Studies in the Ericoideae . I . The genera Eremia and Eremiella

A revision of the genus Eremia, in which seven species are recognized, and o f the m onotypic genus Eremiella is presented. Both genera belong to the Ericaceae-Ericoideae and are endemic in the south-western part o f the Cape Province. The revision necessitated the inclusion o f the monotypic genus Eremiopsis N .E. Br. under Eremia. This is the first in a series on the minor genera o f the Ericoideae in Southern Africa.


H ISTORICAL O U T L IN E
David Don proposed " An A ttem pt at a New Arrangement of the Ericaceae" in 1834.In this paper he pointed out that the generic characters in the family are not so strongly marked, but on that account a subdivision should not be disregarded.He dis approved of the large size of the genus Erica, which he attempted to subdivide into a num ber of m inor groups giving each of them generic status.
Don upheld the genera Calluna and Salaxis of Salisbury and Blaeria of Linnaeus and described the monotypic genus Eremia basing it on Erica totta Thunb.He separated off the rest of the species of Erica into sixteen minor genera equivalent to the sections used by Guthrie and Bolus in Flora Capensis (1905).This arrangement was repeated the same year in his brother's work (G.Don, 1834).
Klotzsch (1838a) upheld Eremia and added E. bartlingiana based on Drege material in Berlin.Rach (1853), in examining the Thunberg herbarium concurred with Klotzsch in treating E. totta and E. bartlingiana as two distinct species.The species was recognized by subsequent authors until N. E. Brown (1905) rightly showed that there was no justification for keeping it separate from E. totta.In the addenda to his slightly earlier work, Klotzsch (1838b) described Eremia parviflora and Eremia recur vat a.The former is now placed under Grisebachia.
The following year Bentham (1839) rearranged the Ericaceae in his work for De Candolle's Prodrom us.He reduced most of Don's genera to sections of Erica and upheld Eremia in a much enlarged form.He placed Eremia in his subtribe Salaxidae characterized by single ovules in each ovary cell.
Section Hexastemon characterized by a 2-celled ovary, 6 stamens and approximate bracteoles (E. lanata).This enlarged and altered treatm ent o f the genus brought together a heterom orphous collection of species some of which bear no relation to each other at specific level.It was repeated in Bentham & Hooker (1876).
Bolus (1894) described Eremia rhodopis and placed it in Eremia as construed by Bentham. Later Bolus (1905) realized the close relationship between the two species in section Poderemia and some species in the genus Erica, section Euryloma, and proceeded to place Eremia rhodopis and Eremia tubercularis into Erica.
Drude (1897) recognized Eremia, but changed its circumscription entirely by including several previously described genera and subdividing the genus into four sections.He retained the sections Poderemia, Eremiastrum and Hexastemon established by Bentham and added the section Grisebachia based on Klotzsch's genus.This latter section included the genera Finckea, Acrostemon and Comocephalus described by Klotzsch.These were subsequently removed by N. E. Brown (1905) and placed under Grisebachia and Acrostemon as separate genera, a view which is upheld by myself.
N. E. Brown (1905) changed Bentham's treatm ent to one without any sections and removed two of the species included by Bentham to other genera, Eremia lanata (Klotzsch) Benth.to Hexastemon lanatus Klotzsch, upholding the change of Eremia tubercularis (Salisb.)Benth.to Erica tubercularis Salisb.by Bolus in the same work (1905).He retained Eremia parvi flora Klotzsch commenting that he had not seen any material, but later (1906) included it in the synonomy of Grisebachia eremioides MacOwan, after examining Klotzsch's type.This view is upheld in the present work.He also placed Eremia bartlingiana under the synonomy of Eremia totta.
Brown's treatm ent effectively removed the dis cordant elements from Bentham's Eremia thus reducing the generic circumscription to include species with 2-or 4-celled uniovulate ovaries and 8 stamens.Phillips (1926) repeated the arrangem ent in Flora Capensis and retained Eremia unaltered.
When dealing with specimens from the Ceres Wildflower Show, C om pton (1935) came across what he regarded as four distinct undescribed species of Eremia and named them E. florifera, E. peltata, E. calycina and E. virgata.This involved no major alterations to the generic circumscription.Compton pointed out the close similarity between Eremia and Erica particularly in regard to his Eremia florifera.Phillips (1944) put forward his proposals for a reclassification of the family in South Africa for the second edition of his Genera (1951).He stressed the character of ovule num ber as the principal criterion in the subdivision of the family.He believed that the very similar looking minor genera were more uniform in their general appearance than the species in the single genus Erica and should be recognized as such.This latter statem ent is quite correct, but that is as far as it goes.The variation found in Erica is almost entirely in size, shape and colour whereas in the minor genera the remarkable variation is in the num ber of parts per floral whorl and their com binations.
With the above two criteria in mind, Phillips proceeded to recircumscribe all of the genera.He retained Eremia purely on grounds of priority and reduced the following genera to synonomy under Eremia: Platy calyx, Hexastemon, Arachnocalyx, Grisebachia, Acrostemon, Simocheilus, Thoracosperma and Aniserica, in fact all those genera with uniovulate ovaries other than 1-celled.
Phillips appeared to ignore other characters and definite close relationships between genera.He reduced Eremiopsis to synonomy under Scyphogyne thus completely obscuring its close relationship with Eremia with which it is combined in this present paper.Further discussion of Phillip's treatment of the other genera will appear under the relevant genera when published in this series.

M O R PH O LO G Y
In habit most of the species of Eremia are erect or prostrate and spreading.E. totta, E. recurvata and E. curvistyla are usually laxly spreading with E. peltata and E. brexifolia erect and almost fastigiate.M asson's record of the latter at " 3 to 4 feet high" seems very unlikely.E. florifera is described by Com pton as erect.E. calycina forms a low compact shrublet.
The branches of all species are never entirely glabrous.There is usually some short pubescence with another type of hairiness admixed, either long and hispid or gland-tipped.
The leaves are all ericoid and are 3-nate in all species except E. brexifolia in which they are 4-nate.In most species, the leaves are erect and imbricate, but in E. recurxata and E. curxistyla they are characteris tically erect-spreading to much recurved.
The flowers of all species are terminal either at the ends of main branches or more often at the ends of short lateral branchlets.These may then be aggregated together to form a congested pseudospike as in E. totta and E. curxistyla.The 3-bracteolate flowers are usually 3-nate or as much as 10-nate in E. recur xata.In E. brexifolia the bracteoles are so enlarged that the inflorescence of 4 flowers becomes involucrate with the bracteoles giving the basic colour.
The calyx in all species is 4-lobed or 4-partite.In E. curxistyla, E. calycina, E. florifera, E. peltata and E. recurxata there is a distinct tubular base, which is most marked in E. calycina.The tube is distinctly narrow with spreading lobes in all of these except E. recurxata.In E. totta the calyx segments are free or very slightly joined at the base.They are entirely free in E. brexifolia.In all species except E. brexifolia the calyx segments are more or less equal.In this species the type is odd in having the adaxial and abaxial segments different in shape, but similar in size.
The number of stamens in all specimens examined was constantly 8 and is im portant in the generic classification.In E. peltata Compton recorded " stamens 8, sometimes but rarely 6" .This I have not found in any of the flowers examined.The record of C om pton's must have been a chance flower.All stamens are free at anthesis and have slender linear or filiform filaments.In E. recurxata the filaments are dilated towards the base.
The anthers are all manifest or just included, the most visible being in E. peltata.The majority of anthers are bilobed and not bipartite with completely free cells.Only in E. curxistyla and E. recurxata are the anthers bipartite, but not as markedly as is found in Eremiella or Grisebachia.
The pollen is of two types.The grains are in tetrads in E. totta and E. florifera and single in the other species.The former condition is typical of the Ericaceae and is shared with Erica, the latter with many of the other minor genera.
The ovary is either 1, 2 or 4-celled, but always with a single subapical pendulous ovule in each cell.In E. totta and E. florifera the ovary has 4 cells as in Erica.In E. peltata, E. calycina, E. recurxata and E. brexifolia the ovary always has 2 cells and is erect and regular with an apical style.In E. curxistyla the ovary is obliquely 1-celled with the style arising from the eccentric apex and curving upwards.In one collection of this species 2-celled regular ovaries were found in about 80% of the flowers.
As in most species of the Ericoideae, the ovary is seated on a nectariferous disc in E. totta, E. florifera, E. curxistyla, E. calycina and E. recurxata.This is absent in E. peltata and E. brexifolia.
The stigma varies from simple in E. totta, E. florifera, E. calycina and E. recurxata to slightly capitellate in E. totta, capitate in E. brexifolia and peltate-crateriform in E. peltata.
M ature fruits and seeds have not been seen in any of the species.When old flowers are examined it is found that the fruits are either very hard and nut-like with mostly soft juicy seeds inside, if they have developed, or they are soft and thinwalled and dis integrate.In the latter case soft juicy seeds occur unlike the hard dry type occurring in Erica.An examination of developing ovaries shows distinct " suture" lines down the locules suggesting loculicidal dehiscence which does not seem to occur.This problem is met with in the majority of m inor genera and needs a considerable am ount of field study to elucidate.

D ELIM ITA TIO N OF THE G E N U S EREM IA
As defined in the present work the genus Eremia is characterized by having 3 bracteoles, 4 sepals which are free or partly fused, a 4-lobed corolla, 8 free stamens and an ovary with 4, 2 or 1 cell with a single ovule in each cell.The im portant character is the stamen number.
When first described by D. Don in 1834, Eremia was based on the single species E. totta with 4 unio vulate cells to the ovary.The uniovulate condition of the ovary served to distinguish it from the genus Erica and still is the only real differentiating character.Klotzsch (1838) altered the generic circumscription by introducing the 2-celled Eremia recurxata.Bentham (1839) added Eremia brexifolia and proceeded to broaden the limits by including Salisbury's Erica tubercularis, which has more than one ovule per cell and also Klotszch's Hexastemon lanatus with 6 stamens.Bolus (1905) included Erica trichroma, Erica tubercularis and Erica rhodopis in the section Euryloma of Erica on the grounds that they have ovaries with cells generally more than 1-ovuled.He stated under Erica trichroma that the ovule number is not con stantly 1.He examined Niven 147 and found 12 ovules in the 4 cells, in Masson s.n. he found 17 ovules, in Schlechter 10278, 17 ovules and in Schlecther 7530, 6 and 7 ovules.In both Erica rhodopis and Erica tubercularis he regarded the ovaries as 4-celled with sometimes only one ovule per cell.
An examination of numerous flowers of my own collection of Erica trichroma showed 12 ovules per cell (Oliver 4176).I also examined my collection of Erica rhodopis (Oliver & Palser 73) and found most of the ovaries had 7 ovules, i.e. one cell had only 1 ovule, and only a few had 8 ovules.In describing Eremia florifera Com pton pointed to its close resemblance to the section Arsace in the genus Erica, particularly to the material that is classified as Erica copiosa Wendl.and occurs in the Ceres district.The ovule number in this species is, however, at least 7 per cell as opposed to the single one in E. florifera.
Eremia totta and E. florifera are the only 4-celled species in the genus and with their 8 stamens are very closely allied to the genus Erica.The overlap between these two species and the Erica spp.mentioned above suggests the possibility of com bining the two genera.This step would, however, complicate the position regarding the relationship between Erica and Philippia and Blaeria.
The basic ovary character of the genus Erica, which at present contains 625 species, is 4-cells with numerous ovules per cell.The very few exceptions to this should in no way be used to bring about the recircumscription of such a large genus.It will thus be necessary in the treatment of the m inor genera to regard some of the slightly overlapping genera allied to Erica as "genera of convenience" .
The 2-celled uniovulate erect ovary found in Eremia peltata, E. calycina, E. recurvata and E. brevifolia is constant in all the material examined.In nearly all the material examined belonging to Eremiopsis curvistyla it was found that the ovary was constantly 1-celled, uniovulate and oblique with a remarkable eccentric curved style.This appeared to be a very good character for distinguishing this taxon from all other Ericoideae.However, an examination of Esterhuysen 29687 showed that the majority of flowers had 2-celled uniovulate erect ovaries o f the Eremia type and a few of the Eremiopsis type.This factor made it impossible to place the material in either Eremia or Eremiopsis with any certainty.
The question then arose whether to keep Eremiopsis curvistyla as a separate monotypic genus with a partial overlap of the basic generic characters or to incorporate it into Eremia.This had to be looked at in the light of the generic differentiating characters used between other genera in the subfamily.It was decided to incorporate Eremiopsis curvistyla into Eremia and regard it as an aberrant reduced form in the process of evolving into a separate genus.Also, to retain a monotypic genus there should be a very distinct discontinuity with no overlap at all.
In Eremia a further close relationship occurs with certain species in the genus Grisebachia and the monotypic genus Eremiella.Grisebachia exhibits very little variation having constantly 4-stamens and bipartite anthers.It is a relatively natural genus of closely related species easily distinguishable from all other genera.There is a remarkable similarity between Grisebachia parviflora and Eremia curvistyla and to some extent G. minutiflora, all of which have a similar sprawling habit and flower shape.E. curvistyla has the bipartite anthers of the Grisebachia type, but the 8 stamens of Eremia.Undoubtedly these species had some closer ancestral relationship in a Grisebachia-Eremia complex.

P H Y T O G E O G R A P H Y
The genus Eremia is endemic in the south-western and southern parts of the Cape Province corres ponding to the Western and Southern Phytogeographical G roups proposed by Weimarck (1941).There is a conspicuous concentration of species in his north-western centre comprising the W interhoek and Cedarberg Subcentres where six of the seven species occur.E. totta, the most widespread and common species in the genus, is spread further south as far as French Hoek and the Hottentots-H olland in Weim arck's South-western Centre.O f particular interest is the northern record of this species from the Heerenlogementberg where very few Ericaceae have been recorded.The species also occurs in the " island floras" on the Piquetberg m ountains, Riebeek Kasteel and Paarl M ountain and on the Paardeberg near Malmesbury.Of the other species in this area, E. peltata is of interest in extending out of the Bokkeveld onto the Bonteberg near Touws River where isolated patches o f Cape flora occur on the mountains of the eastern end of the Little Karoo.
The only representative of the genus outside the above area is E. brevifolia which is restricted to a single locality, Attaquaskloof, in the Outeniqua M ountains about 200 km east of the rest of the genus.This disjunction should be looked at in conjunction with the occurrence slightly further east of the closely related monotypic genus Eremiella.
Perennial woody shrublets, erect up to 30 cm or prostrate and spreading.Leaves 3-nate, in one species 4-nate, erect, imbricate to spreading and recurved.Flowers terminal usually on short lateral branchlets, often forming congested pseudo-spikes.Bracteoles 3, mostly approxim ate, small and incon spicuous to large forming an involucre around the inflorescence, glabrous rarely pubescent, ciliate.Calyx 4-partite or lobed, small to enlarged and conspicuous; segments mostly equal, sometimes in two slightly dissimilar ranks, in one species slightly unequal, glabrous rarely pubescent, ciliate with simple, glandtipped or plumose hairs, in one species sparsely villous never lanate.Corolla 4-lobed, urceolate, campanulate or cyathiform, glabrous rarely puberulous.Stamens 8, free, included or manifest.Anthers bipartite or bilobed, aristate or muticous.Pollen in tetrads or single-grained.Ovary 2-or 4-celled with a single subapical pendulous ovule in each cell, ovoid, or obliquely 1-celled with a single oblique subapical ovule.Nectariferous disc present or absent.Style exserted or included, apical and straight erect in 2or 4-celled ovaries or eccentric and crooked in 1-celled oblique ovaries.Stigma simple, capitate or peltatecrateriform.Fruit a hard apparently indehiscent nut or soft and dry, breaking by decay.
The genus is divided into 2 subgenera on the ovary characters.

Subgenus Eremia
Ovary 4-celled with a single ovule in each cell.Pollen in tetrads.
The subgenus contains two species, E. totta and E. florifera, which possess the characters used in the original circumscription of the genus.
Low, spreading shrublet to 30 cm high, rarely higher.Branches stout, minutely pubescent all over sometimes sparsely so, with sparse or dense, long, spreading hairs which are minutely hispid mostly more so towards their bases and with few to many simple shorter gland-tipped hairs admixed.Leaves 3-nate, spreading or reflexed but curved upwards, 2-3 mm long without the petiole, angular when dried, narrowly elliptic to linear-elliptic, minutely pubescent all over, rarely glabrous, with long, white, spreading hairs becoming subechinate on the sides and abaxial surface, hairs often hispid when young; petiole usually adpressed, up to 0 ,5 mm long, pubescent or glabrous, with short hairs on the margins.Flowers terminal, (2)3(4)-nate on ends of minute axillary branchlets clustered together at the ends of short lateral branchlets to give a congested pseudo spike; pedicels 0 ,5 -1 ,3 mm long, glabrous or puberulous, sometimes shortly glandular pubescent; bracteoles 3, adpressed to the calyx 0 ,7 -2 ,1 mm long and up to 2 mm broad, elliptic to ovate to depressed ovate, acute, rarely obtuse, keel-tipped, glabrous or occasionally puberulous, serrulate-ciiiate or fimbriate with mostly hispid cilia, not glandular, occasionally with a few stout hairs on the keel-tip, white.Calyx equally 4-partite or occasionally slightly joined at the base, reaching from halfway to completely up the corolla-tube; segments 1 ,7-3,5 mm long and 1 ,4 -2 mm broad, elliptic-oblong to broadly elliptic, sometimes obovate, mostly obtuse and cucullate, with or rarely without a keel-tip, glabrous or puberulous, serrulate-ciiiate or fimbrate, the "cilia" often minutely hispid and gland-tipped mostly towards the apex where the glands become subsessile, occasionally with a few hairs on the keel-tip.Corolla 4-lobed, 2,5-3,9 mm long and up to 2,6 mm wide, mostly urceolate with varying degrees of narrow neck from a large inflated markedly 4-winged base, the wings alternating with the sepals, rarely constricted below
A low, spreading shrublet, rarely erect, occurring frequently on dry m ountain slopes from the Vanrhynsdorp district south-wards to the Caledon district, the most widespread and commonest species in the genus.E. totta is distinct in the genus for its large white flowers which turn yellow-brown when dried.Florally it is easily distinguishable in the subgenus by its large free sepals and bracteoles.
Originally described as a species of Erica by Thunberg the species formed the type of D on's new genus Eremia.Despite this latter fact E. totta is not typical of the species included under Eremia in the present revision.
Several authors found grounds for splitting off additional species from E. totta.Bartling described Erica pectinata and Klotzsch Eremia bartlingiana.These were both based on variations in corolla shape and hairiness of the ovary, characters which have subsequently been found to be variable and of no taxonom ic significance.N. E. Brown reduced the latter species to a variety of E. totta, but I have not upheld this.
There is much variation in the hairs on the ovary from very crisped lanate to glabrous, sometimes with only one or two long hairs present.In his type description, Thunberg makes no mention of the ovary.There are two sheets in his herbarium, a and and N. E. Brown states that both specimens on sheet a have lanate ovaries, while specimen /? has only a few hairs.I have chosen the first as the lectotype.
The variation in the pubescence on the sepals and bracteoles appears to be randomly distributed and not correlated with other characters.This is also true of the sepals and bracteoles which vary in shape and size and in their relationship to the length of the corolla.In Schlieben and Ellis 12453 the sepals are broadly ovate and less than half the length of the corolla-tube whereas in Schlechter 10687 they are elliptic and as long as the corolla-tube ( F i g .3).E. totta is the most widespread and common species in the genus, occurring in dry sandy or rocky areas from the Cedarberg to the Stellenbosch m ountains.It is one of the few Ericaceae which have been recorded from the " island" floras of Paarl M ountain, Riebeek's Kasteel and Paardeberg.A very unusual and interesting record is Taylor 3946 from the Heerenlogementberg in the Vredendal district.
Com pton (1935) drew attention to the close resemblance of this species to members of the section Arsace in the genus Erica, especially to what G uthrie and Bolus regard as Erica copiosa Wendl.var.longicauda H.Bol.The only basic difference between E. florifera and the Erica spp. is in the num ber of ovules.In the latter the number of ovules is 7-8 per cell.
The relationship between E. florifera and the genus Erica is in fact closer than its relationship with E. totta.It would seem highly probable that these two species evolved from ancestral Erica stock completely independently.
No locality can be given for the species as it has only been recorded once at the Ceres Wildflower Show in October, 1934.At the flower shows organized in some centres of the south-western Cape, the majority of exhibits are collected in the local district, but there is no guarantee of this as a record.The Ceres District, however, fits into the distribution of the genus with 5 species occurring there.
Ovary 2-celled or 1-celled with a single ovule in each cell, pollen grains single.
A small erect shrublet in habit, the species occurs on dry sandy slopes in the southern part of the Ceres district.E. peltata is a very distinct species easily recognizable in the genus by its peltate-crateriform stigma.
In some characters it is similar to E. brevifolia.Both species are odd in not having the typical ericoid nectariferous disc below the ovary.This, coupled with the dull greenish white colour of their flowers, manifest to semi-exserted anthers with large pores and their stigmas, peltate in E. peltata and capitate in E. brevifolia, suggests that they are both wind pollinated.All other species in the genus are either distinctly white or pink with simple stigmas.
The species is fairly uniform over its distribution range.The only noticeable variation occurs in the anthers, which vary in shape sometimes in the same flower.They sometimes have a remarkable prog nathous base and occasional long colourless hairs of the type found in E. brevifolia.Compton recorded that there were 8 stamens " sometimes but rarely 6" .This I have not seen in any of the specimens I have examined.E. peltata occurs in a distinct part of the distribution range of the genus in the southern and eastern parts of the Ceres district, where it grows on dry sandy, stony mountain slopes.Only in the Cold Bokkeveld around Gydo and Sandberg does it overlap with other species.The species occurs on the Bonteberg north of Touws River where islands of the Cape Flora occur in karroid vegetation types.
The species flowers from September to December.The exact locality of the type specimen is unfor tunately not known as it came from the Ceres Wildflower Show.

E. virgata
In his paper in 1935 Compton described four species.He described E. calycina characterizing it by the large broad calyx segments, which are inflexed and broadly cucullate at the apex and which exceed the corolla-tube in length.The latter statement is, however, not borne out by the type description or figure (cf.Fig. 7 of the present paper) where the calyx is just shorter than the corolla-tube.The former characters could easily apply to one o f the other species he described, E. virgata.This latter species he claimed, was well distinguished in the genus by its erect virgate or tufted growth and its very short adpressed leaves.This, in fact, also applies to E. calycina.There appeared to be no difference between the two species.
As both erect, virgate and compact to spreading shrublets occur with varying degrees of calycine flowers, the two species were combined under the more appropriate name, E. calycina.The first name was also preferred, because the type locality of E. calycina is known whereas the type of E. virgata was recorded from the Ceres Wildflower Show.Low, spreading shrublet.Branchlets slender, some times long and entwining in surrounding vegetation, puberulous with simple or glandular hairs, the grey bark splitting irregularly and flaking off to reveal red-brown wood, with leaves mostly only on the young branchlets.Leaves 3-nate, curved spreading to recurved, crowded, up to 1,7 mm long excluding the petiole, mostly 1,0 mm long, oblong-lanceolate to ovate, somewhat trigonous, flat above towards the petiole, convex towards the apex, sparsely pubescent on upper surface when young becoming roughly scabrid, otherwise glabrous with an apical gland-tipped hair and with 3 or 4 on each side which soon become short setae, sometimes also shortly ciliate; petiole up to 0,5 mm long, adpressed ciliate.Flowers (2) 3 (4)nate on ends of short lateral branchlets, usually pendulous with 1, 2 or 3 branchlets in a whorl forming a congested spike-like group; pedicel very short, about 0 ,1 -0 ,2 mm long, glabrous; bracteoles 3, approximate but spreading, subequal, the median broadly ovate, acute, 1 x 0 ,6 mm, the laterals broadly elliptic, obtuse, oblique, all long ciliate with a few short gland-tipped hairs towards the apex otherwise glabrous, soft scarious, white becoming reddish, sometimes with a green leaf-like apex, keel-tipped.Calyx 4-lobed, divided from two-thirds to three quarters its length, up to 1,8 mm long, cyathiform from a narrow base; tube up to 0 ,5 mm long, quad rangular; lobes up to 1 ,4 x 1 ,3 mm, orbicular to broadly elliptic, curved and spreading upwards from the narrow tube, like the bracteoles soft scarious, white becoming reddish with an acute distinct wide keel at apex, glabrous or pubescent at apex and inside the keel, long-ciliate and with many to a few short glandular hairs towards the apex.Corolla 4-lobed, divided for about a third of its length from 1 ,7 -2 ,5 mm long, obconical to campanulate with a short pouch-like base, sometimes 4-angled with distinct ridges down the lobes, very sparsely pubescent on the lower half only, sometimes with a few hairs up the lobes otherwise glabrous, white; lobes very broad, obtuse or acute, erect with acute interstices, smooth or crenulate.Stamens 8 free; filaments filiform, glabrous; anthers bipartite, 0 ,4 -0 ,8 mm long, included or manifest, haphazardly arranged, narrowly oblong, dorsally attached near the base, scabrid, pale brown with darker red-brown backs, aristate; awns coming off the filament at the point of attach ment, about quarter the length of the cell, scabrid; pore about one-third the length of the cell; pollen grains single.Ovary l(2)-celled, 0 ,7 x 0 ,6 mm, pubescent at apex, rarely completely glabrous; when 1-celled with a single, oblique, pendulous ovule, obliquely ovoid abaxially with the style arising from the eccentric apex of the ovary and curved upwards; when 2-celled with a single pendulous ovule in each cell and apical style; style crooked at the apex markedly so in the bud, included or exserted and becoming straighter, glabrous; stigma simple.F i g .9 A species sprawling in habit, sometimes forming mats, on dry m ountain slopes in sandy rocky places from Bainskloof northwards to Citrusdal and Piquetberg.Brocteole length On ovary structure E. curvistyla is distinct in the genus having an oblique 1-celled ovary and markedly eccentric curved style.In its sprawling habit it is similar to E. recurvata, but is easily distinguishable by its shortly ciliate calyx as opposed to the long hirsute calyx of the latter.
The species is very similar to Grisebachia parviflora (see under Excluded Species) which was originally erroneously described as a species of Eremia.A comparison with the figure of this species to be published later in this series will show the remarkable similarity.The most noticeable difference lies in the generic character of stamen num ber-8 in E. curvistyla and 4 in G. parviflora.Externally, the large bracteoles and sepals of E. curvistyla are very different from the narrower ones of the latter species.The two species are sympatric with G. parviflora having the wider distribution range.
This species was first described by N. E. Brown as a separate monotypic genus, Eremiopsis, characterized by its unusual ovary, oblique, 1-celled with an eccentric markedly curved style.On first examination this condition suggests an abortive 2-celled ovary.An investigation of 14 different collections showed this character to be constant.However, a collection, Esterhuysen 29687 from the northern Cold Bokke veld, when examined was found to have an equal quantity of 1-celled oblique ovaries and 2-celled erect ovaries as in Eremia.
Consideration was given to retaining the species in a separate monotypic genus.With the above definite overlap of generic characters it was felt that separation could not be upheld.The concept of " a genus of convenience" , which will appear several times in the treatment of the Ericoideae, could not be considered in the case of a monotypic genus being kept separate from an already variable genus.flat at base becoming thick and rounded towards apex, at first minutely puberulous above becoming glabrous with some long spreading hairs which may be gland-tipped, also with subsessile glands, apiculate with a long gland-tipped hair; petiole adpressed, shortly ciliate.Flowers 1-10-nate on ends of branches; pedicels up to 1,5 mm long, pubescent sometimes with long white hairs as well; bracteoles 3, subapproxim ate to approximate, usually more or less spreading, usually leaflike, up to 1 ,7 x 0 ,9 mm, the median broader than laterals which may be linearoblong and acute, all pubescent towards the base, long-ciliate, often gland-tipped, also with sessile glands on margins, slightly sulcate.Calyx 4-lobed, joined for quarter half its length, up to 2,3 mm long; lobes up to 0 ,9 mm broad, narrowly ovate-acute to ovate-acute, thinly beset with long soft hairs and with dark sessile to subsessile glands on edge of inner surface, edged with fine short hairs towards the base and finely pubescent, sometimes glandular, very slightly sulcate, forming a cup at the base, often with 4 main ridges opposite each segment and 4 opposite the interstices.Corolla 4-lobed, to 3,5 mm long, conical to campanulate, obscurely to markedly 4-angled, glabrous except for thin pubescence near apex of each lobe and sometimes spread down the ridge, very noticeable in bud stage, white; lobes very broadly rounded, erect.Stamens 8, free, included; filaments up to 1,2 mm long, linear to narrowly oblong-elliptic, glabrous, transparent but dark at the apex, tapering into the sigmoid apex below the anther; anthers up to 0 ,7 mm long, narrowly ovoid, narrowed upwards with contiguous separate cells, bigibbous at the base, minutely scabrid, awned, golden brown; awns narrowly lanceolate, from half to equal the length of the cell, ciliate, pale; pore very small; pollen grains single.Ovary 2-celled, with a single pendulous ovule in each cell, 0 ,5 mm long and broad, broadly ellipsoid to globose-ovoid, obtuse, very sparsely pubescent to pubescent at the apex; style filiform, glabrous, included to shortly exserted; stigma simple.F ig .12. The variation found in E. curvistyla occurs in the size of the flowers and in the shape and size of the bracteoles.Fig. 11 shows the distribution of bracteole sizes (length x breadth) of 15 collections.The median bracteole was included as it is subequal in nearly all of the collections.The specimens are num bered in order of increase in size ratios and are plotted on the map accordingly.
There is a distinct disjunction in bracteole size at line A-B.The odd ratio is of an unusually developed median bracteole and can be ignored.Unfortunately only one of the three specimens is localized, Edwards s.n.from the Olifants River M ountains above W arm baths.W ithout further exact collections in this group no taxonomic recognition can as yet be given to it.In the future, if there are further collections an assessment of this disjunction can be made.
The remainder of the collections are slightly separable along line C -D given two foci X & Y of maximum overlap.This is correlated with a slight distributional grouping as shown in the map.The collection Oliver 4072 (No. 9) is anomalous falling on the wrong side of the line C -D .Pillans 10558 (No. 14) is similarly anomalous.
In his type description N. E. Brown cited three syntypes, but labelled the sheet of Niven s.n. as the type.Niven s.n. is now selected as the lectotype.
Shrublets, sprawling or erect up to 30 cm.Branches rigid, spreading, minutely puberulous with a few glands or rarely gland-puberulous, with later some long white hairs admixed which are sometimes hispid, becoming glabrous and grey, occasionally with distinct ridges below the leaf-bases when young.Leaves 3-nate, erect, spreading-recurved to squarroserecurved, 2 x 1 mm, ovate to broadly ovate, fairly This species can be easily recognized by its longciliate calyx, conical often markedly 4-angled corolla, dilated filaments and anthers with very small pores.In habit and general appearance it is very similar to E. curvistyla.It has no closely related species.E. recurvata does not have much variation.Bentham described the stamens as mostly 8, sometimes seen as 6 or 7.This I have not found in any of the material I examined.
In some specimens from the northern part of the distribution range e.g.Stokoe in SAM 55051 from Krakadow and Esterhuysen 7446 from the Pakhuis Pass area, there is a tendency for the plants to possess more glands on the branches, leaves, bracteoles and sepals.On the branches in particular there are shortstalked glands in addition to the longer gland-tipped hairs.
The Drege neotype seems to be the only collection with narrow linear bracteoles; all others have broader more elliptic leaflike bracteoles, particularly the median bracteole.Klotzsch in his type description describes the bracteoles as linear, thus confirming his examination of the Drege material.However, he describes the ovary as glabrous.This is somewhat odd as the Drege material I have dissected is the most pilose of all the specimens examined.
The holotype in Berlin is no longer extant.Klotzsch gave only " Cederbergen-Drege" in his type description.Numerous Drege specimens were dis tributed to herbaria, some numbered, some with a locality in the Cedarberg and some with a locality which cannot be traced with any certainty (Blaaw berg).The material distributed as " Cedarberg prope Ezelbank" is labelled as Drege 2965 in ten herbaria.Unfortunately most of the specimens are rather scanty.The specimen in Edinburgh is better than most and is chosen as the neotype.
Two distinct forms can be recognized in the available three collections based on the sepal shape:  All three collections come from the same general area, Attaquaskloof, with only Oliver 4128 being an exact locality.Niven 85 matches Oliver 4128 and most probably came from the same population.Without any exact information about the spatial relationship between the two forms A and B no further comment can be made at present.Only when a thorough examination of the A ttaquaskloof m oun tains has been made can the correct taxonomic position of these forms be worked out.
Masson travelled through the A ttaquaskloof in the company of Thunberg in November 1773.
Bentham described Eremia brevifolia in DC.Prodr.and based his description on two specimens, one in De Candolle's herbarium, the other belonging to Lee, both according to Bentham collected by Masson.The specimen in Geneva is, however, Niven 85, not a Masson collection.Thus Bentham unknowingly based his species on two syntypes.
It has been stated above that the Masson and Niven collections can be assigned to two different forms on the shape of the sepals.Bentham's type description describes the calyx as " calycis laciniis oblong-spathulatis" .He could thus have only been referring to the Masson sheet in Herb.Lee, now at Kew.This has been selected as the lectotype.The Niven sheet in Geneva has not been dissected and examined critically to see whether it matches the Niven material in the British Museum or Kew.I accept that they are duplicates.
There is some confusion about the collectors of the early specimens as N. E. Brown (1905) queries the collector of Niven 85.The label on the specimen in the British Museum was in my opinion written by Niven as No. 85.The sheets marked No. 57 are in a different hand and labelled as Fr.Masson, but there appears to be no guarantee of this as the hand writing does not match exactly that on letters in the Kew Archives.N. E. Brown complicated the issue by changing an almost identical label on the type of Grisebachia niveni from Fr. Masson to Niven.
Masson states on the label of the Kew specimen that the plants are 3 or 4 ft high.Those seen by myself on the summit of the Voortrekker Pass were at the most 1 ft (30 cm) high and this was in old established vegetation that had not been burnt for some considerable time.
The taxonomic position of Eremia brevifolia is somewhat obscure.It is placed within the genus Eremia on a summation of characters, but bears no relationship to any of the other species included in that genus.It is the only species in the genus with 4-nate leaves.As has been stated under the phytogeographical section, the species is geographically far removed from the main distribution centre of the genus in the south-western Cape.
The possibility cannot be ruled out that evolution of this species has taken place independently of the rest o f the genus Eremia, both having evolved from ancestral Erica stock by reduction in the cells of the ovary and number of ovules.
In general appearance, there is a remarkable similarity between Eremia brevifolia and Eremiella outeniquae Compton, which occurs on m ountain peaks not far east of Attaquaskloof.Both have 4-nate similar leaves and flowers in small, terminal, capitate clusters where the bracteoles are enlarged and larger than the sepals.Both have a similar habit, although the latter can form low compact shrublets.Both grow in similar open dry places at high altitudes in the Langeberg and Outeniqua M ountains.
The name is the diminutive form of Eremia and refers to the superficial resemblance to some of the species in that genus.
Perennial, woody, shrublets erect up to 30 cm or low and compact about 15 cm high.Leaves 4-nate, erect or spreading.Flowers 4-nate, in terminal erect or semipendulous heads.Bracteoles 3, ad pressed to the flower, subequal to unequal, conspicuous and larger than the sepals, sparsely pubescent and subscarious.Calyx 4-partite, small compared to the corolla, sparsely hirsute and subscarious.Corolla 3-lobed, divided for about quarter of its length, cyathiform, sparsely hirsute.Stamens 6, free, included.Anthers bipartite with the cells completely free and shortly stalked, basifixed, muticous.Pollen grains single.Ovary 3-celled, with a single, pendulous, apical ovule in each cell, broadly ellipsoid and much enlarged in fruit.Nectariferous disc distinct.Style filiform, farexserted, deciduous.Stigma capitellate.
A monotypic genus confined to the Outeniqua and Tzitzikama M ountains o f the Southern Cape.
The genus was described by Com pton very recently in the history of the minor genera of the Ericoideae.He found when identifying iiis collection from Ruyterskop near Mossel Bay that he was unable to place it in any known genus.After summing up the characters he decided that it was worthy o f recognition as a distinct new monotypic genus.
The possession o f 3 bracteoles, a 4-partite calyx, 3-lobed corolla, 6 free stamens with markedly bipartite anthers and a 3-celled ovary with a single pendulous ovule in each cell serves to distinguish Eremiella from all other genera in the Ericoideae.
The most closely related genus is Eremia which differs in having 4, 2 or 1-celled ovaries, 8 stamens and a 4-lobed corolla.To have included this species in Eremia would have meant a further considerable emending of the circumscription of that genus.It was decided to retain Eremiella as a distinct monotypic genus.
In distribution Eremiella is far removed from the majority of species of Eremia which occur in the Western Cape.But the anomalous species, Eremia brevifolia, grows in an area adjacent to the type locality on Ruytersberg and is superficially similar to Eremiella outeniquae.
Eremiella has undoubtedly evolved from some ancestral Erica stock by way of a reduction in the number or floral parts, i.e. 4 to 3-lobed corolla, 8 to 6 stamens, 4 to 3-celled ovary and num erous to one ovule per cell.In this evolutionary pathway Eremia brevifolia could well have been involved, but resulting in a combination of characters which classifies the species as belonging to the genus Eremia.
The single species Eremiella outeniquae forms a woody shrublet, erect up to 30 cm or low and com pact about 15 cm high growing at high altitudes near the summits of a few scattered peaks in the Outeniqua and Tzitzikamma M ountains.
The type locality, Ruytersberg just east o f the Robinson Pass in the Mossel Bay district, was visited to establish the habitat of the species.Here the plants were very localized on dry stony slopes facing south-west and west near the summit.The associated vegetation was very sparse and consisted almost entirely of low scattered restiad clumps.The habitat appeared to be remarkably dry, but no doubt water is regularly deposited in the area by the south-easterly and southerly cloudbearing winds in summer.
Low, compact shrublet up to 15 cm or erect soft shrublet up to 30 cm.Branches slender, hirsute.Leaves 4-nate, up to 2 mm long without the petiole, erect and imbricate to spreading with adpressed petiole, narrowly ovate-oblong to elliptic-oblong, acute, semi-openbacked, hirsute with long simple hairs and sometimes with a few gland-tipped hairs on th e lower edges, ciliate with subsessile glands towards the apex which ends in a gland-tipped hair; petiole up to 0 ,7 mm long ciliate.Flowers terminal, 4-nate, in closely packed heads which are erect or semipendulous; pedicels from very short up to 0,75 mm long, glabrous or pubescent; bracteoles 3, subequal to unequal, approxim ate and clasping the flower, the median 1 ,5 x 1 ,0 mm, elliptic-oblong, the laterals 1 ,5 x 0 ,6 mm, narrowly oblong to obovate, all acute, ciliate and sparsely pubescent on lower half, with a few dark sessile glands near the apex, keel-tipped, subscarious, greenish to reddish pink.Calyx 4-partite, reaching up to the corolla interstices, 1 ,5 x 0 ,5 mm, narrow ly ovate-oblong to oblong from a broader base, ciliate and pubescent, with subsessile glands tow ards the apex, acute, subscarious, greenish to reddish pink.Corolla 3-lobed, divided about £ its length, 2 x 2 mm, cyathiform, sparsely hirsute to sometimes almost glabrous with hairs only on the lobes; lobes broad, obtuse, emarginate, sometimes ciliate towards the interstices.Stamens 6, free; filaments filiform, glabrous; anthers included, bipartite, 0,75 mm long, with the cells obovoid, completely free and shortly stalked, basifixed, minutely scabrid, pale brown, muticous; pore about % the length o f the cell; pollen grains single.Ovary 3-celled, with a single pendulous apical ovule in each cell, about 0 , 5 x 0 ,5 mm, broadly ellipsoid becoming broadly ovoid in fruit, with a nectariferous disc; style filiform, far exserted, glabrous, deciduous, up to 2 ,5 mm long; stigma capitellate.F i g .16, reproduced from C om pton's original drawing.
A low com pact shrublet confined to the summits of a few high peaks in the O uteniqua and Tsitsikama M ountains of the southern Cape.This is the only species recorded in the genus.In the five collections so far made there is very little morphological variation.The only noticeable variation occurs in the leaves, those from the Form osa Peak and Jonkersberg collections being more erect and imbricate than in the type which has more widely spaced and patent leaves.
The four localities in which the species grows occur in two disjunct areas.The three, Ruytersberg, Jonkers berg and Cradockberg, are in the mountain range north of Mossel Bay and George.Stokoe's record from Form osa Peak in the Tsitzikam m a M ountains is 115 km to the east.A search will have to be made of all the intervening high peaks to establish if this disjunction is real or due to a lack o f records.
The collections all possess mature flowers only.Those o f Zinn and Stokoe collected in January have considerably enlarged ovaries.A visit to the Ruyters berg population in mid-November showed that the true flowering period must be from about September to December.
1.-D istrib u tio n o f the genus Eremia.
Erica trichroma and Erica tubercularis are closely related and bear no resemblance to any of the species of Eremia.Erica rhodopis on the other hand has a superficial resemblance to Eremia calycina, but is more closely related to the former two species.It occurs in the Houw Hoek to Hermanus area.It would appear from the above that the key character to use in separating Eremia from Erica both of which have 4-celled ovaries is: Ovules 4 or less per ovary.................... Eremia Ovules more than 4 per ovary............. Erica