Six new species and one new subspecies of Erica ( Ericaceae ) from Western Cape , South Africa

Two new species and one new subspecies belonging in §Ceramia of the genus Erica L. are described from the western half of the province—E. cavartica known from a single collection growing in a cave in the Cederberg. E. amalophylla con­ fined to a shaded rocky overhang and known only from a single collection in the Porterville Mountains and E. cxmosa subsp. grandiflora occurring in the mountains between Ceres and W'orcester. Four new species are described from moun­ tains in the Little Karoo area—E. schelpeorum occurring widespread mostly in dry watercourses in arid vegetation at low altitudes, E. lithophila tound only in dry rock crevices at high altitude in the eastern Swartberg and Kammanassie Mountains, E. umbratica confined to shaded rocky habitats and known only from a few. very localized small populations at low altitude in Meiringspoort, and E. annalis restricted to the arid habitats on lower northern slopes of Kammanassie Mountains. Each is provided with a detailed drawing and a distribution map.

In §Ceramia of the genus Erica L. there are many species which are associated with damp, shaded or wet habitats.Most are soft, low shrublets, either erect and compact or diffuse and sprawling, with long delicate main branches and often with open-backed leaves.The section contains a group of species (the E. planifolia group) which are closely related due to the position of the bract being on the main stem and leaf-like and not recaulescent on the pedicel (bract = leaf, in our terminol ogy), and to a thickened midrib which is distinctly enlarged towards the distal end of the lamina and/or beyond the lamina apex.Within this group there were col lections of herbarium material made by Thomas Stokoe and Elsie Esterhuysen that had not been placed with any certainty within a species and remained unnamed.
The three new taxa have a very similar facies with similar stems, leaves and bract but with differences noted in the corolla size, the presence of anther appendages, the ovary indumentum and the filament shape.They all occur in shaded, damp habitats.
The species is closest to the similar looking E. cymosa E.Mey.ex Benth.For differences between this and the other species in this group see Table 1.
Erica cavartica is known only from the type collec tion made by Thomas Stokoe in the Cederberg (Figure 2).He gave the habitat details to Neville Pillans in the Bolus Herbarium as 4in rock-crevices and on floor of a cave near Crystal Poof.The name is derived from this habitat preference-cavarticus = bom or living in caves (Latin).Pillans noted at the time that the main branches were up to 2xh ft [0.75 m] long and procumbent on large plants but on small plants were erect.
The plants within this alliance of three species are all rather soft and delicate with the leaves thin, flat and open-backed.The name of the species is derived from this feature-amalophylla from amalos = soft, tender, weak; phyllon = leaf (Greek).For species alliances and differences see Table 1.
Erica amalophylla is known only from the type col lection made by Elsie Esterhuysen in the mountains southeast of Porterville which drain into the Twenty Four Rivers system (Figure 2).She noted that the plants were 'growing on shaded sandy floor of overhanging rock shel ter'-a similar habitat description to that of E. cavartica.
This species forms a close alliance with the two new species described above.It differs in having (he bracte oles placed further away from the bract ('/; way up pedi cel to approximate to the calyx and not basal) and the sta mens with appendiculate anthers and filaments not spoon-like at the base (Table 1).
Variation in characters between the populations of E. cymosa are present.The most noticeable is in the size and shape of the corolla.There are however, two clear size groupings which occur in two disjunct distribution areas (Figure 2).These are also coupled w ith a difference in indumentum of the ovary.We have therefore decided to recognize these two regional variants at subspecific level.
Bentham (1839) described the species based on a Drêge collection w hich had been provisionally named E. cymosa by Meyer who worked on and distributed many Drêge collections.Bentham saw many collections from Berlin and usually cited these as 'v.s. sp. in herb.reg.Berol.' Since he did not cite this reference for E. cymosa he did not see his material there and must have relied solely on the rather scant material at Kew w hich he anno tated.He noted in the protologue that he saw no cymes in the material, and unfortunately he proceeded to use Meyer's manuscript name 'cymosa' tor the species.
The typical subspecies occurs in the mountains of the Du Toitskloof/Wellington/Wemmershoek area where it grows in moist places on cliffs (Figure 2).Where we have seen it.the plants grow on a ledge under an over hang in deep shade in a river gorge.Flowering time: October to March, one record in June.Diagnostic features: corolla larger.3.5-5.5 x 3-4 mm; ovary hairy (Figure 4; Table 1).This subspecies occurs on the mountains in the Hex River Range just southwest of Ceres to near Worcester and on the Keeromsberg across the Hex River Valley (Figure 2).These populations are cut off from those of the typical subspecies by the much drier, lowland Worcester-Tulbagh Valley.
It is most similar to Erica hirta, E. perlata and E. tradouwensis in having the long pedicels and flowers with slightly exserted anthers.E. hirta differs in the short spike-like inflorescence (not umbel-like), the hairy corolla and 3-nate leaves, E. perlata in the hairy corolla, the subopen-backed, spreading leaves and the spurred anthers, and E. tradouwensis in the 3-nate leaves which are slightly open-backed, the hairy corolla and the eglandular, long hairs on the stems and leaves.
Erica schelpeorum was stated above as being similar to three other species in other sections.With E. armata it shares the bract on the main axis (non recaulescent), the 4-nate leaves, similar glands and umbel-like inflores cence but that species has larger, hairy flowers, pedicel shorter than the corolla and longer anther appendages.E. mitchellensis, which is similar to E. armata, has a re caulescent bract and inflorescence of only a single whorl.E. strigosa has the non recaulescent bract and glabrous corolla, but differs in having a spike-like inflorescence and short pedicel.This species was first collected in 1954 by Mrs Sybella Schelpe  while on a field trip with her husband.Prof. Ted Schelpe , who was the lecturer in taxonomy for both of us during our student years at the University of Cape Town.We thus take this opportunity of honouring them both in naming this species which was discussed with Sybella shortly before her death.
Erica schelpeorum is widespread on mountains and hills in and bordering the Little Karoo (Figure 6).In all cases where we have seen the species, it occurs in dry, arid areas usually associated with a dry watercourse.Despite the high altitude of the type collection, the plants there were growing in arid renosterveld vegetation on the southern slopes of the northern foothills to the main range of the Great Swartberg.In the Kykoe locality which was shown to us by Jan Vlok and which is the same population as Viviers & Vlok 184 and Schumann 684, the plants were growing in renosterveld vegetation on shale/clay with a thin overlay of quartzite pebbles washed down in a small dry watercourse.
All the species mentioned as being similar in some respects, do not occur sympatrically with E. schelpeo rum, they all occur in the region to the southwest from the Hex River Mountains to the Cape Peninsula east wards as far as the Langeberg, just east of Swellendam.Erica hirta and E. tradouwensis are both low altitude species with the former growing on clay/gravel slopes sometimes on the border between dry fynbos and renos terveld vegetation in the Malmesbury, Stellenbosch and Genadendal areas.
Diagnostic features: shrublets very small with flowers relatively large compared to the plant and leaves; leaves grey; pedicel with long, plumose hairs in upper part; anther appendages narrow; ovules 20-30 per locule, erect; seeds with delicate testa having thin anticlinal walls.
This new species is most closely related to E. kirstenii E.G.H.Oliv.(Oliver & Oliver 2000) which is distin guished by the much larger stouter shrubs, leaves ± twice as long, pedicel long and curved with short plumose hairs all over, ovules ± 60 and not erect and testa cells with thick anticlinal walls.
Erica lithophila occurs on the eastern parts of the Great Swartberg Range in the Meiringspoort area and on the Kammanassie Mountains just to the south where it is confined to high altitudes (Figure 8).The species is easi ly overlooked due to the small size of the plants with very old gnarled plants often only 30 x 30 mm and growing.The habitat is rock crevices in large rocks usually with no soil and north-facing.The largest plant we have seen was growing in some humus in a vertical crack facing east and was open and ± 120 mm in diameter.In these dry, rocky habitats, the shrublets are often cryptic, melting in with the lichen-covered rock.In younger plants the flowers at ± 7 .5mm long, are rather disproportionately large for a plant only 20 mm in diameter.The habitat preference pro vides the specific epithet for the species-lithos = rock (Greek), -philos -loving (Greek).
Erica kirstenii is confined to rocky places on the Klein Swartberg in the region of Ladismith and Seweweekspoort ± 100 km to the west of the distribution range of E. lithophila.
There are several other species that grow in rocky places and that have similar flowers with large, petaloid bracts, bracteoles and sepals edged with teeth or stout plumose hairs-E.taylorii E.G.H.Oliv.& I.M.Oliv.(Oliver & Oliver 2001) and E. modesta Salisb.The for mer occurs in two very disjunct regions, the Cederberg and the Klein and central Groot Swartberg and produces large spreading mats on rocks at high-lying altitudes, whereas the latter occurs along the summit ridges in the Riviersonderend Mountains and forms a taller, erect shrub up to 0.75 m.
This new species shares similarities with a range of viscid-flowered species such as E. carduifolia Salisb., E. nubigena Bolus, E. ixanthera Benth., E. wittebergensis Dulfer, and two undescribed species-one from the Kouga Mountains (Esterhuysen 10673, 27097), the other from the northern side of the Swartberg Pass (Schumann 798 and Vlok 2502).With these it shares long, 3-nate leaves mostly with sticky glands, often spreading and caducous, bract-like leaves at base of flowers and side branches, long pedicel with few to many glands, similar sized and positioned bract and bracteoles, large sessile marginal glands on the sepals, long anther appendages, and ovary with large placenta bearing numerous ovules producing small seeds.It differs in its growth form, being pendulous with thin, very brittle branches, much longer zigzagged and irregularly toothed anther appendages and short-haired ovary.Each of the species has rather specif ic leaves.
Most of the other species grow in shaded, moist places often on southern slopes, E. nubigena and E. carduifolia at high altitudes, E. ixanthera, E. sp.nov.(Kouga) and E. umbratica at a much lower altitude.Erica sp.nov.from the Swartberg Pass is the odd one out as it grows on dry, rocky slopes in open veld on the northern side of the Swartberg and forms erect, woody shrublets.
Erica umbractica has rather viscid flowers which easily collect dirt on them in the very windy conditions that prevail in Meiringspoort.This made Schumann in frustration, refer to this new species as Erica "puer-sordidus"-the dirty boy, when trying to produce perfectly clean material for photographing (Schumann & Kirsten 1992).
Erica umbratica is known only from a few small populations in Meiringspoort (Figure 10) where it grows on very large, shaded rocks, just above the main stream which runs through the poort.It could be postulated to have evolved with the formation of the poort.The plants we have studied grow in cracks and on ledges sheltered under overhangs on a large rocky outcrop where they receive very little direct sunlight.We have thus chosen the name to reflect this habitat, umbraticus = of the shade (Latin).Due to the inaccessible location, material for study is very limited.
The discovery of this rare species was made by Jan Vlok who has an eye for recognizing unusual and inter esting plants-the plants look so unlike an Erica species, and more like some members of the family Asteraceae such as Stoebe.
Diagnostic features: leaves 4-nate; corolla hairy with a distinct swelling at the base; bract and bracteoles small, remote; anthers well exerted, muticous, long and narrow; sparse hairs on filaments and style; pedicel rela tively long, 10-15 mm, with small, sticky, red glands; ovary densely hairy, cylindrical; leaves with very few sclereids.
E. annalis could be related to several long-tubed species that occur in the region-E.wendlandiana Klotzsch, E. densifolia Willd.and E. abelii E.G.H.Oliv.and to a lesser extent E. maximilianii Bolus.The first three species have the narrow, elongated anthers which are included to slight ly manifest, but these are appendiculate, they have the nar row cylindrical ovary but all of them are glabrous, and the hairy corolla.E. wendlandiana has the hairy filaments and style which can be present in E. densifolia but these are lacking in E. abelii.E. abelii differs in having a long bract and bracteoles which are approximate to the calyx, not small and remote.All three species differ from E. annalis in having 3-nate leaves and no similar basal swelling of the corolla.E. densifolia has spike-like synflorescences on the main stems.E. wendlandiana is remarkable in the genus for the large swelling at the base of the style and E. maximilianii differs in having a glabrous corolla and a glabrous, broadly obovoid ovary.This new species is known only from the type local ity where it was found during a survey of water catch ments in the Kammanassie Mountains by a local Nature Conservation officer, Jan de Jaar.The plants are con fined to rock faces, either facing south or north in a sin gle side kloof running east-west.They grow in crevices in the quartzitic rock.On the hot, dry, north-facing cliffs they are the only Erica species present but on the cool er, south-facing slopes they grow on steep sheet-rock with small plants of E. viridiflora subsp.primulina, which is widespread in similar situations in the sur rounding mountains (Oliver & Oliver 2002).The sur rounding vegetation is arid scrub with no elements of fynbos, which only starts appearing ± 300 m higher up the slopes.
The epithet for this new species honours the dis coverer, the Nature Conservation field officer, Jan de Jaar.We have used the word, jaar = year (Afrikaans), annalis = relating to a year (Latin), even though his name is probably derived from De Jager, the hunter (Afrikaans/ Dutch).

FIGURE 5 ,-
FIGURE 5,-Erica schelpeorum.A, flowering branch, natural size: B, stem, with leaves removed; C, leaf-D flower E bracteole-F sepal G stamen side, back and front views; H. ovary; I. ovary opened laterally to show ovules and placentae: J. capsule with one valve removed: K. seed; L, testa cells; M. anther variant with appendages.A-I drawn from the Oliver 11863 & 11879.M from Vlok 2454 Scale bars B-j" M, 2 mm; K, 1 mm; L, 50 pm.