Neotrigonia

The palaeoheterodont genus Neotrigonia is the only remaining genus of the Trigonioida, an order which dates back until the early Palaeozoic [38]. Although members of this order were conspicuous and abundant components of shallow marine communities throughout the Mesozoic world-wide, they suffered major losses at the end Cretaceous mass extinction. The only survivors, which might be considered as “living fossils” [39], [40], were confined to the Australasian and southwest Pacific region waters [41], [42]) where today they are represented by eight species (WoRMS 2014, January 30^th), N. margaritacea (Lamarck, 1804), N. lamarckii (Gray, 1838), N. uniophora (Gray in Jukes, 1847), N. strangei (A. Adams, 1854), N. bednalli (Verco, 1907), N. gemma Iredale, 1924, N. kaiyomaruae Habe & Nomoto, 1976 and N. jacksoni Morrison, 2011. However N. bednalli has been questioned as valid species and it is proposed as junior synonym of N. margaritacea [43], [44]. Aside from their interest as ‘living fossils”, they are also phylogenetically interesting because of their position within the Palaeoheterodonta, as the sister-group to one of the major freshwater clades, the Unionoida [45]–[47,]. The Palaeoheterodonta have traditionally been regarded as the sister-group of the Heterodonta, though more recent molecular analyses [47], [48]) now support the relationship between the Palaeoheterodonta and the Archiheterodonta.

All modern and fossil trigoniids have sub-quadrate, trigonal equivalve and relatively thick shells, with radial ribs frequently adorned by tubercles or scales (Fig.1); the valves are articulated with a relatively massive schizodont dentition. In spite of their highly ornamented shells, modern Neotrigonia are moderately rapid burrowers using a large muscular L-shaped foot with a distinctive ‘toe’ and ‘heel’ (Fig. 1) which is also used to execute vigorous leaping as an escape mechanism from predators [39], [49], [50]. All species are subtidal, living in sediments ranging from poorly sorted sand to sandy mud [51].

The shell of Neotrigonia has been described as being wholly aragonitic, with an outer prismatic layer and inner layers of lenticular ( = columnar) and sheet nacre [7], [52]. An interesting feature of the outer surface of the shell is that each prism in the outer layer begins with a ‘boss’ which is embedded within the periostracum, features which are evident even in fossil material dating back to the Palaeozoic [53]. Taylor et al. [7] hypothesized that these bosses represent the initial spherulite “from which each prism has developed, on the inner surface of the periostracum”. Initial observations by scanning electron microscopy of the outer surface of shells revealed that the bosses appear free of periostracum and so may appear to have pierced through the periostracal layer [36] but it is not clear that these protrusions were not initially covered by a pellicle-like outer periostracum, which led these authors to suggest that they might initially grow within the periostracum, thus being intraperiostracal. It is, therefore, interesting to examine whether these calcified bosses are homologous to the calcified spikes reported in anomalodesmatans. Our specific aims were to carry out a study of the growth and structure of the periostracum in the genus Neotrigonia from its very early formation in the periostracal groove to its complete calcification at the shell margins and to investigate the precise relationship between the calcified bosses and the periostracum and the underlying prisms.

Ethics Statement

Ian Loch (The Australian Museum, Sidney) eased the loan of specimens of Neotrigonia listed in Table 1 under repository label AMS. Professor Gonzalo Giribet (Harvard University) and the BIVAToL project supported by the U.S. National Science Foundation (NSF) Assembling the Tree of Life (AToL) program (DEB-0732854/0732903/0732860) provided access to material and gave permission to use the image in Figure 1 (photographed by Professor Gonzalo Giribet). The specimen was collected in Moreton Bay Marine Park under Permit number QS2005/CVL588 from Queensland Parks and Wildlife Service. Dr. Ana Glavinic (School of Biological Sciences, Flinders University, Adelaide, Australia) sampled and provided the glutaraldehyde-fixed specimen of N. margaritacea from Port Stanvac, under the permit that the School of Biological Sciences has for such work (an S115 Ministerial Exemption Number 9002462) issued by the Primary Industries and Regions SA, Government of South Australia. No specific permission was required since the area is unprotected and the species is neither endangered nor protected. The loans and cessions of specimens were made under CITES guidelines, to which all the involved institutions are adhered. Professors Jan Bonarski and Marek Faryna, and Dr Katarzyna Berent (Institute of Metallurgy and Materials Science, Kraków, Poland) gave permission for the use of the crystallographic data shown in Figure 3F–I, acquired jointly with A. Checa.

Material and Techniques

The material studied consisted of a series of in vivo fixed specimens obtained from two sources (see Table 1 for collection details and localities). Most of the specimens were supplied on loan from The Australian Museum (Sidney) but because these were sub-optimally fixed for use in histological studies, we also obtained an individual specimen of Neotrigonia margaritacea from Port Stanvac, Adelaide, South Australia, fixed in 2.5% glutaraldehyde specifically for this study. This was also supplemented by the addition of dry valves from our own collections. The glutaraldehyde fixed specimen of N. margaritacea, as well as the dry shells, is presently housed at the paleontological collection of the University of Granada.

This suite of material has enabled us to undertake a scanning electron microscope (SEM) study of the shell microstructure of the 5 of the 8 extant species of Neotrigonia (Table 1). Microstructural details were assessed from fractures through different shell areas. Fragments were cleaned for 15 s in an ultrasonic bath and allowed to air dry before SEM (field emission SEM Zeiss LEO Gemini 1530 of the University of Granada) observation after carbon coating (Hitachi UHS evaporator). Additionally the mantle margin of one specimen of each of the species N. margaritacea, N. lamarckii and N. gemma were analysed by transmission electron microscopy (TEM) and optical microscopy (OM). Blocks of mantle, with their associated periostraca, were excised mainly from the ventral areas. The specimens were wholly decalcified by immersion in 2% EDTA for several days before post-fixation in OsO₄ (2%) for 2 h at 4°C, and embedded in epoxy resin Aname Epon 812 (EMS). Semithin sections (1 µm) were stained with different concentrations of toluidine at pH 4.5 in order to better check the characteristics of the vesicles present in the epithelial cells of the outer mantle fold. Sections of Neotrigonia margaritacea were stained with: a) 1% toluidine blue in water with 1% Na₂B₄O₇.10H₂O (Merck 6308) at pH 4.5, which gives an intense blue stain; b) 1% toluidine blue in water with 0.2% acetic acid, at pH 4.5; c) 0.5% toluidine blue in water with 0.2% acetic acid, at pH 4.5; d) 0.2% toluidine blue in water with 0.2% acetic acid, at pH 4.5. Sections were observed with an Olympus BX51 optical microscope. Ultrathin sections (∼ 50 nm) were stained with uranyl acetate (1%) followed by lead citrate. They were observed with TEM (Zeiss EM10C and Zeiss Libra 120 Plus at the University of Granada). Pieces of the mantle margin, removed from different regions, of the glutaraldehyde-fixed specimen of N. margaritacea were also examined by SEM, for which they were critical-point dried, sputter-coated with gold and observed with the SEM JEOL JSM-840 of the University of Málaga.

The presence of Ca²⁺ in the epithelial cells of the outer mantle fold was confirmed by applying a fluorescent indicator (Fluo 3-AM, Sigma 46393) to semithin sections of Neotrigonia margaritacea. To obtain the stock solution, 1 mg of Fluo 3-AM was solubilized in 1 ml of a a buffer solution composed of 25% of Pluronic® F-127 (Sigma P2443) in dry dimethyl sulphoxide (DMSO) (Sigma 276855). In order to load the semithin sections, a work solution was prepared with 1 µl of the stock solution in 20 µl of buffer solution. To eliminate possible autofluorescence of the glutaraldehyde and the epoxy resin, the semithin sections were photobleached by exposing them to light for an extended period of time (∼1 h) [54]. For loading with Fluo 3-AM, the semithin sections were incubated for 1 h at room temperature (22–23°C). The fluorescence images were obtained with a laser confocal microscope (Leica DMI6000) with excitation by argon-ion laser source at 488 nm and maximum emission wavelength at 526 nm.

The crystallography of the prismatic units was analyzed by means of Electron Backscatter Diffraction (EBSD). Samples of Neotrigonia margaritacea were sectioned approximately parallel to the long axes of the prisms and polished on horizontal diamond-impregnated discs (Struers DP-U2 type polisher) with grit sizes 1 and 0.25 µm. Preparation was carried out at the Institute of Metallurgy and Materials Science (IMIM) of the Polish Academy of Sciences in Kraków. We used two sets of equipment. First, we used an Inca Crystal (Oxford Instruments) detector coupled to an Auriga Cross-Beam Station (Carl Zeiss) FESEM (Centro de Instrumentación Científica, Universidad de Granada). To avoid excessive charging, samples were coated with a thickness of 2 nm of carbon in a Baltec MED 020 electron beam evaporator. The second set of equipment (used for most analyses) was a TSL OIM detector coupled to FEI Field Emission Gun (FEG) SEM Quanta 3D microscope of the IMIM. Operation in low vacuum mode made coating unnecessary. A special cone was attached to the SEM pole piece to minimize the so-called “skirt effect” of the primary electron beam and reduce the gas-path length. Analysis software (TSL OIM version 5.3) was used to post-process the EBSD measurements. All data with a confidence index (CI) below 0.1 were removed. For visualization purposes the following cleanup procedure was applied: 1) grain CI standardization, 2) neighbor orientation correlation, 3) neighbor CI correlation.

Shell Microstructure

The shell microstructure for all species examined conformed to the description for all members of the superfamily [7]. The calcareous part of the relatively thick shell is wholly composed of prismato-nacreous aragonite (Fig. 2A, B) [7]. The periostracum is relatively thick (4–7 µm) and persistent, and its continuation can be traced some way down (several tens of microns) into the calcareous part of the shell and forms a series of thick organic sheaths which surround the upper portion of the underlying prismatic layer (Fig. 2C, D). In shells with excellent surface condition, the outer periostracal surface is adorned with pimple-like nanometric, elongate elevations (Fig. 2D, E). More frequently observed, the external surface of the periostracum is studded with regularly spaced crystalline units (∼ 5 µm in diameter and <1 µm in height) (Fig. 2D–I), the bosses described in [7], which sometimes preserve a thin organic cover (Fig. 2D, E) and are totally embedded within the fully grown periostracum (measured thicknesses of 2–4 µm) (Fig. 2C, D, H, J) which wrinkles around their edges (Fig. 2F, H, I). These bosses have polygonal (pseudo-hexagonal) outlines (Fig. 2E–G) and, as previously described [7], [53], represent the initial calcification centres of the prisms of the underlying prismatic layer (Fig. 2B–D, J, K). At lower magnification it is apparent that these crystalline units tend to have a regular distribution in comarginal rows (Fig. 2B, D, F, H–K). The distance between units in a row is more or less the same as between rows and they are completely offset with respect to those of the preceding and succeeding rows, such that the final arrangement of crystalline units is an approximately grid-like reticulate arrangement, with the diagonals along the radial and comarginal directions.

The outer prismatic layer consists of aragonitic columnar non-denticular composite prisms which are between 15–30 µm wide and 60–100 µm long (Fig. 2A–D) and which, in longitudinal section, are shown to be composed of fibres radiating from the prism axis (Fig. 2M). There is a strict one-to-one correspondence between bosses and prisms (e.g., Fig. 2K), except at some ornament features, such as knobs and ribs, where some small prismatic units appear to lack the boss (Fig. 2L). The bosses are fully continuous with the prisms with the transition usually taking place smoothly, through a gentle continuous cone-like enlargement of the boss, which later accelerates its widening at the transition to the prism to produce a shoulder (Fig. 2D). In some cases, however, there is a more or less marked constriction at the transition (Fig. 2C). The whole range of transitions can be found within a single species (e.g., N. lamarckii; Fig. 2C, D).

The prisms are underlain by nacre (Fig. 2A, B, M). The one at the external half thickness is columnar ( = lenticular, [7]) (Fig. 2N), whereas it is of the sheet type internally (Fig. 2O).

Mineralogy and Crystallography

The mineralogy and crystallography have been assessed with EBSD. In all cases analysed of N. margaritacea, the phase maps (Fig. 3A–B) indicated that the prisms and their bosses are entirely made of aragonite. Orientation mapping of the bosses and the topmost part of the prisms was hindered by the fact that the top part of the prisms (including the bosses) was directly in contact with the embedding resin, which produced differential polishing and, as a consequence, the quality of the diffraction patterns was too low quality for a correct indexation. Nevertheless, some insight can be gained from the surface features of the bosses. In most of them there is some degree of natural abrasion and or etching. This reveals that each boss consists of several radially arranged sectors (Figs. 2F, G, 3C–E). Each sector displays a series of parallel lineations at angles of ∼60° or 120° with those of adjacent sectors (Fig. 3C–E). These lineations have also been revealed in the nacre of bivalves and gastropods [55], [56], in the foliated aragonite of monoplacophorans [57] and in the prismatic aragonitic spikes of anomalodesmatan bivalves [34] and have been recently demonstrated by TEM [58] and EBSD [59] to be parallel to the a-axis of aragonite. This therefore suggests that the bosses in Neotrigonia are short prismatic units of aragonite consisting of several crystals (sectors) which are polycyclically twinned on {110}. This is consistent with the angles between lineations of adjacent sectors (Fig. 3C–E).

The EBSD maps made on the more internal part of the prismatic units, where they are in coalescence, indicates that each prismatic unit is composed of tens of fibrous units which run more or less parallel to the long axis of the prism and tend to diverge outwards towards the periphery of the prism (Fig. 3A, B, F, G). In section, these units range in width from 1 to 7 µm. Adjacent fibres in a single prism have different colors in orientation maps (Fig. 3G), which implies that they are independent crystals. Although there is a varied array of orientations, most of the fibres come in blue and green colors, which, according to the color key (Fig. 3H, bottom), implies that the 010 and 100 orientations are perpendicular or at a high angle to the section plane (which approximately contains the long axes of the prisms). This is confirmed by examination of the 001 and 100 pole figures (Fig. 3H). The 001 maxima (i.e. the c-axes) are aligned with the long axes of the prisms, whereas the 100 maxima are scattered with an approximate equatorial distribution, which implies that the a-axes show no preferential orientation.

Although we do not know exactly the position where the prismatic unit changes from being a prism composed of several twinned crystals to a fibrous aggregate, the SEM images obtained under some diffraction contrast show a drastic change at the level where the thick organic membranes disappear (Fig. 3I), which suggests that this is the boundary between the two crystallographic arrangements.

Formation and Growth of the Periostracum and Bosses

It is evident that the crystalline bosses form within the periostracum and indeed they develop within folds of the periostracum even where it is not underlain by mineralized shell (Fig. 2J). The implication of this is that they must develop at an early stage of periostracum formation, necessitating a detailed investigation of periostracal formation in Neotrigonia in order to unravel the processes involved. By far, the best results for mantle fixation were achieved in the glutaraldehyde-fixed specimen of N. margaritacea, on which most observations on initial formation of the periostracum are based. However in most respects the observations made on the sub-optimally fixed N. lamarckii and N. gemma were in agreement, except for some details of the mantle lobes (see below). The excellent fixation of our glutaraldehyde-fixed material permits a detailed description of the histology of the periostracal groove and the forming periostracum.

Formation of the Periostracum and Bosses in Neotrigonia

Our observations for the origin in the basal cell and extrusion of the periostracum along the PG can be summarised and interpreted as follows, with reference to Fig. 12:

1. 1. Neotrigonia has a distinctive spiralling periostracal groove.
2. 2. The periostracum, secreted by the basal cell, is initially no more than a thin pellicle (25–30 nm thick). During the first half turn of the PG there is no appreciable thickening of this layer but it does become folded (Fig. 6A, E). At the same time, the first crystal plaques form below the initial periostracum (which, at this stage, seems to be constituted by the pellicle) (Fig. 9A, B). Given the grid-like reticulate arrangement of bosses, there must be particular areas of the outer mantle surface where cells are specialized for the production of new bosses and which are evenly spaced along the comarginal direction (positions 1 and 2 in Fig. 13). Once a new boss has been produced, the production subcellular site enters a refractory period until it becomes again activated for the production of a new boss. In this way, the bosses become regularly spaced in the longitudinal direction. For the grid-like pattern to become complete intermediate production sites (position 3 in Fig. 13) should become activated alternatively. The observed reticular extensions of the microvilli underlying the forming crystals (Fig. 9A, C–F) must, therefore, migrate in coincidence with these (Fig. 13). This model implies that the production of bosses is under genetic control.
3. 3. The bosses acquire their full width very soon within the early coiled portion of the PG. Their patterning into sectors observed upon etching (Fig. 2E, F, G) allows us to determine that tablets are in fact polycyclically twinned crystals (Fig. 3C–E). By comparison with other similar instances of biogenic aragonite (e.g., spikes of anomalodesmatans [34]) we can infer that the c-axis is perpendicular to the surface of the plaque and that the a-axes run in parallel to the observed lineations.
4. 4. From the beginning of the second half turn towards the exit of the PG, the folds of the DL become united by additions of new material by the cellular microvilli (Fig. 6F). The observed huge reticulate extensions of the microvilli developed by the OMF epithelial cells below each forming crystal are produced by active secretions of vesicles from the Golgi complex throughout the microvilli (Fig. 9C–E). This points to an active transport of calcium (either in the form of ions or as amorphous calcium carbonate [60]) and to the presence of organic compounds (such as phenolic-acid mucopolysaccharide complexes and sulfated glycoproteins). The application of fluorescent indicator fluo 3 AM on semithin sections has shown the presence of Ca²⁺ both in the epithelial cells of the outer mantle fold and in the forming periostracum (Fig. 8). Due to lack of resolution we could not test under the confocal the likely presence of Ca²⁺ in the vesicles opening to the reticulate extensions of microvilli seen under the TEM (Emv in Fig. 9). Jaiswal et al. [61] found that membrane proximal lysosomes are the major vesicles responsible for calcium-dependent exocytosis in nonsecretory cells.

Toluidine blue is known to cause red and purple metachromasia with acid mucopolysaccharide and glycoprotein respectively. Green metachromasia by toluidine blue has been reported, among other instances, in the organic cups of the solenogaster Proneomenia agalaophaeriae and inner cuticle of the chiton Acanthochitona crinitus [62], and in the vitelline cells of monogenetic platyhelminthes Pricea and Protomicrocotyle, and freshly moult epicuticle of the millipedes Spirostreptus asthenus [63] and these authors have shown that this is related to the presence of diphenols. Histochemical examination of the cuticle of Proneomenia showed that the matrix was composed of a glycoprotein complex with highly acidic mucopolysaccharide and low protein contents in which tanning plays little part in stabilization [62]. These authors considered that the cuticle of the Aplacophora is tentatively equated with an early mucoid stage in the evolution of the molluscan shell and it is suggested that secretion of additional protein, followed by hardening by quinone-tanning, are necessary stages before a calcified shell evolved. The presence of phenolic mucopolysaccharide complexes in the epithelial cells of the outer mantle fold of Neotrigonia margaritacea seems to point to a more intense hardening of the periostracum. The brownish metachromasia is usually related to the presence of sulfatide and acidic lipids [64]. In the epithelial cells of Neotrigonia margaritacea, the brownish metachromasia could be due to the link to sulfated glycoproteins with toluidine when acetic acid is added to the aqueous solution. Taking the above into account, the epithelial glands of the outer mantle fold of N. margaritacea may contain a phenolic-acid mucopolysaccharide complex and sulfated glycoproteins, among other secretions, which would be related both to the formation of the organic matrix and hardening of the periostracum.

The distance between the crystals and the tips of the microvilli or of their reticulate extensions (estimated at ∼100 nm) is the so-called extrapallial space, e.g. [14], [65] (Ep in Fig. 9C–E). Since the crystals are completely enclosed within the mantle tissues and in view of the consistency of the estimated thickness, we are confident that the extrapallial space is imaged with its correct dimension. Both the reticulate extensions of the microvilli and the extrapallial space are imaged here for the first time in molluscs.

1. 5. From the third half turn of the periostracal groove onwards, a conspicuous TL begins to appear and rapidly increases in thickness, so as to constitute the bulk of the periostracum (Fig. 9A, F). The DL also thickens probably due to tanning and condensation of the TL, as observed in other bivalves [66]. The crystalline bosses increase in thickness, by keeping pace with the DL, at the same time that they expand in width.
2. 6. During the free periostracum stage, i.e. that portion of the periostracum which spans the gap between the PG and the shell edge, the periostracum slides along the surface of the OMF at the same time that both periostracal layers thicken (Fig. 5D). The TL thickens by secretions of the OMF, whereas the DL thickens by tanning of the TL. When the periostracum leaves the tip of the OMF, the TL ceases to be secreted and its thickness begins to decrease due to continued transformation by tanning into DL. The presence of large and numerous glandular vesicles in the cells of the OMF, which stain greenish or brownish with toluidine blue (Fig. 7), may be related to the secretion of phenolic-acid mucopolysaccharide complex and sulfated glycoproteins that contribute to the formation of the organic matrix and to the tanning process of the TL into DL layer by phenolic secretions.

At the same time, the crystalline bosses continue to thicken and expand until they impinge upon each other acquiring a prismatic morphology (Fig. 11C–H). During growth of the prismatic layer, the prismatic units still continue to be separated by thick organic walls made of DL, which are formed at the expense of the last residues of TL (Figs. 10F, 11H).

1. 7. The prisms continue to increase in length beyond the periostracal walls, while still being separated by intervening spaces. At this time each prism has been transformed into a polycrystalline aggregate of longitudinal fibres (Fig. 3D–G).
2. 8. Finally, the nacreous layer begins to be secreted below the prismatic layer (Fig. 11H). At this point, the spaces between prisms become filled with additional mineral material (Fig. 11I).

Comparison of the Formation of the Periostracum and Initial Crystals in Neotrigonia with Other Bivalves

The architecture of the mantle edge of Neotrigonia is similar to that which has been described for other bivalves. As previously noted [50], there are three, relatively unspecialised mantle folds which may be described using the standard nomenclature of Yonge [67], [68], although there is some morphological variation between species. The presence of tentacles in the middle and inner mantle folds is a general feature in the genus. Although not remarked upon by Morton [50], it is apparent from his figure 2 that his material of N. margaritacea also had a coiled periostracal groove. The presence of a spiral-shaped periostracal groove is unusual but has also been reported in the heterodont cardiids Cerastoderma edule [69] and Acanthocardia tuberculata (unpublished data), as well as in the unionoids Potomida littoralis [24] and Unio pictorum (pers. obs.), the sister group to the trigonoids [47]. In all these cases, the periostracum also coils towards the OMF, but only for less than half a turn. As in other studies [66], [70]–[74], the periostracum originates from a basal cell. Although a few studies on, e.g. Nucula sulcata [73] and Cerastoderma edule [69], have indicated that some bivalves have two basal cells, we have no positive evidence that this is also true for Neotrigonia. The epithelial cells of the MMF fold are columnar in some bivalves species [71] or cuboidal in most of them [72], [75], but in Neotrigonia they are squamous and very stretched out. The cells lining the inner surface of the OMF are columnar and contribute to the formation of the TL of the periostracum, as in most of the studied bivalves [66], [69], [74], [75] (Figs. 5, 7 and 9A).

Our results show that the initial crystalline plaques which constitute the bosses of the prisms and the external part of the outer prismatic layer are entirely intraperiostracal. The previous suggestions that the bosses pierce the outer surface of the periostracum [36] are mistaken, based on imperfectly preserved material where the ultra-thin periostracum covering (in fact, the pellicle) has been lost.

The form of the bosses in Neotrigonia is very similar to those seen in the anomalodesmatans and unionoids, perhaps indicating some homology. There are, however, several characteristics of spike formation that are exclusive to Neotrigonia. Of particular note is that our observations demonstrate the very early onset of calcification within the periostracum of Neotrigonia, deep within the periostracal groove. Detailed investigation of periostracal calcification on other Recent bivalves, e.g. the gastrochaenid Spengleria [35], [37] and many anomalodesmatans [34], [76], indicate that in those groups these structures are initiated rather later in the process of periostracum formation only at the free periostracum stage, within the TL, at its boundary with the DL, after the sheet has exited the periostracal groove. Re-examination of previously published material [24] and other additional material indicates that the bosses of the prisms of Unionoidae (which, if present, are much less protrusive than those of Neotrigonia) form well within the periostracum, most probably only after this has reflected back onto the external surface of the shell. Zieritz et al. [36] also described mineralized spikes in 20 species of Unionoida, which are enclosed very shallowly within the periostracum. Although that study did not include a TEM study of their formation, their appearance on the shell surface [36] is more similar to that of the spikes of anomalodesmatans than to the bosses of Neotrigonia in that they remain well within the periostracum. Two additional, related features which are unique to Neotrigonia, are (1) the strictly regular distribution of initial bosses and (2) the one-to-one correspondence between bosses and prisms of the outer layer, which, together, imply that the number and position of prisms are determined already within the periostracal groove. When, during growth, bosses transform into prims and occupy all the available space, the resulting sizes and distribution patterns of the prismatic units is highly regular. This is in contrast with what is observed in other aragonitic prismatic layers (e.g., those of their close relatives, the unionoids), in which prisms may differ in the position within the periostracum as well as in size (Fig. 14A). Also the density may easily change from one position to another (Fig. 14A, B). This implies that the formation of unionid prisms are not under such an strict genetic control. These differences are responsible for the high degree of selection by competition of prismatic units [77], [78]) (Fig. 14B), which is totally absent in Neotrigonia.

Understanding how calcium carbonate crystallises within an organic layer may have considerable interest outside the realms of molluscan evolution. The ability to secrete a hard shell occurred ‘rapidly’ in disparate metazoan groups as part of the Cambrian explosion [79], [80]. Many of the early biomineralising taxa produced spicules or plates of calcium carbonate, presumably held in an organic sheath (see [81] for a review). Although the sclerites and spicules of many of these early scleritome-bearing organisms are rather complex structures [81], an understanding of how mineralization proceeds within an organic sheet may be instructive in understanding their evolution. Our findings of the very early onset of calcification within the growing periostracal sheet of these bivalves, is perhaps of broader interest as it suggests one model by which biomineralization may proceed from isolated units within an ‘organic’ cuticle, and which subsequently merge to form a recognizable shell.