Siphonochilus aethiopicus (Zingiberaceae): observations on floral and reproductive biology

A taxonomic revision for the Flora of Southern Africa of Siphonochilus (Zingiberaceae) awaits publication (R.M. Smith, RBG, Edin., pers. comm. to ABC, 1987). In this revision the two southern African species, S. aethiopicus (Schweinf.) B.L. Burtt and S. natalensis (Schltr. & Schum.) Wood & Franks will be united. For the purposes of this article this conspecificity is assumed. S. aethiopicus has a distribution in Africa southwards from Senegal and Ethiopia to the Transvaal. Futher south wild populations are not now known, the species having disappeared from the natural flora of Natal. Rhizomes are extenSively used in traditional African medicine and cultural practices. Floral and reproductive biology is poorly documented, but important if cultivated stocks in southern Africa are to be maintained. Monitoring of plants under cultivation and semi-natural conditions did not support the record of polygamy within the species. One type of flower only was produced by an individual rhizome. Bisexual flowers may mature to fruits containing viable seed that will germinate in situ. Ovaries of female flowers did not develop, suggesting apomixis does not operate in seed production. The internal surfaces of the staminodial tubes of bisexual flowers bore scattered glandular trichomes; those of female flowers densely placed multicellular papillae. Bisexual and female flowers are thus strikingly different in detailed structure as well as in general morphological form.


Introduction
must be adopted as the legitimate genus for African elements of Kaempferia L. (Zingiberaceae) . Kam (1980) had recognized these African plants as forming a distinct group distinguished from the kaempferias of Asia by a combination of six well-marked characters .
For southern Africa, Gibbs-Russell et al. (1985) listed two species, S. aethiopicus (Schweinf.) B.L. Burtt and S. natalensis (Schltr. & Schum.) Wood & Franks . Morphologically these species are indistinguishable except that polygamy has been recorded for the latter (flowers bisexual and female, Wood & Franks 1911c: 115), but never for the former where flowers are bisexual only. A reVISIOn of Siphonochilus for the Flora of Southern Africa has been completed and awaits publication (R .M . Smith , RBG , Edin. , pers. comm . to ABC, 1987). In this revision the species named above will be united . For the purposes of this article, this conspecificity is assumed.
There are other species of the genus in Africa, all from further north (Burtt 1982) . These are not relevant here, except to report that for none is polygamy known .
S. aethiopicus has a distribution southwards through Africa from Senegal and Ethiopia. PI owes & Drummond (1976) record it for Mozambique and the eastern highlands of Zimbabwe; Compton (1976) for Swaziland (under Kaempferia ethelae 1.M . Wood) ; Onderstall (1978Onderstall ( , 1984 and Davidson (pers. comm . to ABC, 1987) for the Transvaal. Further south it is not now known to exist in the wild. Earlier authors recorded its presence (as S. natalensis) at Ngoye and Inanda (Wood & Franks, 1911a, c) and there have been tentative suggestions of its possible occurrence in some of the river valleys south of Durban and into the Transkei (Burtt 1982 reporting Sanderson 1868; also unsubstantiated verbal statements).
Rhizomes are important in traditional African medicine and in magi-medicinal cultural practices. Since the middle of the last century there are records of traffic and trading in the plants from sites of occurrence to areas of utilization. Also, transplantation from natural populations to the vicinities of habitations, either for cultivation or for the magical protection of the inhabitants, was commonplace (Wood & Franks 1911c;Webb & Wright 1979, 1982Burtt 1982;Cunningham 1988Cunningham , 1989. Recent findings indicate that the bulk of the material now traded originates from the Transvaal and perhaps further east (Cunningham 1988). Interference with natural populations has made southernmost distributional limits uncertain and depredation has resulted in elimination of the species from the natural flora of Natal and KwaZulu. In consequence, it is important to maintain stocks presently under cultivation and from these and from the natural populations further to the north, to attempt to derive botanical details still unknown. Particularly is information in the field of reproductive biology significant, as, because of the extensive usage of rhizomes in herbal medicine, commercial cultivation will surely be necessary to maintain supplies and to keep prices reasonable. Burtt (1982) expressed the need for a more critical investigation of the stamen less female flowers. It is also important to substantiate' .... that both the female and the perfect flowers were found on the same plant, but not at the same time .... ' (Wood & Franks 1911c: 115).

Plants monitored
Five stands of plants were kept under observation. Table  1, A-E gives details of the known history, present location and treatment under cultivation of these stands, together with individual findings derived from study over variable periods of time, of the plants they comprised. For stand E, clonal relationship of the components is very unlikely; for the remaining stands no inferences may be drawn since no information on the reproductive relationships of the plants within each stand exists.

General procedure
Monitoring of the stands took place at intervals; for stand A inspection was daily, for other stands not so frequent. Flowers and fruits were examined as development took place.
Formal descriptions based on information derived from these examinations are given at the end of the paper, but these cover only features requiring modification or elaboration or that are previously undescribed. For rhizomes bearing bisexual flowers, details of the reproductive cycle and the duration of its phases are S.-Afr. Tydskr. Plantk., 1989, 55(3) recorded. Insufficient material was available to make this worthwhile for stands C and D.

Results and Discussion
Floral structure Bisexual flowers The structure of these flowers has been well described and illustrated (Wood & Franks 1911a, b;Kam 1980;Burtt 1982). Their most conspicuous feature is formed by the petaloid staminodial whorls situated above the thick-walled corolla tube and lining this. The lower half of these whorls forms a split, rounded tube lined by downwardly directed, stalked glandular trichomes that decrease in frequency into the corolla tube (see Figure  1). The upper half expands into the two lateral 'petals' and bilobed 'labellum'. There is also the posteriorly placed stamen.
Some flowers from rhizomes cultivated in Edinburgh developed an additional small lobe between the two main lobes of the 'labellum' (Burtt 1982, Fig. 1G). This additional lobe was not represented in any of the flowers checked by us (stand A, 35 taken at random in 1986, all 61 produced the following year; stand B, 2 taken at random 1986, repeated 1987).
All flowers dissected contained a functional anther, sessile on the broad, petaloid filament, the connective of which projects from the staminodial tube and overarches the anther and stigma. The positional co-ordination of these last-named organs, ensured by enclosure of the upper part of the style in a centrally placed narrow channel in the broad filament, has not been adequately stressed before.
Early descriptions (Wood & Franks 1911a, b) do not mention the two oval, white, densely textured glands compactly set in sterile tissue towards the apex of the ovary. These epigynous glands are illustrated by Kam (1980, Fig. 3E) and by Burtt (1982, Fig. 11, I). We found them always present.

Female flowers
The female flower was illustrated and described by Wood & Franks (1911a, b), but its structural features are less known than those of the bisexual type. We have dissected only six, from different rhizomes, grown under pot culture. It is probable the full range of variability of these flowers is not yet known. All we observed differed markedly in shape from the bisexual type and could be recognized as different on superficial observation, even at a distance. It is notable that flowers from different rhizomes may differ quite markedly in size. There was no difference in construction, however. The unsplit staminodial tube is the most conspicuous feature. Wood & Franks (1911c, p. 115) described it as 'closed almost to the apex, the aperture (was) very narrow ... '. We did not find the tube or the mouth appearing so narrow. The rough papillate lining is worthy of note. The papillae sometimes run in vertical lines; their multicellular construction is quite distinct from the stalked, glandular trichomes of the bisexual staminodial tube (see Figure  1). Wood & Franks (I.c.) described this lining as ' ... a thickened or somewhat furry substance ... '. They also (Wood & Franks 1911a)

Reproductive cycle
The complete reproductive cycle is known only for rhizomes that produce bisexual flowers, but details of the vegetative organs, flowering and the growth cycle apply also to rhizomes that produce female · flowers.

Vegetative organs and growth cycle
Rhizomes grow either on the soil surface, or underground to a depth of 150 mm. Widely spreading lateral roots with marked tuberous swellings are developed. During the unfavourable season stands are not obvious above ground, but with the advent of the spring rains aerial shoots develop, shortly before, or contemporary with, buds. The aerial shoots continue to elongate until after flowering is complete, but after the December solstice growth gradually ceases. Yellowing commences quite suddenly, usually during April, and by May the aerial shoots have collapsed . Pla nts are reported as being remarkably pest free.

Flowering
One of the six important differences by which Kam (1980) differentiated Siphonochilus from Kaempferia was that in the former, the inflorescence axis develops separately from the leafy shoot. This was apparent at flowering, for buds emerged directly from the ground 10-20 mm from aerial shoots. Excavation exposed the rhizome to show the abbreviated , racemose , 4-5flowered inflorescence axis , developed from tissue underlying , but terminal in relation to , that which had produced the nearest vegetative shoot. Buds were visible above ground usually 5-8 days before opening. Expansion into open flowers takes place during the night. Each flower is reported to last only one day (Onderstall 1978;Davidson, pers . comm. to ABC, 1987) but occasional flowers may last till a second morning.
Flowering commences mid to late October , and continues to the end of December. It is unusual for flowers to be developed at other times, but transplanting of rhizomes with buds already set during a previous resting season causes immediate development of these buds .

Pollination
Opinions differ concerning the odour of the flowers , from unperfumed (one person only) to ' ... a haunting, spicy fragrance reminiscent of ginger' (Onderstall 1978). Consensus favours a delicate scent particularly evident in the late evening and morning.
The only insects noted in constant association with the flowers were black dipterids (±2 mm in length) resting (?) in large numbers on the peta loid staminodial lobes .
Presumably this is the 'Mucid Fly' seen by Wood & Franks (1911a) and the insect said to enter the throat of the flower by a Transvaal observer (Davidson , pers . comm. to ABC, 1987). Dissections of deteriorating flowers showed no dead dipterids within staminodial or corolla tubes, but minute translucent to milky-white larvae with black points were frequently present in the stylar channel near the anther. It is not known whether these are a stage in the dipterid life cycle. The downward-pointing glandular hairs of the staminodial tube seem likely to prevent insect escape; nevertheless occasional black ants are reported to creep in and out of flowers grown in Durban .
Unless apomixis operates to produce viable seed, pollination must take place . Self-pollination seems likely since all the fading flowers dissected had pollen in contact with the receptive stigmatic surface . This transference possibly takes place at time of collapse of the staminodial lobes.
Capsule and seed development Ovary enlargement and capsule maturation is rapid . Seed is mature 5-8 weeks after flowering. If the capsules are underground , the soft parts decay leaving a fibrous skeleton which eventually disintegrates; if 285 exposed they dry and dehisce. Some rhizomes flower profusely yet mature few, if any , fruits. The seeds remain viable for at least 1 year.

Seed germination and seedling survival
Seeds germinated in situ. The seedlings penetrated through the decaying fruit walls. Growth was relatively slow and damping-off took place. When individually potted and given reasonable care, seedlings grew rapidly and in the second year of growth were hardly distinguishable, except for rhizome development, from mature plants . Seedling structure is shown in Figure 2. From an early stage the tuberous root(s) are well developed.

Conclusions
Monitoring studies have shown: 1. that rhizomes produce either bisexual or female flowers . No evidence was obtained that suggests both flower types may develop from an individual rhizome. 2. that bisexual flowers set mature fruit and viable seed which germinates either in situ, or when planted under cultivation . The reproductive cycle of rhizomes producing bisexual flowers is thus capable of completion in nature. 3. that female flowers lack a functional male organ and are distinctive in form from the bisexual flowers.
There is size variability in the female flowers produced from different rhizomes. 4. that female flowers do not produce mature fruits.
There was no enlargement of the ovary, which is suggestive that apomixis does not operate in seed production in either the female or the bisexual flowers.
It is important that study of the rhizomes producing female flowers be continued. Few rhizomes were available and these have been monitored for only 4 years . It is also essential that wild populations in the eastern Transvaal be carefully checked for the presence of stamenless flowers. Seed should be studied during development for the possible presence of an aril. Such an outgrowth was not observed in mature seeds.
Eventually karotypical studies should be undertaken. These may reveal information on polygamy within the family .

Bisexual flower and fruit (information derived from stands A & B)
Anther 12-14 mm long, slightly curved, clearly bilobed, dehiscence longitudinal, releasing sticky pollen, grains large , accrescent into clumps. Style ± 65 mm long, free below, eventually passing into a narrow central channel of the filament where it is securely held behind the anther lobes. Stigma immediately above and overarching the anther, the receptive surface directed upwards, hypocrateriform (salver-shaped) in face view with an asymmetric, laterally elongate cavity leading back into the hollow style, obsoletely 3-lobed, the lobes variable in development , never strongly marked, infundibuliform  (funnel shaped) in side view. Ovary subterranean to slightly exposed, a cherry-sized, bright maroon capsule at maturity that , if underground, decays and disintegrates; if exposed dries, darkens and dehisces. Up to 15 fruits produced by 1 rhizome, usually 2-4 only. Seeds up to 30 per capsule, 2-3 mm in diameter, pale cream to brownish; viable for ±1 year: germination period ±2 months.