Electron microscopic studies of the shell fabric of articulate brachiopods show that the triple division of the shell into periostracum, and primary and secondary calcareous layers, is characteristic of most members of the subphylum. The relationship between these layers and the outer epithelium of the mantle controlling their secretion can be studied in living Rhynchonellida and Terebratulida. Each fibre of the secondary layer is coated with protein and secreted by an outer epithelial cell which is moulded to the characteristic shape and stacking of the fibre. The morphology of the exposed parts of fibres can, therefore, be used to identify modifications of the epithelium, like those arising from the spread of muscle bases, and to determine the nature of the outer epithelium in extinct groups. The attitudes of fibres can also be used as vectors to study the growth dynamics of the shell and its internal skeletal features.
The triple-layered shell with orthodoxly stacked fibres was the basic skeletal type of ancestral Rhynchonellida and Terebratulida, and of the Spiriferida, Pentamerida, and Orthida. Modifications include the development of a tertiary calcareous layer in the Koninckinacea (through inhibition of the secretion of protein sheets and the coalescence of fibres across intercellular boundaries to form a coarsely prismatic shell) and the suppression of the secondary calcareous layer in the Thecideidina. The most profound changes affected the Strophomenida. Among the Plectambonitacea, the secondary calcareous layer was made up of orthodoxly stacked fibres, but the primary calcareous layer consisted of lath-like laminae, which were the sole constituents of the shell of all other Strophomenida (including the Triplesiidina). Regular banding of the strophomenide shell, resulting from laminar deposition, with a periodicity of about 0.3, is comparable with that sporadically found in the primary and secondary calcareous layers of living articulates, and is believed to register diurnal variation in secretion.
BRACHIOPODS constitute one of the most intriguing and challenging phyla that can be studied by the palaeontologist. Their recorded history extends back to the early Cambrian, since when they have emerged as two distinct groups, the Inarticulata and the Articulata, with an unrivalled record in marine facies throughout time and space. A wealth of information can therefore be obtained about the course of brachiopod evolution over the last 600 million years especially among the articulates, because their skeletal remains include variously derived outgrowths concerned with the articulation of the shell and the support of muscle and lophophore systems, while the internal surfaces of the valves commonly bear impressions of soft parts even on an ultramicroscopic scale. Unfortunately the opportunities afforded by interrelating such data have not been fully exploited. All too often palaeontologists forget that fossils were once organic systems and that the science requires them to be concerned with the dynamics as well as the relics of past life. The neglect is most obvious in discussions of morphological variation. All changes in form are expressions of differences in growth, and since the mechanics of growth are actually registered in the fabric of the articulate shell, accounts of skeletal features which ignore the processes that led to their development are at least incomplete and may be gravely misleading.
The study of shell structure, then, is a prerequisite to an understanding of changes in skeletal morphology that have taken place during the evolution of articulate brachiopods. The most profitable techniques used in such study are encouragingly simple. They involve the preparation of replicas of the internal surfaces as well as sections of shells (Williams in Williams et al. 1965, p. H253), and their examination under the light microscope supplemented by the electron microscope. Indeed, greater difficulties arise in choosing appropriate material. The surfaces of most fossil specimens, even those with well-preserved shell fabric, are covered by thin skins of calcite crystallites which obscure the original microscopic texture. However, when the skin is a superficial accretion and not a replacement of the outer layers of the shell, it can usually be removed in ultrasonically vibrated detergent fluids. In this manner surface patterns of Lower Palaeozoic species can be made available for scrutiny and comparison with those of Recent shells. Diagenesis and lithification can also affect the shell fabric and might lead to the acceptance of observed patterns, brought about by these processes in extinct groups, as reflecting the state of the living shell. Fortunately, terebratulides and rhynchonellides are not only represented in modern seas but have a long history extending back to the Devonian and Ordovician respectively. It has, therefore, been possible to use fossils belonging to these two orders as standards in estimating the degree of alteration affecting specimens of extinct orders that have been recovered from the same horizon and locality. Such checks have been used for every order found in Palaeozoic and early Mesozoic rocks, and the results are so consistent for all stocks investigated that important deviations are not expected to come to light during more detailed systematic investigations.
The use of living material as a standard in the interpretation of the shell structure of extinct stocks has determined the sequence of discussion in the following account. Proceeding from the rhynchonellides, which provide the simplest model of shell deposition, there follows a comparative study of the fabric characteristic of Recent and past terebratulides. The spiriferides, which became extinct in the Lower Jurassic, are considered next; while the pentamerides are discussed before the orthides because, although they died out much earlier (Devonian compared with Permian) , they show, on balance, closer affinities with the first three orders. Despite their alleged late survival into the Jurassic, the strophomenides are the last ordinal group to receive attention, an acknowledgement of the uniqueness of their typical shell structure. Finally, some reference is made to problematic groups, and to the bearing that the collation of observations on shell structure has on brachiopod evolution.
The growth of the mantle and the concomitant deposition of the triple-layered shell of articulate brachiopods has already been described (Williams 1953, 1966) but is worth a brief review in relationship to Hemithiris psittacea (Gmelin) and Notosaria nigricans (Sowerby) to give current ideas their full perspective.
The principal generative zone responsible for the enlargement of the mantle lining both valves is located at the closure of a groove separating the inner and outer lobes of the mantle edge. Both lobes act like conveyor belts in that new cells added to them at their junction with the groove closure ultimately become part of the inner and outer epithelium (text-fig. 1). In effect the groove and lobes are maintained as morphological features of the mantle because the creation of cells in the circumferential meristomatic region balances the incorporation of those previously proliferated within the main epithelial spreads posterior to the lobes. Thus as outer epithelial cells make way for new ones arising behind them in the groove, they 'move' anteriorly along the inner face of the outer lobe and begin to secrete the periostracum as a continuous, exclusively protein cover (Jope 1967) to the outer plasma membranes. When the cells reach the tip of the outer lobe they start secreting a primary layer1 of calcite crystallites between the periostracum and the underlying plasma membranes. The first stages in calcite nucleation have not yet been observed but they can be deduced from the topography of the external shell surface of Notosaria (PI. 2, figs. 1, 2). This surface is divided by a series of long radiating grooves into strips, each up to 12 µ wide but tending to encroach on one another here and there. Each strip is further divided by grooves that are normally convex anteriorly and about 2 µ apart. Both sets of grooves accommodate the lists and bars of differentially thickened periostracum (Williams 1956, p. 244); and those that radiate, together with every fourth or fifth arcuate one, probably correspond to the lateral and transverse boundaries respectively of rows of cuboidal epithelium. The calcareous surface between a pair of arcuate grooves is usually asymmetrical in profile, culminating anteriorly in a series of rhombohedral peaks about 0.3 µ apart and gently sloping posteriorly towards the culmination immediately behind (PI. 2, fig. 3). The 'contoured' effect, which is so prevalent within these culminations and may even be traced continuously throughout several 9f them, represents a periodicity in the deposition of primary shell. Such layering indicates that the peaks constitute the first-formed rhombohedral seeds of calcite which are secreted between the plasma membranes and the periostracum, and that initially the seeds tend to be concentrated in zones separated by inwardly directed bars of periostracum and isolated from one another by membranous projections (microvilli) attached to the periostracum. As deposition continues, the crystallites grow and overlap one another across intercellular boundaries (text-fig. 2); but, although the plasma membranes become separated from the periostracum by an increasing wedge of calcite, microvilli which may be aligned along rhombohedral angles (PI. 2, fig. 4) continue to permeate the primary layer to give its inner surface a highly porous appearance (PI. 2, figs. 5, 6). Microvillous strands or sporadically exuded trails may even be pinched off from the plasma membranes by growing crystallites (PI. 3, fig. 1), which would account for traces of protein reported in the primary layer, although, in general, that layer may be regarded as the inorganic constituent of the articulate shell. The growth of the rhynchonellide primary layer also indicates that the microvilli are really temporary extensions of the plasma membranes that may come into being or be eliminated by local changes in the rate of calcite deposition. Yet at any given moment, so many microvilli are present (up to 25 per µ2) that the cells are anchored relative to one another on the inner surface of the primary shell.
Text-Fig 1. Stylized longitudinal section of the edge of a valve of Notosaria, showing the origin of outer epithelium and its relationship to primary and secondary shell
Continuous addition of new cells to the tip of the outer lobes, and secretion of calcite by all epithelial cells underlying the primary layer, cause the extension of that layer anteriorly and its thickening posteriorly. However, the deposition of the primary layer does not proceed indefinitely, because at a certain distance from the shell edge the epithelial cells undergo a profound physiological change and start secreting the secondary layer. The slight increase in thickness of the primary layer towards the shell margins of adult Hemithiris suggests that the change-over in cell activity may become delayed as shells grow older, as it certainly is in embayments corresponding to the sites of setal follicles and therefore to the crests of ribs in radially ornamented species (Williams and Wright 1963, p. 19). Nevertheless it can be held as a generalization that as rows of cells come to occupy a given position posterior to the mantle edge, due to the forward extension of that edge by the addition of new cells, they invariably start secreting secondary shell instead of primary: in Hemithiris, for example, the 20th row, located about 1 mm. from the shell edge, is so affected. The regularity with which this transformation in secretory habit occurs explains why the primary shell is a thin layer showing less dramatic changes in thickness than the underlying secondary shell.
The secretory functions of outer epithelial cells controlling secondary shell growth are so much more complex than those of the same cells when they were depositing the periostracum and primary shell that some consideration must be given to the cells themselves before discussing the development of the secondary layer. This layer owes its distinctive appearance to the presence of protein sheaths that segregate the calcite into discrete units (fibres) (text-fig. 2). Both substances are secreted simultaneously on different parts of the outer plasma membrane of each cell (text-fig. 3; PI. 1, figs. 1, 2). Protein is exuded by the anterior part of the plasma membrane and is temporarily attached to an arcuate zone of the membrane by closely spaced terminal pivots of tonofibrils (desmosomes) which are densely distributed in this section of a cell (PI. 1, fig. 3). The exuded protein separates posteriorly from the arcuate attachment zone as a curved sheet with only rare lacunae. Beneath this sheet, calcite is secreted by the posterior zone of the plasma membrane. This two-fold association of protein sheet overlying a calcite fibre is maintained by every epithelial cell, and since the cells are roughly rhombohedral in outline and arranged in alternating rows relative to the shell edge, a regular skeletal pattern results. Thus the terminal face of each fibre is contained in an inwardly convex patch of cell membrane situated behind a semicircle of the same membrane which exudes the outer protein cover of the fibre (text-fig. 4). The inner protein covers which complete the continuous organic sheath of the fibre are secreted by adjacent anterior arcs of the three cells immediately behind. Consequently the inner surface of the secondary shell characteristically shows the terminal faces of constituent fibres as alternating rows of convexly lobate pads, each semicircular in outline anteriorly but tapering posteriorly towards the rc-entrant angle subtended between the curved sides of two flanking fibres belonging to the overlapping row behind (PI. 4, figs. 1., 3, 5). This pattern is referred to as the secondary shell mosaic. It is important in that it represents a protein-calcite cast of the outer epithelium in the manner illustrated in text-figs. 3 and 4. It can therefore be used to interpret the nature of the outer epithelium in fossil shells, and since it is the sole guide to shell growth in extinct stocks, it seems appropriate, henceforth, to describe the origin and nature of the secondary layer in terms of the mosaic.
Text-Fig. 2. Plan of the internal shell surface at the edge of a valve of Notosaria, showing therelationship between the three layers of the shell
Text-Fig. 3. Stylized longitudinal (a) and transverse (b) sections of an outer epithelial cell of Terebratulina secreting a secondary fibre and its protein sheath
The first stages in the development of the mosaic can be seen in any well-preserved shell. Usually, the earliest indication that a cell is changing to secondary processes of secretion is the laying down of secondary calcite in the re-entrant angle between two fibres deposited by immediately older cells (PI. 2, figs. 5, 6). The secondary calcite is initially, at least, in crystaUographic continuity with that of the primary layer but is visually distinguishable as an evenly distributed plaster spreading out over the more porous primary layer. The smoother texture of secondary calcite is due to a sudden reduction in the number of microvillous extensions from the secreting membrane. Other isolated deposits of secondary calcite may appear anywhere within the limits imposed by a cell and soon become continuous with one another. But coalescence is rarely corn· pleted before a curved strip of protein is exuded by the anterior part of a cell and then extended posteriorly to unite with the protein boundaries of adjacent fibres in the row behind. In this manner the base of a fibre remains in physical continuity with the primary layer, yet simultaneously acquires the beginnings of a protein sheath that thereafter maintains the fibre in isolation from all others.
Text-Fig. 4. Plan of the secondary shell mosaic on the floor of a valve of Terebratulina, showing the relationship between exposed fibres and outer epithelial cells
Any attempt to trace the evolution of the shell structure of articulate brachiopods is necessarily influenced by the demonstrable antiquity of the triple-layered shell of living forms. Admittedly the existence of a periostracum in fossil material cannot be proved, but its presence in all surviving brachiopods, inarticulate as well as articulate, indicates that it has always been exuded as the shell cover. [Whether the periostracum in inarticulates is composed of chitin or protein does not affect its morphological relationship with the rest of the shelL] The chief constituents of the calcareous part of the shell, on the other hand, are more or less identifiable throughout time. A comparative study of them shows that some important changes in the secretory regime did occur in the course of brachiopod evolution, although they were largely irrelevant to the final outcome.
In the great majority of articulates the primary and secondary layers have always been sharply differentiated by the development within the latter of a system of interconnected protein sheaths segregating simultaneously deposited carbonate into distinctively shaped and stacked fibres. As has been shown, this arrangement can only be brought about by a 'conveyor belt' system of mantle expansion and uniquely phased changes in the secretory regime of each outer epithelial cell; and since it is characteristic of early orthides and pentamerides, the primitive mantle must have been identical in growth, morphology, and function with that of living rhynchonellides.
Speculation about the origin of the primitive mantle necessitates consideration of the inarticulates, which probably evolved from the same prototype as the articulates. The growth of the inarticulate shell is not well documented but, like the articulate skeleton, it probably always involved the secretion of a periostracum and the enlargement of the mantle by the 'conveyor belt' system (protoplasmic anchors, like strands and caeca permeating the shells of lingulides and craniaceans respectively, preclude other means of growth). It is therefore possible that the mantle of the brachiopod prototype with its peripheral generative zone was initially protected by periostracum alone, and that the secretion of calcium carbonate (or phosphate) beneath that layer represented a later stage in evolution. The preservable inarticulate shell, whether chitino-phosphatic or calcareous, seems always to have been built up by outer epithelium that was capable of secreting alternating mineral and organic layers across intercellular boundaries. This type of shell is likely to have been deposited by the precursor to the primitive articulate mantle. If this were so, the development of the articulate secondary shell would have represented a significant evolutionary step, whereby the plasma membrane of each outer epithelial cell became specialized to secrete both organic and inorganic skeletal material simultaneously rather than alternately. The differentiation of the primary shell is less easily explained. As a mineral layer interpolated between the periostracum and the secondary shell, it can be interpreted as the earliest organized calcareous skeleton to appear during the evolution of the prototype. If this were true, the secretory sequence followed by the outer epithelial cells during ontogeny is an example of physiological recapitulation. But a persistent, well differentiated, wholly inorganic primary layer is unknown among inarticulates (a homologous layer found in the craniaceans, which probably arose after the divergence of that group, cannot be so described). Even among articulates there is evidence to suggest that the layer was less well developed in orthides. Hence the secretion of a distinctive primary shell may have been a comparatively late innovation in the evolution of the articulate shell, and the calcareous skeleton of the earliest stocks may have consisted solely of fibres ensheathed in protein, but crudely fashioned as trigonal prisms with scalenohedral terminal faces and attached directly to the periostracum.
Whatever the first stages in the evolution of the articulate shell were, subsequent changes are obvious enough (text-fig. 27). The orthides, pentamerides, rhynchonellides, spiriferides, and terebratulides form a continuous chain of descent that led to the survival of the primitive mantle in living species and the growth of a remarkably stable triplelayered shell in the great majority of articulate brachiopods. However, at the cellular level, variation in the secretory behaviour of the outer epithelium gave rise to subtle topographic changes in the terminal faces of secondary fibres. The most important of these changes was initially connected with the development of dense concentrations of tonofibrils in the outer epithelium and associated connective tissue. It led to the breakdown of the characteristic arrangement of the secondary fibres by an interruption in the secretion of protein and the coalescence of terminal faces, normally transformed into simple crystal growths, across intercellular boundaries. In orthides, rhynchonellides, and terebratulides, this disruption of secondary shell appears to have been limited to those parts of the shell Oike muscle scars, cardinal processes, dental plates, etc.) that acted as attachment areas for muscle systems. But in pentamerides and spiriferides interruption in the exudation of protein was commonly precursory to the secretion of relatively thick, coarsely crystalline deposits of calcite spreading well beyond the limits of muscle fields. Most of these prismatic deposits were impersistent and accumulated during temporary changes in the secretory regime of cells that controlled the growth of the secondary layer and can only be regarded as variants of that layer. However, in the koninckinaceans (as represented by Koninckina and Cadomella) prismatic deposits constitute a true tertiary layer occurring inwardly of the secondary. The well-defined distribution of this layer makes it likely that it arose through a regular change in the secretory regime of outer epithelial cells as they came to lie at a given distance from the expanding shell margin. In effect, each mature cell was in turn responsible for the secretion of parts of the periostracum, primary, secondary, and tertiary layers; and in some respects the deposition of the tertiary layer can be regarded as a reversion to the biochemical conditions obtaining during secretion of the primary shell. A similar change may also have taken place in some other spiriferides and in some pentameraceans like Gypidula and Sieberella, although the absence of any traces of secondary fabric within the thick internal deposits of prismatic calcite found in such stocks has still to be proved.
Text-Fig 27. Phylogenetic chart of the articulate brachiopods (excluding the Dictyonellidina), showing the distribution of different types of calcareous shell
Despite the novelty of a third calcareous layer, koninckinaceans and other stocks with a similarly differentiated shell fabric are still an integral part of the main articulate plexus because the extra layer appeared as a gerontomorphic feature which did not upset the standard secretory regime. In fact, as far as is known, only three groups, the Strophomenida, Thecideidina, and Dictyonellidina, are sufficiently different in shell structure to require interpretation based on the assumption that radical changes in the secretory activities of their mantles took place.
The derivation of strophomenides is most easily understood. The plectambonitaceans have always been regarded as primitive strophomenides and current understanding of their general morphology and stratigraphic distribution conclusively reaffirms the traditional view. Yet the plectambonitacean shell structure is basically like that of the standard articulate and differs only in the laminar nature of the primary layer. The difference is extremely important for two reasons. First, it represents a fundamental change in the mode of secretion by outer epithelial cells at the mantle edge, although they reverted to the standard regime in the deposition of secondary shell. Secondly, the plectambonitacean primary layer can be homologized with the entire shell of all other strophomenides. Thus a new type of shell structure emerged through two distinct evolutionary changes. The earlier one involved a caenogenetic modification of the activities of the outer mantle lobe of primitive orthides which gave rise to the first plectambonitaceans. Later, certain plectambonitaceans were affected by neoteny, which included the suppression of the secretion of secondary shell, and in this condition were ancestral to the strophomenaceans, davidsoniaceans, and probably the triplesiidines. Perhaps the most disconcerting aspect of this interpretation of strophomenide evolution is that impunctate (davidsoniaceans and triplesiidines) as well as pseudopunctate species diverged from a common ancestor which itself descended from a group that was at least potentially pseudopunctate. Yet I am now convinced that the ultrastructure of the shell must take precedence over pseudopunctation in gauging relationships between groups. In the past too much importance has been attached to this particular condition of the shell despite evidence of its absence in certain strophomenides like Ukoa, early Christiania and pre-Devonian davidsoniaceans. Pseudopunctae with or without taleolae are nothing more than tubercles which occur in orders other than the Strophomenida and, in my experience at least, the alleged uniqueness of pseudopunctae as seen under the light microscope was really an acknowledgement of the unusual properties of the shell fabric containing them.
Less assurance can be summoned in debating the origin of the thecideidine shell because, in contrast to the differentiation of the standard fabric, the shell, like that of most strophomenides, consists of one type of deposit. Even so the ultrastructure is not laminar but much more like the fabric of the primary layer in other living articulates, and it is probable that the thecideidine calcareous shell is single layered like the strophomenide simply because they were both expressions of neotenous descent. If this were so, and the thecideidine shell is a homologue of the orthodox primary layer, other considerations are more likely to favour an ancestry among the spiriferides than the terebratulides.
It is noteworthy that the thecideidines are endopunctate as well as tuberculate, and although the former condition is probably more relevant to problems of ancestry than the latter, neither is particularly helpful. No morphological differences have yet been found between canals identified as endopunctae in enteletaceans, the rhynchonellide Rhynchopora, certain spiriferides, and terebratulides. Clearly endopunctation is not an index of interrelated descent for these several independent groups, and to use the presence of caeca in thecideidines as a guide to their origin is to deny the possibility that endopunctation developed repeatedly during brachiopod evolution.
Little can be said at present about the relationship between the last group, Dictyonellidina, and other brachiopods. The shell fabric suggests that the dictyonellidine shell was deposited in the same way as the primary layer of living rhynchonellides. Hence it conceivably represents a homologue of that layer, as developed in primitive articulates, which became the sole type of deposit making up the dictyonellidine shell through a neotenous failure of mature outer epithelium to secrete protein sheaths. Alternatively, the dictyonellidines are remnants of a group that arose from the brachiopod prototype independently of both articulates and inarticulates.
I am greatly indebted to Mr. R. Reed and the technical staff of the Electron Microscopy Unit of Queen's University, Belfast, for instruction in the preparation and examination of the material discussed in this paper; to Dr. Jean Graham, research assistant in the Department of Geology, for her help in the final drafting of illustrations and the checking of the manuscript; to Professor Gareth Owen of the Department of Zoology, who so willingly gave much needed advice on the interpretation of cell morphology and provided the electron micrographs of Plate 1; and to Dr. A. D. Wright of the Department of Geology for his helpful comments on the manuscript. I am also grateful to Dr. Howard Brunton of the British Museum (Natural History), Dr. G. A. Cooper of the U.S. National Museum, Dr. Harry Mutvei of the Riksmuseum,Stockholm, and Dr. A. J. Rowell of Nottingham University, for so promptly lending me fossil and preserved Recent brachiopods and sanctioning the preparation of sections of this material. Finally, I wish to express my appreciation of a grant from the Queen's University, Belfast, in aid of publication.
Electron micrographs of sections, stained with uranyl acetate and lead citrate, of specimens of Terebratulina caput-serpentis (Linne), originally fixed in formalin and prepared for sectioning by decalcifying, post-osmicating, and embedding in Araldite. Linear scale, at the bottom left-hand corner of each figure, equivalent to 2 µ.
Electron micrographs of single stage negative replicas---cellulose acetate/carbon: shadowed with gold-palladium at 1 in 1. Linear scale, at the bottom left-hand corner of each figure, equivalent to 2 µ.
1 Many palaeontologists continue to use purely descriptive terms for the primary and secondary layers of the articulate brachiopod shell. Following Thomson (1927, pp. 97, 103), outer or lamellar layer and inner, prismatic, fibrous or laminar layer have been used by some students to denote the primary and secondary layers respectively. However, real or apparent differences in the texture of the calcite have led to so arbitrary a use of such terms as to render them meaningless for comparative purposes. The use of 'primary' and 'secondary', on the other hand, enables homologues to be drawn between different groups, because the primary layer (in contrast to secondary deposits) is that which comprises the calcareous shell edge and is normally underlain by the outer lobe of the mantle. Within this context, terms describing the microscopic nature of the calcite making up the shell can be employed unambiguously, although since the fabric may be the same throughout the shell, 'outer' and 'inner' are also inappropriate.
No Supplementary Information available.