Megasporogenesis, Megagametogenesis and Ontogeny of the Aril in Cytisus striatus and C. multiflorus (Leguminosae: Papilionoideae)

Abstract

• Background and Aims There are few embryological reports on wild legumes and even fewer on their seminal appendages. There are no existing studies on the complete ontogeny of these appendages in Cytiseae, a very important Papilionoideae tribe in Mediterranean ecosystems. In this work megasporogenesis, megagametogenesis and aril ontogeny were studied in Cytisus multiflorus and C. striatus, endemics from the western Mediterranean region.

• Methods Ovaries and ovules from flower buds, flowers at anthesis and hand cross-pollinated flowers were sectioned with a rotary microtome and studied under light and fluorescence microscopy.

• Key Results A monosporic Polygonum-type of megagametogenesis is observed in both species but with megasporogenesis characterized by formation of a triad of cells after incomplete meiosis. The original cell wall of the megaspore mother cell and triad, including the transverse walls between the latter, are surrounded by a callose layer that isolates them from the surrounding diploid tissue; this callose layer gradually disappears during embryo sac formation. There are no antipodals in the mature embryo sac. Aril ontogeny starts in pre-anthesis with the formation of the aril primordium, and its normal development will occur only after fertilization, more specifically after endosperm initiation. After fertilization, a reactivation of meristem capacity takes place in the aril cells resulting in slow and sparse growth. Later, this type of development gradually decreases but the aril cells continue to grow by cell expansion, which in the last period of seed development is the only type of growth of the aril. In the mature seed, the seminal appendage acquires an irregular U-shape in transverse section, showing vacuolated cells with a large central vacuole that stores lipids and some proteins.

• Conclusions Meiotic triad formation is due to a failure in meiosis II of the chalazal cell of the dyad. In Cytisus seeds the aril has a funicular origin with predominantly post-fertilization development, but a normal growth of the endosperm is needed for proper aril development.

INTRODUCTION

The Fabaceae (= Papilionaceae) are composed of approx. 12 000 world-wide species (Polhill and Raven, 1981), occurring predominantly in tropical lowlands and the temperate regions, with their representation in the Mediterranean region being very important. Their most distinguishing characteristic is the presence of papilionaceous flowers with monocarpelate ovaries and a legume fruit type. The ovules are anatropous, hemianatropous, amphitropous or campylotropous (Bocquet and Bersier, 1960, cited in Prakash, 1987), and always bitegmic and crassinucellate, although sometimes tenuinucellate ovules have been described (e.g. soybean, Lersten, 2004) and with a zig-zag micropyle (Cronquist, 1981; Prakash, 1987).

Embryologically, the family is characterized by the presence of a monosporic, seldom bisporic (Guignard, 1881; Rembert, 1966, 1967, 1969), megagametogenesis. This begins from a multicellular archesporium (Prakash, 1987), one of whose cells undergoes considerable enlargement and is accompanied by nuclear changes to develop into the megaspore mother cell (MMC), which after meiosis generally becomes a linear or T-shaped megaspore tetrad. However, the presence of triad cells has been described in Cytisus laburnum, Phaseolus multiflorus (Guignard, 1881), Medicago arborea (Cooper, 1933), Ph. vulgaris (Brown, 1917), Robinia pseudacacia and Vicia villosa (Rembert, 1969). Usually, it is the chalazal megaspore that functions and the three micropylar ones degenerate (Prakash, 1987; Lersten, 2004), and sometimes degeneration starts before all four megaspores are completely formed (Lersten, 2004).

In angiosperms, descriptions of callose wall formation during megasporogenesis are common. This carbohydrate surrounds all the four megaspores at first, and sometimes even the megaspore mother cell before meiosis, but later it dissolves only from around the functional megaspore (Lersten, 2004), as in maize (Russell, 1979). Alternatively, the callose wall may be associated only with the transverse walls, as in soybean (Kennell and Horner, 1985; Moço and Mariath, 2003). Nevertheless, observation of callose during megasporogenesis is little documented (Lersten, 2004).

In legumes, embryo sac development is normal (Polygonum-type), with nuclear endosperm and an Asterad-type of embryo development, although Onagrad, Caryophyllad and Solanad types have been also reported (Dickison, 1981; Prakash, 1987) according to Johansen's system (1950). The functional megaspore undergoes three mitotic divisions and becomes an eight-nuclei embryo sac, which shows a classic egg apparatus (one egg and two synergids, sometimes with a filiform apparatus), two polar nuclei that fused before fertilization, and three usually ephemeral antipodals (e.g. Martin, 1914; Reeves, 1930; Johansson and Walles, 1993; Cameron and Prakash, 1994; Faigón-Soverna et al., 2003; Moço and Mariath, 2004). The latter degenerate even before fertilization (Lersten, 2004). Nevertheless, Cameron and Prakash (1994) observed giant antipodal cells in megagametophytes of several genera belonging to two Australian endemic tribes, Mirbelieae and Bossieaeae.

The fertilized ovules give rise to seeds of a broad morphological variety in the family, both in size and in testa characteristics, including colour, embellishment, presence of hilum and lens—all characters sometimes considered from a taxonomic point of view (Gunn, 1981; López et al., 2000 and references therein). Hilar appendages of the seed merits special mention since they are present in at least 12 tribes of Fabaceae s.s. (= Papilionaceae; Polhill and Raven, 1981). This feature is not exclusive to this family because it has been indicated in several unrelated families (Lersten, 2004). According to Corner (1976) the term aril is used to define a structure ranging from fleshy to more-or-less hard consistency that develops from some part of the ovule, or funiculus after fertilization, and surrounds the seed totally or partially. In fact, Corner pointed out several possible origins for the aril (funiculus, exostome, raphe or some other parts), mentioning that the true aril is an outgrowth of the funiculus, while an arillode is an outgrowth of the exostome (see Gunn, 1981). Nevertheless, there is a confusing and variable terminology that has been created through the history of descriptions of the aril (Planchon, 1845; Baillon, 1876; Corner, 1949, 1953, 1976; van der Pijl, 1955; for a concise review see Kapil et al., 1980).

The literature indicates the presence of aril-like appendages of funicular origin in the family (Corner, 1949, 1951; Polhill, 1976; Prakash, 1987). Indeed, Corner (1949) in ‘The Durian theory’ defined the Papilionaceae as an arillate family showing a rudimentary aril. He also pointed out that many papilionaceous species have small, greenish, yellowish or white arils surrounding the hilum as a rim (e.g. Cytisus), and in every papilionaceous seed that he studied he found a rim-aril (even if microscopic) and concluded that every papilionaceous hilum has or had a rim-aril at least, if not a fully developed aril. On the other hand, van der Pijl (1955) pointed out in leguminous seeds the presence of two different attractive layers for animals in zoochorous diaspores (imitation-sarcotesta, and arillus with two subtypes: true aril from the chalazal part and a fleshy funiculus). According to Gunn (1981), while the origins of legume arils are not clearly established, the recommendation is to use the generic term aril. Sometimes the term strophiole has even been used, but this term is equivalent to the lens or ‘boss’, and this confusion has severely limited its use in the modern lexicon of seminal morphology (Gunn, 1981). In legumes from the Iberian Peninsula, Talavera et al. (1999) pointed out the presence of a well-developed seed appendage (strophiole) in only one tribe (Cytiseae) and an inconspicuous one in the species Erophaca baetica of the tribe Astragaleae. Genera such as Cytisus, Pterospartum, Teline, Ulex, Stauracanthus and Cytisophyllum show arilate seeds (López et al., 2000) or strophioles (cf. Talavera et al., 1999), although the presence of small, inconspicuous or vestigial arils has been noted in species belonging to the genus Genista (Cytiseae; Gibbs, 1974) as G. tinctoria.

This paper focuses on two species of the genus Cytisus, C. multiflorus and C. striatus. The first is an endemic shrub of the western Iberian Peninsula, with white flowers and a valvular type of pollen presentation (López et al., 1999; Rodríguez-Riaño et al., 2004). The second is an endemic shrub of the western Iberian Peninsula and north-west Morocco, with yellow flowers and an explosive type of pollen presentation (López et al., 1999; Rodríguez-Riaño et al., 1999a). Both species show campylotropous ovules (Rodríguez-Riaño et al., 1999c) and seeds with a developed appendage (Talavera et al., 1999; López et al., 2000) used by the plant to spread the diaspores by means of ants (Moreno et al., 1992; T. Rodríguez-Riaño, pers. obs.).

Both species show pre- and post-zygotic rejection of self-pollen (Rodríguez-Riaño et al., 1999a, 2004), and it is quite common to observe aborted ovules both with and without an aril in mature fruit. For this reason, in previous works (Rodríguez-Riaño et al., 1999a, 2004), it was hypothesized that those with an aril were really fertilized ovules that later aborted, in which the seminal appendage development was a consequence of fertilization, as opposed to those without, which had never been fertilized. The main aim in the present work was therefore to investigate the origin, location and complete development of the seminal appendage (aril) in the two species. A second aim was to study the megasporogenesis and megagametogenesis in both species and to check whether there are variations with respect to the typical model of the subfamily Papilionoideae, for which most studies have been based on plants of agricultural interest, with few focused on wild species.

MATERIALS AND METHODS

To study megasporogenesis, megagametogenesis and aril ontogeny, flowers at anthesis and flower buds at different stages of development were analysed. Pre-anthesis stages [flower buds ranging from 2·1 to 13 mm long in Cytisus multiflorus (L'Hér.) Sweet and from 1·8 to 20 mm in C. striatus (Hill) Rothm.] were studied on a total of 244 ovules corresponding to 41 flower buds from eight individuals of C. striatus and 349 ovules of 44 flower buds from six individuals of C. multiflorus. The anthesis stage (before pollination) was studied on 22 and 18 flowers and a total of 79 and 77 ovules from seven and five individuals of C. striatus and C. multiflorus, respectively. Finally, post-pollination stages were studied on 122 and 205 flowers and 226 and 355 ovules from five and six individuals of C. striatus and C. multiflorus, respectively. The flowers were hand cross-pollinated according to the method of Rodríguez-Riaño et al. (1999a, 2004), and collected at 24 h intervals during the first 15 d after pollination (DAP), at 48 h intervals until 30 DAP, and at 4–5 d intervals until fruit maturity. Samples were fixed in FAA₉₆ (1 : 8 : 1), stored in 70 % ethanol, then gradually dehydrated in an ethanol series before being embedded in resin (Leica Historesin), and 3-μm thick sections were obtained with a rotary microtome (Leica RM2145). In the post-pollination phases, anatomical aspects of ovule, embryo and endosperm development were scored to determine whether or not the ovule was fertilized.

Sections were stained either with PAS (periodic acid Schiff) and contrasted with Gill no. 3 hæmatoxylin solution for starch and general visualization of tissues (I. Casimiro, Universidad de Extremadura, Spain, pers. com., modified), or with ruthenium red contrasted with toluidine blue O for visualization of tissues (Crivellato et al., 1990, modified), or with PAS and amido black 10B for starch and proteins (Yeung, 1984, modified). Permanent preparations were mounted in Eukitt. All the sections were studied under a light microscope. Some samples were stained with aniline blue for study with fluorescence microscopy (Rodríguez-Riaño et al., 1999a, 2004). To test for the presence of lipids, the aril was removed from mature seeds, rinsed in 0·1 M PBS, pH 7·4 at 4 °C, cryoprotected in 10 % sucrose solution in PBS, soaked in embedding medium (10 % sucrose plus 10 % gelatine in PBS), frozen in isopentane cooled to −70 °C by dry ice, and sectioned at 15 μm with a cryostat (Leica CM1900). Sections were collected on Super-Frost Plus slides (Sánchez-Calderón et al., 2005), and stained with Sudan IV (Johansen, 1940).

To quantify aril development, the width (W) and thickness (T) was measured in longitudinal sections and the number of cell layers of the chalazal part of the aril was counted. This quantification was performed on sections where the vascular bundle was clearly observed ending at the chalaza (Fig. 1A, B).

Fig. 1.

Idealized diagram of two ovules of C. striatus showing the measurements made on the chalazal side of the aril (T: thickness; W: width). (A) Ovule in pre- and post-fertilized phase (initial period). (B) Ovule in post-fertilized phase (intermediate and final periods).

RESULTS

All results refer to both species, except where differences between them are indicated.

Ovule development

Ovule primordia arise from the placental tissue of the ovary in two rows as small protuberances curving towards the stylar end, and they show early rudiments of the integument from epidermal cells (Fig. 2A).

Fig. 2.

Megasporogenesis. Longitudinal sections of ovules of C. striatus (A, B, D, E, H, I) and C. multiflorus (C, F, G). (A) Integument initiation in an ovule primordium. (B) Megaspore mother cell stage (MMC). (C) MMC surrounded by callose. (D) Dyad of meiocytes with the chalazal cell in meiosis II. (E) Triad. (F, G) Triad of cells and transverse walls surrounded by callose layer (photographs of the same ovule, but not consecutive). (H, I) Gradual callose wall disappearance in different stages of embryo sac formation: (H) two-nuclei stages and (I) eight-nuclei stage. ^*, micropylar area of the ovule; black arrowhead, chalazal cell of the dyad; es, embryo sac; fm, functional megaspore; ii, inner integument; m2, intermediate megaspore; m3, micropylar cell; nd, nucellar cells degradation; oi, outer integument. A, B, D, E stained with PAS and Gill no. 3 hæmatoxylin solution, and C, F–I stained with aniline blue; fluorescence microscopy. Scale bars: A–G = 30 μm; H, I = 50 μm.

Megasporogenesis

An archesporial cell divides into some parietal cells and one (sometimes two) sporogenous cell enlarges and becomes the megaspore mother cell (MMC). This is located in the nucellus, partially covered by the developing integuments, and it has a prominent nucleus (Fig. 2B). It is surrounded by callose before meiosis, as shown by fluorescence microscopy (Fig. 2C). The MMC undergoes the first meiotic division to form a dyad (a large chalazal cell and a small micropylar one) and, after this point, the chalazal member of the dyad seems to undergo the second meiotic division, while the micropylar one remains undivided (Fig. 2D) resulting in a linear triad of cells (Fig. 2E). It is important to note that megaspore tetrads were never observed in any sample, even in fluorescence samples, and always the chalazal member becomes the functional megaspore whilst the rest degenerate (Fig. 2E, F). A callose envelope was observed between the protoplasts and the original walls of the triad cells, including on the transverse walls between them, with the degenerating members of the micropylar pole being more strongly stained. The functional megaspore and also the different stages of embryo sac formation (two-, four- and eight-nuclei phases) keep a discontinuous and lightly stained callose wall (Fig. 2G, H), with its disappearance occurring gradually. At the moment of embryo sac initiation, degradation in the nucellar cell in contact with the embryo sac begins to be observed (Figs 2H, I, and 3A, B).

Fig. 3.

Megagametogenesis and mature ovules. Longitudinal sections of ovules of C. striatus (A–E, G, H) and C. multiflorus (F). (A) Two-nuclei embryo sac stage. (B) Eight-nuclei embryo sac. Only three nuclei are shown in this section. (C) Chalazal region of the embryo sac showing ephemeral antipodals. (D) Ana-campylotropous, bitegmic and type I nucellus ovule with a zig-zag micropyle and aril primordium. (E) Mature embryo sac. (F) Polar nuclei not fused. (G) First periclinal division of an aril primordium cell. (H) Periclinal and oblique division of aril primordium cells. an, antipodals; arp, aril primordium; ^*, micropylar area of the ovule; black arrowheads, starch grains; cc, central cell secondary nucleus; ch, chalaza; cv, central vacuole; ec, egg cell nucleus; es, embryo sac; f, funiculus; fa, filiform apparatus; ii, inner integument; mi, zig-zag micropyle; nd, nucellar cell degradation; nu, nucellus; oi, outer integument; pn, polar nuclei; r, raphal area; sy, synergids; white arrowheads, chalazal area of the ovule. All sections stained with PAS and Gill no. 3 hæmatoxylin solution, except F, stained with toluidine blue O and ruthenium red. Scale bars: A, B, F = 30 μm; C = 10 μm; D = 50 μm; E, G, H = 15 μm.

Megagametogenesis

The functional megaspore undergoes three successive mitotic divisions to produce an 8-nucleate megagametophyte. The first two nuclei migrate to the poles of the embryo sac cavity and become separated by a central vacuole (Fig. 3A). These two nuclei divide to form a 4-nucleate embryo sac, and these four nuclei finally divide to produce an eight-nucleate embryo sac, with four nuclei in the chalazal pole and four in the micropylar one, and always with a large central vacuole (Fig. 3B). During the whole of this process, the embryo sac enlarges and the eight nuclei finally undergo reorganization and cellularization. One nucleus from the chalazal pole and another from the micropylar one migrate to the centre of the embryo sac to form the polar nuclei. The three remaining chalazal nuclei form the ephemeral antipodals cells (Fig. 3C), and the three remaining micropylar ones form the egg apparatus, whose cells have most of their volume occupied by a vacuole. In short, both species show a monosporic and initially seven-celled/eight-nucleate Polygonum-type of embryo sac.

Cytisus ovules

Cytisus multiflorus and C. striatus ovules are ana-campylotropous, with a single vascular bundle that crosses the funiculus and ends at the base of the nucellus (chalaza), bitegmic and crassinucellate (type I nucellus, see Nikiticheva, 2002; Fig. 3D). Explicitly, the funiculus is the stalk that attaches the ovule to the placenta while the prolongation of the funiculus running along the future seed and ending in the chalaza is considered to be the raphe (raphal area; Fig. 3D). In the unfertilized ovule the raphe is not clearly distinguishable, but it is after fertilization and during development of the seed (see below). The inner and outer integuments are two-layered, except in the micropylar region where they may become multi-layered and form a zig-zag micropyle. However, because of the curvature of the ovule, the funicular side of the outer integument is far less developed, being practically reduced to the micropylar area, and is thus mostly multi-layered (Fig. 3D). Starch grains occur in nucellar, outer integument and chalazal cells of the ovule.

At the time of the initiation of anthesis (in flowers that have not been pollinated), the mature embryo sac is formed by 4–5 nuclei and four cells (Fig. 3E): the three-celled egg apparatus and central cell. The three-celled apparatus consists of an egg cell with most of its volume occupied by a vacuole situated at the micropylar pole, which displaces the nucleus towards the chalazal portion (near the nucleus of the central cell), and two synergids with a well-developed filiform apparatus each, with their nuclei displaced towards the micropylar pole (Fig. 3E). The central cell consists of two polar nuclei that may be fused (always in C. striatus) or not (only observed sometimes in C. multiflorus; Fig. 3F). In this phase there are no antipodals because in both species these degenerate before anthesis. Starch grains occur in central and egg cells, and are more abundant surrounding the nuclei of both cells, with those of the central cell being larger (Fig. 3E).

After pollination, ovule fertilization was verified by the presence of a zygote, or developed embryo, or the observation of normal endosperm development. This latter process showed the following phases: (a) first endosperm nucleus phase (Fig. 4A); (b) two-nuclei phase with one micropylar and one chalazal nucleus (Fig. 4B); (c) four-nuclei phase forming a rhombus-shaped figure in the embryo sac after the synchronous mitosis of the nuclei (Fig. 4C); and (d) multinucleate phase with randomly dispersed, well-stained nuclei in the embryo sac, with well-defined cytoplasmic connections. These phases were accompanied by the disappearance of starch grains that surrounded the central and egg cells in non-fertilized ovules. Abnormal endosperm development was considered to be characterised by the presence of an undivided first endosperm nucleus (very large nucleus, see Fig. 6D) in conjunction with a developing embryo, by the presence of two-, four- and even eight-nuclei phases but without nuclear reorganization (e.g. with duplication of genetic material but without karyokinesis; Fig. 4E) or, in later phases, by the presence of lightly stained nuclei with poorly-defined cytoplasmic connections.

Fig. 4.

Endosperm development. Longitudinal sections of fertilized ovules of C. multiflorus (A,B) and C. striatus (C–E); (A–D) show normal and (E) abnormal endosperm development. (A) Embryo sac with first endosperm nucleus (5 DAP). (B) Embryo sac with two-nuclei endosperm (5 DAP). (C) Embryo sac with two-nuclei endosperm undergoing synchronous mitosis to become a four-nuclei endosperm phase (5 DAP). (D) Embryo sac with a multi-nucleate endosperm (10 DAP). (E) Embryo sac with two nuclei, one micropylar and the other chalazal, tetra-nucleolate (5 DAP, only the chalazal is shown): only three nucleoli are shown in this section. ^*, micropylar area of the ovule; ce, chalazal nucleus of binucleate endosperm; ced, chalazal nucleus of binucleate endosperm undergoing mitosis; cen, chalazal endosperm nuclei; cet, tetranucleolate chalazal endosperm nuclei; me, micropylar nucleus of binucleate endosperm; med, chalazal nucleus of binucleate endosperm undergoing mitosis; men, micropylar endosperm nuclei; es, embryo sac; fed, first endosperm nucleus undergoing mitosis; zy, zygote. All sections stained with PAS and Gill no. 3 hæmatoxylin solution. Scale bars: A–C, E = 25 μm; D = 50 μm.

In both species, there is a special feature in mature ovules at the anthesis stage. This is the presence of a small protuberance at the funiculus insertion into the ovule body (aril primordium) and opposite the micropylar pole, which in further developmental stages originates the aril (Fig. 3D).

Aril development

Aril development follows a sigmoidal curve (C. striatus, Fig. 5A and C. multiflorus, Fig. 5B), which is divided into two stages: the pre- and anthesis stage, and the post-pollination stage.

Fig. 5.

Aril development. Width and thickness of chalazal area of aril (mean ± s.e.) in C. striatus (A) and C. multiflorus (B). DAP, days after pollination; Es, mature embryo sac; F, fertilization phase; Mgg, megagametogenesis period (two-, four-, and eight-nuclei phases); Msg, megasporogenesis period (from megaspore mother cell to triad phases); P-F, pre-fertilization phase. The crossing-point of the y- and x-axes indicates the beginning of flower anthesis.

Pre- and anthesis stage

This is characterized by the initiation of the aril primordium. During MMC and triad phases there was no aril zone differentiation: the epidermis of the funiculus at the region of the aril initiation was a single cell layer, undifferentiated from neighbouring epidermal cells. Subsequently these neighbouring cells became more vacuolated and more lightly stained than those from the aril zone. Aril zone differentiation started as a punctual area formed by a small group of cells at the funiculus insertion into the ovule proper, and opposite the micropylar pole. Aril initiation occurred at the time of the first mitotic division of the functional megaspore and was characterized by meristematic activity in some cells that undergo periclinal divisions, increasing the number of cell layers (Fig. 3G), in addition to anticlinal ones (only in the peripheral zone of the ovule). This combination of periclinal and anticlinal, together with oblique cell divisions (Fig. 3H) originated an aril primordium with a 2–3 cell layer thickness in C. striatus (Fig. 3D) and 1–2 in C. multiflorus at the time of anthesis. Meristem activity in the aril zone during these stages seemed to be reduced, with only a few cells showing such activity.

Post-pollination stage

This was divided into two phases: pre-fertilization phase, and post-fertilization phase (Fig. 5).

The pre-fertilization phase is a continuation of the anthesis period of reduced meristem capacity, and was characterized by an almost zero increase in cell layer number and thickness in the aril zone (Fig. 5A, B). The duration of the phase depends on the species, about 3–4 DAP in C. striatus and 4–5 DAP in C. multiflorus.

The post-fertilization phase is characterized by a progressive increase of the aril size, and was divided into three periods: (1) initial, (2) intermediate, and (3) final.

1. Initial period. This is characterized by a stepwise, gradual, progressive increase in the meristem activity of the aril primordium cells, verified by an increase in the number of cells undergoing mitosis in all the sections studied, which lasts until the begining of the intermediate phase. This leads to an increase in thickness in terms of the number of cells, but little thickening of the aril. The duration of this period was different for C. striatus (up to 8–12 DAP, Fig. 5A) and C. multiflorus (up to 18–22 DAP, Fig. 5B). The meristem reactivation occurred only when the ovule was fertilized. Otherwise, the aril primordium thickness either did not increase or did so very slightly (C. multiflorus 18 DAP, Fig. 6A). In fertilized ovules the aril underwent a thickness increase from 1 (or 2) to 4–6 (or 8) cell layers in C. multiflorus (20 DAP, Fig. 6B) and from 2 (3) to about 6–8 (10) cell layers in C. striatus (10 DAP, Fig. 6C). In parallel with this increase in thickness, the aril primordium began progresivelly to wrap around the funiculus, developing an irregular U-shaped expansion.

Fig. 6.

Aril development. Longitudinal sections of ovules of C. multiflorus (A, B, D, E, G) and C. striatus (C, F). (A) Penetrated but not fertilized ovule, 18 DAP. (B) Fertilized ovule, 20 DAP. (C) Fertilized ovule, 10 DAP. (D) Fertilized ovule, 18 DAP with a developing pro-embryo but without endosperm development. (E) Fertilized ovule, 40 DAP; inset: detail of aril cells. (F) Aril region of a fertilized ovule, 52 DAP. (G) Aborted ovule, 42 DAP with developed but arrested aril. ar, aril; arp, aril primordium; cc, central cell; e, first endosperm nucleus; en, endosperm nuclei; es, embryo sac; f, funiculus; ii, inner integument; oi, outer integument; pe, proembryo; pt, pollen tube; vb, vascular bundle; white arrowhead, starch grains. All sections stained with PAS and Gill no. 3 hæmatoxylin solution. Scale bars: A–D = 30 μm; E–G = 100 μm; F = 400 μm.

Every fertilized ovule with a normal endosperm development (Fig. 4D) showed a well-differentiated aril (Fig. 6B, 6C). But if the fertilized ovules had any endosperm malformation (for example with no divisions of the first endosperm nucleus), even though they developed a pro-embryo, either there was no reactivation of the aril zone (i.e. there was no differentiated aril; C. multiflorus 18 DAP, Fig. 6D) or there was delayed development.

2. Intermediate period. This is characterized by a decline in meristem activity, shown by a decline in the number of cells undergoing mitosis, together with an increase of aril cell expansion, the latter being faster than in the previous period. The cells begin to show an irregular outline and vacuolization (appearance of a large central vacuole) of their cytoplasm (C. multiflorus 40 DAP, Fig. 6E and inset). This period continued until about 30–32 DAP in C. striatus (Fig. 5A) and about 40 DAP in C. multiflorus (Fig. 5B).

3. Final period. This is characterized by a total loss of meristem activity and the transformation of aril cells from meristem to storage cells. In this period, the aril cells increase considerably in size, acquiring an irregular outline due to the appearance of a large central vacuole occuping practically the entire cell, with the nucleus displaced totally to its periphery. Hence, from then to the mature seed, aril development consists of an increase in cell size and an absence of meristem activity. The duration of this period is from 32–34 DAP to mature seed for C. striatus (Fig. 5A), and from 41–45 DAP to mature seed for C. multiflorus (Fig. 5B). This phase can be divided into two sub-periods: the first until about 45 DAP in C. striatus and 65 DAP in C. multiflorus is characterized by a large increase in cell size that produces a very rapid thickening of the aril's chalazal zone. This increase is due to the growth of the central vacuole and the irregularization of its shape. In the second subperiod, the cell size changes very little relative to the first subperiod (the size of the central vacuole remains unchanged) and the aril remains more-or-less constant in size (C. striatus 54 DAP, Fig. 6F).

Sometimes fertilized ovules were arrested at different stages of development. In these cases, if the arrest occurred in the intermediate or final phases (C. multiflorus, Fig. 6G), a quite well-developed aril was observed, but always of smaller size and with symptoms of cell degradation, with the cells being highly vacuolized (their interior being practically unstained) and generally with lightly stained structures, dominated by shades of pink due to the presence of polysaccharides released during the cell degradation (Fig. 6G).

Mature seed stage

The mature seed aril becomes an asymmetric U-shaped appendage, as seen in transverse section (C. striatus, Fig. 7A and C. multiflorus, Fig. 7B), almost completely surrounding the funiculus. The aril bypasses the micropyle and covers a very small portion of the seed (C. multiflorus, Fig. 7C and C. striatus, Fig. 7D). The aril shows three different regions: an inner region of small, thick-walled cells (Fig. 7A, B; PAS positive and Amido black 10B slightly positive); a median region of elongated cells (Fig. 7A, B; cell wall PAS positive and lumen cell Amido black 10B positive), and an outer region of large cells (Fig. 7A, B; cell wall PAS positive and slightly Amido black 10B positive) with large vacuoles containing droplets of oils, occupying most of the cell cytoplasm (Fig. 7E).

Fig. 7.

Aril development. Arils of C. striatus (A, D, F–J) and C. multiflorus (B, C, E). (A, B) Cross-sections of the aril in developing seeds 66 DAP and 80 DAP, respectively, stained with PAS and amido black 10B. (C, D) Mature seeds. (E) Portion of mature seed aril (outer region) stained with Sudan IV, showing lipid droplets. (F) Hilum region (half of aril has been removed). (G) Hilum region (aril has been removed). (H) Fertilized ovule 32 DAP showing double palisade layer in hilar area between aril and ovule. (I) Detailed chalazal part of the double palisade layer in hilar area of an ovule 28 DAP. (J) Fertilized ovule 60 DAP showing degradation of funiculus. ar, aril; cpa, counter palisade cells; cen, chalazal endosperm nuclei; dpl, double palisade layer; f, funiculus; fs, funicular scar; hg, hilar groove; hi, hilum; hr, hilar rim or corona; ir, inner region; li, lipid droplets; ln, lens; m, area micropylar; mr, median region; or, outer region; pa, palisade cells; pl, palisade layer; r, raphal area; rpa, raphal palisade cells; se, ovule body; sg, starch grains; white arrowheads, chalazal area of aril. (H, I, J) Stained with PAS and Gill no. 3 hæmatoxylin solution. Scale bars: A, B, H = 200 μm; C, D = 1 mm; E = 100 μm; F, G, J = 500 μm; I = 50 μm.

The hilar region consists of an oval hilum that has a hilar groove and is surrounded by the hilar rim or corona, a collar-shaped mound of a lighter colour than the rest of the testa that has lead to it being referred to as the eye or halo (Gunn, 1981), and with the whole being surrounded by the aril. The micropylar area is on one side of the hilum whilst on the other is the raphal area, in which the lens is hidden, as is the hilum, by the aril (Fig. 7F, G).

After fertilization, considerable changes are observed in the transition between the funicular and the raphal zones. This transition, the hilum (Fig. 7F, G), is the only part of the seed that has a double palisade layer (palisade and counter-palisade, Fig. 7H, I) and unmistakably separates the two zones, so that the aril zone will be connected with the zone outside this hilum (funiculus), without there being any physical connection with the inner part (the body of the ovule, the raphal area; Fig. 7H, I). A longitudinal section (Fig. 7J, 60 DAP) of the aril in a nearly mature seed showed that the funiculus dries up and dies in order to facilite the seed's separation from the fruit, leaving an easily recognizable gap (funicular scar; Fig. 7A, B). In this inner zone of the aril, the close relationship that exists between its cells and those of the funiculus can be observed.

DISCUSSION

Ovule development: megasporogenesis and megagametogenesis

The most common developmental pattern of the female gametophyte in Papilionoideae is the monosporic Polygonum-type (Martin, 1914; Reeves, 1930; Johansson and Walles, 1993; Cameron and Prakash, 1994; Moço and Mariath, 2004), but a bisporic type has also been cited in Lupinus, Laburnum anagyroides, Pueraria lobata (see Rembert, 1969 and references therein) and Wisteria sinensis (Rembert, 1967).

The most common pattern of megasporogenesis in Papilionoideae is the formation of a linear, occasionally T-shaped, tetrad of megaspores with the chalazal megaspore being functional. Our results are in agreement with the observations of Guignard (1881) that Cytisus megasporogenesis shows some variations with respect to the common pattern. After the first meiotic division, the smaller micropylar member of the dyad fails to undergo the second meiotic division, with the chalazal member being the only one that undergoes the second division, resulting in a linear triad. This observation has been reported by various workers as the most common model for some other legumes, such as Cytisus laburnum (Guignard, 1881) and Vicia villosa (Rembert, 1969), while there are still others in which this pattern is not typical but rather an alternative to the common pattern (megaspore tetrad formation), as is the case in Robinia pseudacacia (Rembert, 1969). This model is not the only variation found in leguminous plants since, for example, Rembert (1967, 1969) reported a bisporic model in Wisteria sinensis and Robinia pseudacacia (due to a cytokinesis failure in meiosis II of the chalazal member of the dyad), and also Guignard (1881) in Lupinus polyphyllus. Moço and Mariath (2004) noted the presence of bisporic development of the embryo sac in Lupinus and Cytisus, but both the analysis of the results of Guignard (1881) and our own results indicate monosporic development for Cytisus.

The embryo sac formation in both Cytisus species follows the Polygonum-model without any variation from megaspore to eight-nucleate phase, where intermediate (two- and four-nucleate) and final (eight-nucleate) phases show half of the nuclei towards the chalazal end and the other half towards the micropylar end, separated by a large central vacuole that, according to Folsom and Cass (1989, 1990) and Bittencourt and Mariath (2002), limits nuclear movements and thus controls ontogenetic events.

The presence of callose during megasporogenesis is a known but poorly documented fact. Different hypotheses have been formulated to explain its presence. One is that the callose surrounding some megaspores contributes to their abortion, so the functional megaspore should be completely or at least partially callose-free (see Lersten, 2004). A second hypothesis is that callose isolates the haploid cells from the surrounding diploid sporophyte tissue, as has been proposed for microsporogenesis (see Lersten, 2004). A third hypothesis is that callose prevents the entry of pathogens (Heslop-Harrison et al., 1999). Our observations may support the last two hypotheses because from the MMC until triad formation, the cells were completely surrounded by callose, even in the transverse walls between the cells of the triad, as has been observed by Russell (1979) in maize. The callose layer remains even after the beginning of the embryo sac development, but as a discontinuous and lightly stained parietal layer that disappears before embryo sac cellularization.

During its formation, the functional megaspore remains isolated from the nucellus, with there being no flow of energy between the two. But from the beginning of embryo sac formation we observe symptoms of degradation in the neighbouring nucellar cells in contact with it, as has been observed by other workers (Rembert, 1977). This seems to indicate that nucellar degradation may supply raw material and/or energy for the accumulation of starch grains into the embryo sac. We can also infer that energy coming from this degradation is used for embryo sac development, and for this energy transfer to occur from the nucellus to the embryo sac, the callose wall needs to disappear. It should be noted that in prior phases no nucellar degradation was observed in the regions adjacent to the MMC or triad cells, and it may be speculated that there is thus no possibility of energy transfer from the nucellus, although this will occur later in the stages of embryo sac formation. This could indicate that during these phases the energy needed in these earlier processes is stored completely in the MMC. Also, the MMC and triad cells continue to be isolated from the rest of the tissues so that they might be protected against the attack of viruses or other pathogens.

Cytisus ovules

In a previous paper on the floral biology of Papilionoideae, the ovules of the two species of the present study were considered to be basically campylotropous, although this was not studied in depth (Rodríguez-Riaño et al., 1999c). However, with the study of longitudinal sections it can be shown that these ovules are derived ana-campylotropous forms (Johri et al., 1992, and references therein). The ovules show a single vascular bundle that crosses the funiculus and ends at the nucellus base (chalaza), a common feature in most of angiosperms, although in some families (among them legumes) vascular penetration into one or both integuments has been observed (Johri et al., 1992; Danilova, 2002; Lersten, 2004).

There are some aspects of the embryo sac that are worth noting. Both species show the most common organization of the embryo sac in legumes, constituted by an egg apparatus (egg cell and two synergids, the latter with a very conspicuous filiform apparatus at the micropylar end) and a single central cell. There are no antipodals in the receptive embryo sac because these cells are ephemeral, as in most legumes (Guignard, 1881; Martin, 1914; Brown, 1917; Kennel and Horner, 1985; Johansson and Walles, 1993; Faigón-Soverna et al., 2003). The most significant difference in the mature embryo sac organization (anthesis phase) between the two species is the presence of fused or non-fused polar nuclei. In C. striatus they are generally fused, while in C. multiflorus the situation is very variable, from non-fused but paired polar nuclei to completely fused ones with only one nucleolus inside (secondary nucleus of the central cell), or intermediate situations with fused polar nuclei but with two nucleoli. Another noteworthy feature of the embryo sac is the presence of starch grains accumulated in both central and egg cells (Brink and Cooper, 1940; Kennell and Horner, 1985), with those of the central cell being larger than those of the egg cell due to their coalescence. According to Tilton et al. (1984) this starch grain production of the central cell could compensate for the loss of antipodals that would have contributed to the development of the female gametophyte or embryo.

Aril development

Since Gaertner (1788), who was the first to describe the aril as an accessory integument, the use of differing terminology by the various workers who have studied the seminal appendages has led to great confusion in correct naming of each of the different seminal appendages (Planchon, 1845; Baillon, 1876; Corner, 1949, 1953, 1976; van der Pijl, 1955; Kapil et al., 1980).

Boesewinkel and Bouman (1984) defined the hilum as the scar made by the separation of the funiculus from the seed, and the raphe as a prolongation of the funiculus running along the seed and ending at the chalaza. Under these definitions, it can be considered that the area before the hilum is the funiculus and the area posterior to it is the raphe, and our ontogenic study allows us to conclude that the aril primordium is produced in the funicular area. As the aril develops, this is clearly separated from the seed by the presence of a double palisade layer (palisade and counter-palisade) at the point of attachment (hilum). In addition, it is observed in mature seeds that the chalazal area of the aril is not joined to the raphal area, but only to the funiculus and the hilum, leaving the raphal area (from hilum to seed base) without any kind of associated structure. This seminal appendage, possibly in every Cytiseae and specifically in C. multiflorus and C. striatus, can not be termed a strophiole, because this term is defined as a glandular or spongy proliferation limited to the raphal region (see Kapil et al., 1980). Tacking all this into account, the seminal appendage present in the two species in the current study clearly has a funicular and not a raphal origin, which agrees with Corner's (1949) observations, not only for Cytisus but for every legume that he studied.

Studies of the complete ontogeny of the aril in angiosperms are very limited (but see Grear and Dengler, 1976 for Eriosema; Kloos and Bouman, 1980 for two species of Passifloraceae; and Ciccarelli et al., 2005 for Myrtus communis) and absent for Cytiseae. In Passiflora suberosa and Turnera ulmifolia, Kloos and Bouman (1980) distinguished two very different stages in aril development: pre-fertilization and post-fertilization. We have observed these stages in both Cytisus species. The first stage corresponds to the beginning of the aril primordium. This starts during megagametogenesis via periclinal divisions of specific dermal cells situated in the posterior part of the funiculus, on the side opposite the micropyle, increasing the primordium thickness. This increase in the aril primordium thickness in Cytisus (2–3 cell layers in C. striatus and 1–2 cell layers in C. multiflorus) is considerably smaller than that observed by Kloos and Bouman (1980) in the raphal area of Passiflora suberosa (almost half in C. striatus and practically one-third in C. multiflorus). In both Cytisus species, these periclinal divisions spread out towards neighbouring cells, increasing the width of the aril zone by similar amounts (12–17 cells) to those observed by Kloos and Bouman (1980) in P. suberosa (13–15 cells). Also similar, in both Cytisus species the aril primordium cells are initially undifferentiated from neighbouring epidermal cells, but subsequently the latter became more vacuolized. The meristem activity-driven growth of the aril primordium until fertilization occurs is very slow and sparse, and if fertilization does not take place the area remains practically unchanged, and after some days its cells begin to degrade and die together with the ovule.

In both Cytisus species, for the aril to develop completely it is necessary that fertilization takes place, as has been noted previously in general terms (Corner, 1949; Kloos and Bouman, 1980; Johri et al., 1992). In our case, it was confirmed that fertilization reactivates the aril primordium's development, but this reactivation is not complete if there is no normal endosperm development, i.e. if an endosperm is not formed there is no normal aril development. Thus, endosperm development must be responsible for a reactivation of the meristem capacity of the aril cells. This reactivation may be due to the increase in the number of nuclei of the endosperm, i.e. the mitotic processes of these nuclei might have a direct influence on the cells of the aril, activating its meristem capacity (direct reactivation), or the increase in the number of endosperm nuclei leads to an increase in size of the ovule, and it is this increase that leads to a change in the aril cells' meristem capacity (indirect reactivation). In contrast, when endosperm development was null or defective (for instance, when there was no division of the first endosperm nucleus), even though there was development of the embryo (normal or abnormal; T. Rodríguez-Riaño et al., unpubl. res.) the aril primordium cells underwent meristem reactivation (possibly due to the simple fact of the entrance of the pollen tube and subsequent fertilization), but eventually the process was arrested.

The aril primordium development after fertilization is due to the meristem activity of its cells (Kloos and Bouman, 1980), but cell proliferation decreases as aril development progresses, and practically disappears in the intermediate period of development. In the final period of aril development, its growth is due to the increase in cell size only, as in the observations of Kloos and Bouman (1980) for Passiflora suberosa and Turnera ulmifolia, and of Grear and Dengler (1976) for Eriosema. This increase in size involves a functional change in the aril cells, from meristem cells (isodiametric, with a more-or-less central nucleus) to storage cells (irregular shape, large central vacuole and peripheral nucleus).

There exist few differences in aril ontogeny between the two Cytisus species studied. In the pre-fertilization phase, the aril primordium is larger in C. striatus than in C. multiflorus, both in thickness and in width. This is not only due to the larger ovule size in C. striatus than in C. multiflorus (T. Rodríguez-Riaño et al., unpubl. res.), but also to a greater meristem activity in the former species that usually produces a larger number of cell layers. In the post-fertilization phase, the aril primordium development is much slower in C. multiflorus than in C. striatus, which could be due (not only in this phase but also in the pre-fertilization phase) to the different phenology of the two species, the former being winter-flowering with lower temperatures, and the latter spring-flowering (Rodríguez-Riaño et al., 1999b), with winter-flowering corresponding to a slow-down in biological processes (Primack, 1985; Jakobsen and Martens, 1994; Cruden, 2000). This slowing could in part explain the difference observed in endosperm development between the two species (T. Rodríguez-Riaño et al., unpubl. res.), which is slower in C. multiflorus than in C. striatus.

The elaiosome is an ecological term used to describe all fleshy and edible parts of seeds dispersed by ants (Sernander, 1906, cited in Boesewinkel and Bouman, 1984). This structure attracts ants and they remove it easily from the seed, which is hard and inedible, so that the seeds are collected but not eaten. According to this definition, the seed appendage with funicular origin in both studied Cytisus species should be considered as an elaiosome. This appendage has lipids and some proteins: reserve types that were also found by Lisci et al. (1996) in Cytisus scoparius and by Tiano et al. (1998) in Chamecytisus proliferus. As also in C. proliferus, three regions can be distinguished—inner, median and outer—with differing responses to stains, with the outer being the most lipid-rich. Usually, not only lipids, proteins and carbohydrates but also vitamins B₁ and C have been found in elaiosomes (for a review see Beattie, 1985; and Boesewinkel and Bouman, 1984, and references therein), although lipids are the most frequent components either alone (Morrone et al., 2000) or together with starches or proteins (Grear and Dengler, 1976; Johri et al., 1992; Lisci et al., 1996; Mayer and Svoma, 1998; Tiano et al., 1998). The fatty acid composition of the elaiosome is known to vary between populations and determines diaspore removal by ants (Boulay et al., 2006). Usually, for ants, elaiosomes are a high-energy food, and for plants they are seed appendages that serve as bait for myrmecochorous dispersal (Lisci et al., 1996).

As in other Cytiseae (López et al., 2000; Malo, 2004; López-Vila and García-Fayos, 2005 and references therein), the two species studied here show a primary ballistic type of seed dispersal mechanism via explosive dehiscence of the legumes. Indeed, the seeds can be thrown up to 4 m away in Cytisus multiflorus (Moreno et al., 1992). This primary dispersion is assisted by myrmecochory as a secondary dispersal mechanism, which can give more opportunities for the temporal and spatial establishment of the plant (López-Vila and García-Fayos, 2005). Sometimes myrmecochory is the only dispersion mechanism in both species, because the explosive dehiscence does not eject the seeds, which thus remain attached to the open fruit. In these cases, ants climb the plants and harvest them directly from the fruit (T. Rodríguez-Riaño, pers. obs.). These seeds may also fall beneath the plant or around it by barochory, as in other Cytisus (C. scoparius, Malo, 2004). In many species, as observed in both Cytisus species studied here, the elaiosome represents a very small percentage of the total seed weight (4–9 %, Lisci et al., 1996). Several functions have been suggested for the elaiosome, such as: (1) facilitation of dehydration and hydration of the seed; (2) induction of seed dormancy; and (3) a water reserve during germination (Lisci et al., 1996; Tiano et al., 1998). Indeed, in some Cytiseae (e.g. Ulex parviflorus, López-Vila and García-Fayos, 2005; but also in other groups, e.g. Knautia, Mayer and Svoma, 1998), elaiosome removal from the seeds by ants increases the germination percentage compared with control seeds.

In summary, C. multiflorus and C. striatus show a monosporic Polygonum-type of embryo sac with megasporogenesis characterized by the presence of a triad of cells due to a failure in the meiosis II in the micropylar cell of the dyad. The presence of callose seems to be related to the isolation of the megaspore mother cell and the triad from the diploid cells of the nucellus. The disappearance of the callose from the walls of the functional megaspore seems to be necessary for the subsequent development of the embryo sac by allowing passage of nutrient supplies resulting from the degradation of the nucellar cells. The aril of the seed shows a funicular origin. Its development follows a sigmoidal curve, with slow growth in the first period characterized by meristem activity of the cells of the aril primordium and a subsequent gradual change to growth by cell expansion in the later periods. The aril development is stimulated by fertilization and completed by endosperm development. The aril in the mature seed of both species has lipids and some proteins as reserve types.