| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Structure and Development |
2Royal Botanic Gardens, Kew, Richmond, Surrey, TW9 3AB, UK; 3Department of Botany, Natural History Museum, Cromwell Road, London, SW7 5BD, UK
Received for publication April 16, 2002. Accepted for publication July 12, 2002.
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMATERIALS AND METHODSOBSERVATIONSDISCUSSIONLITERATURE CITED |
|---|
Key Words: Alliaceae • Asparagales • CYCLOIDEA • FIDDLEHEAD • flowers • fluctuating asymmetry • Gilliesia • monocotyledons • monosymmetry • pseudocopulation • septal nectary • solid style • transmitting tissue • zygomorphy
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSOBSERVATIONSDISCUSSIONLITERATURE CITED |
|---|
, using molecular data from rbcL, demonstrated that the six American genera formerly placed in a tribe Brodiaeeae of Alliaceae (e.g., Dahlgren, Clifford, and Yeo, 1985
) actually constitute a separate family, for which they resurrected the name Themidaceae. This family has closer affinities to other asparagalean families, including Hyacinthaceae and Aphyllanthaceae (Fay and Chase, 1996
; Fay et al., 2000
; Pires et al., 2001
). Fay and Chase (1996)
and Fay et al. (2000)
further demonstrated that Alliaceae and Amaryllidaceae (59 genera) are closely related families. However, the placement of Agapanthus (nine species; often included in Alliaceae) remains unresolved within this assemblage, occurring either as sister to Alliaceae or as sister to both familes; Agapanthus was therefore retained as a separate family, Agapanthaceae.
|
; Traub, 1963
; Fay and Chase, 1996
; Rahn, 1998
) or even in a separate family, Gilliesiaceae (Traub, 1982
; Ravenna, 2000
). In this paper we follow the classification of Alliaceae into three subfamilies (Allioideae, Gilliesioideae, and Tulbaghioideae), as proposed by Fay and Chase (1996)
. The subfamily Gilliesioideae sensu lato (s.l.) includes all the South American genera of Alliaceae. The representatives of tribe Gilliesieae (Gilliesia and Gethyum) that were included in the molecular analyses (Fay and Chase, 1996
; Fay, Qamaruz Zaman, and Chase, 1997
) were a sister pair, showing relatively little sequence divergence from each other. However, they differ widely in other respects, including floral morphology and chromosome number. One species of Gilliesia reportedly has an alliaceous odor (Traub, 1982
; Rahn, 1998
), but this observation was contradicted by Ravenna (2000)
and was not observed here. Most other Alliaceae have more or less actinomorphic flowers with six stamens and six tepals, but Gilliesieae are remarkable for their zygomorphic flowers exhibiting several unusual features (Engler, 1889
; Reiche, 1893
). Although little is known about the pollinators of Gilliesia, many authors, including Ravenna (2000
, p. 13), have observed that the flower of Gilliesia "much resembles the body of an insect" and suggested that some small butterfly or wasp may be attracted by this similarity. If so, this would make Gilliesia unique among non-orchid monocots in possessing a pollination mechanism associated with mimicry and putatively also with deceit (cf. van der Pijl and Dodson, 1966
), though there may be other examples among eudicots; for example, Kullenberg (1961)
reported pseudocopulation of flowers of Guiera senegalensis (Combretaceae) by a wasp. In Gilliesia, the homologies of the floral structures that have been compared with insect legs, abdomen, and wing cases require exploration.
More detailed examination and evaluation of these data are timely in the context of recent novel ideas about family relationships in higher asparagoids (Fay et al., 2000
). Furthermore, interest in zygomorphy has recently been rekindled by molecular-developmental studies, which demonstrated that (at least in asterid eudicots) zygomorphy is dependent on expression of the single-copy nuclear CYCLOIDEA and DICHOTOMA genes, resulting in abortion of the adaxial stamen(s) (e.g., Coen, 1996
; Cronk and Möller, 1997
; Cubas, Vincent, and Coen, 1999
; Luo et al., 1999
; Citerne, Möller, and Cronk, 2000
; Theissen, 2000
; Endress, 2001
). A similar situation may well characterize zygomorphic monocots, though this has not yet been tested (for review, see Rudall and Bateman, 2002
). This paper therefore has two primary aims: (1) to examine the floral structure of Alliaceae in relation to the systematics of the family and (2) to analyze the morphology and anatomy of the zygomorphic flowers of Gilliesia in the broader context of the evolution of floral symmetry in monocots.
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSOBSERVATIONSDISCUSSIONLITERATURE CITED |
|---|
Agapanthaceae: Agapanthus praecox Willd. subsp. orientalis, HK 1960–5804.
Alliaceae: Allium caesium Schrenk, HK 1987–795; A. cernuum Roth, HK 1978–4208; A. roseum Linn., HK 1982–1231; A. rotundum Linn., HK 1996–1635; A. rubrovittatum Boiss. and Heldr., HK 1972–10301; A. scorodoprasum L. subsp. rotundum (L.) W.T. Stearn, HK 1976–1405; A. sphaerocephalum L., HK 1975–5510; A. subhirsutum L., HK 1981–1020; A. textile A. Nelson and Macbride, HK 1978–3706; A. tolmiei Baker ex Coult. var. platyphyllum (Tidestr.) Ownbey, HK 1957–30010; Gethyum atropurpureum R.A. Phil. (sometimes called Solaria atropurpurea [Phil.] Ravenna), HK 1988–1970; Gilliesia graminea Lindl., K (40636): Wygnauki s.n., 1978, Chile; HK 1977–2342 (originally collected by Dr. J. Alvarez, Chile), HK 1977–1670 (originally collected by Dr. J. Alvarez, Chile); Ipheion uniflorum Rafin., HK 1981–2444; Nothoscordum andicolum Kunth, HK 1986–34; N. gaudichaudianum Kunth, HK 1979–2616; Trichlora sp., K (15205): A.L. Johnson s.n., 1933, Peru; Tristagma sp., HK 1977–674; Tulbaghia ludwigiana Harv., HK 1969–5052; T. violacea Harv., HK 1946–26601.
Amaryllidaceae: Cyrtanthus parviflorus Baker, HK 1978–3088; Galanthus plicatus Bieb. subsp. byzantinus (Baker) D.A. Webb, HK 1978–3174; Eucharis korsakoffii Traub., HK 1981–2831.
Themidaceae: Brodiaea stellaris S. Wats., HK 1988–1329; Triteleia hyacinthina Greene, HK 1960–67031.
Flowers were fixed in formalin-acetic alcohol (FAA) and stored in 70% ethanol. For sectioning, complete flowers or ovaries were embedded in Paraplast (Sigma, St. Louis, Missouri, USA) using standard methods of wax embedding and were serially sectioned using a rotary microtome. Sections were stained in safranin and Alcian blue, dehydrated through an alcohol series to 100% ethanol and then Histoclear (National Diagnostics, Atlanta, Georgia, USA), mounted in Euparal (Asco Laboratories, Manchester, UK), and examined using a Leitz Dialux 20 photomicroscope (Leitz, Oberkochen, Germany). For scanning electron microscope (SEM) examination, fixed flowers and buds were carefully dissected in 90% ethanol, then dehydrated in an absolute ethanol : acetone series (90% ethanol, 30 min; absolute ethanol, 30 min; absolute ethanol : acetone in proportions 50 : 50, 10 min; and finally two steps of acetone, 10 min each). Dehydrated material was then critical-point-dried using a Balzer CPD 020 (Balzer Union, Furstentum, Liechtenstein), mounted onto SEM stubs using double-sided sellotape, coated with gold using an Emscope SC 500 sputter coater (Emscope, Ashford, UK), and examined using either a Cambridge 240 SEM (Cambridge Instruments, Cambridge, UK) (Fig. 2B, C, G) at 15 kV or an Hitachi cold-field emission SEM S-4700-II (Hitachi Co., Tokyo, Japan) at 2–5 kV. A few flowers were examined fresh and uncoated at 1 kV (Fig. 2A, H).
|
|
OBSERVATIONS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSOBSERVATIONSDISCUSSIONLITERATURE CITED |
|---|
). In petaloid monocots, both perianth whorls are normally petaloid and are therefore normally collectively termed tepals (outer and inner), except in orchids, in which the terms petal and sepal are more commonly used. In most Alliaceae, the tepals are often more or less basally connate, sometimes entirely fused into a floral tube, which may be fleshy (e.g., in Tulbaghia ludwigiana, Fig. 5A), or sometimes free (e.g., in Gilliesia).
|
; Muñoz and Moreira, 2000
). In Gethyum (Fig. 1D) and Solaria (the latter illustrated by Muñoz and Moreira, 2000
), although both normally reported as actinomorphic, the outer median (abaxial) tepal is distinctly larger than the other tepals (and therefore labellum-like), and the inner median (adaxial) tepal is often smaller than the two inner laterals. In all other Alliaceae the perianth is more or less radially symmetric.
|
|
In addition to fluctuating asymmetry, another remarkable aspect of the flower of G. graminea is the presence of two different types of appendage that are largely responsible for establishing insect mimicry: (1) two thick structures that extend abaxially from the staminal tube and resemble insect wing-cases surrounding an abdomen (appendage type 1, Fig. 2A, B, E, G) and (2) several long, narrow appendages that surround the staminal tube but are derived from the tepal bases and resemble insect legs (appendage type 2, Figs. 1A–B, 2A–C, E–F). The leg-like appendages are present all round the staminal column, but are longer on the abaxial side; in the Wygnauki material, the leg-like appendages on the adaxial side are longer than those of the two Kew accessions. The two inner lateral tepals may represent wings, and two of the three anthers may represent eyes. Both "legs" and "wing-cases" have a relatively thick epidermis with domed or densely papillate cells (Fig. 2A), in some cases with relatively dark contents, possibly indicating that they are osmophores, though no odor was detected. Since there is no associated vasculature in either of the two types of appendage, a nectar-secreting role appears unlikely. These appendages are outgrowths of the tepal bases and/or basal margins; during earlier stages in floral ontogeny their tepaline origin is clearly visible (Fig. 2E–G). In older flowers at anthesis, parts of the "legs" of G. graminea become fused to the staminal tube (not to the free upper parts of the tepals) (Fig. 4E).
|
; Muñoz and Moreira, 2000
); their illustrations suggest that the flower is less insect-like in these species.
Such appendages are also present in Miersia (Table 2), in which they are often deeply bifid or trifid (Ravenna, 2000
). Similar unvascularized papillate appendages, but much shorter (and unbranched), are also present in Gethyum atropurpureum (contrary to the report by Ravenna, 2000
, but in agreement with the observations of Muñoz, 2000
), in the form of small (free) lobes extending from the bases of the inner tepals at their junction with the staminal tube, sometimes also present at the bases of the outer tepals (Fig. 3D); this position is consistent with a tepaline (rather than staminal) origin.
|
). Filament tubes are otherwise not common in Alliaceae (e.g., Rahn, 1998
), though they are a feature of some members of the sister family Amaryllidaceae (Amaryllidae–Amaryllidinae), in which they often form a "paracorona" (Arber, 1937
; Meerow and Snijman, 1998
). In some other Alliaceae (e.g., Nothoscordum andicolum, N. gaudichaudianum) all six stamens are adnate to the perianth tube, either basally or entirely; for example, in Tulbaghia the filaments are entirely adnate, so that the anthers are sessile on the perianth tube (Fig. 5A). In some species of Allium the three inner stamens (a1, a2, a3), often have (at least basally) flattened filaments, sometimes with appendages (e.g., in A. rotundum, Fig. 3B). In some species (e.g., A. cernuum), the filaments are basally connate and only the three flattened filaments are adnate, resulting in three cups that presumably collect fluid. In Tristagma and Tulbaghia violacea, the filament bases are partially fused to the gynoecium, also forming cups.
Gynoecium
The ovary is tricarpellate, syncarpous, and trilocular in all Alliaceae; these conditions characterize most other higher asparagoids, including Amaryllidaceae, Agapanthus, and Themidaceae. There are generally several ovules per carpel (or only one or two in Allium) and placentation is axile. In Gilliesia the lower (proximal) part of the ovary, although trilocular, shows slight bilateral symmetry (Fig. 4F).
Septal nectaries are present in the upper part of the ovary (Figs. 5C, 6B, E–F), opening as slits around the base of the single style in all Alliaceae examined except Gilliesia and Gethyum, in which septal nectaries are absent (Figs. 4, 5D–F). In Allium, in which the style is basifixed (see below), the septal slits open toward the outside, rather than into the well around the base of the basifixed style (Fig. 6E). Daumann (1970)
previously examined the structure and diversity of septal nectaries in Allium and other Alliaceae, and Ashurmetov and Yengalycheva (1997)
reported variation in the shape of the septal nectary canals within the genus Allium, particular shapes often being characteristic for individual species.
|
).
|
) the style is sunken into the apex of the ovary; that is, the fused stylar regions of the carpels are inserted laterally on the ovary regions of the carpels at anthesis (Fig. 6A, C); this arrangement is termed basifixed or gynobasic and is derived by lobing of the top of the ovary that commences at the same stage as the style starts to elongate (Fig. 3C). A gynobasic style is not present in other Alliaceae examined here (e.g., Ipheion, Fig. 6B), except for a slightly sunken style in Nothoscordum andicolum, confirming observations made on Nothoscordum by Di Fulvio (1973)
. Many species of Allium have a ridged ovary (Fig. 6E), and in some species (e.g., A. scorodoprasum) the margins of the surface ridges are fused to form a well located immediately below the septal nectary openings. |
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSOBSERVATIONSDISCUSSIONLITERATURE CITED |
|---|
) (Fig. 7A–E) is a consistent synapomorphy for Alliaceae s.s., i.e., including Tulbaghia and all genera listed in Table 1, but excluding some former Alliaceae that are now placed in Themidaceae, which have uniformly hollow styles (Fig. 7F). Anderson (1941)
also reported solid styles in Tulbaghia, Nothoscordum, and Allium (and hollow styles in Brodiaea), and Johri (1966)
reported solid styles in Tulbaghia and Nothoscordum (Alliaceae). Both authors also confirmed that hollow styles occur in Amaryllidaceae, but neither examined any other Alliaceae or the phylogenetically contentious genus Agapanthus. This character therefore supports the continued separation of Agapanthus from Alliaceae, because Agapanthus has hollow styles, in common with Amaryllidaceae and all other higher asparagoids, and indeed with most other syncarpous monocots.
Solid styles are generally regarded as characteristic of syncarpous eudicots (Eames, 1961
) and are rare in monocots, with the exception of Alliaceae, some Orchidaceae (e.g., Apostasia and Neuwiedia: Kocyan and Endress, 2001
; Rudall and Bateman, 2002
) and many grasses (e.g., Triticum: Li and You, 1991
); Arber (1934)
referred to the transmitting tissue of grasses as the "stylar core." In hollow styles the transmitting tissue is a single, normally relatively distinct, layer of epidermal cells; pollen tubes grow from the stigma to the ovary along the surface of the canal, normally through a thin layer of mucilage. In solid styles the transmitting tissue is more extensive (cf. Endress, 1994
), and the epidermis has become intimately fused between adjacent carpels; pollen tubes normally grow between the cells of the transmitting tissue, as in Petunia (Sassen, 1974
), or sometimes through the cell walls, as in Gossypium (Jensen and Fisher, 1970
). Lennon et al. (1998)
demonstrated that in solid styles of the eudicot Arabidopsis, pollen tubes adhere to, and travel along, an extracellular matrix. The transmitting tissue of solid styles includes an intercellular substance containing pectin, comparable with the mucilage found in the canal of hollow styles (Labarca, Kroh, and Loewus, 1970
).
Orchidaceae and Alliaceae are not closely related within Asparagales, and Poaceae belong to an entirely different order within the monocots (e.g., Chase et al., 2000
); the presence of solid styles is therefore homoplastic among these families. Presence of solid styles may be a homoplastic mutation, perhaps related to a homolog of the FIDDLEHEAD gene that was identified in Arabidopsis as responsible for regulating contact-mediated epidermal cell adhesion; this gene encodes a protein that is thought to be involved in the synthesis of long-chain lipids that modify the properties of the cuticle (Lolle, Cheung, and Sussex, 1992
; Yephremov et al., 1999
; Sinha, 2000
). The adaptive significance of a solid style is not immediately obvious, since Orchidaceae, Poaceae, and Alliaceae have little in common in terms of reproductive biology (indeed, Poaceae are wind-pollinated). However, in the context of the evolution of syncarpy, Endress (1994)
suggested that a major advantage may be the possibility of an even distribution of pollen tubes among the carpels. This is perhaps a significant benefit in orchids, in which an exceptionally large number of ovules (typically 104–106) is borne on branching placentas; a broad region of transmitting tissue would also facilitate simultaneous growth of a large number of pollen tubes from the single pollen mass (pollinium) that is the basal unit of pollen transfer in most orchids (e.g., Dressler, 1993
). Another possible advantage may be more rapid chemical signalling between stigma and placenta; in many orchids ovule development is retarded at anthesis but triggered by pollination (Dressler, 1993
). Furthermore, apomixis, in some cases dependent on pollination (but not fertilization), is common in both grasses (e.g., Young, Sherwood, and Bashaw, 1979
; Kellogg, 1987
) and Allium, in which antipodal embryos are common (e.g., Kojima, Nagato, and Hinata, 1991
; Kojima and Nagato, 1992
). Apomictic development that is triggered by pollination must rely on rapid movement of chemical signals through a cellular matrix.
Gynobasic style
A sunken (gynobasic) style is formed late in development by lobing of the top of the ovary (Fig. 3C); this is apparently a synapomorphy for the genus Allium, also reported by Di Fulvio (1973)
. This condition, which results in a cup around the base of the style, is unusual for Asparagales, and indeed for other syncarpous monocots, though a gynobasic style is also present in Walleria (Sterling, 1974
) and Cyanastrum (Brummitt et al., 1998
). Both genera are members of the lower asparagoid family Tecophilaeaceae and therefore are not closely related to Alliaceae; in both families this condition is apparently correlated with a substantial reduction in ovule number. There is also a tendency among other Alliaceae to develop (in all the floral organs) excrescences and partial fusion, often resulting in formation of (possibly nectar-collecting) cups. Schaeppi (1939)
described the diversity of filament appendages in Allium and other monocots and discussed their relationship with the corona in Amaryllidaceae. Pistrick, Kruse, and Adler (2001)
reported postgenital fusion of stamen bases and presence of filament appendages in some species of Allium. A similar range of variation in floral structure occurs in the higher asparagoid family Themidaceae, which was formerly included in Alliaceae but was recently segregated on the basis of molecular phylogenetic analysis (Fay and Chase, 1996
). In Themidaceae, filaments are also often adnate and/or connate, often flattened or winged, and the perianth tube is often partially fused, presumably postgenitally, to the gynoecium (Pires et al., 2001
).
Tenuinucellate ovules
The tenuinucellate condition (lacking parietal cells in the developing ovule) is also a consistent synapomorphy for Alliaceae (Stenar, 1932
), though Ashurmetov and Yengalycheva (1997)
reported a "mediocrinucellate" type for Allium, somehow intermediate between the crassinucellate and tenuinucellate types. Tenuinucellate ovules are otherwise rare in the higher asparagoid clade, the only other example being a group of genera within Ruscaceae s.l. (Rudall, Conran, and Chase, 2000
). All other higher asparagoids have crassinucellate ovules (i.e., possess parietal tissue derived from the archespore), including Agapanthus (Stenar, 1933
), Amaryllidaceae, and Themidaceae (Berg, 1978
, 1996
).
Septal nectaries
Ravenna (2000)
suggested that, in addition to the insect-like flower, nectar derived from septal nectaries may be a further attraction to insect pollinators in Gilliesia. However, on the contrary, this investigation demonstrates that septal nectaries are entirely absent from Gilliesia and Gethyum, in contrast to all other Alliaceae. This observation therefore supports a close relationship between Gilliesia and Gethyum. Among related families, septal nectaries are present in Agapanthus (Fig. 5B) and also in Themidaceae, in which they are often relatively extensive (Berg, 1996
; and P. J. Rudall, personal observations). On the other hand, in Amaryllidaceae, which have epigynous flowers, septal nectaries are either entirely absent (e.g., Cyrtanthus) or short supralocular septal nectaries are present around the style base (e.g., Eucharis). Absence of septal nectaries (i.e., complete fusion of carpel margins) occurs only sporadically in Asparagales, in contrast to Liliales, in which absence of septal nectaries and presence of perigonial nectaries represent highly consistent synapomorphies for the order (Rudall et al., 2000
). Septal nectaries are sometimes absent from epigynous Asparagales (e.g., some Iridaceae: Rudall, 2002
; Rudall, Manning, and Goldblatt, in press
), though some other epigynous Asparagales have prominent septal nectaries that produce copious nectar (e.g., the bird-pollinated Doryanthes). They are always absent from "hyper-epigynous" taxa that possess gynostemia (Rudall and Bateman, 2002
).
In Gilliesia and Gethyum there is no evidence of nectar production from another part of the flower (such as the tepals), in contrast to many Liliales, some Iridaceae, and some Orchidaceae. Most nectariferous orchids confine the nectar to at least one tepaline spur. Absence of nectar often indicates an alternative pollination mode; for example, other asparagoids that wholly lack nectaries include several families with vibratile (buzz) pollination and Solanum-type flowers, including some Laxmanniaceae (e.g., Arthropodium), some Hemerocallidaceae, some Tecophilaeaceae, and the orchid Apostasia (Vogel, 1990
; Dressler, 1993
; Bernhardt, 1996
; Kocyan and Endress, 2001
; Rudall, 2001
, 2002
). However, at least in G. graminea, the remarkable appendages that resemble insect legs are putative osmophores (see below).
Mimicry and deceit in Gilliesia graminea
In G. graminea the possession of presumed insect mimicry, presence of putative osmophores, and absence of nectar together may indicate a deceitful pollination mechanism similar to that of some Orchidaceae (cf. van der Pijl and Dodson, 1966
). Little is known about pollinators in Gilliesieae, which mostly flower in early spring (August to October: Muñoz, 2000
; Muñoz and Moreira, 2000
), though Zöllner and Arrigada (1998)
commented that they often flower in late winter when few insects are active. Ravenna (2000)
suggested that the insect-like flower of Gilliesia may attract insect pollinators and highlighted the need for studies of pollination biology to test this hypothesis.
Ravenna (2000)
, in analogizing the Gilliesia flower with an insect body, equated the "abdomen" with the fused staminal ring, the "legs" with digitate appendages of the staminal column, "wings" with "winged excrescences," and the eyes with anthers. This investigation underlines the remarkable insect-like appearance of the flower of G. graminea (Fig. 1A, B), but demonstrates that although the "abdomen/wing cases" are indeed associated with the base of the staminal column, the "legs" are derived from outgrowths of the tepal bases and/or margins and are homologous with similar tepal-derived structures in other Gilliesieae, such as Gethyum (Fig. 3D). Furthermore, despite the lack of an obvious odor, these novel floral structures may function as osmophores; both "legs" and "abdomen" have a relatively thick, domed or papillate epidermis with densely cytoplasmic contents, both characteristic features of osmophores (Vogel, 1990
). Since there is no associated vasculature in either of the two types of appendage, a nectariferous role seems unlikely; there is no evidence of the presence of nectar, and nectaries are normally supplied by abundant vasculature (e.g., Fahn, 1979
). Many orchids have perianth-derived osmophores that sometimes secrete pheromone-like substances. For example, in Platanthera bifolia (the Lesser Butterfly-orchid) the epidermis of the labellum secretes a nocturnal scent, and in several Ophrys species, which are known to produce odors that are attractive to insects, the osmophores on the labellum consist of areas of dome-shaped, papillate, dark-staining epidermal cells (Vogel, 1990
), similar to those of G. graminea. Some Ophrys species (e.g., O. tenthredinifera, O. fuciflora aggregate) have a glandular appendix at the anterior (lower) margin of the labellum, which may be an osmophore (Vogel, 1990
). In many species of Disa the entire labellum functions as an osmophore and is often highly dissected, as in the appendages of Gilliesia montana (illustrated by Muñoz, 2000
) and Miersia (Ravenna, 2000
). Osmophores have not previously been reported in Alliaceae, but in the sister family Amaryllidaceae, flowers of Narcissus are known to emit pollinator-specific volatiles that are probably derived from the colorful corona (Vogel, 1990
; Dobson et al., 1997
). Interestingly, the different corona types of Amaryllidaceae, which are derived from the stamens in the tribe Pancratieae and from the tepals in the tribe Coronatae (Arber, 1937
), also represent novel floral stuctures, perhaps with a similar function to those of Gilliesieae; the corona of some other Alliaceae such as Leucocoryne merits further exploration in this context.
If, as seems likely, insects are attracted to flowers of G. graminea by both visual and olfactory signals, the absence of a nectar reward indicates that the flower may be sexually deceptive. This could make Gilliesia unique among non-orchid monocots, since it is widely believed that pollination by sexual deceit is exclusive to Orchidaceae (e.g., Nilsson, 1992
; Ayasse et al., 2000
). Admittedly, G. graminea is not obligately insect pollinated; it sets seed by self-pollination (and/or apomixis) fairly readily in a winter (February) greenhouse at Kew (Tony Hall, Royal Botanic Gardens, Kew, personal communication, 2002). However, some sexually deceitful orchids (e.g., Ophrys apifera) readily self-pollinate in the field (e.g., Kullenberg, 1961
). Within Orchidaceae, a sexually deceitful strategy appears to have evolved several times, even discounting some groups in which pseudocopulation is suspected but not confirmed, such as the Malaysian epidendroid Porphyroglottis (van der Pijl and Dodson, 1966
). For example, in Australia, the orchidoid Cryptostylis and its allies are sexually pollinated by ichneumon wasps, Calochilus by flower wasps, Caleana by sawflies, and Leporella by ants (van der Pijl and Dodson, 1966
; Dafni and Bernhardt, 1989
; Jones, 1993
; Peakall and Beattie, 1996
; Kores et al., 2001
). The South American epidendroid Telipogon and its allies, in which the large flowers mimic, and are pollinated by, flies of the family Tachidinae (e.g., Dodson and Escobar, 1987
), apparently lack odor, despite the presence of osmophore-like hairy appendages on the column. Among European orchidoids in the genus Ophrys, males of the pollinator species (usually a bee) are attracted to the flower by both chemical and visual cues (Kullenberg, 1961
; Schiestl et al., 1999
; Ayasse et al., 2000
). The similarity of these cues to both the morphology and sex pheromones of the female insects causes the male insect to attempt to copulate with the labellum of the flower; during this process pollinia become attached to its body, and pollination occurs when the insect visits a different flower on the same (or ideally a different) inflorescence.
In addition to the striking similarities between Gilliesia graminea and sexually deceitful orchids, there are several structural differences. For example, in G. graminea pollen is shed as separate grains, rather than as the cohesive pollinia of sexually deceptive orchids, and the landing platform role of the labellum in some orchids (e.g., Ophrys) is replaced by the "abdominal" appendages derived from outgrowths of the tepal margins in G. graminea. Indeed, in all Gilliesieae the largest tepal is invariably the median outer tepal, in contrast to orchids, in which the median inner petal normally forms the labellum. Furthermore, in G. graminea (as in all Alliaceae) the ovary is hypogynous, rather than epigynous as in orchids, and there is no resupination as in most orchids, or fusion of the stamens to the gynoecium as in all orchids (hyper-epigyny sensu Rudall and Bateman, 2002
), although Gilliesieae are unusual for Alliaceae in possessing a staminal tube, which closely connects the stamens with the style.
Floral zygomorphy in Gilliesieae
Although several monocots possess a bilaterally symmetric corolla associated with organ displacement and floral curvature (termed "positional monosymmetry" by Endress, 1999
), floral zygomorphy relating to major structural differences in both perianth and stamens is relatively infrequent in lilioid monocots other than Orchidaceae (Asparagales). Other lilioid monocots that show zygomorphy encompassing more than one floral whorl include Corsia (Corsiaceae–Liliales: Rudall and Eastman, 2002
), Diplarrhena (Iridaceae–Asparagales: Rudall and Goldblatt, 2001
), and Cyanella (Tecophilaeaceae–Asparagales: Simpson and Rudall, 1998
). Commelinid monocots that show this feature include most Zingiberales (e.g., Zingiber) and many Commelinales (e.g., Philydrum). Orchids represent arguably the strongest expression of zygomorphy among lilioid monocots, since in orchids zygomorphy encompasses all three floral whorls and includes both labellum formation and adaxial stamen suppression (reviewed by Rudall and Bateman, 2002
).
Endress (1999
, 2001
) regarded the occurrence of zygomorphy in certain major plant groups as a key innovation that prompted subsequent adaptive radiations. He also postulated that in some large groups with zygomorphic flowers, such as orchids, zygomorphy is maintained by a high degree of organ fusion. In Orchidaceae, zygomorphy and synorganization (hyper-epigyny: fusion of androecium and gynoecium to form a gynostemium) each probably evolved only once, with later modifications resulting in suppression of different stamens in different subfamilies (Rudall and Bateman, 2002
). Zygomorphy and synorganization were merely prerequisites for a combination of subsidiary factors that prompted the extraordinary species richness of the subfamilies Orchidoideae and Epidendroideae, including aggregation of pollen into cohesive pollinia and the development of often nectariferous tepaline spurs. However, none of these factors applies in Alliaceae, in which zygomorphy is not associated with either species richness, complex organ fusion, pollen aggregation, or even (in Gilliesieae) nectar production.
With the exception of Gethyum and Gilliesia, all Alliaceae examined here have six (3 + 3) stamens. Gethyum and Gilliesia have only three (abaxial) stamens (A1, a1, a2), as in the apostasioid orchid Neuwiedia (e.g., Kocyan and Endress, 2001
; Rudall and Bateman, 2002
), but the perianth of Gethyum is only slightly bilaterally symmetric, whereas Gilliesia graminea shows bilateral symmetry in at least four floral whorls: perianth (suppression of the inner adaxial tepal), inner and outer androecial whorls (suppression of three adaxial stamens), and gynoecium (slight bilateral symmetry). In Gilliesia graminea there are no obvious staminodes (except minute vestigial projections), contrary to some reports (e.g., Rahn, 1998
) that may have mistaken the osmophores for staminodes; furthermore, there is no vestigial vasculature for either the three missing adaxial stamens (A2, A3, a3) or the missing median inner tepal. There is an interesting parallel between floral structure of the four south American Gilliesieae (Gethyum, Gilliesia, Miersia, and Solaria: Table 2) and the mainly Southern Hemisphere family Tecophilaeaceae (eight genera: Asparagales), in which flowers of four genera show varying degrees of androecial zygomorphy, from stamens divided into posterior and anterior groups (in Cyanella and Odontostomum) to two or three adaxial stamens being reduced to staminodes (in Zephyra and Tecophilaea) (Simpson and Rudall, 1998
; Rudall, in press
).
Degree of bilateral symmetry may be at least partly correlated with the pseudanthial nature of the alliaceous inflorescence; that is, with a condensed, flower-like presentation (for review of pseudanthia see Classen Bockhoff, 1990
). Floral zygomorphy is common in florets of pseudanthia, and the alliaceous umbel, in which individual flower buds are tightly compressed between the enclosing bracts during development (Figs. 2J, 5G), may be preadaptive for evolution of zygomorphy. Position of flowers in the inflorescence is known to play a role in floral symmetry (e.g., Coen, 1991
, 1996
). Degree of differential development often varies across the umbel, as in pseudanthia of many eudicots (e.g., Asteraceae), causing weak "differential zygomorphy." Flowers of some Alliaceae (e.g., Allium caesium, A. rubrovittatum, A. textile) show slight bilateral symmetry that indicates a degree of differential growth. Pistrick, Kruse, and Adler (2001)
illustrated several different early developmental stages of flowers of some Allium species in which three adjacent stamen primordia are slightly but distinctly larger than the opposing set of three stamens. There is thus ostensibly a transformation series from weak bilateral symmetry reflecting only differential growth in some Allium species (i.e., differential zygomorphy) to strong bilateral symmetry reflecting stamen suppression and perianth modification in Gilliesieae (i.e., structural zygomorphy). Experimental verification of evolutionary–developmental pathways may ultimately be possible if the underlying controls are identified at the genomic level. Gillies et al. (2002)
postulated that capitulum development in Asteraceae is dependent on the expression of a CYCLOIDEA homolog in the production of zygomorphic ray florets. However, thus far CYCLOIDEA has not been isolated from any monocot, though there are parallels between the teosinte branched1 (tb1) locus in teosinte (Zea mays subsp. parviglumis) and the CYCLOIDEA gene of Antirrhinum; both belong to the TCP gene family (Doebley, Stec, and Hubbard, 1997
; Lawton-Rauh, Alvarez-Buylla, and Purugganan, 2000
).
Although the term zygomorphy represents a complex suite of characters, one test of character evolution would be to optimize the proposed transformations onto the most parsimonious phylogenetic tree, in this case resulting from the most recent combined molecular phylogenetic analyses of Alliaceae (Fay and Chase, 1996
; Fay, Qamaruz Zaman, and Chase, 1997
). Structural zygomorphy involving perianth modification and stamen suppression probably evolved only once in Alliaceae (in Gilliesieae), and presumed insect mimicry only in Gilliesia. However, a more detailed phylogenetic investigation of South American Alliaceae, especially the tribe Gilliesieae, is required in order to satisfactorily evaluate floral evolution in this interesting group. For example, since the genus with six stamens (Miersia) has a zygomorphic perianth, it is possible that there has been an atavistic reversal to an (almost) actinomorphic perianth in Gethyum atropurpureum, as has apparently occurred in some Orchidaceae (Rudall and Bateman, 2002
). From their morphology, the other two zygomorphic genera, Miersia and Solaria, are clearly closely related to Gilliesia and Gethyum, but their precise relationships require further investigation. Fay, Qamaruz Zaman, and Chase (1997)
did not include Miersia or Solaria, nor the actinomorphic genera Speea and Trichlora in their combined multigene analysis of Alliaceae, due to lack of available DNA material, and instead focused on relationships among the other South American genera, including Nothoscordum, Leucocoryne, and their relatives. The discovery here of tiny osmophores of tepaline origin in Gethyum atropurpureum (in addition to Gilliesia and Miersia) indicates that they probably occur in all the zygomorphic Alliaceae genera (though not yet confirmed in Solaria); their presence in Miersia, which has six stamens, suggests that these unique structures evolved independently from stamen suppression.
Developmental flexibility in Gilliesieae
If the pollination mechanism of Gilliesia is indeed sexually deceptive, as its visual signals would indicate, the genus differs significantly in its developmental flexibility relative to the better known, independently evolved pseudocopulatory orchids of Europe, South America, and Australasia. Although orchids generate frequent terata (individuals showing profound phenotypic change relative to their parents), and there is indirect evidence that those terata occasionally establish radically novel evolutionary lineages (see reviews by Bateman and DiMichele, 2002
; Rudall and Bateman, 2002
), it has always been assumed that terata within pseudocopulatory lineages could not establish themselves as new species because each species was under strong stabilizing selection imposed by the close coevolutionary relationship with its pollinating insect(s). The mate choice between insect and orchid, a prerequisite for pseudocopulation, is viewed as analogous to the choice made between opposite sexes of the pollinating insect; in other words, pseudocopulatory flowers should in theory be subject to the relatively strong selection pressures inherent in mate choice among insects. This would in most circumstances militate against pronounced floral variation within species, though selection will inevitably be strongest on the primary cues, whether visual, tactile, or olfactory (P. K. Endress, personal communication, 2002).
However, examination of even a few plants demonstrates that floral variation in Gilliesia is exceptional (e.g., Muñoz, 2000
; Muñoz and Moreira, 2000
; Ravenna, 2000
; this paper), both between individuals (cf. Fig. 1E, F) and, to a lesser degree, among early- and late-opening flowers of the same inflorescence (cf. Fig. 1A, E). In G. graminea, we have observed substantial variation in the number, relative size, and orientation of the tepals, often resulting in asymmetry. (The term asymmetry is used here in the correct sense of lacking any symmetry planes, rather than to mean bilateral symmetry, as inappropriately applied in many evolutionary–developmental studies: for reviews of terminology see Neal, Dafni, and Giurfa, 1998
; Endress, 2001
; Rudall and Bateman, 2002
.) Ravenna (2000)
reported reduction in outer tepal number and speculated that this originated by fusion of two lateral tepals; however, in our material of G. graminea, in flowers with five tepals the missing tepal is the inner median (adaxial) one. A more likely scenario is that in bilaterally symmetric flowers (Fig. 1A) the inner median (adaxial) tepal primordium is normally entirely suppressed (or rarely entirely expressed, in occasional flowers with six tepals, reported by Muñoz, 2000
; Ravenna, 2000
), but that in asymmetric flowers it is either partially or entirely fused with a lateral outer tepal primordium. Such partial expression would inevitably result in the types of asymmetry observed here; partial fusion would result in flowers with a bilobed composite adaxial tepal (Fig. 1E), and complete fusion would result in an enlarged composite adaxial tepal (Fig. 1F).
Other authors (e.g., Muñoz, 2000
; Muñoz and Moreira, 2000
; Ravenna, 2000
) have reported variation in tepal number between five and six in at least two species of Gilliesia (G. graminea and G. montana), as well as in the number of stamens that are suppressed relative to the presumed ancestral number of six, although no reduction in stamen number was observed here. Muñoz and Moreira (2000)
reported that in at least two species of Gilliesia (G. graminea and G. montana) there is variation between five and six tepals and between three and four stamens; they further reported that in at least two species of Gethyum (G. atropurpureum and G. cuspidatum) and in Solaria miersioides the number of fertile stamens varies from one to five. Thus, at least two of the floral whorls are apparently unusually variable in Gilliesieae.
The wide range of floral morphs in G. graminea indicates that mutations in a TCP-group gene (cf. Cubas, 2002
) such as CYCLOIDEA or the closely related teosinte branched1 are an improbable explanation for such developmental flexibility. A naturally occurring genetic polymorphism in one of the MADS-box genes similar to the classic BICALYX polymorphism recorded in Clarkia (Onagraceae: Ford and Gottlieb, 1992
) represents a possible explanation, but it is perhaps more likely that a few downstream regulators of a TCP-group gene have been partially or wholly suppressed. Even this interpretation does not adequately explain significant variation observed among early- and late-opening flowers of the same inflorescence (cf. Fig. 1A, E), given that by definition these flowers share a single genome.
Fluctuating asymmetry as displayed in Gilliesia represents an evolutionary phenomenon thus far studied in greater detail in animals (e.g., Lens et al., 2002
) than in plants (e.g., Neal, Dafni, and Giurfa, 1998
). The approach is predicated on the assumption that, in a homogeneous environment, asymmetry in a single organ cannot be due to environmental or genetic factors but must instead reflect minor and presumably random perturbations in intercellular communication or rates of cell growth, elongation, or division, which are normally buffered by homeostatic controls (Palmer, 1996
; Lens et al., 2002
). Extrinsic stresses can increase asymmetry by either increasing developmental noise or decreasing developmental stability (these two opposing phenomena are in practice difficult to distinguish). Fluctuating asymmetry measures the degree of stability in a population by sampling extensively that population and comparing the result against the expectation of a normal distribution of right–left variation about a zero mean (deviations from this condition are likely to reflect genetic causes). Data are far more easily and cheaply acquired, and are more subtle and sensitive to stress, than are traditional estimates of fitness per se, though they are challenging to interpret statistically and are prone to various sampling errors. Such errors are problematic in studying animals, where asymmetry is typically subtle; levels of fluctuating asymmetry in targeted features typically approximate 1% (Lens et al., 2002
).
In contrast, plants in general show rather greater degrees of fluctuating asymmetry, though even among higher plants Gilliesia is unusual in this regard. Plants typically develop under higher stresses, since they lack the ability of higher animals to move to environments of lower stress, and plant developmental systems typically combine higher noise with lower stability, better conditions under which to detect fluctuating asymmetry (cf. Lens et al., 2002
). Drawing largely on the research of Møller and Eriksson (1994
, 1995
) and accepting that there were frequent exceptions, Neal, Dafni, and Giurfa (1998)
generalized that levels of fluctuating asymmetry were lower in zygomorphic than actinomorphic flowers, but that this contrast also extended to their respective leaves, suggesting that pollinator choice may not be the cause of the contrast in levels of variation among conspecific flowers. Pseudocopulatory flowers are normally more dependent on pollinator choice than are other zygomorphic flowers, increasing the likelihood of selection against asymmetry, but only in cases in which outcrossing is ensured.
Conclusions
Examination of the floral structure of Alliaceae highlights several aspects that are of systematic and evolutionary potential; these will be explored further in a publication on the systematics of Alliaceae (M. F. Fay, M. W. Chase, and P. J. Rudall, unpublished data). A gynobasic style and reduced ovule number are probable synapomorphies for the genus Allium. The presence of solid styles and possession of tenuinucellate ovules both represent highly consistent synapomorphies for Alliaceae and support the separation of Agapanthus and Themidaceae from Alliaceae. Absence of septal nectaries (i.e., complete fusion of carpel margins) is a synapomorphy for the sister genera Gilliesia and Gethyum (Gilliesieae); septal nectaries are present in all other Alliaceae. Furthermore, there are several other potential synapomorphies for these and other Gilliesieae, including the presence of petal appendages, bilateral symmetry, and stamen suppression (Table 2).
The four genera of Gilliesieae, particularly Gilliesia graminea, may offer prospective model organisms for investigations of zygomorphy and fluctuating asymmetry. Floral zygomorphy is expressed in a similar manner in these taxa to that of the eudicot Antirrhinum and other Lamiales (e.g., Endress, 1999
), in which the adaxial stamens and tepals are suppressed. This may indicate that a CYCLOIDEA homolog is responsible (cf. Coen, 1996
; Cronk and Möller, 1997
; Cubas, Vincent, and Coen, 1999
; Luo et al., 1999
; Citerne, Möller, and Cronk, 2000
; Theissen, 2000
; Endress, 2001
), although the genetic mechanisms controlling tepal zygomorphy are apparently unstable. Floral asymmetry of the type described here in G. graminea is unusual; more familiar types are left–right asymmetry, enantiostyly, and contorted asymmetry (for review see Endress, 2001
). Gilliesia offers a particularly attractive range of potential variables for the study of fluctuating asymmetry, combining strong variation in both meristic (e.g., counts of tepals, functional stamens, appendages) and metric (e.g., relative widths of tepals, lengths of stamens and appendages) parameters that may or may not be correlated between particular floral whorls. Not only are many relevant traits available for study, but also each inflorescence produces many genetically identical flowers from closely adjacent points of origin at the apex of the flowering stem, thereby minimizing nutritional clines among the flowers. Admittedly, Gilliesia may possess other features, notably putative pheromones (but not nectar), that could influence pollinator choice but lack a definable size or shape, and thus cannot yield data on fluctuating asymmetry. Nonetheless, we believe that Gilliesia would provide an excellent comparison, both with members of other clades that are potential insect mimics and with other zygomorphic flowers that possess contrasting insect-mediated pollination mechanisms (Møller and Eriksson, 1995
) or have undergone the apparently irreversible transition to autogamy (P. M. Hollingsworth, R. M. Bateman, and J. Squirrel, personal observations) with consequent release from selection for features contributing to pollinator attraction.
|
FOOTNOTES |
|---|
101 The authors thank Tony Hall (Royal Botanic Gardens, Kew, UK) for sharing his extensive horticultural expertise, Mélica Muñoz (Curator of SGO Herbarium, Chile) for helpful communications, and Peter Endress (University of Zurich, Switzerland) and J. Chris Pires (University of Madison, Wisconsin, USA) for constructive reviews of the manuscript.
4 Author for reprint requests (p.rudall{at}rbgkew.org.uk
)
|
LITERATURE CITED |
|---|
