Chemical Geology 355 (2013) 13–27

Contents lists available at SciVerse ScienceDirect

Chemical Geology

j ourna l homepage: www.e lsev ie r .com/ locate /chemgeo

Distribution of sulphur and magnesium in the red coral

Daniel Vielzeuf a,⁎, Joaquim Garrabou b,c, Alexander Gagnon d,1, Angèle Ricolleau a, Jess Adkins d,Detlef Günther e, Kathrin Hametner e, Jean-Luc Devidal f, Eric Reusser g, Jonathan Perrin a, Nicole Floquet a

a Aix-Marseille Université, CNRS, CINaM UMR 7325, Marseille, Franceb Institut Ciències Mar, CSIC, 08003 Barcelona, Spainc Mediterranean Institute of Oceanography (MIO) UM 110, CNRS/INSU, IRD, Aix-Marseille Université, Université du Sud Toulon-Var, Marseille, Franced Division of Chemistry, California Institute of Technology, Pasadena, CA, USAe Department of Chemistry and Applied Biosciences, ETH Zürich, Wolfgang Pauli Strasse 10, Hönggerberg, HCI, G113, CH-8093 Zurich, Switzerlandf UBP-OPGC-CNRS, 5 rue Kessler, 63038 Clermont-Ferrand, Franceg Institut f. Geochemie und Petrologie, ETH Zürich, NW E 85, Clausiusstrasse 25, 8092 Zürich, Switzerland

⁎ Corresponding author.E-mail address: vielzeuf@cinam.univ-mrs.fr (D. Vielz

1 Present address: School of Oceanography, Universi98195-5351, USA.

0009-2541/$ – see front matter © 2013 Elsevier B.V. Allhttp://dx.doi.org/10.1016/j.chemgeo.2013.07.008

a b s t r a c t

a r t i c l e i n f o

Article history:Received 31 October 2012Received in revised form 8 July 2013Accepted 9 July 2013Available online 15 July 2013

Editor: U. Brand

Keywords:Red coralMagnesiumSulphurStrontiumMediterranean seaSea water temperature

The concentrations of major and trace elements were measured in red coral skeletons (Corallium rubrum)by electron microprobe (EMP), isotope dilution inductively coupled mass spectrometry (ID-ICPMS) andlaser ablation-ICPMS (LA-ICPMS). The average composition (in mg/kg or ppm) is as follows: Ca:356300 ± 3200, Mg: 29500 ± 2400, Sr: 2600 ± 250, S: 3100 ± 400, Na: 4200 ± 500, K: 140 ± 20, P:140 ± 40, B: 28 ± 4, Ba: 9 ± 1, Fe: 8 ± 3, Li: 4 ± 1, Mn: 1 ± 0.5, Pb: 0.5 ± 0.3, U: 0.08 ± 0.05. In termsof Mg, the compositions of the red coral skeletons range from 9 to 15 mol% MgCO3 with a mean value of12 ± 1%. Concentrations of sulphur are high (approx. 3000 ppm) and among the highest reported in bio-genic calcites. EMP maps (Mg and S) and organic matter (OM) staining show a regular alternation of100–200 μm wide annual growth rings. Combination of these results with a previous study (Marschal etal., 2004) suggests that Mg-rich rings form during the period spring to early fall, while S-rich rings form im-mediately after (late autumn and winter). Elemental mapping by EMP shows an unexpected anticorrelationbetween S and Mg confirmed by LA-ICPMS. This anticorrelation is ascribed to the concomitant presence of Sin the organic matter and the anticorrelation between Mg and OM in the skeleton. However, mass balanceconstraints indicate that in the skeleton sulphur is probably present both as organo-sulphur and structur-ally substituted sulphur. The studied samples of red coral were collected at various locations and differentdepths (8–73 m) where the temperature of the sea water was monitored for long periods of time. Althoughoverall decreases of the Mg/Ca and Sr/Ca are observed as a function of depth (and temperature), the use ofthese ratios as an indication of the sea water temperature (SWT) seems difficult. In addition, a single colonythat grew in an area where the temperature was monitored for 30 years did not register the measured in-crease of SWT of about 1 °C. However, Mg and OM distributions inside skeletons could be good indicators ofvariations of growth rates on decadal time scales and anomalous ‘summer suffering’ events that could beassociated with periods of unusually high SWT conditions. The red coral is thus an example of how growthdynamics (and not temperature alone) affects the chemistry of biominerals.

© 2013 Elsevier B.V. All rights reserved.

1. Introduction

Biominerals display characteristic physical, morphological andchemical hierarchical patterns at length scales from the nanometer tothe centimeter (Lowenstam and Weiner, 1989; Mann, 2001). Physical

euf).ty of Washington, Seattle, WA

rights reserved.

patterns comprise the nature of the polymorph involved in the biomin-eral construction, variations in the crystallographic arrangement ofmodular units within the structure and periodic variations of densityand porous space distribution. Morphological patterns are related tothe shape of the organism, the shapes of the various constitutingunits, the shape of the pores and the morphology of growth fronts.Chemical patterns include variations in major and trace element con-centrations, stable and radiogenic isotope ratios, and organic matterconcentrations and compositions. The physical and morphological pat-terns often display self-similar characteristics; they are important to un-derstand the origin of the mechanical properties of biominerals, thedynamics of biomineral growth, and to find new routes to synthesize

14 D. Vielzeuf et al. / Chemical Geology 355 (2013) 13–27

newmaterials. On the other hand, periodic chemical patterns are essen-tial to understand how biominerals which are adaptative complex sys-tems respond to cyclic biological or environmental forcings. In addition,they represent important clues to understand the principles of bio-assisted crystal growth.

Biominerals are composite materials associating an inorganic part(crystalline or amorphous), organic matter and porous spaces. As aresult, it can be difficult to determine whether an element from thebulk composition is associated with the organic, inorganic part orboth, with important consequences on the interpretation of underly-ing processes. This is the case of magnesium, and more so of sulphurwhich is classically observed as a trace element in biogenic carbon-ates (Lorens and Bender, 1980; Blake and Peacor, 1981; Busenbergand Plummer, 1985; Rosenberg and Hughes, 1991; Dauphin andCuif, 1999; Vander Putten et al., 2000; Cuif et al., 2003; England etal., 2007; Hermans, 2010).

In this article, we present a study of the chemical composition ofred coral skeletons (Corallium rubrum). It complements previousworks on the physical and crystallographic organization of the skele-ton (Grillo et al., 1993; Vielzeuf et al., 2008, 2010). A comprehensivereview of the multi-disciplinary researches carried out on the redcoral of the Mediterranean (ecology, biology, management, trade, his-tory) will be found in the proceedings of the international workshopon red coral science (Bussoletti et al., 2010). The studied sampleswere collected at various places in the Mediterranean, at differentdepths where long-term instrumental records of seawater tempera-ture exist. The concentrations and the spatial distribution of major(Ca and Mg) and trace elements (Li, B, Na, Al, P, S, K, Cr, Mn, Fe, Zn,Rb, Sr, Ba, Pb, Bi, U) were determined. Elemental correlations arepresented and the relationships between chemical patterns, distribu-tion of organic matter, and sea water temperature are discussed.

2. Materials and methods

2.1. Biological samples and long time series of sea water temperature

Most of the studied red coral colonies come from the rocky coastnear Marseille (France) and the Medes Islands (Spain). Coloniesfrom Corsica (France) and Cap de Creus (Spain) were also studied(Fig. 1). Pieces of axial skeletons were analyzed by isotope dilution in-ductively coupled mass spectrometry (ID-ICPMS). On the other hand,axial skeleton sections (Fig. 2) weremounted in epoxy and polished forelectron microprobe (EMP) elemental mapping and chemical analyses,laser ablation inductively coupledmass spectrometry (LA-ICPMS) anal-yses, and backscattered electron (BSE) imaging. Polished sections ofaxial skeleton perpendicular to the axis (Fig. 2) and still surroundedby the dried organic tissues containing sclerites (grains of Mg calcite)were also prepared for EMP imaging.

Long term temperature series are scarce for the coastal waters ofthe NW Mediterranean. The longest series come from the MedesIslands (NE Spain). Weekly temperature measurements were initiat-ed in 1973 and cover a large range of depth levels (0–80 m depth)(Pascual et al., 1995; Salat and Pascual, 2002; Calvo et al., 2011). Like-wise, bi-weekly temperature measurements at different depth levels(0–60 m) are available since 1994 for the bay of Marseille (SOMLIT,Service d'Observation, Institut Pytheas UMS 3470). High resolutiontemperature series (hourly records) have also been acquired since1998 in theMarseille area, NWCorsica (France) and in theMedes Islands(NE Spain) using autonomous temperature recorders (StowawayTidbits). For more information on high resolution temperature seriessee Bensoussan et al. (2010) and www.t-mednet.org. These recorderswere located within benthic communities where red coral coloniesdevelop. Thus, coral samples that grewunder known temperature condi-tions for long durations can be studied to explore their potential as prox-ies for seawater temperature.

2.2. Analytical and preparation methods

In order to obtain comprehensive chemical information on the redcoral skeleton, various techniques with complementary capabilitieswere used. EMP provides good spatial resolution and allows accuratesurface mapping of Ca, Mg, S, Sr, Na, and P. On the other hand,LA-ICPMS has better sensitivity and precision than EMP for minorand trace elements (like Sr, Ba, Li, B). Finally, ID-ICPMS provideshigh precision measurements for selected metal/calcium ratios onbulk samples.

2.2.1. ID-ICPMSIsotope dilution-ICPMS analyses were carried out in the Division

of Geological and Planetary Sciences at Caltech. For theses analyses,red coral branches (5 to 8 mm in diameter) were sub-sampled trans-versely into pieces, each containing a complete circular section of thebranch. The 15–20 mg sub-samples were cleaned prior to analysis.For the cleaning process, samples were separately sonicated andrinsed in trace metal clean water four times over 30 min; sonicatedin dilute (~2%) sodium hypochlorite for 2 h (with one exchange of so-dium hypochlorite after 1 h); followed by another 30 min of sonica-tion and rinsing with clean water. The samples were then dried in aflow bench prior to grinding in a clean mortar and pestle. Portionsof the powders (0.2–1 mg each) were dissolved in trace metal cleannitric acid and analyzed for Mg/Ca and Sr/Ca by an isotope dilutionmethod on a Neptune multi-collector ICPMS. The process involves amixed element spike, an adaptation of the method described byFernandez et al. (2011). Repeated measurements of two consistencystandards during the analysis were used to assess precision, reportedas the 2 sigma std. deviation of more than 4 replicates. A referencestandard was made from a dissolved red coral and another fromdissolved scleractinian (aragonite) deep-sea coral. For both stan-dards, the relative Sr/Ca reproducibility was better than 0.1%, whileMg/Ca reproducibility was better than 0.4%. For a typical red-coralsample, this analytical precision corresponds to ±0.003 mmol/molSr/Ca and ±0.5 mmol/mol Mg/Ca. The average Sr/Ca and Mg/Ca ofthe scleractinian deep-sea coral consistency standard matches thelong-term mean for this standard, as measured by our lab overmore than two years.

2.2.2. LA-ICPMSTrace element concentrations in several spots within two red coral

skeleton samples (Medes1 and Riou-73) were determined using laserablation inductively coupled mass spectrometry (LA-ICPMS) at ETHZürich. The LA-ICPMS system combines a 193 nm ArF Excimer laserwith an ELAN 6100 ICPMS. Details of the instrument and analyticaltechnique are given elsewhere (Günther et al., 1997; Klemme et al.,2005). In this article, some emphasis will be put on sulphur. It isworth noting that the quantification of sulphur with LA-ICPMS re-mains problematic for several reasons (Guillong et al., 2008). Aspecific approach to overcome these difficulties developed by theZurich group has been applied here (Guillong et al., 2008). For thepresent study, the diameter of the laser beam was set to 40 μm. Foreach analysis, the gas background was measured for 30 s and the sig-nals for the samples were acquired for 10–30 s. The carrier gas usedwas 1 L/min helium, which was mixed with 0.75 L/min argon infront of the ICP. Calciumwas used as internal standard and its concen-tration was determined independently by EMP for each sample;internal variations of calcium content within each sample were nottaken into account. The NIST SRM 610 glass was used as externalreference material (Jochum et al., 2011). The two samples usedfor the LA-ICPMS measurements (Medes 1 — 20 analyses andRiou73 — 16 analyses) were collected at depth of 28 m and 73 m, re-spectively to illustrate two contrasted annual sea water temperature(SWT) regimes.

Fig. 2. (a) General view of a colony of Corallium rubrum (sample Cassis2013) covered with its dried organic tissues. (b) Enlargement of a portion of image a after partial removal of theorganic tissues (remnants of dried tissues appear orange). (c) Section of red coral after staining of the organic matrix (sample D12); [Oe] organic envelope, [Sk] skeleton. Inside frameswith numbers refer to the chemical maps displayed in Figs. 7 to 10. These frames are drawn for relative scale purpose and do not represent the exact locations and sizes of the chemicalmaps. (d) Section of red coral after staining of the organic tissues, enlargement of image c. Staining of the coral sections was done by C. Marschal from the Institut Méditerranéen deBiodiversité et d'Ecologie Marine et Continentale (Marseille, France).

Fig. 1. Map of the Mediterranean sea. The location of the studied samples is indicated in the inset. Geographic coordinates of collected samples: Medes Island: 42.050°N, 3.227°E;Cap de Creus: 42.323°N, 3.324°E; Riou Island: 43.174°N, 5.392°E; Scandola, Corsica: 42.382°N, 8.547°E. The red coral (Corallium rubrum) is an endemic species of the westernMediterranean. Due to hotter sea waters, the red coral is rare or absent south of the 35°N parallel and east of the 25°E meridian (Harmelin, 2000).

15D. Vielzeuf et al. / Chemical Geology 355 (2013) 13–27

6 8 10 12 14 16 180

100

200

300

400

MgCO3 (mol%)

S (ppm)

n

n

1000 2000 3000 4000 50000

100

200

300

entire population

single sample

a

b

Fig. 3. (a) Blue: histogram of MgCO3 content for all samples (mol%). n = 5310, Mean =12.0 ± 1.0. Yellow: histogram of MgCO3 content in a single sample (MedesCol5-Clermont)n = 877,Mean = 11.8 ± 1.0. (b) Blue: histogramof sulphur content (in ppm) for all sam-ples n = 3851, Mean = 3200 ± 400. Yellow: histogram of S content in a single sample(Medes1-ETH) n = 1045, Mean = 3200 ± 300. (For interpretation of the references tocolor in this figure legend, the reader is referred to the web version of this article.)

16 D. Vielzeuf et al. / Chemical Geology 355 (2013) 13–27

2.2.3. EMPMost X-ray images of Ca, Mg, and S were obtained on an SX100

Cameca electron microprobe (LMV, Clermont-Ferrand). The defini-tion of the X-ray images is usually 512 × 512 pixels with beamcurrent, counting times, and step interval in the range 30–50 nA,30–50 ms, and 1–5 μm. Three to four images were acquired at thesame time during sessions that lasted ~8 h. At ETH Zurich, andCaltech, electron probe microanalyses and elemental maps wereperformed on JEOL JXA 8200 instruments equipped with five wave-length dispersive X-ray spectrometers (WDS) and an energy disper-sive spectrometer (EDS). All samples were coated with a ~20 nmthick carbon layer. For elemental distribution maps, an accelerationvoltage of 15 kV, a beam current of 10 to 50 nA and a beam diameterof 1 to 2 μm were applied. Quantitative results (spot analyses andanalytical traverses) were also obtained. The compositions in weightpercent oxide were measured as CaO, MgO, SO3 (and SrO). Resultsobtained on the three different EMPs were processed using thesame structural formula calculation scheme. Since C and O were notindependently measured, CO2 was calculated by stoichiometry fromthe concentrations of CaO and MgO, considering that all Ca and Mgcombine into carbonate. The structural formulae were calculated onthe basis of 12 oxygens. Throughout this article, elemental concentrationsare given in ppm, elemental ratios (e.g. Ca/Mg, Sr/Mg) in mmol/mol andthe proportion of MgCO3 in mol%.

2.2.4. Organic matrix stainingThe organic matrix (OM) staining method described by Marschal

et al. (2004) is a powerful tool to locate OM and characterize the po-tential relationships between OM distribution and chemical zoning.Branches of red coral skeleton, either fresh or fixed for 24 h in a 4%formaldehyde solution and then preserved in 95% alcohol, were em-bedded in epoxy resin (Durcupan ACM FLUKA) following standardprocedures, and heated to 60 °C for 24 h. One millimeter thickcross-sections were made using a diamond saw. The sections wereglued (ESCIL D200 SF UV radiation polymerization) onto a micro-scope slide, and polished to obtain 50–100 μm thick slabs. The slabswere decalcified in 2% acetic acid solution for 4–5 h. After decalcifica-tion, special care was taken to avoid breakage of the organic matrixstructure. After a gentle rinse, the slabs were stained with toluidineblue at 0.05% for 10–30 s. Slabs were occasionally stained severaltimes to better reveal the organic matrix rings under the stereomicro-scope (Fig. 2).

3. Results

3.1. Elemental concentrations

Forty-one EMP analytical traverses representing more than 5000analyses have been carried out on 34 different samples. Each analysisrepresents the composition at a specific point location in a specificcoral. Four elements were detected in measurable amounts: Ca, Mg,S, and to a minor extent Sr. The range of composition observed withina single sample will be referred to as the internal compositional var-iability, in contrast to an external variability that corresponds to thevariation of composition observed in different samples.

3.1.1. Magnesium contentA histogram of proportion of MgCO3 in selected analyses (n =

5310) from all of the samples is shown in Fig. 3a (blue histogram).This histogram indicates that the composition of the red coral skeletonsranges from ~9 to ~15 mol% MgCO3 with a mean value of 12.0 ± 1.0%(1s.d.). The molar Mg/Ca ratio varies between 100 and 180 mmol/mol(mean value of 140 ± 10), in good agreementwith the values reportedin the literature (100 to 160 mmol/mol) (Milliman, 1974; Maté et al.,1986; Weinbauer et al., 2000; Hasegawa et al., 2010). To complementthe EMP data, 35 different colonies of red coral from 9 different depths

(from −8 to −73 m) were analyzed by ID-ICPMS. Data are given inTable 1. In these colonies, the Mg/Ca ratio ranges from 119 to137 mmol/mol in good agreement with the EMP values. For a directcomparison, the mean Mg/Ca ratios obtained by EMP are also given inTable 1.

In previous works, it has been shown that the Mg content within asingle sample varies at millimeter and sub-millimeter scale (Weinbaueret al., 2000), and at scales down to the micrometer (Vielzeuf et al.,2008). The quantitative variations of compositions within each sampleshave been determined by EMP analyses with high spatial resolution (afew μm3). The distribution of compositions in the sample MedesCol5 isshown in Fig. 3a (yellow histogram). The MgCO3 content ranges from10 to 14 mol%. Similar range of variation is observed in other samples.Thus, the variation of composition in a given sample (internal variation)is not significantly smaller than the variation of composition within theentire population of red coral skeletons (compare the blue and yellowhistograms in Fig. 3a). For each analytical traverse, themean compositionand its standard deviation were calculated. The mean value of standarddeviations (σ) determined for all these traverses is 0.72 mol% (mean in-ternal variation) to compare with 0.98 mol% for the entire population.Thus, the external variability can be almost completely explained bythe internal variability of composition observed within the samples.

3.1.2. Sr contentStrontium concentrations were measured by EMP in two samples

(RiouSud and Medes1B1). The histogram of distribution of Sr is

Table 1Sr/Ca and Mg/Ca ratios (mmol/mol) obtained by ID-ICPMS and EMP on different redcoral colonies collected at different depths (Riou island).

Sample ID Depth (m) Sr/Ca Mg/Ca Mg/Ca

(ICP-MS) (ICP-MS) (EMP)

S8_1 8 3.565 122.11S8_2 8 3.387 130.25S8_3 8 3.300 129.26S8_4B 8 3.279 124.24Mean value (8 m) 3.383 126.47 133s.d. (1sigma) 0.13 3.92 11S15_1 15 3.284 135.91S15_2 15 3.304 130.51S15_3 15 3.397 132.67S15_4 15 3.225 124.87Mean value (15 m) 3.302 130.99 128s.d 0.07 4.65 9S20_1 20 3.297 126.08S20_2 20 3.331 128.62S20_3 20 3.326 128.26S20_4 20 3.556 136.74Mean value (20 m) 3.377 129.92 131s.d. 0.12 4.68 13S25_1 25 3.342 129.86S25_2 25 3.272 126.60S25_3 25 3.351 129.55S25_4 25 3.408 132.45Mean value (25 m) 3.343 129.616 135s.d. 0.06 2.39 11S30_1 30 3.273 124.68S30_2 30 3.289 122.94S30_3 30 3.440 134.22S30_4 30 3.446 129.89Mean value (30 m) 3.362 127.93 131s.d. 0.09 5.13 10S33_1 33 3.212 120.67S33_2 33 3.291 122.59S33_3 33 3.162 119.80S33_4 33 3.267 126.72Mean value (33 m) 3.233 122.45 129s.d. 0.06 3.08 8S40_1 40 3.182 124.67S40_2 40 3.220 123.96S40_3 40 3.277 123.75S40_4 40 3.275 125.27S40_5 40 3.193 125.02Mean value (40 m) 3.229 124.53 129s.d. 0.04 0.66 8S60_1 60 3.228 121.88S60_2 60 3.313 126.74S60_3 60 3.324 125.30S60_4 60 3.193 119.04Mean value (60 m) 3.265 123.24 126s.d. 0.06 3.46 8S73_1 73 3.207 122.52S73_1 73 3.254 124.36Mean value (73 m) 3.231 123.44 130s.d. 0.03 1.30 7

1000

2000

3000

4000

5000

6000

24000 28000 32000 36000

S (

ppm

)

Mg (ppm)

EMP

LA-ICPMS

y = -0.092x + 5963R = 0.2972

y = -0.085x + 5348R = 0.6382

Fig. 4. Concentrations of sulphur versus concentrations of Mg obtained by EMP on 9different samples (blue dots and blue regression line, 1659 analyses). Concentrationsof sulphur versus concentrations of Mg measured by LA-ICPMS on two samples:Medes 1 (green dots, 20 analyses), Riou-73 (orange dots, 16 analyses). The green re-gression line applies for both the Riou and Medes samples. See also Table 2. (For inter-pretation of the references to color in this figure legend, the reader is referred to theweb version of this article.)

17D. Vielzeuf et al. / Chemical Geology 355 (2013) 13–27

shown in Fig. A1a (online supplementarymaterials). In the first sample,Sr concentrations range from 1970 to 3290 ppm with a mean valueof 2630 ± 250, while in the second sample concentrations aremore scattered with a mean value of 2580 ± 730. Sr concentrationswere also measured by LA-ICPMS in two other samples (Medes1 andRiou-73); the measured concentrations (range: 2370–2870, meanvalue: 2550 ± 140 ppm) are consistent with the EMP values. The Sr/Caratio measured by EMP varied between 2.5 and 4.3 mmol/mol, with amean value of 3.4 ± 0.3 in the first sample (RiouSud) and a meanvalue of 3.3 ± 0.9 in the second sample (Medes1B1). These values arein good agreement with the ratios determined by ID-ICPMS on 35 bulksamples (range: 3.16–3.56, mean value: 3.30 ± 0.09). Note that in thiscase the standard deviation is much smaller than for the EMP analyses.The strontium contents determined in our study are also in good

agreement with the values reported by Weinbauer et al.(2000) (2100–2900 ppm) and Hasegawa et al. (2010) (2500–2700 ppm).

3.1.3. Sulphur contentFig. 3b shows a histogram of distribution of the sulphur content of

the red coral skeletons measured by EMP. The concentrations vary be-tween 1950 and 4640 ppm with a mean value of 3130 ± 390. Thesevalues are consistent with the values determined by LA-ICPMS (Fig. 4,range: 2020–3120, mean value: 2700 ± 300 ppm) and with the sul-phur concentrations previously reported by Dauphin and Cuif (1999)(2870 ± 780 ppm). These are among the highest values of sulphurconcentrations recorded in biogenic calcites (England et al., 2007;Hermans, 2010). As with magnesium, Vielzeuf et al. (2008) exploredthe spatial distribution of S within red coral skeletons through EMP im-ages at different scales (their Figs. 3f and 5e). Sulphur contents vary sig-nificantly within a given sample at scales down to the micrometer. Anexample of histogram of distribution from a single sample (Medes1)is shown in Fig. 3b (yellow histogram). The concentrations vary be-tween 2100 and 4100 ppm with a mean value of 3200 ± 300. Thus,as in the case of Mg, the range of variation in a given sample is notmuch smaller than the variation of composition within the entire pop-ulation. This conclusion applies to all studied samples.

3.1.4. Other elementsLA-ICPMS analyses were performed along radii nearly coinciding

with EMP traverses. Results are given in Table 2. The detected ele-ments are listed from the highest to the lowest concentrations (Na:4200 ± 500, K: 140 ± 20, P: 140 ± 40, B: 28 ± 4, Ba: 9 ± 1, Fe:8 ± 3, Li: 4 ± 1, Mn: 1 ± 0.5, Pb: 0.5 ± 0.3, U: 0.08 ± 0.05 ppm).It is important to note that Na is an important component of theskeleton with concentrations usually higher than those of Sr and S.Other elements detected but close to the detection limit include Al,Cr, Zn, Rb, and Bi.

To summarize, the average composition of red coral skeleton de-termined in this study is as follows (in ppm): Ca: 356300 ± 3200,Mg: 29500 ± 2400, Sr: 2600 ± 250, S: 3100 ± 400, Na: 4200 ±500, K: 140 ± 20, P: 140 ± 40, B: 28 ± 4, Ba: 9 ± 1, Fe: 8 ± 3, Li:4 ± 1, Mn: 1 ± 0.5, Pb: 0.5 ± 0.3, U: 0.08 ± 0.05. For a comparison,the theoretical composition of a 12.5% MgCO3 Mg-calcite is Ca:357400, Mg: 31000, C: 122400, and O: 489200 ppm.

ID-ICPMS

LA-ICPMS

2000

2400

2800

3200

26000 30000 34000 38000

Sr

(ppm

)

Mg (ppm)

y = 0.429x + 1398R = 0.2872

y = 0.045x + 1153R = 0.7972

Fig. 5. Sr vs Mg measured by ID-ICPMS in 35 samples of red coral (black dots). Sr vs Mgmeasured by LA-ICPMS in 2 red coral skeletons (orange: Riou-73, green: Medes1). (Forinterpretation of the references to color in this figure legend, the reader is referred tothe web version of this article.)

18 D. Vielzeuf et al. / Chemical Geology 355 (2013) 13–27

3.2. Correlations between elements

In addition to the Ca–Mg anticorrelation expected from the solid-solution nature of the Mg-calcite, different inter-elemental relation-ships are observed.

3.2.1. Sulphur–magnesiumAs shown in Fig. 4, EMP analyses from several samples show a neg-

ative correlation between sulphur and magnesium. This anticorrelationobtained by high spatial resolution analyses is confirmed by LA-ICPMSinvolving larger sample volumes (~5 × 103 μm3) (Fig. 4). Small differ-ences in absolute values of Mg and S appear between the EMP andLA-ICPMS results; these differences can possibly be ascribed to thefact that EMP analyses are less accurate for elements in low concentra-tions (here sulphur) and that, on the other hand, the internal variationof Ca within the samples was not taken into account for processing theLA-ICPMS data. In spite of the small differences, it is important to notethat both datasets display an almost identical negative slope (Fig. 4).

3.2.2. Strontium–magnesiumData obtained with both ID-ICPMS (bulk compositions) and

LA-ICPMS (in situ analyses) show a positive correlation between Srand Mg (Fig. 5). The two data sets display almost similar slopes but

Table 2Element concentrations determined by LA-ICPMS on two red coral samples. Min: minimumless than 20 (Medes1) or 16 (Riou-73) indicate that some values were under detection lim

Element Isotopic mass Medes1

n Min Max Mean s.d

Li 7 20 2.60 6.18 3.95B 11 20 20.3 37.5 28.5Na 23 20 3420 5410 4208 6Mg 26 20 28400 37400 32645 26Al 27 7 0.375 0.919 0.568P 31 20 82 248 138S 32 20 2020 2990 2518 2K 39 20 109 139 124Cr 53 7 1.45 2.19 1.68Mn 55 20 0.48 2.22 1.17Fe 57 20 5.69 12.20 8.58Zn 66 10 0.57 1.36 0.92Rb 85 10 0.02 0.04 0.03Sr 86 20 2380 2880 2643 1Sr 88 20 2370 2870 2634 1Ba 137 20 7.45 9.45 8.33Ba 138 20 6.99 9.28 8.39Pb 208 20 0.215 0.625 0.368Bi 209 19 0.027 0.104 0.046U 238 19 0.032 0.162 0.095

a slight systematic offset. Here again, the fact that calcium was usedas an internal standard and was assumed constant in a given samplefor the LA-ICPMS measurement can explain part of the offset. On theother hand, while Sr/Ca and Mg/Ca ratios measured by ID-ICPMS areaccurate, the conversion of these ratios into absolute values of Mgand S in ppm requires assumptions (choice of structural formula,structural state of sulphur) that affect the final accuracy. Neverthe-less, the differences in concentrations determined using bothmethods remain small and most importantly, the two datasets dis-play similar positive slopes. Such a correlation between Sr and Mg isnot observed on the data obtained by EMP (Fig. A1b, online supple-mentary materials). We ascribe this discrepancy to inadequate EMPanalytical conditions for low Sr contents. Indeed, EMP analytical con-ditions were optimized for major element determination consideringthat biogenic calcite rapidly decomposes under the beam. Such condi-tions are not suitable for the measurement of elements in low con-centrations (by defocusing the beam and/or using longer countingtimes for instance) and thus lead to lower precision than expectedto resolve the Sr vs Mg variability.

3.2.3. Other elementsLA-ICPMS data allow the observation of other possible relation-

ships between elements. For some elements (e.g. Li and Ba — Fig. 6aand b) the plots against Mg define single trends as observed for S orSr. Other patterns are also observed: the concentrations of K in theRiou sample are significantly higher and do not overlap with the con-centrations determined in Medes1 (Fig. 6c). On the other hand, forthe three elements Na, Mn and B, the trends observed in the two sam-ples are continuous but seem to be characterized by opposite slopes(Fig. 6d to f).

3.3. Spatial distribution of Mg, S, and organic matter in present-dayskeletons

Preliminary data on the spatial distribution of magnesium and sul-phur have been provided by Vielzeuf et al. (2008). Some of these pre-vious results were re-processed and are presented here together withnew series of EMP images. Fig. 7a to c are examples of Mg and S map-ping at relatively low spatial resolution: the darker gray zones corre-spond to low elemental concentrations. Fig. 7a shows the ordereddistribution of Mg and the presence of concentric rippled growthrings with tortuous boundaries. Growth surfaces of red coral skeleton

value, Max: maximum value, s.d.: 1 sigma standard deviation, n: number of analyses. nit or, in some rare cases, anomalous and discarded.

Riou-73

. n Min Max Mean s.d.

1.10 Li 16 2.78 6.82 3.81 1.004.8 B 16 22.3 32.0 26.6 2.7

68 Na 16 3780 4590 4106 20867 Mg 16 27400 31400 29163 14640.191 Al 0 b0.3 b0.3 b0.3

44 P 16 113 205 140 2567 S 16 2730 3120 2936 1179 K 16 152 181 167 100.28 Cr 5 1.31 1.91 1.60 0.240.42 Mn 16 0.39 2.56 1.23 0.541.76 Fe 15 4.85 11.00 7.34 1.810.25 Zn 2 0.40 0.41 0.40 0.010.01 Rb 15 0.03 0.06 0.04 0.01

41 Sr 16 2350 2620 2469 7137 Sr 16 2370 2590 2455 630.73 Ba 16 8.00 11.30 9.37 0.850.60 Ba 16 8.36 10.9 9.43 0.800.118 Pb 16 0.323 1.110 0.686 0.2640.020 Bi 15 0.024 0.047 0.033 0.0060.041 U 16 0.022 0.096 0.048 0.018

2

4

6

8

10

Li (

ppm

)

5

10

15

Ba

(ppm

)

100

120

140

160

180

27000 29000 31000 33000 35000 37000

Mg (ppm)

K (

ppm

)

2400

3200

4000

4800

5600

6400

Na

(ppm

)

1

2

3

Mn

(ppm

)

14

22

30

38

46

27000 29000 31000 33000 35000 37000

Mg (ppm)

B (

ppm

)

a

b

c

d

e

f

y = 0.0003x - 6.239R = 0.5752

y = -0.0001x +13.427R = 0.2302

Fig. 6. Concentrations of Li, Ba, K, Na, Mn, and B plotted against Mg (ppm) in two red coral samples (orange: Riou-73, blue: Medes1). (For interpretation of the references to color inthis figure legend, the reader is referred to the web version of this article.)

19D. Vielzeuf et al. / Chemical Geology 355 (2013) 13–27

are characterized by the presence of micro-protuberances with com-plex morphologies pointing towards the outside (Weinberg, 1976;Grillo et al., 1993; Vielzeuf et al., 2008). The tortuous boundaries be-tween layers observed on the Mg images are traces of previousgrowth fronts. Fig. 7b is an EMP image of the distribution of S onthe same section than Fig. 7a. The image is not as sharp as Fig. 7adue to the low concentration of S, however similar features are ob-served. Fig. 7c results from the processing and superposition of Mgand S images (Figs. 7a and b). The highest concentrations of Mg andS are shown in blue and green, respectively. This image shows that,in general, Mg-rich layers are rimmed (outwards) by S-rich layers(Fig. 7c [A]). However, this qualitative anticorrelation is not perfectand exceptions can be found (e.g. Fig. 7c [B]). Thus, Mg and S areanticorrelated in space, though imperfectly, which is in agreementwith the general trend and the scattering of the data shown in Fig. 4(EMP data). A second set of images at higher spatial resolution(Figs. 7d to f) confirms that Mg-rich layers are rimmed on their exter-nal side by S-rich layers (Fig. 7f [A]).

What about the spatial relationships between organic matrix, Mg andS? OM stained sections of red coral axial skeleton (Fig. 2b) display amedullar region surrounded by concentric growth rings (see also(Marschal et al., 2004; Vielzeuf et al., 2008; Debreuil et al., 2011)).Each OM growth ring is composed of a dark thin band correspondingto a zone of high OM concentration and a wider brighter band where

the OM concentration is low. Fig. 8a is another example of stainedsection at relatively low spatial resolution already discussed byVielzeuf et al. (2008). The Mg and S distributions have been mappedon the same polished section (Fig. 8b and c): the brighter zones oneach image correspond to high concentrations of the element. Thesimilarities of the Mg and OM distributions are striking: both showidentical ring patterns with the same periodicity. The S image(Fig. 8c) shows similar patterns but less obvious than Mg. The natureof the spatial correlation between OM, Mg and S is explored inFig. 8d–f. These three figures correspond to a selected area shown inFig. 8a–c. Fig. 8d results from the superimposition of OM and Mg pat-terns. Thresholding was used to enhance the highest concentrationsof OM (red) and Mg (blue). The areas where OM- and Mg-richzones superimpose appear in purple (white arrows). Purple areasare present but not predominant which points out that OM andMg are anticorrelated. Fig. 8e corresponds to the superposition ofMg (blue) and S (green) images. The areas where Mg- and S-richzones superimpose appear in light blue (white arrows). Here again,such areas are not predominant indicating that Mg and S are alsoanticorrelated. Finally, Fig. 8f corresponds to the superposition ofOM (red) and S (green) images. The areas where OM- and S-richzones superimpose appear in yellow green (white arrows). Such yel-low green areas are widespread pointing out that OM and S arecorrelated.

Fig. 7. (a–b) Electron microprobe maps of Mg and S in a polished section of red coral (after Vielzeuf et al., 2008). (c) Superposition of images a and b after thresholding; zones ofhigh concentrations of Mg and S appear in blue and green, respectively. (d–e) Electron microprobe maps of Mg and S at higher spatial resolution in another section of red coral.(f) Superposition of images d and e after thresholding; zones of high concentrations of Mg and S appear in blue and green, respectively. (For interpretation of the references tocolor in this figure legend, the reader is referred to the web version of this article.)

20 D. Vielzeuf et al. / Chemical Geology 355 (2013) 13–27

3.4. Concentrations of Ca, Mg, S and Sr in the organic tissues

The fact that OM and S distributions are correlated suggests that Sis present in the OM contained inside the skeleton. An indirect way tosupport this hypothesis is to determine whether sulphur is present inthe main organic tissues surrounding the skeleton. Fig. 9 gathers Ca,Mg, S, and Sr electron microprobe images of the organic envelope[Oe] surrounding the skeleton [Sk]. Sclerites which are complexshaped grains of Mg-calcite embedded within the living tissues(Lacaze-Duthiers, 1864; Weinberg, 1976; Grillo et al., 1993; Floquetand Vielzeuf, 2011) appear on the four images (e.g. Fig. 9a [Sc]).Voids corresponding to sections of gastrodermal canals runningalong the skeleton axis appear in black on the images (Fig. 9a [Gc]).Organic tissues contain significant amounts of Ca and Mg thatenhance internal structures such as digitations, filaments, or smallgranules different in size from the sclerites. Like Ca and Mg, sulphuris present in the organic tissues but not uniformly distributed:S-rich granules (Fig. 9c [Sgr]) appear without any relation with theCa-rich granules observed in Fig. 9a. Strontium is present in both skel-etal structures (skeleton and sclerites) and organic tissues; no gran-ules are observed but anomalously enriched Sr zones borderinggastrodermal canals are observed without correlation with Ca, Mg,or S (Fig. 9d [Ez]). Fig. 10 corresponds to another set of elemental im-ages of a portion of organic tissues surrounding the axial skeleton, at ahigher spatial resolution. Most of the previous observations made onFig. 9 apply. Sulphur is detected in both the inorganic (sclerites andskeleton) and organic parts of the coral. Within the organic matrix,three levels of S concentration can be observed (Fig. 10d), a zone oflow concentration or background (zone 1 [Z1]) within which orga-nized patches of high sulphur content appear (zone 2 [Z2]), and afew granules of even higher S concentrations (Fig. 10d [Sgr]). Thephosphorus image shown in Fig. 10c shows a rather different pattern:the biomineral structures (skeleton and sclerites) do not appear on

the image. On the other hand, organized zones of high concentrationsof P are observed in the organic tissues. The highest concentrations ofP coincide with zones of high concentrations of S, Ca and Mg in theorganic tissues and seem to outline the margins of the gastrodermalcanals.

4. Discussion

4.1. The structural position of S in the red coral skeleton

Although sulphur has been known for a long time as a common el-ement in marine and biogenic calcites (Milliman, 1974; Busenbergand Plummer, 1985; Pingitore et al., 1995; Hermans, 2010), its posi-tion in the biomineral structure, i.e. organic vs inorganic, is still mat-ter of discussion. Pingitore et al.(1995) summarized the differentpostulated modes of incorporation of sulphur in carbonates: sulphide(e.g. pyrite), sulphate (e.g. anhydrite) or native sulphur as mineral in-clusions; organo-sulphur in the organic matter; sulfite (SO3

2−) or sul-phate (SO4

2−) substituting to CO32− groups in the carbonate structure;

and sulphate as fluid inclusions. From both analytical and experimen-tal studies, it is now established that sulphur can be part of the calcitestructure as sulphate substituting for carbonates (Takano et al., 1980;Busenberg and Plummer, 1985; Takano, 1985; Pingitore et al., 1995;Kampschulte and Strauss, 2004; Kontrec et al., 2004). Such sulphuris classically referred to as structurally substituted sulphur (SSS)(Kampschulte and Strauss, 2004). The effect of this substitution interms of cell parameters has been discussed by Busenberg andPlummer (1985). On the other hand, Kontrec et al. (2004) proposeda structural model for the position of the sulphate ion in the calcitestructure. Different mechanisms have been proposed to explain theincorporation of sulphur as SSS in calcite. For instance, the introduc-tion of Mg in the calcite may generate lattice distortions which mayfacilitate the incorporation of S in the calcite structure (Kontrec et

Fig. 8. (a to c) Distribution of OM, Mg, and S in a section of red coral (MedesCol5). (a) Distribution of OM revealed by OM staining, dark bands are rich in OM. The white dots showsome characteristic annual growth rings. (b) Distribution of Mg (same section as in a): dark zones are poor in Mg. The white dots show some characteristic annual growth rings.(c) Distribution of S (same area as in a and b): dark zones are poor in S. The black rectangles in a to c indicate the location of the enlarged zone displayed in d to f. The white line in(b) is the location of the EMP analytical cross-section shown in Fig. 13. (d) Enlargement and superposition of the framed areas in a and b after coloring and thresholding; OM- andMg-rich zones in red and blue, respectively. (e) Enlargement and superposition of the framed areas in b and c; Mg- and S-rich zones in blue and green, respectively. (f) Enlargementand superposition of the framed areas in a and c; OM- and S-rich zones in red and green, respectively. (For interpretation of the references to color in this figure legend, the reader isreferred to the web version of this article.)

21D. Vielzeuf et al. / Chemical Geology 355 (2013) 13–27

al., 2004; Cusack et al., 2008). On the other hand, Hermans (2010) no-ticed the importance of ion pairing between sulphates and magne-sium in seawater that could explain the simultaneous incorporationof S and Mg in the calcite structure.

It has also been postulated that sulphur resides in the organic ma-trix of biominerals [e.g. echinoderms (Blake and Peacor, 1981) andother marine skeletal carbonates (Mackenzie et al., 1983)]. For someauthors, sulphur is viewed as an elemental indicator of the organic

matrix (Dauphin and Cuif, 1999; Cuif et al., 2003; Dauphin et al.,2003, 2005). The presence of S in the organic tissues of the red coralas shown above (Figs. 9c and 10d) is not surprising since many mac-romolecules like glycoproteins and glycosaminoglycanes which arecharacteristic of extracellular matrices in marine organisms containthis element [e.g. scleractinian corals (Tambutté et al., 2007),octocorals (Dauphin, 2006), hexacorals (Cuif et al., 2003), mollusks(Marin et al., 2008)]. Concerning the S images of the organic tissues

Fig. 9. EMP chemical maps of calcium (a), magnesium (b), sulphur (c), and phosphorus(d) of a portion of red coral skeleton surrounded by its organic tissues. [Sc]: sclerites,[Oe]: organic envelope, [Sk]: skeleton, [Gc]: gastrodermal canal, [Cagr]: calcium granule,[Ep]: epoxy, [Sgr]: sulphur granule, [Ez]: enriched strontium zone. (Sample L05).

Fig. 10. Calcium (a), magnesium (b), phosphorus (c), and sulphur (d) EMP images of a poorganic envelope, [Sk]: skeleton, [Gc]: gastrodermal canal, [Ep]: epoxy. Z1 and Z2 indicate

22 D. Vielzeuf et al. / Chemical Geology 355 (2013) 13–27

surrounding the skeleton (Fig. 10d), it should be noted that there isalmost no contrast between the organic tissues (zone 2 [Z2]), onone side, and the axial skeleton or sclerites, on the other side. Thisshould not be the case if S was only present in the organic matrix be-cause the mean content of OM in the skeleton is low [about 1.5 wt.%(Allemand et al., 1994; Cuif et al., 2011)]. On the other hand, the~3000 ppm of sulphur reported in the skeleton by various methodscan hardly be accounted for by the organic matrix alone consideringthat (i) the proportion of OM is low, (ii) the richest organo-sulphurcompounds such as cysteine or methionine contain less than 26% byweight sulphur, and (iii) in the red coral, sulphated polysaccharides(e.g. chondroitin that contain about 7% by weigh sulphur) arepredominant (~80%) organo-sulphur components while proteinsrepresent less than 20% (Allemand et al., 1994; Cuif et al., 2011).These considerations suggest that in addition to being part of theOM, S must be present in calcite as structurally substituted sulphur.

In the red coral skeleton, an anticorrelation between magnesiumand sulphur has been noted in this paper. This was unexpectedsince, to our knowledge, all previous studies pointed out a positivecorrelation between these two elements as in modern brachiopods(England et al., 2007; Cusack et al., 2008), bivalves (Rosenberg andHughes, 1991; Wisshak et al., 2009), foraminifera (Erez, 2003), andechinoderms and hypercalcified sponges (Hermans, 2010). An analyt-ical problem can be ruled out since this anticorrelation was systemat-ically observed in the different EMP analytical sessions involvingdifferent samples and since the results are confirmed by LA-ICPMSmeasurements (Fig. 4). This anticorrelation may come from the factthat S is present in the organic matrix of the red coral and that theconcentration of OM within the skeleton is anticorrelated with themagnesium distribution (e.g. Fig. 8d).

The possible presence of SSS raises the question whether anothertype of correlation between S (as SSS) and Mg may exist in the calcitestructure itself? At this stage of our study, we are unable to answerthis question. However a first indication is provided by preliminaryresults on 2 million-year old red coral fossils that lost their organicmatter (Vertino et al., 2010). Contrary to their present-day analogs,these samples display a positive correlation between S and Mg

rtion of the organic envelope surrounding the red coral skeleton. [Sc]: sclerites, [Oe]:areas with contrasted sulphur concentrations as discussed in the text. (Sample L05).

2004 2005 2006 2007 2008

12

14

16

18

20

22

24

0 10 20 30 40 50 60

Month (from Jan 2004 to Dec 2008)

Sea

Wat

er T

empe

ratu

re (

°C)

10m

20m

30m

40m

Fig. 11. Sea water temperature in the Mediterranean measured at various depths (Riou Island) during the period January 2004–December 2008.

Table 3Mean SWT in the Mediterranean as a function of depth (Riou Island, five year periodfrom Jan 2004 to Dec 2008).

Annual 3 coldestmonths

3 hottestmonths

Depth (m) T°C s.d. T°C s.d. T°C s.d.

5 16.8 1.73 13.5 0.45 20.0 2.4310 16.5 1.71 13.3 0.41 19.7 2.2515 16.4 1.64 13.3 0.36 19.4 2.2420 16.2 1.56 13.3 0.35 19.3 2.1525 15.8 1.66 13.4 0.36 18.7 2.3330 15.6 1.33 13.2 0.26 18.5 1.9935 15.3 1.28 12.9 0.22 18.2 2.0640 15.2 1.19 13.1 0.23 17.7 1.98

23D. Vielzeuf et al. / Chemical Geology 355 (2013) 13–27

(Fig. A2, online supplementary materials). Thus, we can reasonablyconsider the possibility that a positive correlation between SSS andMg might also exist in the calcite of present-day red coral, and thatit could be hidden by an overall anticorrelation of Mg and S due tothe presence of S in the OM and the anticorrelation of Mg and OM.Further work is required to ascertain this hypothesis.

Different studies concluded that S is present in both the OM andthe calcite itself (Cusack et al., 2008; Hermans, 2010). Our resultsare in agreement with this hypothesis. The presence of S in two dis-tinct environments (organic and inorganic) has important implica-tions. For instance, S might not be such a good and unambiguousindicator of S-bearing organic molecules in biominerals (Cusack etal., 2008). The situation could be different for phosphorus. Indeed,Fig. 10c shows that less phosphorus is detected in the inorganic struc-tures than in the organic tissues (compare to Fig. 10d for S). Thisobservation may indicate that phosphorus is mostly present in theorganic matrix. This is not surprising since phosphorylation is acharacteristic feature of many proteins associated with biominerals(Borbas et al., 1991; Gericke et al., 2005; Liu and Franz, 2007). Thus,as noted previously by Cusack et al. (2008), phosphorus could be abetter candidate than S as a marker of organic matrix in biominerals.

4.2. Magnesium and strontium as potential proxies of SWT

In their study on the potential use of magnesium and strontium asecological indicators in the red coral, Weinbauer et al.(2000) statedthat the Mg/Ca and Sr/Ca ratios decrease with decreasing tempera-ture which itself decreases with depth, and thus could be used asthermometer and growth indicator, respectively. A temperaturesensitivity of 0.005 ± 0.001 mmol/mol Mg/Ca per degree C hasbeen proposed. More recently, a positive correlation of Mg/Ca withSWT has been inferred for various deep-sea corals and red coral(Yoshimura et al., 2011). The regression line obtained by Yoshimuraet al. is similar to the one reported by Weinbauer et al. (2000). Weexamine below if our samples and measurements support theseconclusions.

4.2.1. Time series of sea water temperature (Riou and Medes Islands)The variation of SWT over time in the Riou and Medes areas

(Fig. 1) has been analyzed by Bensoussan et al. (2010). These resultswill not be discussed in detail here and only the aspects relevant tothe evaluation of the red coral as a potential proxy of the SWT willbe considered. The SWT data collected at Riou during the 5-yearperiod Jan 2004–Dec 2008 are taken as representative of the SWT re-gime in this area andwill be considered. In theWesternMediterranean,the dominant forcings are the strong orographic winds, the sea water

circulation along the coast, and fresh water inputs from local rivers. Inaddition, the Riou Island is located close to a major zone of coastal up-welling. These conditions favor high variances of temperature duringthe summer over the entire water column [(Bensoussan et al., 2010)and refs therein]. During 5 to 6 months of the year, from November toMarch (or April) the difference in SWT over the entire water columnis less than 0.5 °C (Fig. 11) whichmeans that the corals collected at dif-ferent depths live under similar SWT conditions for long periods of timeduring the year. In contrast, temperature differences in the water col-umn reach a maximum in July or August with a difference of about5 °C from 5 to 40 m depths. The difference in monthly mean tempera-tures can reach 8 °C during anomalously hot years such as 2006(Fig. 11). Thus, the thermal SWT gradient is steeper during the stratifi-cation period (i.e. end spring, summer and early fall) than during thehomogenisation periods (late fall, winter and early spring). The annualmean SWT temperature for the 5 year period and the mean SWT tem-perature of the three hottest and coldest months are reported inTable 3 and plotted in Fig. 12. Annual mean SWTs at 5 m and at 40 mdiffer by only 2 °C. It should also be noted that the standard deviationof the temperature at all depths is larger during the hot months thanduring the cold months. Finally, the standard deviation progressivelydecreases with depth. The mean SWT temperature calculated at eachdepth in thewater columnwill be used to find a possible correlation be-tween chemistry of the corals and the SWT.

A second approach to explore the potential of the red coral as aproxy for seawater temperature consists in studying a single old col-ony from a location where the SWT has been monitored during itsgrowth. For this purpose, a ~55 year old colony collected in theMedes Island in January 2002 has been studied. The temperaturemeasurements at Medes (20 m depth) during the period Dec1974–

5

10

15

20

25

30

35

40

Depth (m)

12 14 16 18 20 22

Temperature (°C)

Three hottest months

Three coldest months

Mean annual T

Fig. 12.Mean SWT in the Mediterranean as a function of depth (Riou Island, 5-year pe-riod from Jan 2004 to Dec 2008). Blue: mean annual temperature, Yellow: 3 hottestmonths, Pink: 3 coldest months. (For interpretation of the references to color in thisfigure legend, the reader is referred to the web version of this article.)

MgC

O3

(%)

SW

T (

°C)

11

13

15

17

19

21

23

0 24 48 72 96 120 144 1

2000 1995 1990

8

9

10

11

12

13

14

15

0 500

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17

Rim

Dista

Time (

Growth ring c

a

b

Fig. 13. (a) Variations of MgCO3 along a traverse in a red coral skeleton (Medes5) collectethe traverse is shown in Fig. 8). The number of growth rings has been determined using a m(b) SWT temperature recorded at 20 m depth in the Medes Islands from January 2002 to D

24 D. Vielzeuf et al. / Chemical Geology 355 (2013) 13–27

Jan2002 (Salat and Pascual, 2002; Calvo et al., 2011) are shown inFig. 13b. This temperature series shows a significant increase ofSWT of more than 1 °C during the last 30 years. Indeed, previousstudies of this time series indicate that the warming trend of thenorthwestern Mediterranean during the 1974–2001 period is about0.040 °C/year for surface waters and 0.025 °C/year at 80 m depth(Salat and Pascual, 2002; Calvo et al., 2011). Whether this dramaticincrease of SWT is registered within the red coral skeleton isdiscussed below.

4.2.2. The Mg/Ca and Sr/Ca ratios versus temperatureThe variation of the Mg/Ca ratio (bulk analyses by ID-ICPMS) as a

function of depth and mean SWT is shown in Fig. 14. The black dotscorrespond to the mean Mg/Ca values with an error bar taking intoaccount the dispersion of the data. It should be noted that the Mg/Cavariability within samples collected at the same site and depth islarge. On the other hand, the difference in mean Mg/Ca values for shal-low and deep samples is small. Thus, even if the difference between thehigh and low Mg/Ca ratios above and below 35 m depth is statisticallysignificant, no simple and regular correlation between Mg/Ca ratioand depth/temperature is observed. Thus, the Mg/Ca ratios from bulksamples of red corals do not seem to be accurate proxies of the meanSWT for locations characterized by large seasonal temperaturevariabilities.

The variation of the Sr/Ca ratio as a function of depth is shown inFig. 15. A plateau with relatively high values of Sr/Ca is observedabove 33 m depth and is followed by a plateau with relatively lowvalues of the ratio for depths in the range 33 to 73 m. However, sim-ilar to Mg/Ca, the Sr/Ca variability within different samples from thesame depth is important and the difference between shallow anddeep samples is small. Thus, the Sr/Ca ratio cannot be easily used asa proxy of the mean SWT for locations characterized by large seasonaltemperature variability, either.

from Jan 2002 (left) to Jan 1974

68 192 216 240 264 288 312 336

1985 1980 1975

1500 2000

//

18 19 20 21 22 23 24 25 26 27 28

Core

nce (µm)

month)

ount (year)

d in the Medes Islands in January 2002 (6 μm spacing between points, the location ofosaic of SEM images along the same traverse (Fig. A3, online supplementary materials).ecember 1974.

120

125

130

135

0 10 20 30 40 50 60 70 80

Depth (m)

Mg/

Ca

16.5 16 15.5 15 13.5°C1414.5

y = -0.1152x + 130.3R = 0.2222

Fig. 14. Mg/Ca ratios as a function of depth determined by ID-ICPMS in red coral skel-etons (see Table 1). The temperature scale corresponds to the mean SWT determinedat each depth (Fig. 12). Yellow squares: individual analyses; large black dots: meanvalues with error bars taking into account the dispersion of the data. (For interpreta-tion of the references to color in this figure legend, the reader is referred to the webversion of this article.)

25D. Vielzeuf et al. / Chemical Geology 355 (2013) 13–27

4.2.3. Response of a single colony to external forcingsIt has been shown earlier in this article that growth rings in the

red coral skeleton are marked by regular patterns in Ca, Mg, S, andOM (Figs. 7 and 8). The OM growth rings periodicity evidenced bythe staining method (e.g. Fig. 2) has been time calibrated byMarschal et al. (2004) using adult colonies of known age (20 to22 years) and in situ labeling of colonies with calcein followed bysampling one year later. Both methods demonstrated the annual pe-riodicity of the OM rings. Marschal et al. (2004) conducted a secondseries of experiments to determine the period of the year duringwhich growth bands develop in the skeleton. Twelve colonies collect-ed at 15 m depth were labeled with calcein three times at about 4month intervals in November, March, and July. Nine colonies were re-covered in December about a year after the first labeling. These colo-nies showed that late fall and winter is the period during which thenarrow band with a high concentration of organic matter develops.Marschal et al. (2004) also concluded that late fall and winter is theperiod with the lowest growth rates in red coral. From these observa-tions, it can be concluded that Mg-rich rings develop during the peri-od spring to fall while S-rich rings form immediately after (late falland winter). Thus, OM and Mg are indicators of annual cycles in the

16.5 16 15.5 15 13.5°C1414.5

3,1

3,2

3,3

3,4

3,5

3,6

0 10 20 30 40 50 60 70 80

Depth (m)

Sr/

Ca

y = -0.0024x + 3.380R = 0.2022

Fig. 15. Sr/Ca ratios as a function of depth determined by ID-ICPMS in red coral skele-tons (see Table 1). The temperature scale corresponds to the mean SWT determined ateach depth (Fig. 12). Blue squares: individual analyses; large black dots: mean valueswith error bars taking into account the dispersion of the data. (For interpretation ofthe references to color in this figure legend, the reader is referred to the web versionof this article.)

red coral. However, is it possible to convert a qualitative recordingof annual cycles into a quantitative recording of the SWT?

Fig. 13a shows the MgCO3 content determined by EMP along a tra-verse in a sample collected in January 2002 in the Medes Islands (seealso Fig. 8b). The variations of MgCO3 contents in Fig. 13a weresmoothed using a moving average filter of 5 data points. The fittedcurve shows oscillations that can be ascribed to annual growthrings. However, the identification of growth rings is more reliableon 2D images than along a 1D traverse (Marschal et al., 2004;Vielzeuf et al., 2008). Thus, a continuous series of SEM images(backscattered electrons) collected at high spatial resolution alongdifferent traverses (Fig. A3, online supplementary materials) wasused to identify and count the growth rings as reported in Fig. 13a.The fact that this count and the Mg peaks do not perfectly matchcan be ascribed to the difficulty in identifying the rings along a 1D tra-verse. Nevertheless, even if the match is not perfect, the number ofmajor peaks and growth rings in both patterns is identical acrossthe traverse (within an estimated error of ±10% in the determinationof growth rings). Due to space constraints and for purpose of clarity,only part of the analytical traverse performed on this sample(Medes5) is shown in Fig. 13a. The complete traverse is 5210 μm longfor 51 identified growth rings. This represents a diametral growth rateof about 200 ± 20 μm/year, consistent with the previous estimatesprovided by Garrabou and Harmelin (2002): 240 ± 50 μm year−1

and Torrents (2007):mean value:145 ± 50 μm year−1; minimum andmaximum values: 90 and 270 μm year−1. Thus, our study confirmsthat the radial growth of red coral is slow (Garrabou and Harmelin,2002; Marschal et al., 2004; Torrents, 2007). The portion correspondingto the 28 rings grown during the 28 years of SWTmonitoring is report-ed in Fig. 13a. The EMP analytical section (Fig. 13a) shows that thewidth and the amplitude of the growth rings vary in space and time:sometimes the growths rings are wide and well-marked while inother cases they are difficult to distinguish. Such differences in chemicaloscillations should not show up if compositional oscillations were sim-ply linked to SWTannual variations. Indeed, a reliable proxy is implicitlybased on the assumption that a consistent periodic forcing generates aconsistent periodic pattern. The fact that this is not the case for thered coral can be due to various causes. First of all, the labeling experi-ments carried out by Marschal et al. (2004) indicate variations ofgrowth rate during the year, thus an irregular growth. Second, structur-al discontinuities within the rings of the skeleton observed by Vielzeufet al.(2008) are interpreted as interruptions of growth. Finally, experi-ments carried out in aquaria to test the thermotolerance of C. rubrumshowed that calcification rates decrease dramatically above 21 °C andthat exposure to 25 °C for durations between 9 and 14 days causemor-tality of most samples (Torrents et al., 2008). These facts related to thegrowth dynamics of the red coral may explain the irregular chemicaloscillations observed in the red coral.

It has been mentioned earlier that the Mg-rich bands develop dur-ing the period comprised between spring and early fall. The dynamicsof development within this period is still unknown and new labelingexperiments with higher temporal resolution would be required tobetter characterize the growth dynamics of the red coral. Neverthe-less, growth occurs between the period when temperatures start towarm up (spring) and reach the warmest values (early fall). Theseobservations are consistent with the experimental studies showingan increase in magnesium content of synthetic calcite with respectto temperature (Mucci, 1987; Morse et al., 2007), and also the studiesof high-Mg calcite secreted by marine organisms indicating that theirMg content increases with temperature (Chave, 1954; Ries, 2010).Consequently, in the case of the Medes red coral section displayedin Fig. 13, we would expect an increase in the Mg content from thecenter to the rim since a clear warming trend in SWT is recorded(~1 °C over 30 years). The opposite is observed indicating that thered coral did not register the change of temperature, small in absolutevalue but dramatic in terms of environmental change.

26 D. Vielzeuf et al. / Chemical Geology 355 (2013) 13–27

The observed slight decrease in Mg content as a function of time(Fig. 13a) could be the result of an abnormal decrease of growthrate of the coral. From the analysis of the distance between rings(Fig. 8), it can be seen that the growth rate has been lower for thelast 10 years than for previous periods. This reduction of growthrate mainly affects the width of Mg-rich bands (thus the overall Mgcontent of the annual ring). This decrease could be related to the in-creasing prevalence of summer conditions associated to the observedwarming trend during the last decades in the Medes Islands. Indeed,summer conditions characterized by reduced hydrodynamics, ther-mal stratification and low food availability affect negatively thegrowth of several Mediterranean suspension feeder species such asred coral (Coma and Ribes, 2003; Torrents et al., 2008; Coma et al.,2009). Besides and as mentioned earlier, the fact that calcificationrates in the red coral decrease dramatically during exposure above21 °C (Torrents et al., 2008) supports the hypothesis of the reductionof growth rates during summer months when temperatures reachand exceed this value. Therefore, we conclude that more than a pre-cise indicator of SWT, magnesium in the red coral is a marker of thebiological activity of the organism, as magnesium in the aragonite ofsome scleractinian corals (Brahmi et al., 2012). Thus, the concentra-tion and 2D distribution of magnesium in the red coral might be anefficient way to detect past prolongated periods of ‘summer sufferingevents’ endured by the organism, together with their frequency.

Supplementary data to this article can be found online at http://dx.doi.org/10.1016/j.chemgeo.2013.07.008.

Acknowledgments

Thiswork has been supported by the Centre National de la RechercheScientifique (CNRS) – Institut National des Sciences de l'Univers (INSU)through grant ECLIPSE 2005, INTERRVIE 2009, by the Agence Nationalepour la Recherche through ANR CoRo 2011–2015, and by the CentreInterdisciplinaire de Nanoscience deMarseille (CINaM) through internalgrants. Thiswork is also part of the EuropeanUnion COST action TD0903.D.V. also benefited from a financial support by E.M. Stolper for a threemonth stay at Caltech in 2010.We thank H. Zibrowius for providing gen-erously the samples of fossil red coral, C. Marschal and P. Raffin for sup-plying some of the present-day red coral colonies, J. Pascual and J. Salatfor allowing access to the long-term temperature data series from theMedes Islands (Spain), and N. Bensoussan for his help in the processingof the temperature series from the Riou Island (Marseille, France). Re-views by Andres Rüggeberg and an anonymous reviewer as well as edi-torial handling by Uwe Brand are gratefully acknowledged. This iscontribution ANR CoRo n° 03.

References

Allemand, D., Cuif, J.P., Watabe, N., Oishi, M., Kawagushi, T., 1994. The organic matrix ofskeletal structures of the Mediterranean red coral, Corallium rubrum. Bulletin del'Institut Océanographique de Monaco 14, 129–139.

Bensoussan, N., Romano, J.C., Harmelin, J.G., Garrabou, J., 2010. High resolution charac-terization of northwest Mediterranean coastal waters thermal regimes: to betterunderstand responses of benthic communities to climate change. Estuarine, Coastaland Shelf Science 87 (3), 431–441.

Blake, D.F., Peacor, D.R., 1981. Biomineralization in crinoid echinoderms: characteriza-tion of crinoid skeletal elements using TEM and STEM microanalysis. ScanningElectron Microscopy 224, 321–328.

Borbas, J.E., Wheeler, A.P., Sikes, C.S., 1991. Molluscan shell matrix phosphoproteins:correlation of degree of phosphorylation to shell mineral microstructure and toin vitro regulation of mineralization. The Journal of Experimental Zoology 258 (1),1–13.

Brahmi, C., Kopp, C., Domart-Coulon, I., Stolarski, J., Meibom, A., 2012. Skeletal growthdynamics linked to trace-element composition in the scleractinian coralPocillopora damicornis. Geochimica et Cosmochimica Acta 99, 146–158.

Busenberg, E., Plummer, L.N., 1985. Kinetic and thermodynamic factors controlling thedistribution of SO4

2− and Na+ in calcites and selected aragonites. Geochimica EtCosmochimica Acta 49 (3), 713–725.

Bussoletti, E., Cottingham, D., Bruckner, A., Roberts, G., Sandulli, R., 2010. Proceedingsof the International Workshop on Red Coral Science, Management, and Trade:

Lessons from the Mediterranean. NOAA technical memorandum CRCP-13, SilverSpring, MD 233 (pp.).

Calvo, E., et al., 2011. Effects of climate change on Mediterranean marine ecosystems:the case of the Catalan Sea. Climate Research 50 (1), 1–29.

Chave, K.E., 1954. Aspects of the biogeochemistry of magnesium. 1. Calcareous marineorganisms. Journal of Geology 62, 266–283.

Coma, R., Ribes, M., 2003. Seasonal energetic constraints in Mediterranean benthic sus-pension feeders: effects at different levels of ecological organization. Oikos 101 (1),205–215.

Coma, R., et al., 2009. Global warming-enhanced stratification and mass mortalityevents in the Mediterranean. Proceedings of the National Academy of Sciences ofthe United States of America 106 (15), 6176–6181.

Cuif, J.P., Dauphin, Y., Doucet, J., Salome, M., Susini, J., 2003. XANES mapping of organicsulfate in three scleractinian coral skeletons. Geochimica Et Cosmochimica Acta 67(1), 75–83.

Cuif, J.P., Dauphin, Y., Sorauf, J.E., 2011. Biominerals and Fossils through Time. Cambridgeuniversity press, Cambridge (490 pp.).

Cusack, M., et al., 2008. Micro-XANESmapping of sulphur and its associationwithmagne-sium and phosphorus in the shell of the brachiopod, Terebratulina retusa. ChemicalGeology 253 (3–4), 172–179.

Dauphin, Y., 2006. Mineralizing matrices in the skeletal axes of two Corallium species(Alcyonacea). Comparative Biochemistry and Physiology. Part A, Molecular & Inte-grative Physiology 145 (1), 54–64.

Dauphin, Y., Cuif, J.P., 1999. Relationship between mineralogy and microstructural pat-terns of calcareous biominerals and their sulfur contents. Annales Des SciencesNaturelles-Zoologie Et Biologie Animale 20 (2), 73–85.

Dauphin, Y., et al., 2003. In situ chemical speciation of sulfur in calcitic biominerals andthe simple prism concept. Journal of Structural Biology 142 (2), 272–280.

Dauphin, Y., Cuif, J.P., Salome, C., Susini, J., 2005. Speciation and distribution of sulfur ina mollusk shell as revealed by in situ maps using X-ray absorption near-edge struc-ture (XANES) spectroscopy at the SK-edge. American Mineralogist 90 (11–12),1748–1758.

Debreuil, J., et al., 2011. Specific organic matrix characteristics in skeletons of Coralliumspecies. Marine Biology 158 (12), 2765–2774.

England, J., Cusack, M., Lee, M.R., 2007. Magnesium and sulphur in the calcite shells oftwo brachiopods, Terebratulina retusa and Novocrania anomala. Lethaia 40 (1),2–10.

Erez, J., 2003. The source of ions for biomineralization in foraminifera and their impli-cations for paleoceanographic proxies, biomineralization. Reviews in Mineralogyand Geochemistry 115–149.

Fernandez, D.P., Gagnon, A.C., Adkins, J.F., 2011. An isotope dilution ICP-MS method forthe determination of Mg/Ca and Sr/Ca ratios in calcium carbonate. Geostandardsand Geoanalytical Research 35 (1), 23–37.

Floquet, N., Vielzeuf, D., 2011. Mesoscale twinning and crystallographic registers inbiominerals. American Mineralogist 96 (8–9), 1228–1237.

Garrabou, J., Harmelin, J.G., 2002. A 20-year study on life-history traits of a harvestedlong-lived temperate coral in the NW Mediterranean: insights into conservationand management needs. Journal of Animal Ecology 71 (6), 966–978.

Gericke, A., et al., 2005. Importance of phosphorylation for osteopontin regulation ofbiomineralization. Calcified Tissue International 77 (1), 45–54.

Grillo, M.C., Goldberg, W.M., Allemand, D., 1993. Skeleton and sclerite formation in theprecious red coral Corallium rubrum. Marine Biology 117 (1), 119–128.

Guillong, M., Latkoczy, C., Seo, J.H., Günther, D., Heinrich, C.A., 2008. Determination ofsulfur in fluid inclusions by laser ablation ICP-MS. Journal of Analytical AtomicSpectrometry 23 (12), 1581–1589.

Günther, D., Frischknecht, R., Heinrich, C.A., Kahlert, H.J., 1997. Capabilities of an argonfluoride 193 nm excimer laser for laser ablation inductively coupled plasma massspectrometry microanalysis of geological materials. Journal of Analytical AtomicSpectrometry 12 (9), 939–944.

Harmelin, J., 2000. Le corail rouge de Méditerranée: quelques aspects de sa biologie etde son écologie. In: Morel JP, R.-C.C., Ugolini, D. (Eds.), Corallo di Ieri, Corallo diOggi. Centro Universario Europeo per i Beni Culturali, Ravello. Travaux CentreCamille Jullian, Université de Provence, 25. Edipuglia, Bari., Ravello, Villa Rufolo.

Hasegawa, H., Iwasaki, N., Suzuki, A., Maki, T., Hayakawa, S., 2010. Distributions of traceelements in biogenic carbonate minerals of precious corals by X-ray fluorescenceanalysis. Bunseki Kagaku 59 (6), 521–530.

Hermans, J., 2010. Incorporation du magnésium dans les squelettes calcitiquesdes échinodermes et des éponges hypercalcifiées. Université Libre de Bruxelles(141 pp.).

Jochum, K.P., et al., 2011. Determination of reference values for NIST SRM 610-617glasses following ISO guidelines. Geostand. Geoanal. Res. 35 (4), 397–429.

Kampschulte, A., Strauss, H., 2004. The sulfur isotopic evolution of Phanerozoic seawa-ter based on the analysis of structurally substituted sulfate in carbonates. ChemicalGeology 204 (3–4), 255–286.

Klemme, S., Prowatke, S., Hametner, K., Gunther, D., 2005. Partitioning of trace elementsbetween rutile and silicate melts: implications for subduction zones. Geochimica EtCosmochimica Acta 69 (9), 2361–2371.

Kontrec, J., et al., 2004. Incorporation of inorganic anions in calcite. European Journal ofInorganic Chemistry 23, 4579–4585.

Lacaze-Duthiers, H., 1864. Histoire naturelle du corail, Paris (371 pp.).Liu, L.L., Franz, K.J., 2007. Phosphorylation-dependent metal binding by alpha-synuclein

peptide fragments. Journal of Biological Inorganic Chemistry 12 (2), 234–247.Lorens, R.B., Bender, M.L., 1980. The impact of solution chemistry on Mytilus edulis

calcite and aragonite. Geochimica Et Cosmochimica Acta 44 (9), 1265–1278.Lowenstam, H.A., Weiner, S. (Eds.), 1989. On Biomineralization. Oxford University

Press, Oxford.

27D. Vielzeuf et al. / Chemical Geology 355 (2013) 13–27

Mackenzie, F.T., et al., 1983. Magnesian calcites: low-temperature occurrence, solubil-ity, and solid-solution behaviour. Reviews in Mineralogy 11, 97–144.

Mann, S., 2001. Biomineralization: Principles and Concepts in Bioinorganic MaterialsChemistry. Oxford University Press, New York.

Marin, F., Luquet, G., Marie, B., Medakovic, D., 2008. Molluscan shell proteins: primarystructure, origin, and evolution, current topics in developmental biology, vol 80.Current Topics in Developmental Biology 209–276.

Marschal, C., Garrabou, J., Harmelin, J.G., Pichon, M., 2004. A newmethod for measuringgrowth and age in the precious red coral Corallium rubrum (L.). Coral Reefs 23 (3),423–432.

Maté, P., Revenge, S., Masso, C., 1986. Estudio preliminar de la composicion quimica delcoral rojo (Corallium rubrum L.) de distintas zonas del Mediterraneo espanol.Boletin del Instituto Espanol de Oceanografia 3 (4), 53–60.

Milliman, J.D., 1974. Marine Carbonates. Springer-Verlag, New-York, Heidelberg, Berlin(375 pp.).

Morse, J.W., Arvidson, R.S., Luttge, A., 2007. Calcium carbonate formation and dissolu-tion. Chemistry Review 107 (2), 342–381.

Mucci, A., 1987. Influence of temperature on the composition of magnesian calciteovergrowths precipitated from seawater. Geochimica Et Cosmochimica Acta 51(7), 1977–1984.

Pascual, J., Salat, J., Palau, M., 1995. Evolution de la température de la mer entre 1973 et1994, près de la côte catalane. OKEANOS 95: la Méditerranée, variabilitésclimatiques, environnement et biodiversité, Héliopolis, Montpellier, France.

Pingitore, N.E., Meitzner, G., Love, K.M., 1995. Identification of sulfate in natural car-bonates by X-ray-absorption spectroscopy. Geochimica Et Cosmochimica Acta 59(12), 2477–2483.

Ries, J.B., 2010. Review: geological and experimental evidence for secular variation inseawater Mg/Ca (calcite-aragonite seas) and its effects on marine biologicalcalcification. Biogeosciences 7 (9), 2795–2849.

Rosenberg, G.D., Hughes, W.W., 1991. A metabolic model for the determination of shellcomposition in the bivalve mollusc, Mytilus edulis. Lethaia 24 (1), 83–96.

Salat, J., Pascual, J., 2002. The Oceanographic and Meteorological Station at l'ESTARTIT(NWMediterranean). Tracking Long-termHydrological Change in theMediterraneanSea. Briand F, CIESM Monaco 29–32.

Takano, B., 1985. Geochemical implications of sulfate in sedimentary carbonates.Chemical Geology 49 (4), 393–403.

Takano, B., Asano, Y., Watanuki, K., 1980. Characterization of sulfate ion in travertine.Contributions to Mineralogy and Petrology 72 (2), 197–203.

Tambutté, S., Tambutté, E., Zoccola, D., Allemand, D., et al., 2007. Organic matrix andbiomineralization of scleractinian corals. Handbook on biomineralization. Thebiology of Biominerals Structure Formation, vol. 1. Wiley-VCH, pp. 243–259.

Torrents, O., 2007. Biologie des populations du corail rouge Corallium rubrum (L. 1758)de Méditerranée Nord-Occidentale. Université de la Méditerranée, Marseille(442 pp.).

Torrents, O., Tambutte, E., Caminiti, N., Garrabou, J., 2008. Upper thermal thresholds ofshallow vs. deep populations of the precious Mediterranean red coral Coralliumrubrum (L.): assessing the potential effects of warming in the NW Mediterranean.Journal of Experimental Marine Biology and Ecology 357 (1), 7–19.

Vander Putten, E., Dehairs, F., Keppens, E., Baeyens, W., 2000. High resolution dis-tribution of trace elements in the calcite shell layer of modern Mytilus edulis:environmental and biological controls. Geochimica Et Cosmochimica Acta 64 (6),997–1011.

Vertino, A., Zibrowius, H., Rocca,M., Taviani,M., 2010. Fossil coralliidae in theMediterraneanbasin. In: Bussoletti, E., Cottingham, D., Bruckner, A., Roberts, G., Sandulli, R. (Eds.), RedCoral Science, Management, and Trade: Lessons from theMediterranean. NOAA techni-cal memoradum CRCP-13, Naples, Italy, pp. 94–99.

Vielzeuf, D., Garrabou, J., Baronnet, A., Grauby, O., Marschal, C., 2008. Nano to macro-scale biomineral architecture of red coral (Corallium rubrum). American Mineralo-gist 93 (11–12), 1799–1815.

Vielzeuf, D., et al., 2010. Multilevel modular mesocrystalline organization in red coral.American Mineralogist 95 (2–3), 242–248.

Weinbauer, M.G., Brandstatter, F., Velimirov, B., 2000. On the potential use of magne-sium and strontium concentrations as ecological indicators in the calcite skeletonof the red coral (Corallium rubrum). Marine Biology 137 (5–6), 801–809.

Weinberg, S., 1976. Revision of the Common Octocorallia of the MediterraneanCircalittoral. 1. Gorgonacea Beaufortia. 24 63–103.

Wisshak, M., et al., 2009. Shell architecture, element composition, and stable isotope sig-nature of the giant deep-sea oysterNeopycnodonte zibrowii sp n. from the NE Atlantic.Deep Sea Research Part I: Oceanographic Research Papers 56 (3), 374–407.

Yoshimura, T., et al., 2011. Mg isotope fractionation in biogenic carbonates of deep-seacoral, benthic foraminifera, and hermatypic coral. Analytical and BioanalyticalChemistry 401 (9), 2755–2769.