| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
2 Departamento de Biología, Bioquímica y Farmacia, Universidad Nacional del Sur, San Juan 670, 8000 Bahía Blanca, Argentina; and 3 Biology Department, Morrill South, University of Massachusetts, Box 35810, Amherst, Massachusetts 01003-5810 USA
Received for publication April 8, 1999. Accepted for publication September 28, 1999.
ABSTRACT
Development of the introrse, tetrasporangiate, and normally dorsifixed and poricidal stamens has been studied at the gross morphological and cellular level in ten species of Ericaceae. Microsporogenesis, followed in four species, is normal, with cytokinesis simultaneous, forming tetrahedral tetrads. The tricolp(or)ate pollen is shed as permanent tetrads with each segment two-celled except in Enkianthus in which pollen grains are three-celled monads. Anther-wall development is similar in all four species initially, but no regular pattern of wall development could be recognized thereafter. The tapetum, of parietal origin, is binucleate, glandular, and mainly uniseriate. Viscin threads occur with the tetrads in the three rhododendroid species. A well-developed endothecium appears only in Enkianthus.
Soon after stamen initiation, anthers of nine species invert at the eventual filament–anther junction to become introrse; in Enkianthus inversion occurs close to anthesis. Microsporogenesis starts during early inversion; greater cell elongation on the abaxial side of the young anther completes inversion by the late sporogenous-tissue stage. In Erica and, to a lesser extent Calluna, inversion results from greater abaxial than adaxial increase in cell number and length just above the filament-anther junction. The single vascular strand reflects the degree of inversion. Stamens of six species are appendaged; three have only awns, two only spurs, while one has both. Appendages arise from residual meristems after inversion is completed (or almost so) in all except Enkianthus. Awns develop at what will be the apex at maturity of each anther half. Their length and orientation vary among species. Only in Vaccinium do the awns become hollow (tubules). Spurs, varying in length, shape, and size, arise on the abaxial side from the filament, connective, or thecae.
Key Words: anther-wall development • appendages • Ericaceae • inversion • microsporogenesis • stamen development
The angiosperm order Ericales ranges from the tropics to subarctic and alpine regions of both northern and southern hemispheres. In different classification systems (Bessey, 1915
; Takhtajan, 1969, 1980
; Philipson, 1974
; Dahlgren, 1975
; Thorne, 1976, 1981
; Cronquist, 1979, 1988
) they have been placed close to Primulales, Theales, Dilleniales, Asteridae, or Cornanae depending on the particular characteristic most stressed by the author. Thus, the alliances have been varied, and they emphasize the fact that phylogenetic relationships of the order within the angiosperms are uncertain. Within the Ericales, certain families have consistently been kept together in the order: Ericaceae, Epacridaceae, Clethraceae, Vacciniaceae, Pyrolaceae, and Monotropaceae. The latter three of these are often included within the Ericaceae as subfamilies. Other small families have been added to, or separated from, these central families at various times by different authors: Empetraceae, Diapensiaceae, Cyrillaceae, Lennoaceae, Saurauiaceae, Actinidiaceae, and Grubbiaceae. Molecular and morphological studies by Kron, Chase, and Hills (1991)
, Anderberg (1992, 1993)
, Judd and Kron (1993)
, and Kron and Chase (1993)
have recently confirmed that both Epacridaceae and Empetraceae, along with Pyrolaceae and Monotropaceae, should be included in Ericaceae. Recently (Angiosperm Phylogeny Group, 1998
) Ericales have been more broadly circumscribed to include Primulales and Theales.
The Ericaceae are a large family in the order, ranging from 50 genera and 1350 species (Willis, 1973
) to 100 genera and 3000 species (Stevens, 1971
) or ~120 genera and 4000 species (Luteyn, 1989
) and is essentially worldwide in distribution. Although the genera placed in Ericaceae have been kept together by most taxonomists, their placement within the family has been varied. The subdivision of the family that has been most commonly followed in the past is that of Drude (1897), who recognized four subfamilies: Rhododendroideae, Arbutoideae, Vaccinioideae, and Ericoideae. More recent treatments (Watson, Williams, and Lance, 1967
; Stevens, 1971
; Willis, 1973
; Thorne, 1976
), however, while leaving Rhododendroideae and Ericoideae relatively undisturbed, agree in combining Drude's Arbutoideae and Vaccinioideae in a single subfamily—Vaccinioideae. These latter systems de-emphasize the inferior ovary and berry, which were major characteristics used in separating the vaccinioids from the other subfamilies, in favor of greater emphasis on a variety of other features of corolla, stamens, etc., in which Drude's Vaccinioideae and Arbutoideae show many similarities. The combined subfamily is a large and heterogeneous one, but Stevens (1971, whose classification will be followed) pointed out that "there are no discontinuities sufficient to separate other subfamilies." Within the subfamilies tribal delimitations have been even more variable and certain genera and/or species (such as Wittsteinia, now removed completely from the Ericales and placed in the family Alseuosmiaceae, Epigaea, Chiogenes, Cassiope, etc.) have often been moved from one tribe, or even subfamily, to another.
Thus, although Ericaceae may be considered a rather distinctive and, in general, a recognizable family, many questions about relationships and phylogeny remain unanswered; Ericales s.l. are particularly poorly understood. Taxonomic diagnoses and the battery of embryological features recognized as ericalean (Samuelsson, 1913
; Maheshwari, 1950
; Palser, 1961a, 1975
) do include several aspects of the stamens, but there is remarkably little detailed information about stamen development or microsporogenesis in Ericales. No attempt, particularly at the developmental level, has been made to try to arrive at an understanding of several peculiarities of ericaceous stamens and the distribution of these within the family (or order). In fact, stamens in Ericaceae fail in several respects to correspond to the generalized description of the stamen for the majority of angiosperms presented by Endress (1996)
.
Although gross aspects of stamens of a very large number of species have been described for taxonomic purposes, only a small proportion have been looked at for cellular details and the majority of these have been for temperate species. Generalizations are based on what has been published and may need to be modified when details for a wider variety of species are available.
Insofar as known, the basic progress of microsporogenesis in Ericaceae follows the normal pattern except that, for the most part, the microspores do not separate from one another and the pollen grains are shed as tetrads. There is, however, no clear description of anther-wall development. The few published papers (Chou, 1952
—Gaultheria procumbens; Batygina et al., 1963
—V. myrtillus and V. vitis-idaea; Stushnoff and Palser, 1969
—Vaccinium atrococcum) supply only isolated pieces of information, and these appear conflicting, i.e., wall thickness varying from two to four layers, with the tapetum derived from an inner wall layer or from sporogenous tissue. Another unusual, but long recognized, feature of certain ericaceous anthers is the presence of viscin threads (extremely thin strands) among the pollen grains. These are characteristic of several genera of Rhododendroideae (Matthews and Knox, 1926
; Bowers, 1930
; Copeland, 1943
; Stevens, 1971
; Hesse, 1980, 1986b
; Waha, 1984
; Clemants, 1995
) and have been used in delimiting tribes in this subfamily.
A general idea of variation in size and occurrence of the various peculiar features described below for ericaceous stamens can be obtained by observation of Figs. 1–18.
|
; Palser, 1951, 1954
). A few developmental studies of species in which the anthers always appear introrse, such as Erica cylindrica (Payer, 1857), E. arborea, and Vaccinium vitis-idaea (Artopoeus, 1903
), E. hirtiflora (Matthews and Taylor, 1926
), V. myrtillus (Matthews and Knox, 1926
), and V. angustifolium (Bell and Burchill, 1955
), show that these also invert, although very early in development. Some anthers appear basifixed (as in several Erica species—Palser and Murty, 1967
) and whether these also invert has not so far been determined. Developmental studies have not been carried out in most species, but since mature anthers are introrse and the vascular strand forms a hook in the connective at the point where the filament joins the anther, it is inferred that inversion has also occurred in these (Matthews and Knox, 1926
; Palser, 1958
; Palser and Murty, 1967
; Villamil, 1973
; Villamil and Palser, 1980
).
The Ericaceae are one among only a few families in which stamens are characterized by the presence of various protrusions called appendages. While some ericaceous taxa lack these staminal appendages, their shape, size, and position in the others are varied (see Matthews and Knox, 1926
; Stevens, 1971
). Those appendages called "awns," which are elongations of thecae (anther halves), occur at the apparent apex, two per anther, and can be hollow (in which case they are called tubules) or solid, single or bifurcated, or apparently fused to become one per anther. "Spurs," usually two per anther, appear on the abaxial side of the inverted anther and are solid, cylindrical, or flattened, long or short; they can also occur on the filament or tubule of some Vaccinieae. In addition to spurs and awns, in a few scattered species there are other projections or overgrowths of the base of the thecae (Vaccinium reticulatum.—Palser, 1961b
).
Ericaceous anthers commonly dehisce by apical pores or clefts, only rarely longitudinally. Dehiscence is normally clearly or obliquely introrse but may be so far to the side as to be called latrorse in a few species (Luteyn and Wilbur, 1977
; Luteyn, 1978, 1987
) and is extrorse in Notopora (Stevens, 1971
; Fig. 13). Anther walls in almost all genera lack a differentiated endothecium, and dehiscence has been reported as resulting from the formation of "collapse" tissue (Artopoeus, 1903
) or a tissue variously called "resorption" (Artopoeus, 1903
), "disjunctive" (Matthews and Knox, 1926
), or a "calcium oxalate package" (D'Arcy, Keating, and Buchmann, 1996
). The latter term was applied to and described for only one ericad (Kalmia), although the authors do not suggest that it is involved in dehiscence.
Another unusual feature of anthers, usually associated with several genera of Andromedeae (Matthews and Knox, 1926
; Chou, 1952
; Palser, 1951, 1958
; Villamil, 1973, 1980
; Dorr, 1980
; Villamil and Palser, 1980
) but appearing also in Calluna and several Central and South American Vaccinioideae (J. L. Luteyn, New York Botanical Garden, personal communication), is the occurrence of two pouches of cells containing small crystals (granular or disintegrating tissue) on the abaxial side of the anther, which eventually break down completely. Both dehiscence and granular pouches will be covered in a subsequent paper.
The common presence of appendages and the great variation in their shape, size, and position; the apparently universal inversion of the anther at some time during development; apical dehiscence associated with the absence of an endothecium and presence of collapse and/or resorption tissue; and the occurrence of granular pouches in different positions on the abaxial side in a more limited number of species make the ericaceous stamen intriguingly different from that in the majority of angiosperms. The relationship of these various aspects to one another, and of one expression of a characteristic to another—as the different types of appendages—is unknown. In an attempt to help clarify these unusual and incompletely known characteristics of the ericaceous stamen and to partially rectify the scarcity of information about anther-wall development, a study of stamen development has been undertaken in selected species of the family. Such detailed investigation may contribute some additional information which may help in arriving at an understanding of relationships within the Ericaceae. This paper is the first of two reporting on development in ericaceous stamens; it covers events in early development (anther wall, microsporogenesis, inversion, and appendages), while the second will present later occurrences (dehiscence and granular pouches). The sequential stages in microsporogenesis will be used as a reference point for timing of stages in the different aspects of stamen development studied.
MATERIALS AND METHODS
Ten species of Ericaceae (in the restricted sense, i.e., no Pyroloideae or Monotropoideae included) were selected for study: Rhododendroideae: Rhodoreae—Rhododendron maximum, Phyllodoceae—Kalmia latifolia, Cladothamneae—Elliottia racemosa; Ericoideae: Ericeae—Erica carnea, Calluneae—Calluna vulgaris; Vaccinioideae: Enkiantheae—Enkianthus campanulatus, Andromedeae—Pieris japonica and Gaultheria procumbens, Vaccinieae—Vaccinium stamineum and V. albicans. Table 1 lists the ten species plus their place of collection and collector. Figures 19–28 illustrate an essentially mature whole stamen of each. In selection of the species several things were kept in mind: the various subfamilies and tribes that have been recognized taxonomically, the wide variety of staminal peculiarities found in the family (inversion, appendages, unusual dehiscence, granular pouches, etc.) and the fact that information on development and cellular detail is scanty. The work reported here was specifically developmental. This meant that collection of materials had to start with bud initiation and continue at close intervals thereafter until at least flower opening. Therefore species selected needed to be fairly accessible to New Jersey where the work was being done. The other aspects needed to be correlated with this requirement.
|
|
Weekly collections were started during the summer of the year prior to flowering for Rhododendron, Kalmia, Enkianthus, Pieris, and Erica since floral primordia are initiated at that time. The remaining species form their buds during the season in which they flower; therefore, collections of Calluna, Elliottia, Gaultheria, and Vaccinium stamineum were started as soon as there was the slightest sign of bud initiation. Vaccinium albicans is native to New Guinea and was sent as buds and flowers in fixative; very early stages were not available.
Buds and open flowers were fixed in formalin-acetic acid-alcohol, dehydrated in an ethyl alcohol-tertiary butyl alcohol series and embedded in a 61°C melting point Tissuemat or in Paraplast (56°C). Serial longitudinal and cross sections of buds and flowers at different stages of development were cut, most of the younger ones at 5 µm but some older ones at 7, 8, or 10 µm. The sections were stained with safranin-fast green (extra bluish) and observed under a compound microscope.
Material for observations with a scanning electron microscope (SEM) was dehydrated to 100% ethanol and critical point dried using CO2. Samples were mounted with double-stick tape, sputter coated with gold-paladium, and observed on an Hitachi S450 SEM in the EM laboratory at Rutgers University, New Jersey.
Micrographs obtained with compound and scanning electron microscopes were used to record observations.
RESULTS
Throughout the Results, which are essentially restricted to the current observations, the generic epithet alone will be used for the particular species studied, unless otherwise indicated.
Gross morphology
The wide variation in gross aspects of the anthers has been covered in the Introduction. The particular species selected for study attempted to cover many of those variations. They include species with differences in time of inversion; in point of filament attachment to the anther; in absence or presence of appendages and, if present, the type(s) and their characteristics; in the shape and location of areas and mechanism of dehiscence; and in the presence or absence of granular pouches, where they occur on the anther and their structure. The present Results deal in some detail with the first three of these anther features; the latter two will be covered in a subsequent paper.
Filament characteristics were not covered in the Introduction. Among the present species the filaments are narrow, often broader near the base but abruptly narrowed at the point of origin from the receptacle (Figs. 24, 27,
64). They are dorsiventrally flattened in all species studied except Enkianthus where they are more or less cylindrical and particularly swollen near the base (Figs. 28, 55, 68). A single vascular strand supplies each stamen; it enters the base of the filament, which it traverses to enter the connective. There it bends inward and, in all except Erica, downward (Figs. 50, 58). The filaments are glabrous in Calluna (Figs. 23, 64), Elliottia (Fig. 22), and Erica (Fig. 21); in the other species, except for the short, narrow, basal portion, they are hairy. In Vaccinium albicans and Gaultheria the hairs are long (Figs. 24, 25), usually unicellular, but occasionally two- or even three-celled. They occur from the lower part of the filament to a point near the base of the anther in Gaultheria (Fig. 24), Pieris (Fig. 27), and Vaccinium (Fig. 25). Rather short hairs occur on the lower third of the long filaments of Rhododendron; the shorter filaments have hairs over most of their length (Fig. 20). In Kalmia and Enkianthus hairs are short but occur along most of the filament (Figs. 19, 28). A few multicellular glandular hairs occur scattered on the filament, tubule, and tubule extension of V. albicans (Fig. 25, upper left edge) and occasionally also on the abaxial side of the anther.
|
|
|
|
|
There are basically three patterns of filament elongation in the species studied. (1) There is a gradual increase in length of the filaments throughout stamen development in Gaultheria, Pieris, V. stamineum, and Enkianthus. At maturity, the filaments are more or less straight in side view in all these species. In V. stamineum, the gradual and slight increase in filament length is obscured by the considerable elongation of the tubules that occurs just before dehiscence. (2) In V. albicans and Rhododendron, the filaments initially elongate rather gradually, but show a short spurt of growth just prior to dehiscence; they are also essentially straight in side view. (3) In the remaining species (Kalmia, Calluna, Elliottia, and Erica) the initial relatively slow growth of the filaments is followed by very rapid elongation just prior to dehiscence. The filaments of Erica remain straight in side view. In Calluna, however, the filaments become geniculate (not shown in Fig. 23) just before flower opening. In Elliottia the filaments remain straight, but two small protuberances occur on their adaxial sides near the base of the anther, one on each side of the vascular strand. The most distinctive stamen arrangement is that in Kalmia. Here the upper portions of the anthers are lodged within pockets that form in the corolla early in development. There are ten of these pockets approximately midway between the base and free edge of the corolla (Figs. 29–30); five occur in the midvein area of each petal and five along the line of "fusion" of two adjacent petals. Just prior to dehiscence, the filaments, which curve slightly inward in young stamens, start to turn outward and then upward (Fig. 29) so that the upper end of each anther fits into one of the corolla pockets (Fig. 30), a position achieved before anthesis. When the flower opens, the anthers remain in the pockets of the broadly saucer-shaped corolla and the filaments have a flattened, elongated S-shape.
Tanniniferous substances generally occur in at least some cells of ericalean flowers. The stamens are no exception; in the species studied, tannins occur in greater or lesser amount in the filament, connective, spurs, and solid awns, but not in hollow awns (tubules). Within the anther they occur in all species except V. albicans; their location varies in the rest of the species, but they occur mainly in the epidermis. Large druses are also widespread in the family, and they occur variously placed in the stamens of most species studied, but they are absent in Pieris and in the anther, but not the filament, of Rhododendron.
Anther-wall development and microsporogenesis
The details of wall development and microsporogenesis have been followed in Calluna, Kalmia, Enkianthus, Vaccinium stamineum, and less completely in Pieris and Gaultheria. The pattern of development is similar in all six species, although sepalad stamens develop more rapidly than petalad ones in Enkianthus.
Wall development
It was of particular interest to follow anther-wall development in considerable detail to relate it to the classification system proposed by Davis (1966)
for all angiosperms. Young anthers are somewhat rectangular in cross section (Fig. 31), gradually becoming trapezoidal. In each of the four corners three to six hypodermal cells in cross section enlarge slightly (Fig. 32) to initiate the archesporium of each of the four microsporangia; in longisection approximately the same number of cells differentiates. Each archesporial cell divides periclinally to form an outer primary parietal cell and an inner primary sporogenous cell (Fig. 33). The primary parietal cells are the source of the wall layers; each divides periclinally to form two secondary parietal cells (Fig. 34). Up to this point, all species (including Gaultheria and Pieris) go through the same pattern of wall development; after this, behavior of the secondary parietal cells varies. Both cells may divide periclinally again, or only one of them may do so, either the outer or the inner one (Fig. 35), or they may not undergo any further periclinal division. Most frequently only the outer secondary parietal cells undergo another periclinal division, resulting in a three-layered anther wall: a hypodermal "endothecial" layer (differentiates as a functional endothecium only in Enkianthus) and a middle layer, and a tapetal layer differentiated directly from the inner secondary parietal cells. There are, however, many exceptions to this pattern: quite frequently some of the outer secondary parietal cells do not divide, so that at these points the anther wall consists of only two layers, the "endothecial" and the tapetal one, differentiated from the outer and the inner secondary parietal cells, respectively. Occasionally in Enkianthus, Pieris, and Kalmia, but more rarely, in Calluna and Vaccinium, both inner and outer secondary parietal cells divide at some spots resulting in a four-layered anther wall. The result may be a hypodermal "endothecial" layer, two middle layers, and tapetum (Fig. 36), or more rarely one "endothecial" layer, one middle layer, and two tapetal layers. The formation of these four-layered walls is not always a result of both inner and outer secondary parietal cells undergoing periclinal divisions. Instead, the inner secondary parietal cell may not divide, differentiating directly into tapetum, whereas the outer one divides producing an outer "endothecial" layer and an inner middle layer, and one or a few cells in this latter layer may divide again periclinally forming the two middle layers. Less frequently, the "endothecial" layer is directly differentiated from the outer secondary parietal cell/s, whereas the middle wall layer and tapetum are derived from the periclinal division of only one or a few inner secondary parietal cell/s. More than one or all of these possibilities may occur within a single microsporangial wall. In all six species the tapetum, derived from parietal cells, is supplemented by cells differentiated from the septum and connective tissues in contact with the sporogenous cells thus forming a complete tapetal ring, which may be two cells thick at some spots. The functional tapetum is of the glandular (secretory) type in all ten species. As all this is occurring, anticlinal divisions in both wall and sporogenous cells are also contributing to growth of the anther as a whole.
Microsporogenesis
The primary sporogenous cells give rise to a narrow cylinder of sporogenous tissue by a few longitudinal and more transverse divisions. The sporogenous cells enlarge and eventually become microspore mother cells. In Calluna, Vaccinium, and Enkianthus, the tapetal cells start to separate from the wall layers during the late-sporogenous-tissue to early microspore-mother-cell stages, and are seen in cross section as a ring surrounding the sporogenous tissue or early microspore mother cells. This separation occurs at the microspore stage in Kalmia and is more pronounced toward the connective than near the anther wall. At the microspore-mother-cell stage in all four species the tapetal nuclei start to divide resulting in two-nucleate cells (Fig. 37). As microspore mother cells increase in size and become surrounded by callose (Fig. 37), the tapetal cells also separate from each other and become larger and rather rectangular, very frequently slightly larger in a radial dimension. By this time the other wall layers have been considerably stretched and flattened. Microspore mother cells undergo meiosis, with cytokinesis occurring only after the second meiotic division (simultaneous). Tapetal cells attain their largest size during meiosis I in Calluna, at the dyad stage in Enkianthus, and at the early tetrad (before walls are formed) and tetrad of microspore stages in Kalmia and Vaccinium, respectively. The resulting tetrads are tetrahedral (Figs. 40–44) (occasionally decussate); the microspores do not separate from one another (Figs. 40–44)—except in Enkianthus—and their nuclei occupy a more or less central position in the cell. Again with the exception of Enkianthus, callose starts to disappear at this stage and tapetal cells reassemble to reform what, in cross section, looks like a ring that surrounds the tetrads. It is still free from the anther wall in Vaccinium, against it in Calluna, and, as stated above, just separating from it in Kalmia, but in Enkianthus, tapetal cells are loose, both separated from each other and from the anther wall. As the microspore tetrads develop thick walls, the tapetum in Calluna lies against the much flattened middle wall layers, its cells now stretching along the perimeter of the sporangium; they look fuzzy and very granular, but both nuclei can still be seen in some cells, although with some difficulty. In Vaccinium, vacuoles start to develop within tapetal cells; this is followed by some plasmolysis, the cytoplasm retracting from the inner wall. In Calluna the same occurs slightly later—at the time microspore nuclei migrate to a position closer to the outer wall. Microspores separate from the tetrads in Enkianthus, thus forming single pollen grains (Figs. 39, 45). As the microspores become free from each other, callose disappears, the tapetum returns to its initial position against the anther wall, all of its cells come together (sides touching), and they start to vacuolate. The endothecial and one middle layer are still visible (Fig. 39). Some abnormalities were occasionally found in one or more sporangia of one or more anthers in a flower of Kalmia and Vaccinium; these most commonly affected the tapetum which started to break down very early in development.
Each spore divides excentrically to form a large central vegetative cell and a small generative cell against the outer wall. The generative cell moves away from the wall to a more central position where it is surrounded by vegetative-cell cytoplasm. By this time the tapetal cells in Vaccinium and Kalmia appear rather empty. In Calluna they have mostly disappeared, though a thin film remains against the anther wall. In Enkianthus, in contrast to those three species, small vacuoles appear in tapetal cells but the two nuclei can still be seen. Thickened bars develop along walls of endothecial cells (Fig. 39); each is U-shaped, the opening toward the outer tangential wall; only one side of the U is seen in an anther cross section. Cells of one middle wall layer have enlarged somewhat so that it is still evident at this stage (Fig. 39). By the time pollen grain walls have become ornamented, the tapetum has completely disappeared; the persisting middle wall layer flattens and at maturity only the epidermis and endothecium remain.
As pollen grain walls become thicker and ornamented (ornamentation not studied), each portion of the tetrad is filled with some storage product (not identified), which makes the nuclei difficult to see. The middle wall layers and tapetum become more flattened until essentially only the epidermis is left. The "endothecial" layer seems to be the last one to crush and can still be seen in some places at the time of dehiscence in Kalmia (Fig. 38).
Equatorial colpi are readily visible in pollen of all species studied (Figs. 40–45), and they also show pori within the colpi in most species (Figs. 41, 43, 45), although this could not always be seen. The appearance of the colpi (and pori) is affected by the fact that the pollen grains are in tetrads. Because the colpi of two individual monads coincide where they are in contact (Figs. 40–44), it appears as if a colpus starts on one segment, crosses the line of contact, and reaches its other terminus on the adjacent one. This is not the case. The pore of each separate colpus is adjacent to the point of contact between two colpi (Fig. 41).
Many small globular bodies, assumed to be orbicules because of their appearance and location, were found lining the tapetal layer in Calluna and in the locules of both Calluna and Enkianthus. In addition, they have been seen in Erica, Rhododendron, and Vaccinium albicans.
Viscin threads occur in Elliottia (Figs. 40, 41), Kalmia (Fig. 42), and Rhododendron (Fig. 43). They are first recognizable as a sort of net surrounding the microspore tetrads. Ultimately they are extremely thin threads that hold the grains loosely together, so that, at maturity, when a few grains leave the anther half, the rest are drawn out with them. The viscin threads are usually difficult to see with the light microscope, and often their presence is more evident under the dissecting microscope, since it is easy to see that the grains are held together in an irregular mass, presumably by the threads. They are very clear, however, under the scanning electron microscope (Figs. 40–43).
Enkianthus is the only species that possesses a well-developed endothecium (Fig. 39); it is distinct by the pollen-grain stage. The typical endothecial thickenings occur from the base of the awn downward through the top of the anther, but they are absent from the dehiscence (stomial) zone between the two microsporangia of a theca. The dehiscence zone extends along the upper half of the anther, but below this level lignified wall thickening in cells of the endothecial layer is not of the usual ribbed pattern but is evenly laid down on the inner tangential and radial walls.
No fibrous or ribbed endothecium occurs in the other taxa studied, as is most commonly the case within the Ericaceae. In Kalmia, however, some thick-walled cells occurring locally have been interpreted as remnants of an endothecium. The hypodermal cells in the upper part of the anther, especially toward the center of the anther, have thick inner tangential and radial walls but lack the characteristic ribbed thickenings of a typical endothecium (Fig. 38). Cells in a similar position are less and less thickened toward the base of the anther until, from mid-anther down, they are thin-walled. They remain large and healthy at all levels until dehiscence, but then become crushed against the epidermis from the level of the vascular hook downward.
The epidermal cells of Elliottia anthers have a rather thick outer tangential wall. The hypodermal cells are thin-walled but remain large and healthy until dehiscence; later, they become crushed against the epidermis.
In Rhododendron, epidermal cells around the pore have thick outer tangential and radial walls. Cells of the hypodermis and one or two middle wall layers contain starch grains and remain large and healthy until the anthers dehisce; they then become crushed against the epidermis. No starch occurs in the epidermis, except in the enlarged, thick-walled cells forming the ring around the pore.
When pollen grains are still immature in V. stamineum, but close to dehiscence time, one to two hypodermal layers of the awn become evenly thick walled except directly above the groove between the two microsporangia. These cells are also found in the upper part of each theca, the area they occupy having the general shape of an inverted, lop-sided funnel, since the zone extends lower down and involves more cell layers on the abaxial side (actual position seen at maturity). There are as many as five layers of thick-walled hypodermal cells below the awn, but none close to the spurs; the cells of the latter are also thick walled except for the epidermis. On the adaxial side, cell layers go from three near the top to zero above spur level. Wall thickness decreases commensurately. Gradation also occurs between abaxial and adaxial sides both in the number of cell layers and wall thickness.
Inversion
The anthers of nine of the ten species studied invert very early in development when the bud is 0.5–2 mm long (measurements are all in length and start at the base of the calyx), the length varying with the species. Inversion starts a little before the archesporial-cell stage in Kalmia (Fig. 46) and V. stamineum (Fig. 47) and at some time between archesporial and primary parietal-primary sporogenous-cell stages in Gaultheria, Pieris, Calluna, Elliottia, Erica, and Rhododendron. The actual beginning of inversion could not be determined for V. albicans, having already occurred in the youngest material available. In contrast to these species, inversion in Enkianthus occurs just prior to anthesis.
As already pointed out, the archesporial region has approximately the same size in cross and longisection. The sporogenous cells divide more frequently transversely than longitudinally so that the sporogenous tissue increases more in length than in width. Transverse divisions also occur in cells of the epidermis and wall layers of the anthers. Depending on the relative number of the transverse divisions occurring above or below the point of attachment of filament and anther, this attachment will vary in position at stamen maturity even though it is always abaxial. Thus, in Kalmia, for example, divisions are primarily below the point of attachment, which, as a result, is close to the apex, whereas in Erica and Calluna it is at or very close to the base (Figs. 21, 23) because most divisions are above the junction. In Enkianthus, where enlargement precedes inversion, transverse divisions occur primarily from the top of attachment upward so that the latter is very close to the apex of the anther (Fig. 28) after inversion. For the same reason, also excepting Enkianthus, the two anther halves above the filament-anther junction are separate from one another to a greater or lesser extent, depending on the species, and the anther halves remain joined primarily from the filament-anther junction downward to the base.
The turn from extrorse to introrse position, as determined by the bend in the vascular bundle, is ~150°–180° in all species except Erica and Calluna. The cells on the abaxial side of the stamen primordia elongate more than those on the adaxial side, leading to curvature of the filament-connective junction and, consequently, of the young anthers with their archesporia (Figs. 48–50). Since inversion starts so early, the anthers are never strictly extrorse and the archesporia are initiated essentially one above the other, rather than one internal, one external (Fig. 51). As the anthers turn, the top archesporia become the inner microsporangia and the bottom ones the outer (Figs. 48–50).
In Erica differential growth at the filament-anther junction is less than in the other species, leading to a curvature of the vascular strand of ~45°; thus, there is no vascular hook, only an inward bend (Fig. 54). This limited curvature occurs early and results in the archesporia being obliquely one above the other (Fig. 51). With no further curvature the archesporia would not reach a completely introrse orientation. The latter is accomplished, however, through differential growth of the microsporangia themselves. Through greater growth on the abaxial side (Fig. 52), the microsporangia gradually become more and more vertical, and end up being upright and parallel (Fig. 53). Thus, as in the other species, the initially upper archesporia become adaxial and the lower archesporia abaxial. The base of the adaxial microsporangia remain higher than that of the abaxial ones, however (Fig. 53, 65). Since no transverse cell divisions occur below the point of attachment of filament and anther and all growth is upward, the anther is basifixed (Figs. 21, 54, 65), and the separation between the two anther halves is extremely deep.
In Calluna inversion is accomplished by a combination of cell elongation on the abaxial side of the stamen primordia, as in most other species, and differential growth of the microsporangia themselves, as in Erica. The archesporia differentiate essentially one above the other, but the upper ones are slightly abaxial, thus being rather more extrorse than in the other species. Elongation of abaxially located cells, however, leads to curvature of the filament-connective junction for ~90°, resulting in some of the change of position of the archesporia that occurs in eight of the other species studied. Since this curvature is not enough to place the upper archesporia in an adaxial position that is level with the abaxial archesporia, the base of the outer microsporangia remains lower than that of the inner ones. Simultaneous with this curvature, there is differential growth of the microsporangia themselves similar to that described for Erica. While most transverse cell divisions occur above the attachment of filament and anther, some do occur below this point, and the anther does not appear basifixed (Fig. 23). The downward growth includes the tip of the vascular bundle so that a real but short hook is formed.
In all cases of early inversion, the latter takes place rather rapidly since, by late sporogenous-tissue stage the anthers have reached their final introrse position.
The anthers of Enkianthus remain extrorse (Fig. 55, 68) until just before anthesis. Shortly after the end of meiosis, while some microspores have still not separated from the tetrad, the cells on the abaxial side of the filament just below the anther start to enlarge, forming a slight protuberance (Fig. 56). As those cells continue to enlarge, the anther turns inward (Fig. 57, 58). Early stages of inversion occur while the bud is still closed, but inversion is not complete until the flower opens.
Appendages
Stamens of four of the species studied possess awns—hollow in both species of Vaccinium (Figs. 25, 26), solid in Enkianthus (Fig. 28), and hollow at the base but solid above in Gaultheria (Fig. 24). Three species have spurs—Pieris (Figs. 27, 67), Calluna (Figs. 23, 64), and V. stamineum (Figs. 26, 63). Only the latter has both spurs and awns. Appendages are not characteristic of stamens of Kalmia (Fig. 19), Rhododendron (Fig. 20), Erica (Figs. 21, 65), and Elliottia (Fig. 22). In Elliottia, however, short upward extensions of each anther half (two per anther) and short basal extensions of each microsporangium, thus four per anther, occur. With the exception of Enkianthus, where inversion occurs at anthesis, and Calluna, where the spurs are initiated during inversion (Fig. 76), there is no external evidence of appendages on stamens of the species studied until after the anther has inverted.
|
In Enkianthus meristematic cells at the morphological base of each (uninverted) theca divide transversely to initiate the awn. When the anther has reached the microspore-mother-cell stage, cells in the center of the awn begin to enlarge. The peripheral cells continue to divide until the late microspore stage, at which time divisions stop and elongation starts in the peripheral cells while continuing in the central ones (Fig. 55). When the anther starts to invert, which occurs at the end of meiosis, a group of cells at the tip of the anther (morphological apex, apparent base when inversion is complete) elongate to form an extension of the connective (Fig. 55, 68); this extension remains solid. When the anthers are almost half turned, the walls of the long central cells of each awn start to thicken evenly and the hypodermal cells in the base develop bar thickenings (endothecial). The awns are about the same length as the microsporangia (Figs. 28, 68).
The awns of both species of Vaccinium are initiated by cell division in residual meristems at the anther apex after inversion has been completed (Vaccinium stamineum; Fig. 72). At the late microspore-mother-cell stage in V. stamineum, cells of a few subepidermal layers of the developing awn elongate considerably, while the central ones lag behind resulting in a central depression (Fig. 59, 60) lined by a layer of small cells (Fig. 70). Epidermal cells continue to divide, keeping pace with awn elongation. Elongation of the peripheral cells continues to exceed that of the central ones, so that the depression is accentuated and finally extends about one-quarter the length of the awn. The depression is associated with dehiscence and its behavior will be described in that connection in another paper. The awns, laciniated at the tip (Figs. 26, 62, 63), are approximately two and a half to three times longer than the microsporangia (Fig. 26). Initially, the growth of the awn or tubule of V. albicans is similar to that of V. stamineum (cell division and some elongation), but at the microspore-mother-cell stage, some of the meristematic cells situated on the abaxial side of the awn tip start to elongate producing an awn extension, while the rest remain meristematic and small (Fig. 71). By the tetrad stage, cells of the awn extension have elongated considerably and those that remained meristematic at its base, except for the epidermis, start to form small crystals. The mature awns are about the same length as their extensions, and together they approximate the length of the microsporangia (Fig. 25). In both species the awns become hollow; this, and the formation of small crystals in V. albicans are related to dehiscence and will be covered in another paper.
Spurs also start as small protuberances, but in this case on the abaxial side of the just-inverted stamens (Fig. 72) (partially inverted in Calluna; Fig. 76). In Pieris, in which anthers are characterized by both spurs and granular pouches (Figs. 27, 67), the two appear to be initiated as a common structure (Fig. 77). At late sporogenous-tissue to early microspore-mother-cell stage, a group of residual meristematic cells on the upper abaxial side of each anther half begins to protrude downward and outward as a result of elongation of the cells beneath it (Fig. 77). The extent of this zone is difficult to determine because, as indicated above, it emerges as a single outgrowth with the granular pouch that extends upward. The cells in the center of the downward-protruding region continue their elongation, while the meristematic cells covering them (one to three layers including the epidermis) continue to divide (Fig. 77). Later, division stops and the cells start to elongate. After considerable elongation of all cells, and, therefore, of the spurs, the epidermal cells become papillate and all cells moderately thick walled. The spurs are 8–12 cells in diameter at the point of attachment to the connective where they are largest. They narrow toward the tip and are approximately the same length as the anther (Fig. 27). Granular tissue occurs at the extreme top of the spur (Figs. 27, 67). The time, place, and mechanism of spur initiation in the anthers of V. stamineum are similar to those in Pieris. The points of initiation are close below the developing awns (Fig. 72). The major difference is that the meristematic cells, which occur in a roughly circular group of 8–12 cells in diameter, are carried primarily outward and to the side of the anther half and only slightly downward. The meristematic epidermal cells divide for some time to keep pace with elongation of the central cells (Fig. 73). When the epidermal cells stop dividing and start to elongate, those occupying a lower position in the spur elongate more than those on the upper side so that the spur curves upward (Fig. 74). The epidermal cells become papillate and all cells evenly thick walled (Fig. 75). The final mature spurs are approximately the same length as the anther, not including the awn (Fig. 26).
In Calluna the spurs are initiated by radial elongation of a group of epidermal cells (six to eight in longisection) at the early sporogenous-tissue stage (Fig. 76) before inversion has been completed. These epidermal cells, which involve the lower abaxial side of the anther, the connective region and a small portion of the top of the filament, form a small downward-projecting protuberance. Beneath the epidermal cells there is a layer of a few meristematic cells, which undergo only a few transverse divisions; as a result most growth of the spur results from cell elongation. This is confirmed at maturity when only four to five very long cells can be seen in longisection. The spurs are four cells thick dorsiventrally and 11–12 cells wide, thus accounting for their flatness. The spread is mostly tangential. They may be bifurcated at the tip for a short distance (Figs. 23, 64). Since spur cells are larger in Calluna than in Pieris or V. stamineum, the spurs of Calluna appear larger, although cell number is similar in all three. The spurs in Calluna attain a length at maturity that approximates that of the anther; their cells become evenly thick walled and the epidermis papillate (Figs. 23, 64). Granular tissue occurs at the extreme top of the spurs (Figs. 23, 64) and is considered with granular pouches elsewhere.
The short basal extensions of the anthers in Elliottia become recognizable approximately at the early microspore-mother-cell stage. They are formed in part by elongation of anther-wall cells at the base of each microsporangium but primarily by considerable radial elongation of epidermal cells in this region.
None of the appendages studied is vascularized.
DISCUSSION
Having presented details of some of the peculiar features of ericaceous stamens, it is of interest to consider the distribution of the various characteristics discussed above among taxa within Ericaceae and its close relatives, and in other angiosperm families.
Microsporogenesis
Sporogenous development of the species studied here is similar to that reported earlier for Gaultheria procumbens (Chou, 1952
), Vaccinium atrococcum (Stushnoff and Palser, 1969
), V. nummularia, V. retusum, and V. serratum (Venkateswarlu and Maheshwari Devi, 1973
), several species of the Phyllodoceae (Ganapathy and Palser, 1964
), and eight species of Gaultheria and Pernettya (= Gaultheria) (Cambi, 1999
). Thus, the number of species known to be characterized by simultaneous cytokinesis to form permanent tetrahedral tetrads (plus a glandular binucleate tapetum—a wall characteristic) is increased. Although the permanence of the tetrads is unusual, sporogenous development corresponds closely to that in most angiosperms. Permanent tetrad pollen is typical of all Ericaceae with the exception of Enkianthus and a few Ericoideae (Erdtman, 1952
; Safijowska, 1960
; Palser and Murty, 1967
) in which single grains occur. The Pyroloideae and Monotropoideae have often been included in the family (as does Stevens, 1971
) and are also subfamilies in the more recently expanded Ericaceae. The former, except for Orthilia secunda, also shed pollen in tetrad form. The Monotropoideae (Copeland, 1934, 1937, 1938, 1939, 1941
), however, all have single grains. In families of the order more recently added to the family Ericaceae, such as Empetraceae (Davis, 1966
) and Epacridaceae (Paterson, 1961
), pollen is often shed in tetrad form. In the latter family Smith-White (1955a, b)
has clearly shown that the apparently single grains of several species are actually pseudomonads (cryptotetrads). All other families that have sometimes been associated with Ericales have single grains: Clethraceae (Kavaljian, 1952
), Diapensiaceae (Palser, 1963
), Cyrillaceae (Copeland, 1953
; Vijayaraghavan, 1969
), and Actinidiaceae (Erdtman, 1952
; Vijayaraghavan, 1965
). Outside the Ericales, occurrence of permanent tetrads is characteristic of only a few whole families (Davis, 1966
), while in a few others they occur in some genera or species but not in all (Davis, 1966
; Skvarla, Raven, and Praglowski, 1975
). In some of these families (Leguminosae, Orchidaceae) clusters larger than tetrads occur in several species. Larger clusters have never been reported in the Ericaceae; the closest approach to the many-celled massulae or pollinia is the loose association of a large number of separate tetrads occasioned by the presence of viscin threads. The tricolp(or)ate aperture of pollen grains found in the species here studied agree with those of 56 other species of the family (Erdtman, 1952
) and over 400 species of vaccinioids (J. L. Luteyn, New York Botanical Garden, personal communication). All ericaceous pollen grains so far studied are binucleate (Brewbaker, 1967
) with the exception of Enkianthus in which they are trinucleate (Safijowska, 1960
).
Among the Ericaceae viscin threads have been reported only in the subfamily Rhododendroideae (Bowers, 1930
; Copeland, 1943
; Ikuse, 1954; Maguire, Steyermark, and Luteyn, 1978
; Waha, 1984
; Clemants, 1995
; Judd, 1995
), and this distribution pattern has been corroborated here: viscin threads were found only in Rhododendron, Elliottia, and Kalmia, all of them Rhododendroideae. The only exception so far reported (Waha, 1984
) seems to be one species of Gaylussacia (Vaccinieae). The threads tend to hold the pollen grains loosely together so that those of an anther half are shed as a single large mass or in several moderately large clusters. Copeland (1943)
thought that Kalmia did not have viscin threads, but they were found in that genus by Ikuse (1954) and Judd (1995)
as well as in the present study (Fig. 42). In addition, scanning electron micrographs of Kalmia pollen (Southall and Hardin, 1974
) suggest that there may be some viscin threads with the grains of K. aggregata, K. ericoides, and K. simulata, although the text says that they do not occur except in K. latifolia, a species the authors did not study but cited from Wood (1961)
. As pointed out by Skvarla et al. (1978)
, viscin threads in the Ericaceae occur only on tetrad pollen, are attached to the distal polar surface, and all appear to show a smooth surface morphology and may not be found in all species of a genus. Outside the Ericaceae they characterize the Onagraceae, but in this family they occur on the proximal pole and on both tetrad and single grains. In addition, viscin threads of Onagraceae were shown to be rather complex in structure and to differ among the genera. Other threadlike exinal connections among pollen grains have been reported in Leguminosae (Cruden and Jensen, 1979
), but these are considered to be distinct from viscin threads (Skvarla, Raven, and Praglowski, 1975
). Threads of epidermal origin, and therefore very different from viscin threads in the Ericaceae, occur in anthers of Strelitzia—Strelitziaceae, Monocotyledoneae (Kronestedt and Bystedt, 1981
); they serve a similar function, however.
The small globular bodies seen in microsporangia under the light microscope, and thought to probably be orbicules (Übisch bodies), were found in Calluna, Enkianthus, Erica, Rhododendron, and V. albicans, both next to the tapetum and close to the pollen grains. They have been reported to appear in anthers of "many genera of angiosperms, both monocotyledons and dicotyledons, and some gymnosperms" (Echlin, 1971
), and in fern sporangia (Hesse, 1986a
). Pacini (1997)
reported that orbicules have been observed only with secretory tapeta, which fits with these five species, but he also indicated that they have never been observed with viscin threads. If the bodies seen here are orbicules, they now have been seen with viscin threads (Rhododendron).
Anther-wall development
Reports on the details of anther-wall development in the Ericaceae are almost nonexistent in the literature. The present description, one of the few in which development has been followed in detail, clearly shows that, at least in four species, it cannot be classified in any of the four types of development recognized by Davis (1966)
. While the initial divisions (up to secondary parietal and sporogenous cells) are found here and in Davis' types, subsequent development in the species studied here is rather erratic so that it does not follow a fixed pattern that might be expected of the types in Davis' classification. The same has been found by Cambi (1999)
for eight species of Gaultheria and Pernettya (= Gaultheria). Batygina et al. (1963)
, who did describe anther development in Vaccinium myrtillus and V. vitis-idaea in some detail, reported that tapetal cells originated from sporogenous tissue. Since Davis' classification is predicated on a parietal origin for the tapetum, Batygina et al.‘s description cannot be made to conform to one of her four types of development. There are very few other reports of a sporogenous origin of the tapetal layer (Steffen and Landmann, 1958
), and, in general, Davis' assumption that the tapetum is derived from the parietal layers appears to hold.
In Davis' summary of embryological characteristics of the Ericaceae, anther-wall development is described as being of the Reduced type—neither secondary parietal cell divided; this was based on Davis's interpretation of an illustration of Gaultheria procumbens by Chou (1952)
; there are no details in the text. The statement by Stushnoff and Palser (1969)
that "periclinal divisions appear to be restricted to the layer adjacent to the sporogenous tissue" would suggest a Monocotyledonous pattern of wall development (only the inner secondary parietal cell divided) for Vaccinium atrococcum. Another three Vaccinium species—namely, V. nummularia, V. retusum, and V. serratum—were reported by Venkateswarlu and Maheshwari Devi (1973)
to have anthers whose "wall consists of the epidermis, two middle layers and tapetum," and the latter "is parietal in origin.... " Since their description gave no further details and their drawings did not provide them, it is not possible to determine whether anther wall development conforms to the Mono- or Dicotyledonous type of development of Davis' classification (1966). Ganapathy and Palser (1964)
merely stated that there were three to four wall layers in anthers of the species of Phyllodoceae they examined, and it is impossible to arrive at a classification from the illustrations. Budell (1964)
and Yakobson (1968)
have both studied stamen development in Rhododendron; their descriptions, however, are difficult to follow and therefore development cannot easily be assigned to one of Davis' types.
Davis' classification suggests that species, and to a certain extent even families, may be consistent in their type of wall development. Our observations and those of Cambi (1999)
do not confirm this, but show that at least within 14 species of Ericaceae variation is within species, within anthers, and even within microsporangia. However, characteristics of microsporogenesis and anther wall—such as glandular binucleate tapetum, absence of a fibrous endothecium, simultaneous cytokinesis to form tetrahedral tetrads, and tricolporate binucleate pollen grains shed as tetrads—are essentially consistent throughout the family as far as is known.
Inversion
All species studied here show inversion of the anthers to an introrse position. In all but one this occurs very early in development, in Enkianthus at anthesis. These observations corroborate the few published developmental studies. The first such study was that of Payer (1857), who prepared drawings of Erica cylindrica, a species with dorsifixed anthers. Although he stated that development of the anther was very peculiar and could not be readily understood, he did not specify the nature of the peculiarities. The drawings, however, do show different stages in the inversion of the anther at the gross morphological level. More detailed descriptions were published later by Artopoeus (1903)
for E. arborea and Vaccinium vitis-idaea, by Matthews and Taylor (1926)
for E. hirtiflora, and Matthews and Knox (1926)
for V. myrtillus. Gross morphology has shown that in Arbuteae (Palser, 1954
), as well as in Enkianthus, inversion occurs at anthesis, while in Cassiope (Cassiopeae) partial inversion occurs late in development but before anthesis (Palser, 1951
). No matter how early or late in development these anthe
