Studies in the Ericoideae ( Ericaceae ) . XV . The generic relationship between Erica and

The small African genus Ericinella Klotzsch, with one species in Malawi and SW Tanzania and three in the Eastern Cape is shown to exhibit variability in its delimiting characters that overlap with those of Blaeria L. and Philippia Klotzsch, both of which have recently been included under Erica . It is postulated that the genus, like the two above, is unnatural and polyphyletic. The genus is therefore included under Erica and the relevant nomenclatural changes are provided: Erica amatolensis E.G.H. Oliv., nom. nov. (= Ericinella multiflora Klotzsch), E. passerinoides (Bolus) E.G.H. Oliv., comb, nov., E. hillburttii (E.G.H. Oliv.) E.G.H. Oliv., comb, nov., and E. microdonta (C.H. Wright) E.G.H. Oliv., comb. nov.


INTRODUCTION
The genus Ericinella Klotzsch belongs to the capsular group of African genera of the subfamily Ericoideae which until recently comprised Erica L ., Philippia Klotzsch, Blaeria L. and Ericinella.The genus Philippia has been found (Oliver 1988) to possess a continuous range of variation in the sole distinguishing character be tween the genera, the fully recaulescent bract in some species, albeit only a few compared to Lie 757 species in Erica.Also it was argued that the genus was not a monophyletic unit.The genus was therefore reduced to synonymy under Erica (Oliver 1987(Oliver , 1992(Oliver , 1993a;;Beentje 1990).As a result, the circumscription of Erica became broader with the inclusion of those Philippia spp. with stamen complements differing from the basic eight stamens in most Erica spp.The most aberrant species of this group is the tropical Erica nyassana (Aim & T.C.E.The situation regarding the validity of Blaeria as a dis tinct genus was also examined in the light of its single character difference from Erica of four versus eight stamens.It has been shown (Oliver 1993b) that several species of Erica, apart from those incorporated from Philippia, showed an overlap in the number of stamens.As in the case of Philippia it was shown that polyphyletic origins exist for the various species groups currently in cluded within the genus Blaeria.As a result, the genus Blaeria was reduced to synonymy under Erica (Oliver 1993b(Oliver . 1993c).

ERICINELLA
This genus was described together with many other genera by Klotzsch (1838) in his reclassification of the subfamily Ericoideae.It was based on one species, E. mul tiflora Klotzsch, and was distinguished from other mem bers of the subfamily by a fully recaulescent bract, the zygomorphic calyx condition, and no bracteoles coupled with only four stamens and a 3-locular ovary.The sub sequent describing of Ericinella gracilis Benth.from Madagascar (Bentham 1839) altered the circumscription of the genus to include a 4-locular ovary.This was com pounded by the adding of E. mannii Hook.f. from tropical Africa in 1862.
This broader circumscription of the genus was main tained by various workers in their publications for overall treatments of the family or for large floras (Bentham 1839;D. Oliver 1877;Drude 1897;Brown 1905).
When Aim & Fries (1927a) undertook the first com plete revision of the genus, they retained only three species, the two Cape ones, E. multiflora and E. pas serinoides, and one from tropical Africa, E. microdonta (C.H. Wright) Aim & T.C.E.Fr., which had originally been described as a species of Blaeria.They purified the genus by removing what to them were the two discordant elements to Philippia, P. tenuissima Klotzsch (E.gracilis) and P. mannii (Hook, f.) Aim & T.C.E.Fr.. Bothalia 24,2(1994) Phillips (1926) retained the genus as circumscribed in Flora capensis (Brown 1905) in his The genera o f South African flowering plants, but later (Phillips 1944) com pletely changed his concept of the subfamily.In this work he combined Ericinella with a number of genera not all closely related, namely Philippia, Coccosperma Klotzsch and Thamnus Klotzsch, under Blaeria L. He retained this treatment in the second edition of The genera ... (Phillips 1951).The genus was retained in the same form as used by Aim & Fries (Oliver 1975) and by Ross (1983) in Flora zambesiaca with the single species, E. microdonta.
As currently construed Ericinella is a small genus with the one species, E. microdonta, in tropical Africa and the three species, E. multiflora, E. passerinoides and E. hillburttii, in the Eastern Cape.It is distinguished only by a combination of two characters, the fully recaulescent bract and four stamens.
The fully recaulescent bract is present in a number of Cape genera such as Salaxis Salisb., Coccosperma Klotzsch, Scyphogyne Decne and Nagelocarpus Bullock, but these minor genera have an indehiscent fruit.It is also present in the philippioid components now included within Erica.Aim & Fries (1924) and Ross (1957) relied on anther appendages and the shape of the stigma to distinguish Ericinella from Philippia.These characters are clearly not of any significance in generic delimitation because they are only a reflection of the different pollination syndromes present in the taxa-entomophily versus anemophily, both of which are present in many species of Erica.It has been pointed out (Rebelo et al. 1985;Koutnik 1987;Oliver 1991) that the reduction in flower size, loss of bright colours and the increase in size of the stigma is associated with anemophily in the southern African Ericaceae.To this must be added the loss of anther appendages.
The only difference existing between Ericinella and Blaeria in tropical Africa is the fully recaulescent bract and lack of bracteoles in the former genus.However, Ross (1980) stated that the bract can be inserted anywhere from near the base of the pedicel to close under the calyx in the tropical species of Blaeria [Erica] and that bracteoles are only present when the bract is inserted at or below the middle of the pedicel.This would then give the con dition of a partially recaulescent bract and no bracteoles which does occur in some specimens of E. microdonta that I have examined.

DISCUSSION
In the discussion of the African capsular genera (Oliver 1988) it was stated that the relationships within this group of genera are very close, with generic distinctions being based on the flimsiest of morphological characters.This would indicate that the genera have arisen from some an cestral ericoid stock by two processes: the shifting of the bract and bracteoles from axial and partially recaulescent to totally recaulescent; and by the reduction in stamen number from 8 to 4. Both processes were involved in the formation of the species placed under Ericinella, whereas in Philippia, the first and in Blaeria the second process has taken place.Ross (1957) did not consider the possibility of any close link existing between Philippia and Ericinella, in stead he regarded them as belonging to two separate evolutionary lines in which the zygomorphic calyx, produced by the totally recaulescent bract, has risen in dependently.
Aim & Fries (1924) when considering the phylogenetic origins of the genera Philippia and Ericinella, were unable to give any clear lineage for Ericinella from either Philip pia or Blaeria.They regarded E. microdonta as an outlier of the genus whose origin lay in the region of the Cape Flora.

These arguments are perhaps feasible if one accepts
Ericinella as a natural, well-defined genus.The tropical species shows a clear relationship with the tropical species of Erica that have been placed in the Section Arsace (Aim & T.C.E.Fr. 1927b), Erica arborea L. and the complex of six species included by Ross (1956) under E. kingaensis Engl, as three subspecies.These ericas exhibit reductions in their bract/calyx arrangement and in the number of stamens.They all clearly arose from some ericoid an cestral stock independently of those much further south in the Cape Province.
There is little doubt that the three Cape species are closely related, but not to the tropical species, E. microdonta.The Cape species are allied to several Eastern Cape-KwaZulu/Natal species of Erica which are placed in the section Arsace in the present, rather unsatisfactory, subgeneric classification of the genus.These are Erica dominans Killick, E. dissimulans Hilliard & B.L. Burtt and E. anomala Hilliard & B.L. Burtt.E. dissimulans and E. anomala both show a strong tendency towards the philippioid reduced calyx.The former species also pos sesses 3^4-locular ovaries as does E. dominans, characters which have not previously been recorded in the species.There is also a relationship with the southeastern Cape species complex based on Erica simulans Duller, which is placed in the section Pyronium.In my opinion the genus Ericinella, as it stands, is an unnatural one which is clearly not monophyletic.
The clear and total overlap in the morphological char acters of partial versus total recaulescence of the bract, the four stamens and 3-to 4-locular ovary indicates that there can be no separation between the species currently included under Ericinella and those which were formerly in Philippia and Blaeria and are now included within Erica.This fact coupled with the independent origin of the two groups of species have led me to reduce the genus to synonymy under Erica.Erica amatolensis is most closely related to the more recently described E. passerinoides and E. hillburttii under which the differences are discussed.These differences are summarised in Table 1.
This species is restricted to the higher mountains of the Eastern Cape Province from the Katberg along the Amatola Range to Stutterheim (Figure 1).It forms erect single-stemmed shrubs up to 1.5 m tall on the edges of forest patches or among wtxxiy vegetation on rocky gras sy slopes from 1 (MK)-l 600 m altitude and flowers from October to December.
It is the most widespread of the three Cape species discussed here, but does not appear to be abundant where it does occur.For instance on the Katberg Pass, where most records have been made due to easy access, the plants are few and far between.Over its range there does not appear to be any variation of significance in the char- Testa cells elongate with convoluted edges broader with less convoluted edges ?
McMasters (pers.comm.)noted that the bushes in the Gubu Dam area near Stutterheim were inclined to have a running root system from which new plants were produced.This could well indicate that these plants may survive burning by sprouting from the root system.This would then relate to the condition found in E. hillburttii with its thick woody rootstock and many stems above ground.All the plants that I have seen in the wild pos sessed a single stem with regeneration having to take place via seeds only.
The flowers have very rudimentary nectaries at the base of the ovary and only a slightly obconic stigma.Anemophily was clearly evident in the populations on the Katberg Pass where clouds of pollen were produced when the plants were disturbed.It is surprising therefore that the flowers are a striking pink colour and that the exserted anthers often have appendages, features which are as sociated with entomophily.Insects in the form of bees foraging for pollen only, could perhaps play a minor role in the pollination of this species.

Specimens examined
The true identity of E. passerinoides remains a problem as the type and only collection does not have fully mature flowers.Thus the true shape of the corolla, the position of the anthers at anthesis and the position of the mature style and stigma cannot be assessed.Fortunately an old capsule remaining from the previous flowering season, al though somewhat disintegrated, could be used to reconstruct the characters of the fruit.
Bolus's diaries for that period do not give sufficient details to pinpoint the exact locality.He appears to have collected en route from Graaff-Reinet to Murraysburg, presumably not far off the road.Several searches in the area of the highest point of the road over the Sneeuberg and environs to the north and south have proved unsuc cessful.Another species of Ericaceae.Philippia tristis Bolus (= Erica caespitosa Hilliard & B.L. Burtt), also known from its type collection gathered at the same locality on the same day, has not been re-collected in that vicinity either.As farms now stretch right to the summit of these mountains where burning and overgrazing are very evident, it is likely that both species have been ex terminated in the area.
Aim & Fries (1927a) retained the species with some hesitation as they felt that it was only a young stage of E. amatolensis (E.multiflora).However, a thorough ex amination of the type has revealed a number of characters which point to the distinctness of E. passerinoides from E. amatolensis (see Table 1).Brown (1905) used the corolla shape and degree of exsertion of the style to separate the two Cape species.Aim & Fries on the other hand used the shape of the calyx lobes and leaf length to separate them.These characters are of no real value in distinguishing the species.
Until the material is found again in the wild I am retaining E. passerinoides as a distinct species.
This species forms erect multi-stemmed shrubs up to 1.5 m tall and occurs on rocky, grassy slopes at high al titude in the mountains northwest of Elliott in the Eastern Cape.It flowers from October to December.

E.
hillburttii was discovered as recently as 1983 by Hilliard & Burtt.It is known to date from only one locality, the Bastervoetpad.Accessibility in these moun tains is very limited, therefore additional populations could well be discovered in the future farther north towards Naude's Nek.
The species is easily distinguishable from its nearest ally, E. amatolensis, on a number of characters (Table 1).The most noticeable of these characters in the field are the multi-stemmed habit, greenish yellow colour of the flowers and the included stamens.Examination of a flower will show the relatively short filaments with broad apex on which the awns are decurrent and the lanate ovary.
The total lack of any nectaries below the ovary, the broadly funnel-shaped stigma and the inconspicuous dullcoloured flowers clearly point to anemophily in this species.This syndrome could not be tested in the field because of the wet overcast weather prevailing when the type material was collected.This species is from tropical Africa where it occurs mostly on high ground in Malawi, just getting into SW Tanzania (Figure 2).Ross (1983) recorded the species as forming shrubs up to 3 m tall.Brass, on his various col lections, records the plants as occasional to plentiful, sometimes gregarious, in grassland on the edges of rain forest and on rocky banks of streams subject to flooding.One plant he recorded as a tree 4 m tall. 1 can confirm these observations from my examination of plants on Mt Mulanje.

Specimens examined
Of the species listed here E. microdonta is by far the most variable.The populations in central Africa were regarded as comprising two distinct species, one with a variety (Brenan 1954).The variation in the size and form of the corolla is considerable, the smallest flowers being from the type of E. brassii Brenan and appearing very distinct.However, Ross (1983) after examining a consid erable number of specimens from the whole area regarded the variation in the indumentum on the branchlets, leaf indumentum, leaf size, corolla size and stigma diameter as of no taxonomic significance.He recognized only one species with no varieties, and this view I support.

E.
microdonta possesses well-developed nectaries below the ovary, a poorly developed funnel-shaped stigma and partially exserted anthers with appendages.This would clearly indicate entomophily as the pollination syndrome.During a visit to a few populations on Mt Mulanje in Malawi during 1991, I was unable to record any instance of the mass scattering of pollen from the white flowers, but noted no visitation by any insect.
In correspondence some time ago on the lectotypification of this species, which had not been done for Flora zjambesiaca, Ross recommended that McClounie 55 was the best specimen to select.I therefore attribute the above selection to him.