Gametophyte morphology and ultrastructure of the extremely deep shade fern, Trichomanes speciosum


  • Kittima Makgomol,

    1. School of Biological Sciences, University of Manchester, 3.614 Stopford Building, Oxford Road, Manchester, M13 9PT, UK
    Search for more papers by this author
  • Elizabeth Sheffield

    Corresponding author
    1. School of Biological Sciences, University of Manchester, 3.614 Stopford Building, Oxford Road, Manchester, M13 9PT, UK
      Author for correspondence: Elizabeth Sheffield Tel: +44 (0)161 2753905 Fax: +44 (0)161 2753938
    Search for more papers by this author

Author for correspondence: Elizabeth Sheffield Tel: +44 (0)161 2753905 Fax: +44 (0)161 2753938


  •  The extent to which macro- and micromorphological features might contribute to tolerance of extremely deep shade by Trichomanes speciosum, a member of the filmy ferns (Hymenophyllaceae), is reported here.
  •  Confocal laser scanning, transmission and scanning electron microscopy were used to study the ultrastructure of gametophytes and sporophyte leaves.
  •  Gametophyte filament cells contain numerous small, spherical or ovoid chloroplasts, whereas sporophyte leaf cells have fewer, slightly larger, disc-shaped chloroplasts. The chloroplast grana of gametophytic cells have fewer thylakoids than sporophyte cells, although grana are not numerous in either. Gametophyte filament cell walls resemble those of sporophyte leaf cells, with two or more layers of electron-opaque material and covered in a thin cuticle. Gemma cell wall ultrastructure does not differ from that of gametophyte filament cells; rhizoid cell walls are thick and several-layered.
  •  Neither gametophyte filaments nor sporophyte leaves have chloroplasts of the extreme forms reported for deep shade fern or angiosperm leaves. The success of the fern is attributed to a low metabolic rate and inability of other species to cope with extreme low light.


Killarney Fern (Trichomanes speciosum Willd.), a member of filmy ferns (Hymenophyllaceae), is the only European representative of its predominantly tropical genus. It has long been regarded as one of Europe’s rarest and most vulnerable species (Ratcliffe et al., 1993). A requirement for constantly humid, winter-warm conditions effectively restricts the species to extreme low light environments in the northern and eastern part of its range. In the past decade, this fern has been the subject of intense study. Uniquely among European ferns, it reproduces via branched filamentous perennial gametophytes with specialized asexual propagules (gemmae). These allow the species to persist and disperse independently of the sporophyte generation, eventually forming extensive clonal patches (Rumsey et al., 1999). Long overlooked, not least because of their deep shade habitat, the gametophytes are now known to occur in many more sites than the sporophyte, which is restricted to Europe’s Atlantic fringe. Indeed, the gametophyte extends over several hundred kilometers beyond the current sporophytic range, into Continental Europe (Rumsey et al., 1998b). This inequality in distribution reflects the difference in ecological tolerances shown by the respective generations; the gametophyte is capable of withstanding drier, cooler and darker conditions.

In recent years, large thriving colonies of T. speciosum gametophytes have been found at many sites in Britain and Europe (Rumsey et al., 1998a,b). The plants grow mainly directly on rock faces, often cave roofs, by the sides of streams that are relatively humid, but not wet. The altitude at which they are found in British Isles ranges from near sea level to c. 520 m on the Moel Hebog range, North Wales (Rumsey, unpublished). They occur in habitats subject to mean levels of photosynthetically active radiation (PAR) below 1 µmol m−2 s−1 and are restricted to sunless, sheltered aspects of sites in Britain (Rumsey, 1994; Rumsey et al., 1998a).

Most studies of T. speciosum have been ecological or environmental, and there is scant microscopical or ultrastructural information in the literature published on this species. Johnson et al. (2000) concluded from their studies that the adaptations of T. speciosum to extreme low light included physiological characteristics. They suggested that gametophytes of this fern are able to survive in the British Isles by maintaining a very low metabolic rate, growing at low temperature and making efficient use of light that is available. Our aim was to determine the extent to which macro- and micromorphological features of this fern might contribute to its tolerance of extremely deep shade.

Light microscope studies have documented the size and number of chloroplasts per cell in some plants, and these parameters tend to fall within similar limits for particular taxonomic or ecological grouping. Hornworts, for example, have one chloroplast per cell in gametophytes and two per cell in their sporophytes (Brown & Lemmon, 1990). Single or low numbers of chloroplasts are found in fully differentiated moss leaf cells. Butterfass (1971) investigated chloroplasts in cells of Sphagnum cuspidatum leaves. He reported that the number of chloroplasts increases from the middle of each leaf toward the tip cell, up to at least 16 chloroplasts per cell. Some species of the pteridophyte Selaginella have one or two chloroplasts per cell (Hébant & Lee, 1984) but the number of chloroplasts in other vascular plant cells can reach up to several hundred. In ferns and angiosperms there are normally 40–200 chloroplasts per cell, typically 4–6 µm in diameter (A. Nasrulhaq-Boyce & J. G. Duckett, unpublished data, cited in Nasrulhaq-Boyce & Duckett, 1991). In some species of Hymenophyllaceae, there are 100–200 chloroplasts in leaf cells and each chloroplast is 3–6 µm in diameter (A. Nasrulhaq-Boyce & J. G. Duckett, unpublished data, cited in Nasrulhaq-Boyce & Duckett, 1991).

According to Boardman (1977) there tend to be fewer chloroplasts in leaf cells of shade than sun plants, but the chloroplasts are usually large. Bartels (1965) reported that each cell in one layer of leaves of the shade plant Peperomia metallica contains only four large chloroplasts. Neumann (1973) reported that cells in the palisade parenchyma of this plant as containing giant chloroplasts, 20–25 µm in diameter with large stacks of thylakoids. Nasrulhaq-Boyce & Duckett (1991) described that the upper epidermal cells of the leaves of one extreme-shade tropical fern, Teratophyllum rotundifoliatum have only 4–12 (mean 6.9) large chloroplasts 9.4–27 (mean 17.9) µm in diameter. The lower epidermis has smaller chloroplasts, 5.0–19.5 (mean 7.6) µm in diameter, and 21–83 (mean 53) chloroplasts per cell.

Plants grown at high irradiances tend to have chloroplasts with grana comprising small numbers of thylakoids, sometimes only two per stack. In contrast, chloroplasts of shade plants typically contain many thylakoids per granum; they may reach one hundred or more (Ballantine & Forde, 1970; Boardman, 1977; Duckett, 1986; Sarafis, 1998). For example, some chloroplasts of an extreme-shade angiosperm, Alocasia macrorrhiza, have grana comprising more than 100 thylakoids with a mean around 43 (Anderson et al., 1973). Nasrulhaq-Boyce & Duckett (1991) showed that chloroplasts in the upper epidermal cells of Teratophyllum rotundifoliatum have grana comprising up 280 thylakoids, with a mean around 86; the lower epidermal cells have up to 84, with a mean of 29. Few authors appear to have compared gametophytes and sporophytes of the same species of pteridophyte, although it is clear that there may be dramatic differences in the number and size of chloroplasts and in number of thylakoids constituting grana between chloroplasts of two different tissues of the same plant (Ligrone et al., 1993)

Some extreme shade plants with iridescent blue leaves have specialized chloroplasts (Graham et al., 1993; Lee, 1997). The chloroplasts of the fern Trichomanes elegans, for example, have uniquely structured grana, each containing a small, precise number of lamellae providing the basis of the iridescent appearance. The chloroplasts in other cells of such plants are typical of extreme shade chloroplasts, with thick grana stacks (Graham et al., 1993). T. speciosum grows in even deeper shade than any of the other shade ferns investigated ultrastructurally to date, but a pilot ultrastructural study revealed no signs of extremes of shade chloroplast structure (Raine, 1994). This latter study involved only gametophytes; there are no published studies relating to T. speciosum gametophyte or sporophyte ultrastructure. The first objective of our study was therefore to establish the extent to which sporophyte and gametophyte share ultrastructural characteristics; the second was to document the morphological characteristics of the gametophyte of this extremely deep shade fern.

Materials and Methods

Plant material

Trichomanes speciosum Willd. gametophytes were collected (under licence) from a woodland site near Hebden Bridge, Yorkshire, UK, where long-term ecological monitoring has been performed (Rumsey, 1994). Plants were examined immediately, described as ‘field-collected specimens’, or cultured at a temperature of 20°C ± 2°C on potting compost or agar in the laboratory for stock cultures, described as ‘cultured specimens’. Light was supplied by fluorescent daylight tubes and the containers of plants were covered with filter films (Lee filters: no. 211, Lee filters, Andover, UK) to give a PAR of 5 µmol m−2 s−1 in a 12-h light and 12 h dark cycle. Glass crystallising dishes were used as compost culture vessels. Plants were potted up on a sterilized sand surface overlaid with Levington M2 Compost, and dampened with distilled water. Plants grown in 9-cm diameter Petri dishes on agar followed the method described in Raine & Sheffield (1997): one-tenth strength Hoagland’s medium with 2% sucrose was the basal medium, adjusted to pH 5.9 and solidified with 0.8% bacteriological agar (No.1 Oxoid).

Sporophyte leaf samples were collected from cultivated plants grown at the Experimental Grounds in the University of Manchester in Fallowfield, UK, in the ‘moss house’, a glass house with no supplementary heat or lighting in a position receiving a typical midday summer PAR of 5 µmol m−2 s−1 on a sunny day.

Light and electron microscopy

Field-collected specimens and cultured plants were initially observed and photographed using light and stereo microscopy. The topography and three-dimensional arrangement of filaments were examined via low temperature scanning electron microscopy, using a Cambridge S200 SEM fitted with a Hexland Cryotrans CT 1000 chamber. Material was orientated on the specimen slug with mounting medium (Tissue-tek OCT compound, Sakwa Finetek Europe B.V., Zoeterwode, The Netherlands) and then plunged into slushed liquid nitrogen and transferred under vacuum to the cooled microscope. The specimens were gold-coated, then scanned and photographed with black and white film (Delta) while the microscope was maintained below −175°C. Gametophyte filaments, gemmifers and gemmae were also observed in their fresh, natural state using an environmental scanning electron microscope (Model E-3, Philips Electrosean, MA, USA). Digital images were obtained from specimens in the ESEM chamber set at 8°C with a vapour pressure of 8.5 Torr.

Confocal microscopy

Due to the large number of chloroplasts per Trichomanes cell, visual counting using classical light microscopy produced large errors. Confocal laser scanning microscopy (CLSM) separates densely packed or overlapping clusters of chloroplasts via gradient optical sections, so was used to detect chloroplasts via the natural autofluorescence of chlorophyll. Gametophyte samples from cultured specimens and sporophyte leaves were mounted aqueously on glass microscope slides, coverslips were applied and sealed with nail polish. Specimens were viewed with a Biorad MRC 600, confocal laser scanning microscope, using ×40 (NA 0.91) and ×60 (NA 1.40) oil immersion objectives. The fluorescence was excited with an argon/krypton ion laser at wavelengths of 488, 568 and 647 nm. Emission Filter FR2 was used and the fluorescence intensity operated on the normal power setting. Chlorophyll autofluorescence was detected using 30% transmission. For each optical plane, images were gathered using Kalman averaging integration (to reduce noise) of five cycles. Z-series profiles of between 12 and 26 optical sections were collected at intervals of 2.5 µm. Forty fully differentiated photosynthetic gametophyte cells from different filaments of the same culture were documented (apical and basal cells were excluded). Forty sporophyte leaf cells (lamina areas) from four different leaves of the same plant were used to estimate chloroplast numbers.

The numbers of chloroplasts per cell were estimated by drawing each optical section image from the computer monitor on a transparent film. This allowed identification of repeated images of the same chloroplasts and compensation for overlapping and clustering.

Three-dimensional images of chloroplasts taken with a ×60 objective were used to estimate chloroplast sizes by using PC Image Analysis and Measurement software (Fosterfindlay Associates Ltd, Newcastle Technopole, Newcastle Upon Tyne, UK). The maximum diameter of 10 chloroplasts from each of 10 cells was measured (for both sporophytes and gametophytes).

Transmission electron microscopy

The technique used to fix the specimens was modified from Duckett & Ligrone (1992) by using a mixture of 1% formaldehyde, 2% glutaraldehyde and 1% caffeine in 0.05 M phosphate buffer, pH 6.8. Gametophyte filaments were taken from cultured specimens, sequentially cut on 4% agar, the surface of which had been flooded with buffer, into segments approximately 2 mm long, and rapidly transferred to a vial containing fresh fixative. Gemmae were removed from the stock culture and put into suspension in buffer solution then transferred to an Eppendorf tube for centrifugation. Sporophyte leaf samples were fixed the same as gametophyte filaments. All samples were fixed for 3 h, then washed with two changes of buffer for 10 min at room temperature. The post fixative was 1% osmium tetroxide in the same buffer which was added and left for 1 h at room temperature. Fixed specimens were washed once with buffer, via centrifugation for gemmae. The specimens were dehydrated through a graded ethanol series to 100% over 3 d at room temperature. The material was infiltrated with Spurr’s hard resin for at least 5 d and embedded in fresh Spurr’s hard resin. Blocks were left in the polymerizing oven at 60°C overnight. Sections were cut with diamond or glass knives and mounted on 3 mm diameter copper grids. Thin sections were stained with 2% uranyl acetate in 70% alcohol for 20 min, followed by 0.3% lead citrate in 0.1 M sodium hydroxide for 5 min and were washed in distilled water. Sections were viewed and photographed on a Phillips 400 transmission electron microscope.

Electron micrographs of gametophyte filament and sporophyte leaf transverse sections were analysed to determine the average number of thylakoids per granum. Five median longitudinal sections of fully developed chloroplasts from each of four different filaments of the same culture and leaves of the same plant were analysed. Thylakoids in two granal stacks in each chloroplast were counted.

For all parameters, the significance of differences between gametophyte and sporophyte leaf chloroplasts was evaluated by the Student’s t-test.


Gametophyte morphology

T. speciosum gametophytes are branched filaments, forming mats that are bright green. No differences of overall mat structure or cellular orientation or content were detected between field-collected specimens and cultured specimens. Each filament consists of cylindrical cells, 40–45 µm in diameter and 150–300 µm long (Fig. 1a–c). Rhizoids are brown at maturity, usually unbranched and apparently non-gravitropic (Fig. 1c). Some rhizoids are quite short when gametophytes are grown on agar cultures, but many are as long as in the potting compost cultured specimens or field-collected specimens. Scanning electron micrographs of both living (ESEM) and frozen (LTSEM) specimens show that the surface of filament cells lacks cuticular elaboration and appears smooth, and that each cell narrows towards each end (Fig. 1d). The numerous chloroplasts are arranged in the peripheral cytoplasm of gametophyte filament cells. Fully differentiated photosynthetic cells contain a range from 137 to 323 chloroplasts with a mean of c. 186 (Fig. 2a, Table 1). Living chloroplasts, visualized in the CSLM, are spherical or ovoid with diameters of c. 8 µm, up to 10 µm (Fig. 2b,Table 1).

Figure 1.

Filamentous gametophytes of Trichomanes speciosum. (a) Gametophytic mat (field-collected specimens) to show colour and three-dimensional growth habit (SM, stereomicroscopy). (b) Gametophytes, showing the detailed nature of the mat of branching filaments (LTSEM, low temperature scanning electron microscopy). (c) Gametophytes growing on potting compost with branched filaments and rhizoids (e.g. arrow) (SM). (d) Detail of filaments (field-collected specimens) showing how each cell narrows towards each end (e.g. arrow) (ESEM, environmental scanning electron microscopy). Bars, 500 µm (a–c); 100 µm (d).

Figure 2.

Three-dimensional confocal laser scanning micrographs of photosynthetic gametophyte cells of Trichomanes speciosum. (a) A living filament showing numerous chloroplasts all round the periphery of the cylindrical cells. (b) As (a) showing the spherical or oval shape of gametophyte chloroplasts. Bar, 50 µm (a, b).

Table 1.  Chloroplast features: gametophyte filament and sporophytic (leaf) cells of T. speciosum. All means of gametophyte filament and sporophytic (leaf) cells are significantly different from each other (P < 0.01, Student’s t-test)
PlantsChloroplast no. per cell (n = 40)Chloroplast diameter (µm) (n = 100)Thylakoid no. per granum (n = 40)
RangeMean ± SERangeMean ± SERangeMean ± SE
  1. (SE = Standard error).

Gametophyte filaments137–323186 ± 6.75–108 ± 0.2 7–1511 ± 0.4
Sporophyte leaves 51–141101 ± 4.07–139 ± 0.110–2916 ± 0.6

Gemmae are found on all specimens and consist of 1–20 cells and are typically spindle-shaped but some are bar-shaped (Fig. 3a). Some gemmae have three or four arms; we refer to them as branched gemmae. Spindle-shaped gemmae have two end cells that are smaller than the other cells. They all grow to become perpendicular to the gemmifer. Some spindle-shaped gemmae produce secondary gemmae from the end cells while they are still attached to their gemmifers. Single gemmifers are more common features than paired gemmifers, which are usually found at the end of filaments (Fig. 3b). Gemmifers are short and blunt, with a narrower apex than base (Fig. 3c). The gemmifer cells are 25–40 µm in diameter and usually 40–60 µm long, although cells at the end of filaments occasionally produce gemmifers that are far longer. Gemma dehiscence generates scars on both the detached gemma and the apex of each gemmifer. Light microscopy shows that gemmifer scars and the attachment scars on gemmae are brown regions. Both scars are commonly domed in the centre with a circular pinched margin (Fig. 3c).

Figure 3.

Gemmifers, gemmae and gametangia of Trichomanes speciosum. (a) Filaments bearing typical spindle-shaped gemmae (arrowheads) and a bar-shaped gemma (arrow) (SM, stereomicroscopy). (b) Light micrograph showing a single gemmifer (arrow) bearing a spindle-shaped gemma and paired gemmifers (arrowheads) at the end of a filament (SM) (both samples were from agar cultures). (c) A single short and blunt gemmifer showing the typical domed apical scar. The small collapsed superficial filamentous structures are almost certainly fungal hyphae (field-collected specimens; LTSEM, low temperature scanning electron microscopy). (d) An archegoniophore bearing two young archegonia in the foreground; each has four rows of up to five neck cells. The old archegonia behind those seen here have collapsed (ESEM, environmental scanning electron microscopy). (e) Two wedge-shaped antheridia that have been produced by successive cells on a filament, consisting of two circular jacket cells, two circular cap cells and short single celled stalks (LTSEM). (Figure 3d,e specimens were collected from laboratory cultures that originated from a woodland site near Hebden Bridge, Yorkshire, UK, photographs courtesy of C. A. Raine). Bar, 200 µm (a, b); 50 µm (d); 20 µm (c, e).

Gametangia are rare (see Discussion section) but their production can coincide with production of gemmae. Archegonia are produced on filaments close to the substrate on an archegoniophore, which is a multicellular 3-D structure borne on a short broad filament. Archegoniophores bear between one and eight archegonia, each with four rows of up to five neck cells (Fig. 3d). Antheridia may be found on the same filament and are more frequent than archegonia (see Discussion section). They are sessile or short-stalked and are typically produced on filaments more superficial than those bearing archegonia. Antheridia have two circular jacket cells and two circular cap cells, one of which differentiates to form an operculum (Fig. 3e).

Ultrastructure of gametophytes

Gametophyte filament cells are highly vacuolate. They have numerous small vacuoles in the periphery and single large central vacuole (Fig. 4a,b). The cell walls have thick electron-opaque material that comprises a dense outer stratum and 2–3 layers of less dense inner stratum. The outer walls are covered with a cuticular layer (Fig. 4c). Numerous starch grains are present in the stroma of the chloroplasts (Fig. 5a); plastoglobuli occur singly and in clusters (Fig. 5b). Gametophyte chloroplasts have grana bearing stacks of up to 15 thylakoids with a mean of 11 (Fig. 5c, Table 1). Mitochondrial profiles were rarely seen.

Figure 4.

Transmission electron micrographs of photosynthetic gametophyte cells of Trichomanes speciosum. (a) Transverse section of peripheral cytoplasm of gametophyte cell showing the thick electron-opaque cell wall, chloroplasts (c) with starch deposits and vacuoles (V) containing unidentified electron-opaque material. (b) Longitudinal section of gametophyte cell showing numerous small vacuoles in the thin layer of cytoplasm around the periphery of the single large central vacuole. (c) Detail of the cell wall showing it to be covered with a thin layer of cuticle (arrow). Bar, 2 µm (a, b); 1 µm (c).

Figure 5.

Transmission electron micrographs of photosynthetic gametophyte cells of Trichomanes speciosum. (a) Detail of the cytoplasm in the vicinity of the nucleus (N), showing numerous chloroplasts but no mitochondria. G, granum; S, starch grain. (b) A portion of a chloroplast showing that plastoglobuli (PG) occur singly and clusters in the stroma. (c) Detail of a chloroplast showing typical grana with relatively few thylakoids, and a large area of membrane-free stroma. Bar, 1 µm (a, b); 0.2 µm (c).

The fine structure of gemma cell walls does not differ from that of gametophyte filament cells. The most distinctive feature of gemma chloroplasts was the larger and more numerous starch grains, which occupy most of the interior of each chloroplast (Fig. 6a), but these parameters were not quantified.

Figure 6.

Transmission electron micrographs of a gametophyte gemma cell (a) and rhizoid (b) of Trichomanes speciosum. (a) Longitudinal section showing chloroplasts with numerous large starch grains (S). Nucleus (N). (b) Transverse section showing thick multilayered electron-opaque cell wall, which is covered with a layer of diffusely electron-opaque material, plastids (P) with stacked internal membranes and the large central vacuole (V). Bar, 2 µm (a, b).

Transverse section of rhizoids showed that cell walls are several-layered and very thick; the plastids contain a small number of grana with relatively few thylakoids and no starch grains. Rhizoid cell walls are covered with a layer of diffusely electron-opaque material (Fig. 6b).

Sporophyte leaves

Sporophyte leaves are pinnately compound membranous leaves, somewhat blue–green in colour, and are slightly iridescent in appearance, especially when young. The lamina areas are only one cell layer deep, cells are rectangular in surface view, c. 58 µm wide, and c. 80 µm long. CSLM images show that living chloroplasts in lamina areas of sporophyte leaves are biconcave and disc-shaped, with diameters of c. 9 µm, up to 13 µm. Chloroplasts are found in the cell periphery and range in number from 51 to 141 with a mean of c. 101 (Fig. 7a, Table 1). The cell walls have 2–3 layers or more of electron-opaque material and are covered with a thin layer of cuticle very similar to that on the cell walls of gametophyte filaments (Fig. 7c). The chloroplasts broadly resemble those of gametophytes with plastoglobuli, many grana and large starch grains (see Fig. 7b), but each granum has up to 29 lamellae with a mean of 16 thylakoids per stack (Fig. 7d,Table 1).

Figure 7.

Sporophyte leaf cells of Trichomanes speciosum. (a) Three-dimensional confocal laser scanning micrograph in lamina area showing numerous living disc-shaped chloroplasts with diameters of up to 14 µm. (b) Transverse section of lamina showing the thick, electron-opaque cell wall, many chloroplasts (C) in the peripheral cytoplasm and vacuoles (V) containing electron-opaque material. (c) Detail of cell wall of young leaf showing regularly ordered layers of cellulose microfibrils in cell wall and it to be covered with a thin layer of cuticle (arrow). (d) A portion of a chloroplast showing a typical granum, with more numerous lamellae than found in gametophyte cells. Bar, 50 µm (a); 2 µm (b, c); 0.2 µm (d).


The gametophytes of T. speciosum are solely filamentous in form; species of Trichomanes in other subgenera have two growth forms: a filamentous, branching phase and a blade-like phase (e.g. T. holopterum, see Farrar & Wagner, 1968). The morphology revealed here via light and scanning electron microscopy expands upon previous knowledge of the gametophytes of T. speciosum (Raine, 1994; Rumsey, 1994) and is broadly in line with earlier descriptions of other species in this genus (Stokey, 1940). Although gemmae are generated reliably and regularly by all plants examined, gametophytes of T. speciosum rarely produce gametangia. Rumsey & Sheffield (1996) reported archegonia in less than 10% of gametophytes collected from the field, or after development in laboratory cultures. Antheridia are slightly more readily produced, being present in c. 25% of gametophytes collected from the field. The gametangia are similar in structure to those of other Trichomanes species (Stokey, 1948; Farrar & Wagner, 1968). Despite their role as vegetative propagules, gemmae differ little from filament cells. Starch grains appear to be more abundant in chloroplasts of gemmae, suggesting a role as a metabolic store.

Duckett et al. (1996) reported that rhizoids of gametophytes of the Hymenophyllaceae are very different from sporophytic root and rhizome hairs, in that rhizoids lack chloroplasts, wall banding, a terminal investiture of mucilage and rarely branch in contact with solid substrata. The rhizoids of T. speciosum do contain some chloroplasts, but are otherwise in line with this description. Although some rhizoids of this species are quite short when gametophytes are grown on agar, many are long and some are branched. They obviously differ from those of T. holopterum, which appear to be too short to function in anchorage or absorption in the manner usually attributed to rhizoids (Farrar & Wagner, 1968). The finding that they are not gravitropic is in line with life in their natural habitats, which include floors, walls and frequently roofs of caves. The unidentified electron-opaque material on the outside of rhizoids may be a polysaccharide (cf. rhizoids of T. holopterum, Farrar & Wagner, 1968). Determination of the nature of this substance is ongoing, but functions could include adhesion or exclusion of pathogens.

Whittier & Peterson (1995) reported a lipid layer covering the surface of Psilotum gametophytes, but these plants are subterranean. The typical fern gametophyte is surficial, photosynthetic and flat; it rarely has surface lipid. There are very few reports of a cuticle or lipid layer on the surface of such pteridophyte gametophytes (Sheffield & Farrar, 1988), but cell walls of T. speciosum gametophytes are covered with a thin layer of cuticle. This, in conjunction with the compact, dense nature of the cell wall material of gemmae, gametophyte filaments and sporophyte leaf cells presumably contributes to their ability to survive mild desiccation, despite their one-cell thick structure. The thin layer of cuticle on the cell walls may also prevent waterlogging, as noticed for mosses in situations where protonemata would suffer significantly depressed photosynthesis (Proctor, 1982; Duckett et al., 1998).

The blue iridescence of ferns, such as Danaea nodosa has been attributed to regularly ordered layers of cellulose microfibrils in adaxial epidermal cell walls (Graham et al., 1993). The 2–3 layers of electron-opaque material in cell walls of both gametophyte filaments and sporophyte leaves of T. speciosum resemble the epidermal cell walls of a blue leaf of Selaginella willdenowii (Lee, 1997). The regularly layered cell wall of some cells of T. speciosum might therefore account for the slightly iridescent appearance of some plants in some conditions. T. speciosum could never be described as strongly iridescent, however, and lacks the specialized chloroplasts with parallel lamellae which account for iridescence in T. elegans and some other plants (Graham et al., 1993; Gould & Lee, 1996). The functional significance of iridescence is not clear, but the strong association of leaf iridescence with extreme shade (Graham et al., 1993) makes its relative weakness in T. speciosum puzzling.

Confocal laser scanning microscopy showed clear differences between chloroplasts in gametophytic filament cells and those of sporophytes. However, chloroplast size and number in each cell of both gametophyte filaments and sporophyte leaves are broadly similar to those of other typical ferns. This finding is difficult to reconcile with accepted dogma concerning deep shade plants, in which few, large chloroplasts are reportedly the norm (Boardman, 1977). In the only other study of a deep shade fern, Nasrulhaq-Boyce & Duckett (1991) described each cell of the upper epidermis of Teratophyllum rotundifoliatum as having only c. 7 chloroplasts with diameter c. 18 µm. The habitats of Trichomanes speciosum receive even less light than the deep shade tropical habitats of Teratophyllum rotundifoliatum (Nasrulhaq-Boyce & Duckett, 1991; Johnson et al., 2000). Our expectation was therefore that Trichomanes speciosum would have still fewer, or yet larger, chloroplasts than Teratophyllum rotundifoliatum. As neither proved to be the case, we predicted that each chloroplast might have abundant thylakoids, because the chloroplasts of angiosperm, pteridophyte and bryophyte species, which are either naturally found in shaded conditions or which are grown in darkness, have been described as having grana comprising many thylakoids (Ballantine & Forde, 1970; Boardman, 1977; Duckett, 1986; Nasrulhaq-Boyce & Duckett, 1991; Sarafis, 1998). Plants with low chlorophyll a : b ratios typically have abundant photosystem II, and as this is located in grana, their chloroplasts usually have many grana per given volume of stroma. The grana in the chloroplasts of the T. speciosum gametophytes studied herein, which do have a low chlorophyll a : b ratio (Rumsey, 1994; Johnson et al., 2000), however, have relatively few thylakoids, and significantly fewer than those of sporophyte leaf chloroplasts. Chloroplasts of both gametophyte filaments and sporophyte leaves do not represent either of the extreme chloroplast forms reported for deep shade fern or angiosperm leaves. They have far fewer thylakoids than the leaf cell chloroplasts of Teratophyllum rotundifoliatum, which grows in less deep shade than Trichomanes speciosum. T. speciosum contain a mean of only 11 in gametophytes; sporophyte leaf chloroplasts have c. 16. Neither do T. speciosum chloroplasts in sporophyte leaves resemble choroplasts of T. elegans, which are iridoplasts in upper epidermal cells and typical extreme shade chloroplasts in other cells (Graham et al., 1993).

We conclude therefore that both the sporophyte and gametophyte of T. speciosum, have means to cope with extremely low light that do not involve modifications of chloroplast number or structure. Johnson et al. (2000) suggested that this species might make more efficient use of the light that is available, but this does not take the form of highly modified chlorophyll a : b ratios or, we now know, photosynthetic organelles. The explanation seems to lie largely with the abundance of chloroplasts, their peripheral position, and the optimization of cellular orientation with respect to light (Rumsey, 1994). The cylindrical growth form of gametophyte filaments does give a high surface-area : volume ratio and we could speculate that the rarity of mitochondrial profiles in gametophyte or sporophyte leaf sections implies low respiration rates and therefore low compensation points. We do know that T. speciosum gametophytes grow extremely slowly, and cannot acclimatize to or grow at a PAR above 60 µmol m−2 s−1, and that gemmae produce only c. 1 gametophytic filament cell per original gemma cell when cultured for 2 months at 20°C and a PAR of 5 µmol m−2 s−1 (Makgomol & Sheffield, unpublished). The success of T. speciosum in extreme low light habitats is therefore thought to be due largely to the inability of other species to cope, and to an extremely low metabolic rate.


We are very grateful to Catherine Raine for providing Fig. 3d,e, to Christopher Gilpin, John Hutton and Samantha Newby in the EM Unit for technical guidance on electron microscopy, to Grenham Ireland and Tony Wade for help with confocal microscopy, and to Abdul Sattar for guidance on Image Analysis. Useful discussion with Jeff Duckett and Don Farrar and the valuable comments of Giles Johnson and Fred Rumsey on an earlier draft of this manuscript are gratefully acknowledged. Thanks also go to the staff in the Photography Unit, Ian Miller, Les Lockey and Tony Bentley. K.M. acknowledges financial support from the Royal Thai Government.