| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
2 Laboratory of Systematics, Botanical Institute, Katholieke Universiteit Leuven, Kardinaal Mercierlaan 92, B-3001 Leuven, Belgium; and 3 Bolus Herbarium, University of Cape Town, 7701 Rondebosch, South Africa
Received for publication October 6, 1999. Accepted for publication January 4, 2000.
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONLITERATURE CITED |
|---|
Key Words: Cornales • Escalloniaceae • floral anatomy • floral ontogeny • Montiniaceae • placentation • Saxifragaceae • Solanales • sympetaly
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONLITERATURE CITED |
|---|
). Male inflorescences are described as cymose with several terminal flowers, whereas pistillate flowers are described as terminal and solitary (Takhtajan, 1997
). Two or three genera are currently placed in Montiniaceae: Montinia, Grevea, and Kaliphora (Dahlgren, 1980
; Thorne, 1992
; Skowno, 1996
; Takhtajan, 1997
). Kaliphora, a monotypic Madagascan genus formerly placed in Cornaceae, was included in Montiniaceae by Capuron (1969)
. Thorne (1992)
included a fourth genus, Melanophylla, in Montiniaceae. Takhtajan (1997)
removed Kaliphora as Kaliphoraceae and Melanophylla as Melanophyllaceae. Grevea (with two species) and the monotypic Montinia share a similar wood anatomy (Carlquist, 1989
; Rakouth, 1989
), vegetative anatomy (Ramamonjiarisoa, 1980, cited in Al-Shammary and Gornall, 1994
) and pollen (Milne-Redhead, 1955
; Pastre and Pons, 1973
; Hideux and Ferguson, 1976
). Similar pollen was also reported for Kaliphora by Hideux and Ferguson (1976)
.
The systematic position of Montiniaceae is far from settled. De Candolle (1828)
placed Montinia in the family Onagraceae (Myrtales); this was followed by Bentham and Hooker (1867)
. Later, Montinia was most commonly allied with members of Saxifragaceae sensu lato (subfamily Montinioideae: Engler, 1930
; Schulze-Menz, 1964
), Grossulariaceae (Cronquist, 1981
; Mabberley, 1987
), or Escalloniaceae (Hutchinson, 1973
). The family Montiniaceae was erected by Nakai in 1943 (see Appendix). Carlquist (1989)
found that similarities of wood anatomy of Montiniaceae with Myrtales (viz. Thymelaeceae, although this family is associated in Malvales by most recent systems: e.g., APG, 1998
) were greater than with Rosales. Dahlgren (1975)
first suggested that Montiniaceae are related with Celastrales, although with uncertainty, but later shifted the family to Cornales (Dahlgren, 1980
). An affinity with Cornales was maintained by Thorne (1992)
, while the family was placed in the closely related Hydrangeales by Takhtajan (1997)
. Montinia is reported to have unitegmic ovules (Schulze-Menz, 1964
), but the author does not state where he obtained this information. Unitegmic, tenuinucellate ovules are characteristic of Escalloniaceae and Hydrangeaceae, which differ in this from saxifragalean or rosalean genera. Serological evidence (Grund and Jensen, 1981
), storied wood, and presence of iridoid compounds (Dahlgren, Jensen, and Nielsen, 1977
; Dahlgren, Rosendal-Jensen, and Nielsen, 1981
) support a cornalean affinity of these families and an asterid rather than a myrtalean link (Dahlgren, 1983
).
Recent molecular studies (Chase et al., 1993
; Morgan and Soltis, 1993
; Soltis and Soltis, 1997
; APG, 1998
) indicate that Montiniaceae belong to the asterids. Phylogenetic hypotheses based on cpDNA break up the original Cornales of Dahlgren into several orders belonging to different supergroups. The closest relatives of Montinia, based on rbcL and 18S rDNA data, are Convolvulaceae and Solanaceae, both placed in Solanales (Euasterids I), far away from the Cornales (basal in the asterids). Interestingly, in the study of Cosner, Jansen, and Lammers (1994)
and Fay et al. (1998)
, Montinia appears as sister to Sphenoclea (Sphenocleaceae) and Hydrolea (Hydrophyllaceae), thus forming a sister clade to the Solanales. However, the argumentation for a solanad affinity appears meager besides the molecular data, as Morgan and Soltis (1993)
enumerate more differences than similarities with members of the Solanales.
We investigated the floral ontogeny of pistillate and staminate flowers of Montinia caryophyllacea to question the systematic position of the family. The absence of a corolla tube in Montinia especially demands clarification, given the systematic importance attached to it (e.g., Erbar, 1991
; Erbar and Leins, 1996
). The systematic importance of the floral anatomy has been repeatedly demonstrated by several authors (e.g., Eyde, 1967
; Bensel and Palser, 1975a, b
; Armstrong, 1986
; Ronse Decraene and Smets, 1991, 1999a, b
; Ronse Decraene, De Laet, and Smets, 1998
). A limited morphological study of mature flowers of Montinia has been made by Skowno (1996)
. We studied the floral anatomy of staminate and pistillate flowers of Montinia, in addition to floral ontogeny.
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONLITERATURE CITED |
|---|
). Voucher material is kept at the botanical institutes of the Katholieke Universiteit Leuven (LV) and the University of Cape Town (BOL).
Material was fixed in FAA (85 mL ethanol 70%, 10 mL acetic acid, 5 mL formaldehyde 40%). The buds were transferred to 70% ethanol and dissected under a Wild M3 dissecting microscope. The material was washed repeatedly in 70% ethanol and dehydrated by putting the buds in a 1:1 mixture ethanol-dimethoxymethan (DMM or formaldehyde-dimethylacetal) for 5 min and in pure DMM for 20 min. Buds were critical-point dried using liquid CO2 in the CPD 030 (Balzers, Liechtenstein). The dried material was mounted on aluminium stubs using Leit-C (after Göcke) or double tape and coated with 
180 nm of gold (Spi-moduleTM Sputter coater of Spi Supplies, West Chester, Pennsylvania, USA) before observation with the Scaning Electron Microscope (SEM.) For light microscopy, preanthetic buds were analyzed and customary methods of preparation were used. The material was run through an alcohol as well as an alcohol-tertiary butyl alcohol series and was next embedded in paraffin. Serial sections, 8–11 µm thick, were stained with safranin and counterstained with fast green or cotton blue using the automatic staining machine Varistain 24–3 (Shandon, Runcorn, UK). Camera-lucida drawings and photographs were made under a Leitz Dialux 20 equipped with a Wild MPS 45/51 photoautomat.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONLITERATURE CITED |
|---|
Organogeny
Unless otherwise stated, male and female inflorescences will be considered together. Inflorescences arise terminally. They consist of a single terminal flower, with a few associated flowers arising basipetally on lateral branches. In staminate inflorescences, a number of lateral flowers are associated with the top flower and arise basipetally and spirally in the axil of a bract (Figs. 4, 6, 26). A smaller lateral flower may be visible within a secondary bract (Fig. 4, arrow), but generally there is no further branching and no bracteoles. Lower down the axis lateral branches may be formed with a similar development. The inflorescence is thus basically monotelic and paniculate (sensu Weberling, 1989
). In pistillate plants there are similarly top flowers, but the inflorescence differs from the staminate inflorescence in that the top flower stands alone, except for some isolated cases of a lower flower in the axil of a bract. More generally a number of spirally inserted lateral branches are formed basipetally, each ending with another flower. Bracts subtending the lateral branches become progressively leaflike from top to bottom of the main inflorescence. On each lateral branch a variable number of smaller second-order bracts arise in a spiral sequence approaching a decussate pattern (Figs. 1, 2, 3, 9, 16). The base of the leaves and larger bracts are swollen and contain a large number of multicellular uniseriate hairs on the inside. A first pair of bracts emerges sequentially in transverse position (Fig. 2). While they curve over the growth apex of the lateral shoot, unicellular trichomes start to grow on their upper surface. Another pair of bracts arises higher up in the median region but is obliquely inserted (Figs. 1, 2, 16). A third pair may be initiated in a latero-transverse position, and finally a last one in latero-median position (Figs. 1, 2). In several cases this number was not attained (e.g., Figs. 3, 16), and in an extreme case only the first two bracts were initiated (Fig. 5). As a result it was sometimes difficult to decide at what moment initiation of the flower really starts. The orientation of the terminal flower on the lateral branches relative to the main axis is not orthogonal, but clearly diagonal (Fig. 16). The former case would occur when the last formed bracts are more or less in median position. No true bracteoles are produced. Four sepals emerge sequentially but with a variable initiation sequence. Often the lateral sepals emerge first and successively (Figs. 4, 6, 8, 9), followed by the abaxial median sepal and finally the adaxial median. Very often the initiation of the adaxial sepal is retarded, and this leads to an asymmetric shape of the receptacle and an irregular initiation of the upper floral organs (Figs. 10, 14, 20). In other cases a regular tetramerous arrangement was rapidly attained by the equal development of the floral bud. In some cases growth of the adaxial sepal precedes that of the abaxial (Fig. 5), or one of the lateral sepals is conspicuously smaller than the others (Fig. 14). In other cases the median sepals emerge before the laterals (Figs. 5, 7, 12). The sepals are inserted on the radii relative to the main axis in case there is only one bract subtending the flower, as in staminate inflorescences (Figs. 4, 6, 8). Sepals grow rapidly. The difference in size of the sepals remains visible during later stages of development, and the outer sepals are inserted lower and externally of the median sepals (Figs. 19, 20); aestivation is imbricate. At anther formation the sepals are exceeded in size by the petals and are pushed away from each other. Finally sepals remain visible as small triangular lobes arranged side by side on the periphery of the flower. During later stages of development the sepal bases become lifted by zonal growth linked with the development of epigyny. Sepal lobes are inserted on a shallow rim (Figs. 38, 40) and bear numerous stomata adaxially.
|
|
|
|
|
|
). At maturity the connective is inconspicuous. The anther surface and upper part of the filament consist of papillate cells. No orbicules were found within the anther. Pollen is tricolporate with Pole/Equator index 1, size 32 µm (Fig. 46); the ektexine is reticulate with the surface covered with small spinules (not visible in Fig. 46). Gynoecium initiation starts as a shallow depression surrounded by the stamens at the time they start differentiating anther tissue. The upper portion of the cavity region develops as a girdling primordium (Fig. 21), and finally two horseshoe-shaped carpel primordia are delimited (Figs. 22, 23). The carpel primordia are situated opposite two staminodes, either in latero-median or latero-transversal position.
Differences between staminate and pistillate flowers start late in ontogeny. In staminate flowers carpellary primordia soon stop growing and are rapidly overtaken by the tetrasporangiate anthers (Fig. 27). As the surrounding tissue keeps growing the carpellary lobes are soon absorbed in the receptacle and appear at maturity as two (Fig. 41) or occasionally a single (Fig. 40) shallow crescent-shaped emergence that is fully embedded in disc tissue. In some cases gynoecia are completely resorbed.
In pistillate flowers the style results from upward extension of the early circumferential growth at the top of the gynoecial cavity (Fig. 23). The style forms two distinct centers of growth which extend upwards between the anthers. The basal confluent zone of the style remains very short, and most extension growth occurs within the lobed region, which shows a deep median invagination (Figs. 24, 25, 29). The upper part of each carpel curves outwards and differentiates stigmatic papillae along the slit (Figs. 30, 36). The papillae grow into long stigmatic hairs, which cover the extended U-shaped stigmatic area (Figs. 37, 38). In the upper stigmatic region the two carpels appear thus as four because the styles are deeply split (Figs. 25, 38, 53, N). Two short compressed styles are thus formed with a narrow central canal (Figs. 29, 30). The styles sit on a massive disc covering the inferior ovary. The limits between style and disc are indistinct. The basal part of the style is covered with long rectangular cells, which become smaller and papillate towards the disc. Both style and disc are interspersed with anisocytic deeply sunken stomata and have a deeply striate cuticle (Figs. 38, 39). In all probability this disc functions as a nectary, although we did not test for nectar secretion. Tissue adjacent to the disc surface is darkly staining in section (Fig. 51) and supplied with numerous veins (Figs. 51, 53, 54). In staminate flowers a similar papillate epidermis occurs on the receptacle, also covering the sterile carpel remnants. However, no stomata were found on the disc.
|
Floral anatomy
Pistillate flowers
The pedicel contains a continuous vascular cylinder, which is similar in staminate flowers (Fig. 54A). Just below the base of the flower the vascular tissue breaks apart in several bundles, while a number of traces are given off tangentially and converge to the central pith (Fig. 53A). The bundles rearrange at the periphery in broad masses of vascular tissue. About 8–9 peripheral bundles were counted, but some are highly dissected; two bundles are conspicuously bigger and are formed by the convergence of tissue that has been separated to the center at lower level (Fig. 53B–C). From the latter bundles small traces split off and converge to the narrow ledge separating the two locules (Fig. 53C–D). These bundles, which we name the ventrals, converge to each other at the level of the common placental lobe. They branch profusely and eventually fuse (Fig. 53D–E). At this level the peripheral bundles have split in a higher number of traces running through the ovary wall. Towards the level of ovule insertion the ventrals, or derivatives, branch profusely and provide a horizontally running supply to each ovule (Figs. 47, 52, 53E–F). In longisection, superposed tiers of vascular supply can be seen, corresponding to the supply of the ovules (Fig. 47). At the level of ovule supply the placenta is almost rectangular. The peripheral bundles remain unchanged during most of their course through the ovary wall. Above the level of insertion of the last ovule the placentae separate while the ventral supply continues upwards in each placental lobe (Fig. 53G, H). At this level the peripheral bundles rearrange in two girdles of traces that will be connected to the organs of the flower. More centrally two dorsal traces can be seen, positioned perpendicular to the ventrals; these can eventually be followed down the ovary wall (Fig. 53H). The peripheral traces consist of large petal traces and intervening stamen traces with sepal median traces. At a lower level, the stamen traces and sepal median traces are fused. In between one finds several smaller traces, which could be confused with deposits of tannins but are mostly the supply to the disc and the lateral supply of the sepals (Fig. 53H–I). The lateral sepal supply is either derived from the sepal medians, from the petals, or from independent bundles. The placental lobes diminish in size higher up, becoming shallow emergences before disappearing from sight (Fig. 53I–J). At that level the ventral supply stops short. Next to each of the dorsal traces two marginal traces appear between the smaller disc traces (Fig. 53J). The peripheral sepal median traces give off two lateral traces and at this level the stamen trace is well separated. At a higher level just below the style the central stylar canal is surrounded by a girdle of smaller traces together with the dorsals and accompanying marginals. Slightly higher the dorsals and marginals converge to the slit, while the disc traces spread out below the disc surface (Fig. 53J). The distance between staminode traces, sepal medians, and sepal laterals increases. At the level of separation of the style the dorsal trace ends and the marginal traces continue up the style (Fig. 53K). Slightly higher the style is split apart in two horseshoe-shaped appendages containing two marginal traces each (Fig. 53L–M). At this level stigmatic papillae are visible abaxially. Further up four stigmatic triangles can be seen opposite the petal lobes; each contains one of the marginal traces (Fig. 53N). Within the sepal ring several traces can be seen; higher up only the median remains visible before fading out (Fig. 53K–L). A single trace runs into the petal and splits into three traces and higher up into several traces (Fig. 53J–N).
|
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONLITERATURE CITED |
|---|
) is incorrect, since the inflorescence consists of a terminal branch surrounded by a number of partial inflorescences. Each lateral branch, (including the top flower on the main branch) ends with a single flower. A number of decussately inserted bracts precedes flower initiation (Figs. 1–3). These bracts enclose a flower only exceptionally (Fig. 4). No true bracteoles (prophylls) are present in Montinia. It could be hypothesized that the ancestral inflorescence was a system of cymes with all flowers on the branches reduced, except for the upper one. This characteristic arrangement is partly retained in staminate inflorescences. Similar cases of sterilization, but with the eventual incorporation of the upper bracts as an epicalyx are found in Dipsacales (Hofmann and Göttmann, 1990
; Roels and Smets, 1996
). For example, Patrinia (Valerianaceae) and genera of the Linnaeeae (Caprifoliaceae) have flowers subtended by three pairs of decussately arranged phyllomes.
Both pistillate and staminate flowers possess evidence of the other sex. However, the process of sterilization has gone further in staminate flowers: the gynoecium is limited to small carpellodes with their dorsal traces. There are no traces of placental or ovular tissue. This implies that the onset of sterilization always occurs at the same developmental level, viz. the initiation of carpel primordia. Pistillate flowers have stamens developing until the stage of anther formation prior to meiosis. The trend to dioecy also involves changes in other characters, such as inflorescence development (see above) and pollinator attraction. The trend to unisexuality implies the existence of a progressive process in this case, not a sudden mutation leading to highly different morphotypes. At a certain stage of development genes become arrested in their expression and an organ fails to attain maturity. This may indicate that unisexuality is a phylogenetically recent development in Montinia. Similarly, the acquisition of dioecy occurs at different rates in pistillate and staminate flowers of Rhus hirta (Anacardiaceae) (Gallant, Kemp, and Lacroix, 1998
), although early developmental stages are similar between both sexes. Once a male or female reproductive organ has been sterilized, it either aborts completely in a next evolutionary step or it gains a new function. In the latter case sex determination occurs before the initiation of the flowers, when there are no traces of the other sex. This has been illustrated for Carica papaya (Ronse Decraene and Smets, 1999b
) where the pistillode has the function of a nectary in staminate flowers.
We often observed that the floral apex becomes distorted by the unequal development of the sepals. This leads to a retardation of organ initiation, unidirectional inception of primordia, and eventually to the loss of primordia. Flowers were occasionally trimerous in petals and stamens (Figs. 10, 18). This distortion happened almost exclusively on lateral flowers, suggesting that pressures of bracts are important. Alternatively, loss of organs could also be caused by insufficient nutrient allocation in lateral flowers, which arise at a later stage. Lack of space on the irregular floral apex is related with the loss of some organs. One (often broader) petal stands opposite a sepal (Fig. 18), which suggests that it has become derived by fusion. A similar variation in development has been observed in Chrysosplenium (Saxifragaceae) by Ronse Decraene et al. (1998)
. This case illustrates the potential for the derivation of trimerous flowers from tetramerous forms (reviewed in Ronse Decraene and Smets, 1994
). Interestingly, staminate flowers of Grevea are described as trimerous in the corolla and androecium (Hutchinson, 1967
).
The systematic relationships of Montinia
Montinia, Grevea, and Kaliphora make up the family Montiniaceae. In the introduction we already pointed to several shared characters. However, no floral materials of Grevea and Kaliphora were available for comparison. Descriptions by Baillon (1884)
and Hutchinson (1967)
of Grevea demonstrate important additional similarities in the absence of stipules, the existence of tufts of hairs at the base of the leaves, dioecy with similar pattern of distribution of sexual organs in the unisexual flowers, and subextrorse anthers. The same distinction of cymose staminate flowers and solitary pistillate flowers is described as for Montinia. Grevea differs in the fewer (4–5) erect ovules on parietal (Baillon, 1884;
Engler, 1930
) or axile (Hutchinson, 1967
) placentae. Kaliphora shares unisexual tetramerous flowers (not dioecious) and two recurved styles. However, contrary to Montinia staminodes are absent in pistillate flowers, and there is a solitary ovule in each locule (Hutchinson, 1967
). Other differences are enumerated by Takhtajan (1997)
.
A fourth genus, Melanophylla, placed in Montiniaceae by Thorne (1992)
and sharing a similar pollen (Hideux and Ferguson, 1976
), differs in its bisexual flowers, glandular trichomes, presence of bracteoles, obscure or absent nectary, and single pendulous ovule in each locule (Hutchinson, 1967
; Takhtajan, 1997
).
Trichome anatomy of Montinia has been investigated by Al-Shammary and Gornall (1994). The spaghetti-like trichomatic mass consists of uniseriate eglandular hairs. Flowers do not possess any trichomes. Contrary to their report that trichomes occur on the petiole of the leaf and on the nodes only, we found trichomes on the margin and top of the leaf, and, occasionally, scattered along the main vein on the upper surface. We also found that the trichomes are occasionally bi- to multiseriate, especially at the edges of leaves (Fig. 43). Grevea has the same kind of trichomes scattered all over the leaf surface. No information is available on Kaliphora. Al-Shammary and Gornall (1994)
pointed to an affinity of Vahlia with Montinia, based on similar uniseriate eglandular trichomes, the opposite, exstipulate leaves, the one-trace unilacunar nodes, a bicarpellate ovary with tenuinucellate ovules, and the presence of iridoids. Vahlia appears to be in the same group of asterid families that contains Montinia (Morgan and Soltis, 1993
). However, Vahlia differs in possessing bitegmic ovules (Takhtajan, 1997
). A floral ontogenetic investigation of the monospecific Vahliaceae would be worthwhile. Due to lack of data we will not include Vahlia in our discussion.
Several potential relatives for Montinia have been proposed in the past (see introduction). In the following we will analyze the potential relationships of Montinia with Solanaceae, Sphenocleaceae, Saxifragaceae, Cornaceae, Hydrangeaceae, and Escalloniaceae. A comparison of characters is summarized in Table 1. The study of floral ontogeny of putatively related taxa of Montiniaceae is instructive and can throw light on the relationships of Montinia by the presence of shared synapomorphies. A comparison with the development of other tetramerous flowers can especially give evidence of a comparable initiation sequence. Floral ontogeny of taxa with inferior ovaries originally grouped in a polymorphic Saxifragaceae sensu lato has been carried out in a number of species (e.g., Ribes—Payer, 1857
; Hydrangeaceae—Roels, Ronse Decraene, and Smets, 1997
; Escallonia—Payer, 1857
; Itea—Vandeputte, 1993
; Saxifragaceae—Klopfer, 1973
, Gelius, 1967
; Ronse Decraene et al., 1998
).
|
; Klopfer, 1973
, Ronse Decraene et al., 1998
), including the slow growth (even absence) of the petal primordia, stamens arising on a ring primordium, and the gynoecium, which is not initiated along a depression, but with two free primordia arising on a flattened apex. No circular gynoecial primordium occurs and the ovary never closes completely, leaving a transverse slit separating the two styles (also visible anatomically). Some anatomical similarities with Montinia exist within the core Saxifragaceae (Bensel and Palser, 1975a
) in the presence of compound ventral bundles, separated from the peripheral bundles in the base of the flower, and the arrangement of peripheral bundles in the ovary wall. However, contrary to Montinia, ventrals continue into the style, the placentae often have a longitudinal division on top of the ovary, and the nectary has no vascularization. More differences are listed in Table 1.
Solanales
Morgan and Soltis (1993)
argued that few morphological and anatomical features are shared by Montinia and Solanales and that there are, indeed, several differences (viz. free stamens, choripetaly, inferior ovaries, lack of internal phloem and iridoid production). However, to accommodate Montinia in the Solanales, Chase et al. (1993)
suggested that the sympetalous corolla of Montinia has been secondarily lost. There are several floral ontogenetic investigations on Solanaceae (reviewed by Huber, 1980
). Most studies are limited to one or a few species, except for Huber (1980)
who investigated the ontogeny of 20 species in the family. Solanaceae differ ontogenetically from Montiniaceae in a number of striking features: flowers are always pentamerous with sepals usually arising in a 2/5 sequence and soon becoming connected by interprimordial growth; petals arise (almost) simultaneously and lag behind in development compared to the androecium; all species of Solanaceae are characterized by zonal growth lifting the androecium and petals; the gynoecium emerges as a ring primordium on a flattened floral apex. Later development of the gynoecium is characterized by the development of two horseshoe-shaped carpel primordia separated by a median "bar"-like septum. This is caused by an equal growth of the placental region and the ovary wall. Floral anatomy is highly variable in Solanaceae (e.g., Murray, 1945
; Armstrong, 1986
). However, all taxa lack a cavity in the style and there is a tendency for free ventrals to become fused.
An affinity of Montinia with Sphenoclea (Sphenocleaeae) and Hydrolea (Hydrophyllaceae) has been proposed on the basis of rbcL sequences (Cosner, Jansen, and Lammers, 1994
; Fay et al., 1998
) and confirmed by ndhF sequences (R. G. Olmstead, University of Washington, personal communication). Sphenoclea has traditionally been related with Campanulales (e.g., Cronquist, 1981
; Takhtajan, 1997
) and its relationship with Solanales appears surprising. However, a cladistic study of morphological and chemical characters by Gustafsson and Bremer (1995)
suggests that the Sphenocleaceae occupy a weakly supported basal position in the Campanulaceae. A floral developmental investigation by Erbar (1995)
does not contradict an affinity of Sphenoclea with Campanulales, but shows that Sphenoclea is different from Montinia in a number of features, such as the initiation of an early ring primordium at the onset of petal initiation and the presence of sympetaly. However, it shares an imbricate corolla aestivation and semi-inferior ovary with Montinia; the placentation is axile (apparently axile, but basically parietal in Montinia) with narrow septal connections. Floral anatomical data would be a welcome addition for comparison.
Secondary loss of sympetaly, a necessary step if Montinia is related to Solanales, is rare in the Asteridae. Apart from Rubiaceae where it has been reported in Mastixiodendron and some other genera (Darwin, 1977
), it has been illustrated ontogenetically for Besseya (Scrophulariaceae) by Hufford (1995)
. In Besseya the presence of a corolla tube depends on the expression of zonal growth below the insertion of corolla lobes and filament insertion. However, contrary to Montinia, the corolla is initiated on a "corolla plateau" or ring primordium in Besseya and the development of a tube depends on the later growth of this plateau. Erbar (1991)
and Erbar and Leins (1996)
attached great importance to the early initiation of sympetalous corollas in distinguishing between "early" and late "sympetaly." In early sympetaly "the petals arise on a ring primordium or are connected already at initiation" (Erbar and Leins, 1996
, p. 428); in "late sympetaly" the petals arise separately, and become connected by interprimordial extensions (Erbar and Leins, 1996
, p. 428). In other words the corolla is expressed as the independent initiation of petal primordia (late sympetaly), or the corolla develops as a ring that lacks, at least initially, independent primordia (early sympetaly). Solanales are characterized by late sympetaly; the petal primordia are small and become connected by a lateral connection of the petal bases. Interestingly, this division between early and late sympetaly corresponds roughly with the growth pattern of the corolla: in early sympetaly petals grow rapidly and soon cover the stamen primordia; in late sympetaly petals lag behind stamen development. Moreover, it appears clear that those species termed "early sympetalous" all have a ring primordium at the stage of petal initiation, viz. a collar delimiting the concave floral apex (cf. Roels, 1998
). One should also remark that the stamen primordia arise on the inner slopes or inside the ring primordium (in the case of early sympetaly), while the petal primordia are situated peripherally or on top of it. The limits between a corolla tube proper and the stamen-petal tube common to sympetalous taxa is difficult to discern, viz., is the interprimordial region behind the stamen part of the petal tube sensu stricto or part of the basal meristem common to stamens and petals? This distinction is not made by Hufford (1995)
who describes growth below the corolla lobes and stamens, as well as the confluent growth above the stamen insertion, as "zonal growth." His description is essentially developmental without taking issues of homology into account. We believe that neither the approach of Leins and Erbar of a basic corolla-derived nature of the ring primordium, nor a generalization of a "corolla tube"without distinguishing between its parts is wholly satisfactory. As clearly set out by Nishino (1978)
and Ritterbusch (1991)
, the initial ring primordium ultimately gives rise to both petal primordia and stamen primordia. The corolla tube sensu stricto arises later on top of this ring by lateral extension and linking of marginal meristems of the originally free petal primordia (e.g., Nishino, 1983a, b
), while the ring primordium should merely be interpreted as the early initiation of the stamen-corolla tube or even an early onset of hypanthial growth. While Erbar and Leins (1996
, p. 428) explicitly state that there are two different parts, viz. a stamen-corolla tube and a corolla tube s.s. (sensu stricto), they restrict their ontogenetic observations of early and late sympetaly to the corolla tube s.s., without considering the functioning of the former, merely stating that it "results from the activity of a circular intercalary diffuse meristem under the insertion area of stamens and corolla tube s.str." However, in Montinia this kind of development is absent, and there are no indications that it had a corolla tube in ancestral forms.
Cornales
The broad circumscription of Cornales by R. Dahlgren (1980)
and G. Dahlgren (1989)
includes several families and is based mainly on embryological and phytochemical characteristics, of which the iridoids are a main feature. Besides their presence in the Cornales, iridoids occur also in taxa currently grouped as asterids (Jensen, Nielsen, and Dahlgren, 1975
). In the asterids the distribution of iridoids is linked with the distribution of tenuinucellate and unitegmic ovules and coincides with that of sympetaly. However, Montinia occupies an awkward position in being strictly choripetalous as are most other Cornales. This is also a first unequivocal evidence for unitegmic ovules in Montinia. Similarly, Loasaceae have the same embryological features (Dahlgren, Rosendal-Jensen, and Nielsen, 1981
), but sympetaly is present in several species. Early petal initiation is correlated with the development of circumspherical growth of a ring in Eucnide. Development of a common stamen-petal tube is linked with vertical zonal growth, but fusion of corolla lobes in suprastaminal regions is weakly expressed (Hufford, 1988
). Montinia shares epigyny with various members of Cornales. The occasional occurrence of perigyny in Hydrangeaceae was interpreted as derived from ancestors with complete epigyny (Soltis, Xiang and Hufford, 1995
). Data on the ontogeny of members of the Cornaceae are limited to a few species. Cornus alternifolia and two species of Cornus (C. kousa, C. sanguinea) were investigated to some extent by Payer (1857)
and Roels (1998)
, respectively. As in Montinia, corolla aestivation is imbricate in bud; no ring primordium occurs prior to petal initiation, and petal growth is rapid. No ring primordium develops below the stamen primordia. However, the development of the ovary differs strongly from that of Montinia. All taxa of the Cornales s.s. studied show broad petal primordia on the flanks of a depression, with lateral extensions. Data from floral anatomy (Wilkinson, 1944
; Eyde, 1967, 1988
) indicate that Cornaceae differ from Montinia in several characteristics. Eyde (1967, 1988
) attached much importance to the absence of central vascular traces to delimit Cornaceae from other genera with doubtful position. The placentation in Montinia is basically parietal, but the axial vascular system is strongly developed.
Escalloniaceae and Hydrangeaceae have been connected by some authors (e.g., Escalloniales—Krach, 1977
; Hydrangeales—Takhtajan, 1997
). Montinia shares a scanty vasicentric axial parenchyma with some genera of Hydrangeaceae (Carlquist, 1989
) and some species of Escallonia (Stern, 1974
). A number of ontogenetic similarities exist between Montinia and the Hydrangeaceae (see, e.g., Klopfer, 1973
; Roels, Ronse Decraene, and Smets, 1997
; Roels, 1998
): the formation of a central pitlike depression prior to the initiation of a gynoecial ring primordium, the rapid growth of the petals, the initiation of massive placental bands prior to ovule initiation, the short bifid style (cf. Deutzia), and identical placentation fluctuating from axile below to parietal higher up (fusion of septa in lower part). Early ontogeny of tetramerous Philadelphus resembles Montinia also in the sequential initiation of the two upper bracts or bracteoles with oblique orientation of the floral bud (Roels, Ronse Decraene, and Smets, 1997
). Hydrangeaceae also possess massive (but U-shaped) placentae, a complex composed of cymes as partial inflorescences (e.g., Kirengeshoma, Philadelphus), and rapid initiation of petals. Transfer of protection of flower bud from calyx to corolla is common, contrary to the situation in Solanales.
Philipson (1967)
and Eyde (1966)
report the existence of "axial" strands supplying the ovules in Escalloniaceae and Corokia, respectively; these axial bundles are connected proximally with one or more peripheral bundles and are absent from Cornaceae (Wilkinson, 1944
; Eyde, 1966
). It is interesting that the ventral vascular system is constructed similarly in Montinia. Other similarities of Montinia with Corokia are the fact that lateral traces of the sepals are occasionally derived from the petal bundle and the existence of a short region above the ovules in which the carpel margins are not united into a septum. Flowers of Corokia and Montinia contain tannin cells, which are conspicuously absent from Cornales s.s. However, Corokia does not appear to be related with either Escalloniaceae or Montiniaceae on the basis of rbcL data (Morgan and Soltis, 1993
; Cosner, Jansen, and Lammers, 1994
), or morphological and chemical data (Gustafsson and Bremer, 1995
).
The Escalloniaceae are a poorly known, probably unnatural family containing several tribes with a high intrinsic morphological variability (e.g., Krach, 1976
; Hideux and Ferguson, 1976
). Molecular studies (Soltis and Soltis, 1997
, Xiang and Soltis, 1996
) place Escalloniaceae far apart from Hydrangeaceae and Montiniaceae. Analyses of Escalloniaceae suggest that the family is polyphyletic.
Contrary to Solanaceae and Hydrangeaceae, Escalloniaceae have not been studied ontogenetically, except for the study of Payer (1857)
on Escallonia floribunda. Data presented by Payer mostly accord with those of Montinia: inflorescences are cymose and terminal; there is a rapid growth of free petals with imbricate aestivation (however, descending); and gynoecium development shows numerous similarities. The gynoecium arises as a ring primordium around a central receptacular concavity; there is a similar development of continuous carpel primordia, each ending in a free dorsal carpel part. Escallonia possesses a similar strongly developed nectary with nectarostomata surrounding a stout style, and a number of ovules ranging between 2 and 12. The disc on top of the ovary is vascularized by small branches derived from carpel-wall bundles. The placentation in Escalloniaceae is intruding parietal. As in Montinia, the separation of the placental lobes is perpendicular to the carpels (Bensel and Palser, 1975b
: Polyosma, Escallonia). In Montinia placentation is pseudo-axillary to nearly the top and is connected with the ovary wall by a narrow connection. The placenta itself is massive and bears two lateral rows of interconnected ovules. The placenta breaks apart in the upper half from the intrusion of a broad stigmatic canal. This breaking-up is different from Saxifragaceae (e.g., Bensel and Palser, 1975a
; Ronse Decraene et al., 1998
), as it does not go along the suture of the two carpels. Escalloniaceae and Montiniaceae differ also from Saxifragaceae in that no ventral carpel bundles enter the style (they do so in Saxifragoideae). Escalloniaceae mostly possess unicellular, eglandular trichomes (Escallonia, Forgesia, and Quintinia also possess glandular hairs), but shapes and surface morphology are variable. Interestingly, Grevea shares similar eglandular, uniseriate trichomes with Anopterus (tribe Anoptereae: Al-Shammary and Gornal, 1994
). From their phenetic analysis Hideux and Ferguson (1976)
conclude that the pollen of Montiniaceae corresponds most closely with that of Escalloniaceae.
Montinia shows striking similarities with Polyosma (Escalloniaceae) at the level of floral anatomy. Polyosma differs from other Escalloniaceae in a number of anatomical characters (see Bensel and Palser, 1975b
), which are also shared by Montinia: tetramerous flowers, deeply intruding parietal placentae (postgenitally fused in Montinia except for the upper part), an interconnection between peripheral bundles and central tracheary elements in the pith, compound ventrals running well above the ovule insertion, two placentae with two rows of ovules each, petal traces providing extreme laterals of the sepals and carpel wall bundles, dorsal traces derived from the sepal-plane traces, and a canal extending through the style into the stigma. However, in Polyosma leaves are opposite and fruit development is very different. Polyosma also differs from Escalloniaceae and Montiniaceae in its seed anatomy (Krach, 1976
: presence of starch in the endosperm and small undifferentiated embryo) and pollen morphology (Pastre and Pons, 1973
; Hideux and Ferguson, 1976
). However, similarities with Montiniaceae include a three-layered seed coat and a very small embryo.
Conclusions
Neither ontogenetical nor anatomical evidence presented in this paper (see also Table 1) provides strong evidence for the phylogenetic relationships of Montinia, but the closest affinity of Montinia lies probably with some members of Escalloniaceae. There is almost no morphological support for a sister-group relationship with Solanales as suggested by molecular data. However, the aim of this study was not to compare Montinia with outgroups cladistically, as information on these potential outgroups is highly insufficient. An asterid affinity of both Montiniaceae and Escalloniaceae, as suggested by molecular data, is corroborated in this study. On the one hand, more research on the floral ontogeny of the Cornales sensu Dahlgren (1989)
and of the Escalloniaceae, in particular, is needed, and on the other, more information of a wide array of characters of putative related Hydrolea and Sphenoclea might give a clearer view about a possible relationship with basal members of the Solanales. Also, the questionable affinities of Vahlia should be addressed on the basis of a floral ontogenetic investigation.
|
FOOTNOTES |
|---|
4 Author for reprint requests (e-mail: louis.ronsedecraene{at}bio.kuleuven.ac.be
).
|
LITERATURE CITED |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONLITERATURE CITED |
|---|
APG (Angiosperm Phylogeny Group). 1998 An ordinal classification for the families of flowering plants. Annals of the Missouri Botanical Garden 85: 531–553[CrossRef][ISI]
Armstrong, J. E. 1986 Comparative floral anatomy of Solanaceae: a preliminary survey. In W. G. D'Arcy [ed.], Solanaceae: biology and systematics 9, 101–113. Columbia University Press, New York, New York, USA
Baillon, H. 1884 Un nouveau type aberrant, de Madagascar. Bulletin de la Société Linnéenne de Paris 1, 1: 420
Bensel, C. R., and B. F. Palser. 1975a Floral anatomy in the Saxifragaceae sensu lato. II. Saxifragoideae and Iteoideae. American Journal of Botany 62: 661–675[CrossRef][ISI]<
