Cryptospores and cryptophytes reveal hidden diversity in early land floras

Authors


Abstract

Summary

Cryptospores, recovered from Ordovician through Devonian rocks, differ from trilete spores in possessing distinctive configurations (i.e. hilate monads, dyads, and permanent tetrads). Their affinities are contentious, but knowledge of their relationships is essential to understanding the nature of the earliest land flora. This review brings together evidence about the source plants, mostly obtained from spores extracted from minute, fragmented, yet exceptionally anatomically preserved fossils. We coin the term ‘cryptophytes’ for plants that produced the cryptospores and show them to have been simple terrestrial organisms of short stature (i.e. millimetres high). Two lineages are currently recognized. Partitatheca shows a combination of characters (e.g. spo-rophyte bifurcation, stomata, and dyads) unknown in plants today. Lenticulatheca encompasses discoidal sporangia containing monads formed from dyads with ultrastructure closer to that of higher plants, as exemplified by Cooksonia. Other emerging groupings are less well characterized, and their precise affinities to living clades remain unclear. Some may be stem group embryophytes or tracheophytes. Others are more closely related to the bryophytes, but they are not bryophytes as defined by extant representatives. Cryptophytes encompass a pool of diversity from which modern bryophytes and vascular plants emerged, but were competitively replaced by early tracheophytes. Sporogenesis always produced either dyads or tetrads, indicating strict genetic control. The long-held consensus that tetrads were the archetypal condition in land plants is challenged.

1 Introduction

Over the past 50 yr, the discovery and analysis of fossil spore assemblages dispersed in rocks of lower Palaeozoic age (470–419.2 million yr) have revolutionized our understanding of the colonization of the land by plants. Although there were some pre-1960s pioneers (Lang, 1925; Thomson, 1940; Eisenack, 1944; Naumova, 1953; Radforth & McGregor, 1954; Chibrikova, 1959), the application of palynological methods developed from the 1960s such that it opened up a new window on to early floras (McGregor, 1960, 1961; Richardson, 1960, 1962, 1965, 1967; Chaloner, 1963; Streel, 1964; Allen, 1965; Richardson & Lister, 1969; Richardson & Ioannides, 1973). Before this, research focused on macroscopic remains (i.e. plant leaves, stems, wood, seeds, etc.) from the Devonian period, which yielded a record of early vascular plants that was both diverse and rich in detail (Kräusel & Weyland, 1935; Lang, 1937; Høeg, 1942; Chaloner, 1970). This megafossil record formed the foundation of our understanding of the nature of early land floras, and it defined the temporal framework for investigating evolutionary trends and their broader effects on Earth systems (Banks, 1968, 1970; Chaloner, 1970; Niklas et al., 1985). Palynology extended significantly the stratigraphical and geographical scope of enquiry, and it brought to light a hidden diversity of enigmatic plants. Key early studies showed that plant life on land extended much further back stratigraphically than did the macrofossil remains (reviewed by Gray, 1985; Richardson, 1985; Richardson & McGregor, 1986), but precisely how far is debated in terms of both evidence acceptable to palynologists (Richardson, 1988, 1992; Steemans, 2000; Wellman & Gray, 2000; Taylor & Strother, 2008, 2009; Steemans et al., 2010; Wellman, 2010; Strother et al., 2011) and the fit to calibrated molecular phylogenies (Clarke et al., 2011; Kenrick, 2011; Magallón et al., 2013; Rota-Stabelli et al., 2013). Also revealed was an unanticipated diversity of palynomorphs, some of which are uncontroversially attributable to land plants (e.g. trilete monads, also known as trilete spores; Steemans et al., 2009), whereas the affinities of others (e.g. obligate tetrads, dyads, and alete monads, collectively known as cryptospores) are less clear (Rubinstein et al., 2010; Steemans et al., 2010; Wellman, 2010; Strother et al., 2011) (Fig. 1). Recently, there has also been a shift in thinking about the geographical region of origin of land plants from Laurussia to Gondwana, based on evidence from palynomorphs (Steemans et al., 2010). These discoveries raise two related questions. First, why does the stratigraphical record of palynomorphs precede to such a great extent the record of macroscopic plant remains? Secondly, what sort of plants produced the cryptospores, and how are they related to living species? The answer to the first question lies partly in the nature of the rock record during the early part of the Palaeozoic Era, where significant changes in the proportions of terrestrial and marine rocks are thought to effectively open and close taphonomic windows on to macrofossil and palynomorph evidence (Kenrick et al., 2012), and partly in changes in the preservation potential of the plants themselves. Progress in addressing the second question has relied on the comparative study of spore-wall ultrastructure. The application of transmission electron microscopy to both living and fossil palynomorphs has been critical to identifying features linking some cryptospores to basal living groups of embryophytes (Taylor, 1995a,b, 1996, 1997). At the same time, the discovery of minute, fragmentary, but remarkably well-preserved fossil plants at several sites in the Welsh Basin has furnished direct evidence of the plants that produced the cryptospores (Edwards et al., 1995a, 1996, 1999; Wellman et al., 1998a,b; Wellman, 1999). We review the nature of this fossil evidence, and we explore its implications for our understanding of early land floras, the organisms they contained and their relations to living plants.

Figure 1.

Relationships among major groups of land plants showing the hypothesized broad range of clades to which cryptophytes (cryptospore-producing plants) might belong (shaded oval). As currently envisaged, cryptophytes are a grade of organization potentially encompassing plants spanning the embryophyte to tracheophyte stem-groups as well as stem-group hornworts, mosses and liverworts. Some distinctive groupings are emerging, and here we recognize two (Partitatheca and Lenticulatheca), which we suggest occupy a comparatively derived position within polysporangiophytes. Relationships of living groups depict one widely supported hypothesis based on: charophytes (Timme et al., 2012; Bowman, 2013); liverworts (Heinrichs et al., 2007); mosses (Cox et al., 2010); hornworts (Shaw & Renzaglia, 2004; Villarreal et al., 2010); embryophytes (Qiu et al., 2007; Magallón et al., 2013); fossils (Kenrick & Crane, 1997). The broken lines joining Partitatheca and Lenticulatheca indicate that their phylogenetic positions are hypothesized. Polysporangiophytes are a clade containing living and extinct groups with branched sporophytes (Kenrick & Crane, 1997). Stomatophytes indicate a clade in which stomata are a synapomorphy but are inferred to have been subsequently lost in some extant taxa (Kenrick & Crane, 1997; Ligrone et al., 2012). †, Extinct taxa.

2 History, characters and disparity in cryptospores

Dispersed palynomorphs can be linked to their source plants through comparative morphology with varying degrees of confidence. One of the common characters of early vascular plant fossils is the trilete monad, which is a spore with a sporopollenin-rich wall that formed in a sporangium as a product of meiosis. This spore type is commonly found in situ in megafossils of the Devonian period (Allen, 1980; Gensel, 1980), but in its dispersed form it is present in Lower Silurian palynomorph assemblages (Gray, 1985), significantly predating the oldest unequivocal macrofossil record of vascular plants. This early record of trilete spores, which now extends into the Upper Ordovician (mid-Katian: Caradoc-Ashgill) (Steemans et al., 2009; Saudi Arabia), is widely regarded as a benchmark for the origin of vascular plants, but this interpretation is clouded by the fact that trilete monads are also known to be characteristic of some living bryophytes (Table 1). It is conceivable, therefore, that the early trilete monads might represent a broader grouping within basal embryophytes.

Table 1. Distribution of selected sporophyte and spore features in major basal land plant clades compared with the early cryptophyte fossil Partitatheca
 LiverwortHornwortMossTracheophyte Partitatheca
  1. ✗, absent; ✓, present; (✓), present in some; ✓ (few), rarely present.

Branching sporophyte
Valvate sporangia✓ (few)(✓)(✓)
Stomata
Lamellae in spore wall
Permanent tetrads(✓)
Permanent dyads
Trilete spores(✓)(✓)✓ (few)

As the early Palaeozoic record of trilete spores came under scrutiny, it became apparent that other distinctive palynomorphs formed a significant component of nonmarine assemblages (Gray & Boucot, 1971, 1972; Richardson & Ioannides, 1973; Gray et al., 1974; Pratt et al., 1978; Strother & Traverse, 1979; Miller & Eames, 1982). These took the form of alete monads and other types that were dispersed as tetrads and dyads, some of which were enveloped in a second wall layer. Although the affinities of these palynomorphs could not easily be established, these discoveries heralded a breakthrough towards the understanding of a new facet of early land vegetation. Permanent tetrahedral tetrads were first recognized by Gray & Boucot (1971) from the lower and middle Llandovery (basal Silurian) of western New York State. They initially hypothesized that both permanent tetrads and trilete monads were the products of vascular plants, but they also recognized that a bryophytic origin could not be precluded. In other words, some of the new enigmatic palynomorphs were probably derived from land plants and possibly from bryophyte-like organisms, hinting at a greater diversity of plants (Fig. 1).

Recognition of these diverse palynomorphs as a separate group began to be formalized when Strother & Traverse (1979) provided generic names and diagnoses for the permanent tetrads (including Tetrahedraletes) and dyads (Dyadospora), based on an assemblage from the Llandovery and Wenlock of Pennsylvania. Building on this, Richardson et al. (1984) erected the informal category (Anteturma) Cryptosporites, also known as cryptospores, with the following definition: non-marine sporomorphs, with no visible haptotypic features such as contact areas or tetrad marks. Single grains or monads, “permanent” dyads and tetrads are included (Richardson et al., 1984, p. 116). The prefix ‘crypto’ referred to their unfamiliar configurations and absence of knowledge regarding their producers and affinity. In 1988, he emended his definition to include the separated products of dyads (i.e. hilate monads: monads with a large circular contact area) (Richardson, 1988). Other usages range from the very general, for example, that of Strother & Beck (2000), who regarded cryptospores as a class of organic walled microfossils produced by terrestrial plants (a subaerial equivalent of acritarchs), to formal definitions that seek to eliminate some ambiguities and possible confusions with acritarchs and which also introduce specific notions of affinity with embryophytes (e.g. Steemans, 2000). In this review we adhere to the definition of Richardson (1988), because it is based on straightforward morphological criteria that can be used to recognize a cryptospore and it does not allude to affinities with specific plant groups.

We concentrate on two cryptospore groups based on their configurations, as either tetrahedral tetrads or dyads. Tetrads are always permanent (also known as obligate), meaning that they are dispersed as polyads, whereas dyads are dispersed either as polyads or as monads. In contrast to trilete spores, such monads are alete and possess an extensive proximal contact feature, termed a hilum. The entire permanent polyad of both tetrads and dyads can be surrounded by an envelope, which may be loose or so closely adherent that it is difficult to distinguish. The surface of the envelope may be sculptured, while that of the enclosed polyad is usually smooth (e.g. Velatitetras; Burgess, 1991). Naked polyads (i.e. those lacking an envelope) may have smooth or sculptured surfaces, and the latter is more common in younger Upper Silurian to Lower Devonian rocks.

The sculpture on cryptospores provides an additional suite of characters that enable species to be distinguished. These cryptospore species are regarded as a proxy for species diversity (disparity) in the source vegetation. Detection of disparity through the analysis of palynomorphs with very simple morphology, as exemplified by the smooth (naked) polyads, presents far greater challenges, although recent studies involving thin sections and transmission electron microscopy (TEM) indicate far greater variation in wall characters in dispersed and in situ permanent cryptospores than previously recorded by scanning electron microscopy (SEM) and light microscopy (Taylor, 1995b, 2002; Edwards et al., 1999, 2012b) (Fig. 2).

Figure 2.

Schematic representation of spore wall ultrastructure (not to scale). (a) Taylor (1997). Type I, OD-10-1; dispersed. Upper Ordovician. Type II, CV-3-5; dispersed. Lower Silurian. (b) Taylor (1995a, 1996, 2000). CV-3-1; dispersed. Lower Silurian. (c) (i–ii) Wellman et al.(1998b). Upper Silurian. (i) NMW96.30G.2; as monads in discoidal sporangium. (ii) NMW96.30G.3; as dyads within elongate sporangium. (iii–v) Wellman et al. (1998a). Lower Devonian. (iii) NMW96.11G.6; as dyads within Culullitheca richardsonii. (iv) NMW97.42G.2; as dyads on sporangial cuticle. (v) NMW97.42G.1; as dyads on sporangial cuticle. (d) (i–ii) Edwards et al. (2012a). Lower Devonian. (i) V.68201; within Partitatheca splendida. (ii) V.68184(1); within Partitatheca cf. splendida. (iii) Habgood (2000). NMW99.19G.1; within base of sporangium attached to bifurcating axis. Lower Devonian. (e) Taylor (1996, 2000). OD-2-2; dispersed. Upper Ordovician. (f) (i) Taylor (1995b, 2002). OD1-1; dispersed. Upper Ordovician. (ii, iii) Taylor (2002). Dispersed, Lower Silurian. #1, 75-5c; #2, CV1-17; #3, CV4-3. (iv, v) Edwards et al. (1999); Lower Devonian. (iv) NMW98.23G.5; within irregular spore mass. (v) NMW98.23G.4; within irregular spore mass. (g) (i) Edwards et al. (1999). NMW96.11G.3; within basal sporangia terminating bifurcating axis. Lower Devonian. (ii) Habgood (2000). NMW99.19G.2, within elongate sporangium. Lower Devonian. (iii) Edwards et al. (2012b). V.68196(2); within irregular spore mass. Lower Devonian. (iv) Taylor (2000). Dispersed, Lower Silurian. (v) Edwards et al. (1999). NMW96.11G.4; within discoidal sporangium. Upper Silurian. (h) Edwards et al. (2012b). MNW2012.17G.1; within discoidal sporangium. Lower Devonian. (i) (i, ii) Edwards et al. (2012b). Lower Devonian. (i) NMW2012.7G.3; within valvate sporangium. (ii) NMW99.11G.2; within ? valvate sporangium. (j) Johnson & Taylor (2005); dispersed. Upper Silurian. (k, l) Morris et al. (2011); Lower Devonian. (k) V.68188(1); within discoidal sporangium. (l) V.68196(1); within discoidal sporangium. L, lamina; env, envelope; sp, spongy; gr, granules; v, verrucae.

In initial attempts to increase the number of distinguishing characters in cryptospores, Wellman & Richardson (1993) introduced the terms ‘fused’ and ‘unfused’ to differentiate between tetrads and dyads that had superficial ‘lines of attachment’/sutures between tetrad units (unfused) and those that did not (fused). Thus, for example, the laevigate genus Tetrahedraletes Strother & Traverse (1979) emend. Wellman & Richardson (1993) is a permanent unfused tetrad, while Cheilotetras Wellman & Richardson (1993) is a laevigate, permanent fused tetrad. However, the ultrastructural basis underpinning this distinction is particularly difficult to elucidate in the absence of thin sections, as later appreciated by Richardson (1996c) and Wellman (1996). This introduces a further taphonomic problem because diagenesis and charring can result in homogenization of adjacent walls. A further possibility is that the absence of sutures at the junctions between units may reflect the presence of a thin adherent envelope.

By far the most important advance in elucidating the affinities of Ordovician and Silurian dyads was made by Taylor in his TEM work on the wall ultrastructure of Dyadospora murusdensa Strother & Traverse (1979) emend. Burgess & Richardson (1991) (Taylor, 1995a, 1996; figs 10–13) and Dyadospora murusattenuata Strother & Traverse (1979) emend. Burgess & Richardson (1991) (Taylor, 1997; plates I, II). Variation in ultrastructure was observed within this genus. Of particular note were the spore walls of D. murusdensa and Dmurusattenuata Type I, their inner layers possessing a laminate construction (Fig. 2a,b). Laminae are distinguished from ‘white line centred lamellae’, the latter being fine structures (60–130 Å) that appear early in spore wall development in extant plants. The latter are subsequently covered on both surfaces by sporopollenin layers and fuse together to form tripartite lamellae (Taylor, 2009). Often, but not always, the tripartite structures are occluded by further sporopollenin deposition, forming laminae (Taylor, 2009). Well-preserved stacked laminae were also observed in sections of in situ permanent naked dyads from the oldest known spore masses (Late Ordovician; Saudi Arabia) (Wellman et al., 2003, fig. 2). By contrast, dispersed Ordovician and Silurian tetrads have homogeneous (Fig. 2f ii) or irregularly or weakly striated walls (Fig. 2f i, iii) (Taylor, 2002), except for a mid-Ordovician example in which a highly convoluted single lamina is internal to a thick homogeneous wall (Taylor, 2009). The wall ultrastructure of the dyads and the tetrahedral configuration of the tetrads have been used to posit an hepatic affinity (Gray, 1985). We will further explore this hypothesis based on an analysis of cryptospores found in situ in cryptophyte fossils.

In situ cryptospores

3.1 Introduction

This review is based on very small, almost all charcoalified, fossils (termed ‘mesofossils’) comprising the fertile tips of small plants. They come from two localities in Shropshire, UK. The older assemblage is found in the Downton Castle Formation exposed in a small cliff at the junction of Ludford Lane and Leominster Road, Ludlow, in strata dated as Přídolí (Late Silurian; c. 419 Ma) on the presence of dispersed spores of the tripapillatusspicula Sporomorph Assemblage Biozone (Richardson & McGregor, 1986). The younger occurs in a stream section to the north side of Brown Clee Hill, belonging to the lower part of the Ditton Group, deposited c. 4 million yr later (Edwards & Richardson, 2004). Spores indicate that the horizon belongs to the middle subzone of the micrornatusnewportensis Sporomorph Assemblage Biozone, indicating an early Lochkovian (Lower Devonian) age (Richardson & McGregor, 1986). The quality and extent of preservation (subtending axes are very rare) are such that the diagnoses and characterization of each group described here are based on a concatenation of evidence. However, there is often a consistent association of cryptospore type with sporangial morphology, as summarized below in A–D, allowing the distinction of lineages and insights into the diversity and affinities in early land vegetation. The in situ spores can then be related to those in the dispersed spore assemblage from the locality, sometimes even to varietal level, although it is now evident that species in the same dispersed genus may belong to different lineages. For example, the dispersed spore genus Cymbohilates Richardson (1996a) encompasses two groups; hilate monads (e.g. Cymbohilates allenii var. magnus and Cymbohilates variabilis), the separation products of dyads, which retain an intact hilum, are found within discoidal sporangia (Morris et al., 2011), and alete monads (namely the Cymbohilates horridus complex and Cymbohilates cymosus), produced by the tearing apart of permanent dyads and sometimes possessing partial fragmentary ‘hilate apertures’, are found within valvate sporangia (Edwards et al., 2012a).

3.2 Descriptions of major groups of sporangia containing cryptospores

  1. APermanent polyads in valvate sporangia
    • i.Quadrivalvate sporangia (Figs , )
      Figure 3.

      Quadrivalvate sporangia. All specimens are Early Devonian in age. All scanning electron microscopy (SEM) unless otherwise stated. (a–e) Partitatheca splendida Edwards et al. (2012a). V.68184(1) (holotype). (a) Whole specimen. (b) Stoma at sporangium base. (c, d) In situ spores; Cymbohilates horridus var. splendidus Richardson in Edwards et al. (2012a). (c) Dyad, with spinose elements. (e) Transmission electron microscopy (TEM) section through C. horridus var. splendidus. (f–h) NMW99.19G.1, Habgood, 2000. (f) Bifurcating axis terminating with basal part of sporangium. (g) In situ spores; Cymbohilates cf. horridus. (h) TEM section through C. cf. horridus. (i–k) Partitatheca cymosa Edwards et al. (2012a). V.68212(1) (holotype). (i) Whole specimen. (j) In situ spores; Cymbohilates cymosus Richardson (1996a). (k) Dyad with spinose elements arranged in star-shaped rosettes. (l–o) cf. Partitatheca sp. Morris et al. (2012a). V.68264a(1) and V.68264b(1). (l) Whole specimen. (m) In situ spores; Chelinohilates cf. erraticus (Richardson, 1996a). (n) Fractured section through dyad, with narrow internal wall. (o) Light microscopy of section through dyad, with narrow internal wall. (p–r) NMW97.42G.1, Wellman et al. (1998a). (p) Whole specimen (single valve intact). (q) In situ spores; Dyadospora murusdensa Strother & Traverse emend. Burgess & Richardson (1991). (r) TEM section of Dmurusdensa. Bars: (a, i, l, p) 200 μm; (b, g, j, m–o, q) 20 μm; (c) 10 μm; (d) 50 μm; (e) 1 μm; (f) 500 μm; (h) 2 μm; (k) 5 μm; (r) 250 nm. Panels (a–e) and (i–k) were first published in Edwards et al. (2012a) and are reproduced here with kind permission of Wiley. Panels (f–h) were first published in Habgood (2000) and are reproduced here with kind permission of Wiley. Panels (l–o) were first published in Morris et al. (2012a) and are reproduced here with kind permission of Wiley. Panels (p–r) were first published in Wellman et al. (1998a) and are reproduced here with kind permission of Wiley.

    • ii.Bivalved elongate sporangia (Fig. )
      Figure 4.

      Valvate sporangia with permanent tetrads and dyads. All specimens are Early Devonian in age. All scanning electron microscopy (SEM) unless otherwise stated. (a–d) NMW2012.17G.3, Edwards et al. (2012b). (a) Quadrivalvate sporangium. (b) In situ spores; Cheilotetras sp. (c) Continuous spore wall across junctions between tetrad members. (d) Transmission electron microscopy (TEM) section through homogeneous spore wall, with wedge-shaped ‘plug’ (arrow a) and a thin encompassing layer (arrow b). (e–g) NMW99.11G.2, Edwards et al. (2012b). (e) Elongate bivalved sporangium. (f) In situ spore; Cheilotetras sp. (g) TEM section through spore wall with laminate structure. (h–j) NMW99.19G.2, Habgood (2000). (h) Elongate bivalved sporangium. (i) In situ spores; Velatitetras sp. (j) TEM section through spore; thick envelope with layer of voids; spore wall homogeneous with outward projections. (k–m) Culullitheca richardsonii Wellman et al. (1998a). NMW.96.11G.6. (k) Whole specimen. (l) In situ spore; Dyadospora murusdensa. (m) TEM section through homogeneous spore wall. (n–p) Fusiformitheca fanningiae Wellman et al. (1998a); Xue & Wang (2011). NMW97.42G.4. (n) Whole specimen. (o) In situ spore; enveloped, laevigate permanent dyad. (p) TEM section through homogeneous spore wall. Bars: (a, e) 200 μm; (b, f, i, k, l, n, o) 20 μm; (c) 5 μm; (d, j, m, p) 1 μm; (g) 3 μm; (h) 500 μm. Panels (a–g) were first published in Edwards et al. (2012b) and are reproduced here with kind permission of Elsevier. Panels (h–j) were first published in Habgood (2000) and are reproduced here with kind permission of Wiley. Panel (k) was first published in Edwards (1997) and is reproduced here with kind permission of Springer Science+Business Media. Panels (l–p) were first published in Wellman et al. (1998a) and are reproduced here with kind permission of Wiley.

  2. BDyads and tetrads in discoidal sporangia
    • i.Hilate monads and separating dyads in discoidal sporangia (Fig. )
      Figure 5.

      Lenticulatheca and Paracooksonia with bi-layered monads. (a–c) Late Silurian in age. (d–s) Early Devonian in age. All scanning electron microscopy (SEM) unless otherwise stated. (a–c) NMW96.30G.2, Wellman et al. (1998b). (a) Whole specimen. (b) In situ spores; Laevolancis divellomedia Type A. (c) Transmission electron microscopy (TEM) section through L. divellomedia Type A. (d–o) Lenticulatheca Morris et al. (2011). (d–f, h, i) Lenticulatheca magna. V.68180(1) (holotype). (g, j, k) Lenticulatheca mesodeca. V.68181(1) (holotype). (l, m) Lenticulatheca variabilis. V.68185. (n, o) Lenticulatheca allenii. V.68187(1). (d–f) L. magna. (d) Distal surface. (e) Proximal surface. (f) Specimen margin; spores occurring between two noncellular layers. (g) TEM section through cuticle of L. mesodeca. (h) Spore within L. magna; Cymbohilates allenii var. magnus Richardson (1996a). (i) TEM section through Callenii var. magnus. (j) Spore within Lmesodeca; Cymbohilates mesodecus Richardson in Morris et al. (2011). (k) TEM section through C. mesodecus. (l) Spores within L. variabilis; Cymbohilates variabilis Richardson (1996a). (m) TEM section through Cvariabilis. (n) Spore within L. allenii; Cymbohilates allenii var. allenii Richardson (1996a). (o) TEM section through Callenii var. allenii. (p –s) Paracooksonia apiculispora Morris et al. (2011). (p–r) V.68189(1) (holotype). (p) Whole specimen. (q–r) In situ spores; Streelispora newportensis Richardson & Lister (1969). (q) Proximal surface. (r) Distal surface. (s) TEM section through S. newportensis. V.68192(1). Bars: (a, d, e, p) 200 μm; (b, f) 20 μm; (h, l, n, q) 10 μm; (j, r) 5 μm; (k) 3 μm; (c, g, i, m, s) 1 μm; (o) 2 μm. Panels (a) and (c) were first published in Wellman et al. (1998b) and are reproduced here with kind permission of The Royal Society. Panel (b) was first published in Edwards et al. (1998) and is reproduced here with kind permission of Maney Publishing. Panels (d–s) were first published in Morris et al. (2011) and are reproduced here with kind permission of Elsevier.

    • ii.Permanent sculptured tetrads in discoidal sporangia (Fig. )
      Figure 6.

      Sporangial fragments with permanent tetrads. All specimens are Early Devonian in age. All scanning electron microscopy (SEM) unless otherwise stated. (a–d) NMW2012.17G.2 Edwards et al. (2012b). Discoidal sporangium containing permanent sculptured tetrads. (a) Whole specimen. (b) In situ spores; Acontotetras sp. (c) Rare tetrad splitting apart (arrow), sculptured with microverrucae and vermiculate coni. (d) Light microscopy of a thin section through tetrads, one example showing three members of a tetrad, with thick distal walls (D) and thin, unfused proximal walls (P). (e–l) NMW2012.17G.1 Edwards et al. (2012b). Discoidal sporangium containing permanent sculptured tetrads. (e) Whole specimen, with stomata (arrow). (f) Cross-section through the sporangial wall composed of rectangular cells. (g–j) In situ spores; Acontotetras inconspicuis Richardson (1996a). (g) Tetrad with major ridges and invaginations. (h) Ridge of tetrad with slight furrow (arrow). (i) Tetrad with ridges, sculptured with micrograna to fused biform elements. (j) Variation in sculptural elements between adjacent tetrads. (k) Transmission electron microscopy (TEM) through a tetrad, with thick distal walls and thin proximal walls. (l) Light microscopy of a thin section through tetrads, with thick distal walls and a thin envelope surrounding the entire tetrad, with sinuous lines (arrows a and b) representing the junction between tetrad members. Proximal walls are thin and extend into the lumen (arrow c). (m–r) NMW2012.17G.5 Edwards et al. (2012b). Sporangial fragment containing alete monads. (m) Whole specimen. (n) Sporangial wall with elongate cells. (o) Alete monad with major ridges and invaginations, sculptured with closely spaced coni, microconi and micrograna. (p) Ridge of tetrad with slight furrow. (q) Light microscopy of a thin section through two alete monads, fused together by sculptural elements (arrow). (r) Light microscopy of a thin section through two alete monads, fused together by sculptural elements, with an additional thin layer occurring between them (arrow). Bars: (a, e, n) 200 μm; (b, f, g, o, p) 20 μm; (c, d, h–j, l, q, r) 10 μm; (k) 5 μm; (m) 500 μm. Panels (a–r) were first published in Edwards et al. (2012b) and are reproduced here with kind permission of Elsevier.

  3. CPermanent polyads in sporangia with unknown dehiscence
    • i.Laevigate permanent dyads (Fig. )
    • ii.Permanent tetrads in bifurcating specimens (Fig. )
      Figure 7.

      Fragmentary sporangia containing permanent tetrads. (j–l) Late Silurian in age. All other specimens are Early Devonian in age. All scanning electron microscopy (SEM) unless otherwise stated. (a–c) Grisellatheca salopensis Edwards et al. (1999). NMW94.76G.1. (a). Whole specimen. (b) In situ spore; Cheilotetras sp. (c) Magnification of junctions between tetrad members. (d–f) NMW98.23G.2. Edwards et al. (1999). (d) Bifurcating axis terminated by cup-shaped depression, containing spores. (e) In situ spores: Tetrahedraletes medinensis. (f) Transmission electron microscopy (TEM) section of in situ spores; note separating outer layer. (g–i) NMW96.11G.3. Edwards et al. (1999). (g) Bifurcating axis terminated by the bases of two sporangia. (h) In situ spore: Velatitetras sp. (i) TEM section through in situ spores; note the tri-layered wall, with a central zone of globular material. (j–l) NMW96.11G.4. Edwards et al. (1999). (j) Discoidal spore mass covered with amorphous layer. (k) In situ spore: Velatitetras cf. anatoliensis. (l) TEM section through in situ spores; note tightly adherent envelope and faint striations in the inner wall. (m–p) V.68196(2). Edwards et al. (2012b). (m) Irregular shaped spore mass covered with cuticular layer. (n) In situ spore: Velatitetras sp. (o) Magnification of tetrad surface, enclosed within envelope with superficial sinuous folding. (p) Light microscopy of a thin section through a tetrad, illustrating a thick ‘envelope’ with inward projections separate from the internal spore. Bars: (a, m) 200 μm; (b, o) 5 μm; (c) 2 μm; (d, j) 500 μm; (e, j) 50 μm; (f) 500 nm; (g) 100 μm; (h, p) 10 μm; (i, l) 1 μm; (k, n) 20 μm. Panels (a) and (b) were first published in Edwards et al. (1995b) and are reproduced here with kind permission of Nature Publishing Group. Panels (c–f), (i), (k) and (l) were first published in Edwards et al. (1999) and are reproduced here with kind permission of Wiley. Panels (g), (h) and (j) were first published in Edwards (1997) and are reproduced here with kind permission of Springer Science+Business Media. Panels (m–p) were first published in Edwards et al. (2012b) and are reproduced here with kind permission of Elsevier.

    • iii.Permanent enveloped tetrads in spore masses with sporangial coverings (Fig. )
  4. DAlete monads (Fig. )

A. (i) Permanent polyads in quadrivalvate sporangia (Figs 3, 4)

Distinguishing characters of this group include:

  • sporangial dehiscence into four valves along entire length;
  • sporangial wall comprising at least one layer of large cells;
  • permanent laevigate and sculptured dyads and tetrads;
  • exospore with complex ultrastructure*;
  • occasional stomata on sporangia and subtending axes*;
  • bifurcating axes*.

*Rare occurrences.

The genus Partitatheca Edwards et al. (2012a) was initially erected for quadrivalvate sporangia (Figs 3a,i,l, 4a) with longitudinal files of large cells in their walls and permanent dyads with spinous sculpture (Fig. 3c,d,j,k). We would now extend the genus circumscription to include those containing dyads with murornate (Fig. 3m,n) and laevigate walls (Fig. 3q), as well as permanent tetrads (Fig. 4b,c).

Considering those containing sculptured permanent dyads, four species were initially distinguished from Lower Devonian specimens: Partitatheca splendida, Partitatheca horrida, Partitatheca densa and Partitatheca cymosa (Edwards et al., 2012a), their names reflecting those of spinose varieties of the dispersed spore Cymbohilates complex (Richardson, 1996a) (Fig. 3c,d,j,k). The sporangia of each species differ in overall morphology (Fig. 3a,i), shape of individual valves, presence or absence of stomata (Fig. 3b) and, in one case, possible persistent attachment at the apex (Edwards et al., 2012a, fig. 8A). In one case, an incomplete basal portion of a sporangium containing C. horridus Richardson (1996a) terminates a branching stem (Fig. 3f,g) with occasional stomata (Habgood, 2000).

Most of the in situ spore species have typical spines, variation deriving from the density, degree of fusion and clustering of the spines (Fig. 3c,g). In Ccymosus Richardson (1996a) they are arranged in bizarre star-shaped rosettes of basally fused spines (Fig. 3j,k). Cymbohilates horridus var. splendidus Richardson in Edwards et al. (2012a) occurs within the type species Psplendida and possesses pointed spines that cluster into pairs or small groups or are irregularly distributed (Fig. 3c).

Cymbohilates horridus var. splendidus shows the most complex ultrastructure recorded to date by TEM in Devonian cryptospores (Fig. 3e). Each spore of the dyad is completely surrounded by a relatively electron-lucent homogeneous layer, with darker flecks (possibly representing lamellae) in the mid region, showing complete fusion at the junction marked by a continuous dark line where the proximal surfaces meet (Fig. 2d i). The two fused spores are further enclosed in a complexly layered wall composed of at least five layers. Perhaps the most significant of these is a <40-nm zone of alternating dark and light lines (Figs 2d i, 3e), interpreted as laminae (sensu Taylor, 2009), that is sandwiched between an electron-dense layer that extends into the sculpture and a further narrower internal layer. A sporadically preserved, tightly adhering, outermost electron-lucent layer is also recorded. Spores assigned to Chorridus sensu lato within the bases of sporangia attached to a bifurcating axis (Fig. 3f,g) have a similar overall construction, with a tri-layered wall, but have a further well-developed separating external layer (Fig. 3h) (Habgood, 2000). Along with other specimens of Cymbohilates within Partitatheca, these spores lack a laminate construction.

The disparity of in situ spores within this sporangium type was extended with the discovery of a group of valvate sporangia from the same locality, assigned to Partitatheca sp. and cf. Partitatheca sp. (Fig. 3l), that contain permanent dyads with murornate sculpture (Fig. 3m,n) (Morris et al., 2012a). Such spores are closest to the dispersed spore taxon Chelinohilates erraticus Richardson (1996a). Fortuitous fractures and thin sections indicate a similar overall construction to Chorridus var. splendidus, the two units fused at their proximal surfaces (Fig. 3n), and completely encompassed by an outer, highly folded wall (Morris et al., 2012a).

There is only one example of this group with sporangia that contain permanent ‘unfused’ laevigate dyads. It is a disintegrated specimen (NMW97.42G.1) with one intact valve remaining of a presumed quadrivalvate sporangium (Wellman et al., 1998a) (Fig. 3p). The dyads belong to D. murusdensa (Wellman et al., 1998a). In section the distal spore walls are bi-layered. The inner layer of each dyad member has a homogeneous structure, apart from a line of voids running close to and parallel with the lumen (Fig. 3r), and is continuous around each dyad member (Wellman et al., 1998a). The outer layer encompasses the whole dyad and is variable in thickness, to absent in places. This layer could be a thick envelope or extra-exosporal material (Wellman et al., 1998a) or an outer wall similar to that observed in Chorridus var. splendidus and Chelinohilates erraticus. The nature of the junction between proximal surfaces is unknown, although TEM sections show a thickening of the outer layer where the two units of the dyad meet at the surface (Fig. 3r), similar to that observed in D. murusattenuata type II sectioned by Taylor (1997) (Fig. 2a).

All of the dyads described above were permanent, not only because of the fusion between the proximal surfaces, but also as a result of an outer encompassing wall. Where these spores occur as monads in the dispersed record (Richardson, 1996a), they have been torn apart, rather than separating naturally.

Although our original diagnosis of Partitatheca stated that the in situ spores were dyads (Edwards et al., 2012a), more recent work has revealed a single Lochkovian valvate sporangium containing permanent tetrads (Edwards et al., 2012b). Specimen NMW2012.17G.3 (Fig. 4a) is a small sporangium similar to Partitatheca splendida, containing laevigate tetrads with collapsed and heavily folded distal walls (Fig. 4b). These lack ‘lines of attachment’/sutures between members of the tetrad, the junctions being marked by furrows (Fig. 4c). Thus, they superficially resemble Cheilotetras (Edwards et al., 2012b). Sections show thin homogeneous to granular spore walls, becoming narrower proximally, where they are fused together (Fig. 4d). A wedge-shaped ‘plug’ of material occurs at the junction, similar to that observed in D. murusdensa described above (arrow a in Fig. 4d). A thin outer layer encompasses the whole tetrad (arrow b in Fig. 4d).

A. (ii) Permanent tetrads in bivalved elongate sporangia

Two additional elongate sporangial specimens containing permanent tetrads possess a valvate construction, but are bivalved rather than quadrivalvate.

Lower Devonian specimen NMW99.11G.2 is an elongate, narrow sporangium composed of two valves that are partially separating, with complete apical rounded tips, but fractured proximally (Fig. 4e). Unlike those in the quadrivalvate sporangia, cell walls are less well defined and smaller, although their topography suggests some degree of lateral shrinkage. Stomata were not observed. The in situ spores are laevigate permanent tetrads with collapsed distal walls and no ‘lines of attachment’/sutures (Fig. 4f), but contacts between members of the tetrad are commonly marked by superficial furrows. They are thus most comparable with Cheilotetras. In semi-thin section, the spores have a similar organization to the laevigate permanent tetrads within the quadrivalvate sporangium (NMW2012.17G.3); thin spore walls, narrowing proximally, where they are fused together, with an outer wall that encompasses the whole spore. However, the TEM sections are different, showing gross layering or laminate structures (Fig. 4g).

The second specimen, NMW99.19G.2, from the same locality was originally described by Habgood (2000). This sporangium consists of two valves, with rounded apical tips, that have separated with the valve margin walls curled inwards (Fig. 4h). The outer surfaces are smooth, apart from where cell outlines are faintly visible (Habgood, 2000, fig. 20). They are comparable in size to those of Partitatheca. Stomata were not observed. The nature of the in situ spores was difficult to determine in SEM, but their main characteristics are irregular, flexuous muri, forming broad reticula (Fig. 4i) (Habgood, 2000). TEM sections revealed that these spores are envelope-enclosed fused tetrads, and thus belong to the dispersed spore genus Velatitetras. Their ultrastructure is quite complex, comprising a thick envelope with voids, which is detached from the inner spore units that have homogeneous walls, but outward projections and occasional voids near the lumina (Figs 2g ii, 4j).

B. (i) Hilate monads and separating dyads in discoidal sporangia (Fig.  5 )

Laevigate and sculptured hilate monads occur in abundance in Upper Silurian and Lochkovian palynological assemblages in the Welsh Basin, sometimes outnumbering trilete monads in sample counts (Richardson, 2007), and with an increasing proportion of sculptured forms in the younger rocks. Both forms have been recorded in discoidal spore masses covered by an acellular cuticular layer (Fig. 5a,d) with a central tuft on one surface (Fig. 5e), presumably marking the position of attachment of the subtending axis, although more extensive remains of the latter have never been recorded. In 2011, Morris et al. erected a new genus, Lenticulatheca, to accommodate such fossils. Members are united in the possession of:

  • discoidal presumed terminal sporangia;
  • sporangial wall represented by a laevigate, acellular, cuticular layer;
  • sporangial cuticle with white lines (?lamellae)*;
  • laevigate and sculptured hilate monads;
  • bi-layered exospore/spore wall with variation in relative widths of the layers.

*Rare occurrences.

From the Lochkovian assemblage, the presence of sculpture on the distal wall combined with the nature of the hilum imposes much greater scope for detection of diversity, and allowed Morris et al. (2011) to erect four species in their new genus, Lenticulatheca. All contained species of the dispersed spore Cymbohilates, with one species per sporangium, their names reflecting either distal sculpture or hilate features in the dispersed species. Thus, Lenticulatheca mesodeca contains spores assignable to Cymbohilates mesodecus Richardson in Morris et al. (2011), which are characterized by a smooth hilum and sculpture of equally spaced micrograna (Fig. 5j,k). Lenticulatheca allenii contains Cymbohilates allenii Richardson (1996a), a monad distally sculptured with small micrograna, microconi, microrugulae and microbaculae, but also proximally, with similar, yet even smaller, sculpture (Fig. 5n,o). In Lenticulatheca variabilis, spores belong to Cymbohilates variabilis Richardson (1996a), with regular radial, concentric and tangential folds/muri on the hilum and distal microconi and microbaculae (Fig. 5l,m). Cymbohilates allenii var. magnus Richardson (1996a), found in Lenticulatheca magna, has an irregular reticulum on the hilum with mainly micrograna on distal and subequatorial surfaces (Fig. 5h,i).

In addition to these, there are also sporangia in coeval and Upper Silurian strata that contain laevigate monads (Fig. 5a,b) (Wellman et al., 1998b; = Groups A & E), belonging to the dispersed spore Laevolancis divellomedia (Chibrikova) Burgess & Richardson (1991) (the authors consider the emendment by Breuer et al., 2007 of the younger genus Gneudnaspora Balme, 1988 to include the former to be invalid). The Silurian specimens (NMW97.1G.1; NMW96.30G.2, Group A) show two layers, both homogeneous, in TEM (Fig. 5c), with the outer consistently slightly wider (c.10–20%) than the inner. The two Lochkovian specimens (NMW96.30G.4; NMW97.1G.6, Group E) have less complete margins (Wellman et al., 1998b, fig. 9a,b) and, although placed in the same spore genus as the older ones, their laevigate spores differ in the presence of a low narrow ridge delimiting the hilum (Wellman et al., 1998b, fig. 9f,h,j,k). TEM sections show a bi-layered wall (albeit with reversed staining) (Wellman et al., 1998b, fig. 9 l,n), but the outer layer is much narrower than in the Silurian specimens, their relative dimensions being much closer to those in coeval sculptured forms.

We have no doubt that these specimens belong to the genus Lenticulatheca and that two new species should be erected to accommodate the variation in spore characters. Indeed, it is obvious from Wellman et al. (1998b), where the appellation Laevolancis divellomedia sensu lato was employed, that simplicity in spore morphology masks the disparity of the producers, as further evidenced by elongate spore masses from the Silurian locality.

B. (ii) Permanent sculptured tetrads in discoidal sporangia

The first of two Lower Devonian discoidal sporangia that contain sculptured permanent tetrads is specimen NMW2012.17G.2, which possesses similar sporangial characters to those described above containing hilate monads, with laevigate, acellular cuticular layers representing the sporangial walls (Fig. 6a) (Edwards et al., 2012b). The in situ spores are circular and possess ridges and furrows (Fig. 6b,c) that are suggestive of a tetrad configuration. The junctions between the tetrad members are unclear, with no lines of attachment; therefore, apart from occasional splitting (Fig. 6c), the tetrads are interpreted as permanent. This is also illustrated in thin section, with lack of distinction between the homogeneous distal walls of each tetrad member (Fig. 6d). However, narrow proximal walls are clearly seen extending into the lumen, indicating that, although unfused, the fragility of these walls means it is unlikely that the tetrad would separate successfully. The spores are sculptured with elements ranging from microverrucae, vermiculate/acinoform (biform) coni, with varying degrees of spacing and fusion. They are most similar to the dispersed genus Acontotetras Richardson (1996a); but the sculptural elements are too different from the type species (widely spaced grana and microconi), so they were assigned to Acontotetras sp. (Edwards et al., 2012b).

The second specimen, NMW2012.17G.1, is a fragment of an originally discoidal sporangium, but, unlike the first specimen, possesses a cellular wall (Fig. 6e) (Edwards et al., 2012b). In cross-section it comprises a single layer of rectangular cells (Fig. 6f). Degradation of these cells has produced a honeycomb surface pattern, while towards the centre of the specimen a wrinkled surface is suggestive of a cuticular layer (Fig. 6e). Stomata are observed on the proximal surface (arrow in Fig. 6e). The sporangium contains permanent sculptured tetrads (Fig. 6g,h), with sculptural elements that range from micrograna, verrucae to biform, with varying degrees of spacing and fusion (Fig. 6i,j). The junctions between tetrad members are unclear under SEM, appearing as either a ridge (Fig. 6i) or a furrow (Fig. 6h), with no clear lines of attachment. Based on surface features alone, these spores were also assigned to the dispersed spore genus Acontotetras, specifically the type species Acontotetras inconspicuis Richardson (1996a).

In light microscopy thin section and TEM (Fig. 6l,k), the majority of the spore wall is homogeneous, with a thin outer layer that encompasses the entire tetrad, interpreted as an envelope. The distal spore walls are thick, particularly at the junctions between the tetrad members, the latter represented by a sinuous line, accompanied by two narrow tapering walls that extend freely into the lumen, interpreted as the proximal walls. As with the first specimen, the fragility of these internal walls suggests that the tetrad did not separate naturally, but may have been torn apart.

C. Permanent polyads in sporangia of unknown dehiscence

It is important to note here that the following are not considered to be biological groupings but are united by similar characteristics.

(i) Laevigate permanent dyads

Two specimens from the Lower Devonian comprise sporangia containing laevigate permanent dyads. The first (NMW96.11G.6) comprises a beaker-shaped sporangium with truncated tip, terminating an unbranched axis lacking anatomy (Fig. 4k), named Culullitheca richardsonii Wellman et al. (1998a). In situ dyads are permanent, but show ‘lines of attachment’ (Fig. 4l). The spore walls are homogeneous (Fig. 4m). They are similar to D. murusdensa, but their walls are markedly invaginated distally, whereas in D. murusdensa they are usually inflated.

In the second specimen (NMW97.42G.4), one branch of a bifurcating stem terminates in a fusiform sporangium that tapers distally, named Fusiformitheca fanningiae (Fig. 4n; Wellman et al., 1998a; Xue & Wang, 2011). The superficial cells of the sporangial wall are large and fusiform, the surface limiting the sporangial cavity minutely reticulate. The dyad walls are thin, highly folded and distally invaginated (Fig. 4o) and covered by a thin, closely adherent envelope. In section the walls are homogeneous (Fig. 4p). As the envelope masks the junction between the dyad members, they could belong to either Segestrespora laevigata Burgess, 1991 (an enveloped fused dyad) or Abditusdyadus laevigatus Wellman & Richardson (1996) (an enveloped unfused dyad). It is possible that the sporangium is immature and intact, and therefore, based on sporangial wall cells, could belong to Partitatheca.

(ii) Permanent tetrads in bifurcating specimens

Three specimens from the Lower Devonian with either bifurcating sporangia or subtending axes contain permanent tetrads. The first specimen, named Grisellatheca salopensis Edwards et al. (1999), comprises a short length of smooth axis that terminates in a bifurcating sporangium (Fig. 7a), whose walls are incomplete but show traces of large cells. It contains laevigate fused tetrads with wrinkling across unit junctions suggestive of an envelope and hence affinities with Velatitetras (Fig. 7b,c). Where fortuitously fractured, it was observed that the spore walls lacked layering (Fig. 7c).

The second specimen is a bifurcating axis described by Edwards et al. (1999) (NMW98.23G.2) (Fig. 7d), in which one branch has a terminal cavity containing laevigate tetrads (Fig. 7e) with minute wrinkles in places and with well-defined, slightly sinuous lines of attachment. It is hence attributable to Tetrahedraletes. However, the spore wall is bi-layered, with a very narrow electron-dense, sometimes detached, outer layer and a much thicker electron-lucent inner layer surrounding each member of the tetrad (Fig. 7f).

The third specimen (NMW96.11G.3) is a bifurcating axis terminated by the bases of two sporangia or the base of a bifurcating sporangium (Edwards et al., 1999) (Fig. 7g). Preserved fragments of cells in the sporangium wall are large. The spores are permanent tetrads with an irregular microgranulate sculpture (Fig. 7h), although this sculpture may be a precipitate of tapetal origin, as it is also found on the inside surfaces of the sporangial wall (see section VI ‘Spore wall formation and the nature of the envelope’ for discussion), and a thick electron-dense envelope that is fused to the spore wall (Fig. 7i). They are most closely attributable to Velatitetras. Each unit is surrounded by a three-layered wall comprising a central zone composed of globular material sandwiched between two homogeneous layers.

(iii) Permanent tetrads in spore masses with sporangial coverings

Two specimens are united here as discoidal spore masses with sporangial coverings that contain permanent, enveloped tetrads. Spores of NMW96.11G.4 from the Upper Silurian (Edwards et al., 1999; mislabelled NMW98.23G.1) are enclosed by an amorphous homogeneous layer (Fig. 7j), but lack evidence of any axial attachment. The spores are tetrads, enclosed tightly by an adherent envelope, with an irregularly granular outer surface. Sections indicate that the envelope is slightly more electron-dense than the spore wall and limited externally by a darker line (Fig. 7l). The spore wall shows faint striations reminiscent of laminae following surface contours (Fig. 7l). Spores were assigned to Velatitetras cf. anatoliensis (Fig. 7k) because the envelope sculpture does not exactly match that described for the species (Steemans et al., 1996).

The second specimen (V.68196(2)), from the Lower Devonian (Edwards et al., 2012b), is an irregularly shaped spore mass partially covered by amorphous ?cuticular material (Fig. 7m). The spores are laevigate tetrads enclosed in envelopes with superficial sinuous folding (Fig. 7n,o). In section this envelope possesses inward projections. The spore walls themselves are homogeneous. Spores were assigned to Velatitetras sp. (Edwards et al., 2012b).

D. Alete monads

A single, poorly preserved specimen of a partial sporangium from the Lower Devonian is included here because it is the only specimen we have found that possesses alete monads (NMW2012.17G.5) (Edwards et al., 2012b). The wall comprises two strap-shaped structures with straight, but incurved, margins (Fig. 6m), suggestive of controlled longitudinal dehiscence, although marginal cells do not appear modified as occurs in Partitatheca. These valves, if indeed present, would be of unequal size. However, the sizes and shapes of the predominantly elongate sporangial epidermal cells (Fig. 6n) are similar.

The monads are large (50–76 μm) compared with hilate monads and have strongly folded walls covered in separated to closely spaced coni, microconi and micrograna (Fig. 6o,p). An unusual feature noted in both light microscopy and TEM is the fusion between tips of the sculptural elements on adjacent spores (Fig. 6q) with, in places, a very thin layer connecting the junctions (Fig. 6r). Such spores have not been recorded in dispersed assemblages.

3.3 Discussion on relationships and affinities

Our use of groupings, based on sporangial and spore characters, indicates to some extent our broad conclusions on their relationships. In particular, a major perceived distinction between producers of permanent polyads and those with separated products is emphasized. The relationships among tetrad producers are less clear-cut.

In the various species of Partitatheca, there is a distinct lineage of dyad and tetrad producers that combine characters of tracheophytes and bryophytes (Table 1; Fig. 1). Unique features are the presence of a layer that surrounds the entire polyad and complete fusion of the proximal walls of each polyad member, and thus viable dispersed monads are never produced. The complex ultrastructure of the spore wall that is seen in the older dispersed spore examples implies links with some of the oldest embryophytes and possibly hepatics, although the limited sporophytic branching and the presence of stomata support a more derived affinity (Fig. 1). Indeed, the combination of stomata and dispersed polyads supports the inference that basal stomatophytes possessed distal spore germination.

By contrast, in Lenticulatheca the dyads do not possess enclosing layers and separate within the discoidal sporangia before dispersal. The Lochkovian locality has also yielded similarly constructed sporangia which contain monads with similar sculpture but differ in the possession of trilete marks (Fig. 5p–s) (Morris et al., 2011). Such spores belong to the StreelisporaAneurospora morphon in dispersed assemblages and are particularly significant because they are recorded in subspecies of Cooksonia pertoni (Fanning et al., 1988; Edwards et al., 1995a). A new genus, Paracooksonia, was erected to emphasize differences in sporangial characteristics (Morris et al., 2011). In the basal tracheophyte Cooksonia, a discoidal spore mass is covered at maturity by a uniseriate wall comprising a single layer of cells distally while the multicellular subtending axis widens below the sporangium.

There are thus three groups of small plants with discoidal terminal sporangia united in their possession of monads with a bi-layered wall construction. In each group, distal spore walls range from laevigate to apiculate. It could be argued that differences in the tissues enveloping spore masses could be taphonomic, namely, that decay of wall tissues could result in cuticle-enveloped masses of spores. However, a prominent cuticle is not a characteristic of well-preserved sporangial walls in Cooksonia, and a cellular construction has never been found in the hilate monad group. Nor is there a cuticular lining to the sporangium cavity in Cooksonia pertoni, although it has been recorded elsewhere (Morris et al., 2012b). We suspect that in Paracooksonia and Lenticulatheca the original sporangial walls were fleshy, that is, composed of very thin-walled parenchymatous tissue with low fossilization potential, but surrounded by a thick cuticle. We discount the possibility that the wall was still developing, because the spores are mature. Regardless of these uncertainties, we are convinced that the plants in the Lenticulatheca complex are closely related to the basal tracheophytes (Fig. 1).

The discoidal sporangia containing permanent sculptured tetrads belonging to Acontotetras share some similarities to those in this lineage, including one with a cooksonioid-like sporangial wall (Fig. 6e). However, the tetrads do not possess a clear bi-layered wall construction (Fig. 6d) as observed in the monad equivalents described above. There is a suggestion of bi-layering in one specimen, but this is based on a very thin outer layer that encompasses the entire tetrad (Fig. 6k,l), which was interpreted as an envelope. Without good TEM sections and more specimens we can only tentatively suggest that they were related to this lineage.

The relationships of the remaining sporangia, each represented by a single specimen, are also difficult to assess. Those with two valves (Fig. 4e–j) may well be close to the Partitatheca complex, the relationship finding support in the complex wall construction of NMW99.19G.2 (Fig. 4j) and because the in situ spores are assigned to Velatitetras sp. (Fig. 4i). A second specimen (NMW99.11G.2) contains Cheilotetras sp., but the sporangial wall cells are smaller and far more numerous (Fig. 4e).

Even more difficult are specimens where we have no evidence of dehiscence. Particularly intriguing is Grisellatheca salopensis (Edwards et al., 1995b, 1999), the first mesofossil in which permanent laevigate tetrads, probably best assigned to Cheilotetras sp., were recorded (Fig. 7a–c), Certain anatomical features, including a single putative elater and the possibility that sporangial tissues were embedded in the tips of a bifurcating gametophyte, persuaded of affinities with the hepatics, but better preserved specimens are needed to substantiate this.

The three remaining illustrated specimens are united in the possession of tetrads with ‘ornamented’ envelopes assignable to three species of Velatitetras. However, this does not indicate a close relationship even in the two examples where the outer layer is closely adherent or fused. The specimen shown in Fig. 7(g) comprises the bases of two sporangia with few spores (Fig. 7h), which show complexity in wall ultrastructure (Fig. 7i) that is comparable to those within the Partitatheca complex, leading to the possibility that the fertile regions of the specimen represent the bases of valvate sporangia. By contrast, the walls of the tetrads in specimen NMW96.11G.4 (Fig. 7j,k) are homogeneous (Fig. 7l). The sporangium itself, being discoidal and covered by cuticle, is reminiscent of the Lenticulatheca complex, but the nature of the spores precludes such a relationship. Thus, Velatitetras, as defined by palynologists, was clearly produced by a number of plants of different affinities.

3.4 Dyads versus tetrads

These recent studies indicate that sporangia contain monads, dyads or tetrads, and that fossils of similar architecture contained monads that were produced from either dyads or tetrads. There are never mixtures, suggesting strict genetic control, with the possibility that the same species could produce polyads or monads at different times, be this on the same plant or in different populations. The latter has been attributed to environmental conditions by a number of workers, although there has sometimes been a certain lack of clarity resulting from the use of the terms ‘taxa’ and ‘plant’ (? individual specimen versus clone versus species etc.; e.g. Hemsley, 1994; Richardson, 1996a, 2007; Lavender & Wellman, 2002; Steemans et al., 2012). Lavender & Wellman (2002) had hypothesized that the great abundance of tetrads as compared with trilete spores at a number of Lochkovian localities indicated that the same ‘taxa’ produced different configurations at different times in response to ecological factors, rather than reflecting a different biological origin. This builds on Gray's contention of the importance of tetrads compared with the more easily dispersed triletes in founder populations under stressed environments (Gray, 1985). It is of some interest here that Visscher et al. (2004) reported a proliferation of lycophyte microspore tetrads in the dispersed spore record coincident with the end-Permian crisis, there associated with postulated increased UV radiation. Steemans et al. (2012) extended such comparisons to include dyads and very strongly advocated that the type of spore produced by basal embryophytes depended on environmental factors and ‘probably do(es) not reflect their biological affinities alone.’ Thus, they noted clear morphological relationships between Cymbohilates cymosus Richardson (1996a) and Chelinohilates erraticus Richardson (1996a) (tetrads, dyads, and alete monads) and Cymbosporites stellospinosus Steemans (1989) and Chelinospora cassicula Richardson & Lister (1969) (trilete monad), respectively, and between Cymbohilates disponerus Richardson (1996a) (alete monads and dyads), Acontotetras inconspicuis Richardson (1996a) (tetrad) and Cymbosporites proteus McGregor & Camfield (1976) (trilete monad). Our own work broadly agrees with such pairings between dyads and monads (e.g. Artemopyra brevicosta Burgess & Richardson, 1991 and Emphanisporites protophanus Richardson & Ioannides, 1973; Velatitetras reticulata Burgess, 1991 and Abditusdyadus histosus Wellman & Richardson, 1996), although our ultrastructural evidence links somewhat different species and thus provides biological support for such groupings. In the dispersed spore record there is ultrastructural evidence for close relationships between Upper Ordovician dyad Abditusdyadus histosus (identified as Segestrespora membranifera in Taylor, 2001) and tetrad Velatitetras reticulata, leading Taylor (2000, 2001) to comment that the producers were the same plant or group of plants.

However, we are not so convinced of a developmental relationship between tetrads and dyads. Steemans et al. (2012) suggested that such a relationship might exist between Cymbohilates baqaensis Breuer et al. (2007) and Cymbosporites dammamensis Steemans (1995), because some of the former show a tear in the hilum that is roughly triangular and mimics a trilete mark, with the inference that it too developed from a tetrad. We consider such a shape to be fortuitous and not uncommon in dispersed hilate monads.

Our results on the Cooksonia–Paracooksonia–Lenticulatheca complex are particularly relevant to the hypothesis that ‘plants’, presumably ecotypes, produced different spore configurations under different environmental conditions. That all these taxa grew at the same time is self evident, but the fossils are allochthonous and represent the remains of plants that grew in a range of habitats within the catchment area of the river systems. The small number of the cuticularized sporangia containing trilete spores of the Streelispora–Aneurospora complex (Paracooksonia) and hilate spores belonging to Cymbohilates (Lenticulatheca) compared with the sporangia of Cooksonia pertoni perhaps indicate that Lenticulatheca and Paracooksonia colonized areas further from the depositional area, with Cooksonia pertoni in its vicinity. Alternatively, it could be argued that the latter had more resilient sporangial walls. Consideration of the relative numbers of their spores in the dispersed spore record is not helpful. Members of the StreelisporaAneurospora complex dominate the record but the relative contributions of the two cooksonioid genera cannot be distinguished. However, the numbers of ornamented hilate monads are higher than might be expected from the mesofossils, even allowing for the fact that the spore rain is derived from a wide geographical area. Indeed, Richardson (2007), in an attempt to use palynological and lithological data to gain a better understanding of the regional vegetational distribution in the Welsh Basin in Early Devonian times, had earlier commented on the relatively high percentages of ornamented monads (cryptospores and triletes) compared with laevigate hilate monads in sediments deposited in the flood plains of meandering rivers. This he attributed to the diverse vegetation growing in varied ephemerally wet habitats and contrasted it with the high percentage of laevigate hilate monads which were recovered from sediments deposited on alluvial plains subjected to occasional marine inundation and hence thought to derive from coastal vegetation. Studies on in situ spores as described here will allow further refinement of the data sets and hence deductions on the source plants.

4 Evolutionary overview

Having identified the spores isolated from the majority of fertile specimens from the Lochkovian (see Table 2) and Silurian localities in the Anglo-Welsh Basin, comparisons with the same or closely related taxa in contemporary or earlier dispersed assemblages (Fig. 8) allow deliberations on the history of these early land plants, even in the absence of meso- and megafossils. However, before interpreting the miospore record in this way, it should be recognized that, until relatively recently, few stratigraphical palynologists outside the oil industry recognized reworking problems and hence the pitfalls of accepting ranges of species without question. An excellent example of such reworking occurs in the Emsian sequences of the Whitney Borehole (Richardson & Rasul, 1978, 1979) where species of the genus Emphanisporites occur in reverse order to the established occurrences. However, with care, reworked palynomorphs can be a useful source of dynamic data, providing evidence for basin tectonics. Pulses of uplift in the Caledonian uplands to the north and west of the Anglo-Welsh Basin, a major source of Lower Old Red Sandstone sediments (Allen, 1974, 1985; Allen & Crowley, 1983), caused reworking from proximal landscapes into fluvial systems. Richardson has observed evidence for these pulses of uplift in the dispersed spore record, from influxes of reworked spores that are noted particularly at four stratigraphical levels. The first influx occurred in the lowermost Lochkovian, where species characteristic of the Upper Ordovician and lowermost Silurian frequently show a spatially erratic and partly discontinuous distribution. After a period of quiescence, a second level is noted near the top of the Ditton Formation (late Lochkovian), where the palynological assemblages, which may consist entirely of marine acritarchs, occur within coarse fluvial sediments and are thus inconsistent with their sedimentary environment. Slightly higher in the sequence, these acritarch-rich assemblages are associated with rocks containing pebble-sized clasts bearing marine macrofossils. By the late Lochkovian and into the Pragian, these tectonic pulses occurred with increasing intensity, as reflected in the abundance of reworked spores. In the Senni Formation of the Brecon Beacons (Pragian) (Hassan, 1982) they are derived mainly from Lower Palaeozoic strata and occur commonly in most sediments. Finally, in Emsian sediments of the Witney Borehole, Lower Palaeozoic marine microfossils are so abundant (Richardson & Rasul, 1978, 1979) that they are often the most frequent fossils present.

Table 2. Review of in situ spores found in sporangia from Brown Clee Hill Lochkovian locality
Genus/TypeSpeciesIn situ sporesReferences
Trilete producers
Cooksonia pertoni subsp. pertoniAmbitisporites sp.Fanning et al. (1988); Morris et al. (2012b)
pertoni subsp. apiculispora Streelispora newportensis Fanning et al. (1988); Morris et al. (2012b)
Aneurospora spp.Fanning et al. (1988); Morris et al. (2012b)
pertoni subsp. reticulispora? Synorisporites sp.Habgood et al. (2002); Morris et al. (2012b)
banksii cf. Ambitisporites sp.Habgood et al. (2002); Morris et al. (2012b)
hemisphaerica Ambitisporites sp.Edwards (1996)
Paracooksonia apiculispora Streelispora newportensis Morris et al. (2011)
Aneurospora cf. trilabiataMorris et al. (2011)
Aneurospora sp.Morris et al. (2011)
Tortilicaulis offaeus Apiculiretusispora sp.Edwards et al. (1994); Morris et al. (2012b)
Pertonella sp. Retusotriletes coronadus Edwards (1996)
Tarrantia salopensis ? RetusotriletesEdwards (1996)
Resilitheca salopensis Retusotriletes sp.Edwards et al. (1995)c
Sporathylacium salopense ? ApiculiretusisporaEdwards et al. (2001)
Salopella cf. marcensisAneurospora sp.Edwards et al. (1994)
cf. Horneophyton sp.Emphanisporites cf. micrornatusEdwards & Richardson (2000)
Spheroidal sporangium ? Iberoespora sp.Morris et al. (2012b)
Circular spore mass Emphanisporites sp.Morris et al. (2012b)
Cryptospore producers
Permanent dyads
Culullitheca richardsonii Dyadospora murusdensa Wellman et al. (1998a)
Fusiformitheca fanningiae Laevigate enveloped dyadsWellman et al. (1998a)
Partitatheca splendida Cymbohilates horridus var. splendidusEdwards et al. (2012a)
horrida Cymbohilates horridus var. horridusEdwards et al. (2012a)
densa Cymbohilates horridus complexEdwards et al. (2012a)
cymosa Cymbohilates cymosus Edwards et al. (2012a)
sp. Chelinohilates erraticus Morris et al. (2012a)
Ficoiditheca aenigma ?Morris et al. (2012a)
Separating dyads
Lenticulatheca magna Cymbohilates allenii var. magnusMorris et al. (2011)
allenii Cymbohilates allenii var. alleniiMorris et al. (2011)
variabilis Cymbohilates variabilis complexMorris et al. (2011)
mesodeca Cymbohilates mesodecus Morris et al. (2011)
? sp. Cymbohilates rhabdionus Morris et al. (2011)
Permanent tetrads
Grisellatheca salopensis Velatitetras sp.Edwards et al. (1999)
Discoidal Velatitetras sp. Edwards et al. (1999)
Bifurcating axis Velatitetas sp. Edwards et al. (1999)
Irregular spore mass ? Velatitetras rugulataEdwards et al. (2012b)
Cooksonioid  Acontotetras inconspicuis Edwards et al. (2012b)
Paracooksonioid Acontotetras sp.Edwards et al. (2012b)
Bifurcating axis Cheilotetras sp.Edwards et al. (1999)
Valvate sporangium 1 Cheilotetras sp.Edwards et al. (2012b)
Valvate sporangium 2 Cheilotetras sp.Edwards et al. (2012b)
Irregular spore mass Cheiloteteas sp.Edwards et al. (2012b)
Bifurcating axis  Tetrahedraletes medinensis Edwards et al. (1999)
Ovoid spore mass  Tetrahedraletes medinensis Edwards et al. (1999)
Irregular spore mass Tetrahedraletes sp.Edwards et al. (1999)
Irregular spore mass ? Tetrahedraletes sp.Edwards et al. (2012b)
Alete monads
Valvate sporangium Sculptured alete monadsEdwards et al. (2012b)
Incertae sedis
cf. Sporogonites Crassitate sculptured sporesEdwards (1996)
Figure 8.

Stratigraphical range chart for selected cryptospore and trilete spore taxa for global and Anglo-Welsh Basin occurrences between the Upper Ordovician and Upper Devonian. Compiled from the John Williams Index of Palaeopalynology at the Department of Earth Sciences, Natural History Museum, London, UK.

Issues with reworking aside, the new information on in situ spores is of value not only in reconstructing disparity in Lochkovian vegetation on regional to global scales, where mega- and mesofossil evidence is lacking, but also in tracing the history of its various components. Here we must emphasize that we are concerned only with spores. We cannot infer, because of the possibility of sequential acquisition of other characters, such as stomata and branching sporophytes, that earlier plants were identical to Upper Silurian and Lochkovian representatives. However, this approach involving spores has been adopted for the tracheophytes, in that the presence of trilete spores in the Middle Ordovician (Steemans et al., 2009) is the bench mark for the appearance of the tracheophytes and, as it includes laevigate forms attributed to Ambitisporites Hoffmeister (1959) for the origin of the Cooksonia complex, because Ambitisporites was recorded in Cooksonia pertoni of Přídolí age (Fanning et al., 1988). Their ancestors might be sought in the Ordovician producers of laevigate, nonenveloped tetrads (e.g. the Tetrahedraletes complex) but these were probably derived from a number of taxa, as indeed were species of Ambitisporites. Specimens assigned to Tetrahedraletes differ in the relative dimensions of distal and proximal walls, the degree of fusion of the latter, the extent of indentation at the margins of contact faces, the convexity versus concavity of the distal surface (Strother et al., 1996) and the nature of the shape of the equatorial thickening in section. However, ultrastructural studies have been disappointing (Taylor, 1995b, 2002). Most Ordovician to Silurian specimens examined usually showed homogeneous walls, but sometimes occasionally spongy walls, apart from one with occasional sporadic linear structures parallel to the surface and another where a thick homogeneous wall is external to a highly convoluted single lamina (23–70 nm thick) that surrounds the lumen (Taylor, 2009). Thus, the bi-layered wall that characterizes spores in the Cooksonia pertoni complex is first seen in in situ triletes (Ambitisporites, Rogerson et al., 1993) and dyads (Laevolancis divellomedia type A, Wellman et al., 1998b). Steemans et al. (2012), in attempting to trace such lineages, mentioned that some Tetrahedraletes spores showed a tendency to become detached and Rimosotetras problematica Burgess (1991) is characterized by loosely attached spores. In fact, Gray (1991) suggested that permanent tetrads tend to break apart revealing a proximal triradiate mark. Evidence for probable physical tearing apart of a tetrad comes from Late Ordovician monads, where a tear on the proximal face mimics a trilete mark (Imperfectotriletes (?Ambitisporites Richardson, 1988) Steemans et al., 2000). Imperfectotriletes varvdovae being crassitate (equatorially thickened) is closest to Ambitisporites sp. in Cooksonia pertoni. However, there is no supporting ultrastructural evidence.

The earliest records of sculptured trilete spores in the Mid to Late Ordovician comprise Synorisporites, either with low verrucae or convolute muri, a ?Synorisporites and Aneurospora with grana or small verrucae (Steemans et al., 2009). This suggests the presence of further members of the Cooksonia complex much earlier than previously known (Homerian; Richardson, 1988; Burgess & Richardson, 1991) and, more importantly, the earlier diversification of the presumably vascularized embryophytes (Steemans et al., 2009).

Returning to our hypothesis that the cooksonioid complex contained plants with either separating dyads or trilete spores, the records of hilate monads with a bi-layered wall construction can be traced back only to the Přídolí (Fig. 5a–c; Wellman et al., 1998b), despite the fact that laevigate forms, represented by Laevolancis divellomedia, appeared in the dispersed record in the Early Silurian (Rhuddanian) (Steemans et al., 2000). This is partly because of the lack of sectioning studies on older examples, but also because the range of ultrastructure noted in spore masses and sporangia of L. divellomedia from the Přídolí (Wellman et al., 1998b) indicates diverse affinities of the producers, making it impossible to identify more precisely the older dispersed examples on morphology alone. However, Wellman et al. suggested that the monads L. div-ellomedia were ‘derivatives’ of certain Dyadospora species (obligate dyads), and we would postulate that, based on the bi-layered nature of its relatively simple ultrastructure and the absence of the outer wall that completely surrounds the dyad, the monads were derivatives of D. murusattenuata type II (Taylor, 1997). On such tentative evidence, stem group tracheophytes might be represented in the earliest spore assemblages, although we should emphasize that the earliest examples of Dmurusattenuata with type II anatomy are found in the Lower Silurian. Similar reasoning might be employed to link D. murusattenuata type I with the C. horridus complex (Fig. 2a,d), extending this lineage in which the spores never separated into the Upper Ordovician. However, as with Laevolancis, the types cannot be distinguished from one another by external morphology alone and more sectioning studies of older examples are required.

Considering further this second lineage of plants possessing permanent tetrads and dyads, sculptured forms of the latter, belonging to the C. horridus complex or Cymbohilates cymosus, are only recorded in the Lochkovian. However, one specimen contains permanent laevigate dyads assigned to D. murusdensa, some with incompletely elucidated wall ultrastructure (Fig. 3r; Wellman et al., 1998a) of complexity broadly similar to that observed in dispersed specimens of D. murusattenuata type I from the Upper Ordovician sectioned by Taylor (1997).

Of the permanent tetrads, the laevigate Cheilotetras sp. occurs in two, albeit Lower Devonian, valvate sporangia (Fig. 4a–d, e–g; Edwards et al., 2012b). The genus Cheilotetras is recorded as far back as the early Wenlock (Wellman & Richardson, 1993), but so far has produced disappointing sections, although some with gross layering (Fig. 4g). However, Přídolí trilete monads and tetrads of Cymbosporites echinatus (Richardson & Lister, 1969) were thought close to the C. horridus complex because of similarities in the ornament on distal surfaces. Johnson & Taylor (2005) demonstrated a bi-layered wall construction in which an electron-dense layer extends around the whole spore and a complex outer wall is confined to the equator and distal wall (Fig. 2j). The latter corresponds to the surface which would be exposed in the tetrad. This layer is tripartite, the outer homogeneous and extending into the sculpture, the middle with a spongy texture and the inner with lamellate construction. It thus broadly corresponds to the layer that extends around both units of the permanent dyad in C. horridus. Isolated monads almost invariably show collapsed poorly preserved proximal faces, the trilete mark either absent or comprising very narrow laesurae, often confined to the equatorial regions, or represented only by a tear that mimics its shape, all features suggestive of poorly controlled separation. It is tempting to compare this with remnants of the fused proximal walls noted in torn-apart dyads in C. horridus (Richardson, 1996a, Plate 6, fig. 1), although impossible to pontificate on the viability of the spores in this developmental scenario.

Regarding the enveloped tetrads and dyads, spores belonging to Lochkovian Velatitetras sp. (Fig. 2g ii, iii) within an elongate valved sporangium and dispersed Upper Ordovician forms of Abditusdyadus (Segestrespora) (Fig. 2e) possess thick envelopes, some with internal structure (Fig. 2g ii). If these envelopes are considered homologous to the outer enveloping wall, they show similarities in ultrastructure with the C. horridus complex and D. murusattenuata type I. Therefore, we tentatively hypothesize homology between the thick envelope in these examples and the outer enveloping wall in the C. horridus complex.

Dispersed records show a global decrease in numbers of permanent dyads and tetrads throughout the Silurian, leading to almost complete disappearance in the Lower Devonian (Fig. 8). Anomalous records of permanent tetrads (Filipiak, 2011) and dyads (Breuer et al., 2007) in the Emsian may be a result of reworking. As indicated in section V ‘Variation in the meiotic process’, and based on the assumption of possession of sex chromosomes, there were advantages in terms of reproductive biology for fertilization in gametophytes developed from tetrads. Negative aspects include increased homozygosity and limitations on dispersal. The latter is improved in dyads, but the chances of fertilization are reduced. In both groups their permanent nature would be barriers to germination, especially if an envelope is present. Indeed, Steemans et al. (2012) postulated that changes in frequency of thickness of envelopes declined to facilitate germination. Exit of the germination tube would presumably have been via distal surfaces, where there are no specially modified sites. However, our sections through a spore mass (Fig. 9a) of loosely adhering tetrads (Fig. 9b) show adherence to separation of the thin proximal walls (Fig. 9c,d) and it is possible that, even in the absence of a well-defined trilete mark, the latter sometimes represented by a weakly developed ridge (Fig. 9e,f), the germination tube could have exited here and grown between individual spores to the exterior. Taylor (2002, p. 334) had earlier speculated that such a protected proximal surface ‘might have been “preadapted” to accommodate a specialized zone of weakness’, although he concluded that, in trilete spores, germination via a ‘proximal suture’ must have occurred when that surface became exposed. We therefore conclude that the torn-apart monads noted in dispersed assemblages, with incomplete proximal surfaces, were not viable.

Figure 9.

(a–f) Spore mass with separating tetrads. (g–r) Sporangia containing extra sporal residues. (o–r) Late Silurian in age. All other specimens are Early Devonian in age. All scanning electron microscopy (SEM) unless otherwise stated. (a–f) NMW2012.17G.7, Edwards et al. (2012b). (a) Spore mass covered with acellular layer. (b) In situ spores: loose laevigate tetrads; ? Tetrahedraletes sp./? Imperfectotriletes sp. (c, d) Separation of the tetrads and degradation of walls after application of nitric acid. Arrows indicate weakly developed ridges that resemble trilete marks. (e) Light microscopy of a thin section through a tetrad, showing three tightly adherent units, with fused proximal walls (arrow a) and triangular thickenings where the distal walls are fused together (arrow b). (f) Light microscopy of a thin section through a tetrad, showing four units of a tetrad, with free proximal walls and either partial (arrow a) or complete (arrow b) lines of attachment between units. (g–n) Ficoiditheca aenigma Morris et al. (2012a). (g, i, m) V.68222(1) (holotype). (g) Whole specimen. (h) V.68225a(1). Partial apical dehiscence via triangular-tipped valves. (i) Section through the proximal part of the sporangial wall of the holotype; irregular layer to inside. (j) V.68228(1). Partially complete inflated spherical spores with microgranulate outer surfaces. (k) V.68258(1). Magnification of a spore surface, covered with a cluster of microgranules, composed of smaller spheres. (l) V.68226a(1). ? dyad with invaginated walls and two subequatorial concentric folds, separated by a central furrow. (m) Transmission electron microscopy (TEM) of the tri-layered spore wall in the holotype. The middle electron-dense layer thickens into a wedge where the two units of a dyad meet to form an internal wall. (n) V.68259(1). TEM section through the tri-layered spore wall. (o –r) NMW96.11G.2, Wellman et al. (1998b). (o) Elongate sporangium. (p) In situ spores: Laevolancis divellomedia. (q, r) TEM sections through spores; layered walls bearing granules across the outer surfaces. Bars: (a, o) 200 μm; (b–d, j, p) 20 μm; (e, f, i) 10 μm; (g, h) 100 μm; (k, m, n, q, r) 1 μm; (l) 5 μm. Panels (a–f) were first published in Edwards et al. (2012b) and are reproduced here with kind permission of Elsevier. Panels (g–n) were first published in Morris et al. (2012a) and are reproduced here with kind permission of Wiley. Panel (o) was first published in Edwards (1997) and is reproduced here with kind permission of Springer Science+Business Media. Panels (p–r) were first published in Wellman et al. (1998b) and are reproduced here with kind permission of The Royal Society.

As for the smooth and ornamented hilate monads, recent studies have shown that they are linked to trilete spores, with similar ultrastructure, hence to the parent plants in the Cooksonia complex. The spores and megafossils share a similar stratigraphical distribution across the Welsh Basin, disappearing in the mid to late Lochkovian, which suggests that differences in meiosis and inferred reproductive biology were not responsible for their demise. Instead, it has been noted that there was an increase in numbers of retusoid spores in the dispersed records across the basin at this time, some known to be derived from zosterophylls (Edwards & Richardson, 1996), plus the appearance of megafossils of Zosterophyllum (summary in Edwards & Richardson, 2004), which show a departure from a strictly determinate growth form. This led to the suggestion of competitive displacement (Edwards & Davies, 1990), with the progressive invasion of the plant habitats of the cryptospore producers and rhyniophytes. This radiation of the zosterophylls and higher plant taxa into the Anglo-Welsh Basin during the late Lochkovian may have been related to a shift towards increased rainfall and a wetter climate in the Lower Devonian (Thomas et al., 2006; Hillier et al., 2007; Morris et al., 2012c).

5 Variation in the meiotic process

In modern land plants, the spores are dispersed as single units (monads) whereas the cryptophytes show much greater diversity, including cryptospores dispersed as dyads and tetrads with or without a second enveloping membrane. Hemsley (1994) pointed out that the origin of this diversity could be explained by the nature of the process of meiotic cell division, which when coupled with changes in the timing of the synthesis and deposition of sporopollenin (Blackmore & Barnes, 1987) could give rise to all of the morphologies observed among the cryptospores. Dyads might arise if, during Meiosis I, cytokinesis was followed by separation of the products to form monads, then, depending on the timing of sporopollenin synthesis and deposition during Meiosis II, either permanent dyads with an outer enveloping membrane or hilate monads would result (Fig. 10). Tetrads would form if the products of Meiosis I did not separate and were followed immediately by Meiosis II. Again, the timing of sporopollenin synthesis and deposition during Meiosis II would determine whether monads or permanent tetrads were the end products (Fig. 10). Two lines of evidence strongly imply that these cryptospore types were a normal part of the developmental repertoire of early land plants. First, with few exceptions (e.g. Richardson, 1996b), cryptophyte sporangia never contain mixtures of dyad types or dyads and tetrads, indicating that there was strict genetic control over their development. Secondly, cryptospores were a prominent element of early palynomorph assemblages, persisting for over 60 million yr. Today, normal meiosis in basal land plants follows the tetrad path, resulting in sporopollenin-coated monads, which has led to the unchallenged contention that the obligate tertrahedral tetrad is the plesiomorphic state in land plants (Gray, 1991, 1993). This hypothesis was proposed when the nature of the dyad-producing plants was unknown. Now that they have been documented in cryptophyte sporangia, the possibility that dyads are plesiomorphic in land plants must be given serious consideration. Neither phylogeny nor stratigraphy provides a clear answer. Dyads are almost as abundant as tetrads in early palynomorph assemblages, and like tetrads they have been documented in situ in cryptophyte sporangia; however, the precise phylogenetic relationships of most early cryptospores to major embryophyte clades remain unclear. Stratigraphically, the earliest palynomorph assemblages comprise monads, dyads, and tetrads, with and without an enveloping membrane (Rubinstein et al., 2010; Steemans et al., 2010). So, we cannot say whether the dyad or the tetrad came first. From the perspective of developmental biology, meiosis in modern bryophytes is seemingly a better fit to the tetrad hypothesis. During sporocyte development, the tetrahedral form marking future spore domains begins to develop precociously, with the process of cytokinesis commencing before meiosis (Brown & Lemmon, 2011). There is no indication of separation of cell products during Meiosis I as would be required for dyad formation (Fig. 10). Although informative, these observations tell us only about the plesiomorphic state for crown group embryophytes, whereas the cryptophytes hint at greater diversity in the bryophyte, vascular plant and embryophyte stem groups (Fig. 1). The diversity of cryptospores in the early fossil record testifies to a far greater versatility in meiosis among the basal embryophytes than is evident in living plants.

Figure 10.

Meiosis in tetrad/trilete spore production on the left and hypothetical stages in formation of permanent and separated (hilate monads) dyads on the right. n, ploidy level; c, DNA content of unreplicated haploid genome.

The particular geometry of cryptospores, including their size, the number of units they contain, and the presence or absence of an enveloping membrane, have potential consequences for cryptophyte biology. Many authors have commented that some characteristics of tetrahedral tetrads are typical of founder populations. These include small size and production in large numbers. Tetrad size averaged < 27 μm during the Ordovician, increasing to 40–50 μm and greater during the Llandovery (Gray, 1991), and Wellman et al. (2003) estimated that a small (0.3 × 0.23 mm) sporangial fragment from the Ordovician contained over 2500 tetrads.

Cryptospore geometry is also potentially significant in sexual reproduction, but only for species with gametophyte dioecy. We do not know how gender was expressed in the cryptophytes, but it is estimated that c. 50% of living bryophytes exhibit gametophyte dioecy (Jesson & Garnock-Jones, 2012), in which the male and female sex organs form on different individuals, and gender is established by sex chromosomes that are segregated during meiosis. The obligate tetrad-producing living hepatics such as Riccia, Sphaerocarpos and Cryptothallus are dioecious. Dispersed tetrads therefore contain two male and two female spores, facilitating sexual reproduction through selfing. In dyad-producing cryptophytes, segregation of sex chromosomes during Meiosis I would result in dispersal units of the same sex. Therefore, from the perspective of sexual reproduction, dyads and monads are functionally equivalent, whereas tetrads have the advantage of transporting both sexes to the same place. However, if the cryptophytes were gametophyte co-sexual, they would have been hermaphroditic, and therefore all cryptospore forms were functionally equivalent in terms of sexual reproduction. So, different combinations of spore type and gender expression may have been advantageous to founder populations. For plants with gametophyte dioecy, the obligate tetrad would have enabled intergametophytic selfing (autogamy). For plants that were gametophyte co-sexual, cryptospore geometry is irrelevant because all spores produce hermaphrodites leading to intragametophytic selfing (automixis). It has been argued that selfing in founder populations might explain the evolutionary stasis of the cryptospores themselves (i.e. their conservative form) throughout the Ordovician (Wellman & Gray, 2000; Steemans et al., 2012), because autogamy and automixis lead to increasingly homozygous individuals and inbreeding depression. A further consequence of cryptospore geometry relates to germination. In trilete monads, the trilete feature is a weakness in the proximal face of the cell wall that facilitates spore germination. No such equivalent feature exists in obligate dyads and tetrads. The proximal faces are fused, so germination can only occur through the distal faces. Emergence of a germination tube would have been further hampered by the presence of the outer envelope. Indeed, Steemans et al. (2012) considered that this double hurdle in germination was responsible for near disappearance of dyads and tetrads during the early Silurian, whereas this period was characterized by the proliferation of trilete monads.

The demise of the cryptophytes is still poorly understood in causal terms. Cryptospores were produced well into Devonian times, and occasionally palynomorph assemblages document the proliferation of tetrads (although the majority of these probably derived from tracheophytes), a phenomenon that has been linked to times of environmental stress (e.g. Lochkovian, Lavender & Wellman, 2002). Whereas the number of permanent dyads declined, the diversity of hilate monads, particularly smooth forms, was sometimes almost comparable to the diversity of trilete spores. This persistence of cryptospores for over 60 million yr and their co-occurrence with trilete monads for at least half of this period indicate that the causes of their decline had little to do with their reproductive biology. Ecological changes during the Early Devonian may provide a more fruitful line of investigation (Edwards & Davies, 1990). During this period, vascular plants were diversifying on land, initiating a rapid increase in plant size that led to the development of the first forest ecosystems by the Middle Devonian (Meyer-Berthaud et al., 2010; Stein et al., 2012). The cryptophytes were very small organisms, and they would have fared poorly in competition with much larger vascular plants for light and other resources. The development of a land flora dominated by vascular plants represents a major environmental change with significant consequences for life on land and for key Earth Systems (Kenrick et al., 2012). The demise of the cryptophytes may have been just one of several tangible consequences of the rise of the vascular plants.

6 Spore wall formation and the nature of the envelope

Fig. 2 summarizes the variation in the complexity of ultrastructure in Ordovician to Lochkovian cryptospores. Regardless of this disparity, which we must emphasize is based on mature dispersed and in situ spores, of the four types of wall development summarized by Blackmore & Barnes (1987), we favour sporopollenin deposition on white line centred lamellae, as seen in almost all investigated bryophytes (Brown & Lemmon, 1990; but see Brown & Lemmon, 1984 for Andreaea) and basal tracheophytes (Tryon & Lugardon, 1991), as the pleisomorphic condition, but evidence of such organization can be obliterated by further sporopollenin deposition. However, the homology of the various layers, including the envelope (sometimes called a membrane), merits further discussion. In some spores, the envelope is a discrete layer, which may surround tetrads, dyads or monads, is adherent to or separate to varying extents from the exospore and, apart from folds, lacks ornament. Its inner surface is usually smooth. Although its resilience to decay suggests a similar composition to sporopollenin, its chemistry is unknown and in some specimens (e.g. Acontotetras within a fragmentary discoidal sporangium NMW2012.17G.1; Fig. 6l), there is a differential response to staining from the spore wall. It usually has a homogeneous ultrastructure. Exceptions include the envelopes of Velatitetras and Abditusdyadus/Segestrespora, which are significantly thicker and may not be homogeneous, for example in specimen V.68196(2) (Fig. 7p). The outer surfaces may show folding, often interpreted as an ornament comprising regular to irregular reticulate muri (e.g. Velatitetras cristata, Velatitetras retimembrana and Abditusdyadus histosus). They also show significant separation from the exospore and internal structures such as voids (Fig. 2g ii), or inward projections (Fig. 2g iii). Edwards et al. (2012a) illustrate sections through the C. horridus complex that suggest that the outer sculptured layer with complex ultrastructure, which is usually very closely adherent to the exospore, shows signs of separation and hence is possibly homologous to the thick envelopes of Segestrespora and Velatitetras, possibly with a similar developmental pathway. This would account for the vacuolated layer and inwardly directed projections within certain specimens of Velatitetras sp. (Edwards et al., 2012b).

Although common in early assemblages, enveloped spores decrease in number abruptly near the Aeronian/Telychian boundary, around the time when trilete spores were increasing in numbers (Steemans et al., 2012). Nevertheless, they persist into the Lochkovian, including our own discoveries in the middle subzone of the micrornatus – newportensis (MN) Sporomorph Assemblage Biozone, as well as new species of Abditusdyadus and Velatitetras (Nodospora) retimembrana Wellman & Richardson (1996) in the lower subzone of the MN Biozone.

For the envelope, a number of authors have postulated its origin via a secretory tapetum to form a kind of perispore, although this layer surrounds individual monads in extant mosses and ferns, and a tapetum is not found in liverworts (Gray, 1985; Johnson, 1985), or the remains of the spore mother cell wall (synoecosporal wall; Taylor & Strother, 2008). Here we adopt a more parsimonious approach and suggest that it was produced, as was the exospore wall, immediately outside the plasmalemma, in the case of the tetrad possibly before any cytokinesis, and in that of the permanent dyads, following the separation of the two cells resulting from cytokinesis after the first reduction division (Meiosis 1). This would be followed by further plasmalemma secretions resulting in a usually layered exospore with variation in appearance at maturity depending on the degree of obliteration of original ultrastructure by sporopollenin. This wall would have completely enclosed the developing tetrad or dyad, but following cytokinesis that produced four or two sets of two cells, further plasmalemma activity would have produced a wall around each spore. The trilete and hilate monads with a bi-layered exospore lack both envelope and an enclosing common outer exospore, but would via plasmalemma activity produce first an outer, and then an inner layer around both distal and proximal surfaces after the separation into monads.

These postulated developmental pathways will be discussed in light of the hypotheses of previous workers who suggested inter alia that the envelope is the equivalent of the spore mother cell wall or synoecosporal wall (Taylor & Strother, 2008), or that a polysaccharide special cell wall was secreted by the plasmalemma after the products of the first meiotic division in D. murusdensa (Taylor, 1996), and that detailed comparisons with processes described are from a relatively small number of extant basal embryophytes (Blackmore & Crane, 1988; Villarreal & Renzaglia, 2006; Brown & Lemmon, 2011; Wallace et al., 2011). The traces of an amorphous outer layer in C. horridus (Fig. 3e) (Edwards et al., 2012a) and D. murusattenuata type I (Taylor, 1997) might represent the remnants of the special cell wall.

Returning to the possibility of tapetal activity, although here we discount the possibility that it was responsible for an envelope, our earlier studies on retusoid spores (Retusotriletes) associated with the basal lycophytes (Resilitheca salopensis Edwards et al., 1995c; Sporathylacium salopense Edwards et al., 2001) demonstrated granular material adhering to the spore surfaces and the innermost surface of the sporangium wall. We interpreted these structures as tapetal residues with possible homologies to the Ubisch bodies in pollen grains (Pacini, 1990) and globules and spherules in extant homosporous lycophytes (Tryon & Lugardon, 1991). In our studies on cryptospores we have found similar structures on both sites in Ficoiditheca aenigma (Fig. 9g,i,j,k), a permanent dyad-producing plant (Fig. 9l) from the Lower Devonian locality, with stomatiferous bifurcating axes terminating in sporangia, which, unlike the valvate forms, dehisced distally into a number of small valves (Fig. 9h) (Morris et al., 2012a). The dyad exospore comprised three homogeneous layers with irregular clusters of granules incorporated in the outer layer (Fig. 9m,n). This combination of characters is unique and probably representative of a further distinct clade with grades of organization similar to those in Partitatheca, except for the spore ultrastructure and the putative tapetal activity. Similar structures have also been noted on the surfaces of enveloped tetrads in a Lower Devonian specimen, NMW96.11G.3 (Fig. 7g,h) (Edwards et al., 1999) and, although initially interpreted as an irregular microgranulate sculpture, similar structures were also observed over the inner surfaces of the sporangium wall in association with the large extra-exosporal fragments, thus suggestive of tapetal residues. An earlier, less convincing example is from the Přídolí at Ludford Lane, where spores with adhering clusters of granules (Fig. 9p) occur in an extensively elongate, sporangial fragment that is incomplete at both ends and partially surrounded by a smooth acellular cuticular layer (Fig. 9o). The hilate monads assigned to Laevolancis divellomedia possessed a layered wall, the inner lamellate surrounded by a thicker homogeneous layer with a further very thin layer present over the distal and equatorial region and bearing the granules (Fig. 9q,r). Such granules are not present on the proximal faces as seen in Retusotriletes, and thus deposition in this case must have preceded separation of the dyad. Therefore, these granules are less likely to represent tapetal material in the conventional sense. Nevertheless, this again is a new combination of characters in cryptospore-bearing plants.

7 Conclusions

  1. Our recent studies of exceptionally well-preserved mesofossils from the Silurian–Devonian (Přídolí–Lochkovian) of the Welsh Borderland are beginning to reveal the source plants of many of the spores found dispersed in sediments of that age. We show that the enigmatic cryptospores were produced by minute plants that possessed a combination of features not found together in living species. The fossils are highly fragmentary, so many aspects of their overall morphology, biology and affinity still remain unclear. We propose to term these plants ‘cryptophytes’ in order to recognize a level of organization of basal embryophytes that does not fit within crown group liverworts, hornworts, mosses and vascular plants, and to provide a focus for further research and discussion. Cryptophytes comprise a grade of organization of plants producing cryptospores in sporangia, some of which might belong to the embryophyte stem group (Steemans et al., 2009), whereas others are more closely related to one or more embryophyte crown groups (Fig. 1). It is important to emphasize that cryptophytes are not bryophytes as is frequently implied in the literature (Wellman et al., 2003; Duckett et al., 2006). Such generalizations tend to be taken up unreservedly by the wider botanical community and are highly misleading. For example, Budke et al. (2012), in studying the cuticularized calyptra of Funaria hygrometrica, concluded that the ‘role of the calyptra in desiccation prevention offers a functional explanation for calyptra retention during the 450 Myr of moss evolution…’. There is no evidence of calyptra in the cryptophytes or in any known fossil of the Palaeozoic era. The newly emerging fossil evidence is more in line with the original interpretation envisaged in the pioneering work of Gray & Boucot (1977), who considered cryptospores to have been the products of plants that were proto-hepatics or proto-bryophytes. The cryptospore-producing plants may be bryophyte-like in their size and in some aspects of their morphology (Richardson, 1992; Tomescu & Rothwell, 2006; Steemans et al., 2009; Kenrick et al., 2012), but much still remains to be learned about these as whole organisms, and little is yet known of their biology. Despite the fragmentary nature of the cryptophyte fossils, some distinctive groupings are beginning to emerge (Fig. 1). Partitatheca includes plants with bifurcating axes terminating in valvate sporangia bearing stomata. Spores are dyads with a laminated wall structure belonging to the dispersed cryptospore genus Cymbohilates. Lenticulatheca encompasses discoidal sporangia containing monads that are clearly derived from dyads that are also assignable to Cymbohilates. In addition, detailed comparative studies of cryptospores indicate that several features of the Dyadospora murusattennuata complex (i.e. absence of the outer dyad envelope; a bi-layered wall) might link this group to the vascular plants. Further descriptive work on these mesofossil assemblages, and in particular the application of new, noninvasive imaging methods (i.e. synchrotron microtomography) (Friis et al., 2007), are essential to developing our understanding of the cryptophytes and their relationships.
  2. Research on the Anglo-Welsh Basin provides a detailed regional picture of the evolution of plant life on land. These sediments are the best in the world for a more or less uninterrupted sequence through the Upper Silurian (Přídolí) to Lower Devonian (Pragian), with an excellent biostratigraphical framework. Although direct fossil evidence of the cryptophytes comes from only two sites, there is an extensive and well-documented record of dispersed spores for the region. Major differences have been noted in the palynological assemblages between the Raglan Marl Formation (Přídolí to lowermost Lochkovian) and the Senni Formations (upper Lochkovian to Pragian) (Richardson & Lister, 1969; Hassan, 1982; Richardson et al., 1982). Spores derived from larger vascular plants, mostly zosterophylls (i.e. basal lycophytes), diversified, and this was accompanied by a major decline of the cryptospores (Edwards & Richardson, 2004). This pattern was interpreted in ecological terms as one of local competitive displacement of cryptophytes and rhyniophytes by the larger zosterophylls (Edwards & Davies, 1990).
  3. The patterns emerging from the Anglo-Welsh Basin provide insights into events on a regional scale towards the end of cryptospore evolution. A global perspective over the same and a more extended time interval is reliant solely on the record of dispersed palynomorphs. As there is no corroborating evidence of in situ cryptospores elsewhere, this evidence should be interpreted cautiously. First, sampling of the dispersed palynomorph assemblages is biased towards sites in Laurussia, whereas sampling from Gondwana is particularly sparse (Steemans et al., 2010). Secondly, records of cryptospores predating the mid-Ordovician are controversial (e.g. Strother et al., 2004) because, although generally similar in form and with resistant walls, they lack the regular, well-defined arrangements of later tetrads and dyads. Thirdly, although some early dispersed spore types have been identified in situ in cryptophytes and vascular plants (e.g. Cymbohilates, Streelispora, Aneurospora, Velatitetras and Cheilotetras), many have not. Fourthly, the newly emerging fossil evidence on the morphology of the cryptophytes documented here comes from latest Silurian to earliest Devonian rocks. The key morphological features exhibited by these fossils (e.g. sporangia; and in some sporangial dehiscence, branched sporophytes, and stomata) would have probably evolved sequentially on land and may not all therefore have been present in the cryptophytes of the Ordovician period. Bearing these caveats in mind, spores attributable unequivocally to cryptophytes first appeared in the Middle Ordovician (Dapingian Stage; c. 470 Ma) on the Gondwana Plate, and subsequently in Avalonia, Laurentia, Baltica and finally South America through the Late Ordovician to Early Silurian (Steemans et al., 2010). Diversity was greatest in the Late Ordovician and the earliest part of the Silurian (basal Llandovery), and it decreased through the Silurian. The cryptospores almost disappeared during the Lower Devonian, but plants from within the cryptophyte grade gave rise to the modern lineages of land plants. The demise of the cryptophytes is poorly understood in causal terms, but we suggest that this was related to major ecological changes brought about by the diversification and rapid increase in size of the vascular plants during the Early Devonian leading eventually to the evolution of the first forest ecosystems.
  4. The fossil record of cryptospores testifies to a far greater versatility in meiosis among basal embryophytes than is evident in living plants. This diversity can be explained by the developmental process of meiotic cell division coupled with changes in the synthesis of sporopollenin and the timing of its deposition. Cryptospore geometry exhibits some traits that are characteristic of founder populations, including small size in early forms. Although gametophyte gender expression in cryptophytes is unknown, the dispersal unit of the permanent tetrad always ensures the possibility of sexual reproduction through either autogamy (gametophyte dioecious: intergametophytic selfing) or automixis (gametophyte co-sexual: intragametophytic selfing), whereas this is possible in dyads and monads only when gametophytes are co-sexual. Where gametophytes are dioecious, dyads and monads are functionally equivalent dispersal units, and this combination of gender expression and cryptospore form would not favour population establishment over long distances. Dyads are almost as abundant as tetrads in early palynomorph assemblages, and our finding of dyads in situ in cryptophyte sporangia proves that this cryptospore type derives from land plants and that it was part of their normal meiotic repertoire. This finding challenges the long-held consensus that the tetrad configuration, plus the production of four haploid spores, was the archetypical condition in the pioneering land plants. The possibility that dyads are the plesiomorphic state in land plants should be given serious consideration.

Acknowledgements

This work was funded by a Leverhulme Trust Research Grant, the Gatsby Charitable Foundation and a number of NERC research grants and studentships, which are all gratefully acknowledged. Throughout our researches in Cardiff the technical assistance of Lindsey Axe has been indispensable. We thank John Williams for the use of The John Williams Index of Palaeopalynology at the Department of Earth Sciences, Natural History Museum, London, UK.

Ancillary