A new model for periostracum andshell formation in Unionidae (Bivalvia,Mollusca)

A. Checa

Tissue&Cell

Tissue & Cell, 2000 32 (5) 405–416© 2000 Harcourt Publishers Ltddoi: 10.1054/tice.2000.0129, available online at http://www.idealibrary.com

Abstract The periostracum in Unionidae consists of two layers. The outer one is secreted within the perios-tracal groove, while the inner layer is secreted by the epithelium of the outer mantle fold. The periostracumreaches its maximum thickness at the shell edge, where it reflects onto the shell surface. Biomineralization beginswithin the inner periostracum as fibrous spheruliths, which grow towards the shell interior, coalesce and competemutually, originating the aragonitic outer prismatic shell layer. Prisms are fibrous polycrystalline aggregates.Internal growth lines indicate that their growth front is limited by the mantle surface. Transition to nacre is gradual.The first nacreous tablets grow by epitaxy onto the distal ends of prism fibres. Later growth proceeds onto previ-ously deposited tablets. Our model involves two alternative stages. During active shell secretion, the mantle edgeextends to fill the extrapallial space and the periostracal conveyor belt switches on, with the consequential secre-tion of periostracum and shell. During periods of inactivity, only the outer periostracum is secreted; this formsfolds at the exit of the periostracal groove, leaving high-rank growth lines. Layers of inner periostracum are addedoccasionally to the shell interior during prolonged periods of inactivity in which the mantle is retracted. © 2000Harcourt Publishers Ltd

Keywords: biomineralization, periostracum, shell, microstructure, bivalves, Unionidae

Introduction

A number of studies on periostracum and shell formationhave been conducted in gastropods (see reviews inSaleuddin, 1979; Wilbur & Saleuddin, 1983; Lowenstam &Weiner 1989). For bivalves, there is a good documentationon shell microstructures, but relatively few studies on thebiomineralization process, these being restricted either to

Departamento de Estratigrafía y Paleontología, Facultad de Ciencias,Universidad de Granada, 18071 Granada, Spain

Received 8 March 2000Accepted 19 July 2000

Correspondence to: Antonio Checa, Departamento de Estratigrafia yPaleontología, Facultad de Ciencias, Universidad de Granada, 18071Granada, Spain, Tel.: 958 243201; Fax: 958 248528;E-mail: acheca@goliat.ugr.es

periostracum (e.g., Hillman, 1961; Dunachie, 1963;Bevelander & Nakahara, 1967; Neff, 1972; Bubel, 1973;Saleuddin, 1974; Richardson et al., 1981) or to periostracumand shell formation (Kennedy et al., 1969; Taylor &Kennedy, 1969; Taylor et al., 1969; Nakahara &Bevelander, 1971; Waller, 1980) in several taxa. Detailedstudies were conducted and published (mainly in thisjournal) by Petit et al. (1978, 1979, 1980a, b) and Petit(1981), on fresh-water Unionidae (mainly the genusAmblema). They developed a mechanism for biomineraliza-tion in unionid bivalves, which was later made extensive tothose bivalves with prismato-nacreous shells (Saleuddin &Petit, 1983). It can be summarized as follows:

(1) The periostracum is initiated as a pellicle by basal cellslocated deep in the periostracal groove (between theouter and middle mantle folds). The pellicle is coatedon both sides and increases in thickness, thus formingthe outer periostracum. Growth lines are visible on its

405

406 CHECA

Fig. 1 Petit et al.’s model for shell biomineralization in Amblema (Unionidae). The periostracal groove secretes a thin outer periostracum and a thickermiddle periostracum (inset 1). When this periostracum is extruded, the inner surface of the outer mantle fold secretes a third layer (the inner periostracum)below the middle periostracum (inset 2). The inner periostracum later separates from the middle periostracum and develops folds, which disappear beforethe whole periostracum reflects at the shell margin. During reflection, the middle periostracum becomes vacuolized and forms a cavernous structure calledthe antrum (inset 3). In time, antrum cavities become filled with aragonite and originate the outer prismatic layer. The inner periostracum also foliates andbecomes mineralized, giving rise to the inner nacreous shell. Some free nacreous units, which form within the extrapallial fluid, later adhere to thenacreous layer.

external surface. Still within the periostracal groove, amulti-layered middle periostracum is secreted insidethe outer periostracum by the glycocalyx of thecolumnar epithelial cells lining the periostracal groove(Fig. 1, inset 1).

(2) Upon extrusion from the periostracal groove, an addi-tional layer (inner periostracum) is secreted below themiddle periostracum, by the glycocalyx of thecolumnar cells that line the outer mantle epithelium(Fig.1, inset 2). At a short distance from the periostracalgroove, the inner periostracum separates from the restof the periostracum and becomes highly folded. Thesefolds or loops disappear before reflection of the perios-tracum at the shell edge.

(3) The extruded periostracum (free periostracum stage) israther long and extends from the exit of the periostracalgroove to the shell edge, forming a wide extrapallialspace. During reflection of the periostracum, vacuolesappear in the middle layer, which begin to be filled withspherical subunits of needle-like crystals.Subsequently, these vacuoles enlarge and fuse, givingrise to elongated columns (Fig. 1, inset 3). With time,this system of cavities (antrum) becomes filled withspherical crystalline units, which, in packing, finallygive rise to the characteristic prisms of the outer shelllayer.

(4) The inner periostracal layer also thickens at the shelledge and suffers foliation processes in which nucle-ation of aragonite needles begin; these later pack intotypical nacreous bricks (initial nacre). Organicproteinaceous layers can also be released into the extra-pallial fluid from the inner periostracal layer, serving astemplates for the crystallization of nacreous layers.

These free units may later add to the nacreous layer andcontribute to its growth (Fig. 1, inset 4).

In summary, the periostracum is composed by threelayers. The outer and middle layers are formed within theperiostracal groove, while the inner layer is secreted by themantle epithelium. The middle layer is responsible forsecretion of the outer prismatic shell layer throughprocesses of vacuolization and antrum formation. Similarprocesses create the nacreous layer from the inner perios-tracum.

I have carried out a detailed study of the structure andmutual relationships of the mantle, periostracum and shell,in several species of Unionidae. In the present paper, on thebasis of these new data and observations, I propose a newmodel for periostracum and shell formation, which isintended to be of general applicability to unionid bivalvesand which is drastically different from that proposed forAmblema.

Material and techniques

MaterialLiving specimens of Unio elongatulus Pfeiffer, 1825(Unioninae) and Potomida littoralis (Lamarck, 1801)(Ambleminae) were collected in 1997 at Canal Imperial deAragón (provinces of Navarra and Zaragoza, Spain). Theywere initially fixed in 10% formaline and preserved in 70%ethanol. Quadrangular pieces of the shell margin (usuallythe ventral area) were cut with a small rotating disk (24 mmin diameter). Care was taken not to damage the mantlemargin, which was later cut with precision scissors. In thisway, the entire mantle-shell system could be examined atthe same time.

PERIOSTRACUM AND SHELL FORMATION IN UNIONIDAE 407

Given the importance of the shell as information source,well-preserved valves of other Unionidae were also studied:the Unioninae Unio crassus Philippson, 1788 (river Meuse,Hastières, Belgium) and U. pictorum (Linnaeus, 1758)(Nijvel, Belgium), and the Ambleminae Lamprotula sp.(locality unknown, China) and Caelatura bakeri (Adams,1866) (Nyanza, Kenya).

Optical and scanning microscopySamples of U. elongatulus and P. littoralis were surficiallydecalcified with 6% EDTA for 20 min at room temperaturefor SEM observation (Zeiss DSM 950). They were laterfixed in cacodylate buffered (0.1 M, pH 7.4) 2.5%glutaraldehyde for 4 h at 4°C. Once washed in the samebuffer, they were dehydrated in increasingly concentratedethanols and critical point-CO2 dried. Gold coating in aPolaron 5000 sputtering for 4 min followed.

For optical microscopy, samples were completely decal-cified, fixed and dehydrated according to the above routine.They were included in epoxy resin (Aname Epon 812) afterprevious infiltration and polymerized at 60°C for 24 h.Sections of 1-µm-thickness (made with a Leica Ultracut S)were mounted and stained with 1% toluidine blue.

Treatments were performed in the Centre for ScientificInstrumentation of the University of Granada.

Samples of Unio elongatulus, U. crassus, U. pictorum,Lamprotula sp. and Caelatura bakeri were studied withSEM. Observations were made on outer and inner valvesurfaces (sometimes with the periostracum partly removedwith 5% NaOH), as well as on transversal radial fractures(intact or etched in 1% hydrochloric acid for less than 1min).

In order to study crystallographic properties with polar-ized-light microscope, shell samples of Unio elongatulusand Lamprotula sp. were embedded in epoxy, thin-sectioned to 30–40 µm (with a Buehler Petro-Thin) perpen-dicular to the shell surface along a dorsoventral radius, andpolished (Struers Planopol-V).

Texture analysisThe three-dimensional orientation of crystals in shells ofLamprotula sp., Unio elongatulus and Potomida littoraliswas determined by X-ray diffraction, using a texturediffractometer (Phillips X’pert). The diffractometer bringsthe sample into different orientations and the recordedintensities are proportional to the number of lattice planes{hkl} that are in Bragg orientation. In our case, the Braggdiffraction peaks {001} and {112} were recorded overalmost a full hemisphere (Fig. 12A), from the substratenormal to a maximum angle of 85° from normal. The inten-sity data were registered by rotating in 2.5° steps, and tiltingthe sample 2.5° per complete rotation. Data treatmentincluded absorption and background corrections, and isolinecalculus. The three-dimensional orientation of the corre-sponding poles can be represented in two dimensions by its(Lambert) projection in the equatorial plane (Fig. 12A). The

scattering or degree of preferential orientation of crystalscan be deduced from the width of intensity maxima of polefigures. The smaller the width, the less scattered are thecrystals and the higher the degree of preferential orientation.The broadness of the central maxima of the {001} polefigure will account for the average tilting of the c-axis ofcrystals. The broadness of the maxima of the {112} polefigure indicates how much a- and b- crystal axes are rotatedaround the c-axis. Samples were prepared so that the equa-torial plane of the Lambert projection would be parallel tothe outer shell surface. To obtain values of the two shelllayers, samples were ground manually parallel to theexternal shell surface to the desired level.

All the material is deposited in the Department ofStratigraphy and Palaeontology of the University ofGranada (labelled SPUG).

Observations and Discussion

Outer and inner periostracumThe initial pellicle of Petit et al. (1978, 1979, 1980a) hasbeen easily recognized by SEM and thin sectioning. Theprocess of thickening within the periostracal groove is alsoevident (Fig. 2) and, upon extrusion from the periostracalgroove, the periostracum has an approximate thickness of2–3 µm in Unio elongatulus and Potomida littoralis (Figs 3,6), for normal-sized adult animals (75–85 mm of anteropos-terior diameter), lower values being typical of P. littoralis.These figures are much lower than those reported bySaleuddin and Petit (1983) of 25–50 µm in Amblema, butcoincide roughly with the thickness of their pellicle,inferred from Petit et al. (1979, Fig. 1; 1980a, Fig. 11). As inAmblema, tiny growth lines have been recognized on theouter surface of this layer in the species studied (Fig. 3).

Upon extrusion from the periostracal groove a new layeris secreted below the outer periostracum. This layer iscomposed internally of micron-sized sublayers (Fig. 4) andis clearly non-tanned. It seems to initiate at the very exit ofthe periostracal groove and thickens progressively in aventral direction (Fig. 5; compare Figs. 3 and 6), to attain30–40 µm at the very shell edge in U. elongatulus and20–25 µm in P. littoralis. On the basis of thickness data, thislayer can be confidently identified with the middle perios-tracum of Petit et al. The only plausible interpretation is thatthe inner periostracum is secreted by the inner epithelium ofan extended outer mantle fold. Dunachie (1963) showedhow the inner surface of the outer mantle fold, outside theperiostracal fold, is able to secrete periostracal material inMytilus edulis, but he did not recognize any zonation ofperiostracal layers along this surface. In our samples, thisfold is dramatically retracted even in previously relaxedanimals, and this effect was enhanced during critical drying.However, if, as in all processes of periostracum formation(see, e.g., Saleuddin, 1979), epithelial contact is needed, it

408 CHECA

Fig. 2 Light micrograph showing increased thickness of the pellicle within the periostracal groove in Potomida littoralis. The pellicle is firmly attachedto the outward-facing surface of the periostracal groove. Growth direction to the left. × 192.

Fig. 3 SEM view of a fold formed by the pellicle in Unio elongatulus, after extrusion from the periostracal groove. The area shown is very close (some150 µm) to the exit of the periostracal groove. Correspondingly, the inner periostracum is extremely reduced. Tiny growth lines formed within the perios-tracal groove are visible on the outer periostracal surface. Growth direction to the top left. × 340.

Fig. 4 Light micrograph of the periostracum of a decalcified shell of Lamprotula sp. The outer periostracum is included within the translucent layer andis folded in a complex fashion. The inner periostracum connects directly with the prisms of the outer shell layer (of which only the organic sheaths remain)and is layered internally. Growth direction to the left. × 384.

Fig. 5 Light micrograph of a decalcified shell of Unio elongatulus, showing the free and calcified periostracum stages. The boundary is indicated by anarrow. Folds are formed exclusively by the outer periostracum, just after extrusion from the periostracal groove (not shown). Thickening of the perios-tracum during the free stage is achieved solely at the expense of the inner periostracum. × 48.

Fig. 6 Same specimen and sample as in Figure 2. SEM view of the periostracum at a point closer to the shell edge (some 350 µm) and far from theperiostracal groove (some 1300 µm). The fold consists exclusively of outer periostracum, and the inner periostracum is thick (compare to Figure 2).Growth direction to the top right. × 725.

Fig. 7 SEM view of a partly decalcified shell margin of a fully-grown specimen of Potomida littoralis. A complex bundle of periostracal folds formed atthe shell edge due to continued emission of outer periostracum (sometimes overlying a thin inner periostracum), while shell and inner periostracum secre-tion was not active. Some bundles continue into the shell as intra-shell layers of inner periostracum. Growth direction to the bottom left. × 150.

PERIOSTRACUM AND SHELL FORMATION IN UNIONIDAE 409

Fig. 8 SEM oblique view of a partly dissolved inner periostracum and a fractured outer prismatic layer of Lamprotula sp. Prisms initiate as spherulithswithin the inner periostracum and expand transversely to coalesce. Competition among prisms causes disappearance of smaller units. Growth surfaces arecontinuous among adjacent prisms, indicating synchronized growth. Growth direction to the bottom right. × 340.

Fig. 9 Detail of Figure 7 to show the spherulithic initial part of prisms embedded within the inner periostracum.

Fig. 10 SEM micrograph of the transition between the prismatic and nacreous layers in Lamprotula sp. The internal fibrous structure of prisms is alsoevident. With growth, fibres align progressively with the long axis of the prism. Some prisms that disappear by competition are still visible in the outerpart. Growth direction to the left. × 725.

Fig. 11 Polarized-light micrograph of the outer prismatic layer of Unio elongatulus. Prisms are composed of fibres diverging from the main axis.Divergence becomes smaller towards the nacreous layer (right). Under polarized light, only those crystals with a parallel orientation of their c-axes extin-guish synchronously. Undulating extinction then demonstrates that prisms are polycrystalline aggregates, each fibre being a single crystal. Growth direc-tion to the top right. × 192.

can be assumed that the outer mantle could be extended tothe shell edge to fill most of the extrapallial space andsecrete the inner periostracum (Fig. 17A). This would alsosolve the mechanical problem of folding the periostracum

backwards at the shell edge just prior to the onset ofbiomineralization. In aquarium-maintained unionids themantle margins advance and protrude slightly from the shellin a normal-feeding position.

410 CHECA

Fig. 12A–D A. The pole of a given {hkl} lattice plane ({112} in the example) is the Lambert projection of its Bragg reflection onto the equatorial plane.When the sample is rotated over a full hemisphere, a record of intensities (pole figure) is obtained for the selected {hkl}. B, C, D. Pole figures for the pris-matic and nacreous layers of a shell of Unio elongatulus. Darker colours indicate increasing intensities. The broadness of intensity maxima of {001} polefigures gives an indication of c-axis scattering. It is high in the prismatic layer, which fits in with the fan-like arrangement of prism-forming fibres (seeFigure 10). Scattering is low in the nacreous layer, which implies that monocrystalline tablets have their c-axes evenly oriented (see Figure 15). Fourdefined intensity maxima of {112} pole figure for nacre indicate that, in addition to the c-axis, tablets have parallel a- and b-axes (as in A, the b-axiscoincides with the horizontal axis and the a-axis with the vertical one).

An important difference with respect to Amblema is thatI have found no traces of an innermost third periostracallayer.

Thickness data and layering indicate that the middleperiostracum of Petit et al. would correspond to our innerperiostracum. However, on a positional basis, with

PERIOSTRACUM AND SHELL FORMATION IN UNIONIDAE 411

Fig. 13 Light micrograph of the prismatic and part of the outer nacreous layer of Unio elongatulus. Internal growth lines plunge backwards and arecontinuous across the prismatic layer and into the nacreous layers. An intra-shell periostracal layer (arrow) originates a small prismatic layer. Both theintra-shell layer and the associated prismatic layer wedge out dorsally. Growth direction to the right. × 95.

Fig. 14 SEM micrograph of a partly decalcified shell of Potomida littoralis, showing the contact between the prismatic and nacreous layers. Severalintra-shell periostracal layers (with their associated prismatic sheets) can be seen intruding the nacreous layer. Growth direction to the bottom. × 145.

Fig. 15 Transition between the prismatic and nacreous layer in Lamprotula sp. Nacreous tablets adapt to the diverging arrangement of basal fibres of aprism. This supports the hypothesis that the first nacreous tablets grow by epitaxy directly onto the fibres. Growth direction to the left. × 1450.

Fig. 16 SEM view of the internal growth surface of the shell. Nacreous sheets are arranged step-like. Individual plates are rhombic in outline and, withminor exceptions (arrow), are consistently oriented with their b-axes aligned with the longest diagonal of the rhombi and with the growth direction (upperleft). The a-axis is transversal, along the shortest diagonal. × 950.

formation outside the periostracal groove, Petit et al.’s innerperiostracum and ours would be equivalent. Given the roleplayed by the different layers during calcification (seebelow) and the uncertainty in recognizing secretion sites inPetit et al.’s model, we favour the first possibility.

Periostracal foldsPetit et al. (1979, 1980a) mentioned loops in the innerperiostracal layer in front of the periostracal groove. I findfolds in the same area in U. elongatulus (Figs. 3, 5, 6), andLamprotula sp. (Fig. 4), but formed by the outer perios-tracum. Sometimes multiple, these folds are sealed at their

412 CHECA

Fig. 17A–E A, B. Proposed model for the formation of the periostracum and shell in Unionidae (compare to Figure 1). A. During active shell secretionthe outer mantle fold extends and fills the whole extrapallial space. The periostracal groove secretes the outer periostracum (inset 1), and the inner perios-tracum is secreted by the inward-facing epithelium of the outer mantle fold below the outer periostracum (inset 2). The periostracum reflects at the shelledge, and calcification initiates with the prismatic layer at the shell edge (inset 3) and continues with the nacreous layer a certain distance backwards fromthe margin (inset 4). B. When the mantle enters an inactive period, both the shell and inner periostracum cease to be secreted, while the outer periostracumcontinues to be extruded from the periostracal groove (inset 1). Since the periostracal conveyor belt remains motionless, the outer periostracum folds uponextrusion from the periostracal groove (insets 2a and 2b). Folds may be later sealed basally by an inner periostracal deposit when shell secretion resumes.They are later incorporated onto the shell surface as high-rank growth lines. The degree of retraction of the outer mantle fold is uncertain. C, D. E.Formation of intra-shell periostracal layers. C. During prolonged periods of inactivity, the mantle fold retracts and the outer periostracum detaches fromthe periostracal groove and ceases to be secreted. The free periostracum is expelled from the shell interior, forming a marginal bundle. D. When growthresumes, the periostracal groove and the outer fold epithelium begin to secrete their corresponding periostracal layers at the same time. Therefore, a transi-tory inner periostracum lacking its outer periostracal cover is secreted initially by the outer mantle fold, which is replaced with time by a two-layerednormal periostracum (inset 1). As the outer mantle fold extends, the isolated inner periostracum is transferred to the outward-facing face of the fold and tothe internal surface of the shell (inset 2). E. Once the outer mantle fold is fully extended, calcification proceeds on the inner periostracal layer (inset).

necks by the inner periostracum (Figs. 5, 6). Since they arenot present within the periostracal groove, they must form atthe very exit of the periostracal groove, before or at a veryincipient stage of inner periostracum formation. They arelater incorporated onto the shell surface (Fig. 5) and disap-pear with age due to wear of the periostracum. There arethree ways in which these folds can form. (1) The outer

periostracum folds because it continues to be extruded at aconstant rate from the periostracal groove while the perios-tracal ‘conveyor belt’ ceases to move forwards, in coinci-dence with halts in inner periostracum and shell secretion(Fig. 17B). They form at the site of lowest periostracalthickness and, hence, lowest resistance to folding. (2) Forthe same reasons, folds can form by compression of the free

PERIOSTRACUM AND SHELL FORMATION IN UNIONIDAE 413

periostracum when the mantle margin retracts betweencalcification periods. (3) Formation may result from acombination of these first two ways.

Similar folds are found in P. littoralis, a particularlyintricate bundle of loops being visible at the shell edge of afully-grown specimen (Fig. 7). Folds consist of outer perios-tracum, whether or not sealed basally by a drasticallythinned (3–4 µm) inner periostracum. This fits in withhypothesis (1) above since this pattern implies that growthstopped for an extended period, while the outer perios-tracum continued to be expelled from the periostracalgroove. The presence of a thin inner periostracum associ-ated with some bundles could imply restricted growthbursts. The pattern is so complex that it is difficult to discernindividual folds and also which periostracal layer isinvolved in their formation. In their third dimension, foldsare roughly parallel to the margin and, although they some-times merge with adjacent folds, they reach long extensions.Their regularity and marginal extension point to possibility(1) above. They are the growth lines usually discerned onthe shell surface at low magnification, and are unrelated tomicroscopic growth lines of the outer periostracal surface(Petit et al., 1980a, Fig. 11; Saleuddin & Petit, 1983, Fig.15; Fig. 3). The location and aspect of folds in Potomida isso similar to those referred in Amblema (Petit et al. 1979,Fig. 5, 1980a, Fig. 14) that it is tempting to conclude that theorigin is the same in both cases. The above-mentioned, dras-tically thinned or absent inner periostracum in Potomida,prevents the periostracal layer from thickening appreciablyafter extruding from the periostracal groove. If the speci-mens examined by Petit et al. were also fully grown, theycould not appreciate coarsening of the periostracum duringthe free stage and easily arrive at the conclusion that themiddle periostracum was secreted within the periostracalgroove.

Intra-shell periostracal layersIn some shells, the periostracum is interrupted from time totime, and it is relieved by a new periostracal band, whichinitiates inside the outer prismatic layer and, on reaching theshell surface, bends forwards to continue the former shellprofile. That part of the periostracum internal to the shell ishere called intra-shell periostracum (Figs. 7, 13, 14). Theseintra-shell layers are more common near the adult shell edgeand indicate growth halts related to reduced growth ratetypical of maturity. Intra-shell layers consist exclusively ofinner periostracum and wedge out towards the shell interior(Fig. 13). Given the mode in which the inner periostracum isformed, this implies that contact with the secretory surfaceof the outer mantle fold was progressively more prolongedfrom the interior to the exterior of the shell. An isolatedinner periostracal band can conceivably be formed if thewhole periostracum were, for some reason, lost and secre-tion of the outer (at the periostracal groove) and inner (at themantle fold surface) periostracum reinitiated suddenly andconcomitantly. In this way, the inward-facing surface of the

outer-mantle-fold epithelium begins to secrete an isolatedinner periostracum (since the outer periostracum is stillemerging from the periostracal groove). On sliding alongthe fold surface the extreme of this periostracal layer closestto the mantle edge soon leaves the secretory fold epitheliumand remains extremely thin, but the zones of the mono-layered periostracum which subsequently reach the mantleedge thicken progressively. Maximal thicknesses werereached when the mono-layer was replaced by the two-layered normal periostracum progressing forwards along themantle-fold surface (Fig. 17D). To explain how the mono-layered periostracum attaches to the internal shell surface Iassume that the outer mantle fold is retracted within theshell and that the mono-layer moves from the inward- to theoutward-facing surface of the outer mantle fold. In this way,it could reach a point of the internal shell surface, somewhatback from the aperture, and be progressively added to theinternal shell surface when the outer mantle fold progressesforwards (Fig. 17D, E). Secretion by a retracted mantle foldis reasonable since, as commented on above, intra-shelllayers represent prolonged growth halts. Under these condi-tions, detachment of the periostracum from the periostracalgroove and from the fold epithelium could occur (Fig. 17C),so that, when growth resumed, the whole periostracum hasto be fabricated de novo. The detached old periostracum canalso be expelled from the shell interior forming periostracalbundles similar to those observed in P. littoralis (Fig. 7),which connect with intra-shell periostracal layers.

Prismatic layerThe outer prismatic layer initiates with spheruliths near thebase of the inner periostracum, but still within it (Figs. 8, 9).During the bivalve’s life this layer must have behaved in aviscous fashion, since no cracks associated to spherulithgrowth have been found. Spheruliths grow inwards andextend beyond the inner boundary of the periostracum. Oncoalescing, they become polygonal and transform intoprisms (Fig. 8). A similar process was described by Tayloret al. (1969) and Waller (1998), and is inferred from thedescriptions and illustrations of the UnionidaeMargaritifera (Mutvei, 1991) and Anodonta (Ubukata,1994). At this point, geometric selection (Taylor et al.,1969) causes the disappearance of smaller units, so that onlya few evenly-sized prisms become fully grown (Fig. 8, 10).This process was interpreted by Ubukata (1994) as beingdue to competition among prisms, in a way similar to thatobserved in inorganic crystal growth.

Petit et al. (1979, 1980a, b) assumed that crystals werepreceded by the formation of vacuoles in the inner perios-tracum, which enlarged and fused to form a cavernous struc-ture, called the antrum (see above). Once fully formed, theantrum consisted of hollow organic columns with hexagonaloutlines, exactly representing the template for the subse-quent prisms. Spheruliths nucleate and pack to finally fillthe whole volume of preformed prisms. This model isuntenable for the following reasons:

414 CHECA

(a) Aragonitic unionid prisms were defined as simpleprismatic by Kennedy et al. (1969) and Taylor et al. (1969),and also by later authors (e.g., Watabe, 1988). Ubukata(1994) was the first to recognize their composite (non-denticular) character, since they are typically composed ofsmaller units which radiate from the main axis and areinclined inwards (e.g., Taylor et al., 1969; Carter, 1980).The composite character of prisms is evident under SEMobservation of intact fractured shells (Fig. 10) and in petro-graphic thin section. Polarized-light examination of thinsections reveals that prisms are polycrystalline, each fibrebeing a single crystal (Fig. 11). Also, {001} pole figures ofthe prismatic layer reveal that scattering of c-axes of fibresis great (Fig. 12B), but there is no random orientation, asimplied by the spherulith hypothesis of Petit et al. Withgrowth, fibres align progressively with the long axis of theprism, until a few, slightly diverging, large-sized unitsremain (Fig. 10, 11). None of the techniques reveal traces ofsecond-order spheruliths associated with prisms. In addi-tion, it is worth pointing out the possibility that thespherulithic structures illustrated by Petit et al. (1979, Figs.10, 15; 1980a, Figs. 19, 21, 22) formed during sample treat-ment, since the fixatives they appear to have used may, insome circumstances, induce artifactual crystallization (seeWatabe, 1990). In any case, it is clear that their spherulithsdo not correspond to prism microstructure.

(b) Prisms display internal growth lines, which runacross adjacent prisms, indicating synchronized growth(Figs. 8, 13). Growth lines (or surfaces) plunge towards theshell interior away from the margin (Fig. 13). These areorganic-matter-rich surfaces, which remain as transversalsheets after decalcification. Similar growth lines were illus-trated in other unionids by Tevesz & Carter (1980), Watabe(1984), Wilbur & Saleuddin (1983) and Ubukata (1994), aswell as by Petit et al. (1980a, b). The latter authors inter-preted these structures as fused vacuoles. Formation ofprisms in independent cavities, as the antrum hypothesisimplies, could not explain synchronized growth. Thesegrowth surfaces rather reflect the successive positions of theexpanded mantle (see above) during inward growth ofprisms.

(c) Prisms usually curve downwards to accommodate thechange in curvature of the growth surfaces (Figs. 11, 13).The mode in which antrum cavities would be formed is notclear, but, if it were by tension, prisms would always bestraight.

(d) The hexagonal outline of prisms and their mutualcompetition is not explained in the antrum hypothesis andresult from mere crystallographic processes (e.g., Ubukata,1994; Checa & Rodriguez-Navarro, in prep.).

Vacuole and antrum formation are interpretations, prob-ably inspired on the compartment hypothesis of Towe &Hamilton (1968), Bevelander & Nakahara (1969) and Towe(1972) for the formation of nacreous tablets in somebivalves, and later rejected by Erben & Watabe (1974).From our observations, it is clear that a similar hypothesiscannot be applied to the prismatic layer in the species

studied. These include three Ambleminae and it is hardlyconceivable that Amblema has such a drastically differentcalcification mode.

The thickness of the prismatic layer is clearly related tothe thickness of the inner periostracum and to shell size. In ashell of U. elongatulus the values of inner periostracum andprismatic layer thicknesses, measured under SEM, are (inµm): 3.5/47, 9/80, 20/133 and 35/241, for a range of anteri-oposterior diameters of 15.5 to 72 mm. Intra-shell perios-tracal layers are comprised exclusively by innerperiostracum (see above) and have the ability to extend pris-matic deposits partly into the nacreous layer (Figs. 13, 14).These periostracal strips wedge out towards the shell inte-rior, in coincidence with reduction in prism size. The rela-tionship of inner periostracum and prisms and of the relativethicknesses of both structures is, therefore, evident. As inthe laboratory (in which gels enhance or inhibit nucleation),it is conceivable that the gel-like inner periostracum in someway controls not only the presence but also the density andsize of prisms.

Nacreous layerAccording to Petit et al. this layer is initiated by cleavageand vacuolization of their internal periostracum, in a waysimilar to their middle periostracum in relation to the pris-matic layer. As mentioned above, the third periostracal layerof these authors is clearly absent in our samples.

At the base of prisms, there is a several-micron-thickshell layer, transitional to the nacre, which was previouslyrecognized by Petit et al. (1980b, figs. 1, 2, 14). In this area,prism fibres begin to septate transversely to transform intostacked nacreous (Figs. 10, 15). Stacking is not so evidentdeeper into the nacreous layer. A likely explanation is thatnacreous plates begin to nucleate by epitaxy on the mostdistal ends of fibres. This is supported by small changes inorientation of the first nacre crystals on adjusting to thesomewhat diverging orientation of basal fibres (Fig. 15).

SEM observation and texture analysis of unionidnacreous tablets reveal that they are evenly orientedmonocrystals (Figs. 10, 15, 16). Their c-axis is perpendic-ular to the tablet plane (Fig. 12C), the b-axis perpendicularto the margin (i.e., in the growth direction) and the a-axisparallel to the margin (Fig. 12D). Similar pole figures werefound by Hedegaard & Wenk (1998) in the pterioidPinctada margaritifera. Texture analysis fits in with theobserved rhombic morphology of growing crystals, whichalign their longest diagonal (b-axis) with the growth direc-tion (Fig. 16). They differ clearly in outline and pole-figurepattern from pseudo-hexagonal twinned crystals of nacrefound in other molluscs (Wise, 1970; Mutvei, 1977, 1978;Hedegaard & Wenk, 1998). Their monocrystalline natureand their common crystallographic orientation precludes thepossibility that nacreous bricks formed by spherulith coales-cence (see, e.g., Saleuddin & Petit, 1983, p. 227).

The nacreous layer extends along the inner shell surfaceand constitutes most of the shell thickness (e.g., 89% in anadult shell of U. elongatulus). The different lamellae of

PERIOSTRACUM AND SHELL FORMATION IN UNIONIDAE 415

nacre are arranged into steps and grow simultaneously (Fig.16), as has been reported in Amblema (Petit et al. 1980a)and in other bivalves (Bevelander & Nakahara, 1969;Taylor et al., 1969; Wise, 1970; Erben & Watabe, 1974;Lutz & Rhoads, 1980; Nakahara, 1991). Nacreous tabletsnucleate on previously formed crystals and epitaxial growthexplains the common orientation of nacreous tablets(Fig. 16).

As with the prismatic layer, the position of the mantlesets the boundary of the forming nacreous lamellae, andtheir growth front advances when new room is provided bythe dorsally-inclined outer surface of the mantle epitheliumin its forward movement during growth (Fig. 13).

Conclusions

The main conclusions on the mode in which the Unionidaeproduce their periostracum and shell are the following (Fig.17A):

(1) The periostracum of Unionidae consists of two layers,an outer, thin layer, which is secreted within the perios-tracal groove, and an inner, thicker layer, which isadded by the inward-facing surface of the outer mantlefold (Fig. 17A, insets 1, 2).

(2) Calcification initiates within the inner periostracumwith spheruliths, which, during early growth, protrudefrom the periostracum and coalesce, transforming intocomposite prisms. Prism growth is synchronized andthe common growth surface is constrained by themantle epithelium (Fig. 17A, inset 3). Prisms extend inlength as the dorsally plunging mantle epitheliummoves forward during bivalve growth.

(3) Nacreous tablets begin to nucleate directly on the endsof prism fibers by epitaxy (Fig. 17A, inset 4).Transition from composite prisms to nacre is probablyinduced by merely crystallographic processes (Checa &Rodríguez-Navarro, in prep.).

The existence of folds formed by the outer periostracum andpresent throughout the shell surface implies that calcifica-tion would be a periodic activity, interrupted by refractoryepisodes. The cycle is two-phase: (a) The mantle marginextends to the shell margin and its outer fold fills the extra-pallial space and touches the inner surface of both the freeperiostracum and the shell growth front. The conveyor beltmovement of the periostracum and shell system would beswitched on, at the same time that the outer mantle foldwould add new sub-layers to the inner periostracum and tothe forming shell (Fig. 17A). (b) Inner periostracum andshell cease to be secreted and only the outer periostracumcontinues to be extruded from the periostracal groove (Fig.17A, inset 1), with the subsequent formation of folds (Fig.17B, insets 2a, b). The physiological origin of these cyclesis not clear, but has a period intermediate between thosegrowth lines internal to the prisms of the outer shell layer orthose developed on the periostracal surface within the

periostracal groove and annual or semi-annual growth halts.The formation of intra-shell periostracal layers (see

above) can be explained with a similar model. They indicatethe onset of biomineralization after prolonged periods ofinactivity and retraction of the outer mantle fold. Uponmantle retraction, the outer periostracum detaches from theperiostracal groove and ceases to be secreted. The looseperiostracal band is expelled from the shell interior, forminga marginal bundle (Fig. 17C). When shell growth is to reini-tiate, the internal surface of the outer-mantle-fold epithe-lium begins to secrete an isolated inner periostracum. Thislayer slides along the fold surface with subsequent growthand is transferred to the outward-external surface of theouter mantle fold and to the internal surface of the shell(Fig. 17D, inset 2). At the same time, it is being replaced bythe two-layered normal periostracum progressing forwardsalong the internal fold surface (Fig. 17D, inset 1). Themono-layered periostracal sheet adheres progressively tothe inner shell surface as the outer mantle fold extendsforwards (Fig. 17D, inset 2). Once the mantle fold is fullyextended and reaches the shell edge, prisms grow on thesubstrate provided by the internal periostracal layer(Fig. 17E, inset). Alternatively, as proposed by Dunachie(1963) for Mytilus edulis, the inner surface of the outermantle fold could also secrete inner periostracum whenexposed to seawater, after detachment of the old perios-tracum. Calcification may reinitiate once the periostracumis regenerated and the extrapallial space is sealed off fromthe environment.

In our model, the periostracum has a more reduced rolein shell configuration, in comparison to that in Petit et al.’smodel. While the outer periostracum serves only a tight-sealing function for the extrapallial space, there is a clearconnection between the inner periostracum and the initia-tion of the prismatic layer. Gels in laboratory may enhanceor inhibit nucleation and the gel-like inner periostracummight be responsible for the unusual size of unionid arago-nitic composite prisms. The fact that both the inner perios-tracum and shell are secreted concomitantly by the outermantle fold points to this connection. The rest ofmicrostructural features of the unionid shell may be attribut-able to the activity of the mantle and to the crystallographicproperties of mineral matter.

ACKNOWLEDGEMENTS

Special thanks are given to Juan de Dios Bueno-Pérez(Unidad de Servicios Técnicos, Universidad de Granada),for his technical assistance, and to Drs Serge Gofás(Muséum National d’Histoire Naturelle, Paris, and,Departamento de Zoología, Universidad de Málaga) andRafael Araujo (Museo Nacional de Ciencias Naturales,Madrid) for providing specimens. Texture analysis waskindly carried out by Dr. Alejandro Rodríguez-Navarro(Savannah River Ecology Lab, University of Georgia). Thecritical suggestions of an anonymous reviewer have signifi-cantly improved the paper. This study has been financed by

416 CHECA

project PB97–0790 of the DGESIC (Ministerio deEducación y Cultura) and by the Research Group RNM0178 (Junta de Andalucía).

REFERENCES

Bevelander, G., and Nakahara, H. 1967. An electron microscopic study ofthe formation of the periostracum of Macrocallista maculata. Calc.Tiss. Res. 1, 55–67.

Bevelander, G., and Nakahara, H. 1969. An electron microscope study ofthe formation of the nacreous layer in the shell of certain bivalvemolluscs. Calc. Tiss. Res. 3, 84–92.

Bubel, A. 1973. An electron microscopic study of periostracum formationin some marine bivalves. 1. The origin of the periostracum. Mar. Biol.20, 213–220.

Carter, J. G. 1980. Guide to bivalve shell microstructures. In: Rhoads, D.C. and Lutz, R. A. (eds) Skeletal growth of aquatic organisms.Biological records of environmental change. Plenum Press, New York,645–673.

Dunachie, J. F. 1963. The periostracum of Mytilus edulis. Trans. Roy.Soc. Edinburgh 65, 383–411.

Erben, H. K., and Watabe, N. 1974. Crystal formation and growth inbivalve nacre. Nature 248, 128–130.

Hedegaard, C., and Wenk, H.-R. 1998. Microarchitecture and texturepatterns of mollusc shells. J. Moll. St. 64, 133–136.

Hillman, R. E. 1961. Formation of the periostracum in Mercenaria merce-naria. Science 134, 1754–1755.

Kennedy, W. J., Taylor, J. D., and Hall, A. 1969. Environmental andbiological controls on bivalve shell mineralogy. Biol. Rev. 44,499–530.

Lowenstam, H. A., and Weiner, S. 1989. On Biomineralization. OxfordUniversity Press, New York.

Lutz, R. A., and Rhoads, D. C. 1980. Growth patterns within themolluscan shell: an overview. In: Rhoads, D. C. and Lutz, R. A. (eds)Skeletal growth of aquatic organisms. Biological records of environ-mental change. Plenum Press, New York, 203–254.

Mutvei, H. 1977. The nacreous layer in Mytilus, Nucula and Unio(Bivalvia). Calc. Tiss. Res. 24, 11–18.

Mutvei, H. 1978. Ultrastructural characteristics of the nacre in somegastropods. Zool. Scripta 7, 287–296.

Mutvei, H. 1991. Using plasma-etching and proteolytic enzimes in studiesof molluscan shell ultrastructure. In: Suga, S. and Nakahara, H. (eds)Mechanisms and phylogeny of mineralization in biological systems.Springer-Verlag, Tokyo, 157–160.

Nakahara, H. 1991. Nacre formation in bivalve and gastropod molluscs.In: Suga, S. and Nakahara, H. (eds) Mechanisms and phylogeny ofmineralization in biological systems. Springer-Verlag, Tokyo,343–350.

Nakahara, H., and Bevelander, G. 1971. The formation and growth of theprismatic layer of Pinctada radiata. Calc. Tiss. Res. 7, 31–45.

Neff, J. M. 1972. Ultrastructural studies of periostracum formation in thehard shelled clam Mercenaria mercenaria (L.). Tissue & Cell 4,311–326.

Petit, H. 1981. Survol de la minéralisation chez les Unionidae. Haliotis 11,181–195.

Petit, H., Davis, W. L., and Jones, R. G. 1978. Morphological studies onthe mantle of the fresh-water mussel Amblema (Unionidae): scanningelectron microscopy. Tissue & Cell 10, 619–627.

Petit, H., Davis, W. L., and Jones, R. G. 1979. Morphological studies on

the periostracum of the fresh-water mussel Amblema (Unionidae):light microscopy, transmission electron microscopy, and scanningelectron microscopy. Tissue & Cell 11, 633–642.

Petit, H., Davis, W. L., Jones, R. G., and Hagler, H. K. 1980a.Morphological studies on the calcification process in the fresh-watermussel Amblema. Tissue & Cell 12, 13–28.

Petit, H., Davis, W. L., and Jones, R. G. 1980b. A scanning electronmicroscopy study of the inorganic and organic matrices comprisingthe mature shell of Amblema, a fresh-water mollusc. Tissue & Cell 12,581–593.

Richardson, C. A., Runham, N. W., and Crisp, D. J. 1981. A histologicaland ultrastructural study of the cells of the mantle edge of a marinebivalve, Cerastoderma edule. Tissue & Cell 13, 715–730.

Saleuddin, A. S. M. 1974. An electron microscopic study of the formationand structure of the periostracum in Astarte (Bivalvia). Can. J. Zool.52, 1463–1471.

Saleuddin, A. S. M. 1979. Shell formation in molluscs with special refer-ence to periostracum formation and regeneration. In: Van der Spoel,S., Van Bruggen, A. C. and Lever, J. (eds) Pathways in malacology.Bohn, Scheltema and Holkema, Utrecht, The Netherlands, 47–81.

Saleuddin, A. S. M., and Petit, H. 1983. The mode of formation and thestructure of the periostracum. In: Saleuddin, A. S. M. and Wilbur, K.M. (eds) The Mollusca, Vol. 4. Academic Press, New York, 199–234.

Taylor, J. D., and Kennedy, W. J. 1969. The influence of the periostracumon the shell structure of bivalve molluscs. Calc. Tiss. Res. 3, 274–283.

Taylor, J. D., Kennedy, W. J., and Hall, A. 1969. The shell structure andmineralogy of the Bivalvia. Introduction. Nuculacea-Trigonacea. Bull.Br. Mus. Nat. Hist., Suppl. 3, 1–125.

Tevesz, M. J. S., and Carter, J. G. 1980. Environmental relationships ofshell form and structure in unionacean bivalves. In: Rhoads, D. C. andLutz, R. A. (eds) Skeletal growth of aquatic organisms. Biologicalrecords of environmental change. Plenum Press, New York, 295–322.

Towe, K.M. 1972. Invertebrate shell structure and the organic matrixconcept. Biomineralisation 4, 1–14.

Towe, K.M., and Hamilton, G. H. 1968. Ultrastructure and inferred calci-fication of the mature and developing nacre in bivalve molluscs. Calc.Tiss. Res. 1, 306–318.

Ubukata, T. 1994. Architectural constraints on the morphogenesis of pris-matic structure in Bivalvia. Palaeontology 37, 241–261.

Waller, T.R. 1980. SEM of the shell and mantle in the order Arcoida(Mollusca, Bivalvia). Smithsonian Contr. Zool. 313, 1–58.

Waller, T.R. 1998. Origin of the molluscan class Bivalvia and aphylogeny of major groups. In: Johnston, P. A. and Haggart, J. W.(eds) Bivalvia: an eon of evolution. University of Calgary Press,Alberta, 1–45.

Watabe, N. 1984. Chapter 25. Shell. In: Bereiter-Hahn, J., Matoltsky, A.G. and Richards, K. S. (eds) Biology of the integument. 1. Springer-Verlag, Berlin, 448–485.

Watabe, N. 1988. Shell structure. In: Trueman, E. R. and Clarke, M. R.(eds) The Mollusca, Vol. 11. Academic Press, New York, 69–104.

Watabe, N. 1990. Chapter 4. Calcium phosphate structures in inverte-brates and protozoans. In: Carter, J. G. (eds) SkeletalBiomineralization: patterns, processes and evolutionary trends, vol. 1.Van Nostrand Reinhold, New York, 35–44.

Wilbur, K. M., and Saleuddin, A. S. M. 1983. Shell formation. In:Saleuddin, A. S. M. and Wilbur, K. M. (eds) The Mollusca, vol. 4.Academic Press, New York, 235–287.

Wise, S. W., Jr. 1970. Microarchitecture and mode of formation of nacre(mother-of-pearl) in pelecypods, gastropods, and cephalopods.Eclogae Geol. Helv. 63, 775–797.