An ultrastructural study of the differentiation of the spermatozoid of Equisetum

The motile male gamete of the Archegoniatae is amongst the most highly differentiated of all plant cells. Its development involves profound and highly integrated changes in the morphology and distribution in most of the organelles of the earlyspermatid. Spermatogenesis thus provides an excellent system to study the capacity of organelles to differentiate and their interrelationships within a highly specialized cell.

In recent years biflagellate plant spermatozoids have received much attention at the ultrastructural level and there are detailed accounts of many aspects of spermatogenesis in bryophytes (Carothers & Kreitner, 1967, 1968; Diers, 1967; Paolillo, 1965; Paolillo, Kreitner & Reighard, 1968a, b) and charophytes (Pickett-Heaps, 1968; Turner, 1968, 1970). The structure of multiflagellate spermatozoids however, is far less well known. Accounts exist of the mature spermatozoids of Pteridium (Bell, Duckett & Myles, 1971; Duckett & Bell, 1971; Manton, 1959), Marsilea (Rice & Laetsch, 1967) and Zamia (Norstog, 1967, 1968) and the origin of the centrioles has been investigated in Marsilea and Zamia (Mizukami & Gall, 1966). Although the presence of some of the characteristic features of biflagellate spermatozoids (e.g.the specialized mitochondrion associated with the motile apparatus) have also been demonstrated in multiflagellate spermatozoids (Duckett & Bell, 1969, 1971), no study of the differentiation of these gametes at the ultrastructural level has hitherto been made.

The metamorphosis of all archegoniate spermatozoids involves the development of an extreme form of cellular asymmetry. This process is intimately connected with the formation of a complex microtubule system which functions as the skeletal component of the motile cell in the absence of a rigid cell wall (Duckett & Bell, 1971). The present paper is concerned with the origin and development of this microtubule system and its relationship to the nucleus and flagella during the differentiation of the spermatozoid of Equisetum. This group of pteridophytes was chosen because here the microtubule system appears to be more extensive than in all others (Duckett & Bell, 1969).

Particular attention is also given to the ultrastructural characteristics of the various layers of the multilayered structure (Heitz, 1959, 1960) in an attempt to shed some light on the conflicting reports over the number and nature of these in bryophytes (Carothers & Kreitner, 1967, 1968; Paolillo, 1965; Paolillo et al. 1968a). Originally described from the spermatids of a variety of bryophytes (Heitz, 1959, 1960) this remarkable organelle, lying between a mitochondrion and the flagellar bases, has now been detected in cycads (Norstog, 1967, 1968), ferns (Bell et al. 1971), Equisetum (Duckett & Bell, 1969) and the alga Coleochaete (MacBride, 1968). However, to date, no detailed account has been published of its morphology and development in a multiflagellate spermatozoid.

The various stages in spermatozoid ontogeny in Equisetum are illustrated by diagrammatic reconstructions from serial sections. In view of the extreme complexity of the mature male gamete of Equisetum this is considered separately elsewhere (J. G. Duckett, in preparation).

Gametophytes of Equisetum hyemale L. were grown in axenic single spore cultures on Moore’s medium (Moore, 1903), solidified with 1·5% agar. General culture techniques are described elsewhere (Duckett, 1970a, b, 1972a;,Duckett & Bell, 1969).

Tips of actively growing male branches containing antheridia in various stages of development were fixed in 1 % glutaraldehyde (Taab Laboratories, Reading, England) in 0·05 M phosphate buffer (pH 6·9) for 1-2 h at 4 °C. Following overnight washing in buffer the material was postfixed for 1 h at 4 °C in 2 % osmium tetroxide, and subsequently dehydrated in acetone. From pure acetone the material was transferred to propylene oxide and thence to Durcupan ACM (Fluka AG, Switzerland.)

The embedded material was first sectioned at 4 μm and the sections, mounted in resin without accelerator, were searched for antheridia in different stages of development. The appropriate sections were then remounted for thin sectioning by a method described elsewhere (Woodcock & Bell, 1967). Thin sections were cut with a diamond knife, on a Porter Blum ultramicrotome, stained for 30 min with saturated uranyl acetate solution followed by 30 min with basic lead citrate (Reynolds, 1963) and examined in a Siemens Elmiskop I.

The antheridia of Equisetum are produced either on tips of lamellae or on extensions of the cushion region of the gametophyte (Duckett, 1970 a, 1972 a). Each male branch has a meristem which produces new antheridia continuously, so that any one branch bears a range of antheridia in different stages of development. The antheri-dial initial is a superficial cell which divides periclinally, the outer cell and its derivatives forming the jacket of the antheridium and the inner the spermatogenous tissue. This latter cell divides by consecutive synchronous mitoses (Fig. 6A, B) until 2561024 cells are formed. The cells resulting from the last division are the spermatids, which undergo metamorphosis into the spermatozoids.

Throughout the early cell generations the spermatogenous tissue is composed of angular thin-walled cells with large central nuclei containing conspicuous nucleoli (Fig. 6c). The cells derived from the last mitosis, the spermatids, are however clearly distinct from their predecessors: nucleoli have disappeared (Fig. 6D) and the protoplasts round off at the corners (Fig. 6E). This marks the beginning of the general recession of the protoplast from the cell wall, a process which continues throughout spermatogenesis. The synchrony seen in the earlier division stages of spermatogenesis persists throughout the metamorphosis of the spermatids, despite the loss of protoplasmic connexions between them.

Half-differentiated spermatids may be distinguished by the asymmetrical position of their nuclei (Fig. 6F), which then gradually lose their circular outline and undergo elongation and coiling. Antheridia containing mature spermatocytes are readily recognizable by their well rounded protoplasts (Fig.6 G). In some planes of sectioning the coiled nuclei and spiral ‘blepharoplasts’ (Sharp, 1912), whose complex nature is described later, can be distinguished.

In Equisetum the earliest progenitor of the motile apparatus apparently appears just before the last antheridial mitosis. This is the blepharoplast, a spherical structure about 0·6μm in diameter, composed of numerous closely packed subunits whose structure is difficult to ascertain (Turner, 1966). Soon after the final division is complete the subunits of the blepharoplast gradually become more distinct and then begin to separate. The newly freed subunits are short cylinders about 0·15 μm. in diameter and 0·3 μm long (Figs. 7-9). Initially their substructure is difficult to make out, but in transverse sections regularly arranged tubules about 27 nm in diameter may be discerned (Fig. 8). These early centriolar, or procentriolar structures, some 80-120 in number, are at first oriented in several directions in a localized area of the cytoplasm often surrounded by conspicuous Golgi bodies (Fig. 9). The plastids of spermatids at this stage have poorly differentiated internal lamellar systems and are surrounded by regular sheets of rough endoplasmic reticulum (Fig. 7).

Further development of the early spermatids is characterized by the gradual separation of the centrioles from each other and their individual tubules becoming more clearly stained (Fig. 10). The centriole of Equisetum is a short cylinder about 200 μm in diameter and 300 μm long. The wall is composed of 9 triplet fibres, each about 27 nm in diameter, inclined at an angle to the edge of the centriole. The lumen contains a clearly defined cartwheel structure, a characteristic feature of the basal bodies of the flagella of the mature spermatozoid.

At this stage a band of microtubules each about 27 nm in diameter suddenly appears in the vicinity of the centrioles (Fig. 10). These are regularly spaced, with a centre-to-centre separation of about 35 nm, and run parallel to one another in a single plane. They are the first sign of the microtubular skeleton of the motile gamete and form the outer layer of the multilayered structure.

Fig. 11 shows the general disposition of the organelles of the spermatids at this stage. The large, spherical, centrally placed, nucleus, lacking a nucleolus, is surrounded by numerous proplastids, mitochondria and small vacuoles. Along one edge of the microtubular band, henceforth referred to as the anterior edge, since it forms the anterior edge of the microtubular skeleton of the mature gamete, the inner or lamellar layers of the multilayered structure soon become distinct (Figs. 11-15). On the side of the lamellar layers away from the microtubular band lies a mitochondrion (Figs. 11-15). In sections perpendicular to the axes of the microtubules (Fig. 12) the lamellar layers appear as an electron-opaque strip. However, in sections oblique to the axes of the microtubules but perpendicular to the axis of the lamellar layers and to that of the subjacent mitochondrion the characteristic columnar profile is revealed (Fig. 13). Grazing sections through the multilayered structure (Fig. 14) indicate that from the outset the lamellar layers are parallel plates aligned at about 35—45° to the axes of the microtubules, an angle which remains constant throughout the metamorphosis of the spermatids.

Initially the lamellar layers form a compact structure not much more than 1μm long and 0·5 μm wide and the associated mitochondrion is approximately oval. The microtubules are relatively few in number (less than 50), short (less than 2 μm) and do not extend far beyond the lamellar layers. However, the lamellar layers and mitochondrion soon increase in length (Figs. 14, 15) and eventually form a curved strip about 15·20μm long extending almost the diameter of the spermatid (Figs. 14-20). From this point the length of the lamellar strip remains constant throughout the metamorphosis of the spermatid.

Scattered sections more or less perpendicular to the longitudinal axis of the lamellar strip give the impression that the microtubular band is composed of a variable number of microtubules. Serial sectioning, however, reveals that this results from the spatial relationship between the microtubular band and the lamellar strip (Fig. 1). The tip of the lamellar strip (termed anterior because this is invariably its position in the mature gamete) is overlain only by the most anterior of the microtubules. Consequently, sections of this region show a narrow band only a little exceeding the width of the lamellar strip. Since the microtubules extend posteriorly at an angle of about 40o to the longitudinal axis of the strip, continued sectioning inevitably shows an increase in the width of the band and in the number of its microtubules, until a maximum of about 150 is reached. Although this is the maximum which has been recorded in any one section, from the total length of the lamellar strip it is estimated that the band contains over 300 individual tubules, each about 3·5 μ m in length at this stage. Serial sectioning also reveals that the number of plates in the lamellar strip diminishes from about 50-70 at the anterior end to less than 10 at the posterior.

Sections illustrating this stage in spermatogenesis are shown in Figs. 16-20, and their approximate positions in relation to the whole multilayered structure are indicated in Fig. 1. In Fig. 16 the microtubular band is sectioned twice, once obliquely through the tubules at the anterior most tip of the lamellar strip and once (enlarged in Fig. 17) almost perpendicular to the axes of the tubules where the band contains about 60 tubules. In Fig. 18 the microtubular band is again sectioned twice, once near the anterior tip of the lamellar strip and perpendicular to its axis, thus revealing the columnar profile (enlarged in Fig. 19), and once about halfway along the lamellar strip perpendicular to the axis of the microtubules (enlarged in Fig. 20). In the former case there are about 35 microtubules in the band and in the latter about 130.

By this time the centrioles have become regularly spaced out along the microtubular band with their axes approximately parallel to those of the microtubules (Fig. 20). They have also begun to elongate (Fig. 19) as the first stage in the formation of flagella from their distal ends and are now about 1-2 μm. in length. Transverse sections through their distal ends reveal the stellate interior profiles (Fig. 20) characteristic of plant flagella (Manton, 1964).

During the early stages of spermatid differentiation described above the multilayered structure and associated mitochondrion lie approximately mid-way between the nucleus and cell membrane and have no precise relationship with the nucleus (Figs. 16-20). However, soon after the lamellar layers reach their full length close association with the nucleus takes place (Fig. 2A, B). The microtubular band becomes closely adpressed to the nuclear envelope (Fig. 21). The lamellar layers and subjacent mitochondrion run along the nuclear envelope with one edge touching it and the other projecting into the cytoplasm. Where they touch the nuclear envelope a shallow ridge appears. The anterior most tip of the multilayered structure lies free in the cytoplasm and just behind this an anterior projection or beak soon appears in the nucleus (Figs. 22, 24). The tubules of the microtubular band grow in length and the lamellar strip and mitochondrion gradually begin to separate from the nucleus, beginning at the anterior tip. At the same time the nuclear beak is drawn out along with the elongating microtubules. The section of this stage in Fig. 24 is approximately parallel to the long axis of the lamellar strip and elongate mitochondrion. In Fig. 22 the section is roughly perpendicular to the one in Fig. 24 along the plane a = b, and shows the anterior tip of the multilayered structure lying freely in the peripheral cytoplasm. Fig. 23 shows the microtubular band closely adpressed to the nuclear envelope with the lamellar strip and mitochondrion running along its anterior edge. Also associated with the anterior edge of the microtubular band, but on its external surface, lies a strip of osmiophilic material with a regularly banded substructure.

Concomitant with the association of the multilayered structure and the nucleus the flagella rapidly grow out from the distal ends of the basal bodies and soon protrude from the general surface of the cell membrane (Figs. 22-24). Numerous membranous vesicles are present outside the cell membrane in their vicinity (Fig. 24).

The nucleus also moves from its central position in the spermatid to the margin, from which it is separated by the microtubular band and the layer of cytoplasm containing the basal bodies (Figs. 23, 24). Where the nucleus is not invested by the microtubular band its envelope is often highly convoluted (Fig. 24) and very difficult to distinguish from abundant and highly irregular arrays of rough endoplasmic reticulum.

At this time, much of the cytoplasm contains abundant endoplasmic reticulum closely associated with polysomes (Fig. 25). The envelopes of the plastids, which are beginning to accumulate starch, are indistinguishable from rough endoplasmic reticulum (Fig. 25). Other regions of the cytoplasm (Fig. 26) are highly vesiculate as a result of intense Golgi activity. These vesicles coalesce and discharge their contents outside the cell membranes of the spermatids. Due to this, the volume of the spermatids’ cytoplasm gradually decreases and the cell membranes gradually recede from the cell walls.

The anterior beak of the nucleus becomes more and more pronounced and more and more of the anterior rim of the multilayered structure separates from the anterior edge of the nucleus due to the elongation of the microtubules (Fig. 3). The nucleus becomes comma- and then crescent-shaped. Since the microtubules are oriented at about 40o to the longitudinal axis of the lamellar strip, the elongation of the microtubules not only brings about the separation of the nucleus and the lamellar strip but also results in the displacement of the lamellar strip laterally relative to the nucleus. At the comma-shaped stage (Fig. 3 A, B) the anterior third of the lamellar strip extends laterally beyond the anterior tip of the nucleus. By the time the nucleus has become a crescent (Fig. 3C,D) only the posterior third of the lamellar strip remains above the anterior part of the nucleus. During the process the lamellar strip, formerly forming a wide arc of less than half a gyre, begins to coil up, beginning at its anterior tip (Fig. 3 C, D). This is most probably due to the more rapid growth of the microtubules in the anterior part of the band. Ultimately, by the time the nucleus is crescent-shaped, the lamellar strip forms three quarters of a gyre.

Representative sections from a series through a mid-spermatid with a commashaped nucleus are shown in Figs. 27-30. The lamellar strip and subjacent mitochondrion form a strip running roughly parallel to, but separate from, the anterior edge of the nucleus. Fig. 30 clearly reveals that a considerable portion of the multilayered structure has now become displaced from the nucleus. In addition to the elongation process which separates the lamellar strip from the nucleus, the microtubules have also grown posteriorly to ensheath the posterior part of the nucleus (Fig. 29) and separate it from the cell membrane. During the lateral and anterior separation of the lamellar strip from the nucleus the region of the microtubular band with flagella external to it gradually comes to lie anterior to the nucleus. At the stage of the comma-shaped nucleus, by which time the flagella are fully formed, over three quarters of the basal bodies, which formerly overlay the nucleus, now lie anterior to it. In the mature gametes all the basal bodies are situated anterior to the nucleus.

The later stages in the metamorphosis of the mid-spermatids are considered in detail elsewhere (J. G. Duckett, in preparation). They involve further coiling and separation of the nucleus and lamellar strip until in the mature spermatocyte the nucleus is coiled in about one and a quartergyres and the lamellar strip two and a half gyres.

The gross morphology and interrelationships of this organelle with other components of the spermatid have already been described. Here the detailed substructure of the various layers will be considered. The clarity of this varies with the stage of maturation of the spermatids. The whole structure is most distinct in mid-spermatids and the descriptions which follow all refer to this stage.

The outermost component is the microtubular band. The individual tubules of this have an outside diameter of 26-27 nm, an inside one of about 13-14 nm and a centre-to-centre spacing of about 32 nm. Where the tubules are situated outside the lamellar layers, keels, about 15 nm wide and 10 nm deep, are present (Figs. 31, 32). Within the lamellar layers, which are separated from the microtubular band by an electron-transparent zone about 20 nm wide, 3 distinct strata may be distinguished. The outer 2, each about 30 nm in height, consist of a uniform series of electronopaque plates with a centre-to-centre spacing of 12-14 nm, embedded in a transparent matrix (Figs. 32, 33). The plates nearer the microtubular band are less distinct in outline than nearer the mitochondrion and the effect of this is to give a line of discontinuity in the profiles of the lamellar layers both when these are cut perpendicular to the long axes of the plates (Figs. 32, 33) and obliquely (Fig. 31).

The innermost stratum of the lamellar layers is about 15 nm thick and separated from the third layer by a thin and somewhat discontinuous layer of electron-opaque material (Figs. 32, 33). Although details of its substructure are difficult to make out they are perhaps best interpreted as small tubules (Fig. 32). These are sometimes open at the bottom and tend to alternate with the plates of the second and third layers.

In sections oblique to the long axis of the lamellar strip this fourth layer appears as a thin but continuous electron-dense strip (Fig. 31). The overall thickness of the multilayered structure is about 120-130 nm.

One additional component of the multilayered structure, present from its inception is an area of electron-dense (‘osmiophilic’) material associated with the external anterior rim of the microtubular band (Figs. 23, 27-30, 31, 33). Sections parallel to the long axis of the lamellar layers indicate that this is a continuous strip of material extending along the whole of the anterior edge of the microtubular band and with a substructure of regular darker and lighter bands with a centre-to-centre spacing of about 30-35 nm.

A drawing depicting a 3-dimensional interpretation of the multilayered structure is shown in Fig. 4 and a similar reconstruction which also incorporates the flagella, nucleus and cell membrane is presented in Fig. 5. The proximal ends of the basal bodies contain cartwheel structures and 9 triplet fibres which are scarcely angled towards the edge of the cylinders (Fig. 33). More distally the lumina contain stellate structures (Fig. 33), and the triplets gradually change to doublets by the loss of their outer fibres. External to this the typical 9 + 2 flagellum structure is found. A conspicuous feature of the cytoplasm surrounding the basal bodies is the lack of ribosomes (Fig. 33).

Comparison of the metamorphosis of the spermatids of Equisetum with the same process in other groups not only reveals many similarities in the nature of the organelles taking part and their spatial interrelationships but also sheds considerable light on the probable functions of many of the highly specialized components of motile plant gametes. Those differences which do exist appear to be closely related to the final form of the mature spermatozoid in the various groups.

Although the ultimate origin of the blepharoplast apparently de novo in the spermatid mother cells of Equisetum (Sharp, 1912; Turner, 1966) has yet to be investigated in detail, the present work makes it possible to recognize the full significance of the observations made by Sharp (1912) on this structure using optical microscopy. The ‘blepharoplast granules’ described by Sharp as originating from the breakup of a spherical blepharoplast in the young spermatids and later giving rise to the flagella are clearly equivalent to the centrioles. Sharp then states that a ‘beaded thread-like blepharoplast’ is formed by the fusion of the granules and gradually extends in an arc half way around the nucleus. This ‘thread’ almost certainly corresponds to the lamellar layers of the multilayered structure together with the subjacent mitochondrion, and the ‘beads’ to the basal bodies arranged in a single layer on the outer surface of the microtubular band. The present description of the separation of the lamellar strip from the nucleus confirms Sharp’s observation that during ‘spiralization’ the blepharoplast and nucleus become widely separated. However, whereas Sharp failed to detect any connexion, except apparently undifferentiated cytoplasm, between the 2 structures, the present study clearly shows that the nucleus and lamellar strip remain linked by the microtubular band, a feature clearly beyond the resolving power of the light microscope.

At the ultrastructural level the breakup of the blepharoplast in Equisetum appears to be identical with the same phenomenon in other archegoniates with multiflagellate spermatozoids. In Marsilea, Zamia (Mizukami & Gall, 1966) and homosporous ferns (Vazart 1964; J. G. Duckett, in preparation) the centrioles all arise from short cylindrical procentriolar structures with 9 subunits in their walls and indistinct cartwheel structures in their lumina.

The young centrioles of Equisetum appear to be identical with those of bryophytes (Moser & Kreitner, 1970), homosporous ferns (Tourte & Hurel-Py,1967) andcharophytes (Pickett-Heaps, 1968; Turner, 1968). However, in Equisetum, during the transformation into flagellar basal bodies the 9 triplet fibres lose their angled orientation to the edge of the cylinders, whereas in Pteridium (Duckett & Bell, 1971), cycads (Norstog,1967, 1968) and charophytes (Pickett-Heaps, 1968; Turner, 1968) this is retained in the motile gametes. In contrast to cycads (Norstog,1967, 1968) and homosporous ferns (Manton, 1959; Duckett & Bell, 1971) there is no evidence of a basal plate in the flagella of Equisetum. The helical structure surrounding the 2 central fibres of moss (Paolillo, 1967) and Anthoceros (J. G. Duckett, in preparation) flagella are absent in Equisetum, but in common with all other archegoniate spermatozoids so far examined there is no trace of flagellar ornamentation in Equisetum.

Although the occurrence of the multilayered structure has been described from a variety of archegoniate gametes (Duckett & Bell, 1969) there are no published accounts of the early stages in its ontogeny. In Equisetum, homosporous ferns (Tourte & Hurel-Py, 1967) and bryophytes (Paolillo et al. 1968 a) the lamellar layers initially form a shortly rectangular structure hardly exceeded by the microtubular band. However, whereas in the latter group the lamellar layers remain this shape, in ferns and Equisetum they rapidly increase in length to form a thin strip. The maximum dimensions of the lamellar layers and the total number of microtubules found in the mature spermatozoids are established at this stage.

The close alignment which then takes place between the microtubular band and the nuclear envelope suggests that the 2 are bound firmly together and that the elongating microtubules form the structural framework responsible for the coiling and elongation of the nucleus. In all other archegoniate spermatozoids nuclear metamorphosis is similarly connected with microtubular sheaths. Likewise, the development of pyriform nuclei in fungal zoospores is associated with regularly arranged skeletal microtubules (Heath & Greenwood, 1971; A. R. Hughes & J. G. Duckett, in preparation).

As in ferns (J. G. Duckett, in preparation) nuclear metamorphosis in Equisetum involves gradual separation of the lamellar strip and elongate subjacent mitochondrion from the nucleus, beginning at the anterior tip.

The microtubular band in Equisetum, which contains over 300 individual tubules, is much wider than in any other plant spermatozoid. The maximum numbers of microtubules recorded from other groups are 27-30 in Chara (Moestrup, 1970; Pickett-Heaps, 1968), 21 in Nitella (Turner, 1968), 17 in Marchantia (Carothers & Kreitner, 1967), 14 in Anthoceros (J. G. Duckett, in preparation), about 16 in Sphaerocarpos (Diers, 1967), 10-12 in Polytrichum (Paolillo et al. 1968 a), about 25 in Marsilea (Rice & Laetsch, 1967), and about 130 in Pteridium (Duckett & Bell, 1971). The width of the band is clearly not related to the number of flagella, since Marsilea with multiflagellate gametes has approximately the same number of microtubules as the biflagellate spermatozoids of charophytes. It is more likely that width is related to the shape of the nuclei in the mature spermatozoids. Bryophytes, charophytes, and Marsilea all have rod-like nuclei, whereas those of Equisetum have a twisted and slightly elongated pear shape. The nuclei of Pteridium spermatozoids are intermediate between the 2 extremes and the number of microtubules is also intermediate.

The present study of Equisetum clearly demonstrates that the lamellar layers of the multilayered structure consist of parallel plates of electron-opaque material embedded in a transparent matrix, just as in Marchantia (Carothers & Kreitner, 1967) and Pteridium (Bell et al. 1971). Similarly, the angle of 35—45° between the axis of these plates and that of the overlying microtubules in Equisetum is the same as in ferns and bryophytes. The presence of keels on the microtubules of Equisetum where these overlie the lamellar layers has also been recorded in Marchantia (Carothers & Kreitner, 1967). In the differentiating spermatids of Equisetum 3 strata can be distinguished in the lamellar part of the multilayered structure. The innermost layer, consisting of indistinct tubules, smaller and more delicate than those of the microtubular band, appears to correspond to the innermost layer of the ‘Vierergruppe’ or ‘Dreiergruppe’ of bryophyte spermatids (Heitz, 1959, 1960; Carothers & Kreitner, 1967; Paolillo, 1965; Paolillo et al. 1968 a). In Equisetum the lamellae external to this can be divided into an outer and an inner half by the clarity of the plates. Just as in the motile spermatozoids of Pteridium (Bell et al. 1971) the outer part of each plate is less distinct than the inner, which results in a line of discontinuity parallel to the inner and outer surfaces of the lamellar strip. In Marchantia spermatids (Carothers & Kreitner, 1967) and mature spermatozoids of Zamia (Norstog, 1967) not only are the plates of the 2 strata of equal clarity but are also separated by a horizontal membrane. However, in mature spermatocytes of Equisetum a similar membrane is also present and the innermost stratum of the multilayered structure no longer consists of tubules but is a uniform strip of electron-opaque material (J. G. Duckett, in preparation). In Anthoceros (J. G. Duckett, in preparation) the lamellar region consists of 3 strata in the young spermatids, but just prior to its disappearance as the gametes mature the plates become occluded with osmiophilic material and distinct layering is no longer visible. Thus the general picture which emerges is that, additional to the considerable ontogenetic changes which take place in their structure, the lamellar strata also exhibit distinct phyletic differences.

Although it is difficult, at present, to ascribe a precise function for the lamellar regions of the multilayered structure, it is noteworthy in Equisetum, Zamia (Norstog, 1967, 1968) and Pteridium (Bell et al. 1971) that it persists in the motile spermatozoid and in Pteridium also enters the egg at fertilization (Duckett & Bell, 1972).Thus it could be argued that it has a mechanical role in holding the microtubules of the overlying band together. However, it is absent in the motile gametes of bryophytes (Paolillo et al. 1968) and has not been detected in Marsilea (Rice & Laetsch, 1967) nor in most algal spermatozoids. Perhaps it is an ancient feature retained in a few primitive groups of plants with a role in the highly co-ordinated synthesis of regularly spaced microtubules. This hypothesis is presently being investigated using inhibitors specific to microtubule formation.

The electron-dense material associated with the anterior rim of the microtubular band in Equisetum is similar to that which occurs in this region of the spermatozoids of bryophytes (Paolillo et al. 1968b), ferns (Manton, 1959; Duckett & Bell, 1971) and Chara (where it is known as manchette adjunct, Pickett-Heaps, 1968). Pickett-Heaps (1970) has suggested that it is a ‘microtubule organizing centre’, involved in a special type of microtubule synthesis where the exact spatial configuration of the tubules may be important. However, from this study of Equisetum, it is by no means certain that the elongating microtubules are actually being synthesized in this region: during the separation of the lamellar strip from the nucleus anteriorly, the microtubules are also extending around the posterior part of the nucleus. Further studies, for example treatment of differentiating spermatozoids with colchicine (Turner, 1970), are clearly required to try and pinpoint the sites of microtubule formation and thus elucidate the possible role of the osmiophilic strip in the process. The persistence of this layer in the mature gametes of Pteridium (Duckett & Bell, 1971) and Equisetum long after microtubule elongation has ceased, perhaps indicates that it is a structural substance holding the band together. The regular banded substructure of the osmiophilic layer adds further support to this notion, since it is highly reminiscent of the kinetodesmal fibres and striated flagellar roots of a variety of algae and protozoa (Allen, 1971; Dingle, 1970; Hoffman, 1970; Leadbetter, 1969; Munn, 1970; Pitelka, 1968). Here these provide mechanical support for the flagellar apparatus and thereby serve as a means for distributing throughout the motile cells the stresses generated by flagellar action.

In mosses (Paolillo et al. 1968 a), ferns (Duckett & Bell, 1971) and Equisetum, but apparently not in the alga Coleochaete (MacBride, 1968) or Zamia (Norstog, 1967, 1968) the lamellar layers of the multilayered structure are associated with a specialized mitochondrion. This is formed by the fusion of mitochondria in the young spermatids of bryophytes (Paolillo, 1965; Paolillo et al. 1968 a) and mitochondrial fusion is well documented in animal spermatogenesis (Favard & Andre, 1970) and in zoospore formation in some fungi (Cantino & Mack, 1969). Turner (1966) suggested that the long mitochondrion in Equisetum also resulted from the coalescence of several shorter ones soon after the last antheridial mitosis. However, the present study clearly shows that a short mitochondrion, associated with the lamellar layers at their inception, elongates along with these layers in the young spermatids. Similarly, in ferns (J. G. Duckett, in preparation) there is no evidence that the long mitochondrion associated with the anterior border rim arises by fusion of ovoid mitochondria in the young spermatids. However, whether the spiral mitochondrion in the spermatozoids of Marsilea (Rice & Laetsch, 1967) arises by elongation of a single mitochondrion or coalescence of several shorter ones still requires investigation. It is of interest to note that whilst the mitochondria of animal spermatozoids are always closely associated with the flagella, in archegoniates the microtubular band lies between these 2 components.

During the differentiation of the spermatids of Equisetum, Pteridium (Duckett & Bell, 1971) and charophytes (Pickett-Heaps, 1968) there is intense Golgi activity, accompanied by progressive separation of the cell membrane away from the original cell wall, the area between the 2 probably being filled, as in bryophytes, with mucopolysaccharides (Vian, 1970; Vian, Barbier & Crepin, 1970). From a consideration of the well established role of the Golgi in polysaccharide formation and transport in other situations (Clowes & Juniper, 1968; Northcote & Pickett-Heaps, 1966) it seems highly likely that this organelle also performs the same function during spermatogenesis.

A characteristic feature of plant spermatogenesis is the accumulation of starch within the plastids of the spermatids and the ubiquity of amyloplasts in the mature male gametes indicates that they may perform some essential function. Although it has often been suggested that the starch may be required as an energy source, there is no evidence that it is mobilized during swimming (Rice & Laetsch, 1967). Alternative hypotheses are that the starch with a specific gravity of 1 -6 stabilizes the gametes in some way or enables them to perceive gravitational forces. Further supporting these ideas is the fact that the large multiflagellate gametes of Pteridium (Duckett & Bell, 1971), Marsilea (Rice & Laetsch, 1967), Zamia (Turner, 1966) and Equisetum all have several amyloplasts, whereas biflagellate charophytes (Pickett-Heaps, 1968; Turner, 1968) and bryophytes (Diers, 1967; Paolillo et al. 1968 a, b) have only from one to four.

At the beginning of starch accumulation, the envelopes of the spermatid plastids of Anthoceros (J. G. Duckett, in preparation) Pteridium (J. G. Duckett, in preparation) and Equisetum become closely associated with rough endoplasmic reticulum and at one stage in Equisetum the plastid envelope becomes indistinguishable from the endoplasmic reticulum. It is noteworthy that where similar associations have been seen elsewhere, as for example in potato tuber buds (Marinos, 1967), resin canal cells (Wooding & Northcote, 1965 a), and differentiating phloem (Wooding & Northcote, 1965 b, c) the plastids were also beginning to accumulate starch.

The author thanks Professor P. R. Bell, University College, London, for critical reading of the manuscript and Messrs D. A. Davies and J. Mackey for technical assistance. The financial support provided by a Science Research Council Research Fellowship is gratefully acknowledged.