A stable ultrastructural pattern despite variable cell size in Lithothamnion corallioides

Recent advances on the mechanism and pattern of calcification in coralline algae led to contradictory conclusions. The evidence of a biologically controlled calcification process, resulting in distinctive patterns at the scale of family, was 10 observed. However, coralline calcification process has been also interpreted as biologically induced, because of the dependency of its elemental composition on environmental variables. To clarify the matter, five collections of Lithothamnion corallioides from the Atlantic Ocean and the Mediterranean Sea, across a wide depth range (12-66 m), have been analyzed for morphology, anatomy, and cell wall crystal patterns in both perithallial and epithallial cells, to detect possible ultrastructural changes. L. corallioides shows the alternation of tiers of short-squared and long-ovoid/rectangular cells along the perithallus, 15 forming a typical banding. The perithallial cell length decreases according to water depth and growth-rate, whereas the diameter remains constant. Our observations confirm that both epithallial and perithallial cells show primary (PW) and secondary (SW) calcite walls. Rectangular tiles, with the long axis parallel to the cell membrane forming a multi-layered structure, characterize the PW. Flattened squared bricks characterize the SW with roundish outlines enveloping the cell and showing a zigzag and cross orientation. Long and short cells have different thickness of PW and SW, increasing in short cells. 20 Epithallial cells are one up to three flared cells, with the same shape of the PW and SW crystals. Despite the diverse seafloor environments and the variable L. corallioides growth-rate, the cell walls maintain a consistent ultrastructural pattern, with unaffected crystal shape and arrangement. A comparison with two congeneric species, L. minervae and L. valens, showed similar ultrastructural patterns in SW, but evident differences in the PW crystal shape. Our observations point to a biological control rather than an induction of the calcification process in coralline algae and suggest a possible new morphological 25 diagnostic tool for species identification, with relevant importance for paleontological applications. Finally, secondary calcite, in the form of dogtooth crystals that fill the cell lumen, has been observed. It represents a form of early alteration in living collections which can have implications in the reliability of climate and paleoclimate studies based on geochemical techniques.

3 tool to define the phenotypic expression of genotypic information (Auer and Piller, 2020). The compelling evidence of a biological control over calcification in coralline algae was provided by the identification of family-specific cell wall ultrastructures. Epithallial cells in the genus Lithothamnion show crystal units as thin rectangular blocks (Auer and Piller, 2020). Seasonality, including seawater temperature oscillations and photoperiod, is considered one of the main factors affecting the growth-rate and the biomineralization process (Steller et al., 2007;Kamenos and Law, 2010;Vásquez-Elizondo and 70 Enríquez, 2017), which may influence the ultrastructural pattern.
The identification of CCA in present-day integrative taxonomy is based on genetic methods coupling with the morphological description and measurement of diagnostic features. Species identification in the fossil record is, on the contrary, merely based on the preservation of morphological taxonomic characters. Consequently, the identification of valuable morphological characters as a tool for the definition of the paleontological species represents an important challenge. CCA are well 75 represented in the fossil record and L. corallioides has been reported in the Pliocene of Spain (Aguirre et al., 2012) and in the Pleistocene of Southern Italy (Bracchi et al., 2014).
This study is aimed at describing the ultrastructural mineralogical features of L. corallioides from different geographic settings (northeastern Atlantic Ocean and Mediterranean Sea) and across a wide bathymetric interval (12-66 m) to test if the ultrastructural pattern preserves under different environmental conditions, and therefore can be considered as an evident sign 80 of a biologically-controlled mineralization. Moreover, the identification of specific ultrastructural pattern could be considered as a valuable tool for species identification to be used also in paleobiology. L. corallioides has been targeted because of its wide distribution, both geographically and bathymetrically, and its occurrence in the fossil record.

Materials and methods
For this study, we considered five collections (Fig. 1, Tab. 1), from two different geographic settings: the Atlantic Ocean and 85 the Western Mediterranean Sea, sampled by scuba-dive or grab at different depth ranging from 12 m in Morlaix Bay (France), down to 66 m (Pontine Islands, Italy). All specimens have been collected alive. Table 1 reports locations and dates of sampling.
To highlight possible ultrastructural differences among the same genus, two additional collections, already identified as Lithothamnion minervae (Basso, 1995) and Lithothamnion valens Foslie 1909 have been considered. These collections have been sampled alive from Egadi Islands (Italy), during July 2016, at 103 and 86 m of water depth respectively. 90 All samples have been air dried, sheltered from sunlight. Once dried, they have been stored in plastic boxes with silica to avoid any decay and transported to the laboratories of the University of Milano-Bicocca.

Coralline sample preparation
Samples have been prepared for Scanning Electron Microscope (SEM) imaging. Altered, badly preserved or encrusted branches have been discarded. Only the branches showing a shiny surface have been picked from all collections, each controlled under a Stereo Microscope, and cleaned manually by removing epiphytes and other encrustations. Each sample, composed of multiple branches, has been cleaned in an ultrasonic bath in distilled water for 10 minutes and air-dried. The 100 branches were then placed in small cylindrical plastic boxes with a base diameter of 1". Branches have been piled up and aligned to obtain multiple layers. The samples were embedded in Epofix resin for SEM analyses, which was stirred for 2 minutes with a hardener (13% w/w), and they were left to harden for one day at room temperature. Samples have then been cut normal to the multiple layers by using a IsoMet diamond wafering blade 15HC, along the direction of branch growth. The number of branches per sample are indicated in Table 1. Moreover, two additional samples (L. minervae and L. valens) have 105 been prepared for SEM observations (n = 10 branches), by breaking them with a small chisel. Both longitudinal and surface sections have been selected for SEM observations.

Scanning Electron Microscope
For SEM imaging, the surfaces of embedded samples have been polished by using different sizes of silicon carbide, cleaned ultrasonically in distilled water for 10 minutes and air-dried. Samples mounted directly on stub have been simply chrome-110 coated. SEM images have been taken with a Field Emission Gun Scanning Electron Microscope (SEM-FEG) Gemini 500 Zeiss, and a Tescan VEGA TS 5136XM. Standard magnifications for SEM images were established (~2500X, ~5000X, ~10000X, ~20000X and ~30000X), to describe comparatively and measure growth bands and cells, the morphology of Mgcalcite crystals, and the main features of perithallial and epithallial cell walls. A rigorous control over cell orientation is required to represent, describe, and measure in 2D the main features of a 3D structure such as cell calcification (both PW and SW). 115 Longitudinal axial sections of branches are a standard representation for calcareous red algae, allowing for subsequent visual comparison (Woelkerling, 1988;Quaranta et al., 2007;Burdett et al., 2011). Surface tangential sections are useful to describe the epithallial cells. Transverse or oblique sections are useful to describe qualitatively the three-dimensional aspects and organization of calcite crystals composing both PW and SW. Description of the cell wall structure follows the nomenclature of Flajs (1977), presenting the primary (PW) and secondary (SW) calcifications of the wall. Some authors refer to PW as 120 interstitial calcite (Ragazzola et al., 2016) or interfilament calcite (Nash and Adey, 2017;Nash et al., 2013Nash et al., , 2015. Cell dimensions have been measured as reported in Figure 2 (n=10 per sample), exclusively on longitudinal sections (Fig. 2).
Separation among adjacent filaments was not always obvious (Fig. 2c). In such cases, PW of adjacent cells has been measured in total (green line in Fig. 2c) and then half of the total was attributed to each cell. In the text we use the term ultrastructure to identify the singular crystal into specific layer of the cell wall, and ultrastructural pattern to indicate the combination and 125 mutual organization of crystals in layers of the cell wall.

Statistical analyses
Spearman and Pearson's correlations were used to test the statistical relationship between the cell measurements in both long and short cells, including morphometry and cell wall thickness. The linear correlation between the mean cell lengths and the sampling depths was measured by Pearson's coefficient, as well. One-way ANOVA and the Kruskal-Wallis test respectively 130 followed by the Tukey's test and the Dunn's test for post-hoc analysis was used to compare the cell measurements among sampling sites and to evidence the differences between group means and medians. All statistical analyses were run in R 3.6.3 software.

Growth rates
Growth rates were estimated under light microscope by measure a linear transect on the longitudinal section and counting how 135 many growth-bands of fourth order sensu Foster (2001) intercept. The growth-rate has been calculated by dividing the length of the transect by the number of growth-bands.
Once cut, all samples show the same micromorphology ( Fig. 3a-i, Supplement 2), with the constant occurrence of an easily detectable banding due to the alternation of series of short and long cells ( Fig. 3a, b) in bands of the fourth order corresponding to 1 year (Foster, 2001). No reproductive structure (conceptacle) was detected. 145 Along the perithallus, long cells are ovoid to rectangular in shape ( Fig (Fig. 6b). The bricks form a sort of envelope around the cell (Figs. 6b, 7e, f) showing sometimes a zigzag and cross orientation (Fig. 7g, h). 6 Different thicknesses characterize PW and SW of short and long cells (Figs. 3,5,7; Table 2, Supplement 2) in longitudinal 165 medial sections. Both PW and SW of short cells are generally thicker than in long cells (Table 2), even if the thickness does not show a correlation with sampling bathymetry. SW thickness ranges between 1.52±0.65 µm (Pontine) and 2.20±0.45 µm (Elba) in short cells, and between 0.57±0.14 µm (Morlaix) and 1.26±0.42 µm (Santa Caterina shoal) in long cells (Table 2).
Both PW and SW show fibrils (Borowitzka, 1982) forming a dense network in support of the mineralization (Fig. 7d).
The epithallus is formed by one up to three flared cells in longitudinal sections (Figs. 3g-j, 5f, 7c), always mineralized, with 170 some exceptions in the top distal surface (Fig. 3 g-j). The cell wall shows the same ultrastructural features of the perithallial cells, with both PW and SW mineralized (Figs. 3i, j, 5f, 7c).
Basing upon image analysis and time of collection, the calculated growth-rate ranges from 0.10 mm/yr in Pontine to 0.13 mm/yr in both Morlaix and Egadi.
The two additional collections of L. valens and L. minervae (Fig. 8) are characterized by both PW and SW (Fig. 8a, b, e, f). 175 The SW shows, in both species, an ultrastructural pattern like the one described for L. corallioides , with oriented bricks with rounded margins variably orientated, only apparently elongated and radial to the cell lumen in longitudinal sections (Fig. 8b, f). If observed into the cell lumen, where the cell membrane is lost, SW shows bricks with different orientation and sometimes a zigzag and cross orientation (Fig. 8d, g).
On the contrary, PW shows a different shape and arrangement of crystals, which are not characterized by the tiles of L. 180 corallioides observed in Figures 5d and 6a. Calcite crystals are squatter and more granular (Fig. 8a) or with irregular shape (Fig. 8h). One interesting aspect is that both samples show the occurrence of secondary calcite in form of dogtooth crystals filling the cell lumen (Fig. 8b, c, e).

Statistical analyses on morphological parameters
The differences in the long cells' morphometry and wall thickness among sampling sites are statistically significant for each 185 measured parameter (p<0.05; Supplement 1) (Fig. 9). Interestingly, the long-cell length of the deepest sample from Pontine Isl. (66 m depth) is lower than the others (p<0.001) (Figs. 4,9;Supplement), while in the shallowest sample collected in Morlaix (12 m depth) cells are remarkably longer (p<0.001) (Figs. 4, 9; Supplement 1).
In short cells, significant differences result only for cell ( Fig. 9) and lumen lengths, and cell PW (p<0.05; Supplement). The shortest cells are observed again in the sample from Pontine Isl., differing from the one collected in Morlaix (p<0.01) ( Fig. 9; 190 Supplement 1), which outstands for the highest values. On the contrary, the cell diameter slightly varies among sites, showing significant differences just in long cells (p<0.01; Supplement 1) (Fig. 9) Although banding is reported for all samples, elongation decreases with increasing depth, showing a strong inverse correlation 195 in long cells (p<0.01; r=0.98) (Fig. 9). The same trend is also observed in short cells, although with non-significant values (p=0.09) (Fig. 9).

Discussion
Properly oriented longitudinal and transverse/oblique sections are mandatory to obtain a precise comparative description of the main morphological features of CCA. Multi-scale approaches are also relevant, among which the ultrastructural pattern 200 may represent a new powerful and strategic diagnostic tool (Figs. 3-7).
Recent studies based upon genetic identification exclude the occurrence of other Lithothamnion species in the maerl of Morlaix Bay (Carro et al. 2014;Melbourne et al., 2017). Based upon this identification and considering the macroscopic features, the thallus pattern, the microanatomy, the morphology, and the morphometry of cell walls (Figs. 3, 5-7, Table 2), we identified the samples from Morlaix Bay as belonging to the species L. corallioides. 205 We compared the Mediterranean specimens with the Atlantic L. corallioides, and upon corresponding morphology, anatomy, and ultrastructure (Figs. 3, 5-7, Table 2), we considered them as conspecific.
The perithallus of L. corallioides clearly shows the alternation of growth bands of third and fourth orders (Fig. 3a, b), in agreement with Foster (2001). Fourth order bands represent the annual cycling, whereas third order ones represent seasonal variations and can be firstly distinguished by an evident chromatic change due to the different calcification thickness between 210 long and short cells (Foster, 2001). In our samples, the banding (third and fourth orders) is easily recognizable (Fig. 3), and both long and short cells length decrease across depth (Figs. 4,9), as expected, mirroring a decrease in growth-rate, whereas the diameter variation is significantly lower (Fig. 4, Supplement 1). Giraud and Cabioch (1979) observed that a cell wall fracture in L. corallioides shows a layer of radial calcite crystals (SW) separated from its neighbor by a different sheet composed of tangential crystals (PW). A discontinuity that coincides with the 215 fibrillar matrix observed in sections of decalcified material marks the limits of adjacent cellular frames (Giraud and Cabioch, 1979). Our results match with the observation of these authors in longitudinal sections (Figs. 3, 5, 7), although the discontinuity between adjacent cells is not easily detectable because of the complete mineralization.
Auer and Piller (2020) built a morphological tree based upon the observation of different ultrastructural patterns in epithallial cells, which match with the phylogenetic tree at family level. For Hapalidiaceae, and in the Lithothamnion-type epithallial 220 ultrastructure, they observed the occurrence of PW with primary crystals formed along the middle lamella and the SW with secondary rod-shaped crystals, also presenting fan-like structures. The samples studied in the present work show PW and SW both in perithallial and epithallial cell walls (Figs. 3, 5, 7). Both are apparently composed of rod-like crystals in longitudinal sections. However, in longitudinal sections that are locally tangential to the PW, the apparent rods reveal to be the longest and thinnest side of variably oriented rectangular tiles (Figs. 5, 6a). The tiles that envelop the cell (Fig. 5d, f) are the basic 225 ultrastructural elements forming the PW. Differently, apparent rods of the SW reveal to be squared and relatively flat bricks with rounded margins, as observed at the cell lumen without membrane or exactly at the contact between SW longitudinal section and cell lumen (Figs. 6b, 7). Crystals in longitudinal sections of SW are radial to the cell lumen, in agreement with Giraud and Cabioch (1979), and appear formed by small grains fused together (Figs. 3, 7b, f). They can also show a zigzag and cross orientation (Fig. 7g, h) like the fan-delta structure described by Auer and Piller (2020). 230 8 Therefore, our findings agree with the results of Auer and Piller (2020), although providing a more detailed description of the ultrastructural pattern of L. corallioides (Fig. 6).
Despite the different environmental conditions, likely occurring at the different sampling sites and depths, the ultrastructures of both PW and SW seem conservative and detectable in all samples. Therefore, the ultrastructures and the ultrastructural pattern are not dependent on environmental controls. 235 However, L. corallioides shows variable cell elongations (Table 2) and growth-rates, both decreasing according to sampling depth (Fig. 9), and a variation in PW and SW thickness, generally greater in short than in long cells, unrelated to depth ( Table   2, Supplement 1). These features possibly represent the effect of the different environmental conditions in which it lives, that do notaffect the ultrastructural pattern.
As defined by Lowenstam (1981), a biological controlled biomineralization is recognized in organisms with extensive control 240 over their mineral formation, resulting in well-ordered mineral structures with minor size variations and species-specific crystal habits, as we detected in L. corallioides. Despite some analogies with the observation of Nash et al. (2019), our findings support that the biomineralization process in CCA is biologically-controlled (Borowitzka, 1984;Cabioch and Giraud, 1986), rather than induced (de Cervalho et al., 2017;Nash et al., 2019). The two additional samples, L. minervae and L. valens, show distinct styles of PW calcification, and this is extremely interesting for its application in Paleontology. Ultrastructures and 245 ultrastructural pattern possibly represent a powerful tool for morphological species identification. Further investigation is needed to clarify the validity of this hypothesis in other genera/species. Finally, the occurrence of calcite in form of dogtooth crystals filling the cell lumen (Fig. 8b, c, e) is an exceptional finding.
The voids of the cell lumens allowed the development of calcite crystals which in term of shape, size and pattern are completely different from the original ones forming the cell wall calcification. It represents a form of extremely early alteration, possibly 250 diagenesis, in collections that were alive at the time of sampling. The phenomenon of earliest diagenesis, as already observed in Holocene and live Scleractinia corals (Nothdurft and Webb, 2008;Rachid et al., 2020), has implications in the reliability of climate and paleoclimate studies based on geochemical techniques, also when applied to recent collections.. Therefore, the possible occurrence of dog-tooth calcite must be carefully checked when selecting coralline algae samples for geochemical investigations. 255

Conclusions
We define the cell-wall ultrastructural pattern of L. corallioides as follows: -perithallus with evident banding as the result of the alternation of series of short-squared and long ovoid/rectangular cells; -epithallus with one up to three flared cells; -same and consistent ultrastructural pattern of the cell walls both in perithallus and epithallus, with PW and SW calcite walls 260 always present; -PW characterized by rectangular tiles; -SW characterized by flattened squared bricks with roundish outlines; 9 -long and short cells have similar diameter, with different thickness of PW and SW, resulting mainly in a thicker PW and SW in short cells. 265 The variable cell elongation, decreasing according to depth, and producing an evident banding, never affects the ultrastructural pattern, that maintains the same arrangement also under different growth-rates. These findings support that the CCA calcification process seems to be biologically-controlled rather than induced. The comparison with other Lithothamnion species highlights differences in the mineralization pattern of PW. Therefore, the ultrastructure of the cell wall in CCA results to be a promising new diagnostic tool for species identification with important potential application in Paleontology. Lastly, 270 an early alteration phenomenon, at the scale of ultrastructures, has been identified for the first time in living coralline algae.  h) the fracture shows detail of crystals composing the PW (white arrow) apparently composed of granules. SW is characterized by elongated crystals in longitudinal section (black arrow). Scale bar = 1 µm. 500 23 24 Figure 9: Correlation plots showing the relationship between sampling depth and cell lengths, measured in both long and short cells. Pearson's correlation significance at p<0.05. Pink is for Morlaix Bay (France, 12); green is for Egadi islands (Italy, 40); red is for Santa Caterina shoal (Italy, 40); yellow is for Elba Island (Italy, 45); blue is for Pontine islands (Italy, 66). 505 25 Tables