| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Systematics |
Natural History Museums and Botanical Garden, University of Oslo, POB 1172 Blindern, 0318 Oslo, Norway
Received for publication October 29, 2002. Accepted for publication January 16, 2003.
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONLITERATURE CITED |
|---|
Key Words: Allium • embryo sac • embryology • endosperm • seed development • taxonomy • Themidaceae • Triteleia
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONLITERATURE CITED |
|---|
, p. 790) concluded that the cormous, inodorous New World genera of the Alliaceae (then often referred to as the Brodiaea group, later as the tribe Brodiaeeae [Traub, 1972
]) had little taxonomic affinity with the bulbous, onion-scented (Saghir et al., 1966
) genera of that family. Embryological data from four genera of the Brodiaea group in the old sense, namely from Muilla maritima (Berg and Maze, 1966
), Brodiaea section Coronariae (Berg, 1978
), Dipterostemon and Dichelostemma (Berg, 1996
), are in support of Mann's view.
When their data from rbcL sequencing pointed strongly in the same direction, Fay and Chase (1996)
resurrected and recircumscribed the family Themidaceae Salisb., as a family containing the Brodiaea group in the old sense, and a few additional genera (see Rahn, 1998b
). The Themidaceae, according to Fay and Chase (1996)
, probably are most closely related to the Hyacinthaceae. However, the assumption that Themidaceae is sister to Hyacinthaceae recently has been questioned, after more DNA regions have been analyzed (Fay et al., 2000
; Pires, 2000
, p. 71; Pires and Sytsma, 2002
, p. 1353).
Much disagreement has existed with regard to the number, circumscription, and mutual relationships of genera within the Themidaceae (Keator, 1968
; Niehaus, 1968
, 1980
; Berg, 1996
; Pires and Sytsma, 2002
). Fay and Chase (1996)
appealed for further studies of variation within and between the genera of this family in order for infrafamilial divisions to be defined. Berg (1996)
, on the basis of embryology, advocated resurrection of the genus Dipterostemon Rydberg for Dichelostemma capitatum A. W. Wood.
Pires (2000)
, Pires et al. (2001)
, and Pires and Sytsma (2002)
provided new and most valuable biosystematic and molecular phylogenetic data for a majority of species and for all genera of the Themidaceae, finding a lack of affinity with the Alliaceae and recognizing four major clades: (1) the Milla complex, (2) the Brodiaea–Dichelostemma–Triteleiopsis complex, (3) the Triteleia–Bloomeria–Muilla clevelandii complex, and (4) the Androstephium–Muilla complex.
With regard to embryology, the literature on Triteleia contains only one brief report, namely of the presence of raphides and an helobial endosperm in T. peduncularis Lindley (Stenar, 1949
).
This study presents embryological data for six species of Triteleia. The data are comparable to those previously presented for Muilla maritima (Berg and Maze, 1966
), Brodiaea section Coronariae (Berg, 1978
), and Dipterostemon and Dichelostemma (Berg, 1996
). The data contribute further towards the understanding of generic relationships and phylogeny within the Themidaceae and towards the evaluation of relationships of this family with the Alliaceae and the Hyacinthaceae.
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONLITERATURE CITED |
|---|
Small buds, ovaries from larger buds and flowers, and parts of fruit were preserved in Belling's modified Navashin fluid and dehydrated with tertiary butyl alcohol. Paraffin sections (10 µm) were cut on a rotary microtome and stained with safranin and fast green.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONLITERATURE CITED |
|---|
) into nectar channels on the ovary outside (Fig. 1). Specialized cells in the ovary wall produce raphides (Figs. 2, 14). The placentae are covered by stigmatoid tissue (Figs. 1, 17). The ovules are arranged in two rows in each of the three ovarian locules. Ovule orientation within the ovary in early stages is approximately dorsal pleurotropous (Fig. 9), in the sense of McLean and Ivimey-Cook (1956
, p. 1392), but soon becomes more or less dorsal hypotropous (Fig. 1).
|
|
90° (Fig. 16, cf. Fig. 7). At the one-nucleate embryo sac stage, the ovule has become anatropous (cf. Fig. 8). At the time of fertilization, the ovule is short and thick, anatropous, bitegmic, and crassinucellate (Fig. 17). There is no distinct funiculus. No raphides occur in the ovule. The inner integument is two cells thick, except at the thickened apex, where the inner integument may become 3–4 cells thick (Figs. 23, 25). The inner integument closes above the nucellus at the time when the functional megaspore germinates (Fig. 22) or shortly thereafter (cf. Fig. 8).
After fertilization, the cells of the inner integument elongate with the ovule to become up to 15–20 times as long as thick. However, the cells remain normal, i.e., not enormously enlarged as in some other Themidaceae. The thickness of the inner integument does not increase, nor do its cells divide periclinally. It remains thin and two-layered (Figs. 25–27) until it becomes compressed towards seed maturity.
The outer integument soon becomes 3–4 cells thick, except at the very apex where it remains thinner (Figs. 4, 23). The cells divide and enlarge as the ovule grows, but the number of layers never goes beyond four. The cells of the outer layer become thick-walled to act as the major protective part of the seed coat. The outer integument does not participate in the formation of the micropyle (Fig. 25), but will eventually close around the ovule sometime after fertilization (cf. Fig. 11).
The vascular strand has fully differentiated sieve tubes and vessels before fertilization (cf. Fig. 13).
Nucellus
At an early stage, the nucellus consists of a one-layered epidermis surrounding a central core, 4–5 cells in width (cf. Fig. 28). The nucellar epidermis stays one cell thick during embryo sac development. Its cells remain more or less isodiametric because of anticlinal divisions. At the time of fertilization, the nucellar epidermis is made up of cells rich in cytoplasm and with large nuclei (Fig. 25, cf. Fig. 13). It is definitely not penetrated by the enlarging embryo sac. Even the young endosperm is completely enclosed by the persistent nucellar epidermis (Figs. 6, 26).
|
The main growth of the nucellus during early embryo sac development is due to cell divisions within the central core of cells. Most of the growth, however, occurs in the subepidermal cells. During initial nucellar elongation, the originally subepidermal cells become obliquely elongated, with most of the later divisions in these cells being more or less periclinal (Figs. 21–24). Later, as the nucellus grows relatively more in thickness, most divisions occur subepidermally. Diagonal rows of rapidly enlarging nucellar cells is the most characteristic feature of the T. laxa ovule at the mature embryo sac stage (Fig. 25), as well as at early stages of endosperm development (Figs. 18, 26, 27). The bulk of the nucellus at these stages is made up of very large cells that are poor in cytoplasm and separated by large intercellular spaces. Only towards the periphery and the base are the nucellar cells densely packed, smaller, richer in cytoplasm, and still dividing. The innermost part of the original central core of cells eventually comes to form a strand of relatively small, narrow cells with slightly thickened walls between the embryo sac base and the chalaza (Fig. 25, cf. Figs. 12, 13).
At the time of fertilization, the nucellus is short and thick, approximately as broad as long (Fig. 17). It continues to surround the growing endosperm for a considerable period of time (Fig. 27).
Archesporium and megasporogenesis
The subepidermal archesporial cell divides to form a megaspore mother cell and a parietal cell. Already, before initiation of meiosis in the megaspore mother cell, a parietal tissue of four cells has formed (Fig. 3, cf. Fig. 28). A few additional divisions occur in the parietal tissue of some ovules (Fig. 23). All parietal cells become compressed and resorbed by the developing embryo sac before fertilization (Fig. 25).
A case of two embryo sacs in the same ovule indicates the rare presence of two functional archesporial cells. The two sacs in question were enclosed within the same inner integument. However, they were separated from each other by a double layer of nucellar cells (Fig. 2); apparently, each sac has a separate nucellus developing around it. It is difficult to imagine how such an arrangement could occur, if the two sacs came from two megaspores of the same tetrad.
The megaspore mother cell produces a tetrad of megaspores in the usual manner (Figs. 3, 19–21). Linear tetrads (Figs. 7, 22, 23) appear somewhat more common than T-shaped tetrads (Fig. 21).
Embryo sac development is monosporic. Normally, the chalazal megaspore is functional, and the upper three degenerate (Figs. 4, 23, 24). In exceptional cases, one of the other megaspores functions (cf. Fig. 47), and the chalazal one degenerates.
|
Three successive nuclear divisions (Figs. 24, 25) produce an embryo sac of the Polygonum type. The mature embryo sac (Figs. 5, 17, 25) is longer than broad and of typically ovoid shape. Synergids, egg cell, and antipodals are as described for Muilla maritima (Berg and Maze, 1966
). The polar nuclei fuse before fertilization. The large secondary nucleus lies close to the antipodal cells. The latter begin to degenerate before fertilization, but remain long into endosperm development, as three cells empty of contents except for a centrally located, strongly staining, structureless mass (Figs. 6, 26, 27).
Fertilization and endosperm development
Papillate stigmatoid tissue on the placenta (Figs. 1, 17) leads the pollen tube to the micropyle, which is formed by the inner integument alone. Also, at the time of fertilization, the integumentary cells bordering the micropyle attain a papillate shape (Fig. 25, cf. Fig. 12). Pollen tubes were seen in micropyles, but stages showing the actual fertilization were not obtained.
The primary endosperm nucleus starts to divide before the zygote. The first nuclear division takes place in the chalazal end of the endosperm cell. This division is followed by the formation of a transverse wall, which cuts off a smaller chalazal chamber from the larger micropylar one. Free nuclear divisions occur in both chambers. That is, endosperm development is helobial (Figs. 6, 18). Initially, nuclear divisions within the endosperm are synchronous (Fig. 26), but after four nuclei have formed in each chamber, development slows down in the chalazal chamber (Fig. 27) to stop at the final number of eight, sometimes possibly only four. As more nuclei form in the micropylar chamber, divisions become less synchronous, spreading in a wave-like fashion from the micropylar end of the endosperm towards the chalazal.
While most of the micropylar endosperm chamber is still occupied by a large central vacuole, cell walls begin to form between the nuclei in its chalazal end (cf. Fig. 10). Somewhat later, walls are also laid down in the micropylar end adjacent to the developing embryo between the nuclei that are dispersed in a thin layer of cytoplasm along the chamber wall (cf. Fig. 11). From these initial areas, peripheral wall formation spreads throughout the micropylar endosperm chamber. Simultaneously, but more rapidly as the expansion of the ovule slows down, cell formation proceeds into the middle part of the chamber, gradually reducing the size of its central vacuole.
Embryo
The first division of the zygote is transverse and produces a proembryo consisting of a terminal cell and a basal cell. The basal cell divides slightly before the terminal cell. The division is transverse, producing two cells, m (towards the terminal cell) and ci (cf. Figs. 34, 44). Division of the terminal cell often is observed to be in prophase as the basal cell division reaches anaphase. The terminal cell divides longitudinally to form the cells, q. The four-celled proembryo, consequently, has the cells disposed in three tiers: q, m, and ci.
Cell m divides longitudinally, while cell ci divides transversely to form the single-celled tiers, n' and n (cf. Fig. 45). The two terminal-cell derivatives, q, divide longitudinally to form the quadrant group of cells. Consequently, the eight-celled proembryo has its cells disposed in four tiers: q, m, n, and n'.
The subsequent divisions were not worked out. The proembryo soon becomes differentiated into a several-celled suspensor and a globular embryo proper (cf. Fig. 11). It was clear, however, that the basal cell, through its derivatives m and n, contributed substantially towards the embryo proper.
Embryo development in T. laxa, consequently, conforms to the Asterad type of embryogeny (Johansen, 1950
).
Triteleia hyacinthina
In T. hyacinthina the number of ovules per locule is between four and seven. The ovules are arranged as in T. laxa, but tend to remain horizontal somewhat longer, i.e., to the two-nucleate stage of embryo sac development (Fig. 12). Initially, the ovule is almost straight, but soon begins to curve (Figs. 7–9, 28–30) to become anatropous before the time of fertilization. It is crassinucellate, with the micropyle formed by the inner integument only (Fig. 31). There is no funiculus, and the ovule contains no raphides.
Integuments are similar to those of T. laxa. The two-layered inner integument (Figs. 8, 30) closes above the nucellus at the one-nucleate embryo sac stage or shortly thereafter. Its tip then thickens to become 3–4 cells wide and 5–6 cells high, completely covering the nucellus (Fig. 31). The cells of the two-layered part elongate but do not grow in thickness. This part of the inner integument becomes compressed during endosperm growth. The thickened tip persists, staining dark red by safranin when the young embryo is developing (Fig. 11). The outer integument starts as a lateral protrusion from the base of the inner integument at the megaspore mother cell stage (Fig. 28). At the one-nucleate embryo sac stage, the outer integument is three cells wide and extends approximately to the apex of the nucellus (Figs. 8, 30). It continues to lengthen, and at the four-nucleate embryo sac stage, it is four cells wide at the base, three cells wide at mid-length, and extends beyond the nucellus (Fig. 31). By the time the endosperm has become cellular, the outer integument is four cells thick throughout, except at the very tip, which now extends beyond the inner integument (Fig. 11). The cells of the inner three layers are now large and highly vacuolated. The outer layer is made up of more narrow cells that are heavily cutinized.
The nucellus of T. hyacinthina is similar to the nucellus of T. laxa, both in earlier (Figs. 28–30) and in later (Figs. 31–32) stages. Its epidermis stays one-layered with only an exceptional periclinal division (Fig. 31) occurring before fertilization. It is not penetrated by the embryo sac. Its subepidermal, oblique cells (Fig. 28) divide periclinally (Fig. 30), eventually resulting in the massive pear-shaped nucellar tissue that gives the ovule its form. The nucellar apex cells persist in a somewhat compressed form below the inner integument beak (Fig. 11), long after all other nucellar cells have been crushed and resorbed by the growing endosperm.
As in T. laxa, parietal cells are present and the embryo sac is of the Polygonum type (Figs. 7, 28–33). The parietal cells become compressed and absorbed by the four-nucleate (Fig. 31) to eight-nucleate embryo sac. The upper dyad cell is always considerably smaller than the lower, and T-shaped megaspore tetrads (Fig. 30) are as common as linear tetrads (Fig. 29). Apparently, the smaller the upper dyad cell, the greater the chance of a T-shaped tetrad. In this species, the lowest megaspore functions as usual (Fig. 30). Only one possible exception was observed. Synergids of this species have a well-developed filiform apparatus (Fig. 33).
Antipodal cell remains are discernible until well after cell walls have formed in the peripheral part of the endosperm.
The endosperm is of the helobial type (Fig. 10). Cellularization of the micropylar endosperm chamber proceeds as in T. laxa. Early embryogenesis produces a four-celled proembryo with the four cells arranged in three tiers (Fig. 34).
Triteleia dudleyi
Triteleia dudleyi has in each ovarian locule two rows of three ovules. Sometimes one of the two rows has three ovules, the other two, i.e., normally between 15 and 18 ovules per ovary. The ovules bend downwards to become hypotropous before meiosis takes place in the megaspore mother cell.
Otherwise, T. dudleyi is similar to the other species: raphides occur in the ovary wall, parietal cells are present (Figs. 35–36), the embryo sac is of the Polygonum type, the endosperm is helobial. The nucellus is short and thick (Fig. 36), with a persistent apex. The inner integument closes above the nucellus prior to the two-nucleate embryo sac stage (Fig. 36).
At the onset of lateral growth in a young nucellus, oblique walls in the subepidermal layer have produced a cylindrical stack of cells, each of which dip strongly towards the base of the ovule at its inner margin (Fig. 35). Later, as these oblique hypodermal cells have divided more or less periclinally, oblique rows of cells have been formed (Fig. 36), producing the special nucellar thickening so characteristic of all species of Triteleia.
Triteleia ixioides
Triteleia ixioides is similar to T. laxa and T. hyacinthina. The ovary is trilocular with septal nectaries and raphides (Fig. 14). Six to ten ovules are arranged in two rows in each ovary chamber. The ovule is anatropous, crassinucellate, bitegmic, short, and thick (Figs. 13, 37). Parietal cells are present, but their number appears restricted to two: only two parietal cells on top of each other are seen in both median (Figs. 38, 40) and transmedian (Fig. 41) sections.
The inner integument is characteristically thickened at the apex (Fig. 40), closes above the nucellus between the two-nucleate and the four-nucleate embryo sac stage (Figs. 12, 40, 41), and is the only integument forming the micropyle (Fig. 42).
The nucellus grows in thickness, as in all other species of Triteleia, through more or less periclinal divisons (Figs. 41, 42) in the oblique to radiating hypodermal cells of the young nucellus (Fig. 40). The persistent nucellar epidermis is permanently one cell thick at its apex (Figs. 12, 13, 38–44). However, quite a few periclinal divisions do occur in the more basal part of the nucellar epidermis approximately at the time of fertilization (Fig. 42). Such divisions increase in number during endosperm development (Fig. 43), adding to the thickness growth of the nucellus. Epidermal thickness growth within the basal part of the nucellus after fertilization appears more pronounced in this than in the other species of Triteleia.
There is a tetrad of megaspores, which in T. ixioides is always linear. Rarely, more than one megaspore of a tetrad develops. An ovule in which the two spores in the middle of the tetrad are both enlarged and vacuolated, i.e., germinating, is shown in Fig. 38. Another ovary contained two neighboring ovules each with two developing embryo sacs. In the one (Fig. 39), a four-nucleate sac and a two-nucleate sac had developed, apparently, from the two lowermost spores of the tetrad. The embryo sac is of the Polygonum type and does not penetrate the nucellar apex (Figs. 12, 38–42).
The endosperm is helobial (Fig. 15). The young endosperm is enclosed by the nucellus (Fig. 43). Nuclear divisions in the endosperm initially are synchronous. Eight nuclei are produced within the chalazal chamber before this chamber begins to degenerate. This chamber does not participate in the formation of the endosperm proper. The number of free nuclei in the micropylar chamber is between 50 and 100 when the first cell walls are laid down in its chalazal end. Wall formation then proceeds as in T. laxa.
The two-celled proembryo becomes four-celled by a longitudinal division of the terminal cell and a transverse division of the basal cell (Fig. 44). The eight-celled proembryo has the cells arranged in four tiers: n', n, m, and q (Fig. 45), as described for T. laxa. The T. ixioides embryo, too, apparently develops according to the Asterad type.
Triteleia hendersonii
This species is essentially similar to the other species: there are 4–6 ovules in two rows in each of the ovarian chambers. Septal nectaries and raphides occur in the ovary. The inner integument is permamently two cells thick (Fig. 47). Parietal cells are present, but in larger numbers than in the other species (Figs. 46–47). A linear tetrad produces a Polygonum type embryo sac, often from a megaspore other than the lower (Fig. 47). The nucellus grows in thickness from oblique subepidermal cells, in characteristic Triteleia fashion (Figs. 46–47), the nucellar epidermis is one cell thick, and the nucellar apex is not penetrated by the enlarging embryo sac.
As in T. ixioides, the nucellar epidermis regularly shows periclinal divisions in the basal half of the nucellus at the time of fertilization.
Post-fertilization stages were not included.
Triteleia crocea
Triteleia crocea possesses only four ovules per ovary locule, but otherwise is essentially similar to the other species: raphides are present in the ovary wall, the inner integument is permanently two cells thick (Fig. 48), the outer integument becomes four cells thick, the nucellus is short and thick with a persistent apex and thickening growth from the subepidermal cell layer (Fig. 48). The endosperm is helobial: stages (micropylar nuclei/chalazal nuclei) of 1/1, 2/2, 4/2, and 8/4 were observed. Initial embryogenesis is as described for T. ixioides (cf. Fig. 44).
The nucellus appears somewhat larger in this species than in the others, namely about twice as long as the embryo sac (Fig. 48).
Pre-flowering stages were not included.
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONLITERATURE CITED |
|---|
) species of Triteleia were included in the present study, but not all stages were present in all species. No difference in basic embryological characteristics were found. All species studied possessed septal nectaries, raphides in the ovary wall, an anatropous, crassinucellate ovule with a micropyle formed only by the inner integument; a short and thick nucellus that is not penetrated by the embryo sac, which thickens from its subepidermal layer of oblique cells and which is without periclinal divisions in the apical part of its epidermis; an inner integument of normal-sized (i.e., not greatly enlarged) cells, which permanently stays two cells thick; parietal cells within the nucellar apex; a monosporic embryo sac of the Polygonum type; and an endosperm of the helobial type.
The six species studied represent all three sections into which Hoover (1941
, p. 76; cf. Lenz, 1975
) subdivided the genus: section Triteleia (T. laxa, T. crocea), section Hesperoscordum (T. hyacinthina, T. hendersoni), and section Calliprora (T. ixioides, T. dudleyi). The four clades recognized within Triteleia by Pires and Sytsma (2002)
show no correspondence with Hoover's sections. However, only Pires and Sytsma's Montane clade (T. montana and T. lemmonae) is not represented in my material. Both from a morphological and from a molecular point of view, the genus seems well represented by the species selected for this investigation. I feel confident making the assumption that all species of Tritelea share the embryological characteristics that were found in the six species studied. This sharing very probably also includes an embryo development of the Asterad type, because T. laxa and T. ixioides, the two species in which this type of embryo development was observed, belong in different sections/clades, and T. hyacinthina, which shows early embryogeny stages compatible with an Asterad type, belongs in the remaining of Hoover's three sections.
Triteleia peduncularis, the only previously studied species, belonging to the Triteleia section according to Hoover (1941)
and to the laxa clade according to Pires (2000)
, follows the described pattern: raphides are present in the ovary wall and the endosperm is of the helobial type (Stenar, 1949
).
Comparison with other Themidaceae
Embryologically Triteleia is similar to Muilla maritima in all characteristics studied, except for one (Table 1): the epidermis in the apical part of the nucellus is permanently one cell thick in Triteleia, while periclinal divisions produce a more or less two-layered epidermis in the nucellar apex of Muilla maritima prior to fertilization (Berg and Maze, 1966
, Fig. 14).
|
, Fig. 20), greatly enlarged cells in the inner integument, and a nuclear endosperm (Berg, 1996
, Fig. 50). However, they are similar to Triteleia in having a unilayered epidermis in the nucellar apex.
Dichelostemma and Brodiaea also differ from Triteleia by the possession of a multilayered inner integument. The fact that Dipterostemon is similar to M. maritima and Triteleia in having the primitive character of only two cell layers in the inner integument (Table 1) has been discussed elsewhere (Berg, 1996
).
No comparisons can be made, at present, with Androstephium, Bloomeria, Muilla (except M. maritima), and Triteleiopsis, because of a total lack of embryological data.
Comparison of Themidaceae, with Hyacinthaceae and Allium
In Table 2 the Themidaceae are compared to the Hyacinthaceae sensu Speta (1998a)
and Allium, the most typical representative of the Alliaceae sensu Rahn (1998a)
, with regard to five embryological characteristics that are constant throughout the Themidaceae.
|
. Raphide bundles were considered taxonomically important in distinguishing the raphide-lacking "Lilioideae" from the raphide-possessing "Scilloideae" (Schnarf, 1948
, p. 261). The absence of raphides in some monocotyledons should be considered as derived states, according to Dahlgren and Rasmussen (1983
, p. 296).
2. Ovules are anatropous in the Themidaceae, mostly anatropous, rarely campylotropous (Cave, 1974
) in the Hyacinthaceae, and campylotropous in Allium. "Nearly anatropous" ovules have been recorded from Leucocoryne of the Alliaceae (Cave, 1939
). According to Dahlgren (1991, pp. 120–121)
, "anatropous ovules are distributed throughout the angiosperms and are usually regarded as representing the plesiomorphic state. Transitions between anatropous and campylotropous ovules occur within a number of independent lines."
3. Parietal cells are present in the Themidaceae and the Hyacinthaceae, absent in Allium. The absence of parietal cell/tissue was regarded as a typical feature of the "Allieae" already by Stenar (1932)
. Agapanthus, which possesses parietal tissue, as well as an anatropous ovule (Stenar, 1933
) and a rhizome, was recently removed from the Alliaceae to the Amaryllidaceae (Fay and Chase, 1996
). The presence of a parietal cell (or tissue) is the primitive state in angiosperms, a view clearly presented by Schnarf (1931
, p. 259) and never disputed since (see Dahlgren and Rasmussen, 1983
, p. 329; Dahlgren, 1991
). Parietal tissue is present in all basal angiosperms (except for a few exceptional nested clades) (Endress and Igersheim, 2000)
.
4. The embryo sac is of the monosporic Polygonum type in the Themidaceae. It is of the bisporic Allium type in all normally reproducing, amphimictic species of Allium (see Berg and Maze, 1966
, p. 149). Within the Hyacinthaceae, the embryo sac type varies to an unusual degree. Most often, the embryo sac is of the Polygonum type (Speta, 1998a
, p. 265), but no less than five other embryo sac types have been recorded from the genera Hyacinthoides and Scilla (Table 2). However, "there can be little doubt that the Polygonum-type embryo sac is primitive in Hyacinthaceae" (Svoma and Greilhuber, 1988
, p. 179) and in Hyacinthoides and Scilla as well (Svoma and Greilhuber, 1989
, p. 598). Dahlgren and Rasmussen (1983
, p. 329), regarded the Allium type of embryo sac as synapomorphic for many Alliaceae: that is "many" probably because Dahlgren included the Themidaceae in the Alliaceae (see Dahlgren, Clifford, and Yeo, 1985
).
5. Only scant information is available for embryogeny (see Berg, 1996
, p. 798). The Asterad type occurs in both Themidaceae and Hyacinthaceae (cf. Johansen, 1950
, p. 244), while both Asterad and Onagrad type embryos have been reported from the Alliaceae. "Either the Asterad or possibly the Onagrad type of embryogeny must be regarded as ancestral in monocotyledons" (Dahlgren and Rasmussen, 1983
, p. 332).
In conclusion, the Themidaceae are embryologically very similar to the Hyacinthaceae and very different from Allium.
Taxonomic conclusions
(1) Embryology supports a monophyletic Triteleia in comparison to the other embryologically known genera of Themidaceae. Triteleia exemplifies Cave's rule: "the embryological characters of the species within a genus are constant" (Cave, 1953
, p. 140). (2) Since only six of 12–14 Triteleia species have been studied embryologically, any evaluation based on embryology of the four clades recognized within Triteleia by Pires and Sytsma (2002
, p. 1351) would be premature. The slight variations described earlier, e.g., in number of ovules per ovary, the period in which ovules remain horizontal, the size of the upper dyad cell (T. hyacinthina), the number of parietal cells (T. ixioides, T. hendersonii), the number of periclinal epidermal divisions in the basal part of the nucellus (T. ixioides, T. hendersonii), and the relative length of the nucellus (T. crocea), have no pattern and appear as species characters at the most.
(3) Triteleia and Muilla maritima are quite closely related, because of their overall embryological similarity. These two genera seem to belong within the same major part of the family. However, the presence of periclinal cell divisions in the apical nucellar epidermis of M. maritima and the absence of such divisions from Triteleia indicate a certain taxonomic distance between the two taxa and gives support to the view expressed by Pires (2000)
, Pires et al. (2001)
, and Pires and Sytsma (2002)
, namely, that M. maritima belongs to another clade than Triteleia. On the other hand, periclinal divisions in the nucellar apex epidermis of M. maritima might just be another variation at the species level, because no other member of the genus Muilla is known embryologically. That M. clevelandii, unlike M. maritima, is included in the Bloomeria-Triteleia clade by Pires and Sytsma (2002
, p. 1351) makes the situation more intriguing. A comparison of embryology, divisions in nucellar apices in particular, amongst Muilla species is definitely needed.
(4) Embryology is strongly in support of the view that Triteleia should not be considered a subgenus of Brodiaea (e.g., Jepson, 1951
; Munz and Keck, 1959
), but a genus in its own right, as maintained by, e.g., Greene (1886)
, Hoover (1939)
, Keator (1968)
, Niehaus (1968)
, Hickman (1993)
, and Pires (2000)
. The many and important embryological differences that exist between Triteleia on the one hand, and Brodiaea, Dichelostemma, and Dipterostemon on the other, are absolutely incompatible with the one-genus view. However, embryology indicates more. Even when accepted as a distinct genus, Triteleia was considered closely related to Brodiaea and Dichelostemma (including Dipterostemon) because of the shared presence of syntepalous flowers (see e.g., Niehaus, 1968
, p. 99). Number and importance of embryological differences indicate that Triteleia belongs in a major part of Themidaceae distant from the major part that holds Brodiaea, Dichelostemma, and Dipterostemon. This conclusion agrees fully with results from molecular studies (Pires, 2000
; Pires et al., 2001
; Pires and Sytsma, 2002)
.
(5) Phylogenetically, Muilla and Triteleia both are more primitive than the Dipterostemon-Dichelostemma-Brodiaea group, because the former two genera exhibit embryological character states that are presumably more primitive (cf. Sporne, 1956
) than the corresponding states exhibited by members of the latter group (Table 1): two cell layers in the inner integument vs. several (except Dipterostemon), a nucellus that is not penetrated by the embryo sac vs. penetrated, normal-sized cells in the inner integument vs. greatly enlarged cells, and an endosperm of the helobial type vs. nuclear (cf. Schnarf, 1931
, p. 259; Wunderlich, 1959
, p. 226; Dahlgren and Rasmussen, 1983
, pp. 330–332). This phylogenetic conclusion is supported by non-embryological evidence (Table 1): Muilla and Triteleia exhibit, for instance, the primitive state of 3 + 3 fertile stamens vs. 3 (except Dipterostemon), as well as simple nectar protection, mostly by means of filaments of functional stamens (sometimes widened or appendaged), vs. complex nectar protection. Muilla, in addition, is primitive in having free tepals.
(6) The Themidaceae are not closely allied to Allium, as was believed until quite recently (e.g., Dahlgren et al., 1985
, p. 196). The new information on Triteleia embryology presented in this paper strengthens the embryologic argument against such a relationship (see Berg and Maze, 1966
; Berg, 1978
, 1996
). Thus, the embryological evidence regarding affinity to the Alliaceae agrees with the molecular, morphological, and biochemical data offered by Fay and Chase (1996)
as their basis for resurrection of Themidaceae and by Pires and Sytsma (2002)
in their phylogenetic evaluation of the Brodiaea complex.
Alliaceous odor (see Saghir et al., 1966
) is considered one of the most characteristic qualities of the genus Allium (e.g., Traub, 1972
). It is present, also, in most other genera of the Alliaceae (Rahn, 1998a
). Alliaceous odor does not occur in any Themidaceae, nor in any Hyacinthaceae (Table 2). A statement by Fay and Chase (1996
, p. 441) to the effect that Androstephium of the Themidaceae possesses alliaceous chemistry is based on erroneous information given by Greene (1890
, p. 57). Androstephium breviflorum does not have an alliaceous odor (personal observation; see also Abrams, 1923
, p. 380; Pires, 2000
, p. 3), and neither does A. coeruleum, to judge from descriptions in which alliaceous odor, if present, in all probability would have been mentioned (e.g., Britton and Brown, 1896
; Correll and Johnston, 1970
).
In characteristics where the Themidaceae and the Alliaceae differ, the character state considered plesiomorphic among monocotyledons occurs in the Themidaceae.
(7) The Themidaceae, at first, seem to be rather closely related to the Hyacinthaceae, to judge from the great similarity in embryological characteristics exhibited by the two families (Table 2). However, absolutely all embryologic similarities here discussed between the Themidaceae and the Hyacinthaceae are in characters considered plesiomorphic within the monocotyledons (cf. Dahlgren and Rasmussen, 1983
). Thus, the taxonomic interpretation of embryologic similarities in this case meet with difficulties similar to those in the interpretation of the results of analyses of additional DNA regions (Fay et al., 2000
; Pires, 2000
, p. 71; Pires and Sytsma, 2002)
: the closest relatives of Themidaceae might possibly occur amongst other of the higher Asparagalean families.
Counting heavily against a close relationship between the Themidaceae and the Hyacinthaceae are two morphologic differences, namely the nature of the underground organ and the nature of the inflorescence (Table 2).
The underground organ in practically all Themidaceae is a tunicated corm (see Smith, 1930
; Moore, 1953
; Keator, 1968
). The underground organ of Triteleiopsis is different and somewhat peculiar, but still a corm (Hoover, 1941
, p. 99). The Themidaceae definitely do not possess "transitional structures" between corms and bulbs, as suggested by Dahlgren and Clifford (1982
, p. 59). In all Hyacinthaceae, the underground organ is a bulb (cf. Speta, 1984
, 1987
, 1998b
). The report of rhizomes in Schoenolirion and Chlorogalum (Speta, 1998a
) seems erroneous. Sherman and Becking (1991
, p. 133) say that Schoenolirion has a bulb, although somewhat unusual, and Chlorogalum is bulbous according to several sources (e.g., Hickman, 1993
). The presence of a corm is a synapomorphy for the Themidaceae (Fay and Chase, 1996
, p. 447), the presence of a bulb is a synapomorphy for the Hyacinthaceae, and both underground organs serve the same purpose (Dahlgren and Rasmussen, 1983
, p. 279) in similar environmental conditions. From an ecological/adaptive point of view, a bulbous clade most unlikely would evolve into a cormous clade living under the same conditions, or vice versa, i.e., unless the concept of transference of function sensu Corner (1958)
is given credence.
The inflorescence is cymose (sympodial) in the Themidaceae, most often in the form of a cymose umbel. The Hyacinthaceae infloresence is racemose (monopodial), in the form of a simple or compound raceme, rarely a spike (Table 2). This difference is fundamental. Were the racemose spikes and racemes of the Hyacinthaceae to become condensed into umbels, those umbels would be racemose or true umbels, not cymose umbels.
|
FOOTNOTES |
|---|
|
LITERATURE CITED |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONLITERATURE CITED |
|---|
Berg R. Y. 1978 Development of ovule, embryo sac, and endosperm in Brodiaea (Liliales). Norwegian Journal of Botany 25: 1-7
Berg R. Y. 1996 Development of ovule, embryo sac, and endosperm in Dipterostemon and Dichelostemma (Alliaceae) relative to taxonomy. American Journal of Botany 83: 790-801[CrossRef][ISI]
Berg R. Y. J. R. Maze 1966 Contribution to the embryology of Muilla, with a remark on the taxonomic position of the genus. Madroño 18: 143-151
Britton N. L. H. A. Brown 1896 An illustrated flora of the United States, Canada, and the British Possessions, vol. I. Charles Scribner's Sons, New York, New York, USA
Buchner L. 1949 Vergleichende embryologische Studien an Scilloideae. Österreichische Botanische Zeitschrift 95: 428-450[CrossRef]
Cave M. S. 1939 Macrosporogenesis in Leucocoryne ixioides Lindl. Cytologia 9: 407-411
Cave M. S. 1953 Cytology and embryology in the delimitation of genera. Chronica Botanica 14: 140-153
Cave M. S. 1974 Female gametophytes of Chlorogalum and Schoenolirion (Hastingsia). Phytomorphology 24: 56-60
Corner E. J. H. 1958 Transference of function. Botanical Journal of the Linnean Society 57: 33-40
Correll D. S. M. C. Johnston 1970 Manual of the vascular plants of Texas. Texas Research Foundation, Renner, Texas, USA
Dahlgren G. 1991 Steps toward a natural system of the dicotyledons: embryological characters. Aliso 13: 107-165
Dahlgren R. M. T. H. T. Clifford 1982 The monocotyledons: a comparative study. Academic Press, London, UK
Dahlgren R. M. T. H. T. Clifford P. F. Yeo 1985 The families of monocotyledons: structure, evolution and taxonomy. Springer-Verlag, Berlin, Germany
Dahlgren R. M. T. F. N. Rasmussen 1983 Monocotyledon evolution: characters and phylogenetic estimation. Evolutionary Biology 16: 255-395
Endress P. K. A. Igersheim 2000 Gynoecium structure and evolution in basal angiosperms. International Journal of Plant Sciences 161: 211-223[CrossRef]
Fay M. F. M. W. Chase 1996 Resurrection of Themidaceae for the Brodiaea alliance, and recircumscription of Alliaceae, Amaryllidaceae and Agapanthoideae. Taxon 45: 441-451[CrossRef][ISI]
Fay M. F. et al 2000 Phylogenetic studies of Asparagales based on four plastid DNA regions. In K. L. Wilson and D. A. Morrison [eds.], Monocots: systematics and evolution, 360–371. CSIRO Publishing, Collingwood, Victoria, Australia
Greene E. L. 1886 Studies in the botany of California and parts adjacent. I. Some genera which have been confused under the name Brodiaea. Bulletin of the California Academy of Sciences 2: 125-144
Greene E. L. 1890 Analogies and affinities. III. Pittonia 2: 51-57
Hickman J. C. [ed.] 1993 The Jepson manual: higher plants of California. University of California Press, Berkeley, California, USA
Hoover R. F. 1939 A definition of the genus Brodiaea. Bulletin of the Torrey Botanical Club 66: 161-166[CrossRef]
Hoover R. F. 1941 A systematic study of Triteleia. American Midland Naturalist 25: 73-100[CrossRef]
Huber H. 1969 Die Samenmerkmale und Verwandtschaftsverhältnisse der Liliifloren. Mitteilungen der Botanischen Staatssammlung München 8: 219-538
Jepson W. L. 1951 A manual of the flowering plants of California. University of California Press, Berkeley, California, USA
Johansen D. A. 1950 Plant embryology: embryogeny of the spermatophyta. Chronica Botanica, Waltham, Massachusetts, USA
Keator R. 1968 A taxonomic and ecological study of the genus Dichelostemma (Amaryllidaceae). Ph.D. dissertation, University of California, Berkeley, California, USA
Lenz L. W. 1975 A biosystematic study of Triteleia (Liliaceae). I. Revision of the species of section Calliprora. Aliso 8: 221-258
Lenz L. W. 1976 The nature of the floral appendages in four species of Dichelostemma (Liliaceae). Aliso 8: 383-389
Mann L. K. 1959 The Allium inflorescence: some species of the section Molium. American Journal of Botany 46: 730-739[CrossRef][ISI]
McLean R. C. W. R. Ivimey-Cook 1956 Textbook of theoretical botany, vol. 2. Longmans, Green and Co., London, UK
Moore H. E., Jr. 1953 The genus Milla (Amaryllidaceae–Allieae) and its allies. Gentes Herbarum 8: 261-294
Munz P. A. D. D. Keck 1959 A California flora. University of California Press, Berkeley, California, USA
Niehaus T. 1968 A biosystematic study of the genus Brodiaea (Amaryllidaceae). Ph.D. dissertation, University of California, Berkeley, California, USA
Niehaus T. 1980 The Brodiaea complex. Genera Brodiaea, Triteleia, Dichelostemma; family Amaryllidaceae. Four Seaons (Tilden Regional Park, Berkeley, California, USA.) 6: 11–21
Pires J. C. 2000 Biosystematics and molecular phylogenetics of Brodiaea (Themidaceae) and related lilioid monocots. Ph.D. dissertation, University of Wisconsin, Madison, Wisconsin, USA
Pires J. C. M. F. Fay W. S. Davis L. Hufford J. Rova M. W. Chase K. J. Sytsma 2001 Molecular and morphological phylogenetic analyses of Themidaceae (Asparagales). Kew Bulletin 56: 601-626[CrossRef]
Pires J. C. K. J. Sytsma 2002 A phylogenetic evaluation of a biosystematic framework: Brodiaea and related petaloid monocots (Themidaceae). American Journal of Botany 89: 1342-1359
Prychid C. J. P. J. Rudall 2000 Distribution of calcium oxalate crystals in monocotyledons. In K. L. Wilson and D. A. Morrison [eds.], Monocots: systematics and evolution, 159–162. CSIRO Publishing, Collingwood, Victoria, Australia
Rahn K . 1998a Alliaceae. In K. Kubitzki [ed.], The families and genera of vascular plants. III. Flowering plants. Monocotyledons: Lilianae (except Orchidaceae), 70–78. Springer-Verlag, Berlin, Germany
Rahn K . 1998b Themidaceae. In K. Kubitzki [ed.], The families and genera of vascular plants. III. Flowering plants. Monocotyledons: Lilianae (except Orchidaceae), 436–441. Springer-Verlag, Berlin, Germany
Saghir A. R. B. L. K. Mann M. Ownbey R. Y. Berg 1966 Composition of volatiles in relation to taxonomy of American alliums. American Journal of Botany 53: 477-484[CrossRef][ISI]
Schnarf K. 1931 Vergleichende Embryologie der Angiospermen. Verlag Gebrüder Borntraeger, Berlin, Germany
Schnarf K. 1948 Der Umfang der Lilioideae im natürlichen System. Österreichische Botanische Zeitschrift 95: 257-269[CrossRef]
Sherman H. L. R. W. Becking 1991 The generic distinctness of Schoenolirion and Hastingsia. Madroño 38: 130-138
Smith F. H. 1930 The corm and contractile roots of Brodiaea lactea. American Journal of Botany 17: 916-927[CrossRef][ISI]
Speta F. 1984 Zwiebeln—versteckte Vielfalt in einfacher Form. Linzer Biologische Beiträge 16: 3-44
Speta F. 1987 Die verwandtschaftlichen Beziehungen von Brimeura Salisb.: ein Vergleich mit den Gattungen Oncostema Rafin., Hyacinthoides Medic. und Camassia Lindl. (Hyacinthaceae). Phyton 26: 247-310[ISI]
