Mineralogical aspects of asteriscus of goldfish (Carassius auratus) consisting of vaterite

Abstract

The fine structure of goldfish (Carassius auratus) asterisci, which consists of vaterite—a metastable polymorph of anhydrous calcium carbonate—has been investigated and compared with inorganically synthesized vaterites, using electron microscopy. This is as a first step in elucidating the mechanism of polymorph selection of calcium carbonate in the biomineralization of otoliths. X-ray and electron diffraction analyses suggested that there was no marked difference in the unit cell parameters, supercell structure, or stacking disorder features between the asteriscus vaterite and the synthesized vaterites. Although the sizes of the vaterite single crystals in the asteriscus are considerably larger than those in the synthesized ones, both show mosaicity, or crystal aggregates with small misorientations, implying that this character is an intrinsic property of the vaterite structure. The asteriscus consists of slender elements radiating from the central region of the asteriscus, with the elements extending normal to the c-axis, suggesting that the polymorph was selected at the asteriscus initial growth stage.

INTRODUCTION

As anhydrous calcium carbonate (CaCO₃) formed at ambient temperature and pressure, three polymorphs—calcite, aragonite, and vaterite—are known. Among them, vaterite is metastable and has a higher solubility than the other polymorphs (Gal et al., 1996). For instance, vaterite, initially precipitated from a supersaturated solution, transforms into stable calcite as the saturation index of calcium carbonate in the solution decreases (Kim et al., 2005). Accordingly, vaterite is rarely observed as a geological mineral.

Another characteristic feature of vaterite is that its crystal structure is not as precisely determined as those of the other two polymorphs. Because vaterite single crystals of good quality are not available, several structural models have been proposed to explain the powder X-ray diffraction (PXRD) patterns. Kamhi’s hexagonal sub-cell (a = 4.13, c = 8.49 Å, P6₃/mmc) (Kamhi, 1963) proposed in the early stage has Ca ions at the origin (0, 0, 0) and half height (0, 0, 0.5) of the sub-cell, and carbonate ions statistically distributed between the Ca atomic planes normal to the c-axis, with the triangle of three carbonate oxygens oriented parallel to the c-axis. This sub-cell explains the location and intensity of the major sharp peaks in the PXRD pattern. However, the true unit cell is thought to be larger because of the ordered arrangement of carbonate ions between the Ca planes. Furthermore, high-resolution transmission electron microscopy (HRTEM) images confirmed the disappearance of periodicity along the c-axis, forming dense stacking disorder (Kabalah-Amitai et al., 2013). This ordering of the carbonate ions in the a_i-plane perpendicular to the c-axis and the disorder along the c-axis produce several weak and broad reflections in the PXRD pattern.

Biominerals are inorganic materials formed by living organisms, such as components of hard tissues including bones, teeth, shells, and outer skeletons. Calcium carbonate is one of the major constituents of biominerals; for example, mollusc shells generally consist of calcite and/or aragonite, and the selection of the two polymorphs is strictly controlled by the species and/or shell microstructures. Although geological vaterite is rare, its occurrence has been frequently reported as a biomineral. These include the spines of ascidians (Kabalah-Amitai et al., 2013), non-lustrous abnormal parts of freshwater pearls (Qiao et al., 2007), and spines of succulent plants in the Saxifraga genus (Wightman et al., 2018).

Otoliths, which are massive hard tissues composed of calcium carbonate, are located in the inner ears of vertebrates, including fish. The function of otoliths is to detect the body’s tilt, acceleration, and sound (Popper et al., 2005). Most osteichthyes (bony fish) have three types of otoliths, namely lapillus, sagitta, and asteriscus, from cranial to caudal. While the lapillus and sagitta are composed of aragonite, the asteriscus is composed of vaterite (Pracheil et al., 2019). In sturgeons, otoliths are composed of vaterite and calcite (Chakoumakos et al., 2016; Pracheil et al., 2017).

The selection mechanism of polymorphs in the otolith formation process has not yet been elucidated; however, organic matter may be involved. For example, the addition of extracts from the asteriscus to a system synthesizing calcite induced vaterite (Falini et al., 2005; Ren et al., 2013). By knocking out a protein named ‘starmaker’ in sagitta which is originally composed of aragonite, it changed into calcite (Söllner et al., 2003), suggesting that the protein might play a role in the polymorph selection in otoliths. However, the ultimate goal of elucidating the polymorph selection mechanism in biomineralization is to understand the interactions between organic and inorganic matter at the molecular level. For this purpose, the fine structures of the inorganic side, calcium carbonate crystals forming the otoliths in the present study, should be characterized in detail. Historically, structural research on fish otoliths has focused on the sagitta, investigating its morphology for the classification of fish and the layered structures formed by additional growth to understand their life history. In contrast, there are no reports on the characteristics of vaterite in the asteriscus and how it differs from synthetic vaterite. Therefore, this study aimed to understand the mineralogical aspects of the goldfish (Carassius auratus) asteriscus, particularly by comparing the constituting vaterite with inorganically synthesized vaterite.

MATERIALS AND METHODS

Materials

The species of fish selected for the study of asteriscus was goldfish (Carassius auratus) because they are easily obtainable and raised and their genomes have already been determined (Chen et al., 2019). Asterisci were taken from the heads of 3-5 cm goldfish (no more than one year after birth) obtained from a commercial dealer. After dissecting and removing the otoliths, they were washed with water and 99.5% ethanol and dried in air. To prepare thin sections, asterisci embedded in epoxy resin were ground to a thickness of 0.03 mm. During this process, the ground surface was often covered with an instant adhesive to prevent crystals from falling out. Finally, the surface of the cross section was polished with colloidal silica. Two synthetic vaterite samples were obtained. One (termed S-1) was synthesized by mixing 30 mM CaCl₂ and 30 mM NaHCO₃ at 30 °C, prepared and donated by Dr. Yoshino of the University of Tokyo (Kamiya et al., 2004). Another (S-2) was synthesized by mixing 1M CaCl₂ and 1M K₂CO₃ and homogenizing them with ultrasonic waves. This was prepared and donated by Mr. Mori of the University of Tokyo (Mori et al., 2009). S-2 precipitated from a far more supersaturated solution than S-1. Both synthetic vaterites appeared as fine white powders that were ready for PXRD without grinding.

Methods

The asteriscus samples were etched with a sodium hypochlorite solution (available chlorine 5%) for 1 h to remove organic matter from the surface. The etched asterisci and two synthetic vaterites were coated with Pt-Pd (5 nm thick) by an E-1030 thin film coater (Hitachi) for scanning electron microscopy (SEM). SEM observations were performed using an S-4500 SEM (Hitachi) equipped with a cold field-emission gun at an acceleration voltage of 2-15 kV. The other asterisk samples were powdered in an agate mortar and placed in a nonreflective silicon sample holder for PXRD analysis. PXRD patterns were obtained by a RINT-Ultima+ diffractometer (Rigaku) with Cu Kα radiation emitted at 40 kV and 30 mA and monochromatized using a Ni-foil filter. A 1° divergence slit and an 8 mm anti-scatter slit were adopted. The scan rate was 1° (2θ)/min and data were collected at every 0.01° (2θ). A silicon-strip X-ray detector (Rigaku D/teX Ultra2) was used to record the diffracted X-rays. The cell parameters were calculated using PDXL ver. 2.8.4.0 (Rigaku) and the widths of the reflection peaks were estimated using JADE ver. 6 (Materials Data Inc.).

The crystal orientations of the vaterite in the asterisci were determined using an EBSD system (Oxford HKL Channel 5) equipped with a low-vacuum (LV) SEM (JEOL JSM-6510LV). The accelerating voltage and probe current were 10 kV and 70-90 µA, respectively; the step size for the EBSD acquisition was 1 µm; and the LV-SEM chamber pressure was 10 Pa. Electron-transparent thin films were prepared for TEM examination using an FB-2100 focused ion beam (FIB) system (Hitachi). The sample surfaces were locally coated with a tungsten film to prevent Ga implantation and the final film thicknesses were 100 to 400 nm. A JEM-2010 TEM (JEOL), operated at 200 kV, was used to image the fine structures and obtain the selected area electron diffraction (SAED) patterns. The crystal orientations obtained by TEM were plotted as pole figures using Stereonet 11 (Allmendinger et al., 2011; Cardozo and Allmendinger, 2013).

RESULTS

XRD analysis

Figure 1 shows the PXRD patterns of the asterisci and two synthetic vaterites, together with the pattern calculated using Kamhi’s sub-cell (1963). All the patterns consisted of two diffraction peak types: intense and sharp peaks and weak and broad peaks (Fig. 1b). All intense and sharp peaks in the patterns from the asteriscus and synthetic vaterites could be explained by the sub-cell, and no calcite or aragonite peaks were detected except for S-2. There were no significant differences in the positions of these peaks between the asteriscus and synthetic vaterites. The cell parameters of the Kamhi sub-cell derived from these peaks are as follows:

• a = 0.41282(2), c = 0.84617(4) nm, c/a = 2.050 for asteriscus,
• a = 0.41292(3), c = 0.84677(7) nm, c/a = 2.051 for S-1,
• a = 0.41333(3), c = 0.84771(6) nm, c/a = 2.051 for S-2,

indicating that the values are almost identical between the asteriscus and S-1. The cell parameters for S-2 were slightly larger, which may be due to the incorporation of impurity ions (K and/or Cl) from the solution during the synthesis under extremely supersaturated conditions. A distinct difference between these PXRD patterns is the half-widths of the sharp peaks. The full widths at half maximum (FWHM) of the 002 reflection for the asteriscus, S-1, and S-2, after removing the Kα₂ component using the software and subtracting the FWHM intrinsic to the instrument, were 0.050, 0.122, and 0.207° (2θ), respectively, and those of 100 reflections were 0.091, 0.185, and 0.301° (2θ), respectively. The half-width was S-2 > S-1 > asteriscus, reflecting their different coherent scattering domain (CSD) sizes, as described in the TEM results below. Another difference is that the 00l peaks of the asterisci were considerably higher than those of the synthetic samples, suggesting a preferred orientation owing to the anisotropic shape of the asteriscus powder.

Figure 1. (a) PXRD patterns from the powdered asteriscus and two synthetic vaterites, together with the calculated one using Kamhi’s sub-cell model. The indices for the sharp peaks are based on the sub-cell. (b) A portion of the patterns to show weak and broad peaks originated from a superstructure cell, probably with ordered arrangement of carbonate ions in the a_i-plane and broadened by dense stacking faults along the c-axis.

In addition to the peaks that could be attributed to the sub-cell, weak and broad peaks originating from a larger true cell with an ordered arrangement of carbonate ions in the a_i-plane and stacking disorder along the c-direction were observed (Fig. 1b). We cannot compare these with the calculated ones because the true structure of vaterite is still debatable (Christy, 2017), but these peaks are mostly similar among the three samples, suggesting that the real structure of vaterite beyond the sub-cell is almost identical between the vaterite in the asteriscus and the synthesized vaterites, which is also confirmed by the TEM results described below.

Microscopy of the asteriscus

The goldfish asteriscus has a disk-like shape (Figs. 2a and 2b). The proximal surface—where it attaches to the otolith membrane—is roughly concave, whereas the distal surface is convex. Figures 2c and 2d show thin sections of the asteriscus almost parallel (c) and perpendicular (d) to the disk, observed using a polarized light microscope. Except for the domain in the upper left corner of Figure 2c, the vaterite birefringence color suggests that the asterisci grew radially from around the central region. In the upper-left domain, such a radiation contrast was not observed and the birefringence was weak irrespective of the polarization direction.

Figure 2. (a), (b) SEM images of the goldfish asterisci observed from (a) the proximal and (b) the distal sides. Optical micrographs of thin sections of the asterisci (c) parallel and (d) perpendicular to the disks, taken with crossed Nicols. Note that the asterisci in these figures are all different individuals.

EBSD mapping was performed on a thin section parallel to the disk in Figure 2c (Fig. 3). Around the edges in the top and center areas, EBSD patterns of sufficient quality could not be obtained, probably because around the edge in the top polishing was not perfect, and around the center the crystal size was too small. In other areas, the patterns were properly indexed with the crystallographic parameters of Kamhi’s sub-cell model. Firstly, the inverse pole figure (IPF) images revealed the presence of a rather monotonous domain in the upper left part (A in the IPF-X image in Fig. 3) and other domains (B to H) consisting of slender elements of 10-20 µm in width extending radially from around the asteriscus central region. These features correspond to the polarized optical micrograph shown in Figure 2c. The pole figure for each domain shows the concentrations of the c and a_i axes directions, indicating that the crystal orientations were roughly aligned inside the domain (Fig. 3). In domain A, the IPF-Z image and pole figure show that the c-axis is almost perpendicular to the disk, and the three a_i-axes are located around the equator. In the other domains with slender elements, the IPF images and [001] pole figures show that the c-axis was normal to the direction of the extension of the slender elements. In other words, their extension directions were close to one of the three a_i-axes. Considering its exceptionality, domain A is likely formed by abnormal nucleation and growth during the initial growth stage.

Figure 3. (Top) IPF images constructed from EBSD measurement of the thin section in Figure 2c. The colors in the images correspond to the zone axes of vaterite parallel to X (horizontal), Y (vertical), and Z (normal to the paper) directions. The black color means that interpretable Kikuchi patterns could not be obtained from the areas. The whole area was divided into domains A to H based on the similarity of orientation, as shown in IPF-X. The rectangle in IPF-X image and its enlarged one at the bottom-left indicate the location of the area (the bold white bar in the enlarged image) where the thin-film specimen for TEM analysis was picked by FIB. (Bottom) Pole figures of upper hemisphere projection showing the distribution of [001] and <100> zone axes in A to H domains. The color scales and associated numbers show multiples of uniform distribution (m.u.d.).

A thin film specimen for TEM analysis was prepared using FIB from the asteriscus in Figure 2c (the sampling location is shown in the IPF-X image in Fig. 3), as it contained the cross-section of two adjacent slender elements identified in the IPF images. From the entire thin film area (Fig. 4a), the SAED pattern along <210> was obtained by tilting the specimen within ±20° in the TEM, and the directions of the a_i- and c-axes were calculated using the orientation of the pattern and tilt angle (Fig. 4b). The results in Figure 4b—where the directions of one of the a_i-axes closest to the north pole (expressed as [100] in the Fig. 4b) and the c-axis at a number of selected areas in the film are shown—indicate that the entire specimen contains approximately two areas with different orientations, especially with respect to the c-axis. This feature probably corresponds to the two adjacent elements identified by EBSD. However, as shown in the bright-field images in Figures 4c and 4e, the diffraction contrast indicates that the single crystal is smaller than the element. Moreover, these single crystals appear perpendicular to the c-axis, considering the corresponding SAED patterns along <210> (Figs. 4d and 4f). The slender elements with a thickness of ∼ 10 µm identified in the EBSD mapping have a mosaic structure composed of single crystals with a small misorientation.

Figure 4. TEM images and SAED from the thin film of asteriscus, prepared by FIB. (a) Bright-field (BF) image of the almost whole area, which contains two adjacent slender elements in Figure 3. Circles in the image indicate the points where the crystal orientation was determined by SAED. The points inside one element are colored with orange, and those inside another element are colored with light blue. The possible boundary of the two elements is drawn with the broken line. (b) Crystal Orientations of vaterite obtained from SAED taken from 37 points and displayed as a stereo projection. The direction of the north pole corresponds to the normal of the thin film. [100] is indicated by circles, and the [001] by triangles. The points that belong to the different elements are colored with orange and light blue, respectively. (c) An enlarged BF image of the rectangular C area in (a), after slight tilting to make single crystal areas distinct by diffraction contrast. (d) SAED pattern from the darkened crystal in (c). (e) An enlarged BF image of the rectangular area D in (a), after slight tilting. (f) SAED pattern from the darkened crystal in (e). (g) Lattice image viewed along <210>, showing dense stacking faults along the c-axis. (h) An image to show that the direction of the contrast of the stacking faults is changed across the crystal boundaries.

The features of these SAED patterns are similar to those reported by Mugnaioli et al. (2012). Between the 00l and 12l diffraction rows with sharp spots, which can be explained by Kamhi’s sub-cell, two diffraction rows with denser spots and streaks on the rows are observed. These additional rows originate from a $\sqrt 3 \times \sqrt 3$ superstructure from Kamhi’s sub-cell (Christy, 2017). Moreover, the intense streaks on the rows originate from dense stacking faults along the c-axis, visible in the lattice image along this direction (Figs. 4g and 4h).

Synthetic vaterite microstructure

The synthetic vaterite grains precipitated from a lower supersaturated solution (S-1) have a shape like a thick convex lens (Figs. 5a and 5b) or an aggregate of them (Fig. 5c). The bright-field TEM images of the cross-section of a convex-lens grain prepared by FIB showed a patched contrast (Fig. 5d), suggesting that the grain is not a single crystal. However, the dispersion of the crystallographic orientation inside the grain was within a certain range when plotted as a stereo projection (Fig. 5e) by obtaining an SAED pattern from several points, as we did for the asteriscus. Figure 5e shows that the c-axis is almost parallel to the axis of the lens. Moreover, <100> (expressed as [100] in Fig. 5e) is dispersed within a range of ∼ 20°, smaller than 60° expected for uniaxially oriented aggregation along the c-axis. In other words, the grain has a single crystal character with mosaicity, which produces the patched diffraction contrast shown in Figure 5d. The patch size in Figure 5d corresponds to that of a single crystal or the CSD for the diffraction phenomena, and this size is considerably smaller than that in the asteriscus (Fig. 4), which probably reflects the difference in the peak width in the PXRD pattern (Fig. 1). The SAED pattern along <210> (Fig. 5f) is almost identical to that of the asteriscus (Figs. 4d and 4f), showing extra diffraction rows from the supercell, and the features of the stacking faults are also similar (Fig. 5g).

Figure 5. (a)-(c) several SEM images of synthetic vaterite (S-1), showing a grain morphology like a convex lens or an aggregate of them. (d) TEM BF image of the cross-section of a convex-lens shaped grain prepared by FIB. (e) Oriental distribution of [100] and [001], obtained from SAED patterns taken from a number of points in the grain in (d), and displayed as a stereo projection. (f) SAED pattern from the point f in (d), corresponding to that of vaterite along <210>. (g) Lattice image viewed along <210>, showing dense stacking faults along the c-axis.

The synthetic vaterite particles precipitated from the extremely supersaturated solution (S-2) exhibit an isotropic morphology with an uneven surface (Figs. 6a and 6b). Cross-sectional TEM imaging of single particles thin foiled using FIB showed that voids were common inside the particles, and it appeared that the whole particle consisted of smaller nanoparticles, corresponding to the surface unevenness (Fig. 6c). The SAED pattern from the whole particle (Fig. 6d) is explained by vaterite viewed along <100>, but all diffraction spots are considerably arced, indicating that the particle consists of small crystallites with close crystal orientations, showing a ‘mosaic’ character. The small size of the crystallites reflects the widest peak width in the PXRD pattern of this sample compared to the others.

Figure 6. (a) and (b) SEM images of the particles of another synthetic vaterite (S-2), showing a rather isotropic morphology. (c) TEM BF image of a single particle thinned by FIB. Opaque material around the particle is tungsten coating to prevent Ga ion implantation. (d) SAED pattern from the whole of the particle in (c). The pattern is explained by the reciprocal lattice of vaterite viewed along <100> but all spots are arced, indicating mosaicity in the particle.

DISCUSSION

A major theme in biomineralization research is how organisms are involved in the formation of inorganic materials and induce characteristic structures of biominerals. These characteristic structures can appear at various levels within the structural hierarchy of biominerals, including crystal structures, defects and local structures, crystal shape and orientation, and the texture of polycrystals. As an example of the distinguishing crystal structures of biominerals, Pokroy et al. (2007) showed that the cell parameters of aragonite in some molluscan shells are slightly different from those of geological aragonite, which they attributed to the introduction of lattice strain induced by organic matter embedded in the crystals. As examples of crystal defects, Suzuki et al. (2012) and Kogure et al. (2014) found that the twins in the aragonite of some shells were much denser than those in geological aragonite. Okumura et al. (2010) showed that calcite crystals in the prismatic layer of certain molluscan shells are not simply crystalline but have a mosaic property with organic matter located at small-angle grain boundaries.

Regarding vaterite in the goldfish asterisci revealed here, mineralogical characteristics distinct from synthetic ones were not clearly observed, except for the size of the single crystals or CSD. The cell parameters for the sub-cell of vaterite determined by PXRD were almost identical, especially between vaterite and S-1. The superstructure and stacking disorder appearing in PXRD, electron diffraction, and high-resolution lattice imaging were also similar between the asteriscus and the two synthetic vaterites. The present study revealed mosaicity with a small misorientation, instead of the formation of a large single crystal, in both natural and synthetic vaterites, although their single crystal sizes are considerably different. The difference in crystal size may be attributed to the degree of supersaturation—the asterisci grew far slower at lower supersaturation than during synthesis. The formation of a mosaic structure without the formation of a large single crystal may be an intrinsic characteristic of vaterite, which is absent in calcite and aragonite.

Biogenetic and synthetic vaterites with such mosaic features may correspond to mesocrystals proposed in recent literature (e.g., Cölfen and Antonietti, 2005). However, in our interpretation, the term mesocrystals includes not only mosaicity with small crystal misorientation, but also their forming scheme by oriented attachment of pre-existing fine crystallites (De Yoreo et al., 2015). There was no evidence that the asteriscus was formed from preexisting crystallites. In contrast, the radiation structure (Figs. 2 and 3) and intricate boundaries of the single crystals (Fig. 4) cannot be explained simply by the oriented attachment of the preexisting crystallites.

Polarized microscopy and EBSD mapping indicated that vaterite in the goldfish asterisci grew radially from the central region. Some studies on the sagitta—an otolith composed of aragonite—have found that aragonite fibers also grow radially from the center (Parmentier et al., 2007; Stolarski et al., 2023). In contrast, all otoliths in sturgeons are composed of single crystals of small granular vaterite bound together by calcite (Chakoumakos et al., 2016). Therefore, the growth process of the goldfish asterisci is closer to that of the saggita found in various fish species, despite the differences in crystal polymorphs. From the EBSD analysis, the slender elements growing radially exhibited an extension normal to the c-axis. Conversely, the convex lens-like grains of the synthetic vaterite have a c-axis along the lens axis. This type of growth, where the a_i-plane or c-face of the vaterite expands, has been commonly reported in studies on synthesized vaterite. Wang et al. (2015) synthesized vaterite using a gas diffusion method in the presence of sodium citrate and sodium dodecyl benzenesulfonate. The resulting crystals displayed a tablet-like shape with hexagonal faces corresponding to the c-face of vaterite. Similarly, Hu et al. (2012) reported that vaterite assumes a tablet- or petal-like shape, consisting of minute hexagonal crystals. In this case, the hexagonal faces corresponded to the c-face. Accordingly, the expansion along the a_i-plane is a crystal habit of vaterite, and the orientation of the crystals in the asteriscus is considered to be the result of geometrical selection. However, some studies have shown that the vaterite grows along the c-axis. According to Kabalah-Amitai et al. (2013), the elongation direction of the sea squirt spicule is along the c-axis of the vaterite.

Accordingly, it is natural to suggest that the selection of vaterite in goldfish asterisci is determined by its initial formation stage. Therefore, detailed future analyses of the vicinity of the center of the otoliths are required to understand the polymorph selection mechanism.

ACKNOWLEDGMENTS

We are grateful to Dr. Toru Yoshino and Mr. Yota Mori for providing synthetic vaterite specimens. We also thank the two anonymous reviewers for their valuable comments which improved the original manuscript considerably. This research was supported by the JST SPRING (Grant Number JPMJSP2108) and JSPS KAKENHI (Grant Numbers JP19H05771, JP22H01340, and JP23H00339).

REFERENCES

• Allmendinger, R.W., Cardozo, N. and Fisher, D.M. (2011) Structural geology algorithms: Vectors and tensors in structural geology. Cambridge University Press, Cambridge.
•  Cardozo,  N. and  Allmendinger,  R.W. (2013) Spherical projections with OSXStereonet. Computers and Geosciences, 51, 193-205.
•  Chakoumakos,  B.C.,  Pracheil,  B.M.,  Koenigs,  R.P.,  Bruch,  R.M. and  Feygenson,  M. (2016) Empirically testing vaterite structural models using neutron diffraction and thermal analysis. Scientific Reports, 6 (July 2018).
•  Chen,  Z.,  Omori,  Y.,  Koren,  S.,  Shirokiya,  T., et al (2019) De novo assembly of the goldfish (Carassius auratus) genome and the evolution of genes after whole-genome duplication. Science Advances, 5, eaav0547.
•  Christy,  A.G. (2017) A Review of the Structures of Vaterite: The Impossible, the Possible, and the Likely. Crystal Growth and Design, 17, 3567-3578.
•  Cölfen,  H. and  Antonietti,  M. (2005) Mesocrystals: Inorganic superstructures made by highly parallel crystallization and controlled alignment. Angewandte Chemie International Edition, 44, 5576-5591.
•  De Yoreo,  J.J.,  Gilbert,  P.U.,  Sommerdijk,  N.A.,  Penn,  R.L., et al (2015) Crystallization by particle attachment in synthetic, biogenic, and geologic environments. Science, 349, 6247.
•  Falini,  G.,  Fermani,  S.,  Vanzo,  S.,  Miletic,  M. and  Zaffino,  G. (2005) Influence on the formation of aragonite or vaterite by otolith macromolecules. European Journal of Inorganic Chemistry, 1, 162-167.
•  Gal,  J.Y.,  Bollinger,  J.C.,  Tolosa,  H. and  Gache,  N. (1996) Calcium carbonate solubility: A reappraisal of scale formation and inhibition. Talanta, 43, 1497-1509.
•  Hu,  Q.,  Zhang,  J.,  Teng,  H. and  Becker,  U. (2012) Growth process and crystallographic properties of ammonia-induced vaterite. American Mineralogist, 97, 1437-1445.
•  Kabalah-Amitai,  L.,  Mayzel,  B.,  Kauffmann,  Y.,  Fitch,  A.N., et al (2013) Vaterite crystals contain two interspersed crystal structures. Science, 340, 454-457.
•  Kamhi,  S.R. (1963) On the structure of vaterite CaCO₃. Acta Crystallographica, 16, 770-772.
•  Kamiya,  N.,  Kagi,  H.,  Tsunomori,  F.,  Tsuno,  H. and  Notsu,  K. (2004) Effect of trace lanthanum ion on dissolution and crystal growth of calcium carbonate. Journal of Crystal Growth, 267, 635-645.
•  Kim,  I.W.,  Robertson,  R.E. and  Zand,  R. (2005) Effects of some nonionic polymeric additives on the crystallization of calcium carbonate. Crystal Growth and Design, 5, 513-522.
•  Kogure,  T.,  Suzuki,  M.,  Kim,  H.,  Mukai,  H., et al (2014) Twin density of aragonite in molluscan shells characterized using x-ray diffraction and transmission electron microscopy. Journal of Crystal Growth, 397, 39-46.
•  Mori,  Y.,  Enomae,  T. and  Isogai,  A. (2009) Preparation of pure vaterite by simple mechanical mixing of two aqueous salt solutions. Materials Science and Engineering C, 29, 1409-1414.
•  Mugnaioli,  E.,  Andrusenko,  I.,  Schüler,  T.,  Loges,  N., et al (2012) Ab initio structure determination of vaterite by automated electron diffraction. Angewandte Chemie - International Edition, 51, 7041-7045.
•  Okumura,  T.,  Suzuki,  M.,  Nagasawa,  H. and  Kogure,  T. (2010) Characteristics of biogenic calcite in the prismatic layer of a pearl oyster, Pinctada fucata. Micron, 41, 821-826.
•  Parmentier,  E.,  Cloots,  R.,  Warin,  R. and  Henrist,  C. (2007) Otolith crystals (in Carapidae): Growth and habit. Journal of Structural Biology, 159, 462-473.
•  Pokroy,  B.,  Fitch,  A.N. and  Zolotoyabko,  E. (2007) Structure of biogenic aragonite (CaCO₃). Crystal Growth and Design, 7, 1580-1583.
•  Popper,  A.N.,  Ramcharitar,  J. and  Campana,  S.E. (2005) Why otoliths? Insights from inner ear physiology and fisheries biology. Marine and Freshwater Research, 56, 497-504.
•  Pracheil,  B.M.,  Chakoumakos,  B.C.,  Feygenson,  M.,  Whitledge,  G.W., et al (2017) Sturgeon and paddlefish (Acipenseridae) sagittal otoliths are composed of the calcium carbonate polymorphs vaterite and calcite. Journal of Fish Biology, 90, 549-558.
•  Pracheil,  B.M.,  George,  R. and  Chakoumakos,  B.C. (2019) Significance of otolith calcium carbonate crystal structure diversity to microchemistry studies. Reviews in Fish Biology and Fisheries, 29, 569-588.
•  Qiao,  L.,  Feng,  Q.L. and  Li,  Z. (2007) Special vaterite found in freshwater lackluster pearls. Crystal Growth and Design, 7, 275-279.
•  Ren,  D.,  Feng,  Q. and  Bourrat,  X. (2013) The co-effect of organic matrix from carp otolith and microenvironment on calcium carbonate mineralization. Materials Science and Engineering C, 33, 3440-3449.
•  Söllner,  C.,  Burghammer,  M.,  Busch-Nentwich,  E.,  Berger,  J., et al (2003) Control of crystal size and lattice formation by starmaker in otolith biomineratization. Science, 302, 282-286.
•  Stolarski,  J.,  Drake,  J.,  Coronado,  I.,  Vieira,  A.R., et al (2023) First paleoproteome study of fossil fish otoliths and the pristine preservation of the biomineral crystal host. Scientific Reports, 13, 1-11.
•  Suzuki,  M.,  Kim,  H.,  Mukai,  H.,  Nagasawa,  H. and  Kogure,  T. (2012) Quantitative XRD analysis of {110} twin density in biotic aragonites. Journal of Structural Biology, 180, 458-468.
•  Wang,  Y.Y.,  Yao,  Q.Z.,  Li,  H.,  Zhou,  G.T. and  Sheng,  Y.M. (2015) Formation of vaterite mesocrystals in biomineral-like structures and implication for biomineralization. Crystal Growth and Design, 15, 1714-1725.
•  Wightman,  R.,  Wallis,  S. and  Aston,  P. (2018) Leaf margin organisation and the existence of vaterite-producing hydathodes in the alpine plant Saxifraga scardica. Flora: Morphology, Distribution, Functional Ecology of Plants, 241, 27-34.