Androecium and floral nectaries ofHarungana madagascariensis (Clusiaceae)

s~lsant

tematics and Evolution

P1. Syst. Evol. 178:179-194 (1991)

© Springer-Verlag 1991 Printed in Austria

Androecium and floral nectaries of

Harungana

madagascariensis ( Clusiaceae) L. P. RoNsE DECRAENE and E. SMETS Received March 25, 1991; in revised version August 23, 1991

Key words: Angiosperms, Clusiaceae, Harungana madagascariensis. - Androecium, stamenpetal complex, floral nectaries, floral development, floral anatomy. Abstract: The mature flower of Harungana madagascariensis (CI-IOm¥) POIR. has an androecium of five antipetalous fascicles, consisting of four stamens each. The stamen fascicles alternate with five indented nectary scales. A SEM-study of the floral development, as well as a study of the floral anatomy was carried out to understand whether the nectariferous scales represent staminodia or are receptacular in nature and consequently whether or not the androecium of Harungana, and the Clusiaceae in general, is originally diplostemonous. The five petals originate by the splitting of petal-stamen complexes. Next the upper part of each complex differentiates basipetally in four stamens. The stamens remain fascicled and are lifted on a long stalk at maturity. Five carpel primordia are initiated united in a low ringwall. The five nectary scales appear after carpel inception and develop an external morphology reminiscent of anthers. The floral anatomy reveals an independent origin of sepal median traces and common sepal lateral traces, free petal traces, stamen fascicle traces and alternating vascular tissue which supplies the nectaries. The petal-stamen complexes are the result of a retardation in petal inception, linked with the absorption of petal tissue into the stamen primordia. The development of the stamen fascicles is discussed; it is suggested that they are of a secondary nature and do not appear as a reduction from a multistaminate androecium. The external morphology and vascular anatomy of the scales speaks in favour of a staminodial nature. The comparison with some other species of the Clusiaceae gives evidence of a diplostemonous ancestry of the androecium.

Harungana madagascariensis (CI-IOISy) Poem is a typical member o f the family Clusiaceae or Guttiferae, which comprises about 50 genera and 1200 species (CRoNQUIST 1981). The study of the floral development can provide interesting data about trends occurring in the family. Some species of Clusiaceae (incl. Hyperiaceae) have been studied ontogenetically in the past: e.g., Hypericum calycinum L., H. elodes L., H. perforatum L., and H. aegypticum L. by HW,MER (1918), Hypericum aegypticum, and H. hookerianum WIGHT & ARN. by LHNs (1964a), H. perforatum by SAan'LER (1973), and H. androsaemum L., H. perforatum, and H. calycinum by PAYER (1857). Other species have been neglected to a large extent. Only observations of MoNcuR (1988) on Rheedia madruno (H.B.K.) PLANCn. & TPdANA and Garcinia spec. can be mentioned.

180

L.P. RONSEDECRAENE& E. SMETS:

The development of the androecium is a feature of much interest and can have a wide application in taxonomy (RONSE DECRAENE & SMETS 1987; RONSE DECRAENE 1988, 1989, 1990). The position of Clusiaceae in Theales (R. DAHLGREN 1980, 1983; G. DAHLGREN 1989; CRONQUIST 1981; THORNE 1983) has been widely accepted by systematists on the basis of several characters, such as the centrifugal androecium, bitegmic ovules, axile placentation, etc. The centrifugally developing stamens is herein one of the most important features in determining the position of the Clusiaceae. The androecium of Clusiaceae is highly variable and has been interpreted by several authors following different floral theories. According to the telome theory (WILSON 1937, 1942; KAWANO 1965; VAN HEEL 1966; STEBBINS 1974), the development of antipetalous stamen fascicles is an important step in a reductive evolution from telomes. These authors interpret the primitive androecium of Clusiaceae as made up of antipetalous fascicles, each containing numerous stamens, which become progressively reduced. Examples are known where the external stamens of each fascicle are represented by staminodes. So, the androecia with the lowest number of stamens are necessarily derived. Another interpretation has been presented by PAYER (1857), SACHS (1874), HOFMEISTER (1868), EICHLER (1878) and ENGLER (1925) and later by LEINS (1964a, b, 1971, 1975) and RONSE DECRAENE& SMETS (1987), i.e. that the androecium has been derived from an oligomerous structure (haplostemonous or diplostemonous). Examples are known of species bearing ten filaments fused in a short column (e.g., Renggeria MEISSN., Quapoia AUBL.: ENGLER 1925). Nouhuysia LAUTERB. is tetramerous and has eight free stamens; it is not possible to decide from the description in ENGLER (1925) if the stamens belong to two distinct whorls or not. The androecium is thus often highly polyandrous. However, the multitude of stamens can arise by different patterns of inception which are in most cases strongly related to the initiation and further development of the petals. Sometimes the petals arise as free primordia before the stamens appear (SATXLER 1973, LEINS 1964 a, PAYER 1857, SACI-IS 1874). On the other hand, LEINS (1964 a) describes a stamenpetal complex for Hypericum aegypticum, which is formed after the independent origin of the petal and stamen primordia. The more usual development of numerous stamens in the family starts from large complex alternisepalous primordia. First, petal primordia become differentiated by a transversal slit on each stamen-petal complex (HIRMER1918, HOFMEISTER 1868, RONSE DECRAENE 1989); petal development is often slow if not fully retarded. The upper part of the complex primordium is large and develops further independently of the petal. Stamens usually remain confined to strictly delimited fascicles. In other cases no such distinct antipetalous primordia are present, but a whole ringwall (circular primordium) is said to develop (e.g., in Hypericum L. sect. Brathys, Garcinia L.: PAYER 1857, HIRMER 1918). Stamens are initiated centrifugally on the flanks of the ringwall. The formation of a ringwall is usually associated with a trend of complete dioecy in Garcinia L. (BAILLON1873). PAYER (1857) places a link between a ringwall and the inception of five independent complex stamen primordia. He observed in Hypericum sect. Brathys that five primordia become united into a large circular primordium. Garcinia anomala PLANCI-t. & TRIANA also has a continuous ringwall; G. rnultifloraCHAMP. on the other hand has four large antipetalous fascicles

Androecium and nectaries of Harungana

181

Fig. 1. Floral diagram of Harungana madagascariensis

(BAILLON 1873). Clusia ovigeraPLANCH.& TRIANA has four centrifugally developing stamen sectors on a large ringwall (pers. obs.). Harungana madagascariensis has been described as having an androecium of antipetalous fascicles alternating with sterile appendages (BAILLON 1875, Fig. 1).. The nature of these appendages is uncertain; they could represent staminodes of a second reduced stamen whorl (in the opinion of EICHLER and ENGLER), or just be receptacular outgrowths. LEINS (1964 a) similarly described nectaries in Hypericum aegypticum, but stated that these were receptacular emergences (,,sekund/ire Bildungen des Blfitenbodens"). VistulaVAND. and Psorospermum SPACH. bear also five antipetalous stamen bundles alternating with five scales (BAILLON1875) and most closely resemble Harungana. The present study was undertaken to acquire more knowledge about the nature of the androecium and the floral nectaries in the Clusiaceae: - Whether the idea of an ancestral diplostemony is founded; - whether the second whorl of appendages is staminodial or receptacular in nature. LEINS (1964 a) already urged to study the floral development of Vistula in order to understand the real nature of the appendages alternating with the stamen fascicles. The study of a species which closely resembles Vismia in its floral morphology can help to unravel this question. Material and methods

Floral buds of Harungana madagascariensis were collected in the tropical greenhouses at the National Botanical Gardens of Belgium in Meise. Buds were immediately fixed in FAA and later dissected with a Wild M 3 dissecting microscope. Preparation for SEM followed the methods described in RONSE DECRAENE(1989). Buds were critical-point dried using a CPD 030 (Balzers), sputtered with a gold-layer and observed with a Philips 501.B SEM in Meise. Photographs were taken on Agfa-pan 25 using the oscillophot M 20 of Steinheil. Some material was also prepared for light microscopy. Buds were run through an alcohol and alcohol-tertiary butylalcohol series, paraffin-embedded and sectioned 11 ~tm thick. Next they were stained with safranin and fastgreen and observed under a Leitz Dialux 20 microscope. Photographs were taken with a Wild MPS45/51 Photoautomat. Voucher material (RoNsEDECRAENE409 Lm) is kept at the Botanical Institute of Leuven. Results Floral development. Flowers of different age are grouped in small, tightly clustered heads. These grow later into terminal compound cymes. Five sepals are incepted

182

L.P. RONSEDECRAENE& E. SMETS:

in a rapid 2/5 sequence and do not differ much in size (Fig. 2 a). They grow rapidly and enclose the bud completely. At maturity they are visible as small valvate lobes, which are persistent in fruit (Fig. 5 c, d). The receptacle remaining after sepal initiation takes the shape of a flattened star (Fig. 2 b, c). The angles between the sepal interstices swell up into large protuberances which become differentiated as common stamen-petal primordia (Fig. 2 b, c). Petal primordia are inserted between the free sepal appendages after separation from the upper half of the complex primordium by a median slit (Fig. 2 c: arrows). The petal primordium grows rapidly into a crescent-shaped primordium, but its development is soon strongly retarded (Figs. 2 d-3 c). Petals become progressively larger (Fig. 3 f) and take over the protective function of the sepals before anthesis; at maturity they are reflexed and bear long trichomes on the upper surface (Fig. 5 c, d). Aestivation is imbricate (Fig. 1). The remaining upper half of the complex primordium is slightly larger than the petal primordium and grows into a hemispherical structure (Fig. 2 d-f). Development of tissue in the upper primordium does not extend further than the lateral limits of the petals. This upper half of the complex primordium is also of a complex nature and splits into three stamen primordia (Fig. 3 a--c). The central stamen is situated slightly higher than the laterals and appears before them. Finally another stamen primordium is initiated centrifugally opposite and below the central stamen (Fig. 3 d, e). These four stamens make up a stamen fascicle. The primordia elongate a little and become broader at the top. Two thecae are differentiated as two globular structures (Fig. 3 e, f). Development of the anthers is successive and is connected with the difference in age of the stamen primordia (Fig. 3 e, f). The four primordia of each fascicle develop slender filaments, but they are elevated by a common stalk (Fig. 4 d, f). Soon the long ribbon-like stalk, carrying the four stamens on top, is growing high above the gynoecium (Figs. 4 f and 5 c, d). At maturity the stalks are reflexed against the petals and prevent the anthers from shedding pollen on the styles (Fig. 5 c, d). At the time of differentiation of the three upper stamens on the androecial complex primordia, the gynoecium is incepted as five congenitally fused carpels (Fig. 3 b, c). This is visible as a central star-like depression with five arms, surrounded by a low crenellated border. The fused carpel wall grows upwards and five septa extend towards the centre of the apex in an irregular manner (Figs. 3 e and 4 a, c). Their margins touch each other and enclose the central area (Fig. 4 c). Each carpel becomes enclosed by the localized growth of tissue alternating with the septa and produces a slender style. Adaxially long grooves are visible. The upper part of each style becomes globose and differentiates into stigmatic tissue (Figs. 4 f and 5 a). At maturity the stigmas are reflexed outwards, lying in the interstices of the androecium (Fig. 5 c, d). The placentae produce two ovules each, one on each side of the septa. Ovules are anatropous and ascendant. Two integuments develop around the nucellus (Fig. 5 b). When carpel margins start to meet on their ventral side, slight protuberances are seen between the stamen fascicles well below the base of the gynoecium (Figs. 3 e and 4 a: arrow, b). The stamen fascicles are already in a state of anther differentiation when these protuberances develop into two cylindrical lobes reminiscent of thecae (Figs. 4 d-f and 5 c). At maturity the two lobes are slightly bent outward and fill all space between the stamen fascicles (Figs. 4 f and 5 c, d). However, occasionally

Androecium and nectaries of Harungana

183

Fig. 2. Floral development of Harungana madagascariensis, a Successive inception of sepals in a 2/5 sequence (numbers indicate order of appearance), b The sepals (S) are differentiated and surround a five-angled apex on which stamen-petal complexes (A-P) become differentiated; four sepals removed, c Differentiation of a stamen primordium and a petal primordium by the formation of a slit (arrows) on the complex primordium, d Further differentiation of a lower primordium into a petal (P) and an upper primordium into a primary complex androecial primordium (A). e Idem; sepals (S) removed, f Inception of a five-angular gynoecium (G) on the apex and lateral extension of the five primary androecial primordia (A). - Bars: 20 ~tm

184

L.P. RONSE DECRAENE& E. SMETS:

Fig. 3. Floral development of Harungana madagascariensis, a Detail of a developing stamen fascicle above a developing petal (P). Note the successive inception (numbers). Bar: 14 ~tm. b Inception of five continuous septa and three stamen primordia on each complex primordium. Note the difference in size of the petals and the associated stamen fascicle. Bar: 40 ~tm. c Later stage of development of the septa (se); four petals removed. Bar: 40 pro. d Detail of the inception of a lower stamen primordium (arrow) on each primary primordium; petals removed from this stage on. Bar: 14 ~tm. e Lateral view of a flower bud with centrifugal differentiation of anthers on each fascicle and inception of a nectary primordium (arrow); sepals (S) and petals (P) removed. Bar: 40 ~tm. f Adaxial view of stamen group with their associated petal (P). Bar: 20 ~tm

Androecium and nectaries of Harungana

185

Fig. 4. Floral development of Harungana madagascariensis, a Lateral view of the flower. Note the stamen development and the location of the nectary (arrow). Bar: 25 ~tm. b Detail of the development of a nectariferous protuberance (arrow) below the carpel in alteration with the stamen fascicles (F). Bar: 20 ~tm. c Apical view of flower bud at the same stage of development. Note the starlike slit of the gynoecium. Bar: 40 ~tm. d Development of a common filament (F) and growth of nectaries (N). Bar: 120 ~tm. e Detail of nectary with unusual protuberance (arrow). Bar: 70 ~tm. f Lateral view of nearly mature flower. Note the fascicles and two-horned nectaries. Bar: 200 ~tm

186

L.P. RONSE DECRAENE& E. SMETS:

Fig. 5. Floral development of Harungana madagascariensis, a Apical view of nearly mature stage of development. Note the globose stigmas (ST). Bar: 250 gin. b Lateral view of developing ovule and inception of inner (I 1) and outer (I 2) integuments; carpel wall (W) partially removed. Bar: 20 btm. c Lateral view of mature flower. Note the external position of the nectaries (arrow) and the curved stamen fascicles (F); one petal (P) removed. Bar: 500 gm. d Apical view of mature flower. Note the pilose petals (P), small sepals (S) and the long stamen fascicles (F) alternating with the nectaries (arrow). Bar: 450 gm a third smaller lobe was seen between the two larger lobes (Fig. 4 e: arrow). The surface of the scales is rugate and bears occasionally some stomata. The morphological nature of these appendages is difficult to discern but the floral anatomy can help to unravel this problem (see below). Floral anatomy. The floral pedicel consists of a continuous ring of vascular bundles surrounding a parenchymatous pith, and a large-celled cortex (Fig. 6 a). At the base of the sepals ten bundles arise at almost the same level and branch off to the periphery (Fig. 6 b). Five of them represent sepal median traces running to the areas which start to bulge out, and the five other traces split into two units between the bulges; they represent the c o m m o n sepal lateral traces (Fig. 6 b: arrow). Sepal tissue is rapidly separated from the main body and contains the sepal median traces with on either side one of the sepal lateral traces (Figs. 6 e and 7 a, b). Petal traces become demarcated at five sectors alternating with the sepals (Fig. 6 c-f). Alternating with them patches of small cells with large nuclei become separated from the stele (Fig. 6 c, d). The tight mass of cells runs towards the periphery for

Fig. 6. Floral anatomy of Harungana madagascariensis: successive transverse sections from the base upwards, a Transverse section through the pedicel. Bar: 200 lain. b Departure of ten traces. The traces to Sepal laterals start to bifurcate (arrows). Bar: 200 ~tm. c Detail at the separation of the sepals with inception of petal traces (P) and patches of nectary traces (arrows). Bar: 120 ~m. d Section at slightly higher level. Note the petal traces (P) and separation of vascular tissue to the nectary (arrows). Bar: 120 ~m. e View at separation of sepals (S), petals (P) and nectaries (N). Departure of stamen fascicle traces (arrows). Bar: 200 ~tm.fDetail of the nectary base (N) linked with vascular tissue (arrows) and departure of stamen fascicle trace (A). Bar: 150 ~m

188

L.P. Ro~s~ DECRAENE& E. SMETS:

a while but ends abruptly in the base of the nectary (Figs. 6 fand 8 a). They represent the vascular supply of the nectaries. No tracheary elements were visible and the cells resembled not fully developed vascular tissue. These darkly staining cells tend to ramify before running over into the compact mass of the nectary (Fig. 6 f). At a slightly higher level five stamen fascicle traces (stamen trunk bundles) are given off opposite the petal traces (Figs. 6 f and 7 a). Although the petals are fused basally with the stamen fascicle, traces arise independently (Fig. 7 c-e). Petals become separated as flattened fan-like structures before the lateral parts start spreading as two winged bodies around the nectaries (Fig. 7 a-e). Petal bundles split in numerous traces at the level of separation from the stamen fascicles (Fig. 7 c). Stamen trunk bundles arise as one or two bundles per fascicle (Fig. 6 e); higher up these eventually reunite and become splitted in three traces, the middle one being the largest (Fig. 7 e, f). These run up into the different anthers. The more external anther of the tetrad receives another smaller vascular trace from the middle trace at a much higher level. Nectaries become separated from the stele at the same level as the stamen fascicles (Fig. 7 b, c). They are cordate to kidney-shaped in outline (Figs. 7 b, c, and 8 a). Higher up they are visible as two flattened lobes (Fig. 7 d, e). The nectarial tissue consists exclusively of densely packed darkly staining cells with large nuclei (Figs. 6 f and 8 a). At the level of separation of the nectaries five patches of vascular tissue remain in the stele opposite the sepals (Fig. 7 a); these become reorganized in a five-angular ring. The five angles, which are opposite the nectaries differentiate as five dorsal traces and a pair of marginal traces (Fig. 7 b, c). More inwardly a second ring is differentiated (Fig. 7 c). The outline of the five carpels become visible at this stage. Higher up five locules can be seen opposite the dorsal traces. The marginal traces unite in single traces and are situated opposite five septa. The septa bear basally inserted placentae with two ovules in each locule (Figs. 7 d, e, and 8 b). The central mass of vascular tissue has become confined to a number of ventral traces which branch off smaller traces to the bases of the ovules (Fig. 7 d, e). Each carpel contains two ovules, which are surrounded by two integuments (Figs. 7 d and 8 b). Low in the gynoecium placentation is axile (Fig. 7 d); higher up the septa become separated from each other in the central area of the gynoecium (Fig. 1), running progressively over into the stylar canal (Figs. 7 e, f, and 8 b). Within the whole gynoecial tissue long vesicular canals were observed, their inner walls being lined

Fig. 7. Floral anatomy of Harunganamadagascariensis:successive transverse sections from the base upwards, a Separation of fan-like petals (P) and reorganization of central vascular tissue in five groups, b Separation of stamen fascicle traces and nectaries, c Separation of the gynoecium. Note the abaxial fusion of the stamen fascicles with the petals and the fivecarpellate gynoecium, d Dorsal traces; arrows point to marginal traces, d Appearance of locules and axile placentae. Note five pairs of ovules on the placentae, d Dorsal traces, v ventral traces, m fused marginal traces, e View of the lower area of the stylar canal. Note the bitegmic erect ovules. Division of stamen fascicle traces in three traces (arrows). fView in the upper part of the locules. Note the imbricate arrangement of the carpels and the three-traced stamen fascicles. - Bars: 200 ~tm

Androecium and nectaries of

Harungana

189

190

L.P. RoNse DECRAENE& E. SMETS:

Fig. 8. Floral anatomy of Harungana madagascariensis, a Longitudinal section with a lateral view of the nectary (N). Note the vascular tissue (arrow), the gynoecial wall (G) and the sepal median trace (S). Bar: 110 ~tm. b Detail of the gynoecium in the upper part of the locules. Note the bitegmic ovule and the vesicular canals bordered with trichomatic or papillous cells. Bar: 80 ~tm

with trichomatic or papillous cells (Figs. 7 c-f and 8 b). Their function is unknown to us.

Discussion Stamen-petal complexes. A stamen-petal complex (or common stamen-petal primordium) can be found in certain species of the Clusiaceae and is related to a developmental retardation of the petals. Petal inception is usually inconspicuous and primordia are developing slowly. The lag between the inception of petals and the androecial primordia can be reduced, giving the impression of a single inception of a stamen-petal complex. Three possibilities are found in Clusiaceae: 1. the petals arise before the complex stamen primordia; 2. the complex stamen primordia arise simultaneously with the petals; 3. the complex stamen primordia arise before the petal primordia, which seem to be split off as appendages on the common primordia. In this way the petal primordia are absorbed into the larger primary primordia. Observations on the floral anatomy reveal that petals and stamen primordia

Androecium and nectaries of Harungana

191

may become fused externally while their vascular bundles are running independently at the same time. This demonstrates once again that evolutionary trends of the vascular tissue can run at a different rate than what evolves externally (cf. PvPd 1951, CARLQUIST1969). Stamen fascicles. It is clear from observations of the congenitally fused complex stamen primordia of some species of Hypericum that each complex behaves as an entity. PAYER (1857), HIRMER (1918), LEINS (1964a), and SATTLER (1973) clearly demonstrated that four of the five complex antipetalous stamen primordia may occasionally become coalescent in two pairs. The result is the differentiation of three fascicles, two larger inserted between the petals and one smaller opposite a petal. In that case stamen primordia are clearly independent from the petals, as the latter five primordia arise separately. In several unrelated taxa a trimerous gynoecium often induces a reduction of the stamens which take a position alternating with the carpels. This may be caused by simple loss of two stamens of a whorl (e.g., Caryophyllaceae: STERK 1970) or by fusion of two pairs of (complex) primordia into two units (Molluginaceae: BATENBURG & MOELIONO 1982, Ericaceae: N~si-uNo 1988, Dilleniaceae: DICKISON 1971, Clusiaceae: LziNs 1964 a). In five-carpellate species, the number of (complex) stamen primordia is never reduced, but consists of five primordia alternating with the carpels (cf. this study). KAWANO (1965) accepts a progressive reductive trend for the androecium of the subfam. Moronoboideae. The clustering of numerous free stamens to form fascicles of stamens is a first step; next the fascicles become fused and the number of stamens per fascicle is reduced. However, the regular inception of complex primordia in Harungana and the grouped appearance of the stamens in a fascicle deny the interpretation that the androecium is primitively polymerous. The fusion of numerous, originally free, stamens in fascicles and the consequent fusion of these fascicles, linked with a reduction in stamen number, is a process that is very complicated to visualize. To decide whether stamens are free or fused in a continuous unit is often not possible unless one studies the ontogeny. This has usually been omitted by those advancing a reduction of telomes. In all cases studied in literature the numerous stamens develop on separate complex primordia. If no interprimordial growth takes place within the fascicle, the stamens are considered to be free; if the stamen primordia are lifted by a common meristem, they are considered to be fused. The number of stamens per fascicle can vary considerably. In Harungana this number is relatively low and fixed (four) but in other taxa a greater variability can be found (e.g., Hypericum perforatum: SATTLER 1973). The fact that there are fewer stamens per fascicle, or the fact that the last initiated stamens of the fascicles are not developed beyond the stage of staminodes is no evidence to consider such an androecium to be more advanced than androecia containing higher numbers of stamens. The individual stamen inception on each complex primordium tends to follow a strict developmental pattern running successively downwards (see also LEINS 1964 a, SATTLER 1973, RONSEDECRAENE 1989). The process of multiplication of the stamens runs without intermission throughout the development of the rest of the flower. It is consequently possible that a need for a rapid development of the flower prevents all (i.e. the younger outer) stamens to reach complete maturity. The presence of outer nectariferous staminodes can also be of prior ecological importance.

192

L.P. RONSEDECRAENE~; E. SMETS:

The antisepalous appendages. The morphological nature of the nectariferous appendages is a point of controversy because there are arguments pointing for, but others against a staminodial nature. The antisepalous scales arise relatively late in ontogeny on the receptacle (at the time of differentiation of the anthers of the inner stamens and long after the gynoecium has become differentiated). This suggests a receptacular nature (nectaria axialia: SMETS 1988, SMETS ~¢ CRESENS 1988, see also RONSE DECRAENE t~¢ SMETS 1991). Indeed, LEINS (1964 a) considers the nectaries in Hypericum aegypticum as secondary emergences of the receptacle. It is probable that these nectariferous appendages, which are found alternating with stamen fascicles, should be homologized with the Harungana nectariferous scales. On the other hand, the time of initiation cannot be a sole reason for considering a structure as receptacular, for tissue can remain meristematic for a very long time. For example, in Samolus valerandi L. (Primulaceae, unpubl, obs.) the staminodial whorl emerges very late, long after the gynoecium has been incepted. In addition to this, the development of these appendages in Harungana has some resemblance with the inception of anthers. Two halves are separated by a median longitudinal groove, much in analogy with the pollen sacs on the thecae. Later growth of the antisepalous scales is very conspicuous. The occasional appearance of a third protuberance stresses this resemblance. As the appendages are vascularized by traces from the floral stele like those of the petals, sepals or stamens, they can also be regarded as morphological entities comparable to the other floral organs. The traces to the appendages arise before the separation of the antipetalous fascicle traces, suggesting a transformed antisepalous stamen whorl. Vascular tissue develops as a functional unit related to the physiological activity of the organ it vascularizes. The vascular tissue of the nectaries of Harungana contains no xylem elements, which is expected considering their function as secretory organs. It would be wrong to define these traces as relictual elements; indeed, their differentiation is related with the phylogenetic transformation of stamens into staminodes which take on a nectar-secreting function. So, the existence of undeveloped traces to the nectariferous appendages in clear alternation with the stamen fascicles, the outline of the nectarial organs reminiscent of anthers, and their position are irrefutable arguments for a staminodial nature (nectaria androphyllomina, type staminodialia: SMETS 1989). Moreover, the nectaries may not be considered as gynoecial emergences (nectaria gynoecialia: see SMETS 1988, SMETS & CRESENS 1988), because they are not associated with the gynoecium ("associated" = "completely or incompletely homologizable with and/or localized on"). The ontogeny and vasculature are the major arguments against a gynoecial nature. Thus the nectaries of Harungana madagascariensis must not be confused with the floral nectaries of, e.g., Salicaceae or Apocynaceae which possess typically persistent ("nectaria persistentia" versus "nectaria caduca", see SMETS 1986), gynoecial nectaries. We thus firmly believe that nectar-secreting appendages of Harungana madagascariensis (and probably other Clusiaceae as well) represent a whorl of nectariferous staminodes and not receptacular outgrowths. This is an additional argument for considering the ancestors of Clusiaceae as diplostemonous (see RONSE DECRAENE • SMETS 1987). If the nectaries are homologizable with staminodes, we

Androecium and nectaries of Harungana

193

believe that they arose by transformation of a whorl of antisepalous stamens (a diplostemonous condition). Other examples are known where the antipetalous stamens become increased while the antisepalous stamens remain single or become staminodial (see, e.g., BRow~ 1935, RONSE DECRAENE 1990). This illustrates the easy transition between different androecial configurations, which are all linked by a basic diplostemony. RONSE DECRAENE & SMETS (1987) postulated that diplost e m o n y is ancestral for this group of plants. The occurrence of diplostemony, together with polymerous androecia, in related taxa of Theales stresses this link. This is also confirmed by examples of species with ten or eight stamens (see introduction). We thank Prof. Dr E. PETIT, director of the National Botanical Gardens of Belgium in Meise for permission to collect material and for the use of the SEM. We are endebted to Mrs CATI-IERINERONSE DECRAENEfor drawing the floral diagram. Our thanks also go to Mr MARCELVERttAEGEN for his technical assistance with the SEM. Financial support from the N F W O - Kredieten aan navorsers is gratefully acknowledged. R e f e r e n c e s

BAILLON,H., 1873: Du genre Garcinia et de l'origine de la gomme-gutte. - Adansonia 10: 283-298. 1875: Hyp6ricac6es. - Histoire des Plantes 55: 379-383. BATENBURG,L. H., MOELIONO,B. M., 1982: Oligomery and vasculature in the androecium of Mollugo nudicaulis LAM. (Molluginaceae). - Acta Bot. Neerl. 31: 215-220. BROWN, E. G. S., 1935: The floral mechanism of Saurauia subspinosa ANT~. -- Trans. Proc. Bot. Soc. Edinburgh 31: 485497. CARLQUIST,S.~ 1969: Towards acceptable evolutionary interpretations of floral anatomy. - Phytomorphology 19: 332-362. CRONQUIS%A., 1981: An integrated system of classification of flowering plants. - New' York: Columbia University Press. DAHLGREN,G., 1989: The last Dahlgrenogram. System of classification of the dicotyledons. - In TAN, K., (Ed.): Plant taxonomy, phytogeography and related subjects, pp. 249260. - The Davis and Hedge Festschrift: Edinburgh University Press. DAHLGREN, R., 1980: A revised system of classification of the angiosperms. - Bot. J. Linn. Soc. 80: 91-124. 1983: General aspects of angiosperm evolution and macrosystematics. - Nordic J. Bot. 3:119-149. DICKIsoN, W. C., 1971: Comparative morphological studies in Dilleniaceae VII. Additional notes on Acrotrema. - J. Arnold Arbor. 52: 319-333. EICHLER, A. W., 1878: Blfitendiagramme II. - Leipzig: Wilhelm Engelmann. EN~LER, A., 1925: Guttiferae. - In EN~LER, A., PRANTL, K., (Eds.): Die nat/irlichen Pflanzenfamilien 21, pp. 154-237. - Leipzig: Wilhelm Engelmann. HIRMER, M., 1918: Beitrfige zur Morphologie polyandrischer Blfiten. - Flora 110: 140192. HOVMEISTER, 1868: Allgemeine Morphologie der Gewfichse. Handbuch der physiologischen Botanik I, 2. - Leipzig: Wilhelm Engelmann. KAWANO,S., 1965: Anatomical studies on the androecia of some members of the GuttiferaeMoronoboideae. - Bot. Mag. (Tokyo)78: 97-108. LEINS,P., 1964 a: Die friihe Blfitenentwicklungvon Hypericum hookerianum WIGHT• ARN. und H. aegypticum L. - Ber. Deutsch. Bot. Ges. 77:112-123. 1964b: Das zentripetale und zentrifugale Androeceum. - Ber. Deutsch. Bot. Ges. 77: 23-25. -

-

-

194

L.P. RONSE DECRAENE& E. SMETS: Androecium and nectaries of Harungana

1971: Das Androecium der Dikotylen. - Ber. Deutsch. Bot. Ges. 84: 191-193. 1975: Die Beziehungen zwischen multistaminaten und einfachen Androeceen. - Bot. Jahrb. Syst. 96: 231-237. MONCVR, M. W., 1988: Floral development of tropical and subtropical fruit and nut species. An atlas of scanning electron micrographs. - Melbourne: CSIRO. NISHINO, E., 1988: Early floral organogenesis in Tripetaleia. - In L~INs, P., TUCKER, S. C., ENDRESS, P. K., (Eds.): Aspects of floral development, pp. 181-190. - Berlin: J. Cramer. PAYER, J. B., 1857:Trait6 d'organog6nie compar6e de la fleur. - Paris: Victor Masson. PURI, V., 1951: Role of floral anatomy in the solution of morphological problems. - Bot. Rev. 17: 472-534. RONSE DECRAENE, L. P., 1988: Two types of ringwall formation in the development of complex polyandry. - Bull. Soc. Roy. Bot. Belgique 121: 122-124. - 1989: The floral development of Cochlospermum tinctorium and Bixa orellana with special emphasis on the androecium. - Amer. J. Bot. 76: 1344-1359. 1990: Morphological studies in Tamaricales I: floral ontogeny and anatomy of Reaumuria vermiculata L. - Beitr. Biol. Pflanzen 65: 181-203. SMETS,E., 1987: The distribution and the systematic relevance of the androecial characters oligomery and polymery in the Magnoliophytina. - Nordic J. Bot. 7: 239-253. 1991: The floral nectaries of PoIygonum s.1. and related genera (Persicarieae and Polygoneae): position, morphological nature and semophylesis. - Flora 185: 165-185. SACHS, J., 1874: Lehrbuch der Botanik. - Leipzig: Wilhelm Engelmann. SATTLER, R., 1973: Organogenesis of flowers, a photographic text-atlas. - Toronto, Buffalo: University of Toronto Press. SMETS, E., 1986: Localization and systematic importance of the floral nectaries in the Magnoliatae (Dicotyledons). - Bull. Jard. Bot. Nat. Belgique 56: 51-76. 1988: La pr&ence des "nectaria persistentia" chez les Magnoliophytina (Angiospermes). - Candollea 43: 709-716. 1989: The distribution and systematic relevance of caducous nectaries and persistent nectaries in the Magnoliophytina (angiosperms). - Acta Bot. Neerl. 38: 100. CRESENS,E., 1988: Types of floral nectaries and the concepts "character" and "character-state"-a reconsideration. - Acta Bot. Neerl. 37: 121-128. STEBBINS, G. L., 1974: Flowering plants - Evolution above the species level. - Cambridge: Harvard University Press. STERK, A. A., 1970: Reduction of the androecium in Spergularia marina (Caryophyllaceae). - Acta Bot. Neerl. 19: 488-494. THORNE, R. F., 1983: Proposed new realignments in the Angiosperms. - Nordic J. Bot. 3:85-117. VAN HEEL, W. A., 1966: Morphology of the androecium in Malvales. - Blumea 13: 177394. WILSON, C. L., 1937: The phylogeny of the stamen. - Amer. J. Bot. 24: 686-699. 1942: The telome theory and the origin of the stamens. - Amer. J. Bot. 29: 759-764. -

-

-

-

-

-

-

-

Address of the authors: L. P. RoNsE DECRAENE and E. SMETS,Laboratory of Systematics, Botanical Institute, Catholic University of Leuven, Kardinal Mercierlaan 92, B3001 Heverlee (Leuven), Belgium. Accepted August 23, 1991 by F. Er~RENI~ORFZR