Comparison of the Calcareous Shells of Belemnitida and Sepiida: Is the Cuttlebone Prong an Analogue of the Belemnite Rostrum Solidum?

: The microstructure of the rostrum solidum of Jurassic belemnites is compared with that of Sepia cuttlebones, in order to examine possible convergences in their style of growth. For this study, transmitted and polarized light, cathodoluminescence, epifluorescence, scanning electron and backscattered electron microscopy have been employed. Despite differences in the primary mineralogy of the studied belemnites and sepiids, calcite and aragonite, respectively, many similarities have been observed between the microstructure of the belemnite rostra and the prong of Sepia cuttlebone: (1) In both, crystals start growing from successive spherulites, from which crystals emerge radially towards the apex and the external walls, displaying internally microfibrous texture. (2) Both display concentric growth layering, comprising an alternation of organic-rich and organic-poor layers, which, in turn, is traverse by the radially-arranged micro-fibrous crystals. (3) The highest organic matter content and porosity have been observed along the apical area of the Sepia prong, similarly to that interpreted for belemnite rostra. The strong convergences observed suggest that the growth of belemnites occurred similarly to that of the prong of sepiids and that the Sepia prong is the analog of the belemnite rostrum. Additionally, non-classical crystallization processes are proposed to be involved in the formation Sepia endoskeleton. cathodoluminescence and epifluorescence of cuttlebones and belemnites. M.I.B. etching of belemnites with


Introduction
In addition to their usefulness for biostratigraphical studies, belemnite rostra are commonly used for palaeoclimatic, palaeobiological and palaeobiogeographical studies based on their geochemical data or on their morphological features . However, no agreement exists regarding whether belemnites were originally porous [1,2,[37][38][39][40][41][42], or whether the original porosity was of minor importance [43,44], although this issue has important implications for reconstructing, for example, their buoyancy, size and ontogenetic strategies and palaeoenvironments. Some authors have proposed low porosity values for belemnite rostra [3,7,29]. However, Veizer [2] and Spaeth [38] proposed up to 10% and 20% of primary porosity, respectively, for the belemnite rostrum, Ullmann et al. [41] up to 40% of porosity at the apical line, Hoffmann et al. [42] proposed 50-90% of porosity in belemnite rostra, and Benito et al. [31] proposed porosity values ranging from 20 to 60% in the apical area of the rostrum. In this regard, Benito et al. [31] also suggested that the analysis of the prong of Sepia cuttlebone could be useful for understanding the original structure and porosity of belemnite rostrum.
The fossil Order Belemnitida is included in the Subclass Coleoidea, though most recent coleoids have reduced their internal skeleton substantially or even entirely [36,45,46]. For example, it has been interpreted that the chitinous squid gladius with septa-like layers in cone situated at the posterior end of the proostracum represents a vestigial phragmocone of ancient coleoids that has lost the siphuncle and gas-filled chambers, and that the "rostrum situated apically from the cone is homologous to the rostrum of ancient phragmocone-bearing coleoids" [47].
Despite the differences between their primary mineralogy, previous works have compared belemnites with cuttlebone, looking for homologies [65][66][67][68][69][70][71][72]. The prong or spine has been also named as rostrum by Naef [65], Dauphin [70] and Bizikov [73]. Abel [66] and Krymgol'ts [67] interpreted that the spine was not homologous to the belemnite rostrum. Naydin [68] observed clear homologies between some parts of the cuttlebone and the belemnite. For this author, the phragmocone, the fork (or ventral process) and the innermost layer of the dorsal shield are related to the belemnite phragmocone. Naydin [68] also interpreted that the middle and outermost layers of the dorsal shield and the prong (or spine) correspond to the belemnite rostrum. Barskov [69] interpreted that the middle layer of the dorsal shield corresponds to the belemnite rostrum, whereas Dauphin [70] interpreted that the outer layer of the dorsal shield and the spine corresponded to the belemnite rostrum. Naef [65] suggested that the belemnite rostrum and the Sepia prong are analogous organs and Rexfort and Mutterlose [48] proposed that "the sepioid rostrum, which is homologous to the belemnite guard, is purely organic". Additionally, the prong (or spine) has been associated with the middle layer of the dorsal shield [69,74], the outer layer of the dorsal shield [70] or seen as constituting an additional layer [73,75]. On the contrary, Fuchs [71] has interpreted that the rostrum is absent in Sepiida and that there is not an equivalent structure to the belemnite rostrum. In other cases, the comparison between Sepia and belemnites has been focused for understanding the belemnite palaeoecology from the study of stable isotopes of O and C on Sepia officinalis [48,76,77]. Finally, other authors have compared belemnite rostra with squids. Arkhipin et al. [47] indicated that the rostrum of the gladius of some squids (onychoteuthids, omnastrephids and gonatids) is homologous to the rostrum of belemnites Hibolithes and Pachyteuthis.
In this study, we compare the microstructure of the rostrum solidum of Jurassic belemnites Hibolithes, Belemnopsis and Passaloteuthis with that of cuttlefishes (Sepia officinalis and S. orbignyana) in order to examine if their style of growth is similar and check if the growth of the Sepia prong is similar to those of belemnite rostrum and can be considered as an analogous of the belemnite rostrum solidum.

Methods
This study is based on the analysis of 18 specimens of well-preserved Jurassic belemnite rostra (12 Hibolithes, 3 Belemnopsis and 3 Passaloteuthis; Figure 1) with that of 15 specimens of cuttlefishes (nine Sepia officinalis and six S. orbignyana; Figure 2) (see supplementary material for details, Table  S1). The belemnites analyzed come from two outcrops located in the Betic Cordillera (South Spain). Passaloteuthis was recorded at La Cerradura section (lower Toarcian, Lower Jurassic, External Subbetic; see location on Figure 1 of Reolid et al. [78]) and Hibolithes and Belemnopsis were recorded at Pozo Cañada Section (Oxfordian-lowermost Kimmeridgian, Upper Jurassic, External Prebetic; see location on Figure 1 of Benito et al. [31]). Specimens of cuttlefishes come from the Alboran Sea (Western Mediterranean) and the Cantabrian Sea, in the case of S. officinalis, and from the Alboran Sea, in the case of S. orbignyana. Polished and uncovered thin sections (30 and 400-500 µm thickness) were prepared for each belemnite rostrum (one longitudinal and one transversal to the growth of the belemnite rostra) and, when possible, one for each cuttlefish rostrum and prong which, in turn, were embedded in epoxy resin to make possible thin section preparations. The longitudinal sections were lateral and not oriented dorso-ventrally, except in one thin section. Ultra-thin sections (<15 mm thick) were also obtained after re-polishing three thin sections. Thin sections were studied and photographed under transmitted light (TL), polarized light (PL), cathodoluminescence (CL), and epifluorescence (FL) microscopy at the Universidad Complutense (Madrid): For CL study, a Technosyn cold cathodoluminescent unit MK4 at 20-24 kV with 350-400 mA beam current was employed; the FL study was carried out using an incident-light FL Nikon Y-FL epifluorescence system coupled to an Eclipse 6400 POL petrographic microscope. An UV (340-380 nm) excitation filter (400 nm dichroic mirror and 420 nm barrier filter) and a blue light (450-590 nm) excitation filter (505 nm dichroic mirror and 520 nm barrier filter). Specimens and thin sections are deposited in the collections of the Department of Geology of the University of Jaén. In terms of petrography, all the studied belemnite rostra are well-preserved and display transparent appearance under TL and lack of luminescence [30,31], despite the apical area and some growth layers of belemnites being diagenetically altered and displaying cloudy appearance and bright to dull luminescence (see [30] and Figure 6 in Benito et al. [31]). In this sense, Stevens et al. [33] have observed alternation of blue and orange luminescent growth lines in translucent brood cases of Argonauta and in belemnite rostra, which they interpret as primary. However, these authors also observed luminescent areas displaying cloudy appearance (under TL microscopy) in belemnite rostra, which they interpret as diagenetic in origin, in accordance with interpretations made by other authors [30,31,44,57]. Secondary electrons and backscattered scanning electron microscopy (SEM and BSEM, respectively) was performed using a JEOL JSM-820 electron microscope at the Universidad Complutense (Madrid, Spain) and a SCI Quanta 400 scanning electron microscope at the Centro de Instrumentación Científico-Técnica of the Universidad de Jaén (Spain), in both thin sections and in fresh-cut belemnites and cuttlefishes. Subsequently to examination, samples were etched after immersion in a 25% solution (Merck) of glutardialdehyde maintained at a pH of 4.0 for 12 h, in the case of thin sections, and 24 h in the case of fresh cut samples. Glutardialdehyde, CH2(CH2CHO)2, is known to react with proteins [79]; moreover, because glutardialdehyde is easily oxidized to glutaric acid [80], it has been also benefited from the etching effect of the acid [35]. After treatment with glutardialdehyde, the slabs were washed with water and ethanol and then dried and analyzed under SEM.

Belemnitida
Belemnite shells are composed of three main parts: phragmocone, primordial rostrum and rostrum proper [71]. Longitudinally, they are also divided into proostracum, rostrum cavum and rostrum solidum (Figure 3). For checking the wide terminology for shell elements of belemnites [36,71]. Specimens of Hibolithes (Family Belemnopseidae) are characterized by an elongated and hastate rostrum, with the maximum transverse diameter at the posterior region, near the apex, a circular transverse section, and a shallow ventral groove limited at the anterior half of the rostrum. Belemnopsis (Family Belemnopseidae) presents a hastate and depressed rostrum, with maximum transverse diameter slightly posterior of the midpoint, an asymmetrical transverse section with ventral side inflated, and a deep broad ventral groove. The Toarcian specimens of Passaloteuthis (Family Passaloteuthididae) are characterized by a robust rostrum (with well-developed primordial rostrum), with cylindrical outline and a rounded to slightly compressed transverse section, without ventral grooves. The orthoconic phragmocone of Hibolithes and Belemnopsis opens at an angle of ca. 20° with the initial chamber of the protoconch (<0.6 mm diameter) completely sealed by a closing membrane and delimited by a slight constriction (Figures 3 and 4A). The phragmocone of Passaloteuthis opens at an angle of ca. 28°. The siphuncle penetrates the septa proper along their ventral margins ( Figure 4). The phragmocone is enveloped by two calcitic external layers, the primordial rostrum (uni-layered and thin) and the rostrum proper (multi-layered and thick), as described by Fuchs [71] and Doguzhaeva et al. [81]. The primordial rostrum of these belemnite genera was originally aragonitic and contained abundant non-biomineralized components (probably chitin [81]). The rostrum solidum was originally composed of low Mg-calcite [2,7,44,47,[57][58][59]. The belemnite rostrum solidum, when observed under the petrographic microscope in both transversal and longitudinal sections, is formed by radially arranged prismatic calcite crystals, which emerge from the apical line, and display a concentric pattern ( Figure 5), commonly interpreted as growth layers or rings [30,31,44,82,83], and undulose extinction under polarized light (PL) microscopy ( Figure 5B,E). The rostrum solidum is also characterized by a regular spherulithic prismatic microstructure [31] formed by spherulites of around 250 µm in diameter that successively develop along the apical line and from which calcite crystals radially emerge and diverge, as is observed in the longitudinal sections of the rostra under SEM, as well as under PL microscopy in the ultra-thin sections ( Figure 6). The inner parts of the spherulitic centers are composed by dense concentrations of small calcite crystals (<1 µm in the nuclei of the spherulithic centers). Progressively, calcite crystals become larger and more elongated and radially distributed as hemispheric fans ( Figure 6B,C). Towards the outer parts of the rostra, crystals become longer, thicker and prismatic in shape ( Figure 6).  The detailed examinations of the calcite prisms also allow to differentiate an internal microfibrous texture, which is observable under TL, PL, FL, SEM and BSEM (Figure 7), and is enhanced after etching with 25% glutardialdehyde ( Figure 8) [31,44]. The concentric pattern, which may be also observed at different scales, is traversed by the radial pattern (Figures 5, 7 and 8B) and is enhanced under FL microscopy and BSEM ( Figure 7C,G-H). Some concentric layers (10-50-µm thick, generally) display an internal radial micro-fibrous texture, weak FL and an overall light grey colour under BSEM. These layers are separated by very thin concentric layers (<5 µm thick), which display intense FL and an overall dark grey colour under BSEM. Benito et al. [31] also observed that single calcite crystals of belemnite rostra are composite and are composed of two sectors, which are in optical continuity ( Figure 9A-J): an inner sector, fluorescent, with relatively low optical relief under TL microscopy, and a dark-grey colour under BSEM, which may have an internal "patchy" appearance ( Figure 9G-I), and an outer sector, non-fluorescent, with relatively high optical relief under TL, and a light-grey colour under BSEM. These sectors may be also observed after etching thin sections with 25% glutardialdehyde ( Figure 9K-L). The composite nature of belemnite crystals has been also suggested or interpreted by other authors, such as Richter et al. [84] and Hoffman et al. [36], respectively, for other Jurassic and Cretaceous belemnite species (Belemnitella, Cylindroteuthis, Gonioteuthis, Megateuthis, and Pachyteuthis). Additionally, Benito and Reolid [30] and Benito et al. [31] observed that the concentric layers and sectors of the crystals displaying intense FL also had higher Mg, Na and S concentrations than those showing weak or no FL. showing the micro-fibrous texture of the crystals (black arrows) and the concentric layering. Under FL, the concentric layering is enhanced: some layers display weak FL and micro-fibrous texture; they are separated by very thin and intense FL layers, in which the microfibrous texture is no evident. (D-F) Transversal fresh-cut sample (CE-25-04) under SEM. Prismatic crystals of calcite internally display a micro-fibrous texture (E,F, red arrows). Perpendicular growth layering is also observable. Note that microfibers traverse the concentric growth layering (F, white arrows). (G-H) Longitudinal (G) and transversal (H) thin sections (specimens PC-31-04 and PC-37-01, respectively) observed under backscattered (B) SEM. The micro-fibrous texture is observable as fibers displaying alternating light-and dark-grey colours. Note that the light-grey fibers traverse the concentric growth layers displaying an overall dark-grey colour (white arrows), which correspond to the thin layers that displays intense FL (see C). Note also that, in transversal sections (H), the microfibrous texture is not observable in the inner part of the rostra (in the apical area) but progressively becomes more evident towards the external wall of the rostra.  perpendicularly to their c-axis and they are observed as an equigranular mosaic of crystals. Outwards, towards the external wall, crystals are cut tangentially to their c-axis and become elongated. Note also that the concentric layering is enhanced under FL microscopy and that layers become thinner towards the external walls. (D-F) Detailed images of (A-C). Note the triangular shape of crystals that comprise sectors displaying relatively low optical relief (D,E) and intense FL (F) (red arrows) and sectors displaying relatively high optical relief (D,E) and none-to-weak FL (F) (purple arrows). (G-I) Detailed photographs showing the composite nature of calcite crystals under TL (G), PL (H) and FL (I) microscopy (specimen PC-11-31). Triangular-shaped crystals comprise an inner sector (red arrows), which displays an overall relatively low optical relief (D,E), intense FL (F) and an internal "patchy" appearance (black arrows). Note that the "patches" also displays a triangular shape and higher optical relief than the surrounding calcite. The outer sector of the crystals (purple arrows), which is in optical continuity with the inner sector (red square in (H)), displays relatively high optical relief (D,E) and non-to weak FL (F). Note the crystal boundaries (purple arrows and red square in (G) and (H)). Note also that the triangular "patches" of the inner sector has an identical optical relief and FL than the outer sector of the crystals. (J) Same thin section as in (A-F), observed under BSEM. The triangular inner sectors of the crystal displaying intense FL are dark-grey in colour (red arrows), and the outer sectors of the crystals, displaying non-to weak FL, are light-grey in colour (purple arrows). (K,L). Same thin section as in (A-G), observed under BSEM after etching with 25% glutardialdehyde. Note that the triangular shape of crystals and the concentric layering (white arrows in (K)) is enhanced under etching. The triangular areas comprise an inner sector, which has been largely etched (red arrows), and an outer sector where dissolution is less pervasive (purple arrows). The largely etched concentric layers, which would correspond to the intense FL concentric layers, are thin (white arrows) compared to those where etching is less pervasive.

Sepiida
The cuttlebone of Sepia is composed of aragonite, coated with a thin layer of organic matter, and encased by an epithelium also called the cuttlebone sac [48,85]. The cuttlebone is composed of three different parts, the phragmocone, the dorsal shield and the prong (Figures 3 and 10).
The phragmocone contains many spaced septae (around 400 in Sepia officinalis [48]), connected by small undulating transverse pillars (undular plates [69], or palisades [44]) ( Figures 10A,B and 11). The septae are connected to the inner mineralized layer of the dorsal shield and extend rearward from it ( Figure 11B,C). The septae are composed of two layers: the anterior (~9 µm) is of prismatic structure with fine and densely arranged parallel prisms, which are perpendicular to the septum; the posterior layer is lamellar (~15 µm; Figure 11F).  (SP-01). The red square on G represents the area shown in (I). Note the concentric growth pattern of the prong (red arrows in (H)), and the high intra-skeletal porosity (white arrows in (G) and (I)), which is enhanced in the apical area. Spherulites are developed in the inner part of the prong (yellow arrows), displaying radially arranged aragonite crystals and undulose extinction. Towards the outer wall of the prong, aragonite crystals become longer and radially arranged, displaying undulose extinction and traversing the concentric layering (red squares). Modified from Figure 7 of Benito et al. [31], who erroneously assigned this specimen to S. orbignyana. The analysis of septae under BSEM allowed to recognise that the lamellar layer is richer in organic matter than the prismatic layer. In detail, the lamellar layer is composed of an alternation of organic-and aragonitic-rich sublayers ( Figure 11G). The transverse pillars are made up of laminae composed of small randomly oriented aragonite crystals [69] that, under BSEM, show an internal alternation of organic-rich and organic-poor laminae ( Figure 11E). In S. officinalis, the heights of the transverse pillars range between 490 and 150 µm, meanwhile in S. orbignyana they range between 195 and 80 µm. According to Florek et al. [86], septa are richer in β-chitin and pillars in aragonite. Organic films are also observed between septae ( Figure 11D,H,I).
The chambered cuttlebone has a dual function as a support and as buoyancy regulation through liquid and gas adjustments, which requires an open structure that is pressure-resistant while maintaining a constant volume [87]. The porosity of the cuttlebone of S. officinalis is 93% [88], although Sepia species that live in deeper habitats had thicker septa and less space between pillars than shallow species [89].
The dorsal shield ( Figures 10A-E and 12) contains high organic matter content and has an important mechanical role by increasing the flexural strength of the cuttlebone at the same time that has high porosity and high permeability [87,88]. In the case of S. officinalis, the phragmocone contains 10% of organic matrix and the dorsal shield 30-40% [86,88]. Dorsal shield occupies the dorsal region of the cuttlebone, above the phragmocone, and comprises three mineralized layers [69,70] (Figures 3,  10C-E and 12). The outer layer consists of convex outwards hemispheres composed of radially arranged acicular crystals of aragonite, which display undulose extinction under PL microscopy and concentric layering of alternating non-to intense FL ( Figures 10C-E and 12B-G). The hemispheres constitute the tubercles in the surface of the dorsal shield of S. officinalis but, in S. orbignyana, tubercles are poorly developed. The thickness of the outer layer is variable along the dorsal shield, ranging between 315 and 411 µm and the contact with the middle layer is stepped in the proximal parts of the cuttlebone in S. officinalis. Over the outer mineralized layer of the dorsal shield, there is an organic granulated layer (170 µm maximum thick), which is highly porous and displays intense FL ( Figures  10C-E, 11B,C and 12D).
The middle layer of the dorsal shield ( Figure 10C-E) presents a laminated structure ( Figure 12H-J) composed of organic-rich lamellar sublayers (1.5 µm) alternating with prismatic sublayers of aragonite (4-5 µm). The distribution of organic matter is clearly observed under BSEM ( Figure 12C). The thickness of the middle layer is variable along the dorsal shield and displays a stepped distribution with a thinner middle layer of about 50-60 µm, and it becomes progressively thicker, reaching 95-130 µm. In addition, the middle layer is thinner in the distal parts of the cuttlebone.
The prong or spine (also called mucron by Cuif et al. [90]) is a solid tooth-like continuation of the middle layer of the dorsal shield, according to Barskov [69] (Figures 3, 10A,F-H and 13). Dauphin [91] proposed that the Sepia spine consists of the same three layers which form the dorsal shield, with the inner and middle ones being very thin. In our samples, we have not observed these layers in the spine from S. orbignyana and S. officinalis. The prong is comparatively better developed in S. orbignyana (7.6-11.2% of the total length) than in S. officinalis (1.5-2.7% of the total length) ( Figure 2). In transversal sections, the prong is composed of alternating concentric laminae of aragonite crystals, displaying radial arrangement, and organic-rich and very thin laminae ( Figure 13B). In a close view, the aragonite crystals are composed of clusters of small crystalline units ( Figure 13B) with rod-like appearance. To the outer parts of the prong, aragonite crystals are progressively more elongated and become parallel (Figure 14), displaying a similar microstructure to that observed in transversal sections of the belemnite rostra (compare Figures 7E,F,H, 13B and 14).   In longitudinal sections, the prong is also composed of concentric laminae, which are parallel and younger towards the prong surface, similarly to that observed in longitudinal sections of belemnite rostra (compare Figures 5D-F, 9H, 10F-I and 13C-E). The highest porosity has been observed along the area where laminae have their maximum curvature (c.f. the apical area). In detail, under FL microscopy, concentric laminae of the Sepia prong are composed of alternating intense and non-to weak FL laminae ( Figure 10H); under SEM relatively thick and organic-poor (aragonite-rich) laminae (2-5-µm-thick in the apical area) alternate with relatively thin and organic-rich laminae (0.5-1-µm-thick) ( Figure 13D,E and 15). The overall thickness of the laminae decreases from the curved apical area to the straight lateral margins, similarly to that observed in belemnite rostra (compare Figures 5D-F, 9H, 10F,I and 13C-E). Some laminae in the apical area pinch out laterally and disappear ( Figure 15A). Moreover, in longitudinal sections of the prong of S. officinalis, spherulites of radially arranged acicular crystals of aragonite have been observed in the inner area of the prong ( Figure  10F,I). However, it has not been possible to observe the arrangement of spherulites properly, because of the difficulty in obtaining a section completely parallel to the apical area of the prongs. Aragonite crystals, which display undulose extinction under PL microscopy, become progressively longer towards the external wall of the prong, crossing-cut through different concentric laminae, similarly to that occurring in belemnite rostra (compare Figures 5E, 6C, 7A-C,G and 10I). However, under SEM it can be observed that radially arranged crystals are internally composed of clusters of nannometersized aragonite crystalline units ( Figure 13E,F). In S. orbignyana, organic fibers (500 nm length, 60 nm diameter) are observed among the aragonite crystalline units, which may be aligned, displaying a poorly developed concentric layering ( Figure 13F). In the case of S. officinalis, concentric thin organic films (<50 nm), comprising organic fibers, which are projected out, are observed interlayered with the aragonite crystalline units ( Figure 15B-D).

Discussion: Is the Cuttlebone Prong an Analogue of the Belemnite Rostrum?
The analyses of belemnites and cuttlebones show that biomineralization was strongly controlled by the organism during its stepping growth. In the case of Sepia, the organic matter is present everywhere in the phragmocone, dorsal shield and prong, always in alternation between organicrich and organic-poor layers and sublayers, which may be observed at different scales (Figures 11,12  and 15). Despite the difficulties related to preservation of organic matter and the incidence of diagenesis, this type of growth architecture is also observed in the belemnites, as indicated by petrographic features (Figures 5 and 7-9; see the results section and Benito and Reolid, [30] and Benito et al. [31] for details) and by the distribution of Mg, Na and S contents, which are higher in the intense FL layers than in the none-to-weak FL layers [30,31].
The prismatic structure of the inner layer of the dorsal shield of the cuttlebone has been interpreted as equivalent to the inner prismatic layer of the phragmocone wall in belemnites (primordial rostrum sensu Fuchs [71]). The structure of the middle layer of the dorsal shield and of the prong combines organic-rich and organic-poor aragonitic elements, such as has been deduced in the belemnite rostrum after SEM and FL examination [30,31]. The outer layer of the dorsal shield cannot be compared with the rostrum solidum of belemnites, but this type of radially arranged microstructure is observed in other cephalopods, such as the Nautilus external shell [69] and the brood case of Argonauta [33,92,93].
In belemnite rostra, as indicated in the results section, calcite crystals start growing from the spherulites, which are distributed along the apical line, displaying a regular spherulitic-prismatic microstructure [31]. The spherulitic-prismatic structure of belemnite rostra was firstly described by Bandel et al. [43] in other specimens of Hibolithes and, after Benito et al. [31], has been confirmed in Neohibolites minimum and Gonioteuthis rostra [33,36].
Crystals become progressively longer and radially distributed until, towards the external wall of the rostrum, they convert into parallel prismatic crystals (Figure 6), which, in turn, displays an internal micro-fibrous texture (Figure 7). Benito et al. [31] interpreted that the slight variation in the orientation of the c-axis of calcite crystals during progressive growth could be the cause of the undulose extinction that is observed in belemnite rostra under PL microscopy. The slight changes in c-axis direction between neighboring calcite crystals have been subsequently confirmed in Neohibolites minimum [33]. Similarly, in longitudinal sections of Sepia, an equivalent arrangement of aragonite crystals, emerging from spherulites, becoming progressively longer and radially arranged to the outer wall of the prong, and displaying undulose extinction, has been also observed ( Figure  10). Moreover, in transversal sections of the prong, a micro-fibrous texture, similar to that observed in belemnite rostra, has been also recognized (compare Figures 7F and 14). In this regard, Yancey and Garvei [56] have reported radial-prismatic and spherulitic-like aragonite crystal arrangements in Anomalosaepia specimens.
The concentric layering of belemnite rostra also resembles that observed in the prong of Sepia orbignyana and S. officinalis (compare Figures 7, 9H, 10F,I and 13C-E). In the studied belemnite rostra, the concentric pattern comprises an alternation of radially arranged layers, displaying weak FL, and very thin layers without radial structure, displaying intense FL. The alternation of weak and intense FL concentric layering has been interpreted as an alternation of organic-poor and organic-rich layers, respectively [30], in agreement with interpretations previously made by Saelen [44] in Neohibolites and Belemnellocamax. In the case of the Sepia prong, consecutive and concentric layers of aragonite crystals with radial arrangement are interlayered with organic-rich and non-radially distributed laminae. Additionally, in both belemnite rostrum and Sepia prong, it is observed that calcite and aragonite crystals, respectively, which are in optical continuity, cross-cut the organic-rich concentric layers (compare Figures 7, 9A-C, 10I and 14).
The presence of original organic matter in belemnites and other fossil shells has been observed or suggested by several authors [1,30,31,33,40,44,47,55,82,[94][95][96][97][98][99][100]. In this regard, Benito et al. [31] also observed that single calcite crystals of belemnite rostra are composite (Figure 9; see the results section), interpreting that the FL was activated by organic compounds, which were incorporated into the belemnite skeleton during progressive growth, preferentially in the apical area and in some concentric layers. This interpretation is in agreement with that given by Florek et al. [40] and Dunca et al. [99], who observed an excess of carbon in Jurassic belemnites, mainly in the apical line [40], and interpreted that organic matrix was involved in the formation of belemnite skeleton.
Based on these findings, Benito et al. [31] interpreted that belemnite rostra were originally porous (20-60% of porosity) and that crystals were formed during two different growth stages: the fluorescent sectors of the crystals formed during belemnite skeletal growth, as a product of biologically controlled mineralization, and the non-fluorescent sectors precipitated subsequently as a postmorten cement, filling the intra-and inter-crystalline porosity. In this regard, Benito et al. [31] argued that if calcite overgrowths would have precipitated while belemnites were alive, the density of the rostrum would have increased, changing their buoyancy and gravity center and making them too heavy to support the swimming animal's horizontal living posture, as suggested by Spaeth [38]. Thus, Benito et al. [31] highlighted that caution should be taken when analysing geochemically the belemnite rostra for palaeoenvironmental studies, in agreement with interpretations previously made by Benito and Reolid [30] based on geochemical data obtained from the studied belemnites. This assessment has been reinforced by Hoffmann et al. [42] and Stevens et al. [33], based on data obtained in Belemnitella mucronate and Megateuthis gigantea and in Neohibolites minimum, respectively. Although it has not been possible to observe current porosity or pristine organic matter in belemnite rostra, the presence of high organic matter content together with the high porosity observed in the Sepia prong, mostly along the "apical area" (Figures 10F,I and 13C-E), is in agreement with the interpretation that belemnite rostra were originally porous [1,2,31,[37][38][39][40][41][42]. The areas displaying the highest porosity, probably related to the higher organic matter content [40] and its decay, such as the apical line and growth rings displaying intense FL, were more susceptible for diagenetic alteration [30,31]. In this regard, the apical line commonly displays a cloudy appearance and bright orange-reddish luminescence due to diagenetic alteration [30,31,33,38,40,43,44,99]. Based on that, and on the high organic matter content observed along the apical line, Stevens et al. [33] proposed that the apical line of belemnite rostra is an original feature, most probably of organic origin. However, this is not in agreement with observations made by Bandel et al. [43] (1984), Benito et al. [31] and in this work (see Figure 6, Figure 5 of Bandel et al. [43], Figures 3A-F and 5A-D of Benito et al. [31]). As shown in these figures, primary spherulites are observed along the apical line, despite its cloudy appearance and bright luminescence, suggesting that the apical line was not fully organic, but an organic-rich area [31,40,43,99], where diagenetic fluids could easily percolate, because of its relatively higher original porosity, favoring the breakdown of the organic matter and the diagenetic alteration. This is also in agreement with observations made in the apical area of the Sepia prong ( Figure 10F,I), which is composed of spherulites and displays high porosity, and with observations made by Yancey and Garvei [56], who reported that "the prong growth axis and the prong's median fissure of Anomalosaepia is less mineralized than other areas of skeleton, suggesting a high proportion of non-mineralized organic material that decayed after death.
Thus, the strong growth convergences observed between the belemnite rostra and the Sepia prong (Table 1) suggest that growing of belemnites occurred similarly to the growth of the prong of sepiids and that the Sepia prong is the analog of the belemnite rostrum solidum.
BR: Concentric pattern is formed by an alternation of thin layers, displaying intense FL (organicrich) and layers displaying internal radial micro-fibrous texture and weak FL (organicpoor) (Figure 7). SP: Concentric pattern is formed by an alternation of thin and organic-rich laminae and organic-poor layers comprising radially arranged aragonite crystals ( Figures 10I and 13).

Non-classical mineralization
Non-classical crystallization processes are proposed to be involved in the formation of Sepia endoskeleton Cuif et al., [90], Benito et al., [31] Another question to address is how to interpret, in the Sepia prong, the fact that single aragonite crystals are internally composed of aligned clusters of nanometer-sized crystalline units, which are iso-orientated and include abundant organic matter inclusions (Figures 12, 13F and 15C,D). These structures resemble those observed in calcite crystals of the studied belemnite rostra ( Figure 7E,F) and those reported by Cuif et al. [90] in belemnites (see Figure 7.42 of Cuif et al. [90]). These authors noted that radial prims of belemnites were internally composed of clusters of nanometer-sized isoorientated crystalline units, suggesting that fusion of their crystal lattices formed the broader units under a non-classical crystallization process. Non-classical crystallization has been proposed as the mechanism for the biomineralization of living and fossil organisms, such as some echinoderms, corals, foraminifera and mollusks, including belemnites [31,90,[101][102][103][104]. Thus, the observations made in Sepia also suggest that non-classical crystallization is probably involved in the growth of their cuttlebone. These observations also suggest the possibility that small, biologically controlled and isoorientated aragonite crystalline units were fused into larger crystals, leaving intra-and intercrystalline pore spaces, similarly to what has been interpreted for belemnites [31,90].

Conclusions
The analyses of Jurassic belemnites (Hibolithes, Belemnopsis and Passaloteuthis) and cuttlebones (from Sepia officinalis and S. orbignyana) show that biomineralization of these shells was strongly related with the presence of organic matter in their structure. The organic matter in the cuttlebones is present in the phragmocone, dorsal shield and prong, always in alternation between organic-rich and organic-poor layers. In the case of belemnite rostra the organic matter is not directly observed, but the epifluorescence and BSEM pattern, the distribution of Mg, Na and S content and the pattern observed when rostrum solidum is etched with glutardialdehyde, indicates that organic matter was incorporated into the belemnite skeleton during progressive growth, preferentially in the apical area and in some concentric layers. Similarly, organic-rich and organic-poor (aragonite-rich) concentric layers of aragonite crystals are interlayered in the Sepia prong.
In belemnite rostra, calcite crystals start growing from spherulitic centers of growth, developed along the apical line, from which calcite crystals emerge and diverge becoming larger and prismatic towards the external wall of the rostra. These calcite crystals traverse the concentric growth layering of the rostra, display undulose extinction, and internal micro-fibrous texture. This microstructure resemble that observed in the prong of Sepia.
Additionally, the presence of high organic matter content together with the high porosity observed in the Sepia prong, mostly along the "apical area" is consistent with the interpretation that belemnite rostra were originally porous, that the areas displaying the highest porosity (the apical area and the organic-rich concentric layers) were more suitable for diagenetic alteration, and that caution should be taken when analysing geochemically belemnites for performing palaeoenvironmental interpretations.
The strong growth convergences observed between the belemnite rostra and the Sepia prong suggest that growing of belemnites occurred similarly to the growth of the prong of sepiids and that the Sepia prong is the analog of the belemnite rostrum solidum.
Finally, the fact that single aragonite crystals of the Sepia prong are internally composed of aligned clusters of nanometer-sized crystalline units, which are iso-orientated and contain abundant organic matter, suggest non-classical crystallization, as has been previously interpreted for other living and fossil organisms.