Scanning electron microscopy (SEM) was necessary for the observation of nacre tablets, which are only a few µm in circumference and around 500 nm thick. Our observations of the growing surface of the pearl oyster shell provided information about changes in the shape of the tablets. At a depth of 7 m, nacre tablets were elongated hexagonals (Figure 1) in the first and second year. At a depth of 20 m, the tablets were regular hexagonals (Figure 2). Figure 3 shows the elongation of the nacre tablets between a depth of 30 m (second year) and a depth of 39 m (first year) before the tablets became rhomboid. Figure 4 shows the different shapes of the nacre tablets as a function of depth. A change in the long axis in the growth direction was observed. Because of their shape, the growing tablets do not come into contact with each other as soon as for the other shape. The tablets can thus continue to grow and are finally bigger. This is related to the way tablets grow following a Voronoi process [9].

About 10 measurements of δ 18 O were performed in the aragonite part of the shell, in the inner nacreous layer to make sure that the shell had been deposited during the course of the experiment. δ 18 O values varied from −2.48 to −0.94‰ (average −1.86 ± 0.50 ‰, n = 10) for the shell deposited at a depth of 39 m, from −2.65 to −0.83‰ (av. −1.70 ± 0.67‰, n = 13) for a depth of 20 m, from −2.96 to −0.02 ‰ (av. −1.59 ± 0.96 ‰, n = 12) for a depth of 7 m, and from −3.09 to −1.09‰ (av. −2.38 ± 0.63‰, n = 9) for the control.

The δ18O in carbonate is determined by the δ18O of the water and by the temperature at which the carbonate is deposited. In biogenic carbonates, there is also an additional effect called the “vital effect”. This effect varies with the organism [10]. To our knowledge, to date there has been no study of the influence of depth on the δ18O of shell. Indeed, even if at the depths of culture used in this study, no influence on the isotopic fractionation coefficient would be expected, depth could have an impact on the “vital effect”. The prevailing environmental conditions at the time of the experiment (2004) resulted in temperature homogenization (24 °C), whatever the depth of the lagoon. Consequently, observed variations in the δ18O of the shell could only be due to the influence of depth. At equilibrium, δ18O can be calculated using the formula of Grossman and Ku [11] modified by Goodwin et al. [12]:

T(°C) = 20.6 − 4.34 (δ18Oaragonite − (δ18Owater − 0.2))

where δ 18 Oaragonite and δ 18 Oshell are expressed relative to SMOW and PDB, respectively. Here, we assume that the δ 18 O of seawater is 0‰. With a temperature of 24 °C, the δ 18 O at equilibrium is −0.98‰. The measured average δ 18 O of the shell ranged between −1.59 ± 0.96 ‰ and −2.38 ± 0.63 ‰. Therefore, all the values are depleted in 18O relative to the equilibrium value, i.e., the so-called vital effect.

The δ18O values in each shell displayed wide variability, between 1.54 and 2.94‰ (Figure 5a). Despite this variability, the average δ18O for each shell remained similar at the four different depths considered here (Figure 5b). The only difference between the depths was that δ18O variability appeared to decrease with depth (Figure 5a). Thus the change in the shape of the nacre tablets had no detectable effect on the δ18O of the shell. The fact that depth appears to play no significant role in the δ18O of the shell has major implications for paleoceanographic reconstructions.

The thickness and size of the nacre tablets was measured using scanning electron microscopy. Thickness was measured on SEM images of shell fractures not on the presented SEM pictures (pictures not shown). We measured the 10 most recently laid down layers close to the surface of the shell and the 10 oldest layers of the same shell as internal control. The oldest layers had been secreted during the period before the experiment. We then calculated the difference between the two nacre thicknesses as a percentage of the thickness before the experiment began. The thickness of the tablets decreased by between 16 and 30% on average (Figure 6). In all cases, the change in the environment resulted in a reduction in the thickness of the tablets. The biggest variations were observed at the two extreme depths: close to the surface (7 m) and far from the surface (39 m), where conditions are extreme. The second year the reduction in thickness was more important at 7 and 20 m depth than the first year.

At a depth of 39 m, the shape of the tablets changed from hexagonal to rhomboid, which resulted in a change in size (Figure 7). The size of the tablets increased from 7 to 21 μm2 with depth. These observations have to be correlated to the growth mechanism of nacre tablets [13]. Tablet formation is the result of a highly dynamic biomineralization process during which organic materials and carbonate precursor phases are delivered to the site of nacre tablet formation from the overlying mantle with a high degree of spatial control. Modifications of the environment like the depth have a direct impact on the shell growth at the scale of tablets.

It has been known for many years that altered gravity, hyper- as well as hypogravity, can influence mineralized structures. Statoliths in Aplysia californica decrease in size under hypergravity [14]. Also, for cichlid fish larvae, hypergravity results in a significantly smaller size of otoliths compared with controls in 1 G. No morphological differences with respect to the general shape of the otoliths and the overall pattern of daily incremented ring layers could be found [15]. Marxen et al. [16] studied the influence of microgravity on the embryonic development of the freshwater snail Bomphalaria glabrata and they observed a slower development than under standard conditions. In older embryos a decreased mineralization of the shell was detected. In both cases the shell formation proceeded normally. In our study we applied hypergravity and observed a modification of the mineralized structure at the tablet scale.

We studied one-year-old Pinctada margaritifera oysters grown in their natural habitat, at a depth of 12 m, in carefully monitored conditions at Marutea Marine Station, French Polynesia. We then kept the pearl oysters at several different depths at the same station for one week. The oysters were reared on down-lines as they were growing normally. We did not remove them from the line until the end of the experience to avoid any other stress. Only the depth of the line was changed. We used from 10 to 20 pearl oysters for each depth, depending on the numbers of the oysters attached to the line. This experiment was repeated one year later: the first year at depths of 7, 20 and 39 m, and in the second year at depths 7, 20 and 29 m, due to different diving conditions. After one week, the sample oysters were fixed in 70% ethanol. Five shells were analyzed for each depth.

The water content was gradually decreased and replaced by increasingly ethanol-rich solutions (3 day baths in 80%, 90% and 100% ethanol, respectively). After complete dehydration, the shells were opened, allowing the soft tissue of the mantle to be separated from the growing nacre. The shell samples were then air dried. After gold coating, observations were performed at the Electron microscopy service of Life Sciences laboratory, National Museum of Natural History (Paris, France) with a JEOL JSM-840A scanning electron microscope (SEM) operating at 15 kV.

A piece of each shell grown at a different depth was cut perpendicularly to the growth axis. The samples were mounted in epoxy and polished down to 1-µm with diamond paste. Ion probe oxygen isotope analyses of the aragonite part of the shell were performed at CRPG-CNRS (Nancy) using a Caméca ims 1270. The analytical settings were the same as those described in Rollion-Bard et al. [17]. Briefly, a primary beam of Cs+ ions with an intensity of about 8 nA was focused on a spot approximately 20 µm in diameter. A normal incidence electron flood gun was used to compensate for sample charging during analysis. The measurements were performed in multi-collection mode using two off-axis Faraday cups. Typical intensities were 5 × 106 counts/s (cps) on the 18O− peak. The acquisition sequence consisted of 20 cycles of 3 s duration each.

During analysis, the internal precision was about ±0.1 ‰ and the external reproducibility, based on repeated measurements of an in-house aragonite reference material (Arg, δ 18 OPDB = −7.18 ± 0.11‰, [18]) was ±0.25 ‰ (1 σ). The reference material was measured regularly throughout the analytical session.