| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Structure and Development |
Laboratory of Plant Systematics, Institute of Botany and Microbiology, Katholieke Universiteit Leuven, Kasteelpark Arenberg 31, B-3001 Leuven, Belgium
Received for publication August 7, 2003. Accepted for publication January 8, 2003.
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMATERIAL AND METHODSRESULTSDISCUSSIONLITERATURE CITED |
|---|
Key Words: Ericales • floral ontogeny • morphology • Primulaceae • Samolus • scanning electron microscopy (SEM) • Theophrastaceae
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIAL AND METHODSRESULTSDISCUSSIONLITERATURE CITED |
|---|
, namely Theophrastaceae, Myrsinaceae and Primulaceae, are currently included in the enlarged Ericales (e.g., Savolainen et al., 2000
; Soltis et al., 2000
; APG, 2003
). Analyses of molecular data showed that the classical subdivision of Primulales is artificial. Theophrastaceae are still supported as a monophyletic group, but Myrsinaceae as well as Primulaceae appear to be paraphyletic (Anderberg et al., 1998
; Källersjö et al., 2000
). The genus Maesa Forssk., formerly the single genus of the subfamily Maesoideae of Myrsinaceae, was lifted out of this family and given a family rank by Anderberg et al. (2000)
on the basis of morphological as well as molecular data. Maesaceae are sister to the other families of the former Primulales. Theophrastaceae are, together with their sister taxon Samolus L., a second lineage to diverge (Källersjö et al., 2000
). Hence, Samolus was transferred by Källersjö et al. (2000)
to Theophrastaceae.
Floral characters have proven to be valuable in defining relationships within the former Primulales: the separation of Maesa from Myrsinaceae was also supported in floral ontogenetic results (Caris et al., 2000
). Furthermore, there is floral ontogenetic evidence (P. Caris, personal observations) to support the generic interrelationships of the Myrsinaceae-Primulaceae complex as proposed by Anderberg et al. (1998)
and Källersjö et al. (2000)
. In the present paper we compare the floral ontogeny of six species from the major Theophrastaceae genera to the ontogeny of Samolus valerandi L. in order to find possible morphological support for the incorporation of Samolus into Theophrastaceae.
The Theophrastaceae are a neotropical family of about 100 species. The two largest genera are Clavija Ruiz & Pav. (ca. 50 species) and Jacquinia L. (ca. 40 species). Deherainia Decne. and Theophrasta L. are the other commonly recognized genera of the family. The monospecific genus Neomezia Votsch was added to this list by Votsch (1904)
. Ståhl (1993)
derived a sixth genus from Jacquinia nemophila Pittier, namely Votschia Ståhl, with a single species from northeastern Panama.
Theophrastaceae are characterized by their alternate, often pseudoverticillate leaves. All members of the family are shrubs or small trees, with extraxylary foliar sclerenchyma as a common characteristic. This sclerenchyma is arranged in subepidermal layers or bundles and was thought to be a synapomorphy of the family.
Flowers of Theophrastaceae are arranged in terminal or axillary racemes, rarely reduced to a single flower. Beneath each flower a small bract can be found, often adnate to the pedicel. The sympetalous flowers are actinomorphic, tetramerous, or pentamerous, and they have a persistent calyx and a firm and waxy corolla, both with an imbricate aestivation (Ståhl, 1990a
). Theophrastaceae are reported to be characterized by the absence of bracteoles, the presence of staminodes opposite the sepal lobes and fused to the corolla, antepetalous stamens that develop from common petal-stamen primordia, and anthers with extrorse dehiscence, typically containing calcium oxalate (Mez, 1903
). Except for the latter character, however, these are not exclusive features of Theophrastaceae when compared to other families of the former Primulanae. With Sapotaceae, Theophrastaceae share the presence of staminodes; the firm, thick corolla with a well-developed tube; and the extrorse dehiscence of the anthers, but they differ by the absence of latex and the structure of the gynoecium (Mez, 1903
).
Sattler (1962)
described the division of the common primordia within Primulales using the principle of the "variablen Proportionen." Depending on the relative moment on which the separate primordia appear and depending on the growth rate of the primordia, different situations can occur. The situation in Clavija macrophylla (Link) Radlk. is considered to be a marginal case (Sattler, 1962
), with two primordia appearing from the start; Deherainia smaragdina Decne. and C. aff. elliptica Mez are the subsequent species in a series of Primulales members, in which the presence of common primordia is gradually getting more and more distinct.
Staminodes develop in antesepalous position, on a whorl outside the stamen whorl. In all species studied, staminodes are born from the areas in between the petals and they become fused to the corolla tube. Staminodes that are fused to the corolla tube are also present in Lysimachia section Seleucia and in most species of Samolus (Ståhl, 1990b
).
The circumscription of the genera mainly depends on floral characteristics. Staminodes of Theophrastaceae vary considerably and they seem particularly useful morphological characters to distinguish among genera (Ståhl, 1990b
).
Deherainia can be distinguished from the other genera by its large, green flowers, which appear singly or with a few together. It has rather inconspicuous staminodes. Flowers of Jacquinia are white or orange-red, bisexual, with flattened, petaloid staminodes inserted at the top of the corolla tube, and free stamen filaments adnate to the corolla base. They are usually arranged in terminal racemes (Mez, 1903
). In Clavija, the pale yellowish to orange flowers are often unisexual. In male flowers one can find a rudimentary gynoecium, while female flowers possess free stamens with anthers that remain closed during anthesis (Ståhl, 1991
). Filaments of staminate and hermaphrodite flowers are fused into a staminal tube (Ståhl, 1990b
). The staminodes are large, gibbous, and more or less ovoid structures, often set with glandular hairs. Smets (1988)
described these glandular staminodes as nectaria caduca. However, no nectar has been observed in the flowers so far (Ståhl, 1991
). In Theophrastaceae, nectar secretion has only been reported from flowers of Jacquinia (Vogel, 1986
). In Clavija, the leaves are clustered in more or less distinct pseudo-whorls on the upper part of the stem and the flowers are most abundant among and just beneath them (Ståhl, 1990b
, 1991
). The trees are cauliflorous or flowers are placed in lateral racemes in the axils of the leaves. Theophrasta and Neomezia have a similar arrangement of the leaves. In Theophrasta the laterally born racemes are found strictly suprafoliar. Reduced leaves occur along the stem as spiny scales (Mez, 1903
; Ståhl, 1990b
). The flowers are pentamerous and bisexual. The anthers are produced at the apex to form a triangular appendage (Ståhl, 1987
). Theophrasta possesses large staminodes and glabrous fruits, while Neomezia can be recognized by its rather inconspicuous, triangular staminodes and puberulous fruits; like Theophrasta it possesses long anther appendages (Ståhl, 1990b
). Votschia resembles Jacquinia in having similar staminodes, and in its corolla shape it resembles Deherainia, but it differs from these genera with regard to the exine of the pollen grains and the petiolar vascularization (Ståhl, 1993
). Jacquinia and Votschia both possess flattened, petaloid staminodes, fused with the corolla at the mouth of the tube. These staminodes are somewhat smaller than the petals, and both whorls may play a role in attracting pollinating insects.
Seven species were examined for the ontogenetical study. Clavija species are hermaphrodite or more frequent sexually polymorphic (gynodioecious, polygamous, androdioecious, or dioecious). We studied the floral development of the gynodioecious C. latifolia (R.&S.) K.Koch, and the androdioecious C. ornata D.Don.
Deherainia smaragdina has axillary flowers, occurring solitary or with a few together (Mez, 1903
). They have a short campanulate corolla and liguliformous or triangular staminodes inserted at the top of the corolla tube, which are fairly small when compared to the staminodes of the other genera (except Neomezia).
Jacquinia aculeata (L.) Mez has reduced inflorescences of 2–5 flowers (Borhidi and Muñiz, 1978
). Another species of Jacquinia at which we looked is J. macrocarpa Cav. The morphological variation encountered in this species with orange- red flowers is mainly geographically correlated (Ståhl, 1989
). Källersjö and Ståhl (2003)
suggested a division of Jacquinia into two genera, one (Jacquinia) with the white-flowered species, the other one (Bonellia) with mostly orange-red-flowered species. Jacquinia aculeata would still belong to the genus Jacquinia, whereas J. macrocarpa should be transferred to Bonellia. As a result, the variation within the genus Jacquinia is taken into account in the present study.
The pentamerous, bisexual flowers of Theophrasta americana L. are arranged in 7 cm long, many-flowered racemes. The petals are united to three-quarters of their length to form a large corolla tube with suborbicular petal lobes (Mez, 1903
).
Samolus is a (sub)cosmopolitan genus of perennial or suffrutescent herbs (Mabberley, 1997
). It consists of about 15 species and has its main distribution in the Southern Hemisphere (Källersjö et al., 2000
). Samolus valerandi is cosmopolitan. It possesses small, white flowers aggregated in a terminal raceme. Samolus has traditionally been placed within Primulaceae, though in a separate subfamily Samoleae. It differs from the other members of the Primulaceae by its semi- inferior ovary, a character state that can only be found in one other genus of the former Primulales, namely Maesa, which takes a basal position with respect to the whole primuloid group. According to Thenen (1911)
Samolus deserves a separate position within Primulaceae, because the secondary vascular traces of the calyx are never reduced; in the corolla, however, they may be reduced due to the strong development of the primary vascular traces. The latter are in most Primulaceae unbranched, but not in Samolus.
|
MATERIAL AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIAL AND METHODSRESULTSDISCUSSIONLITERATURE CITED |
|---|
The floral buds were dissected in 70% ethanol under a stereomicroscope (Wild M3; Leica Microsystems AG, Wetzlar, Germany) equipped with a cold light source (Schott KL 1500; Schott-Fostec LLC, Auburn, New York, USA). Next, the buds were washed twice for 5 min with 70% ethanol before being exposed to a mixture (1 : 1) of 70% ethanol and DMM (dimethoxymethane) for a further 5 min. Afterwards, the material was placed in pure DMM for 20 min, and critical point dried using liquid CO2 with a BAL-TEC CPD030 (BAL-TEC AG, Balzers, Liechtenstein). Finally, the dried material was mounted onto stubs using Leit-C and gold-coated with a sputter coater (SPI Supplies, West Chester, Pennsylvania, USA). Observations were made using a JEOL JSM-5800 LV scanning electron microscope (JEOL, Tokyo, Japan) at the National Botanic Garden in Meise (Belgium).
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIAL AND METHODSRESULTSDISCUSSIONLITERATURE CITED |
|---|
|
|
|
Deherainia smaragdina
The calyx develops unidirectionally, with the first sepal abaxially on the flower primordium in the axil of the bract, but not exactly on the median axis (Fig. 4A). Afterwards, two lateral sepals are initiated and finally two sepals are initiated on the adaxial side of the flower. All sepals differ in size, suggesting that they all develop consecutively (Fig. 4B). The other floral whorls are initiated more or less at the same time. By horizontal growth between the calyx and the center of the floral primordium, the floral apex becomes separated from the calyx (Fig. 4C, D). It forms a pentagonal shape (Fig. 4C) and the lateral margins seem to grow out centrifugally while the primordia of the inner whorls (corolla, androecium and staminodes, and gynoecium) are initiated (Fig. 4D). All these whorls develop unidirectionally, towards the axis (Fig. 4E). The unidirectional development starts between the first and the second sepal. On the periphery of the floral apex five petal primordia develop and almost at the same time, stamen and staminode primordia arise. Soon after the appearance of primordia on the adaxial side of the flower, a gynoecial ring primordium is formed in the center of the flower in a central depression (Fig. 4E). While the outer whorls are developing, the young ovary becomes closed, leaving a slit on the top, the orientation of which is not fixed (Fig. 4F–H, L). Underneath the petals, the stamens start differentiating and an anther becomes visible on an initially short filament (Fig. 4G–K). The staminodes alternate with the petals and common zonal growth under petals and staminodes results in the formation of a corolla tube to which the staminodes become adnate (Fig. 4H, I). The truncate anthers are curved inwardly. The thecae are clearly bilobed, each lobe representing a pollen sac (Fig. 5G). The cells on the surface of the anthers elongate, giving the anther a specific, papillose texture, except on the upper side, above the stomium (Fig. 5G). All thecae and pollen sacs take a radial position, with the longitudinal slits directed to the outside of the flower, so that anthers are extrorse (Fig. 5G). The flowers of D. smaragdina are protandrous: firstly, the stamens lie close to the style, but after a while they bend outward (Fig. 5H), disclosing the discoid stigma, still with a slit in the middle (Fig. 5I).
|
|
Jacquinia
The yellowish-white flowers of J. aculeata develop in few-flowered racemes on long, slender pedicels that arise from the axils of bracts at the end of the branches (Fig. 6A, B). Sepals develop in a two-fifths spiral (Fig. 6C). In J. macrocarpa, the inflorescences are few-flowered, normally with up to five orange-red flowers. Calyx development is unidirectional, starting from the abaxial side of the flower primordium (Fig. 7A). As in Deherainia smaragdina the zone inside the calyx starts to grow horizontally and the floral apex becomes separated from the calyx (Figs. 6D, E, 7B). Its form is a little concave, and in J. aculeata, five petal primordia are initiated on the outer side (Fig. 6E). While the outer rim under the petal primordia is growing out, stamen primordia can be observed on its inner side, in antepetalous position (Fig. 6F, G). Even before all stamen primordia have been initiated, a more or less ringlike gynoecium primordium develops from the center of the flower bud (Fig. 6F, G). At this point, the petals start to grow hiding underlying parts (Fig. 7F, G). When they are removed, the inwardly growing stamen primordia can be observed. Staminode primordia are initiated on the inner side of the developing corolla, and they alternate with the petals (Fig. 6H). In J. macrocarpa, individual petal, stamen, and staminode primordia originate simultaneously on the outer rim of the concave apex (Fig. 7B, C). In both species, the ring primordium of the gynoecium enlarges (Fig. 7D, E) and the ovary is formed, bearing a short style (Figs. 6H, I, 7F, H, I, K). The capitate stigma has 2–4 terminal divisions (Fig. 6I). As the staminodes grow larger, the stamens differentiate and dithecal anthers are formed, which are obtuse at their apex (Figs. 6H–J, 7J). The thecae of an anther are curved outwards, and they are squeezed in between the petals (Fig. 6J). In an apical view, the thecae are heart-shaped, with the tips of the hearts pointing to the outer side of the flower (Fig. 6J). At this stage the filaments start to grow and they lift up the anthers in the developing flower. In J. aculeata, the flattened and very broad filaments are fused together, up to half of their length, to form a well-developed stamen tube, adnate to the lower part of the corolla tube (Fig. 6K, L). In J. macrocarpa, the filaments of the stamens are fused at the very base into a short tube, which is adnate to the base of the corolla (Fig. 7O, P). As in Deherainia the texture of the surface of the anthers changes while they elongate to their full length. The extrorse anthers open with two longitudinal slits (Fig. 6K). Inside the gynoecium, the placenta is formed by the remaining part of the floral apex (Figs. 6M, 7L). Initially, this placenta has a conical shape. Numerous ovule primordia become visible on the placenta in a downward spiral, starting close to the top (Fig. 6N). Ovules are anatropous, and in J. aculeata they are more or less arranged in three whorls (Fig. 6O). The rotund, sterile top of the placenta will expand considerably, and some capitate trichomes can be observed on it (Fig. 6O). On the older placenta, the ovules lie tightly arranged. Inside the gynoecium of J. macrocarpa several whorls of ovule primordia are initiated, much more (Fig. 7M) than in J. aculeata. In mature flowers the flattened staminodes have developed into large petaloid organs that are adnate to the corolla and are situated in between the petal lobes at the mouth of the corolla tube (Figs. 6P, 7N).
|
|
|
|
|
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIAL AND METHODSRESULTSDISCUSSIONLITERATURE CITED |
|---|
described the sepals as virtually free. According to our observations one might as well call them virtually connate. Our results clearly show that after the initiation of the sepal primordia, the floral apex enlarges significantly, while the sepals are developing. Next, the other floral organs appear on the floral apex. We observed a similar development in all genera studied, although less distinctly in Clavija. Because of the longer plastochron between the development of the free sepals and the development of the other floral organs on the apex, which could be interpreted as a kind of anthophore, the sepals seem to be united in a tube. As a result, the calyx is "pseudo-synsepalous." We choose not to speak of a true calyx tube in Theophrastaceae because of the long plastochron and the resulting receptacular region between the sepals and the remaining part of the floral apex. It is, in fact, due to the presence of this region that the individual sepals appear to be connected.
The short tube at the base of the sepals might as well be called a hypanthium instead of a calyx tube. A hypanthium can be described as a floral cup developing from common zonal growth at the base of (congenitally fused) perianth (and androecium) parts. Whether this tube should be interpreted as (partially) receptacular remains an arbitrary decision. Any tubular structure supporting the calyx plus the corolla lobes and sometimes also the stamens is considered to be an hypanthium. When the dorsal flanks of the carpels are included in the formation of the hypanthium as well, it is called a gynoecial hypanthium and the result is an inferior ovary (Leins, 2000
). In Theophrastaceae, the enlarged part between the sepals and the floral apex might be interpreted as a very short and early developing hypanthium bearing free sepals and the secondarily formed apex, on which the other floral parts develop.
The former Primulales are characterized by the presence of common primordia, from which both petals and stamens develop. Contradictory to this general belief, however, we did not observe common primordia in the genera of Theophrastaceae. Petals, stamens, and staminodes develop from separate primordia directly from the floral apex, but more or less at the same time. It could be argued that Theophrastaceae form a primitive line in primuloid groups in which the congenital fusion between the petal and the stamen primordia is not yet fully completed. Sattler (1962)
observed the development of five common primordia in Clavija aff. elliptica. These primordia become immediately united at the base. Just before the development of the gynoecium a petal primordium appears at the dorsal side of the common primordium. Meanwhile the staminodia develop in between the common primordia (Sattler, 1962
). In C. macrophylla the common primordia are split up into a stamen primordium and a petal primordium from the beginning (Sattler, 1962
). Still according to Sattler (1962)
the petals of Deherainia smaragdina develop in a very early stage at the outside of the common primordium. Sattler (1962)
finds it reasonable to doubt the presence of common primordia in this species. All depends from where the borderline between separate primordia and floral apex is drawn. Conclusively, we may say that, in Theophrastaceae, the presence of common primordia can at least be doubted. If common primordia are present, the top of the primordium will develop into a stamen, and the petal arises on the outside. According to our observations, however, separate primordia are present from the start in all Theophrastaceae species investigated.
The staminodes of Clavija are fused with the inner parts of the petals at the mouth of the bowl-shaped flower. These basal inner parts of the petals grow inwardly in a later stage of the ontogeny, and they become adnate to the staminodes to form a continuous staminodial ridge. In Theophrasta a similar situation can be found. However, in this genus the thick ridge that is formed by the inner parts of the petals and the staminodes is placed at the base of the well-developed corolla tube. In both genera, the emergences of the petals resemble the staminodes and together they hide the underlying part of the flower.
The gynoecium of all Theophrastaceae species investigated for this paper develops from a ring primordium at the center of the floral apex. Some authors have made suggestions about the number of carpels. Mez (1903)
constructed a flower formula for the family in which he stated that the gynoecium is composed of two or possibly three carpels. Ståhl (1985)
also remarked that the unilocular ovary might consist of two carpels. However, in our results no sign of the number of carpels has been observed in the early developmental stages of the gynoecium. The irregular division of the stigmatic surface might suggest a carpel number, but the gynoecium is not built by a postgenital fusion of separate carpels; it always develops from a ring primordium. The ovary is often ovoid and bears a short style, usually with a truncate, (sub)capitate, or discoid stigma.
The free-central placenta starts as a spherical outgrowth of the floral axis, becomes conical, and ovule primordia arise on it in a downward spiral. Usually there are many ovules, but in Clavija their number is often restricted to less than 30, or only a few, especially in hermaphrodite flowers. The ovules are bitegmic and tenuinucellate. They are set closely together and not embedded in the placental tissue. On top of the placenta a sterile tip was observed, which is blunt and rounded. In several flowers of Jacquinia aculeata trichomes were observed on the sterile tip of the placenta. Their precise nature and function is unknown to us, but most probably, these are intra- ovarian trichomes. According to Dickison (1993)
placental trichomes are known from several families (e.g., Araceae, Styracaceae, Urticaceae, Euphorbiaceae). However, the occurrence of intra-ovarian trichomes within angiosperms is uncommon. Within Ericales, they are at least present in one Sapotaceae species, in part of the Rhododendron L. species, and in several styracaceous genera (Dickison, 1993
). Possibly they function in directing pollen tube growth (Dickison, 1993
). It seems appropriate to assume that they have a similar function in Jacquinia. Because they occur on the sterile tip of the placenta, it could be possible that they link the free-central placenta to the stylar canal. It is unclear if something is secreted by these structures in Jacquinia.
The particular development of the calyx and a secondarily formed apex, as described here by us for the Theophrastaceae, can easily be compared with the observations and conclusions on the placenta of Primulaceae as presented by Grégoire (1935)
. In some anomalous flowers of Primula, a flower or even a complete inflorescence develops from the top of the placental column, from which one might conclude that the median part of the placenta is, in fact, the top of the floral axis. It is important to note, however, that the flowers that develop from the placenta never possess a calyx (Grégoire, 1935
). It appears that a floral top, "sommet floral" as Grégoire (1935
, p. 298) calls it, develops after the sepals have been formed. This floral top should be compared to what we have called the remaining part of the floral apex. Indeed, from this top, the petals, stamens, and the gynoecium arise. According to Grégoire (1935)
the sepals of the Primulaceae have an origin, similar to that of leaves on a vegetative axis, before the transformation of the top to a floral primordium. He stated that the central placenta of the Primulaceae shows the same constitution as a floral primordium. The placenta represents the median part of a typical floral primordium and develops as a protuberance at the base of the ovary. The meristematic zone, which produces petals, stamens, and the ovary on a primitive floral primordium, will produce here ovules on the placenta. It is easily understood that the sepals, which develop like leaves on a vegetative top, and which do not arise from a floral top, are never found on the flowers that arise from the placenta (Grégoire, 1935
).
The habit and distribution of Theophrastaceae and Samolus are clearly different. We found some important similarities and differences with respect to the flower morphology and floral ontogeny as well, on which we will briefly comment here.
In Samolus the flowers are placed in corymbs, panicles, or racemes at the end of generative stems, while Theophrastaceae usually have flowers that are arranged in terminal or lateral racemes. Flowers of Samolus are pentamerous and hermaphrodite, with antepetalous stamens and a semi-inferior ovary. Each flower is subtended by a bract, which is raised to the middle of the pedicel (Pax, 1889
). As a result, we get a situation very similar to that of the Theophrastaceae, where bracts are adnate to the pedicel as well. Bracteoles are lacking.
While we did not observe a well-developed calyx tube within Theophrastaceae, Samolus is clearly synsepalous: the sepals are laterally fused into a short tube. Nevertheless, it could be argued that this is the result of the perigynous position of the outer whorls. On the floral primordium sepals and common primordia develop on the periphery, and already in a very early stage of development intercalary growth at the base of the peripheral rim results in a distinct depression at the center of the floral apex. Hence, a perigynous flower develops, in which the borderline between the calyx tube and the gynoecial hypanthium is difficult to observe. According to Payer (1857)
the sepals in Samolus are initiated in a normal two-fifths sequence, and they develop free from each other. According to Soukup (1972)
, however, the sepals are fused to the middle and adnate to the ovary. This is the result of the particular development of the gynoecium, which we will discuss later. The calyx has a quincuncial imbricate aestivation, but transitions towards a cochlear pattern, characteristic of the Theophrastaceae, are regularly found (Sattler, 1962
). Likewise, the petals have a quincuncial imbricate aestivation, but by inclusion of the second petal, a cochlear pattern may develop. Unlike Theophrastaceae, Samolus does have distinct common primordia on which petal primordia develop on the abaxial side. From the top of the common primordium a stamen develops. According to Pax (1889)
, the thin petals develop as dorsal appendages of the stamens. By common zonal growth at the base of the common primordia a short stamen-corolla tube develops.
The most obvious floral character that is shared by the Theophrastaceae and Samolus is the presence of a staminode whorl outside the stamen whorl. However, staminodes can be found sporadically in other primuloid genera as well. In Primulaceae a rudimentary antesepalous stamen whorl is only rarely present. Normally, the outer stamen whorl is completely reduced. When staminodes do occur, they develop late in the ontogeny. In Samolus, staminodes develop as liguliform structures at the mouth of the campanulate corolla, alternating with the petals. According to Sattler (1962)
the development of the staminodes is comparable to that of Clavija aff. elliptica, namely a development from the sinuses in between the petal primordia. However, our observations on Theophrastaceae show that staminodes do not appear after the corolla has been initiated, but more or less simultaneously. As a result, staminode primordia develop directly on the floral apex and not from the sinuses in between petal primordia on a corolla ring primordium. Of course, the staminode primordia alternate with petal and stamen primordia. When analyzing our observations of Samolus, we agree with Sattler (1962)
on the development of staminodes in this genus. Indeed, in Samolus a ring primordium is initiated, on which sepal primordia and common petal-stamen primordia arise secondarily. It is only much later, after the initiation of the gynoecium, when separate petals and stamens have developed and anthers are getting formed, that staminodes are initiated on a whorl corresponding to the position of an antesepalous stamen whorl. Afterwards, the staminodes grow out while the anthers are differentiating, and together with the stamens they are lifted up by the growth of the common stamen-corolla tube, getting a position at the mouth of this tube. So, in their development, staminodes of Theophrastaceae on the one hand and of Samolus on the other hand differ with respect to the moment of initiation and the fact that development occurs independently of stamen and petal development in Theophrastaceae. In all cases where staminodes are present in primuloid groups, the staminode whorl might be interpreted as an abortive antesepalous stamen whorl, although there certainly is some morphological and ontogenetical variability regarding the origin, development, and appearance of staminodes in primuloids.
In Samolus the whorl of staminodes is on the basis of the vascular anatomy currently interpreted as a reduced antesepalous stamen whorl. According to Thenen (1911)
the presence of secondary vascular traces in the corolla of Primulaceae cannot be used as an argument to support the hypothesis of a reduced antesepalous stamen whorl. These secondary vascular traces diverge above the corolla tube towards two neighboring petals. They are positioned in between two petals, and consequently, they could be interpreted as vascular traces that belong to a reduced antesepalous stamen whorl. Nevertheless, secondary vascular traces can also be observed in the calyx, and moreover, secondary traces remain present in genera that possess staminodes, such as, for instance, Samolus. If the secondary vascular traces were reminiscent to a reduced stamen whorl, then they would be lacking in a species with staminodes, which are homologous to the antesepalous stamens. However, this is not the case. On the contrary, the reduction of staminodes, for example in S. ebracteatus H.B.&K., goes along with the reduction of secondary vascular traces instead of enhancing the development of these traces (Thenen, 1911
).
In Samolus, the development of the gynoecium is completely different to that of Theophrastaceae. To a certain extent, this might be a consequence of the semi-inferior position of the ovary. According to Payer (1857)
, the gynoecium of Samolus develops from two distinct parts. Shortly after the appearance of the stamens, a depression is formed at the center of the flower at a certain distance from the stamen whorl. From the zone between this depression and the base of the stamens, a circular primordium develops that rapidly grows to form a sort of chimney tube above the depression. In Theophrastaceae, the gynoecium develops from a ring primordium that arises in the depression at the center of the flower. According to Payer (1857)
, the gynoecium of Samolus consists of an inferior part that is formed by the receptacular cup and a superior part, being the short style that develops from the circular primordium.
Our results confirm this description. The peripheral region of the floral apex enlarges while sepals and common primordia are initiated on it. Consequently, the floral apex becomes concave, with a depression at the center in which in a later stage the placenta will arise. In fact, an incomplete gynoecial hypanthium is formed, lifting up the peripheral floral parts, i.e., sepals, petals, staminodes, and stamens. Normally, the formation of a gynoecial hypanthium results in a completely inferior ovary, but in Samolus, the hypanthium does not fully develop, and a semi-inferior ovary will eventually characterize the flowers. While the hypanthium is getting shaped, the central depression, i.e., the actual ovary, is covered completely by centripetal growth from the inner basal side of the common primordia. At the center of the resulting gynoecial cap the style develops, bearing a capitate stigma. This cap offers sufficient protection to the underlying free central placenta, making a completely inferior ovary redundant. Moreover, this cap is nectariferous at the periphery. Other nectariferous ericalean groups (e.g., Vaccinioideae; P. Caris, personal observations) in which an evolution towards inferior ovaries can be observed possess a nectary disc, which evolved from the gynoecial base (in case of superior ovaries) towards the top of the gynoecium (in case of inferior ovaries), surrounding the style. In primuloid families obvious nectary discs are absent and nectar is often released by nectarostomata on the flanks of the ovary. The development of the gynoecial cap could be a solution to deal with the problem of nectar presentation, when a transition towards (semi-)inferiority occurs. A similar nectary disc is present in the only other perigynous genus of the former Primulales, namely Maesa (Caris et al., 2000
), which has been placed at the base of the primuloid clade by recent analyses (Källersjö et al., 2000
) as well. Hence, a basal position of the genus Samolus could be justified when considering this character. Mez (1902)
agreed with Bartling (1830)
to put Maesa in a separate group together with Samolus, which should be better lifted out of Primulaceae. Nevertheless, he pointed to some major differences between both genera, namely the woody habit of Maesa, the presence of bracteoles, and the different fruit type (Mez, 1902
). Apart from these, Maesa has a lower number of ovules, which are partly embedded in the placenta and separated from each other by placental tissue (Caris et al., 2000
).
When removing the calyx of Samolus for our observations, we noticed that the wall of the ovary was torn off together with the sepals. The connection with the gynoecial cap is lost, being an additional argument in favor of our observations that two separate regions are responsible for the building of the gynoecium. In fact, we removed the gynoecial hypanthium on which the short calyx tube and the free parts of the sepals are inserted. The calyx is persistent on the developing fruit, while the corolla, together with the staminodes and the stamens, is shed. The fruit is an ovoid or globose capsule, opening by five valves. These valves could be interpreted as reminiscent to five carpels, although no individual carpel primordia have been observed in the ontogeny, neither by us nor by Payer (1857)
. This pentamerosity of the gynoecium has never been observed in Theophrastaceae, in which the fruit is a berry or a drupe, instead of a capsule.
In Samolus, like in Theophrastaceae, ovules are bitegmic and tenuinucellate, anatropous or tending to campylotropous, and not embedded in the placenta; they occur in several whorls on the placenta.
Although Samolus may share some morphological characters with Primulaceae, these might as well be homoplasious. Because the separation of primuloid genera on the basis of their distribution and habit (i.e., tropical, woody Theophrastaceae and Myrsinaceae on the one hand and essentially temperate, herbaceous Primulaceae on the other hand) has proven to be artificial, the herbaceous habit of Samolus cannot be used to keep the genus in Primulaceae. Apart from that, the flower size and the thin corolla texture of Samolus, typical of Primulaceae, may be linked to the short flowering period, which in turn might be linked to the more temperate distribution (cf. Källersjö et al., 2000
). The semi-inferiority of Samolus and therefore the different gynoecium development separates the genus from all other Primulaceae. The resulting fruit, which is in both cases a capsule, opening by five valves, might have evolved independently in both lineages.
When Samolus is compared to Theophrastaceae, the presence of staminodes is a notable character that is easily considered to be a morphological synapomorphy of the Samolus- Theophrastaceae clade. However, many differences can be found between both lineages. First of all, there is the different habit: although the distinction between woody and herbaceous species appears to be artificial, the woody habit and neotropical distribution still characterize the Theophrastaceae sensu stricto (s.s.). Next, common primordia are obviously present in Samolus, whereas in Theophrastaceae, petals and stamens develop from separate primordia. The development of the inner whorls on a secondarily formed floral apex in Theophrastaceae, and the resulting pseudo-synsepaly, was not observed in Samolus. The latter, however, has a semi-inferior ovary and a particular gynoecium development, which differs completely from Theophrastaceae. The fruit in Samolus is a many-seeded capsule and the mature seeds are, unlike seeds of Theophrastaceae, rather small and angular, as in Primulaceae. Secretory cavities are absent from the Theophrastaceae, but they are found in the roots and vegetative parts of Samolus species (Anderberg and Ståhl, 1995
). The typical indumentum of Theophrastaceae, with immersed glandular hairs on different parts of the flower, is lacking in Samolus. Apart from the presence of (possibly nonhomologous) staminodes and the similar placenta and ovules, little floral ontogenetic evidence can be found to support a close relationship between the genus Samolus and the Theophrastaceae. On the other hand, Theophrastaceae s.s. constitute a very well-supported monophyletic group, and it may be clear that several synapomorphies can be found to support this group.
So it seems that no unambiguous morphological synapomorphies can be found that justify a position of Samolus in either Primulaceae or Theophrastaceae. Källersjö et al. (2000)
correctly stated that including Samolus in one of these families may be considered morphologically aberrant, but they nevertheless argue for a position within Theophrastaceae.
Samolus has always taken a rather isolated position within Primulaceae, mainly due to its perigynous flowers and the presence of staminodes. Based on the semi-inferior ovary, Rafinesque (1820 fide Reveal, 2003
) established the Samolaceae, but this family name has only rarely been adopted (Reveal, 2003
).
Actually, the genus can be separated from Myrsinaceae and Primulaceae on the basis of molecular data as well: Myrsinaceae and Primulaceae, with exception of Maesa and Samolus, show two deletions in the ndhF gene, whereas Samolus is characterized by an insertion in this gene (Källersjö et al., 2000
).
Furthermore there might be a molecular synapomorphy for the Theophrastaceae s.s. as well, since all genera that have been examined so far show a three base-pair insertion in the ndhF gene (Källersjö et al., 2000
). Sequence data of the chloroplast gene rbcL prove the monophyly of the Theophrastaceae s.s., but generic interrelationships in the family remain unresolved (Källersjö et al., 2000
).
Anderberg and Ståhl (1995)
carried out a cladistic analysis of Primulales, based on morphological characters. In their initial, unweighted analysis, Samolus appears to be the sister to all other Primulaceae, but after successive weighting it got a position higher up in the cladogram, near the genera of the Primuleae; Theophrastaceae appear to form a monophyletic group. In a more recent analysis of Primulales, based on rbcL sequences (Anderberg et al., 1998
), Theophrastaceae remain monophyletic, but both Myrsinaceae and Primulaceae become paraphyletic; Samolus has a basal position (jackknife = 0.55) in the Myrsinaceae-Primulaceae clade (excluding Maesa). A drupe appears to have evolved in woody members of the group, a transformation that can be observed in other plant families as well (Anderberg et al., 1998
).
In the combined molecular and morphological analyses of Källersjö et al. (2000)
, Theophrastaceae s.s. are supported at 100%. Theophrasta is sister to a group formed by Clavija and Jacquinia. The relationship between Theophrastaceae and Samolus is only 60%. A combined analysis of three chloroplast genes without morphological data results in a 90% sister group relationship (Källersjö et al., 2000
).
So it appears that, depending on the data analyzed, the position of Samolus differs. The genus is found to be the sister taxon of the Myrsinaceae-Primulaceae complex, of Theophrastaceae s.s., or it stands in a trichotomy with these two groups. Support is, generally speaking, low to very low, except in a combined analysis of ndhF, rbcL, and atpB (Källersjö et al., 2000
). Samolus shares several morphological characters with Primulaceae. As a result, morphological data reduce support for a Samolus-Theophrastaceae relationship, although in some combined analyses a sister group relationship can be observed.
Källersjö et al. (2000
, p. 1340) mention that "actual close evolutionary relationships will be obscured by differences in aspect." They report that morphological characters may show a wide range of variation caused by evolution in response to selective pressure. To a certain extent, this might be the case, but nonetheless, evolutionary relationships are recognizable in ontogenetic characters and developmental studies certainly have proven to be useful in recognizing homologies. Variation in floral morphology in Ericales sensu lato (s.l.) is considerable and seems to be very useful for classification at high taxonomic levels (cf. Caris et al., 2000
, 2002
). It seems to us that morphology is indispensable in defining synapomorphies and increasing the utility of phylogenetic classifications.
Topical results that suggest a close relationship of Samolus and Theophrastaceae s.s. are not (yet) robust and the monophyly of Theophrastaceae s.l. is not at all well supported. For instance, the results of ITS sequence analysis carried out by Martins et al. (2003)
are not in agreement with the topology as suggested by Källersjö et al. (2000)
and confirmed by Anderberg et al. (2002)
. Hitherto, no unambiguous morphological characters can be found to consign Samolus to either Primulaceae or Theophrastaceae. If additional molecular results confirm the position of Samolus at the basis of the Theophrastaceae clade, it will be better to raise the genus to family level, as was also suggested by Källersjö and Ståhl (2003)
. Then, the highly supported monophyly of Theophrastaceae s.s. is guaranteed, and moreover, there will be morphological support to distinguish between the resulting five primuloid families.
|
FOOTNOTES |
|---|
2 E-mail: pieter.caris{at}bio.kuleuven.ac.be
|
LITERATURE CITED |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIAL AND METHODSRESULTSDISCUSSIONLITERATURE CITED |
|---|
