Studies in the Ericoideae ( Ericaceae ) . XIV . Notes on the genus

Recent investigations of all the recorded southern African species of Erica have been undertaken to establish the characters of importance in generic delimitation in the subfamily Ericoideae. As a result several anomalies have been found in the circumscription and identification of certain species and in the nomenclature of others. The corrections and alterations are published here.


INTRODUCTION
The last full revision o f the southern African members o f the genus Erica was provided in Flora capensis by Guthrie & Bolus (1905).Subsequently numerous new spe cies have been added by many workers with a major con tribution being the abbreviated revision of the genus by Dulfer in 1965.In this work Dulfer did much to sort out the nomenclature and references to our species in the lit erature.Unfortunately he did not examine the complete collections in any of the major herbaria housing material from southern Africa, namely BM, BOL, K. NBG, PRE and STE.His revision also lacked descriptions of the spe cies.
Since the revision by Guthrie & Bolus (1905) some 180 additional species have been added which has made it increasingly difficult to gain an overall view of the genus.Within recent years we have come across anoma lies in the circumscription of certain species, in the appli cation o f nam es and in the identification o f certain collections.These records and observations are rational ized and published in this paper.

OVARY COMPLEMENT
The genus Erica is generally characterized by having a 4-locular, many ovuled ovary that forms a dehiscent capsule.Guthrie & Bolus (1905) recorded that the ovary was mostly 4-celled, very rarely 8-celled.They noted under the species treatments that the 8-celled condition occurred in only E. perspicua Wendl.and E. verticillata Berg, and 4 -8 in E. propendens Andrews.
During our detailed examination of all 657 species of Erica, several additions to this list of species having non standard ovaries have been recorded: E. pinea Thunb.
The ovary complement of this species was never men tioned by Guthrie & Bolus so we presume they regarded it as having the standard 4-locular ovary.We have found the species has 8-locular ovaries, but some collections from the Hemianus area exhibit 6-loeular ovaries.

Erica alexandri
Until recently this species was known from only three collections from sandy flats in the Paarl area: the type collection by Alexander-Prior of 1847, a second collection of his in K, not cited in Guthrie & Bolus (1905) and a third one by F.H. Cooper in 1922.The spread of farms and pine plantations has destroyed almost all natural veg etation on sandy flats in the area.Searches for the species in likely places in the Paarl area have not met with suc cess.
The species was, however, recently found growing in a wetland near the Wemmershoek sawmill east of Groot Drakenstein towards Fransch Hoek.These plants match those of Alexander-Prior and Cooper exactly and are the only known surviving population of the species.E. alexandri is strikingly similar to E. acockii, which is known only from a few collections made in the 1930's from the Brackenfell/Kraaifontein area and is listed as ex tinct in the South African Red Data Book (Hall & Veldhuis 1988).It has not been found since, in spite of searches of the area, including the type locality, pointed out to one of us (EGHO) by Acocks in 1965.It must be assumed that the species has been destroyed by housing development and alien acacias.
E. acockii grew in very similar conditions to E. al exandri, namely sandy flats with some indication of water seepage just below the surface.This factor may be rele vant to the fact that both species flower(ed) in the driest months of the year, February through to early May.The localities of the current and former populations of the two species are only about 40 km apart.
Morphologically the two species are very similar de spite the fact that they were placed in two different sec tions in the genus, E. alexandri in section Pachysa and E. acockii in Section Pseuderemia.Both have heads of mauvish pink sticky flowers with manifest dark anthers, long gland-tipped hairs on the leaves, stems, pedicels, bracts, bracteoles and sepals.There are only some slight differences between the collections from the two areas.The only clear difference lies in the leaves of E. acockii which are shorter and more rounded than those of E. al exandri.Surprisingly the type collection of E. acockii has obovoid-obconical corollas with more erect lobes as op posed to the urceolate corollas w ith spreading to retlexed lobes in the other collections of E. acockii and of E. al exandri.

Erica auriculata/E. greyi
Both these species were based on a single specimen and published in the same work (Guthrie & Bolus 1905).The type of E. auriculata had only a few flowers which were still in bud, and according to the authors, anther appendages unlike any other species.On account of its puberulous flowers Guthrie & Bolus (1905) placed the species in Section Ephebus.The type of E. greyi had nu merous flowers in 6-8-flowered heads, and the authors made no reference to any distinctive feature of the anthers.They placed it in Section Pseuderemia.
Comparison of the two types showed that their anthers are identical.Other characters of flowers and vegetative parts also matched, and there is no doubt that the two collections belong to the same species.
No other collections of these two species have been made and a search for them in the more exact locality given by Schlechter has not produced any plants.Unfor tunately the whole area was burnt in 1991 and thus it will be several years before further searches will be worth un dertaking as it would appear that the plants are reseeders.

Erica ceniciflora/E. sphenanthera
Both of these species have been known only from the type collections.They were both placed in Section Didymanthera on their possession of long tubular flowers with exserted anthers.An examination of these collections has shown characters not previously noted in assessing their true relationships, namely the shape of the anthers with minute decurrent appendages and fringed corolla lobes.These characters allied the two species to E. grandiflora which is highly variable in the position of its anthers which may be included to exserted even further than other species of Section Didymanthera.
There are clearly no characters to separate the above two species from the older E. grandiflora and so they are here reduced to synonymy under the older epithet.

Erica esterhuyseniae
When Compton described this species he provided a Latin diagnosis and full English description, but did not cite a type.However, he stated that there were clearly two distinct varieties, var. a tetramera and var.b trimera, for which he provided the diagnoses and holotypes.Accord ing to the current Code this would automatically imply that there was a typical variety var.esterhuyseniae with the same type as the species name.Com pton's application o f the term 'Var.a' would sug gest that he intended this variety to be the typical variety but that he supplied the incorrect epithet for it.We are therefore replacing Com pton's name with the correct au tonym and regarding the holotype of his var.tetramera as the type o f the species name.African Botany 7: 193 (1941);Dulfer: 75 (1965).Type: Uniondale Div., Kamanassieberg (Mannetjieberg), 1 700 m, 1 Feb. 1941, Compton 10557 (NBG!).

Erica ft I (form is
Guthrie & Bolus (1905) recorded the species as having a variable number of stamens, namely usually 8 but often 7-4.They noted that 4-stamened specimens were 'tech nically Blaeria and not E rica'.In a recent paper (Oliver 1993) this species was discussed in detail in connection with the relationship between Erica and Blaeria.An ex amination of all the available collections of E. filiformis showed that some collections have 4-stamened flowers, others 8-stamened flowers, with only one collection show ing any intermediate numbers.
In the same work Bolus (1905) described the variety, var.maritima Bolus, from a single collection made in the Agulhas area.An examination of the type showed that the 4-stamened flowers look superficially very similar to those of E. filiform is, but do not possess the 4-locular, multiovuled ovary typical o f Erica.Instead the flowers have 2-, rarely 1-locular ovaries with a single ovule per locule, a feature characteristic of certain minor genera.The ma terial was in fact an isosyntype of Acrostemon schlecliteri N.E.Br.!

Erica gillii
Until recently the type at Kew was the only known collection of this species with none apparently occurring in southern African herbaria.Various recent collections, Oliver 4127, Schumann 793 and Vlok 2499, made in the Attaquas Kloof area have turned out to be this species when compared with the type.Also an investigation of the type sheet of Erica rhodantha Guthrie & Bolus in Section Polycodon showed that two collections were mounted on the sheet, one being the type, Galpin 3706, from Garcia's Pass and the other being a collection of E. gillii, Bolus 11339, also from Garcia's Pass.A duplicate of the Bolus collection turned out to be included with E. rhodantha in STE.The two species are superficially very similar, but may be distinguished on their anthers, black with a large apical ridge in E. gillii, typical of the Section Melastemon, versus brown and not ridged in E. rhodantha.This latter species was also poorly represented in herbaria with, surprisingly, only a few old collections from the Garcia's Pass, an area which has been very well collected.It has also recently been turned up on the lowest northern slopes of the Langeberg west of Garcia's Pass (Oliver 10242).

Erica leptostachya
The type and only authentic material attributable to this name came from the cultivated collections of William MacNab.Many of M acNab's collections of heaths culti vated in Edinburgh in the early 1800's were known to have originated from material collected at the Cape, par ticularly through the efforts of James Niven.So when de scribing this species Guthrie & Bolus noted 'though its origin is somewhat uncertain, it is most probably South African'.A close examination of the fragment of the type in BOL showed that the material identified in herbaria as E. leptostachya did not match the type which was recog nized as being identical to specimens in STE of the Eu ropean species.Erica scoparia L. This species is charac terized by small, creamish green, wind-pollinated flowers with large red stigmas.The flowers are borne on absolute brachyblasts in the axils of foliage leaves up the main branches, thus giving the appearance of a pseudoraceme.Dulfer (1965) identified two sheets of unlocalized ma terial in W as clearly belonging to this species, but also did not realise the European connection.Bothalia 24.1 (1994) Dulfer (1963) described a variety of Guthrie & Bolus' species as var.glabra Dulfer, based on a recent collection from Mariepskop in the Transvaal.He noted that his type material had originally been identified as E. leucopelta Tausch var.ephebioides Bolus.Bolus' type also came from Mariepskop. as do the type and paratype of another species described by Dulfer, E. merxmuelleri Dulfer.The specimens of the latter species appear to us to be E. natalitia Bolus.
Dulfer's taxon can clearly not be part of the European E. scoparia.Until such time as we can sort out the tax onomic problems of the E. leucopelta/natalitia/ merx muelleri complex this variety must be left unplaced, if indeed it is a sound taxon.

Erica longisepala
When describing E. longisepala, Guthrie & Bolus (1905) placed it in the long-tubed Section Pleurocallis and allied it to E. grandiflora.They commented that the spe cies was a link between the Sections Pleurocallis and Hermes.Both of these sections are characterised by a pseudospicate inflorescence and are separated only on the size of the flowers.Guthrie & Bolus noted that their spe cies was related to E. parilis Salisb. in Section Hermes.
Since their publication much material of E. parilis and E. longisepala has been collected which shows a consid erable amount of variability in the size of the flowers from as short as 5 mm up to 25 mm.There is no clear demar cation o f a boundary between the two species.Esterhuysen (1963) noted that these two species were synonymous, but Dulfer (1965) did not follow this.We confirm Esterhuysen's opinion with the following synon ymy.

Erica minutissima
An examination of a portion of the type which Bolus fortunately acquired from Berlin has revealed that the specimen is a monstrosity having a double calyx, no co rolla and no stamens.Bolus noted on his sheet that there were two sheets of this 'species' in Berlin, labelled as locality 83, 'Stellenbosch Umgebung Somerset West'.In BOL there is a collection of Zeyher (?Ecklon) which matches the Berlin fragment exactly, but is labelled as moist places at Bloemendal.The material in our opinion is referrable to E. quadrangular is which occurs sporadi cally in the area and which must have been very common before the spread of housing and farms.E. m inutissim a K lotzsch ex Benth.: 691 (1839); Guthrie & Bolus: 223 (1905);Dulfer: 107 (1965).Type: in montibus Hottentots Holland, Ecklon & Zeyher s.n.(B.holo.f;BOL, fragm.!).

Erica monadelphia
The name of this species is problematic in having no relevance to the species.The anthers of the species are not joined together into one unit as they so clearly are in the anomalous E. embothriifolia Salisb.
There has been much confusion with the publication dates of Andrews' species, particularly those in the first volume of his folio edition.Coloured engravings o f heaths, which appeared in the four bound volumes pub lished from 1802 onwards.This led Dulfer (1965) to change the authorship, as given by Bentham ( 1839) and G uthrie & Bolus (1905), from A ndrew s (1802) to Willdenow (1799).
Andrews produced his first drawings in separate parts of three loose plates each, before he had them bound as the first volume.Some, but unfortunately not all, of the plates in the first volume of his work are dated.Recently an excellent set of the first 22 parts of his Coloured en gravings was found in the library of the Farm Vergelegen, Somerset West.From this set the publication date of E. monadelphia was June 1, 1797 in part 8.A note on the Andrews' publications on heaths is being prepared for publication.

Erica priorii
Until recently this species was known only from the type collection in Kew, collected by Alexander-Prior near George in 1847.Some recent collections, Vlok 1120 & Schumann 590 from Karatara Forest Reserve, were com pared with the type and found to be this species.Subse quently a collection made by Keet in 1920 from Spitzkop, Knysna and placed under incertae and one by Taylor from the Outeniqua Pass in 1962 placed under E. coarctata Wendl., have been found to be this species.
The fresh material from Vlok has shown us that the speeies is a distinct one favouring moist, humic, condi tions on steep, southern slopes of the Outeniqua Moun tains.The shrubs grow to 1.2 m tall and bear bright pink flowers in dense pseudospikes towards the ends of the branches.

E. recurvata
This name has been maintained in all major revisions of the family with the citation o f no collections from the wild, only material cultivated in Europe during the early 1800's.From the scraps of cultivated material available for us to examine we find that the material is clearly a lush cultivated form o f E. cumuliflora Salisb.with its dis tinctive corolla lobes.These lobes are relatively long, subspathulate and suberect forming windows around the base of the corolla interstices and edged with short hairs in the window areas.Only two species o f Erica possess these peculiar lobes, E. cumuliflora and E. genistifolia Salisb.The former has 4-nate leaves and denser heads of 5-12 flowers whereas the latter has 3-nate leaves and fewer flowers (3-4) per inflorescence.E rica cum uliflora Salisb. in Transactions of the Linnean Society 6: 336 (1802); Benth.: 657 (1839); Guthrie & Bolus: 237 (1905); Dulfer: 112 (1965).Type: Hottentots Holland.M ulder s.n.(K, holo.!).
Erica revoluta/austrovem a W hen going through the collections of incertae in BOL.PRE and STE, several specimens were found that matched two recently described species, the Transvaal ma terial being E. revoluta in Section Arsace and the Natal material being E. austroverna in Section Pyronium.An additional specim en was found filed under E. woodii Bolus in Section Chlorocodon.This led to an investigation of the two species.

Erica solandra/setulosa
In 1963 Dulfer described the variety mollis under E. solandra Andrews based on a collection from Seven Weeks Poort.We subsequently found material growing on the Rooiberg in the Little Karoo and realized that the growth form and habitat were quite unlike those of E. solandra which is known only from the Outeniqua Moun tains near George.Several other collections from the Seven Weeks Poort area matched Dulfer's material.
During an investigation of all the species in the Section Pseuderemia, it was found that a distinct new species could be described, E. ingeana E.G.H. Oliv.(Oliver & Oliver 1991) and that Dulfer's variety was not part of E. solandra and also constituted a distinct separate species.Later during our survey of all Erica species the material attributed to var.mollis was found to match a collection by Elsie Esterhuysen from the extreme western end of the Langeberg and which she had tentatively placed under E. setulosa Benth.E. setulosa is known only from Bentham's type in Kew, a specimen collected without precise locality by Niven in the 1790's and fragments of which were given to H. Bolus and are now housed in BOL.A detailed examination of these fragments showed that they match the Esterhuysen collection in m ost respects.They also indicate that Dulfer's variety is indeed E. setulosa.
The placing of E. setulosa under the Section Ephebus must have been based solely on the possession of the hairy corolla which fact overlooked the head of flowers similar to those found in Pseuderemia which also has some spe cies in which the corolla is hairy.This clearly indicates a rather tenuous distinction between the two sections.E rica setulosa Benth. in De Candolle, Prodromus 7: 682 (1839); Guthrie & Bolus: 123 (1905); Dulfer: 69 (1965).Type: Cape Colony, Niven s.n.(K, holo.; BOL, fragm.!).

Erica turmalis
Salisbury based his E. turmalis on one of the many collections supposedly made by the enigmatic I. Mulder.Salisbury is the only botanist to have cited this collector, as Jac.Mulder (.Salisbury 1796) and I. Mulder (Salisbury 1802), but nowhere is there a collection of his labelled as such.There is no Salisbury material bearing the name 'turmalis' in either K or BM.Salisbury described his spe cies as 4-5-anthered and noted that it had the facies of his E. bruniifolia.A collection by Mund in K with a frag ment in BOL, determined as this species by Bentham, is clearly part of the E. cordata Andrews complex.Until such time as an authentic specimen can be located we are Bothalia 24,1 (1994) removing this name from the list of currently accepted southern African species and placing it in the list of 'spe cies non satis cognitae'.

Erica umbelliflora
This species has remained uncollected since it was de scribed in 1839.An examination of the material cited by Bentham revealed that E. umbelliflora is, in fact, a distinct and well-known species in the southern Cape and that all the collections assignable to this species were filed under two recently described species.
The current sectional subdivision of the genus Erica is a complex and unfortunately unnatural system which has led to a number of irrelevant new species being described (see several cases cited above).The circumscription of some oi the sections has been rather vague and with the arbitrary placing of species in these sections true relation ships have inevitably been overlooked in the describing oi new taxa.This has been the case with two species, E. manifesto Compton and E. ionii H.A. Baker.

E.
umbelliflora was placed by Guthrie & Bolus in the Section Pachysa because they assumed that the flowers were viscid.Compton looked in the Section Gypsocallis tor species allied to his E. manifesta on the grounds of slightly exserted stamens and a pseudospicate inflores cence.Baker placed his E. ionii in Section Pyronium on account of the exserted stamens and terminal flowers.In all three cases the true relationship of the taxa was not possible to ascertain because of their diverse placement in the genus.
The variability of the flowers is such that they may be viscid because of the glands on the calyx, may have in cluded to exserted stamens and have the flowers arranged from terminal on short lateral branchlets to aggregated in a pseudospicate arrangement.
The collections from Natal and Swaziland tend to have the leaves more revolute and therefore narrower, whereas most of the collections from the eastern Transvaal have broad open-backed leaves.How ever, the type of E. revoluta, Wilms 90S, and Keel STE 16476 have the narrow leaves of the Natal material and both come from near Lydenburg.E rica revoluta (Bolus) L.E.Davidson in Journal ol South African Botany 51: 71 (1985).E. subverticil laris Diels ex Guthrie & Bolus var.revoluta Bolus: 224 (1905);