flowers in Reyemia.

An examination of phylogenetic relationships in Manuleae and Selagineae by Kornhall, Heidari and Bremer (2001), based on the plastid gene regions ndhF and trnL, confirmed the close relationship between Reyemia and Zaluzianskya. This analysis, although including just four species of Zaluzianskya plus R. chasmanthiflora, provided the first indication that the latter was, in fact, nested in Zaluzianskya. A subsequent, more intensively sampled study of relationships within Zaluzianskya by Archibald, Mort and Wolff (2005) used a combination of nuclear ITS and plastid rps16 and trnL-F gene regions from a sampling of 23 species of Zaluzianskya, plus the second species of Reyemia, R. nemesioides, to produce the first broad phylogenetic analysis of the genus. This study confirmed that Reyemia is deeply nested within Zaluzianskya, as sister to the clade comprising those members of section Holomeria (Benth.) Hilliard & Burtt with four fertile stamens. The placement is entirely consistent with the morphological similarities between the two genera noted by Hilliard (1994) and supplemented by additional states identified by Archibald et al. (2005), viz. filaments decurrent to the base of the corolla tube and a ligulate stigma.

These findings led Archibald et al. (2005) to conclude that the differences in floral morphology between the two genera are no greater than exists amongst other species of Zaluzianskya and that Reyemia should be merged with Zaluzianskya.

The entire corolla lobes in the former species of Reyemia and in members of Zaluzianskya sect. Holomeria supports the close relationship between the two groups suggested by the molecular analysis of Archibald et al. (2005). There are indications, however, that sect. Holomeria may not be monophyletic as currently defined because Zaluzianskya benthamiana was retrieved as a member of sect. Zaluzianskya subsect. Zaluzianskya. This species (plus Zaluzianskya diandra which was not included in the analysis) are unique in sect. Holomeria in having just two stamens, raising the possibility that the entire corolla lobes in Z. benthamiana and the other members of sect. Holomeria are convergent. This possibility, combined with the distinctive inflorescence and flowers of the former members of Reyemia, favours the retention of the two species as the separate section Reyemia within Zaluzianskya.

We implement this finding here in preparation for the forthcoming regional flora of the Karoo region. We also select an appropriate neotype from the same locality to replace the holotype of Zaluzianskya nemesioides in the Berlin Herbarium which has been lost.

We consulted the relevant literature and implemented the necessary nomenclatural conclusions. Type material is cited from the Berlin Herbarium (B) and Compton Herbarium, South African National Biodiversity Institute, Cape Town, South Africa (acronyms after Holmgren, Holmgren & Barnett 1990).

Zaluzianskya F.W.Schmidt, Neue und Seltene Pflanzen: 11 (1793), nom. cons., non Zaluzianskia Necker (= Marsilea L.). Type species: Zaluzianskya villosa F.W.Schmidt.

Sect. Reyemia (Hilliard) J.C.Manning & Goldblatt, syn. et stat. nov. Reyemia Hilliard in Edinburgh Journal of Botany 49: 297 (1992). Type species: Reyemia chasmanthiflora Hilliard = Zaluzianskya chasmanthiflora (Hilliard) J.C.Manning & Goldblatt.

Zaluzianskya chasmanthiflora (Hilliard) J.C.Manning & Goldblatt, comb. nov. Reyemia chasmanthiflora Hilliard in Edinburgh Journal of Botany 49: 297 (1992). Type: South Africa, [Northern Cape, Williston], Farm Annexe Kransfontein 721, 02 Sept. 1986, Cloete & Haselau 172 (NBG [as STE], holo.).

Zaluzianskya nemesioides Diels in Botanische Jahrbücher fur Systematik, Pflanzengeschichte und Pflanzengeographie 23: 482 (1896). Reyemia nemesioides (Diels) Hilliard: 297 (1992). Type: South Africa, [Northern Cape], Hantam Mountains, Meyer s.n. (B†, holo.). Neotype, selected here: South Africa, Northern Cape, Calvinia, gravel flats below Hantamsberg, 15 Sept. 1996, Goldblatt & Manning 10525 (NBG, neo.; MO, iso.).