vesicles in guard-cell walls and their possible roles

J. Cell Sci. 38, 83-95 (i979)

83

Printed in Great Britain © Company of Biologists Limited

VESICLES IN GUARD-CELL WALLS AND THEIR POSSIBLE ROLES IN THE STOMATAL MECHANISM ROBERT JOHN CORK AND BRENDA J. NELMES Astbury Department of Biophysics, University of Leeds, Leeds LSz gJT, U.K.

SUMMARY A variety of vesicular inclusions have been observed in guard-cell walls. They have been seen in a number of species and are mostly membrane-bound. They appear confined to the upper and ventral walls of the guard cell. A number of possible origins and functions are discussed. They may be involved in wall deposition or cuticle formation possibly in a role similar to ectodesmata. They also may serve to increase the wall-plasmalemma interface to enhance the movement of ions during the stomatal mechanism.

INTRODUCTION

The stomatal mechanism is based on the translation of osmotically controlled turgor pressure changes into movement of the guard-cell walls. This is dependent on the specialized architecture of the guard cells. It is therefore probable that an ultrastructural study of guard-cell walls will give information which may be useful in analysing the mechanics of stomata. Considering the number of published reports on stomata, there are relatively few ultrastructural studies of guard cells. Only a very small number contain any details of the wall structure (Setterfield, 1957; Singh & Srivastava, 1973; Palevitz & Hepler, 1976; Peterson, Firminger & Dobrindt, 1975). The guard-cell walls are thought to have a radial arrangement of cellulose microfibrils which assists in the mechanism (Aylor, Parlange & Krikorian, 1973; Zeigenspeck, 1954). Otherwise they are regarded as thickened primary walls (Setterfield, 1957)-

The results presented here are part of a study of guard-cell ultrastructure. It is hoped that the details of guard-cell wall structure will be useful in resolving some of the problems encountered by the variety of mathematical and engineering models proposed for stomata (Aylor et al. 1973, 1975; Cooke, Debaerde, Rand & Mang, 1976; Delwiche & Cooke, 1977; Demichele & Sharpe, 1973, 1974; Shoemaker & Srivastava, 1973).

R. J. Cork and B. jf. Nelmes MATERIALS AND METHODS The species examined included: Chlornphytiim antheridium (Spider plant); Sedum telephium (Orpine); Sambucus nigra (Elder); Narcissus pseudonarcissus (Daffodil); Scilla nonscripta

(Bluebell); Pelargonium sonale (Geranium); and Taraxacum officinah (Dandelion). Lower epidermal strips were prepared for electron microscopy by one of the following methods. (1) Fixation in 3 % glutaraldehyde in phosphate buffer (002 M, pH 7-0) for 3 h at 4 CC. Three 10-min washes in buffer followed by posttixation in 1 % osmium tetroxide in 002 M buffer pH 70 for 1 h at 4 °C. After a rinse in buffer the material was dehydrated in an ethanol series and then processed for embedding in propylene oxide at room temperature. Embedding was in Spurr resin, polymerized at 70 °C for 8 h. (2) Removal of pectic substances by treatment for 3 h in 10% pectinase in a 1 % aqueous solution of peptone (Chayen, 1952) at room temperature, followed by fixation and embedding as in (1). (3) Staining for pectins with hydroxylamine and sodium hydroxide followed by alcoholic hydrochloric acid with alcoholic ferric chloride (Gee, Reeve & McCready, 1959). The modifications of Albersheim & Killias (1963) were followed to adapt this stain for electron microscopy. After washing the stained material was dehydrated and embedded as in (1).

10 Jim

Longitudinal

Paradermal

Transverse

Fig. 1. Diagram of 3 possible views of a stoma indicating positions of walls, thickenings and vesicular regions (stippled areas), dio, dorsal wall; hv, lower wall; p, pore; 5, stomatal ledge; MM, upper wall; vw, ventral walls. Arrow points to exterior of

Fig. 2. Paradermal section, Clilorophytum guard cell. Glutaraldehyde/osmium fixation, stained with uranyl acetate and lead citrate. Electron-lucent 'blotches' (v) seen along inner edge of ventral wall (to). Some elaboration of the plasmalemma (pi) is seen extending into an electron-lucent space (s) between wall and cytoplasm (c). x 54300. Fig. 3. Transverse section, Clilorophytum guard cell. Corner of upper wall with the ventral wall. Treated with pectinase and stained with uranyl acetate and lead citrate. Large vesicles in upper wall thickening and smaller ones along outer edge of ventral wall near cuticle (CM). Arrow points towards exterior of leaf, i.e. out of pore (p). x 30000. Fig. 4. Paradermal section, Chlorophyhim guard cell. Inner part of central ventral wall. Treated and stained as Fig. 3. Large vesicles, many with 'tails' forming network, x73200. Fig. 5. As Fig. 4, but outer part of central ventral wall. Smaller vesicles in electrondense matrix, x 73 200.

Vesicles in guard-cell walls

w

W

86

R. J. Cork and B. J. Nelmes

(4) Removal of pectic substances by treatment with 1% pectinase in pH3'5 phosphate buffer at 40 °C for times varying from 4 to 24 h (Deshpande, 1976). After treatment the material was rinsed in distilled water then dehydrated and embedded as in (1). (5) Some material was treated with DAB to localize microbodies (Frederick & Newcomb, 1971). Otherwise these cells were stained and fixed as in (1). Following embedding, all blocks were trimmed and then sectioned on an LKB 4800A or Reichart OmU2 microtome using a diamond knife. Sections were picked up on 400-mesh uncoated copper grids and post-stained with either uranyl acetate and lead citrate or 10 % KMnO4. Sections were examined on a Philips EM200 at 40 or 60 kV. OBSERVATIONS Chlorophytum Quite often when examining guard-cell walls, electron-lucent 'blotches' are seen. In paradermal sections they are seen along the inner edge of the ventral wall (Fig. 1). At high magnification these 'blotches' are seen to be vesicles of various shapes and sizes. They vary from elliptical to filamentous and sometimes appear as elliptical vesicles with long 'tails' (Fig. 2). The roughly circular vesicles have an average diameter of around 50-60 nm in normal osmium/glutaraldehyde-fixed material. The long vesicles have an average width of 15-18 nm and may be up to 1 /tm in length. In general the wider they are the shorter they are. When the walls are treated to remove most of the pectic substances (treatment (2)), areas of vesicles are revealed (Fig. 3). These vesicles are surrounded by a matrix of electron-dense material not removed by the treatment. These vesicular regions of the wall are found only in defined areas; the upper thickened wall and the ventral wall bordering the upper edge of the pore (Figs. 1, 3). Thus, in paradermal sections through the upper part of the guard cell, vesicles appear in 2 regions; the inner thickened part of the ventral wall (in fact the upper wall) (Fig. 4) and along the edge of the ventral wall by the pore (Fig. 5). The vesicles in the latter region tend to be smaller, more oriented and in a more electron-dense matrix (Figs. 5, 3). They are very elongated in transverse sections (Fig. 3) and almost uniformly elliptical in

Fig. 6. A: Paradermal section Chlorophytum guard cell through upper wall. Elliptical vesicles fanning out from pore, aligned to main microfibril direction. Treated with pectinase and stained with 10% KMnO.t. x 15000. B: Similar section stained with ferric chloride/hydroxylamine and poststained with uranyl acetate and lead citrate. Vesicles still visible but not electron-dense matrix, x 55200. Fig. 7. Longitudinal section, Chlorophytum guard-cell upper wall. Treated as Fig. 6. Bands of vesicles of differing sizes, mostly in thickened part of wall. Layer of densely staining membranes (tn) bounding vesicular area, x 23700. Fig. 8. Paradermal section, Chlorophytum guard cell, junction of ventral and upper walls. Treated as Fig. 6 but for longer time in pectinase. Transversely sectioned fibrils in ventral wall (vw) adjacent to more obliquely sectioned fibrils of upper wall (MZO). Irregular membrane-bound vesicles in upper wall (arrows), x 12 780. Fig. 9. Transverse section, Chlorophytum guard cell, corner of upper and ventral walls. Treated as Fig. 8. Membrane-bound vesicles (arrows) associated with densely staining area of wall, x 17040.


88


paradermal sections (Fig. 5). These regions of vesicles appear in some sections to be continuous with the cuticle (Fig. 3). The vesicles in the upper wall are much larger and elliptical in both transverse and paradermal sections (Figs. 3, 4). They often seem to have 'tails' and these sometimes appear to form a network (Fig. 4). It is noteworthy that in these treated walls the usual wall material is invisible to the electron microscope and the only regions seen are those with vesicles. When the pectin-cleared walls are stained with permanganate (treatment (4)), the wall as a whole becomes visible as permanganate is a stain for cellulose fibrils (Deshpande, 1976). The regions of vesicles are still clearly seen and often the vesicular region is more electron-dense than the surrounding wall (Fig. 6A). The vesicles in tangential sections through the upper wall (Fig. 6 A, B) are mostly elliptical but oriented so that their long axes are parallel to the predominant direction of the microfibrils. It often appears that these vesicles form part of a loose network in the thickening of the upper wall. In longitudinal sections the regions of vesicular wall are clearly seen (Fig. 7). Again, the main area of vesicles is the middle band of the upper wall. In some sections this band appears to be bounded by a layer of densely staining membranes or fibrils (Fig. 7). The vesicles in these treated walls appear to be larger than those in osmium/glutaraldehyde-fixed walls, being about 100 nm in diameter. If the cells are treated for longer periods with pectinase a very skeletal wall is seen (Figs. 8, 9). The vesicles in these walls appear as irregular membrane-bound sacs (Figs. 8, 9), often associated with an irregular network of cellulose microfibrils and densely staining areas of wall. Sedum Vesicles are clearly seen in the Sedum guard-cell upper wall (Fig. 10). They are also seen near the upper cuticle (Fig. 11). At high magnifications the vesicles in both regions are rather irregular bodies with a dark limiting border (Figs. 10-12). The vesicles near the cuticle appear to merge with the cuticle in some areas (Fig. 11). The cuticle itself seems to be composed of discrete globules (Fig. n ) . Figs. 10-12. Transverse section of Sedum guard cell. Incubated with DAB, fixed with glutaraldehyde/osmium and poststained with uranyl acetate and lead citrate. Fig. 10. Central region of upper wall, irregular vesicles with dark limiting borders. Fig. n . Similar vesicles nearer cuticle, some appearing continuous with cuticle composed of globules (arrows), x 85000. Fig. 12. Array of rectangular vesicles in central part of upper wall, x 222000. Figs. 13-16. Paradermal sections, daffodil guard cells, ventral wall. Glutaraldehyde/ osmium fixation, post-stained with uranyl acetate and lead citrate. Fig. 13. Wall ingrowth with membrane-bound vesicles, x 55200. Fig. 14. Membrane-bound vesicles with cytoplasmic contents, x 55000. Fig. 15. Cytoplasmic inclusion containing membrane-bound vesicles, x 55000. Fig. 16. Wall ingrowths enclosing area of cytoplasm. The cytoplasmic inclusion appears continuous with the main cell contents and is enclosed by a membrane possibly an extension of the plasmalemma. x 49500.


89

90

R. J. Cork and B. jf. Nelmes

In some sections the vesicles are grouped together as parallel filaments running circumferentially in the upper wall (Fig. 12). These filamentous vesicles are composed of shorter rectangular segments (Fig. 12). Daffodil In daffodil guard cells the vesicles appear to be of a different form. They are almost all membrane-bound (Figs. 13-15) and are usually found in areas of cytoplasm which are enclosed in the ventral or upper wall (Figs. 14-16). In some sections the vesicles and cytoplasmic inclusions seem to be continuous with the plasmalemma and main cell contents (Fig. 16). In some regions the 'blotchy' type of vesicle as seen in Chlorophytum and Seduvi is also visible. Elder Initial studies of a few sections indicate that vesicles are present. They seem to be especially associated with the cuticle on the upper guard-cell wall. The vesicular wall forms an intermediate band between the usual wall and the cuticle (Fig. 17). Bluebell Very small, flattened, membrane-bound sacs are seen localized in the upper guardcell wall (Fig. 18). Subsidiary cells On a few occasions vesicles have been observed in walls other than the ventral/ upper wall of the guard cells: (a) in a region where the subsidiary cell wall has a bend some rectangular vesicles are seen in the wall (Fig. 19); and (b) surrounding an area of subsidiary cell wall, which appears to have been damaged or is part of some sort of pore, are 2 'plates' of wall with many vesicles similar to those seen in guard cells (Fig. 20).

Fig. 17. Paradermal section, elder guard cell, through upper wall and cuticle. Glutaraldehyde/osmium fixation, post-stained with uranyl acetate and lead citrate. Vesicular wall forming band between cuticle and wall proper, x 72600. Fig. 18. A: Longitudinal section, bluebell guard-cell upper wall. Glutaraldehyde/ osmium fixation, post-stained with uranyl acetate and lead citrate. Membranous inclusions in layers, x 55200. B : A membranous sac. X171000. Fig. 19. Paradermal section, Sedum subsidiary cell wall. Incubated with DAB, fixed with glutaraldehyde/osmium, post-stained with uranyl acetate and lead citrate. Vesicles at bend in wall, aligned to microfibril direction, x 36600. Fig. 20. Paradermal section, Chlorophytum epidermal cell wall. Pectinase-treated, stained with 10 % KMnO 4 . Vesicular plates enclosing constricted epidermal cell wall, x 12 780.


92


DISCUSSION

Although these vesicles have not been reported in previous examinations of guard cells, very similar structures have been reported in the outer epidermal-cell wall (O'Brien, 1965). There are a number of reasons for thinking they are not merely artifacts. Firstly, they are seen, albeit in a variety of forms, in a number of species examined. They have not, however, been confirmed in all species studied, e.g. they have not been seen in Dandelion. Vesicles are not always visible in all appropriate sections but they are seen in enough sections to warrant the belief that they represent real structures in the walls of some guard cells. Secondly, they are seen by using a variety of fixation and staining techniques. This argues against them being the product of some chemical or physical effect during preparation for the electron microscope. Thirdly, most of the preparative techniques have been used previously to study other types of cell wall, e.g. epidermal cells, collenchyma, xylem, etc. No vesicles have been reported as a result of these techniques and there seems no reason why they should cause vesicles specifically in guard cells. Finally, the vesicles are seen only in the upper and ventral walls of the guard cells. The rest of the walls appear to be composed of cellulose microfibrils in a matrix, similar to most other cell walls. It would seem reasonable to assume that if the vesicles were artifacts, they would be seen in all walls, not just in certain defined areas. It is possible that the vesicles are the spaces between a loose array of cellulose microfibrils. This could be the result of stresses pulling the fibrils apart so that they develop a trellis configuration (Boyd & Foster, 1975). Guard cells do elongate considerably during development and they are also subject to stress and strain during stomatal movement. However, it seems unlikely that such spaces would be membrane-bound, and this is often the case with the observed vesicles. Also if they were just spaces the vesicles should always run with their long axes parallel to the main microfibril direction. While this is often the case (Figs. 6, 3) it is not always so (Fig. 4). Studies of tilted sections are not conclusive, but do indicate that the elongated and elliptical vesicles represent different views of a sausage-shaped body. There are a number of other possibilities for these structures. They could be separate membrane-bound packages, transported across the plasmalemma and into the wall. They may remain stationary in the wall or they could pass through it to the cuticle or the environment. From some views it also seems possible that the vesicles are part of a network of vesicles and tubules extending from the plasmalemma into the wall. It is, of course, possible that there is more than one type of vesicle and that all of these possibilities are represented. A variety of vesicular bodies have been seen associated with the plasmalemma of guard cells (Sanchez, 1977; Chabot & Chabot, 1977; Srivastava & Singh, 1972; Ziegler, Shmueli & Lange, 1974; Thomson & Dejournett, 1970). It has never been resolved whether these are involved in import or export of materials to or from


93

the cytoplasm. Srivastava & Singh (1972) suggested that they might be involved in cutin migration across the plasmalemma. They proposed that the cutin passed through the wall porosities to be laid down at the cuticle. Chabot & Chabot (1977) described a vesicular region in the cuticle and upper wall of Abies epidermal cells. It seems plausible that the cutin may be transported in these vesicles and discharged at the cuticle (Fig. u ) . This would give rise to a vesicular cuticle-wall interface (Fig. 17). It may be that these vesicles are involved in the transport of other cell-wall materials into the wall. A variety of components have been reported for guard-cell walls, including lignin (Chabot & Chabot, 1977), callose (Peterson et al. 1975) and silica (Hayward & Parry, 1973). The vesicles might carry or hold these components or their precursors in the wall. If the vesicles are transporting materials through the wall, they could perhaps be regarded as a form of ectodesmata. These pathways through the wall have been the subject of much debate and no satisfactory structure has been ascribed to them (Merkens, DeZoeten & Gaard, 1972; Hall, 1967; Schb'nherr & Bukovac, 1970). They are often thought to be no more than artifacts caused by some special property of the wall or cuticle in localized areas. They are said to be quite transient in the wall, and this could be explained if they were vesicles moving through the wall. Ectodesmata are often associated with guard cells (Schonherr & Bukovac, 1970), but plasmodesmata are thought to be absent or uncommon (Singh & Srivastava, 1973; Thomson & Dejournett, 1970; Sanchez, 1977). The lack of plasmodesmata is interesting when comparing guard cells with other types of cell which have similar physiology and function. The active transport of ions is important in the stomatal mechanism; it is also thought to play a role in the functions of salt gland cells and transfer cells. Both of these types of cell have a cytoplasm very similar to guard cells, except that they have numerous plasmodesmata. The walls of transfer cells are characterized by a labyrinth of wall protuberances which extends into the cytoplasm and gives a large surface area for the plasmalemma (Gunning, Pate & Briarty, 1968). Salt glands also have wall protuberances (Thomson & Lui, 1967) as do the digestive glands of the Venus flytrap (Schwab, Simmons & Scala, 1969). In all these cell types the complex wall-cytoplasm interface is thought to assist in the transport of substances and ions across the highly invaginated plasmalemma. These cells also have an electron-lucent space between plasmalemma and wall proper (cf. Figs. 2, 13). Browning & Gunning (1977) suggested that this is an artifact when seen in transfer cells. It was thought that it was caused by the swelling of a wall component during fixation and embedding. However in other cells (Thomson & Lui, 1967; Schwab et al. 1967) it is thought to be an integral part of the interfacial apparatus which plays a role in secretion. In guard cells this labyrinth is much less pronounced, but it may provide a complex surface for the plasmalemma across which ion transport takes place. A difficulty with this hypothesis is that the vesicular wall is only observed in walls which have no adjoining cells, i.e. they are bordered on one side by the external environment. Therefore it seems unlikely that ions would be transported through these walls. Membranous inclusions have been reported in some types of cell wall and these 7

CEL 38

94

R. J- Cork and B. J. Nelmes

may provide one clue to why the upper thickened guard-cell wall has them. In pea internodes treated with 2,3,5-triiodobenzoic acid (Bouck & Galston, 1967) membranous sacs are seen in the thickened walls. This is because wall deposition continues across the plasmalemma but cell elongation is inhibited. Thus the excess membrane material is incorporated into layers in the thickened walls. A similar wall structure is seen in the developing sporangia of Phytophthora parasitica (Hemmes & Hohl, 1969), where the wall develops in a non-elongating cell. These membrane inclusions are similar to those seen in some guard-cell walls especially those in defined layers (Figs. 7, 13, 18). This may be caused if the thickening of the upper wall is laid down after the guard cell has stopped enlarging. In conclusion it must be said that none of these theories fully explains all the observations. It is probable that these vesicles have more than one origin. More study, especially histochemical localization, would be useful in elucidating the role of these vesicles in the stomatal mechanism. This work wasfinancedby a grant from the Science Research Council. We are indebted to Messrs R. White, A. Hick, L. Child and W. D. Brain for their technical assistance and we should also like to thank Professor R. D. Preston for his inspiration and encouragement.

•RENCES RSHEIM, P. & KILLIAS, U. (1963). Histochemical localization at the electron microscope el. Am. J. Bot. 50, 732^745. )R, D. E., PARLANCE, J-Y. & KRIKORIAN, A. D. (1973). Stomatal mechanics. Am. J. Bot. , 163-171. )R, D. E., PARLANGE, J-Y. & KRIKORIAN, A. D. (1975)- Comment on some recent models stomatal mechanics. J. theoret. Biol. 54, 395-397. :K, G. B. & GALSTON, A. W. (1967). Cell wall deposition and the distribution of cytoismic elements after treatment of pea internodes with the auxin analog 2,3,5-triiodobenzoic d. Ann. N. Y. Acad. Sci. 144, 34-48. ), J. D. & FOSTER, R. C. (1975). Microfibrils in primary and secondary wall growth yelop trellis configurations. Can. J. Bot. 53, 2687-2701. VNING, A. J. & GUNNING, B. E. S. (1977). An ultrastructural and cytochemical study of : wall membrane apparatus of transfer cells using freeze substitution. Protoplasma 93, 26. !OT, J. F. & CHABOT, B. F. (1977). Ultrastructure of the epidermis and stomatal complex Balsam fir {Abies balsamea). Can. J. Bot. 55, 1064-1075. rEN, J. (1952). Isolation of cells by pectinase. Nature, Lond. 170, 1070-1072. CE, J. R., DEBAERDE, J. G., RAND, R. H. & MANG, H. A. (1976). Finite-element shell alysis of guard cell deformation. Trans. Am. Soc. agric. Engrs 19, 1107-1121. VICHE, M. J. & COOKE, J. R. (1977). An analytical model of the hydraulic aspects of imatal dynamics. .7. theoret. Biol. 69, 113-141. 'ICHELE, D. W. & SHARPE, P. J. H. (1973). An analysis of the mechanics of guard cell )tion. J. theoret. Biol. 41, 77-96. ICHELE, D. W. & SHARPE, P. J. H. (1974). A parametric analysis of the anatomy and ysiology of the stomata. Agric. Meteorol. 14, 229-241. [PANDE, B. P. (1976). Observations on the fine structure of plant cell walls. I. Use of rmanganate staining. Ann. Bot. 40, 433-437. ERICK, S. E. & NEWCOMB, E. H. (1971). Ultrastructure and distribution of microbodies leaves of grasses with and without C02-photorespiration. Planta 96, 152-174. M., REEVE, R. M. & MCCREADY, R. M. (1959). A stain for pectin. J. Agric. Food Client. 1, 34-38.


95

GUNNING, B. E. S., PATE, J. S. & BRIARTY, L. G. (1968). Specialized 'transfer cells' in minor veins of leaves and their possible significance in phloem translocation. J. Cell Biol. 37, C7-Ci3. HALL, D. M. (1967). The ultrastructure of wax deposits on plant leaf surfaces. II. Cuticular pores and wax formation. J. Ultrastruct. Res. 17, 34-45. HAYWARD, D. M. & PARRY, D. (1973). Silicon distribution in barley. Ann. bot. 37, 579-591. HEMMES, D. E. & HOHL, H. R. (1969). Ultrastructure changes in directly germinating sporangia of Phytophthora parasitica. Am.J. Bot. 56, 300-313. MERKENS, W. S. W., DEZOETEN, G. A. & GAARD, G. (1972). Observations on ectodesmata

and the virus infection process. .7. Ultrastruct. Res. 41, 397-405. O'BRIEN, T . P. (1965). Note on an unusual structure in the outer epidermal wall of the Avena coleoptile. Protoplasma 60, 136-140. PALEVITZ, B. A. & HEPLER, P. K. (1976). Cellulose microfibril orientation and cell shaping in developing guard cells of Allium: the role of microtubules and ion accumulation. Planta 132, 71-93PETERSON, R. L., FIRMINGER, M. S. & DOBRINDT, L. A. (1975). Nature of the guard cell wall in leaf stomata of three Ophioglossutn species. Can. J. Bot. 53, 1698-1711. SANCHEZ, S. M. (1977). The fine structure of the guard cells of Helianthns annus. Am. jf. Bot. 64, 814-824. SCHONHERR, J. & BUKOVAC, M. J. (1970). Preferential polar pathways in the cuticle and their relationship to ectodesmata. Planta 92, 189-201. SCHWAB, D. W., SIMMONS, E. & SCALA, J. (1969). Fine structure changes during function of the digestive gland of Venus flytrap. Am. J. Bot. 56, 88-100. SETTERFIELD, G. (1957). Fine structure of guard-cell walls in Avena coleoptiles. Can. J. Bot. 35, 79!-794SHOEMAKER, E. M. & SRIVASTAVA, L. M. (1973). The mechanics of stomatal opening in corn leaves. J. theoret. Biol. 42, 219-225. SINGH, A. P. & SRIVASTAVA, L. M. (1973). The fine structure of pea stomata. Protoplasma 76, 61-82. SRIVASTAVA, L. M. & SINGH, A. P. (1972). Stomatal structure in corn leaves. J. Ultrastruct. Res. 39, 345-363THOMSON, W. W. & DEJOURNETT, R. (1970). Studies on the ultrastructure of the guard cells of Opuntia. Am. J. Bot. 57, 309-316. THOMSON, W. W. & Lui, L. L. (1967). Ultrastructural features of the salt gland of Tamarix aphylla. Planta 73, 201-220. ZIEGENSPECK, H. (1954). Das Vorkommen von Fila in radialer Anordnung in den Schliesszellen. Protoplasma 44, 385-388. ZIEGLER, H., SHMUELI, E. & LANGE, G. (1974). Structure and function of the stomata of

Zea mays. I. The development. Cytobiologie 9, 162-168.

{Received 23 February 1979)

7-2