Structure and distribution of chalky deposits in the Pacific oyster using x-ray computed tomography (CT)

Oysters are unusual among bivalves in that they possess chambers, often filled with soft, chalky calcite, that are irregularly scattered throughout the shell. Because the function of these so-called chalky deposits is still unclear, evaluating the growth and distribution of chalk is important for elucidating the ecological function of this unique shell trait. Specimens of the Pacific oyster Magallana gigas, an oyster well known for chalk expression, were grown in Bodega Harbor, Bodega Bay, CA. At the end of an 11 month growing period, specimens were culled and selected animals were submitted for x-ray computed-tomography imaging. Three-dimensional reconstructions of oyster shells were used to assess the overall distribution of chalk, and also to better understand the relationship between chalk and other structures within the shell. Results indicate that chalky deposits underly sculptural features on the shell exterior, such as external ridges and changes in growth direction, and also that there is a relationship between chalk formation and oyster processes of cementation. Overall, chalk is useful for a cementing lifestyle because it enables morphological plasticity needed to conform to irregular substrates, but also acts as a cheap building material to facilitate rapid growth.

Shape of typical shells. M. gigas specimens CG1-A01, CG1-A02, CG1-A03, and CG1-A04 were chosen for CT imaging because they possessed a shell shape and other morphological characteristics that were commonly observed in the oysters grown in Bodega Harbor (Fig. 2). On average, shells submitted for scanning were 59.0 ± 6.2 mm (n = 5) from umbo to commissure (i.e. shell length), and 46.4 ± 8.1 mm wide (n = 5), where width was measured at the widest point perpendicular to the shell length. Inflation of shells, measured as the maximum distance left valve and the hinge line between the umbo and the commissure, averages 14.2 ± 3.1 mm (n = 5).
Valves were approximately 2-3 mm thick near the commissure before tapering to a fine edge, but were thicker towards the umbo, which was 5-6 mm thick in most specimens. The number and size of ridges expressed on shell exteriors varies widely amongst the oysters. For example, CG1-A01 displays six ridges, 1-2 mm in width, that are easily identifiable on the shell surface ( Fig. 2A,B). In contrast, CG1-A02 does not have any ridges that contact underlying shell layers, but has pronounced growth breaks where the overhanging shell layer terminates in ridges (Fig. 2C,D). CG1-A03 has just three ridges on the shell exterior, and only has distinct growth breaks near the commissure (Fig. 2E,F). Ridges within a single specimen also vary in length, and may extend over the full length of the shell, or end closer towards the umbo. Changes in growth direction occur both as growth breaks, which result in higher inflation of valves, and as a change in the primary growth direction of the valve. The latter phenomenon yields a curved shell shape, which is exemplified by specimen CG1-A02 that has a crescent shaped umbo, indicating a change in growth direction early in ontogeny (Fig. 2C).  www.nature.com/scientificreports/ Shape of recemented shells. Shells that recemented onto a hard substrate, such as another oyster or cage mesh material, display much greater variability in shell morphology than their counterparts that remained unattached throughout ontogeny. The commissures of animals that recemented are still generally ovate, but may also possess outgrowths that make the shell margin outline highly irregular (Fig. 2J). In addition, while the commissure of specimens CG1-A01 through CG1-A04 lie primarily in the same plane, the shell margin of CG1-A05 does not and the contour of the commissural margin may be convex or concave. Finally, recemented shells tend to retain the shape of the object onto which they were growing. For example, CG1-A05 retains the waffle-like pattern of the cage mesh onto which it had recemented (Fig. 2I). This texture, a result of growing onto cage mesh, can also be seen in specimen CG0-A01 (Fig. 2K,L). Though CG0-A01 was not submitted for CT scanning, it represents an excellent example of the morphological plasticity exhibited by specimens that grew onto another substrate. Cross sections of the suture zones of recemented oysters, specimen CG1-005 and two additional pairs of animals, showed that while chalk was abundant where oysters grew onto one another, there was no chalk laid down outside of an individuals' periostracum ( Fig. 3).
3D visualization of shells. The calcitic folia are a major component of these oyster shells, and define the structure of the external features. For example, folia define growth breaks (Fig. 4), and ridges are supported by complex structures composed of folia, creating the sculptures present on the shell exterior (Fig. 5). Three-  . Smaller chalky deposits are present irregularly throughout shells, including on the interior shell surface. Significant chalk formation occurs in the umbo (Fig. 6), beneath ridges (Fig. 5), and underlying growth breaks (Fig. 4) and other changes in growth direction, such as the curved umbo in CG1-A02. Chalky deposits in the umbo are generally lens shaped and are fully surrounded by folia, apparently sealing them within the shell (Fig. 6).

Discussion
Morphology and distribution of chalk and folia The timing of chalk deposition in relation to folia precipitation has been a long standing question in regards to chalky deposit formation. For bivalves in general it is well recognized that elongation is achieved by the addition of new material in the spiral direction (i.e. to the commissural margin), and shell thickening occurs by the addition of carbonate to the interior shell surface, beneath (i.e. interior to) old www.nature.com/scientificreports/ shell material 17 . Chalky deposits observed in 3D reconstructions were found interior and ventral to folia, and most were completely sealed within the shell by surrounding folia. This was true for chalk found in the umbo, as well as chalky deposits that formed directly beneath growth breaks (Figs. 4,6), though it remains unclear if calcification continues in chambers after they are sealed, as has been hypothesized previously 23 . Although not common, examination of 3D reconstructions also revealed the presence of chalky deposits that were not completely sealed within the shell. These appear to be the result of a small amount of chalk that was precipitated onto folia at the commissure, which was not completely sealed when the next layer of folia was deposited, leaving a small wedge of chalk exposed to the shell exterior (Fig. 5C). The position of folia as exterior to chalk indicates that folia, not chalk, is precipitated as the outermost shell layer that underlies the periostracum, and consequently provides structural support to the shell, as has been noted in previous work 29 . In addition, the deposition of folia produces external features, such as ridges, growth breaks, and other changes in growth direction, thus creating depressions on the interior shell surface that are subsequently filled with chalk. The pattern of chalk formation beneath growth breaks and other external features confirms previous observations that chalk fills in depressions on the interior surfaces of shells 18,20,24,26,27,30,31 , but also provides context for how these depressions form. For oysters, growth breaks represent a significant change in growth direction, and are important in the context of chalk formation because these features are often associated with chalky deposits. Therefore, a more complete understanding of how growth breaks form can help elucidate the function and timing of chalk growth. Based on the observations regarding the relationship between chalk and folia, there are three possible models for shell formation at growth breaks, which all have different implications for the function of chalk in oyster shells, and also the behavior of the mantle during shell formation more generally (Fig. 8).
Model 1 First, the mantle lays down the periostracum (the outermost organic layer of the shell) which is followed by deposition of a thin layer of folia directly beneath, towards the shell interior. Next, the mantle separates from the previously formed shell, and without retracting significantly, begins forming a layer of periostracum and folia at an oblique angle to the previous growing plane, creating a growth break. This new geometry creates a depression on the interior shell surface, directly below where the second layer of periostracum and folia branched off from the first. This depression is subsequently filled with chalk. As the oyster continues to thicken and extend its shell outward, this chalky deposit beneath the growth break is sealed within the shell with a new layer of folia. In Model 1, the growth break is caused by the mantle detaching from the initial layer of periostracum, which may be caused by breakage of the periostracum at the commissure, or some other interruption (Fig. 8: row 1).  Figure 5. View of specimens as observed in 3D-visualizer used to exemplify chalk as it relates to external radial ridges. See Fig. 4 for a description of the individual panels. Here, red arrows indicate the position of chalky deposits under external radial ridges. The black arrow in C indicates a small chalky deposit near the umbo that was not completely sealed by folia, and remains exposed to the exterior environment. www.nature.com/scientificreports/ Model 2 First, the mantle lays down the periostracum, which is followed by deposition of a thin layer of folia directly beneath the external organic layer. Next, the mantle detaches from the currently forming layer of periostracum and retracts to a point along the previously formed shell. Then, the mantle begins forming a new layer of periostracum and folia, inward from the previous layer, at an angle oblique to the previously growing plane. This new geometry creates a depression on the interior shell surface, directly interior to where the second layer of periostracum and folia branched off from the first. This depression is subsequently filled with chalk. As the oyster continues to thicken and extend its shell outward, this chalky deposit beneath the growth break is sealed within the shell by a new layer of folia. In Model 2, the growth break is also caused by the mantle detaching from the initial layer of periostracum, which may be caused by breakage of the periostracum at the commissure, or some other environmental interruption. This scenario differs from Model 1 in the behavior of the mantle. In Model 2, the mantle retracts and builds new shell material inward from the previous growing plane. In contrast, in Model 1, the mantle does not retract; rather, it moves toward the interior space between valves and precipitates the next layer while remaining extended ( Fig. 8: row 2).
Model 3 First, the mantle lays down the periostracum, which is followed by deposition of a thin layer of folia directly beneath the external organic layer. Next, the mantle precipitates a small mound of chalk that extends to the outermost extent of the previously grown shell. The mantle detaches from the currently forming layer of periostracum and retracts to a point along the previously formed shell. Then, a new layer of periostracum and folia is precipitated interior to the initial chalk mound, and extends past this chalky deposit, thus leaving the chalk exposed to the shell exterior. This new growth break also creates a depression on the interior shell surface, directly below the point where the first folia, the umbo end of the chalk mound, and the second layer of folia and periostracum intersect. This space is subsequently filled with chalk, and as the oyster continues to thicken and extend its shell outward, the chalky deposit beneath the growth break is sealed within the shell. Here, the formation of the growth break is not driven by the separation of the mantle from the first layer of periostracum. Instead, the change in growth direction is a result of the formation of the chalk mound, which directs new growth at an angle oblique to the previous growing plane. For shells observed in this study, growth breaks are not associated with chalky deposits (i.e. the initial chalk mound described here) that are exposed to the shell exterior, as in Model 3. Therefore, we can exclude Model 3 as an explanation for growth breaks, and the change in shell growth direction immediately following a growth break (Fig. 8: row 3).
Models 1 and 2 are more difficult to differentiate based on shell morphology alone because as described, they result in the same, or very similar, set of features, and the major differences between these two models concerns the behavior of the mantle during growth break formation. However, other lines of evidence can be used to parse out which model is more likely the to produce the shell features observed. For Model 1, the mantle does not retract and instead moves away from the previous growing plane by pivoting away from the old shell. Therefore, the new layer of periostracum and folia should be the same length as the initial layers, because the mantle has www.nature.com/scientificreports/ remained fully extended. In contrast, if the mantle retracts (Model 2) and builds the second layer of periostracum outward from old shell material, the newer layers will be shorter (i.e. terminate closer to the umbo than the first layer of folia) until enough growth has occurred to extend new folia past the old shell material. This second pattern has been observed in the cross section in Fig. 1D. Therefore, Model 2, which includes mantle retraction, is the most likely representation of how growth breaks form and how the mantle behaves in this context. In addition, Model 2 agrees well with previous work in which oyster mantles were directly observed to contract during shell formation and cementation 32 . In an experiment in which the shell margin of the left valve of an oyster was removed, the mantle subsequently began forming a new layer of periostracum on old shell material, dorsal to the commissure ( 32 , Fig. 7B). This illustrates that the mantle does retract in order to form new periostracum and shell outward from old shell material consistent with Model 2. Additional observational work, such as that performed by Yamaguchi 32 , coupled with careful measurements to establish how far back the mantle retracts during growth break formation, would yield definitive evidence to support Model 1 or Model 2. Chalky deposits and cementation Chalky deposits were observed to be extensive at sites where shells reattached to a substrate, such as where oysters cemented to other individuals or cage mesh material (Figs 2L, 7). For oysters that recemented onto cage material, the periostracum was the outermost layer, indicating it was laid First, the mantle lays down the periostracum, followed by an initial layer of folia. Next, the mantle separates from the previously formed shell (black arrow), and without retracting, begins forming a new layer of periostracum and folia, creating a growth break. Last, a chalky deposit is formed in the interior depression resulting from the previous two steps. Model 2 (row 2): First, the mantle lays down the periostracum and the initial layer of folia. Next, the mantle retracts (black arrow), and forms a new layer of periostracum followed by the next layer of folia. Last, chalk is precipitated in the interior depression resulting from the previous two steps. Model 3 (row 3): First, the mantle lays down the periostracum and the initial layer of folia.
Next, a small chalky deposit forms near the commissure. The new layer of periostracum and folia is formed at an angle oblique to the previous growing plane because of the presence of the chalky deposit. Last, another chalky deposit is formed on the interior depression resulting from the previous two steps. Schematic of shell and mantle adapted from Yamaguchi 32 . www.nature.com/scientificreports/ down first to conform to the irregular surface. Similar observations were also made by Yamaguchi 32 , who also proposed that the mantle is an active participant in shell cementation by pressing newly made periostracum into the substrate, thus promoting attachment. Therefore, chalky deposits are involved in cementation at least to the extent that chalk allows the oyster to conform to an uneven substrate while maintaining a favorable internal space. Interestingly, Harper 33 recorded the presence of a chalky deposit outside of an M. gigas shell, between the periostracum and the substrate. If the presence of chalk between the shell and the substrate is a common occurrence, then this would indicate that chalk plays a more central role in cementation than simply conferring the ability to conform to an uneven substrate. In the present study, three pairs of oysters that had recemented to one another were sectioned and examined for these chalky deposits occurring outside of the periostraca (Fig. 3). However, none were found. Although this may indicate that a larger sample size is required to survey for this phenomenon, we propose that external chalky deposits, such as that found by Harper 33 , were aberrant. This indicates that chalk is generally not directly used for oyster attachment. But, how was the chalk observed by Harper 33 emplaced between the periostracum and the substrate?
While the mantle pressing of the periostracum to the substrate may be one component of the cementation process for M. gigas 32 , a mantle-produced organic-inorganic adhesive plays a central role in attaching oysters to the substrate. Harper 33 hypothesized that the organic fraction of the cement is able to move through the permeable periostracum to fill the space between this layer and the substratum, thus attaching the oyster. There is additional evidence to suggest that mantle secreted organics play a central role in determining crystal morphology in molluscan shells [34][35][36][37][38][39] . Thus, if the mantle-secreted organic precursors to chalky calcite were inadvertently leaked through the porous periostracum, this may explain the present of chalk in a space that would normally be occupied by cement 33 . While this remains a speculative hypothesis, additional work characterizing the organic constituents of shell layers, and addressing the ability of these organics to mobilize through the periostracum, would help to resolve these questions.
Functional ecology of chalk Data presented here indicate that chalk is associated with morphological variability in M. gigas, and is useful for cementation because it allows the animal to conform to an irregular substrate while maintaining a favorable internal space. Although many other bivalves are able to achieve a cementing lifestyle without the use of chalk, the high porosity of chalk produces a shell structure that is much less dense than surrounding folia 19,40 . In addition, chalk formation is associated with rapid growth rates, and very large, thick shells 10,23 . This indicates that chalk, in addition to being useful for cementation and accommodating morphologic plasticity, enables rapid shell growth by acting as a relatively cheap building material during shell construction. Future work comparing the strength or durability of cementation in oysters that produce chalk versus oysters and other cementing bivalves that do not (e.g. Chamids) would provide additional insight into the tradeoffs associated with using this material to aid cementation. Last, although recent work suggests that folia and chalk are precipitated synchronously 20,41 , open questions remain regarding the timing of folia versus chalk deposition within the shell. Additional experiments employing both marking techniques (e.g. calcein) and stable isotope geochemistry would help to determine the temporal relationship between chalk and folia, particularly as it relates to the formation of growth breaks and other features discussed here.

Methods
Study organism, oyster culturing, and environmental setting. For this study, chalky deposit formation was investigated in M. gigas, which is commonly farmed on the coast of California. All specimens of M. gigas obtained for this experiment were spawned by by Hawaiian Shellfish LLC. In nature, oyster larvae undergo a metamorphosis during which they settle to the seafloor as spat (juveniles) and cement themselves to a hard substrate, which may either be a rocky substrate or another oyster 42 . Aquaculture practices regularly sort and separate juvenile oysters by size, yielding singles (i.e. individual oysters that are unattached to other individuals) that are suitable for commercial purposes. When settled juveniles reached a size of 1.4 mm (+/− 0.2 mm) they were placed on ice and shipped overnight to the Starbird Mariculture, Inc facility in Bodega Harbor, Bodega Bay, CA (38 • 19'41.56"N, 123 • 03′22.61″W). Oysters were grown in the Floating Commercial Upweller System (FLUPSY) at Starbird Mariculture Inc. until they reached a size of 4-5 mm (approximately 4-7 months old). At this time, oysters used for this experiment were separated and moved to mesh cages that were secured to a dock in Bodega Harbor immediately adjacent to the FLUPSY and all specimens were initially singles (i.e. unattached to other individuals). Oysters were monitored and cleaned of epibionts approximately every 2-3 weeks. The permits necessary for outplanting oysters and using the following protocols were obtained from the California Department of Fish and Wildlife.
Bodega Harbor is a relatively shallow bay on the California coast that is protected from wave action by two parallel jetties at its entrance. The harbor is well ventilated and is almost completely flushed during each tidal cycle. Bodega Harbor receives little freshwater input from April to November, and is periodically exposed to nutrient rich water during strong upwelling events 43 . Computed tomography. Oyster specimens that were submitted for computed tomography (CT) scanning were moved from the FLUPSY to synthetic mesh cages on June 19, 2016. The five specimens listed in Table 1 were sacrificed on May 5, 2017 and submitted for CT analysis thereafter. Specimens CG1-001 through CG1-004 were singles that were selected because they displayed shell morphology that was representative of the rest of the population. Over the course of this experiment it became apparent that oysters were able to secondarily cement to a hard substrate long after the initial settlement phase, including to other oysters and the cages themselves. CG1-005 was selected as an example of two singles that cemented onto one another and cage material, yielding a specimen with two individuals. A separate oyster cohort was outplanted prior to the primary group for a separate experiment and reared under the same conditions until animals were culled. One specimen from this group, Scientific RepoRtS | (2020) 10:12118 | https://doi.org/10.1038/s41598-020-68726-4 www.nature.com/scientificreports/ CG0-A01, was used for the present study because it also exemplified the effect of cementation on shell morphology. However, this specimen was not submitted for CT-scanning because it was too large for the scanner. X-ray tomographic images were obtained on the Zeiss Xradia 520 Versa (Carl Zeiss Microscopy, Inc.) with the Flat Panel extension at the UC Davis Center for Molecular and Genomic Imaging (CMGI). This produced 46 um x 46 um x 46 um voxels. Center for Active Visualization in the Earth Sciences (http://KeckCAVES.org) open source software for 3D visualization was used to view scans, specifically Vrui 44 and 3DVisualizer 45 . Recemented shell cross sections. To characterize attachment of oysters that recemented to one another (i.e. non-singles), specimen CG1-005 and two additional oysters randomly selected from the same cohort that were not submitted for CT scanning were sectioned through the suture zone between two individuals. This was done using a Buehler IsoMet low-speed saw. The cut surface was ground using a 600 grit diamond wheel, and the shell-half was mounted onto a large slide (51mm × 75mm) using Hillquist AB thin section epoxy. Shells were then re-sectioned using Buehler PetroThin thin section saw, resulting in mounted shell sections that were approximately 500-700 µ m thick. Sections were digitized using a scanner.