New species of Erica ( Ericaceae ) from the Cape Province

Three new species of Erica L. (Ericaceae) are described from the southern parts of the Cape Province. E. amicorum E.G.H. Oliver, an endemic marsh species from the Langeberg near Riversdale, is one of the most distinctive species in the genus on account of its almost free reflexed petals, reflexed stamens and exserted gynoecium. The mat-forming E. tegetiformis E.G.H Oliver is allied to E. senilis Klotzsch ex Benth. and E. eriophoros Guth. & Bol., all occurring at high altitudes in the Cold Bokkeveld and Cedarberg. E. cunoniensis E.G.H. Oliver is allied to the E. squarrosa Salisb./£. gysbertii Guth. & Bol. complex and is very localized in the mountains of the western part of the Caledon Division near Rooi Els.


INTRODUCTION
The genus Erica in southern Africa has recently been increased in size with the inclusion of the 15 species of Philippia Klotzsch (Oliver 1987(Oliver , 1988) and now contains some 654 species.As such it is by far the largest genus in the flora of southern Africa, indeed the continent of Africa.The considerable increase in collections of plants from the region of the Cape Flora during the last three to four decades has produced a wealth of new and interesting records.These include a number of distinct undescribed species of which three are described here.

TYPE
The possession of almost free corolla lobes is very rare in the subfamily Ericoideae.It occurs in a few schizopetalous cultivars of European heathers.In southern Africa only four species share such a divided corolla, i.e.Erica lanuginosa Andr., E. occulta E.G.H. Oliver and E. bodkinii Guth.& Bol., but not to the same extent as in E. amicorum.Also they do not have the lobes reflexed in the mature flowers.This condition renders the flowers unique in that the complete androecium and gynoecium are exserted, thus giving the impression of an erica flower in which the corolla has been eaten away by some insect.The mature flower is also unique in having those stamens alternating with the corolla lobes totally reflexed between the lobes.The stamens are also rather unusual in being joined together at the base into a thickened collar.Fused or partially fused stamens are now no longer rare in the genus Erica because of the recent inclusion of the genus Philippia into Erica (Oliver 1987(Oliver , 1988)).
Another striking feature of E. amicorum is the arrange ment of the flowers which can be borne in a variety of ways on the branches from simple 1-3-nate terminal florescences to 1-flowered florescences on lateral absolute brachyblasts (Figure 2).These may then all be aggregated together into a loose synflorescence on a lateral branch (Figure 3).When the main branches are long, up to 1 m, they bear numerous lateral branches which themselves bear various combinations of florescences.In some cases the simple 3-flowered terminal florescence can look very similar to a terminal synflorescence.Examination of a florescence will show that flowers alternating with the leaves belong to a simple florescence (Figure 2A) whereas those opposite the leaves are terminal on absolute brachyblasts (Figure 2B).

E. cunoniensis was first collected by Thomas Stokoe in
1936.I first saw the species in 1966 as several scattered shrubs on the dry northern slopes of Buffelstalberg.The mountain was visited again in 1969 when a single plant, just past flowering, was all that could be found.At the time it was thought that the plant may have been a hybrid between E. gysbertii and E. massonii L.f., both common on these slopes but only flowering later in summer.Natural hybrids between very different-looking putative parents have previously been recorded in Erica (Oliver 1977(Oliver , 1986)).Lower down the slopes on an earlier occasion I had noted a plant of clearcut hybrid origin between E. mas sonii and E. cygnea.
A subsequent search failed to locate any plants of the new species until a concerted effort was made together with members of the A Team of the Botanical Society of South Africa.The plants were eventually found on the southern slopes, thus indicating that the original plants had been chance strays.On the southern side of Buffelstalberg the plants were growing in a single population on a cool moist loamy slope with a south-western aspect.Subsequent examination of the pollen showed it to be normal and well developed.
FIGURE 2.-Two similar-looking inflorescences commonly found in £ amicorum: A, a simple 3-flowered florescence condensed from the schematic diagram Al; B, a compound synflorescence consisting o f three 1-flowered florescences (in each florescence there are two aborted flowers) condensed from the schematic diagram in Bl.Zigzag lines represent expanded absolute brachyblasts.
Museum and Kew, which could form part of the type collection by Masson, is similar to that of E. gysbertii, but unlike several collections which have been identified as E. squarrosa in Cape herbaria.I refer to this latter material as Species A, which has been collected from the Hottentots Holland Mountains between Landrostkop and the Triplets.It has been recorded as a low compact shrublet with flowers 'bril liant shiny crimson, dark at the throat'.In the flower colour Species A is similar to E. cunoniensis but it has globoseurceolate corollas with a very narrow mouth and the anthers are glabrous, free and situated at the mouth of the corolla.This, coupled with the open nature of the flowers and semiconnate anthers, would indicate a totally different pollinator for E. cunoniensis.However no pollinators were observed by Rebelo and myself on Buffelstalberg.The pollinator for E. gysbertii and E. squarrosa is a Guth.& Bol. in Flora capensis 4: 90 (1905).type: Cape, Stellenbosch Div., on the western foot of the Hottentots Holland Mountains, 200 ft., Guthrie 3654 (BOL!).