Three cryptic new species of Aristea ( Iridaceae ) from southern Africa

Field work in southern Africa over the past several years has resulted in the discovery of three new species of the subSaharan African and Madagascan genus Aristea Aiton, which now comprises some 53 species. Aristea has a pronounced centre in southern Africa and a centre of diversity in the winter rainfall zone o f the subcontinent, where all three new species occur, one extending eastward into the adjacent southern edge of the summer rainfall zone. All three novelties have been collected in the past but were confused with related species. A elliptica (subgenus Eucapsulares), confused in the past with A. pusilla (Thunb.) Ker Gawl., has a more robust habit, usually with 4 or 5 flower clusters per flowering stem, pale blue flowers, smooth ellipsoid seeds with flattened surface cells, and pollen shed as monads, whereas A. pusilla usually has 1-3 flower clusters per flowering stem, dark blue flowers, pollen shed as tetrads, and globose seeds with faint foveate sculptur­ ing and colliculate surface cells. A. nana (also subgenus Eucapsulares), known from few collections, and also confused with A. pusilla or A. anceps Eckl. ex Klatt. has the unbranched and leafless flowering stem of the latter but has large green floral spathes, flowers borne on long pedicels, and lacks a leaf subtending the single terminal flower cluster in contrast to nearly sessile flowers in A. pusilla and A. anceps, and in the latter, dry, rusty spathes. A. cistiflora (subgenus Pseudaristea) closely resembles A. teretifoha Goldblatt & J.C.Manning but has linear to narrowly sword-shaped leaves and ± secund flow­ ers with the outer tepals only slightly smaller than the inner and with small, dark brown markings at the bases of all the tepals. In contrast, A. teretifolia has narrower, sometimes terete leaves and flowers held upright with the outer tepals notice­ ably smaller than the inner and bearing dark markings covering the lower half, whereas the inner tepals are unmarked.


INTRODUCTION
In the course of field work in southern Africa, three undescribed species of the sub-Saharan African and Mada gascan genus Aristea Aiton have come to light.All are spring-flowering species native to the eastern half of the southern African winter rainfall zone and adjacent south ern edge of the summer rainfall zone.All have been col lected before but have been confused with known species similar in vegetative or floral morphology.Comparison of the taxonomically critical features of the genus, in cluding seeds and pollen grains (Goldblatt & Le Thomas 1997;Goldblatt et al. 2004), has substantially aided in distinguishing two of them. A. elliptica and A. nana, both members of subgenus Eucapsulares: section Eucapsu lares (taxonomy following Goldblatt & Le Thomas 1997).In contrast, details of the flower have shown that A. cistiflora, of subgenus Pseudaristea.differs from the closely related A. teretifolia Goldblatt & J.C.Manning, although its other vegetative and fruiting features accord closely with the remaining members of the subgenus (Goldblatt & Manning 1997a).All species have been examined live in the field as well as in the herbarium.With the addition of these three novelties, Aristea com prises an estimated 53 species.Seven species occur in Madagascar (Goldblatt 1991(Goldblatt , 1995a) ) and about 18 in tropical and eastern southern Africa (Weimarck 1940;Vincent 1985), one shared with Madagascar.There are 33 species in the southern African winter rainfall zone (Goldblatt & Le Thomas 1997;Goldblatt & Manning 1997a. b), four of which are shared with eastern southern Africa.
In the descriptions that follow, we ignore the homolo gies of the subtending foliar bracts of the inflorescence, a binate rhipidium.and call the outer two bract members spathes and those enclosed within them bracts, the latter always smaller than the spathes.The individual inflores cence units, which vary in number and arrangement on the flowering stem, are simply termed flower clusters.
Distribution and ecology: occurring in Western and Eastern Cape, from Robinson's Pass to the Baviaanskloof, on sandy and rocky sandstone slopes, in arid, mar ginal fynbos.
A dwarf species, mostly reaching less than 120 mm in stature, Aristea nana has until recently rarely been col lected and then has usually been confused with similarly low-growing A. pusilla (Thunb.)Ker Gawl.When we came across the species in the wild in 2003, growing together with A. pusilla near Joubertina, in Eastern Cape, on a slope burned the previous summer, direct compari son of the two in full bloom made it clear that these were different species, despite their similar dwarf growth form and virtually identical flowers.They differ slightly in their leaves, those of A. nana having a glaucous bloom and wider translucent margins, whereas the pale green leaves of A. pusilla are softer textured.The flowering stem of A. nana is almost always unbranched and bears leaves only near the base.The terminal intemode is sev eral times longer than the rest of the stem and terminates in a single inflorescence of two or rarely three flowers.Particularly striking are the we 11-developed pedicels of the flowers, 10-12 mm long, and ovoid ovary, 4-6 mm long, quite different to the short pedicels, ± 2 mm long, of A. pusilla and triangular-columnar ovary, 12 mm long.
Subsequent examination of the pollen of the two species from the Joubertina site showed that Aristea pusilla has pollen shed in tetrads, the monads with operculate apertures, as described by Goldblatt & Le Thomas (1997).Pollen of A. nana, however, is shed singly, and the grains are dizonasulculate, having two smooth, welldefined apertures at opposite ends of the grain.Once we had determined that A. nana was a distinct species, we made a concerted effort in 2004 to establish its range.We found it to be common in the Long Kloof and valleys to the north from Avontuur to Joubertina.We were also able to identify additional collections of A. nana in herbaria, where the earliest collections that we have found are those made by R.D. Bayliss in 1974 and mixed with A. pusilla.Later collections in herbaria have consistently been misidentified as A. pusilla.
Despite their superficial similarity, Aristea nana and A. pusilla are probably not closely related.Instead, we believe that A. nana is most closely allied to the eastern southern African A. abyssinica Pax (currently including A. cognata N.E.Br.(Goldblatt 1995b) and A. anceps Eckl.ex Klatt.These two species also have unbranched, flattened and broadly winged flowering stems with an extended upper intemode, thus bearing leaves only near the base (Weimarck 1940).In addition, A. nana, A. abys sinica, and A. anceps sometimes produce a short stem held close to the base of the plant as well as normal extended flowering stems, a feature not before reported.Some examples are collections of A. abyssinica from KwaZulu-Natal (Goldblatt & Manning 9720, MO) and Limpopo in South Africa (Goldblatt & Porter 11954B, MO) and Zimbabwe (Chase 3650, MO) and A. anceps (Barker 6991, NBG; Compton 20288, NBG).
Unlike Aristea nana, however, both A. abyssinica and A. anceps have a small subterminal leaf subtending the terminal flower cluster or terminal pair of flower clus ters, and rarely a second subterminal leaf 10-20 mm below the flower clusters.A. anceps also differs in having the inflorescence spathes and bracts ± dry at flowering time, whereas those of A. nana are green.Spathes and bracts of A. abyssinica are green with broad dry margins at flowering time, later becoming entirely membranous.They also have subsessile flowers (pedicels are 2-3 mm long in A. anceps), whereas those of A. nana have pedicels up to 20 mm long at flowering time, extending to 15 mm in fruit.Moreover, both A. abyssinica and A. anceps have pollen grains with the apertures obscured by masses of exine (Goldblatt & Le Thomas 1997).The grains were thus termed sulculate; subsequent examina tion using transmission electron microscopy showed that the apertures are either zonasulculate or disulculate, in either case with a thick intine indicating the location of the aperture (Goldblatt et al. 2004).Pollen grains of A. nana differ significantly from those of its apparent rela tives in having smooth apertures, unique for a species of section Eucapsulares.Whereas nearly mature capsules of A. nana are known, mature seeds are not, making impossible comparison of the latter character, important in determining relationships in Aristea (Goldblatt et al. 2004).Aristea elliptica Goldblatt & A.PD old, sp.nov.
Distribution and ecology: occurring in Eastern Cape, from the Zuurberg Mountains as far east as the Fish River Mouth, mainly on sandstone slopes and often on rock outcrops, in fynbos or grassy fynbos.Aristea elliptica most closely resembles the diminu-very different and, as Weimarck noted, it is difficult to tive southern Cape species, A. pusilla, and was included accept these as belonging to the same species, in what Weimarck (1940) called A. pusilla subsp.robus-Weimarck's decision to treat the two as subspecies was tior Weim.That taxon is based on a painting of a dark largely based on the presence of apparent intermediates, blue-flowered plant called A. pusilla in C urtis' s which he did not enumerate, in the Uitenhage and Port Botanical Magazine (Ker Gawler 1809).In the absence Elizabeth areas.Vincent (1985) who also recognized A. of an associated specimen, however, it is impossible to pusilla subsp.robustior does list intermediates but some establish its identity with confidence.We believe it rep-of these specimens are subsp.pusilla and others are A. resents a well-grown specimen of A. pusilla but is not spiralis (L.f.) Ker Gawl.(Marsh 660, PRE, from distinct from that species, which has the dark blue flow-Franschhoek Pass) or A. afT.pauciflora Wolley Dod ers clearly evident in the painting.However, some col-(Oliver 5472, PRE, NBG, from Bailey's Peak, Klein lections from Eastern Cape referred to subsp.robustior Swartberg Mtns), both of which localities are outside the by Weimarck (1940) and later by Vincent (1985), appear range of both A. elliptica and A. pusilla.Measurements given by Vincent for subsp.robustior mostly do not apply to A. elliptica.
While most of these so-called intermediates are mere ly more robust Aristea pusilla, the taller Eastern Cape plants referred to A. pusilla subsp.robustior represent a second species, recognized at first by its larger size and associated broader stem and leaves, the latter firmer and more leathery than those of A. pusilla.More careful examination of these plants shows that they have ellip soid seeds, unique in Aristea, with a smooth surface and surface cell outlines plane or weakly colliculate (Figure 2A, B).The seeds lie in a single row in each locule, as they do in the closely related A. ecklonii Baker, A. ensi folia Muir and A. pusilla, but unlike the horizontally packed seeds in these three species, those of A. elliptica are loosely arranged and often oriented obliquely to the long axis of the capsule in a zig-zag arrangement with only their tips touching (Figure IB).In this arrangement every alternate seed is more or less parallel.These plants also have pollen grains shed singly (Table 1) and they are bisulculate, with two large apertures covered with exine.
Both the seeds and pollen contrast starkly with those of Aristea pusilla and its ally A. ensifolia.Seeds of these species are depressed-globose to shortly columnar (Figure 2C) (flattened dorsally and ventrally due to pres sure from adjacent seeds) with shallow foveate sculptur ing and surface cell outlines colliculate to tuberculate (Goldblatt et al. 2004).The seeds are vertically stacked, in capsules similar to those of A. elliptica but (14-) 20-25(-30) mm long.Pollen of A. pusilla is shed in tetrads (Goldblatt & Le Thomas 1997) and the monads are operculate.We have now sampled six populations of A. pusilla from across its entire range and confirm tetrads in all of them (Table 1), whereas four populations of A. elliptica examined for the feature have disulculate pollen grains shed singly.The latter pollen type conforms to the pattern in most members of section Eucapsulares (Goldblatt & Le Thomas 1997;Goldblatt et al. 2004).
Typical Aristea pusilla is a dwarf plant, usually 80-150 mm high, with fairly soft-textured, pale green leaves mostly 2-4 mm wide, and flowers slightly small er than those of A. elliptica, with outer tepals 1 2 -1 4 x 7 mm, and inner 9-11 x ± 9 mm versus outer tepals 16-18 x ± 7 mm, and inner ± 16 x ± 9 mm in A. elliptica.While A. pusilla occurs on both loamy clay and sandy slopes and extends from Swellendam in the west along the coast and in the Long Kloof to Grahamstown in the east, A. elliptica favours rocky sandstone slopes, often occurring in quartzite outcrops and is known from the Zuurberg at Grahamstown and eastward to the Fish River Mouth.The ranges of the two species overlap in the Grahams town area.Any confusion between the two is most like ly the result of depauperate specimens of A. elliptica being mistaken for robust plants of A. pusilla.Apart from the pollen and seed differences, A. elliptica can be sepa rated by the leaves, mostly 4-5 mm wide, leathery and somewhat glaucous, the flowering stem usually bearing at least four flower clusters, the lowermost sometimes short-stalked, and the inflorescence spathes (25-)30-38 mm long.Capsules of A. elliptica are elongate, (20-) 24-30 mm long, and contain ellipsoid seeds in a single vertical row, with the long axis of the smooth seeds oblique to the long axis of the locules as explained above.
The spathes of A. pusilla are 16-20 mm long and the stem 1.5-2.0mm wide.
Distribution and ecology: Western Cape, on the lower southern slopes of the Langeberg Mtns, in peaty sandstone soil, flowering only after fire or clearing of the veld.Evidently first collected by T.M. Wurts in 1952, and only a few times since then, we found Aristea cistiflora in 2003, in the spring after a wildfire on the Langeberg near Swellendam, when it became clear that it was an undescribed species.Although Wurts 326 is stunted and only about 100 mm tall, the large, well-pressed flowers are identical to those of the type collection.Interestingly, in 1979 the South African botanist, Dr E.G.H. Oliver noted on the Wurts specimen, 'probably a new species, not described as material inadequate'.His prediction has proved correct.
Aristea cistiflora is apparently most closely related to A. teretifolia Goldblatt & J.C. Manning (Goldblatt & Manning 1997a), which also has unequal tepal whorls, with oblique outer tepals smaller than the inner and bear ing dark basal marks.Aristea teretifolia is distinguished by its linear to terete leaves, up to 2 mm wide, flowering stems bearing at most two lateral inflorescences, and by the slightly smaller flowers with the the outer tepals ± 20 mm long and the inner 24-28 mm long, thus substantially longer than the outer, which are abruptly constricted at the base.In A. cistiflora the leaves are narrowly sword shaped, mostly 2.5-3.0 mm wide, the flowering stems bear up to six lateral flower clusters and the inner tepals are only 1-2 mm longer than the outer, which taper gradually toward the base.Aristea cistiflora also has larger capsules than A. teretifolia, 3 5 ^0 mm long versus 20-30 mm long.In general appearance A. cistiflora is most likely to be confused with A. cantharophila Goldblatt & J.C.Manning but this species has subequal, symmetrical tepals with dark markings at the base of both whorls, dark filaments and exceedingly long cap sules, 60-85 mm long.
As in other members of subgenus Pseudaristea (Goldblatt & Le Thomas 1997), the pollen grains of Aristea cistiflora are dizonasulculate and have reticulate exine and smooth apertures.Recognition of A. cistiflora brings the total number of species in subgenus Pseu daristea to eight.