Sr/Ca and Mg/Ca vital effects correlated with skeletal architecture ina scleractinian deep-sea coral and the role of
Rayleigh fractionation
Alexander C. Gagnon a,⁎, Jess F. Adkins b, Diego P. Fernandez b,1, Laura F. Robinson c
a Division of Chemistry, California Institute of Technology, MC 114-96, Pasadena, CA 91125, USAb Division of Geological and Planetary Sciences, California Institute of Technology, MC 100-23, Pasadena, CA 91125, USAc Department of Marine Chemistry and Geochemistry, Woods Hole Oceanographic Institution, Woods Hole, MA 02543, USA
Received 29 November 2006; received in revised form 15 May 2007; accepted 3 July 2007
[email protected] (L.F. Robinson).1 Present Address: Department of Geology and Geophysics, University of Utah, 135 S. 1460E, Room 719, Salt Lake City, UT 84112, USA.
Please cite this article as: Gagnon, A.C. et al. Sr/Ca and Mg/Ca vital effects correlated with skeletal architecture in a scleractinian deep-sea coraland the role of Rayleigh fractionation. Earth Planet. Sci. Lett. (2007), doi:10.1016/j.epsl.2007.07.013
2 A.C. Gagnon et al. / Earth and Planetary Science Letters xx (2007) xxx–xxx
ARTICLE IN PRESS
1. Introduction
Deep-sea corals are a new tool in paleoceanographywith several promising characteristics. Their opticallybanded, aragonitic, and unbioturbated skeletons providecentury long records of deep ocean change with thepotential for sub-decadal resolution. High concentrationsof uranium allow for accurate independent calendar ageswith ±1%, or better, precision (Cheng et al., 2000).Together these features of deep-sea corals provide for anew archive in paleoclimate with the potential for icecore-like resolution in the deep ocean. In an attempt toutilize this new archive, both Sr/Ca and Mg/Catemperature proxies, long applied to surface corals(Smith et al., 1979; Beck et al., 1992; Mitsuguchi et al.,1996), have been recently investigated in deep-sea corals(Shirai et al., 2005; Cohen et al., 2006). However, a directtemperature relationship in corals from both surface anddeep environments is complicated by the overprint ofphysiological processes during biomineralization, termedvital effects (Weber and Woodhead, 1972).
Most thoroughly investigated in surface corals, Me/Ca vital effects are evidenced by: (1) large differencesbetween temperature calibrations for different speciesand even between individuals within the same species(Correge, 2006); (2) differences in temperature sensi-tivity between corals and inorganically precipitatedaragonite (Kinsman and Holland, 1969; Mucci et al.,1989; Zhong and Mucci, 1989; Dietzel et al., 2004;Correge, 2006; Gaetani and Cohen, 2006); (3) theapparent growth dependence of Me/Ca temperaturecalibrations (de Villiers et al., 1994) including similarsensitivity to both light and temperature inducedchanges to growth rate (Reynaud et al., 2006); (4)spatial variability of Me/Ca and stable isotope proxiesacross coral skeletal features (Cohen et al., 2001;Allison et al., 2001; Adkins et al., 2003; Allison andFinch, 2004; Shirai et al., 2005; Cohen et al., 2006;Robinson et al., 2006; Meibom et al., 2006a,b); and (5)differences in Me/Ca-temperature calibrations in thepresence or absence of photosymbionts (Cohen et al.,2002). While Me/Ca temperature proxies continue toprovide important information on past climate, quanti-fying and understanding the processes that cause non-temperature related variability in Me/Ca paleotherm-ometers is a major goal of paleoceanography.
The growth environment of the deep-sea is a virtuallyconstant “culture medium” during a coral's lifetime,outside specific regions that experience rapid ventila-tion. In properly chosen modern deep-sea coral samples,the observed variability of a proxy, isolated from aknown and constant environmental background, is
Please cite this article as: Gagnon, A.C. et al. Sr/Ca and Mg/Ca vital effectsand the role of Rayleigh fractionation. Earth Planet. Sci. Lett. (2007), doi
attributable entirely to vital-effects. Thus, deep-seacorals are an ideal model system for the study of vitaleffects. As an additional advantage, deep-sea coralsgrow far from sunlight and lack the photosymbiontstypical of surface corals, enabling calcification to beinvestigated uncoupled from photosynthesis. Similarpatterns of Me/Ca heterogeneity in both surface anddeep-sea corals imply some commonality in vital effectmechanism and suggest results from deep-sea coralsmay be applicable to corals in general (Sinclair et al.,2006). In this study we investigate the magnitude,pattern and mechanism of Sr/Ca and Mg/Ca vital effectsin the scleractinian deep-sea coral Desmophyllumdianthus (formerly D. cristagalli) a globally distributedspecies with a large available fossil coral collection.
Vital effects often correlate with skeletal structuralfeatures, relating spatially resolved geochemical measure-ments to the biologically controlled process of skeletonformation. The skeletons of both surface and deep-seacorals are composed of the same basic architectural units.In transverse petrographic microsections of a deep-seacoral, two different crystal morphologies are clearlyevident by visible light microscopy (Fig. 1). Acicular, orneedle-like, bundles of crystals radiate from regions ofsmall-disorganized granular crystals or Centers ofCalcification, COCs. In this study, the term COC refersto interior regions of the coral skeleton typified by smallirregular or fusiform crystals, following the terminologyof Gladfelter (1982, 1983). InD. dianthusCOCs combineinto a 30–100 μm wide optically dense central band.Crystal morphology is related to the process of skeletalgrowth with the initial precipitation of small fusiformcrystals followed by the extension of needle-like crystals,as hypothesized by Gladfelter (1982) and recentlydocumented in living surface corals (Raz-Bahat et al.,2006), although the relative growth rate of these twocrystal forms is still a topic of research.
There is growing evidence that Mg/Ca variability isstrongly correlated with skeletal architecture. A study ina surface coral shows Mg/Ca doubles in COCs whenmeasured by SIMS with a 30 μm spot (Meibom et al.,2006a). Smaller scale sampling, at ∼0.4 μm using aNanoSIMS instrument, reveals small regions where Mgincreases by an order of magnitude (Meibom et al.,2006b). Sr/Ca heterogeneity also correlates with skeletalstructure in surface corals, with higher Sr/Ca ratiosreported in COCs (Cohen et al., 2001; Allison et al.,2001; Allison and Finch, 2004; Meibom et al., 2006a).In deep-sea corals, Cohen et al. (2006) used SIMS tomeasure Sr/Ca and Mg/Ca across the thecal wall of thespecies Lophelia pertusa, observing increased Mg/Cacoincident with, and extending beyond, most opaque
correlated with skeletal architecture in a scleractinian deep-sea coral:10.1016/j.epsl.2007.07.013
Fig. 1. Coral skeletal architecture. (a) Sampling for petrographic thin sections begins with removal of “pie-shaped” wedges containing at least oneexert septa (S1) from the calice in the transverse plane, and subsequent polishing to ∼100 μm thickness. The transverse plane is perpendicular to thebody axis, roughly cutting the calice into an ellipse. Analysis in this plane gives the view of looking down the body axis toward the oral surface— asseen in figure inset. (b) Transmitted light photomicrograph, in negative, of thin section 47407A. White regions are optically dense with an obviouscentral band in the middle of septa S1. (c) Positive transmitted light view of a single septa, the dark, optically dense central band is surrounded byacicular (ac), or needle like, crystals in the outer septa. (d) Crystal morphology, granular (gr) irregular crystals making up the centers of calcification(COCs), and surrounding acicular crystals, is evident in this etched SEM image of a septum. (e) A 10 s of micrometer thick section viewed in positivecross polarized light. In addition to granular crystals, the central band appears to be composed of many sphere like bodies (sp) stacked together.Acicular crystals (ac) show c-axis alignment.
3A.C. Gagnon et al. / Earth and Planetary Science Letters xx (2007) xxx–xxx
ARTICLE IN PRESS
bands, with Sr/Ca depleted by a smaller relative amountin the same regions. For the deep-sea coral genus Car-yophillia, maps of elemental distribution by electro-nprobe X-Ray microanalysis suggest Mg and Srheterogeneity may correlate with COCs (Shirai et al.,2005). In D. dianthus, the target of this study, previouswork with other proxies show very depleted δ13C andδ18O within the central band (Adkins et al., 2003) and aregion of low uranium content that follows but extendsbeyond the central band (Robinson et al., 2006). The
Please cite this article as: Gagnon, A.C. et al. Sr/Ca and Mg/Ca vital effectsand the role of Rayleigh fractionation. Earth Planet. Sci. Lett. (2007), doi
pattern of tracer heterogeneity suggests skeletal archi-tecture is a key factor in understanding vital effects.
We examine Mg/Ca and Sr/Ca across skeletalfeatures in D. dianthus using micromilling to samplewith lateral spatial resolution of 10s of micrometers.Micromilled samples allow the use of a rapid isotope-dilution-ICP-MS method to determine precise andaccurate metal ratios. We also use electronprobe X-raymicroanalysis to analyze Mg/Ca heterogeneity at finescale. We compare our measurements to inorganic
correlated with skeletal architecture in a scleractinian deep-sea coral:10.1016/j.epsl.2007.07.013
4 A.C. Gagnon et al. / Earth and Planetary Science Letters xx (2007) xxx–xxx
ARTICLE IN PRESS
experiments, to previous measurements in deep-sea andsurface corals, and interpret our results within the con-text of models for vital effect mechanism.
Geochemical modeling of biogenic carbonatesattempts to explain tracer abundance, tracer heterogene-ity, and correlations between tracers using chemical andphysical principles within a biological framework.McConnaughey (1989a,b) explained correlated δ13Cand δ18O results in corals with an elegant model forisotope behavior paving the way for the systematic studyof vital effects. More recently, an equilibrium model ofstable isotope vital effects was proposed to explainobserved departures from a linear correlation betweenδ13C and δ18O in corals (Adkins et al., 2003). Severalgroups have expanded the geochemical approach ofcombined measurements and models to study Me/Cavital effects in biogenic carbonates (Elderfield et al.,1996; Cohen et al., 2001, 2002; Russell et al., 2004;Sinclair, 2005; Bentov and Erez, 2006; Sinclair and Risk,2006). Rather than develop a complete numerical modelof biomineralization to complement our measurements,we attempt to test for the presence of an importantparameter in most vital effect models, the “openness” or“closedness” of the calcifying system to metal ions, andthe implied mechanism of Rayleigh fractionation.
Most models of biomineralization involve a privilegedspace closed to the external environment to some extent,sometimes termed the calcifying fluid or extra-cellularfluid, where a coral isolates seawater and chemicallydrives precipitation. Anytime there is a closed system anda tracer is discriminated against or preferentially incor-porated during coprecipitation, Rayleigh fractionationwill occur. Even without a complete understanding of theunderlying processes that determine this partitioning, aRayleigh process is predicted and should be considered.Rayleigh fractionation has been applied previously toexplain single element Me/Ca behavior in foraminifera(Elderfield et al., 1996), and has been recently incorpo-rated as a key mechanism along with calcifying solutionmanipulation and temperature effects to explain Me/Cavariability in a surface (Gaetani and Cohen, 2006) anddeep-sea coral (Cohen et al., 2006).
Rayleigh fractionation predicts a quantitative relation-ship between different Me/Ca ratios in an expressionpreviously developed to understand the evolution ofmagma melt-systems (Pearce, 1978; Albarède, 1995).Correlated Mg/Ca and Sr/Ca data from micromilledsamples are used to test the role of Rayleigh fractionationin Me/Ca vital effects across skeletal structural featuresand place this mechanism in the context of other tracervital effects. By examining a key component of manybiomineralization models, our work aims to improve the
Please cite this article as: Gagnon, A.C. et al. Sr/Ca and Mg/Ca vital effectsand the role of Rayleigh fractionation. Earth Planet. Sci. Lett. (2007), doi
understanding of vital-effect mechanisms during calcifi-cation to help better interpret Me/Ca temperature proxies.
2. Materials and methods
2.1. Coral samples
Five individuals of recent D. dianthus were loanedfrom The Smithsonian Institution National Museum ofNatural History (Fig. 2). Two individuals from the sameSouth Pacific location, Smithsonian Collection number47407, are the focus of most of this study. In situ tem-perature at the location and depth of 47407 is estimatedto be 3.6±0.6 °C from the Levitus 1994 dataset (Levitusand Boyer, 1994).
2.2. Micromilling
Exert septa of two individuals from the same SouthPacific location were selected for obvious and widecentral bands (samples 47407A and 47407B). Wedgesof coral are removed to allow analysis in the transverseplane — equivalent to a top view down the body axis.The corals are then mounted on glass slides with epoxy,ground to a thickness of 200–400 μm on a 30 μmdiamond wheel, sonicated without further polishing, andimaged (background grey-scale images in Fig. 3).A computer controlled micromill (Merchantek) is usedto sub-sample parallel with banding to a depth of∼200 μm, starting from one side of the septa andmoving across. We believe growth in the transverseplane outward from the central band represents a growthaxis, with the oldest material deposited as COCs in thecenter of the septa. However, the coral grew in nearconstant conditions with largely time-invariant environ-mental parameters, and the sampling method is designedto clearly separate different skeletal regions rather thanto follow a particular growth axis. Dashed vertical linesin Fig. 3 mark successively milled regions and corre-spond to individual Me/Ca data points. The sup-plemental information (S-1) contains a completedescription of sampling locations. For each data point∼100 μg of milled powder is dissolved in 5% tracemetal clean HNO3 (Seastar).
2.3. Isotope-dilution mass spectrometry
The isotope dilution ICP-MS method used in thisstudy allows rapid and accurate determination of Mg/Caand Sr/Ca ratios. Unlike previous approaches, likethe work of Lea and Martin (1996) which uses acombination of internal and external standardizations to
correlated with skeletal architecture in a scleractinian deep-sea coral:10.1016/j.epsl.2007.07.013
Fig. 2. Collection locations of the recent deep-sea coral D. dianthus samples used in this study. Most of the analyses presented in this paper focus ontwo individuals from South Pacific location 47407.
5A.C. Gagnon et al. / Earth and Planetary Science Letters xx (2007) xxx–xxx
ARTICLE IN PRESS
determine Me/Ca ratios, our method is a completeisotope dilution method, as will be described in detail ina separate paper. Briefly, a combined 25Mg–43Ca–87Srenriched spike is added to a subaliquot of each sample.With a combined spike, Me/Ca ratios are determinedaccurately without an exact knowledge of the spikevolume, simplifying analysis and limiting uncertainty.Isotope ratios are measured in analog mode on aThermoFinnigan Element, a single collector magneticsector ICP-MS. The 24Mg/25Mg and 48Ca/43Ca ratiosare corrected offline for the interference of the doublycharged species 48Ca++ and 86Sr++. Measurement of acombined Mg–Ca–Sr matrix matched standard ofknown isotopic ratio before and after each sample isused to correct for instrumental mass fractionation. Mg/Ca and Sr/Ca metal ratios are calculated using theisotope dilution relationship and assuming naturalisotopic abundance of Mg, Ca and Sr in the coral. The16 sub-samples of 47407B were analyzed twice eachover roughly a 30 h period. For a few hours near themiddle of analysis, instrumental mass fractionationdrifted dramatically. Even though replicate samplesmeasured both during and outside the period of highdrift compared well, all data collected during the highdrift period were disregarded (9 of 36 measurements).Mg/Ca and Sr/Ca measurements for each sample aretabulated in the supplemental information (S-2).
The reproducibility (external error) of our method isassessed by the repeated measurement of a dissolveddeep-sea coral consistency standard. Over three months,
Please cite this article as: Gagnon, A.C. et al. Sr/Ca and Mg/Ca vital effectsand the role of Rayleigh fractionation. Earth Planet. Sci. Lett. (2007), doi
11 measurements of the consistency standard resulted inMg/Ca and Sr/Ca ratios that varied by1.3% and 2.1%respectively (2σ standard deviation). As part of prelim-inary work, three other individuals of coral were analyzedusing an identical method except with an uncalibratedspike, yielding less precise relative Mg/Ca ratios.
2.4. Electronprobe microanalysis
Using a JEOL JXA-8200 electron probe X-raymicroanalyzer, Ca, Mg and Sr abundances were measuredin a separate polished septal piece of a coral also collectedat Smithsonian sample location 47407.Multiple “lines” aremeasured perpendicular to banding in wavelength disper-sive mode (WDS) with a 15 keVelectron beam at a currentof 30 nA focused to a 10 μm spot. Total measurement timeper spot is 90 s, including two backgrounds. Calcite,strontianite and dolomite are used as standards and ZAFmatrix correction is applied. Due to its low abundance indeep-sea coral (∼2mmol/mol), magnesiummeasurementsare very close to the detection limit of our method. On anygiven day, it is not always possible tomeasure theMgpeak.Sr is more abundant thanMg in our coral (∼10mmol/mol)and therefore easier to quantify with the electron probe.
3. Results
Mg/Ca is over twice as high in the optically densecentral band, as compared to the surrounding septa(Fig. 3a–b). The outer septal region is characterized by a
correlated with skeletal architecture in a scleractinian deep-sea coral:10.1016/j.epsl.2007.07.013
Fig. 3. Isotope-dilution ICP-MS measurements of Mg/Ca and Sr/Ca across skeletal features in two D. dianthus septa. (a and b) Mg/Ca ofmicrosamples from septa 47407B and 47407A, respectively, overlaid on negative transmitted light photomicrograph to show sampling locations inrelation to skeletal structural features. Mg/Ca ratios double coincident with the optically dense central band. On this scale, ±2σ error bars are smallerthan the line thickness. (c and d) Sr/Ca of microsamples across 47407B and 47407A, respectively, with smaller and less structured variability thanMg/Ca. Notice that the range of Sr/Ca is much smaller relative to variations in Mg/Ca. Error bars represent ±2σ external error. Septa areapproximately 1.5 mm across. The optically dense central white band, corresponding to COCs, is ∼110 μm across. See supplemental information(Tables S-2a and S-2b) for metal ratio data.
6 A.C. Gagnon et al. / Earth and Planetary Science Letters xx (2007) xxx–xxx
ARTICLE IN PRESS
Mg/Ca of ∼1.5 mmol/mol while the central band isgreater than 3 mmol/mol. The relative Mg/Ca ratios ofthree other D. dianthus individuals collected fromseparate locations are also enriched up to two-foldwithin the central band (Fig. 4). Different amounts ofsmoothing, the incorporation of both central band andouter septal material in the same milling line, mayexplain differences in the relative enrichment betweeneach coral. Outside the central band, Mg/Ca variabilityis ∼18% (2σ standard deviation). Also in this region,Mg/Ca ratios increase with decreasing Sr/Ca ratios.
Unlike Mg/Ca, Sr/Ca ratios show less structure andvary by a smaller relative amount across the coral septa,with a standard deviation of less than 5% (2σ, Fig. 3c–d). The average Sr/Ca of the two corals 47407A and47407B, 10.56 and 10.62 mmol/mol, respectively, arestatistically identical, given our analytical uncertainty.Comparing all data by skeletal region rather than be-tween individual corals, Sr/Ca within the central band
Please cite this article as: Gagnon, A.C. et al. Sr/Ca and Mg/Ca vital effectsand the role of Rayleigh fractionation. Earth Planet. Sci. Lett. (2007), doi
varies significantly less than the surrounding skeleton(F-test, N98% loc, n=23 and 8) with a mean value in thecentral band of 10.6±0.1 (2σ standard error of themean). The standard deviation of Sr/Ca in the centralband is similar to methodological external error and mayrepresent instrumental variability rather than the lowerlimit of coral Sr/Ca variability.
Electron probe results, although qualitative, agree withmilling, showing magnesium enrichment across fourcentral bands (Fig. 5). Volcano like structures, up to20 μm in size, grow during sampling with the electronprobe. Similar observations, attributed to beam damage,have been imaged and described for other biogeniccarbonates (Gunn et al., 1992). Defocusing the beam andreducing beam intensity failed to eliminate the volcano likestructures. In the future, use of amore thermally conductivecoating such as gold or silver (Smith, 1986) may diminishbeam damage. In an experiment to determine the effectof beam damage, we monitored Ca and Sr counts every
correlated with skeletal architecture in a scleractinian deep-sea coral:10.1016/j.epsl.2007.07.013
Fig. 4. Relative Mg/Ca ratios in several corals in relation to structural features. (a) Coral 48739, again showing a nearly 2-fold increase in Mg/Cacoincident with the central band. A region of optically dense material removed prior to milling is marked with a large “X”. (b and c) Relative Mg/Ca incoral 48738 and 85080 also increase in the central band. Although the relative increase in Mg/Ca is less than two-fold in these two corals, theconsistent observation of increased Mg/Ca in the central band was observed in all the corals examined in this study. Relative Mg/Ca ratios weredetermined in reference to a particular microsample within the septa identified as a upside-down triangle in each plot. Error bars noted by blackvertical lines.
7A.C. Gagnon et al. / Earth and Planetary Science Letters xx (2007) xxx–xxx
ARTICLE IN PRESS
5 seconds for the duration of a typical collection. Countsdrift by less than 4% as beam damage occurs. Whilelimiting the quantitative power of the electron probe, thismagnitude of drift should not obscure the mapping of highMg regions assuming Mg and Sr drift are comparable.Unfortunately, the small relative variation of Sr/Ca in thecentral band, as measured by ICP-MS, is the roughly thesame magnitude as beam damage effects, limiting the useof the electron probe method to determine Sr/Caheterogeneity, results which are dominated by noise andnot included here.
4. Discussion
4.1. Magnitude and pattern of Me/Ca vital effects andthe inorganic reference frame
To quantify the absolute magnitude of vital effects,the partitioning of elements between seawater and coralshould ideally be compared to the thermodynamicpartitioning of elements between seawater and abioticaragonite, and then to kinetic effects. Unfortunately,reaching thermodynamic equilibrium in precipitationexperiments has proven difficult; see, for example, thediscussion in Gaetani and Cohen (2006). Therefore,
Please cite this article as: Gagnon, A.C. et al. Sr/Ca and Mg/Ca vital effectsand the role of Rayleigh fractionation. Earth Planet. Sci. Lett. (2007), doi
coral data are typically compared to inorganic experi-ments with kinetic or other effects which may or maynot be present during coral precipitation. Determining avalid inorganic reference frame to compare with coral isa topic of ongoing research.
Fortunately, there is general agreement, over a rangeof experimental conditions and methodologies, betweenmost studies on the partition coefficient of Sr for in-organic aragonite as a function of temperature (Kinsmanand Holland, 1969; Mucci et al., 1989; Zhong andMucci, 1989; Dietzel et al., 2004; Gaetani and Cohen,2006), where the partition coefficient from a solution ofconstant composition is defined as:
DaragSr ¼ Sr
Ca
� �aragonite
= SrCa
� �seawater
ð1Þ
As thoroughly discussed by Gaetani and Cohen(2006), the low partition coefficient of Mucci et al.(1989), by 16% in comparison to the consensus fromother studies, may be the result of kinetic effects. Amajorcontrol of precipitation rate in inorganic systems is thesaturation state of the solution: (Ω=[Ca2
+][CO32−] /ksp).
Mucci et al. precipitated aragonite from constantcomposition solutions with saturation states ranging
correlated with skeletal architecture in a scleractinian deep-sea coral:10.1016/j.epsl.2007.07.013
Fig. 5. Mg/Ca increases in optically dense bands as determined by electron probe X-ray microanalysis in a thin section of D. dianthus fromSmithsonian location 47407. White horizontal lines mark path of analysis.
8 A.C. Gagnon et al. / Earth and Planetary Science Letters xx (2007) xxx–xxx
ARTICLE IN PRESS
between 2.3 and 5. This is significantly lower than the Ωof N20 in Kinsman and Holland (1969), Ω between 10and 31 for Dietzel et al. (2004), andΩ between 40 and 90in Gaetani and Cohen (2006). It is unlikely that coralreach the high saturation states found in most of theabove mentioned inorganic experiments; possibly mak-ing the data of Mucci et al. more applicable tounderstanding coral vital effects. However, it is alsopossible that effects other than kinetics may cause thedifference between Mucci et al. and the other studies. Infact, Mucci et al. and Zhong and Mucci (1989) report thepartition coefficient for strontium is independent ofprecipitation rate over the factor of 14 range in preci-pitation rates explored by their study. Gaetani and Cohenrecognize the results of Mucci et al. and Zhong andMucci, but conclude that the range of low precipitationrates explored in that study is too small to see thedifferences obvious between Mucci et al. and the high Ωstudies. Here, we choose to use the larger and morecomplete dataset of highΩ partition coefficients for Sr tocompare with deep-sea corals, although future work todetermine the DSr
arag at slow precipitation rates over arange of temperatures is certainly necessary. Using thecombined empirical data of Kinsman and Holland(1969), Gaetani and Cohen (2006), and Dietzel et al.(2004) and assuming a Sr/Caseawter of 8.6 mmol/mol (deVilliers, 1999), the Sr/Ca of inorganic aragonite can bepredicted as a function of temperature (solid line and 2σerror envelope in Fig. 6).
The range of Sr/Ca measured from coral microsam-ples both within and outside the central band agree withthe expected Sr/Ca of inorganic aragonite precipitated atthe temperature of coral growth, qualified by the errorenvelope of the inorganic relationship. Sr/Ca variabilityin the central band is smaller and statistically differentfrom that of the outer septa. Inorganic behavior and lowvariability in the central band are consistent with
Please cite this article as: Gagnon, A.C. et al. Sr/Ca and Mg/Ca vital effectsand the role of Rayleigh fractionation. Earth Planet. Sci. Lett. (2007), doi
minimal vital effects for Sr/Ca in this region, while thevariability in the outer septa suggests the influence ofvital effects. Similar mean Sr/Ca values in both thecentral band and outer septa ofD. dianthus are in contrastto the results of Cohen et al. (2006) for another species ofdeep-sea coral, L. pertusa, where opaque bands in thetheca are associated with decreased Sr/Ca relative to thesurrounding skeleton. In surface corals, Cohen et al.(2001) explain both differences in Sr/Ca variabilitybetween the COCs and the surrounding skeleton as wellas different mean Sr/Ca ratios for these skeletal regions bythe influence of photosynthesis. In an elegant test of thishypothesis, Cohen et al. (2002) observed higher Sr/Cavariability outside of COCs in a symbiont containingcoral as compared to an aposymbiont individual of thesame species. We also observe higher Sr/Ca variabilityoutside COCs, but unlike the surface coral data we seesimilar means. Our data suggests that some of thedifference in Sr/Ca variability between the COCs andthe surrounding skeletonmay be ubiquitous in corals bothwith and without photosymbionts.
Unlike Sr/Ca, there is a two-fold difference in Mg/Cabetween the central band and outer septa of D. dianthus,following a similar pattern to skeletal features found insurface corals (Meibom et al., 2006a,b), a commonalitynoted by Sinclair et al. (2006). As the deep-sea corals inthis study grew in near constant environmental condi-tions, the spatially structured 1.5 mmol/mol variation inMg/Ca is a clear indication of vital effects and places alower limit on the magnitude of these effects. Theaverage Mg/Ca of D. dianthus compares well to therange of Mg/Ca found by Shirai et al. (2005) from bulksampling of two different genus of deep-sea coral; butthe results of Cohen et al. (2006) in Lopehelia pertusaare significantly higher, ranging from 2.6 to 4.3 mmol/mol. Cohen et al. (2006), measure smaller spot sizesthan our method, and the larger range in Mg/Ca they
correlated with skeletal architecture in a scleractinian deep-sea coral:10.1016/j.epsl.2007.07.013
Fig. 6. Sr/Ca of D. dianthus compares well with inorganic aragonite. Black box marks total range of all D. dianthus samples where width representsestimated uncertainty in growth temperature. Gray box is smaller range of the optically dense central band. Solid line is an error weighted regressionof the combined data from Kinsman and Holland (1969) (diamonds), Dietzel et al. (2004) (squares), and Gaetani and Cohen (2006) (circles) assuminga Sr/Ca of seawater of 8.6 mmol/mol. Dashed lines mark the 2σ error envelope of the regression. The slope of this inorganic relationship is 44±6(μmol Sr) (mol Ca)−1 °C−1. The data of Mucci et al. (1989) is also plotted. For the experiments of Kinsman and Holland and Dietzel et al., the largestsource of error is the variability between replicate precipitation experiments conducted at the same temperature, represented by 2σ error bars. InGaetani and Cohen each temperature represents a single precipitation experiment. Since Gaetani and Cohen and Kinsman and Holland used similarmethods, the average error of Kinsman and Holland was used to estimate the replicate error of Gaetani and Cohen for the purpose of the weightedlinear regression.
9A.C. Gagnon et al. / Earth and Planetary Science Letters xx (2007) xxx–xxx
ARTICLE IN PRESS
observe may represent skeletal variability with lessspatial averaging than our method; however, the highermean Mg/Ca measured by Cohen et al. (2006) ascompared to Shirai et al. (2005) and this study cannot beexplained by averaging and may represent differencesacross species of deep-sea coral or differences in growthenvironments between the studies.
It is difficult to determine the correct inorganicreference frame for magnesium in aragonite. Gaetaniand Cohen (2006) were able to establish a clear temper-ature relationship for Mg/Ca in aragonite, an importantand significant achievement. However, assuming aseawater Mg/Ca of 5.1 mol/mol, the data of Gaetani andCohen predict aragonite should have a Mg/Ca ratio of∼10 mmol/mol compared to the range of 1.5 to 3 mmol/mol found in D. dianthus and 4 to 5 mmol/mol Mg/Caratios typical of surface corals. To explain the Mg/Caratios observed in the surface coral Diploria and in thedeep-sea coral L. pertusa consistent with these highinorganic Mg/Ca results, Gaetani and Cohen (2006) andCohen et al. (2006) invoke a 3-fold to 8-fold reduction ofMg/Ca in the calcifying fluid with respect to seawater.Although there are no direct measurements of Mg/Ca inthe calcifying fluid, the energetic cost of manipulatingMg/Ca by such a large amount seems prohibitive withouta clear biological purpose (Zeebe and Sanyal, 2002). IfMg/Ca is not manipulated by a large amount, the in-
Please cite this article as: Gagnon, A.C. et al. Sr/Ca and Mg/Ca vital effectsand the role of Rayleigh fractionation. Earth Planet. Sci. Lett. (2007), doi
organic results of Gaetani and Cohen (2006) suggesteither that there is a large vital effect of an absolute offsetin Mg/Ca between corals and inorganic aragonite or thatthe conditions of the inorganic experiment are not a validreference frame for coral growth. The only othersystematic study on magnesium partitioning in aragonite,the inorganic experiments of Zhong and Mucci (1989),find a Mg/Ca of 3.5 mmol/mol at the one temperatureinvestigated, 25 °C, roughly half the result of Gaetani andCohen (2006). Unlike the inorganic behavior of Sr/Ca, aconsensus of supporting studies over a range of experi-mental conditions does not exist for Mg/Ca. The largedifference between Zhong and Mucci and Gaetani andCohen suggest further research on the coprecipitation ofMg in aragonite over a range of experimental conditions isa high priority towards understanding inorganic behavior,and by extension, vital effects.
4.2. Correlated Me/Ca ratios and Rayleighfractionation
The correlated variability of multiple tracers can be astrong constraint on vital effect mechanism, as recog-nized and exploited by Sinclair (2005) in surface coralsand more recently by Sinclair et al. (2006) for Mg/Caand U/Ca in deep-sea corals. A tracer–tracer plot of Sr/Ca vs. Mg/Ca in D. dianthus using the combined data
correlated with skeletal architecture in a scleractinian deep-sea coral:10.1016/j.epsl.2007.07.013
10 A.C. Gagnon et al. / Earth and Planetary Science Letters xx (2007) xxx–xxx
ARTICLE IN PRESS
sets of coral 47407A and 47407B (Fig. 7) shows twotrends: (1) outside the central band, Sr/Ca decreaseswith increasing Mg/Ca in a relationship that is roughlylinear but may show some curvature, and (2) inside thecentral band, corresponding to the centers calcification(COCs), variable and enriched Mg/Ca is associated withconstant Sr/Ca. We test for the presence of a Rayleighprocess in these skeletal regions.
Rayleigh fractionation as applied to individual tracemetal behavior has been discussed thoroughly elsewhere(Mcintire, 1963; Elderfield et al., 1996). Models of coraltrace element variability incorporating Rayleigh frac-tionation as a key mechanism have also been recentlyproposed (Gaetani and Cohen, 2006; Cohen et al., 2006).Here we develop a simple and general relationship formultiple Me/Ca ratio behavior during Rayleigh fraction-ation, similar to the approach of Albarède (1995), andcompare this to our data set.
Following Elderfield et al. (1996), the Sr/Ca ofaragonite precipitated at any point during a Rayleighprocess from a closed solution can be calculated fromthe initial solution composition assuming a constanteffective partition coefficient:
SrCa
� �coral
¼ DCoralSr
SrCa
� �SolO
FDCoralSr �1 ð2Þ
where the extent of precipitation is defined as:
FuCaCao
� �sol
ð3Þ
The effective partition coefficient, DSrCoral relates the
Sr/Ca of the coral skeleton to the surrounding seawaterand is an empirical measure that integrates the thermo-dynamics of coprecipitation, kinetics, binding byorganic molecules and all other process acting on Srunder the specific growth conditions of the coral (Morseand Bender, 1990). Mg/Ca can be described similarly toEq. (2), where Sr/Ca is replaced by Mg/Ca andDSr
Coral byDMgCoral. Since Mg/Ca and Sr/Ca are linked by the extent
of precipitation, F, the expressions can be combined,eliminating F, yielding a linear log–log relationship:
lnSrCa
� �¼ DSr � 1
DMg � 1
� �ln
MgCa
� �
þ lnSrCa
� �O
� DSr � 1DMg � 1
� �ln
MgCa
� �O
� �
ð4ÞUnder the conditions of closed system precipitation
with constant partition coefficients and any initial fluidcomposition, tracer–tracer behavior is predicted to
Please cite this article as: Gagnon, A.C. et al. Sr/Ca and Mg/Ca vital effectsand the role of Rayleigh fractionation. Earth Planet. Sci. Lett. (2007), doi
follow Eq. (4). Testing if correlated tracer behaviorfollows the functional form of Eq. (4) is a general test forRayleigh behavior that does not require knowledge ofinitial solution conditions or partition coefficients. In theouter septal region of D. dianthus, the log–log tracerplot is linear, with an R2 of ∼0.6, (Solid line, insetFig. 7), consistent with a Rayleigh mechanism. Equallyimportant, the behavior of Mg/Ca and Sr/Ca in thecentral band cannot be explained by Rayleigh fraction-ation. In the deep-sea coral L. pertusa, Cohen et al.(2006) also observe decreasing Sr/Ca with Mg/Ca whichthey explain by a model emphasizing Rayleighfractionation. Indeed, while the plot of tracer–tracerbehavior in Cohen et al. (2006) is roughly linear it mayshow some curvature toward low Sr/Ca and high Mg/Ca, as predicted by Eq. (4). Recently, Sinclair et al.(2006) showed that relative U/Ca and Mg/Ca ratios fromboth surface and deep-sea corals follow a power lawrelationship and explained this behavior with end-member mixing. As an alternative explanation toendmember mixing, a Rayleigh process also predicts apower law relationship between U/Ca and Mg/Ca, andmay explain some of the correlated observations.
In Eq. (4) the slope of the log–log relationship isdetermined by the partition coefficients of each metaland the intercept is influenced by both the initialsolution composition and the partition coefficients. Theslope in Eq. (4) is predicted to be relatively less sensitiveto variations in DMg
Coral than DSrCoral, since DMg
Coral is ≪1,driving the denominator of the slope very close to 1 evenover a large relative range of DMg
Coral. The low relativesensitivity of a Rayleigh process to variations in DMg
Coral
has two implications, (1) the tracer–tracer behaviorpredicted by a Rayleigh process is robust to smallvariations in DMg
Coral and (2) it is difficult to determine aprecise DMg
Coral by inverting the fitted slope and using anestimate of DSr
Coral.Instead of inverting partition coefficients from a fit of
the log–log relationship, a predicted Rayleigh relationshipfor tracer–tracer behavior can also be generated fromprescribed effective partition coefficients and initialsolution composition. For this forward modeling, we usean initial Mg/Ca ratio matching the seawater value of5.1 mol/mol. We also make the functional assumption thatthe lowest Mg/Ca ratios measured in our corals correspondto precipitation from the initial unfractionated solution(F=1), implying a DMg
Coral of 2.75×10−4. We choose thispartition coefficient because it describes our coral measure-ments well and does not imply a dramatic modification ofthe Mg/Ca of the calcifying fluid; however, it is sig-nificantly lower than the partition coefficient for Mg ininorganic aragonite as determined by Gaetani and Cohen
correlated with skeletal architecture in a scleractinian deep-sea coral:10.1016/j.epsl.2007.07.013
Fig. 7. Correlated Sr/Ca–Mg/Ca vital effects are consistent with Rayleigh fractionation in the outer septa, but not in the central band. Data points with2σ error bars are measurements from both corals 47407A and 47407B, identified by skeletal location. Inset is a log–log plot of outer septa results witha linear model (bold black line) fit to the data (R2=0.6). The fit parameters of the linear model in the inset were used to plot the Rayleigh Model in thelarger plot (bold black line labeled “Rayleigh Fit”). As an alternative to fitting the data, Rayleigh behavior can be forward modeled with prescribedparameters. In one scenario, initial Mg/Ca is set to a seawater value; initial Sr/Ca is enriched by ∼3% compared to seawater; DMg
Coral is set to2.75×10−4, consistent with the lowest Mg/Ca measurements; and DSr
Coral is set to equal the partition coefficient of inorganic aragonite, 1.24±0.4. Theresults of this calculation are shown as the grey solid line in the main plot with the dashed error envelope representing a propagation of uncertainty inthe inorganic strontium partition coefficient. While most of the discussion in this paper assumes that we sample instantaneous Rayleigh precipitant,the grey solid and dashed lines in the inset are the predicted composition of both instantaneous and accumulated solid, respectively, for the aboveforward Rayleigh model scenario, with relatively tight agreement.
11A.C. Gagnon et al. / Earth and Planetary Science Letters xx (2007) xxx–xxx
ARTICLE IN PRESS
(2006). As described above, uncertainty in DMgCoral has a
small effect on the slope of the tracer–tracer relationship,making our forward model conclusions relatively insensi-tive to this functional assumption.
Given the above assumptions, two Rayleigh fraction-ation scenarios can explain the observed Mg/Ca vs. Sr/Ca relationship of the outer septa. If the abiotic aragonitepartition coefficient, DSr
Arag, which matches the partitioncoefficient of the central band, is also taken as anestimate for DSr
Coral in the outer septa, then the correlateddata are consistent with a Rayleigh process and a non-seawater initial Sr/Ca ratio of 8.8 mmol/mol, anenrichment of less than 3% (solid grey line and dashederror bounds Fig. 7). Evidence of active Sr pumpingexists from cultured surface corals in a study withchemical inhibitors, Ferrier-Pages et al. (2002), al-though, as noted by Sinclair and Risk (2006), the resultsof this study are also consistent with general inhibition ofcalcification and may simply evidence Sr coprecipitationrather than active Sr pumping. If Sr pumping is present,this first Rayleigh scenario is consistent with an initialmanipulation of the calcifying fluid followed by closedsystem behavior during precipitation. Changes tosolution composition prior to precipitation only affectthe intercept of the log–log relationship, shifting the
Please cite this article as: Gagnon, A.C. et al. Sr/Ca and Mg/Ca vital effectsand the role of Rayleigh fractionation. Earth Planet. Sci. Lett. (2007), doi
whole plot vertically in log–log space, but otherwiseleaving the slope and sign of the tracer–tracer correlationunchanged. In a second scenario, initial Sr/Ca is assumedto match seawater. In this case, our data are consistentwith a DSr
Coral in the outer septa of 1.28, within 2σ of theinorganic relationship but different from theDSr
Coral of thecentral band. Our data support the presence of Rayleighfractionation in the outer septa but are unable todistinguish between these two scenarios.
The above expressions ((1)–(4)) describe the com-position of instantaneous precipitate, or the mostrecently formed precipitate at a given F. As it is unclearwhether our analytical method samples instantaneous oraccumulated solid, both must be considered. Theindividual Me/Ca ratio of accumulated precipitate,integrated from the start until a given extent of pre-cipitation, (from 1 to F) can be derived (Doerner andHoskins, 1925; Elderfield et al., 1996). Even though thecombined tracer–tracer relationship for accumulatedsolid is non-linear in a log–log plot, Me/Ca correlationsare nearly identical between accumulated and instanta-neous solids using the same partition coefficients andinitial solution composition (solid and dashed lines ininset of Fig. 7). The most significant difference intracer–tracer behavior between instantaneous and
correlated with skeletal architecture in a scleractinian deep-sea coral:10.1016/j.epsl.2007.07.013
12 A.C. Gagnon et al. / Earth and Planetary Science Letters xx (2007) xxx–xxx
ARTICLE IN PRESS
accumulated solids is the sensitivity to F, with the samepaired Sr/Ca and Mg/Ca ratios corresponding to largerextents of precipitation in the accumulated solid than forinstantaneous solid. F values are sensitive to the choiceof partition coefficients and initial solution composition,which, combined with the difference in F betweenaccumulated and instantaneous precipitant, means thatF is underconstrained by our data set. Qualified bythis large uncertainty, the largest extent of precipitationin our data, corresponding to the most depleted Sr/Caratios and the most enriched Mg/Ca ratios in the outersepta, are consistent with an F value of roughly 0.8 forinstantaneous precipitation and an F value of 0.6 foraccumulated precipitation.
In the above discussion of the Rayleighmechanismweassume a system closed to all metal cations during pre-cipitation. However, active Ca2+ pumping in surfacecorals has been demonstrated in culture experiments(Tambutte et al., 1996) and is supported by the discoveryof a plasma membrane Ca-ATPase localized to thecalcifying region (Zoccola et al., 2004). The effect ofactive calcium pumping on metal/calcium ratios duringprecipitation can be evaluated with a simple analyticalmodel that assumes constant rates of pumping andprecipitation (see supplemental information S-3 for aderivation and discussion). Provided that the rate ofpumping is less than the rate of precipitation, this simplemodel predicts a linear relationship between Sr/Ca andMg/Ca in a log–log plot with a negative slope, similar to acompletely closed Rayleigh process. On the other hand, ifpumping and precipitation occur at the same rate or ifpumping outpaces precipitation, the model predicts a Mg/Ca to Sr/Ca relationship with a positive slope, unlike ourdata. Calcium transport prior to precipitation has the po-tential to modify Me/Ca ratios in the initial calcifyingfluid; however, Ca-ATPase driven calcium-proton ex-change is effectively alkalinity pumping and perturbs[Ca2+] relatively little while still increasing the saturationstate of aragonite dramatically through the speciation ofthe carbonate system. Together, small changes in initial[Ca+2] and a rate of calcium pumping less than the rate ofprecipitation may result in initial Me/Ca ratios close toseawater that then progress along a Rayleigh curve, con-sistent with both the presence of active calcium transportand our data. The constraint of a closed system withregard to non-calcium metal cations does not apply todissolved inorganic carbon (DIC), which can diffusethrough bio-membranes as CO2(aq). A Me/Ca Rayleighprocess is therefore consistent with the equilibrium stableisotope vital effect model of Adkins et al. (2003) whereδ13C is set by net transmembrane CO2(aq) flux, and δ
18O isset by the pH of the calcifying fluid.
Please cite this article as: Gagnon, A.C. et al. Sr/Ca and Mg/Ca vital effectsand the role of Rayleigh fractionation. Earth Planet. Sci. Lett. (2007), doi
4.3. Me/Ca vital effect mechanism in the central band
Large variations in Mg/Ca are uncoupled from Sr/Cain the central band suggesting that Rayleigh fraction-ation is not the dominant process influencing Mg and Srcorrelation in this region. Understanding this vital effecthas implications beyond deep-sea corals, as increasedMg/Ca in COCs appears to be a common feature in bothsurface and deep-sea corals (Meibom et al., 2006a,b;Sinclair et al., 2006). Possible mechanisms for high Mg/Ca in the central band can be evaluated with our data set.
A surface entrapment model predicts precipitationrate dependent behavior of a tracer when growth rate andnear-surface solid diffusion are closely matched (Wat-son, 2004). This model assumes chemical potentials atthe surface and in the bulk-solid differ for a tracer,resulting in a difference of chemical composition as thesolid “traps” more or less of these endmembers. Forsurface entrapment to explain the enriched Mg of thecentral band, assuming rapid precipitation occurs in thisregion and traps more of a surface-like composition, thearagonite surface must be enriched in Mg compared tothe bulk crystal. Sr/Ca, however, is not observed tosystematically vary with Mg/Ca in the central band.Within the surface entrapment framework, this impliesthat either the bulk and surface composition of aragonitehave identical strontium activities, or the balance of near-surface diffusion and precipitation rate are differentenough between Sr and Mg such that Sr is insensitive togrowth rate over the range of rates that occur during coralgrowth. While our data are not a direct test of the surfaceentrapment model, they do highlight key model para-meters and predictions if this process is at work inD. dianthus. Furthermore, if surface entrapment explainsthe uncoupled variability of Mg/Ca and Sr/Ca in thecentral band, the dominance of Rayleigh fractionationin the outer septa may suggest crystal growth rates andconsequently surface entrapment effects do not varysignificantly in the outer septa.
There is evidence for organic material in coral skeletonsthat may help regulate biomineralization (Cuif et al., 2003;Puverel et al., 2005). It is possible that the “organic matrix”or another biomolecule is responsible for enriched Mg inthe central band. A biological component that selectivelybinds Mg but not Sr seems possible given the difference inpolarizability between these two cations. This putativebiomolecule would have to be localized within the COCs,however, the existence and magnitude of Sr vs. Mg traceelement selectivity by organic components within the coralskeleton has yet to be demonstrated.
While far from a comprehensive list of all processesaffecting magnesium, certain conclusions can be made.
correlated with skeletal architecture in a scleractinian deep-sea coral:10.1016/j.epsl.2007.07.013
13A.C. Gagnon et al. / Earth and Planetary Science Letters xx (2007) xxx–xxx
ARTICLE IN PRESS
High Mg/Ca in the central band is not dominated byRayleigh fractionation; may be the result of surfaceentrapment, given key constraints; and, biomoleculesmay be at work, although it is impossible to quantita-tively assess their role without further characterization.
4.4. Uranium/calcium and a continuousbiomineralization model
Differences in Me/Ca behavior between the outersepta and central band raise the possibility of fundamen-tally different biomineralization mechanisms in theseregions. However, previous measurements of U/Caargue against this. In fission track experiments onD. dianthus coral sections similar to ones used in thisstudy, there is a greater than three-fold range in [U]across the coral skeleton (Robinson et al., 2006). Thepattern of this variation is crucial: the region of lowuranium follows but extends well beyond the centralband, with [U] largely unchanged across the interfacebetween the central band and the outer septa. If thecentral band were under the control of a fundamentallydifferent mechanism than the outer septa, it seemsunlikely that [U] behavior would remain unchangedacross the interface of these two mechanisms. The pat-tern of [U] suggests a continuous general mechanismdescribing vital effects and biomineralization. Conclu-sions about uranium can be generalized to other tracers tothe extent that they are influenced by the same processes.Unfortunately the mechanism of U/Ca vital effects is stillan open question. Experiments suggest DU
arag varies by∼3 fold due to pH and precipitation rate (Meece and
Fig. 8. Large Mg/Ca vital effects in D. dianthus compared to extrapolated s(1996), which compares well with recent coral culture experiments of Reynau(1998). Calibrations are marked as solid lines within the temperature rangetemperature range.
Please cite this article as: Gagnon, A.C. et al. Sr/Ca and Mg/Ca vital effectsand the role of Rayleigh fractionation. Earth Planet. Sci. Lett. (2007), doi
Benninger, 1993; Russell et al., 2004), although therelative contributions of these factors and the mechanismof influence are a topic of current research.
4.5. Prospect for a Me/Ca paleothermometer inD. dianthus
Agreement between the mean Sr/Ca of D. dianthusand the abiotic Sr/Ca temperature relationship at a singletemperature point is promising, if preliminary, for thedevelopment a Sr/Ca paleothermometer. Individualdeep-sea corals collected from a range of oceantemperatures, following the approach of Shirai et al.(2005), and analyzed with improved precision isnecessary to determine if a Sr/Ca paleothermometer inD. dianthus can be developed. Improved precision isnecessary because the external error of our current Sr/Camethod corresponds to roughly 5 °C using the slope ofthe relationship between Sr/Ca and temperature forinorganic aragonite. Since Sr/Ca in the central band ofD. dianthus exhibits low variance, with our resultspossibly limited by analytical error rather than coralvariability, localized sampling in this region shouldyield the most precise test of a Sr/Ca paleothermometer.Outside the central band, vital effects may still present achallenge to the development of a Sr/Ca deep-seapaleothermometer in D. dianthus, as the slope of theinorganic Sr/Ca temperature relationship is shallow andeven small vital effects translate into large differences intemperature.
The impact of vital effects on the Mg/Ca paleotherm-ometer can be estimated, in the absence of clear
urface coral data. Surface coral calibrations: (line a) Mitsuguchi et al.d et al. (2006), circles. (line b) Fallon et al. (1999), (line c) Sinclair et al.of the respective experiments, and dashed lines over the extrapolated
correlated with skeletal architecture in a scleractinian deep-sea coral:10.1016/j.epsl.2007.07.013
14 A.C. Gagnon et al. / Earth and Planetary Science Letters xx (2007) xxx–xxx
ARTICLE IN PRESS
inorganic data, by comparing D. dianthus with surfacecorals (Fig. 8)). From Mg/Ca-temperature relationshipscalibrated in surface corals (Mitsuguchi et al., 1996;Sinclair et al., 1998; Fallon et al., 1999; Reynaud et al.,2006) the total ∼2 mmol/mol range of Mg/Ca inD. dianthus corresponds to ∼10 °C. Surface coral tem-perature calibrations correspond best with the outersepta of our deep-sea corals. While the slope of the Mg/Ca temperature relationship, as determined in surfacecorals, is steeper than that of Sr/Ca, the presence of largevital effects can introduce a false and difficult to separatetemperature-like signal if different proportions of skel-etal regions are sampled during analysis.
5. Conclusion
We determined accurate and detailed co-located Sr/Ca and Mg/Ca ratios across different skeletal regions intwo individuals of the deep-sea coral D. dianthusthrough a combination of micromilling and isotopedilution—ICP-MS. Overall, Sr/Ca variability wasrelatively small, with Sr/Ca in the optically densecentral band varying significantly less than in thesurrounding skeleton. The mean Sr/Ca of all skeletalregions generally agree with the predicted Sr/Ca ofinorganic aragonite. This finding combined with theagreement in mean Sr/Ca ratios between two individualsfrom the same location is a promising preliminary resulttowards the development of a deep-sea coral Sr/Capaleothermometer. Since Sr/Ca in the central bandexhibits small variance, sampling localized to thisregion should yield the most precise test of a Sr/Capaleothermometer.
Unlike Sr/Ca, mean Mg/Ca varies dramaticallybetween different skeletal regions. Coincident with theoptically dense central band, Mg/Ca was at least3 mmol/mol, more than twice that of the surroundingskeleton. This result appears to be general, as relativeMg/Ca ratios of three other D. dianthus individualscollected from separate oceanographic locations alsonearly double within the central band. The difference inthe mean Mg/Ca of the central band and surroundingskeleton implies an ∼10 °C temperature signal, whencalibrated via Mg/Ca-temperature relationships devel-oped in surface coral. A large non-environmental effect,or vital effect, can obscure and complicate application ofa Mg/Ca paleothermometer.
While complicating interpretation of a paleotherm-ometer, vital effects are useful to help understand themechanism of biomineralization. Outside the central band,Mg/Ca increases with decreasing Sr/Ca. This relationshipcan be explained by Rayleigh fractionation as shown by a
Please cite this article as: Gagnon, A.C. et al. Sr/Ca and Mg/Ca vital effectsand the role of Rayleigh fractionation. Earth Planet. Sci. Lett. (2007), doi
linear tracer–tracer relationship in a logarithmic plot. To beconsistent with our accurate measurements of Mg/Ca andSr/Ca, Rayleigh fractionation can occur from an initialsolution where Me/Ca ratios match seawater, provided thatthe effective strontium partition coefficient (DSr
Coral) differsbetween the outer septa and central band. Alternatively, ifDSrCoral is the same throughoutD. dianthus and matches the
abiotic partition coefficient, our data is consistent with aninitial solution enriched in Sr/Ca by ∼3% compared toseawater. Rayleigh fractionation implies a closed systemduring precipitation, at least with respect to trace or minormetal cations. In the central band, Me/Ca ratios aredominated by a non-Rayleigh process with our dataconstraining a number of possible explanations for vitaleffects in this region. That Rayleigh fractionation cannotexplain the large variability in Mg/Ca within the centralband is equally important as our evidence for the presenceof Rayleigh fractionation in the outer septa. Given thewidespread feature of enriched Mg/Ca in COCs, themechanism of this process is an important target of futureresearch.
Acknowledgements
Special thanks to Dr. Stephen Cairns and TheSmithsonian Institution National Museum of NaturalHistory for allowing the analysis of the deep-sea coralspecimens in this study. A.C.G. would like to thank D.Rees of the California Institute of Technology forcontinued scientific advice and guidance. Comments byGlen Gaetani and Anne Cohen as well as two anonymousreviewers contributed to improving this manuscript.
Appendix A. Supplementary data
Supplementary data associated with this article canbe found, in the online version, at doi:10.1016/j.epsl.2007.07.013.
References
Adkins, J.F., Boyle, E.A., Curry, W.B., Lutringer, A., 2003. Stableisotopes in deep-sea corals and a new mechanism for “vitaleffects”. Geochim. Cosmochim. Acta 67, 1129–1143.
Albarède, F., 1995. Introduction to Geochemical Modeling. Cam-bridge University Press, Cambridge, pp. 36–38.
Allison, N., Finch, A.A., 2004. High-resolution Sr/Ca records in modernPorites lobata corals: effects of skeletal extension rate andarchitecture. Geochem. Geophys. Geosyst. 5, Q05001. doi:10.1029/2004GC000696.
Allison, N., Finch, A.A., Sutton, S.R., Newville, M., 2001. Strontiumheterogeneity and speciation in coral aragonite: Implications forthe strontium paleothermometer. Geochim. Cosmochim. Acta 65,2669–2676.
correlated with skeletal architecture in a scleractinian deep-sea coral:10.1016/j.epsl.2007.07.013
Cohen, A.L., Layne, G.D., Hart, S.R., Lobel, P.S., 2001. Kineticcontrol of skeletal Sr/Ca in a symbiotic coral: implications for thepaleotemperature proxy. Paleoceanography 16, 20–26.
Cohen, A.L., Owens, K.E., Layne, G.D., Shimizu, N., 2002. The effectof algal symbionts on the accuracy of Sr/Ca paleotemperaturesfrom coral. Science 296, 331–333.
Cohen, A.L., Gaetani, G.A., Lundalv, T., Corliss, B.H., George, R.Y.,2006. Compositional variability in a cold-water scleractinian,Lophelia pertusa: new insights into “vital effects”. Geochem.Geophys. Geosyst. 7, Q12004. doi:10.1029/2006GC001354.
Correge, T., 2006. Sea surface temperature and salinity reconstructionfrom coral geochemical tracers. Palaeogeogr. Palaeoclimatol.Palaeoecol. 232, 408–428.
Cuif, J.P., Dauphin, Y., Doucet, J., Salome, M., Susini, J., 2003.XANES mapping of organic sulfate in three scleractinian coralskeletons. Geochim. Cosmochim. Acta 67, 75–83.
de Villiers, S., 1999. Seawater strontium and Sr/Ca variability in theAtlantic and Pacific oceans. Earth Planet. Sci. Lett. 171, 623–634.
de Villiers, S., Shen, G.T., Nelson, B.K., 1994. The Sr/Ca-temperaturerelationship in coralline aragonite— influence of variability in (Sr/Ca) seawater and skeletal growth-parameters. Geochim. Cosmo-chim. Acta 58, 197–208.
Dietzel, M., Gussone, N., Eisenhauer, A., 2004. Co-precipitation of Sr2+
and Ba2+ with aragonite by membrane diffusion of CO2 between 10and 50 °C. Chem. Geol. 203, 139–151.
Doerner, H.A., Hoskins, W.M., 1925. Co-precipitation of radium andbarium sulfates. J. Am. Chem. Soc. 47, 662–675.
Elderfield, H., Bertram, C.J., Erez, J., 1996. Biomineralization modelfor the incorporation of trace elements into foraminiferal calciumcarbonate. Earth Planet. Sci. Lett. 142, 409–423.
Fallon, S.J., McCulloch, M.T., van Woesik, R., Sinclair, D.J., 1999.Corals at their latitudinal limits: laser ablation trace elementsystematics in Porites from Shirigai Bay. Jpn. Earth Planet. Sci.Lett. 172, 221–238.
Ferrier-Pages, C., Boisson, F., Allemand, D., Tambutte, E., 2002.Kinetics of strontium uptake in the scleractinian coral Stylophorapistillata. Mar. Ecol., Prog. Ser. 245, 93–100.
Gaetani, G.A., Cohen, A.L., 2006. Element partitioning duringprecipitation of aragonite from seawater: a framework forunderstanding paleoproxies. Geochim. Cosmochim. Acta 70,4617–4634.
Gladfelter, E.H., 1982. Skeletal development in Acropora cervicornis:I. Patterns and calcium carbonate accretion in the axial corallite.Coral Reefs 1, 45–51.
Gladfelter, E.H., 1983. Skeletal development in Acropora cervicornis:II. Diel patterns of calcium carbonate accretion. Coral Reefs 2,91–100.
Gunn, J.S., Harrowfield, I.R., Proctor, C.H., Thresher, R.E., 1992.Electron-probe microanalysis of fish otoliths — evaluation oftechniques for studying age and stock discrimination. J. Exp. Mar.Biol. Ecol. 158, 1–36.
Kinsman, D.J.J., Holland, H.D., 1969. Co-precipitation of cations withCaCO3-IV. Co-precipitation of Sr
2+ with aragonite between 16 and96 °C. Geochim. Cosmochim. Acta 33, 1–17.
Please cite this article as: Gagnon, A.C. et al. Sr/Ca and Mg/Ca vital effectsand the role of Rayleigh fractionation. Earth Planet. Sci. Lett. (2007), doi
Lea, D.W., Martin, P.A., 1996. A rapid mass spectrometric methodfor the simultaneous analysis of barium, cadmium andstrontium in foraminifera shells. Geochim. Cosmochim. Acta60, 3143–3149.
Levitus, S. and Boyer, T., 1994. World Ocean Atlas, Volume 4:Temperature. U.S. Department of Commerce, Washington, D.C.
Mcconnaughey, T., 1989a. 13C and 18O isotopic disequilibrium inbiological carbonates: I. Patterns. Geochim. Cosmochim. Acta 53,151–162.
Mcconnaughey, T., 1989b. 13C and 18O isotopic disequilibrium inbiological carbonates: II. In vitro simulation of kinetic isotopeeffects. Geochim. Cosmochim. Acta 53, 163–171.
Mcintire, W.L., 1963. Trace element partition coefficients— a reviewof theory and applications to geology. Geochim. Cosmochim. Acta27, 1209–1264.
Meece, D.E., Benninger, L.K., 1993. The coprecipitation of Pu andother radionuclides with CaCO3. Geochim. Cosmochim. Acta 57,1447–1458.
Meibom, A., Yurimoto, H., Cuif, J.P., Domart-Coulon, I., Houlbreque,F., Constantz, B., Dauphin, Y., Tambutte, E., Tambutte, S.,Allemand, D., Wooden, J., Dunbar, R., 2006a. Vital effects in coralskeletal composition display strict three-dimensional control.Geophys. Res. Lett. 33, L11608. doi:10.1029/2006GL025968.
Meibom, A., Cuif, J.P., Hillion, F.O., Constantz, B.R., Juillet-Leclerc,A., Dauphin, Y., Watanabe, T., Dunbar, R.B., 2006b. Distributionof magnesium in coral skeleton. Geophys. Res. Lett. 31, L23306.doi:10.1029/2004GL021313.
Mitsuguchi, T., Matsumoto, E., Abe, O., Uchida, T., Isdale, P.J., 1996.Mg/Ca thermometry in coral-skeletons. Science 274, 961–963.
Morse, J.W., Bender, M.L., 1990. Partition-coefficients in calcite —examination of factors influencing the validity of experimentalresults and their application to natural systems. Chem. Geol. 82,265–277.
Mucci, A., Canuel, R., Zhong, S.J., 1989. The solubility of calcite andaragonite in sulfate-free seawater and the seeded growth-kineticsand composition of the precipitates at 25-degrees-C. Chem. Geol.74, 309–320.
Puverel, S., Tambutte, E., Pererra-Mouries, L., Zoccola, D., Allemand,D., Tambutte, S., 2005. Soluble organic matrix of two Scleracti-nian corals: partial and comparative analysis. Comp. Biochem.Physiol., Part B Biochem. Mol. Biol. 141, 480–487.
Raz-Bahat, M., Erez, J., Rinkevich, B., 2006. In vivo light-microscopic documentation for primary calcification processes inthe hermatypic coral Stylophora pistillata. Cell Tissue Res. 325,361–368.
Reynaud, S., Ferrier-Pages, C., Meibom, A., Mostefaoui, S., Mortlock,R., Fairbanks, R., Allemand, D., 2006. Light and temperature effectson Sr/Ca and Mg/Ca ratios in the scleractinian coral Acropora sp.Geochim. Cosmochim. Acta. doi:10.1016/j.gca.2006.09.009.
Robinson, L.F., Adkins, J.F., Fernandez, D.P., Burnett, D.S., Wang, S.L.,Gagnon, A.C., Krakauer, N., 2006. Primary U distribution inscleractinian corals and its implications forU series dating.Geochem.Geophys. Geosyst. 7, Q05022. doi:10.1029/2005GC001138.
Russell, A.D., Honisch, B., Spero, H.J., Lea, D.W., 2004. Effects ofseawater carbonate ion concentration and temperature on shell U,Mg, and Sr in cultured planktonic foraminifera. Geochim.Cosmochim. Acta 68, 4347–4361.
16 A.C. Gagnon et al. / Earth and Planetary Science Letters xx (2007) xxx–xxx
ARTICLE IN PRESS
Sinclair, D.J., 2005. Correlated trace element “vital effects” in tropicalcorals: a new geochemical tool for probing biomineralization.Geochim. Cosmochim. Acta 69, 3265–3284.
Sinclair, D.J., Risk, M.J., 2006. A numerical model of trace-elementcoprecipitation in a physicochemical calcification system: appli-cation to coral biomineralization and trace-element ‘vital effects’.Geochim. Cosmochim. Acta 70, 3855–3868.
Sinclair, D.J., Kinsley, L.P.J., McCulloch, M.T., 1998. High resolutionanalysis of trace elements in corals by laser ablation ICP-MS.Geochim. Cosmochim. Acta 62, 1889–1901.
Sinclair, D.J., Williams, B., Risk, M., 2006. A biological origin forclimate signals in corals — trace element “vital effects” areubiquitous in Scleractinian coral skeletons. Geophys. Res. Lett. 33,L17707. doi:10.1029/2006GL027183.
Smith, M.P., 1986. Silver coating inhibits electron microprobe beamdamage in carbonates. J. Sediment. Petrol. 56, 560–561.
Please cite this article as: Gagnon, A.C. et al. Sr/Ca and Mg/Ca vital effectsand the role of Rayleigh fractionation. Earth Planet. Sci. Lett. (2007), doi
Tambutte, R., Allemand, D., Mueller, E., Jaubert, J., 1996. Acompartmental approach to the mechanism of calcification inhermatypic corals. J. Exp. Biol. 199, 1029–1041.
Watson, E.B., 2004. A conceptualmodel for near-surface kinetic controlson the trace-element and stable isotope composition of abiogeniccalcite crystals. Geochim. Cosmochim. Acta 68, 1473–1488.
Weber, J.N., Woodhead, P.M., 1972. Temperature dependence ofoxygen-18 concentration in reef coral carbonates. J. Geophys. Res.77, 463–473.
Zeebe, R.E., Sanyal, A., 2002. Comparison of two potential strategiesof planktonic foraminifera for house building: Mg2+ or H+
removal? Geochim. Cosmochim. Acta 66, 1159–1169.Zhong, S.J., Mucci, A., 1989. Calcite and aragonite precipitation from
seawater solutions of various salinities — precipitation rates andovergrowth compositions. Chem. Geol. 78, 283–299.
Zoccola, D., Tambutte, E., Kulhanek, E., Puverel, S., Scimeca, J.C.,Allemand, D., Tambutte, S., 2004. Molecular cloning and localiza-tion of a PMCA P-type calcium ATPase from the coral Stylophorapistillata. Biochim. Biophys. Acta, Biomembr. 1663, 117–126.
correlated with skeletal architecture in a scleractinian deep-sea coral:10.1016/j.epsl.2007.07.013
