Light Microscopy.

Polarized Light Microscopy: Standard thin sections (∼ 30 μm) were prepared using HERMES water grinding papers (successively P1200/P2500 and P4000 grain sizes) and observed in transmission using polarized light, as well as polarized and analyzed light (using crossed polarizers).

Fluorescence Microscopy: For epifluorescence observations, sample sections were polished using P1200, P2500, and P4000 HERMES water grinding papers, 3 and 1 μm Buehler diamond polycrystalline suspensions, and aluminum oxide suspension (∼ 300 nm size). Surfaces were carefully rinsed with deionized water. Two samples were then softly etched for 10 s in a formic acid solution (0.1 wt.%) with glutaraldehyde (5 wt.%), and colored with acridine orange for 30 min (20 mg mixed in 95 ml of water and 5 ml of methanol). Observations were made using a Zeiss Universal microscope equipped with x6 (NA = 0.2) and x25 (NA = 0.6) fluorite objectives, a mercury lamp, a 365 nm excitation filter (UV), and a 420 high-pass emission filter. For observation of acridine-stained samples, a 435 nm excitation filter (blue) and a 510 nm high-pass emission filter were used, as the emission wavelength of this fluorophore varies from green to red depending on its environment (see details in the discussion section).

Confocal Laser Scanning Microscopy: Samples were prepared as for epifluorescence microscopy. Observations were carried out on an Olympus FV-1000 inverted confocal microscope (located at LIONS laboratory, CEA-Saclay, France). The sample surface was excited with a green He-Ne laser (λ = 488 nm) and emission light was collected through a band pass BA 560 IF filter. 20x (Olympus UPLSAPO, NA = 0.75) and 40x (Olympus UPLFLN, NA = 0.75) objectives were used, with a confocal aperture set at 80 and 130 μm, respectively. All scans were focused ∼ 2 μm under the sample surface, to avoid any potential surface contamination or polishing remains.

Raman mapping was conducted using a WITec alpha300R (WITec GmbH, Germany) confocal Raman microscope. Scans were performed using a piezoelectric scanner table with a maximum scan range of 200μm × 200μm and a minimum lateral step size of 4 nm and a vertical step size of 0.5 nm. The sample surface was excited with a green diode laser (λ = 532 nm), and the Raman signal was collected with a dedicated ultra-high throughput spectrometer (UHTS 300, WITec, Germany), using a grating, 600/mm and 500 nm blaze. A Nikon 100x (NA 0.9) objective was used. The spectral analysis and imaging processing was performed using the WITecProject software (version 2.04, WITec GmbH, Germany). Peak positions were determined using the “Mulipeak Fitting 2” routine of IGOR Pro using a Gauss shape for the fitting (version 6.11, WaveMetrics, Inc. USA). To identify calcite, we used calcite peak T_c at 155 cm⁻¹ (translation mode) and L_c at 282 cm⁻¹ (librational mode) and the two internal modes (in-plane band ν₄ at 711 cm⁻¹ and symmetric stretch ν₁ at 1085 cm⁻¹).

Scanning Electron Microscopy.

For SEM, several freshly broken, unetched radial shell fragments were prepared. A staining assay was conducted by putting one of the fragments in a sealed container next to an OsO₄ solution (4% vol. in H₂O) for 1h. This compound is commonly used to stain organic components in TEM preparations. Even though no obvious staining was observed in our case, contact with the vapor was still found to perform very faint etching at a submicrometric scale. Observations were carried out with two microscopes: 1/ a Philips XL30 SEM using a secondary electrons collector (located at IDES UMR 8148). In this case, the SEM was operated at 25 keV and 10 mm working distance and the samples were coated with Au/Pd. 2/ a Carl Zeiss Ultra Plus Field Emission SEM using a secondary electrons collector (located at the mineralogy department of NHM, London). This FEG-SEM was operated at 1 keV and a 4 mm working distance with an aperture of 10 μm. Samples were not coated.

Electron Microprobe.

Samples were included in epoxy resin, cut, and polished using HERMES water grinding papers (grain size P1200/P2500 to P4000) and Buehler diamond polycrystalline suspensions (3 μm and 1 μm). Surfaces were carefully rinsed with deionized water and then carbon-coated. Ca, Mg, Sr, and S distributions were mapped using wavelength-dispersive spectrometry on a Cameca SX100 electron microprobe (at the Natural History Museum of London, UK). The microprobe was operated at 15 kV accelerating voltage with a 20 nA specimen current.

Atomic Force Microscopy.

For AFM observations, samples were polished using HERMES water grinding papers (grain size P1200/P2500 to P4000), Buehler diamond polycrystalline suspensions (3 μm and 1 μm), and finally aluminum oxide suspension (300 nm). The obtained surface was very gently cleaned using a diluted formic acid solution (0.1 wt.%) with glutaraldehyde (3 wt.%) for 1 s, to remove pollution or dust generated by polishing. As shown in existing studies [66], no major artifact is induced by such a process. AFM observations were carried out on a Veeco Dimension 3100 operated in tapping mode, in which height and phase error signals are recorded. The phase error signal gives insights into surface properties: the phase lag between the period of oscillation imposed to the cantilever and the recorded period, at each point, is due to interactions between the tip and the sample surface. Strong phase-lag contrasts therefore reveal contrasts in elastic deformation or surface adhesion effects.

Modern Patella vulgata.

The organization of the shell margin cross-section is well illustrated using PLM (Fig 4a). The M+3 layer is thin and displays a radial crossed-foliated pattern and the m+2 layer makes up the bigger part of the shell, with a very regular concentric crossed-foliated pattern. In polarized light, growth increments are very well marked and regular, and perfectly continuous through consecutive first order lamellae (Fig 4b). The use of crossed polarizers (Fig 4c) allows us to determine the orientation of the first lamellae. The simultaneous extinction of one out of two lamella is evidence that these lamellae present a similar orientation, whereas two consecutive lamellae have distinct orientations (different intensities). Second order lamellae are also clearly visible (black arrows), as well as the individual third order rods (white arrow), when they are faintly disorientated within an extinct first order lamella.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 4. Microstructural organization of modern Patella vulgata calcite crossed-foliated outer layers.

a) PLM view (polarized and analyzed light) of a radial thin section. Outer surface is at bottom. b) PLM view (polarized light) showing the growth increments (green arrows) between two consecutive 1^st order lamellae (I). c) PLM view (polarized and analyzed light) of three consecutive 1^st order lamellae (I), displaying the preserved alternate orientation one lamella on two. 2^nd order lamellae are visible (black arrows), as well as the individual, faintly disoriented 3^rdorder rods (white arrow). d-g) Electron micropobe maps. d) Backscattered image of the scanned area, showing the alternate 1^st order lamellae. e) Distribution of Mg content, layered following growth layers. f) Distribution of Sr content, displaying very faintly marked growth layers. g) Distribution of S content, faintly marking the crossed-foliated structure, strongly marking the growth layering. h-k) SEM images of a radial, unetched, freshly broken section. h) Several consecutive 1^st order lamellae (I) (SEM). i) Limit between two 1^st order lamellae (I), showing the change of orientation of its constituting 3^rdorder rods (white arrow) or slats (FEG-SEM). j) 2^nd order lamellae, composed of superimposed rows of 3^rdorder units (SEM). k) Surface view of three consecutive 3^rdorder slats within a second order row, separated by faint, punctuated limits (white arrows) and showing an inner texture (FEG-SEM). l-m) FEG-SEM images of a radial freshly broken section etched by OsO₄ vapor, revealing organic membranes that separates each 3^rdorder rod (white arrows). n-q) AFM scans. n) Phase image of the contact between two 1^st order lamellae. Green arrows mark a growth increment. o-p) Height and phase images of 2^nd order lamellae within a 1^st order lamella. q) Phase image of several 2^nd order lamellae, separated by a seemingly continuous membrane (blue arrow). Some ovoid sub-units (white arrow) can be seen, constituting the lamellae.

https://doi.org/10.1371/journal.pone.0137162.g004

On electron microprobe maps, the distribution of some minor elements strongly marks growth layering, perpendicular to the first order lamellae, which are easily identified in the backscattered electrons image of the scan (Fig 4d). In particular, Mg (Fig 4e) and S content (Fig 4g) display marked contrast, with thin (enriched or depleted) growth bands, whereas Sr content distribution shows much fainter variations (Fig 4f). No correlation can be found between minor element contents in the three distribution maps. In both Mg and S content distribution maps, slight contrast can be observed between consecutive first order lamellae, one out of two lamella, and anti-correlated: the lamellae with higher S content also contain less Mg (and vice-versa). Mn and Ba were also measured but are present in fairly low quantities and homogeneously distributed in the shell, and therefore the maps are not shown here.

SEM views of freshly broken sections illustrate the classical organization of crossed-foliated microstructural type. Between each first order unit (Fig 4h) the third order elongated rods or slats (Fig 4i) dip in opposite directions with a constant angle, and second order lamellae are composed of superimposed rows of third order units (Fig 4j). On a surface view of a second order lamella, a complex and very fine pattern can be distinguished within its third order rods. Moreover, the rods are separated from each other by a faint, very thin, and seemingly punctuated limit (Fig 4k). On the broken section, in contact with OsO₄ vapor, the very soft alteration of second order lamellae surfaces enhances the mineral/organic contrast, revealing fine sheets separating each third order unit (Fig 4l and 4m).

On AFM views, growth increments are well marked and perfectly continuous between adjacent first order lamellae (Fig 4n). Second order lamellae are visible (Fig 4o and 4p) and are also separated by a thin rim (Fig 4q), displaying a strong contrast in phase image. This contrast would result from a higher interaction between the tip and the sample surface at the border of the lamellae, and highlights the presence of a more visco-elastic material: most probably, organic sheets of meshes. If the delimitation of the third order slats is not so clear, we observed the presence of ovoid sub-units of irregular size and shape and highlighted by faint visco-elasticity contrasts (Fig 4q).

Fossil Patella sp.

The m+3 layer is thicker than in modern specimens and the transition from the radial to the concentric crossed-foliated pattern is progressive (Fig 5a). In polarized light, the first order lamellae are well marked, but the regular growth increments observed in modern Patella vulgata are not (Fig 5b). Instead, fractures irregularly divide first order lamellae, along the direction where growth increments should be observed, and they are not interconnected to adjacent lamellae (Fig 5c). However, the alternate orientation of first order lamellae is preserved, as seen in polarized and analyzed light (Fig 5c), and even their second and third order patterns can be observed, and exactly match the modern shell features (Fig 4c). For some given positions of the crossed-polarizers, faint crystallographic dis-orientations are visible close to first order lamellae borders (black arrows). These are not structure-related, and may correspond to regions of recrystallization of the mineral phase at a very fine scale.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 5. Microstructural organization of fossil Patella sp. calcite crossed-foliated outer layers.

a) PLM view (polarized and analyzed light) of a radial thin section. Outer surface is on bottom. b) PLM view (polarized light) of several consecutive 1^st order lamellae (I). c) PLM view (polarized and analyzed light) of the same area, displaying the preserved alternate orientation one 1^st lamella on two. Black arrow marks a region of possible recrystallization. d-g) Electron micropobe maps. d) Backscattered image of the scanned area, showing the alternate 1^st order lamellae. e) Distribution of Mg content, strongly marking the growth layering. f) Distribution of Sr content, faintly marking the growth layering. g) Distribution of S content, enriched along growth layers but also within the canaliculi left by microboring organisms (white arrow). h-k) FEG-SEM images of a radial, unetched, freshly broken section. h) 1^st order lamellae (I). i) Limit between two 1^st order lamellae (I), showing the change of orientation of its constituting 3^rdorder slats. j) Surface view of 3^rd order slats forming one second order row. k) Surface view of one 3^rdorder units, displaying angular-shaped sub-units. l) SEM image of abnormally thick 2^nd order units, probably composed of several fused lamellae. m-o) AFM scans. m)Height image within a 1^st order lamella. n) Phase images of its constituting 3^nd order rods. o) Phase image of the 3^nd order rods, showing that they are composed of ovoid sub-units (white arrow).

https://doi.org/10.1371/journal.pone.0137162.g005

On electron microprobe maps, minor element distributions are very similar to modern Patella vulgatashell distributions: they follow the growth layering well, with strong contrasts in Mg and S content (Fig 5e and 5g) and slight differences in Sr content (Fig 5f). Canaliculi left by microboring organisms were observed on the outer surface of the shell. Such canaliculi can cut quite deep into the shell layer (Fig 5d, 5e, and 5f) and appear as zones depleted in Mg and Sr. Quite surprisingly, they are still rich in S, therefore probably still filled with organic components from microborers. As in the modern shell, Ba and Mn contents are low and evenly distributed, so their maps are not shown here.

SEM views highlight the generally well-preserved and easily recognizable crossed-foliated organization at all microstructural levels: first order lamellae are clearly delimited (Fig 5h), composed of third order slats dipping in opposite directions (Fig 5i). In a few places however, some uncommon features can be observed: close to the boundary of first order lamellae (Fig 5j), the fracture of the shell highlights some angular-shaped constituents within third order rods, with a 120° angle typical of rhombohedral idiomorph calcite crystals (Fig 5k). The complex and fine pattern observed on the surface of second order lamellae in the modern specimen is missing, as are the limits separating third order slats. Moreover, the second order lamellae often appear to be unusually thick, as if fused together (Fig 5l). Etching or staining using OsO₄ vapor remained unsuccessful.

On AFM scans, third order rods can be easily identified (Fig 5m), but the second order lamellae are not so obvious. In particular, the strong visco-elasticity contrast found at the boundary of second order lamellae in modern Patella vulgata,is not visible in phase images on the fossil specimen (Fig 5n), where no clear separation can be seen between second order rows. Ovoid sub-units are nonetheless still present, irregularly shaped and make up the third order units (Fig 5m).

Modern Patella vulgata.

The canaliculi left by microboring organisms, attacking the outer sides of shells, display a fluorescence response when subjected to UV excitation (365 nm, Fig 6a). The shell itself, however, does not display a strong epifluorescence under the same UV excitation, except for some well-marked growth increments (Fig 6b). The acridine orange staining procedure works well and its binding to specific organic molecules reveals microstructural features (435 nm excitation) not visible with autofluorescence only. On the outer side of the shell, the canaliculi formed by microboring organisms are strongly marked by an orange-red fluorescence under blue excitation, while the crossed-foliated pattern of layers m+3 and m+2 is marked by a fainter green (sometime tending towards orange) fluorescence color (Fig 6c). This highlights the alternate organization of first order lamellae (Fig 6c), but also their third order rods dipping in opposite directions (Fig 6d). Growth increments are also particularly marked in the m+3 layer (Fig 6c), and are perfectly continuous through crossed-foliated lamellae (Fig 6d).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 6. Epifluorescence and laser scanning microscopies of modern Patella vulgata calcite crossed-foliated outer layer.

a) Canaliculi left by microboring organisms (white arrow) on the outer side of the shell (UV excitation). b) Inner border of the m+2 layer (UV excitation). c) Acridine orange-stained sample (blue excitation). White arrow marks micro-borer canaliculi on the outer side. d) Same as, focused in the m+3 layer. e) Zoom between two 1^st order lamellae (I), white arrows mark a growth increment. e) Confocal image of natural fluorescence of the sample under 488 nm excitation. White arrows mark a growth layer. f) Raman spectra extracted from following scan. g)Confocal Raman map showing the distribution of the ratio of calcite peaks L_c (librational mode, 282 cm⁻¹) / ν₁ (symmetric stretch, internal mode, 1085 cm⁻¹), which highlights changes of crystallographic orientations of calcite between 1^st order lamellae (I). h) Map of the distribution of 1134 cm⁻¹ band of polyenes. i) Map of the distribution of 1525 cm⁻¹ band of polyenes. j) Composite map of i) (in blue) and j) (in red) showing anti- and co-localization of polyenic molecules. (k) Map showing the intensity distribution of the background intensity (between 2400 cm⁻¹ and 2500 cm⁻¹) related to the fluorescence of the sample.

https://doi.org/10.1371/journal.pone.0137162.g006

On unstained samples, growth increments become visible when their autofluorescence is induced by excitation with a blue laser (488 nm) under confocal laser scanning microscope (Fig 6e); at this wavelength, a strong fluorescence from boundaries of the first order lamellae can also be observed. This fluorescence appears to correspond to seemingly continuous organic sheets, or membranes (Fig 6e).

Using confocal Raman microscopy, calcite is confirmed as the only mineral phase (Fig 6g). However, the Raman spectra also shows two peaks (at 1134 cm⁻¹ and 1525 cm⁻¹) (Fig 6f), which are not related to the calcite spectra, but characteristic of Resonance Raman (RR) spectra (respectively C-C and C = C stretching peaks), from molecules presenting a central polyenic chain (as in β-carotene). Many of the pigments widely found in the colored parts of mollusk shells are polyenes, either isolated or bound to other molecules, and can be accurately documented using RR Spectroscopy [71]. Indeed, due to the resonance coupling effect with the laser wavelength, even a small amount of pigments in the sample (down to 10⁻⁸ M) [72] can induce detectable peaks, leading to some structural characterizations of the molecules [73]. Here, these polyene peaks can be finely correlated to the microstructural pattern using the ratio of calcite peaks L_c/ν₁, as this ratio is related to the crystallographic orientation of the lattice [74]. The mapping of this peak ratio highlights the alternate shift of crystallographic orientations between consecutive first order lamellae (Fig 6g). The distribution of polyene peaks at 1134 cm⁻¹ and 1525 cm⁻¹ marks the relative distribution of polyenic molecules richer in C-C and C = C bonds respectively. These molecules are both distributed in growth increments, perpendicular to first order lamellae (Fig 6h and 6i), reflecting the cyclic secretion of these compounds by the mantle during shell growth. A composite in false color of maps at 1134 cm⁻¹ (in blue) and 1525 cm⁻¹ (in red) shows the co-localization of these peaks within the growth layers and their anti-localization (only blue color) within the membranes between the first order lamellae (Fig 6j). The polyenic molecules present in the membranes therefore display a higher ratio of C-C / C = C bonds than the polyenes present in the growth layers. Following the method developed in [74, 75], a range of the spectrum without pronounced Raman peaks (the range between 2400 cm⁻¹ and 2500 cm⁻¹) is used to map the fluorescence distribution throughout the mapped area. This map reflects the distribution of all the organic molecules with an epifluorescence response to a 532 nm excitation laser light (Fig 6k). Their distribution also varies following the growth layering, but it does not match the distribution of polyenic molecules (Fig 6j).

Fossil Patella sp.

As in modern Patella vulgata, the imprint of the canaliculi left by the microboring organisms can still be observed on the outer side of the shell under UV light (Fig 7a). But all the fossil shells display much stronger epifluorescence under UV light than the modern samples and hardly reveal any microstructural features (Fig 7b). Moreover, again under UV light, a brownish coloration of the outer side of the shell is often visible, sometimes limited to a thin layer that progressively disappears (Fig 7c), and sometimes affecting a thicker part of the shell with clear limits (Fig 7d). The acridine orange staining procedure remains unsuccessful for marking any microstructural feature, but still marks some of the microborer canaliculi on the outer side of the shell (Fig 7e). When excited with a blue laser (488 nm) under confocal laser scanning microscope, an autofluorescence response can be observed in some growth layers (they may be slightly more irregularly spaced than in modern samples). However, no fluorescence signal corresponds to the position of the membranes between first order lamellae (Fig 7f).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 7. Epifluorescence and laser scanning microscopy of fossil Patella sp. calcite crossed-foliated outer layer.

a) Canaliculi left by microboring organisms (white arrow) on the outer side of the shell (UV excitation). b) Inner border of the outer layer (UV excitation). c-d) Faint (c) and strong (d) changes of natural fluorescence hue of the shell at the contact with the sediment. e) Acridine orange-stained sample (blue excitation). white arrows mark some micro-borer canaliculi on the outer side. f) Confocal image of natural fluorescence of the sample under 488 nm excitation. White arrows mark a growth layer. g) Raman spectra extracted from previous scan. h)Confocal Raman map showing the distribution of the ratio of calcite peaks L_c (librational mode, 282 cm⁻¹) / ν₁ (symmetric stretch, internal mode, 1085 cm⁻¹), which highlights changes of crystallographic orientations of calcite between 1^st order lamellae (I). i) Map of the distribution at the wavenumber that should correspond to 1134 cm⁻¹ band of polyenes. j) Map showing the intensity distribution of the background intensity (between 2400 cm⁻¹ and 2500 cm⁻¹) related to the fluorescence of the sample.

https://doi.org/10.1371/journal.pone.0137162.g007

Raman confocal microscopy also confirms that calcite is the only mineral constituent (Fig 7g) and still displays an alternate shift of crystallographic orientations between consecutive first order lamellae (L_c/ν₁ peak ratio, Fig 7h). However, no peak corresponding to the polyenes can be found in the mean scan spectra (Fig 7g) nor in the map at 1525 cm⁻¹ (or 1134 cm⁻¹) (Fig 7i). The Raman spectra shows a higher background (Fig 7g), caused by a higher fluorescence of the sample. When mapped using background intensity between 2400 cm⁻¹ and 2500 cm⁻¹ (Fig 7j), the fluorescence signal still follows growth layering, although more irregularly than in modern Patella vulgata. Organic components seem to be better preserved in some areas than in others.

The disappearance of organic membranes.

The question of the occurrence of structural membranes within inter-crossed mollusk shell layers is still under debate, and of the utmost importance, as they may play an active role in the formation of these microstructural patterns [91]. Structural membranes can also have major implications for the modalities of the taphonomic/diagenetic evolution of these mollusk shell layers. The presence of sheets between second and/or third order lamellae in crossed-lamellar aragonite has long been reported in some bivalve shells, such as Tridacnidae [64], Glycymerididae,Cardiidae or Corbiculidae [92]. In other cases, these sheets are not reported, such as for the gastropod Lottia kogamogai (Patellogastropoda, Lottiidae) [93]. As for the boundaries between adjacent first order lamellae in crossed-lamellar microstructure, some authors consider that there are no membranes [64]. However, some indications suggest the presence of a non-continuous organic mesh in-between [74, 91]. Moreover, it has been shown that organic macromolecules extracted from limpet shells does impact the nucleation and the growth of aragonite crystals formed during in vitro experiments [94]. It is therefore important to validate the presence (or absence) of organic meshes or membranes structuring the crossed-lamellar microstructure, as the molecules forming these intricate organic patterns could be actively involved in the formation of such a complex 3D architecture.

In this respect, our finding of very similar membranes in crossed-foliated layers of modern Patella vulgata shells is quite consistent, as this structure is basically the calcitic counterpart of crossed-lamellar. An organic framework is indeed observed, separating third order slats within a second order row (Fig 4k–4m), and second order rows from each other (Fig 4p–4q). Moreover, seemingly continuous membranes are also visible at the boundaries of first order lamellae, but were only observed with scanning laser microscopy (Fig 6e–6h), and not on SEM views (Fig 4i). The membranes display strong autofluorescence under 488 nm excitation, and Resonance Raman maps indicate that they contain small amounts of polyenes (Fig 6j).

No similar framework of organic sheets or meshes is found in fossil Patella sp. shell, nor between first order (Fig 7f and 7i), second order (Fig 5n–5o) or third order (Fig 5j) lamellae. The disappearance of these structuring organic membranes is a first step in the taphonomic evolution of the shell. This can easily be explained by the fact that the molecules composing these pure organic meshes are not strongly linked to the mineral phase, unlike the organic matrix located within the mineral phase. This matrix is incorporated within the crystal at a molecular level [55]. The meshes are therefore more easily destabilized and liable to disappear first during fossilization processes such as hydrolysis, etc. The very classical example of Pinna illustrates this mechanism: the early degradation of the organic sheath surrounding the prisms leads to their dissociation and to the degradation of the prismatic layer. This process can occur very soon after the death of the animal and often only leaves the nacreous layer as fossil remain (in fact, this early diagenetic process even begins during the animal’s lifetime) [95].

Here, the degradation of the organic network can be correlated with the diagenetic evolution of the mineral phase. The disappearance of organic molecules separating second order rows would enable the observed fusing of several second order rows together (Fig 5l). Some modifications only tend to be visible close to the boundaries between first order lamellae, such as localized crystal dis-orientation spots in PLM (Fig 5c), or unusual angular-shaped components of third order rods (Fig 5k), whereas the bulk of the lamellae is better preserved (Fig 5n–5o). They could therefore concur with the disappearance of organic sheets separating first order rows.

Alteration/preservation of organic molecules.

Membranes are not the only organic components affected. Whereas the depigmentation of the shell is visible to the naked eye, the disappearance of polyenic molecules distributed in the growth increments is confirmed at the micrometer scale (Fig 7i). The polyenic molecules represent a small percentage of the total organic content of the shell: indeed in a modern shell, their distribution does not match the distribution of fluorescent molecules (Fig 6k). However, Raman spectroscopy allows us to detect them with high sensitivity and these compounds therefore represent valuable markers to monitor the diagenetic evolution of fossil specimens.

The organic molecules that interact with acridine orange and were distributed in growth increments/third order rods in modern samples (Fig 6d), were probably degraded as no acridine fluorescence could be detected in the fossil shell. The interpretation of the fluorescent color caused by acridine orange staining is not straightforward. It is a cationic dye, which in solution emits in the green region (533 nm) when in monomer form, and in the red region (656 nm) in polymer form [96], the shift being caused by a dye-dye coupling effect. The emission color therefore depends on the distance between the stained molecules [97]. When acridine molecules are bound to neighboring target molecules, or neighboring sites on the same molecules, they display an orange-red fluorescence; when bound to isolated compounds, so that the dye molecules are weakly or not coupled with one another, the fluorescence is green [98]. Acridine orange is nonetheless commonly used [97, 99] as a dye for various compounds displaying available negative binding sites (such as glycoproteins) [100]. Here, due to the complexity of the organic matrix composition, we cannot draw conclusions on the characterization of the molecules stained by this protocol. A wide spectrum of fluorescence (green, orange, and red) is observed. But these organic compounds were nonetheless sufficiently modified or even completely disappeared during the taphonomic evolution of the fossil Patella sp. shell, so that they no longer bind acridine orange molecules.

A substantial fraction of the organic matrix is preserved and still displays autofluorescence under laser excitation (488 or 532 nm). These molecules are preserved in situ, still distributed in some well-marked growth layers, confirming that the organic corteges underwent differential diagenesis. But, as shown using the autofluorescence of organic compounds in confocal or CRM, many growth layers have been worn away. Any sclerochronological studies based on internal growth layering (daily/sub-daily growth increments) detected by means of organic components (fluorescence, etchings, etc…) are therefore prone to biases; the differential diagenesis of the organic matrix also rules out the use of the bulk biochemical composition of these fossil Patella sp. shells as palaeo-proxies.

Mollusks and human behavior.

MP/MSA (Aterian) groups seem to have occupied the Témara caves ¨intensively¨ during the isotopic stage 5 (120-75 ka BP), undoubtedly in connection with periods of high sea levels when the caves where near the shore, as is the case today [17]. This location would have facilitated the exploitation of mollusks. Fresh limpets gathering would be confirmed at El Harhoura 2 Cave by the presence of notches and the good preservation state of limpet shells (no dissolution/recrystallization, no bioerosion and no abrasion/fragmentation aspects). However, notches have also been observed on Patellidae used as tool in Asturia dating from the Magdalenian [101]. In absence of micro-wear analyses (as it was made by Cuenca et al [101]), it is impossible to exclude the possibility that the Patella sp. shells from El Harhoura 2 have also been used as tools. Thus these notches may result both from their collecting and/or their use as tool. Anyway, these notches on fossil specimens are morphologically very similar to those obtained by collecting modern Patella vulgata with a knife.

Furthermore, the good preservation of limpet shells indicates that a very short period of time occurred between the gathering of the mollusks on the seashore and their deposition and burying in the cave. In addition, this preservation suggests that the sediment of level 8 is a favorable local micro-environment for mollusk shell conservation until the present time, unlike other archaeological sites, such as Die Kelder Cave [102]. Thus, at El Harhoura 2, limpet shell preservation allows us to document human behavior. Aterian groups of El Harhoura 2 (level 8) seem to have focused their selection on Patellidae, but it remains to be established whether these taxa were chosen among others, or whether they were the main available taxa in the local environment. The reconstruction of the geomorphology of the local coast is in progress, in order to identify coastal morphology during the cave occupations. The example of the neighboring cave of El Mnasra (unit 8, also dated to OIS 5) shows that the Aterian populations of Témara have exploited varied terrestrial (diverse ungulates and tortoises) and marine (mollusks) resources [11, 17, 27]. In addition, different activities were carried out on-site (lithic production, butchery, terrestrial faunal and mollusk consumption, use of fire, bone tools, Nassarius beads, and worked pigments). These results are similar to those from Southern Africa, such as Pinnacle Point Cave [33–35], Yserkfontein [36] or Blombos [103–105].

The apparent good preservation state of the limpet shells from El Harhoura 2 could suggest that they were well suited to further studies focusing on human behavior. Sclerochronological investigations, for example, could lead to the identification of occupation seasons at the site and help to establish a mobility model for MP/MSA groups in North Africa; a question still under debate. Indeed C. Marean [33, 34] suggests that the first southern AMH planned their visits to littoral areas according to tidal cycles, in turn dependent on lunar cycles. Thus, during spring tides MSA groups could have exploited the shore for a longer period without danger from waves. But the faint diagenetic degradations evidenced in the present study tend to bias such investigations and severely hinder these perspectives.