Mg/Ca, Sr/Ca, and stable-isotope (δ18O and δ13C) ratio profiles from the fan mussel Pinna nobilis: Seasonal records and temperature relationships

Authors


Abstract

[1] We present new annually resolved δ18O, δ13C, Mg/Ca, and Sr/Ca ratio records for two shells of the fast growing Mediterranean fan mussel Pinna nobilis, collected from proximal Spanish coast sea grass meadows. The relationship between the potential geochemical proxies and ontogenetic and environmental controlling factors is investigated. Specifically, the use of shell Mg/Ca and Sr/Ca ratios as potential calcification temperature proxies, the latter calculated from measured shell δ18O values, has been assessed. The δ18O cycles along the growth axis indicate that our P. nobilis specimens are ∼10.5 and ∼4.5 years old. Shell Sr/Ca ratios do not exhibit any consistent interannual cyclicity and are not correlated to temperature. A subtle ontogenetic effect on shell Mg/Ca ratios was observed during the first 4.5 years of recorded growth but was highly evident during the organism's later growth years. In P. nobilis shells, different mechanisms influence ontogenetic variation in shell Mg/Ca and δ18O records. Shell Mg/Ca ratios from the first 4.5 years of growth correlate significantly to temperature, in a best fit relationship described by the equation Mg/Ca = 17.16 ± 1.95 * exp(0.022 ± 0.004 * T). P. nobilis shell Mg/Ca records therefore are a valid temperature proxy only during an early growth phase. For the same range of temperatures, shell Mg/Ca ratios in P. nobilis are approximately 1/3 lower than those reported for inorganic calcite but 3 to 4 times higher than in another bivalve species, Mytilus trossulus, and 4 to 16 times higher than in foraminifera. We suggest these offsets are due to a higher degree of similarity between seawater and calcification-fluid composition in P. nobilis than in other bivalves and foraminifera. The observed shell Mg/Ca ratio change per °C of 2.2% also is lower than that observed for inorganic and other biogenic calcites. Our findings strongly support taxon- and species-specific Mg/Ca–temperature relationships for bivalves and other calcifying organisms. An appreciation of the physiology and calcification mechanisms of any biogenic carbonate archive therefore is paramount prior to the application of stable-isotope and element/Ca ratio proxies for paleotemperature reconstructions.

1. Introduction

[2] The use of marine biogenic carbonates, i.e., foraminifera, corals and molluscs (mainly bivalves), as archives of paleoenvironmental information is fundamental to our ability to reconstruct past environmental conditions over a wide range of spatial and temporal scales. The potential utility of mollusc shells as paleoenvironmental archives is due to their incremental deposition, such that they possess in their shell geochemical composition a temporal record reflective of ambient conditions during growth [Jones, 1983; Richardson, 2001]. Marine bivalves also are widely distributed throughout the oceans, from the tropics to the polar regions and from coastal estuarine waters to the deep ocean, displaying a range of growth rates and longevity. Bivalve shell geochemistry thus provides the potential to reconstruct high temporal resolution records of environmental conditions over a wide range of spatial and also temporal scales, e.g., decades, centuries, or even millennial timescales if a sclerochronological approach is taken [e.g., Jones, 1983; Weidman et al., 1994]. Compared to other biogenic carbonate archives there have been relatively fewer studies that have investigated bivalve mollusc shell geochemistry, especially robust validation studies of likely geochemical proxies. In this study we present new geochemical data for two specimens of the fast-growing fan mussel Pinna nobilis, and consider these data in the context of ontogenetic and environmental variability, i.e., metabolic and temperature relationships.

[3] Traditionally, stable-isotope ratios have been one of the most powerful tools used for reconstructing climatic and oceanographic parameters and for understanding environmental processes. It is a common view that the oxygen-isotope composition (18O/16O ratios expressed as a δ18O value) of marine mollusc shell carbonate is deposited at, or near, isotopic equilibrium with the seawater solution from which precipitation occurs and hence reflects a combination of ambient temperature and the oxygen-isotope composition of the seawater medium (δ18Ow) [e.g., Epstein et al., 1953; Wefer and Berger, 1991]. As a consequence, bivalve shell δ18O records have been used to reconstruct seawater temperatures [e.g., Elliot et al., 2003; Kennedy et al., 2001; Klein et al., 1996a; Schoene et al., 2004; Williams et al., 1982]. However, δ18O values lower than that predicted for equilibrium precipitation have been observed periodically in some bivalve species with calcite shells [e.g., Klein et al., 1996a; Krantz et al., 1984; Owen et al., 2002; Tan et al., 1988]. Such differences have been attributed to kinetic effects during calcite precipitation and thus it is apparent that species-specific considerations of such processes must be made if the veracity of temperature reconstructions is to be believed. Such an issue is particularly problematic for extinct species, especially those with no existing analogues.

[4] The interpretation of carbon stable-isotope ratios (13C/12C ratios expressed as δ13C values) in marine biogenic carbonates is more complex than for δ18O values. Several authors have assumed isotopic equilibrium between bivalve shell δ13C values and the carbon-isotope composition of seawater dissolved inorganic carbon (δ13CDIC). This relationship has been used to trace seasonal variations of δ13CDIC [Andreasson and Schmitz, 1998], surface-ocean primary production [Krantz et al., 1987], upwelling events [Killingley and Berger, 1979] and water column stratification [Arthur et al., 1983]. Other studies, however, have demonstrated that kinetic and metabolic factors influence the δ13C record in biogenic carbonates, in addition to δ13CDIC [e.g., McConnaughey, 2003]. There is clear evidence for the incorporation of carbon with a metabolic origin into the shells of some bivalves [Klein et al., 1996b; Owen et al., 2002; Putten et al., 1999; Tanaka et al., 1986], including the fan mussel P. nobilis [Kennedy et al., 2001]. An increase in the incorporation of metabolic carbon with increased age and body size has also been documented in some species [Kennedy et al., 2001; Lorrain et al., 2004] and kinetic effects have been observed to influence P. maximus shell δ13C values [Owen et al., 2002]. Considerations of kinetic and metabolic effects, in addition to environmental conditions, therefore are paramount when interpreting δ13C records obtained from extant and extinct mollusc specimens.

[5] The use of Mg/Ca and Sr/Ca ratios in biogenic carbonates as geochemical proxies has grown rapidly in the last decade and complements the more traditional δ18O and δ13C proxies. However, the factors controlling the trace element composition of biogenic carbonates generally are much less well understood than those governing stable-isotope ratios. The importance of elemental/Ca ratio paleothermometry has encouraged active research into the evaluation of biological and environmental (e.g., temperature, salinity and pH) factors which may influence the incorporation of Mg and Sr into biogenic carbonates. These studies have focused on foraminifera [e.g., Elderfield and Ganssen, 2000; Lea et al., 1999; Nurnberg et al., 1996], corals [e.g., de Villiers et al., 1995] and to a lesser extent bivalves [Klein et al., 1996a, 1996b; Lorens and Bender, 1980; Putten et al., 2000; Stecher et al., 1996]. In M. trossulus, shell Mg/Ca ratios have been shown in a field-based experiment to provide an accurate estimate of seawater temperature with an accuracy of 1.5°C [Klein et al., 1996a]. In the mussel M. edulis, however, a breakdown in the relationship between shell Mg/Ca ratio and seawater temperature has been observed and attributed to a metabolic effect, potentially hampering the use of this proxy for temperature reconstructions [Putten et al., 2000]. Sr/Ca ratios in M. trossulus and M. edulis shells are suggested to be mainly controlled by metabolic effects with a secondary salinity influence [Klein et al., 1996b; Putten et al., 2000]. Consequently, additional investigations of those factors influencing bivalve-shell element/Ca ratio geochemistry are required now.

[6] P. nobilis are large (length up to 1 m) and relatively long-lived (over 20 years) bivalve molluscs indigenous to the Mediterranean. Shells from the genus Pinna have an outer prismatic calcite layer with a thinner aragonite nacreous inner layer [Watabe, 1988]. They occur in sheltered coastal areas, usually among sea grass meadows between 0.5 and 60 m water depth with approximately equation image of the shell buried in the sediments [e.g., Moreteau and Vicente, 1982]. In the Spanish Mediterranean coastal region, P. nobilis exhibits a fast growth rate during the first years and can reach up to 65 cm length over 12–15 years [Kennedy et al., 2001; Richardson et al., 1999]. P. nobilis has been shown to calcify under oxygen-isotope equilibrium with surrounding seawater, confirming the potential for high temporal resolution paleotemperature reconstruction studies to be performed using the first 4–5 years of shell growth [Kennedy et al., 2001]. After this period, a reduction of growth rates causes severe time averaging of the shell record reducing its utility as a temperature archive latter in life. Seasonal cycles in δ13C, also recorded along the shell, displayed a clear ontogenetic trend toward more negative values caused by increasing incorporation of metabolic carbon into the shell with increased age [Kennedy et al., 2001]. For element/Ca ratios, a preliminary study of juvenile P. nobilis found a significant correlation between Mg/Ca and Sr/Ca ratios and calcification temperature estimated from measured shell δ18O, for a period that corresponded to the first annual cycle [Richardson et al., 2004]. The combination of fast growth rate and prior high resolution documentation of consistent ontogenetic changes in δ18O and δ13C values in P. nobilis provides a template that can be used to aid interpretation of new Mg/Ca and Sr/Ca records. Finally, P. nobilis shells have the potential to enable reconstruction of Mediterranean oligotrophic and semiarid environmental coastal settings once the potential geochemical proxy records are understood better.

[7] In this study we investigate shell Mg/Ca and Sr/Ca ratios in the fan mussel P. nobilis as potential temperature proxies, as well as the significance of any ontogenetic metabolic factors, through contemporaneous sampling and measurement of elemental and stable-isotope ratios. More specifically, our aim is to derive empirical element/Ca ratio to calcification temperature calibrations to enable paleotemperature reconstructions. Comparison of our new observations to published relationships for other biogenic carbonate archives further allows us to make an assessment of any taxon- and bivalve species-specific aspects of element/Ca to temperature relationships.

2. Methods

2.1. Field Sampling and Location

[8] Two P. nobilis shells were live collected by divers on September 1995 at ∼5 m depth from Posidonia oceanica meadows on the south east Spanish coast, from two different locations, Aguamarga (Ag4a) and Villaricos (V4b), separated from each other by ∼70 km (Figure 1). Shell Ag4a measured 53 cm (Figure 2a) while shell V4b measured 40 cm (Figure 2b).

Figure 1.

The Spanish Mediterranean coast with the location of the two sampling sites, Aguamarga and Villaricos.

Figure 2.

Pinna nobilis specimens: (a) Ag4a from Aguamarga and (b) V4b from Villaricos. Shell features: shell margin (M); spines (S); shell growth axis (A); umbo (U); sampling grooves (G).

2.2. Shell Preparation and Drilling

[9] Any fouling of the shell surface was gently removed by scrubbing with a soft brush. Powder samples were then collected by milling the external calcite shell layer along the main axis of growth, from the umbo toward the shell margin, each sample taken from within individual growth bands (Figure 2). Samples were collected at spacings ranging from 0.5 to 1 cm using a 0.4 mm wide steel carbide burr (Minerva Dental Ltd) attached to a hand-held dental drill. In the region closest to the umbo the calcite layer was very thin or worn completely away and so could not be sampled. Thus the first milled samples were spaced 5 cm and 1 cm away from the umbo in specimen Ag4a and V4b, respectively. Depth of milling was controlled carefully in order to avoid drilling into the inner nacreous aragonite layer. For the 7 cm closest to the shell margin in shell V4b spine samples were removed sequentially instead of milled powder samples. These spines are conspicuous over the outer shell surface close to the shell margin (Figure 2) and can be sampled as they are an extension of the outer calcite layer, being composed of the same mineralogy and possessing the same stable-isotope composition as the corresponding external calcite shell layer [Kennedy et al., 2001]. Milled powder or spine samples then were split into replicate aliquots for elemental and stable-isotope ratio analyses.

2.3. Stable Isotope Analysis

[10] Stable oxygen- and carbon-isotope ratios were analysed using a PDZ-EUROPA Geo 20-20 dual-inlet stable-isotope ratio mass spectrometer with an automated carbonate system (CAPS) operated at 80°C. Data are reported in per mil (‰) deviations relative to VPDB (Table 1). The overall analytical precision for δ18O and δ13C of calcite samples was 0.03‰ and 0.02‰, based on analyses of an internal laboratory standard run concurrently with all samples.

Table 1. Shell Geochemical and Calcification Temperature Data
ShellDistance From Umbo, cmδ18O, ‰Calcification Temperature, °Cδ13C, ‰Mg/Ca, mmol/molSr/Ca, mmol/molCommentsa
  • a

    Identifies to which samples the text refers as the first 4.5 years and last 4.5 years of specimen Ag4a.

Ag4a5.001.6013.73.0023.50.97First 4.5 years
Ag4a5.500.7217.52.6625.31.01First 4.5 years
Ag4a6.000.3519.22.5629.51.16First 4.5 years
Ag4a7.000.9216.62.5926.71.08First 4.5 years
Ag4a8.001.3514.82.4524.71.20First 4.5 years
Ag4a8.752.3810.52.4920.31.17First 4.5 years
Ag4a9.751.9212.43.0521.21.06First 4.5 years
Ag4a10.750.9916.32.5826.21.22First 4.5 years
Ag4a11.750.6717.72.5324.91.12First 4.5 years
Ag4a12.750.3119.32.5526.11.12First 4.5 years
Ag4a13.750.0620.52.4327.21.24First 4.5 years
Ag4a14.750.2319.72.3026.21.23First 4.5 years
Ag4a15.751.2215.42.2224.21.11First 4.5 years
Ag4a16.751.5513.92.1923.21.21First 4.5 years
Ag4a17.752.0511.82.4420.81.21First 4.5 years
Ag4a18.751.2215.32.6423.71.15First 4.5 years
Ag4a19.750.9616.42.5124.31.22First 4.5 years
Ag4a20.750.3319.22.2426.01.25First 4.5 years
Ag4a21.750.1420.12.2225.11.20First 4.5 years
Ag4a22.800.3519.22.2024.51.13First 4.5 years
Ag4a23.750.8317.02.0422.61.17First 4.5 years
Ag4a24.752.2611.02.1321.31.20First 4.5 years
Ag4a25.751.6613.52.3521.11.23First 4.5 years
Ag4a26.750.7217.52.0325.21.25First 4.5 years
Ag4a28.25−0.2221.81.7526.71.36First 4.5 years
Ag4a29.25−0.0521.01.6824.71.19First 4.5 years
Ag4a30.251.3314.91.9021.21.16First 4.5 years
Ag4a31.251.9012.52.0521.31.12First 4.5 years
Ag4a32.250.6218.02.1325.21.38First 4.5 years
Ag4a33.250.2719.51.7025.31.26First 4.5 years
Ag4a34.10−0.2421.91.6525.01.21First 4.5 years
Ag4a35.401.2415.31.8021.81.18First 4.5 years
Ag4a36.601.5214.11.9421.41.25Not used
Ag4a37.201.1615.61.8822.31.22Not used
Ag4a38.40−0.0220.81.2325.01.24Not used
Ag4a39.601.6213.61.7821.11.21Not used
Ag4a40.102.0811.71.9420.41.15Not used
Ag4a40.601.7413.12.0421.81.20Last 4.5 years
Ag4a41.201.0815.91.6425.11.20Last 4.5 years
Ag4a41.800.6417.91.3823.71.20Last 4.5 years
Ag4a42.200.3019.41.5924.01.22Last 4.5 years
Ag4a42.801.0616.01.4421.51.16Last 4.5 years
Ag4a43.301.5414.01.7422.11.08Last 4.5 years
Ag4a43.90   20.31.08Last 4.5 years
Ag4a44.400.6018.11.1622.91.12Last 4.5 years
Ag4a44.900.0620.51.1024.91.18Last 4.5 years
Ag4a45.401.4514.31.3321.61.10Last 4.5 years
Ag4a45.901.5114.11.5721.11.22Last 4.5 years
Ag4a46.301.6113.71.5522.01.26Last 4.5 years
Ag4a46.802.0112.01.8621.71.17Last 4.5 years
Ag4a47.401.7013.31.6724.71.24Last 4.5 years
Ag4a47.801.0116.31.3823.41.18Last 4.5 years
Ag4a48.401.3714.71.6122.31.10Last 4.5 years
Ag4a48.801.1915.51.5722.61.21Last 4.5 years
Ag4a49.400.8517.00.9425.01.17Last 4.5 years
Ag4a49.900.7417.40.75  Last 4.5 years
Ag4a50.400.8417.01.3826.41.18Last 4.5 years
Ag4a50.800.9916.31.7820.91.21Last 4.5 years
Ag4a51.401.6213.61.5322.11.25Last 4.5 years
Ag4a51.901.6013.71.3223.71.24Last 4.5 years
Ag4a52.400.6317.90.7424.91.17Last 4.5 years
V4b1.001.9312.32.9922.80.83 
V4b2.001.0516.12.8726.60.97 
V4b3.000.4918.52.7227.81.06 
V4b4.000.3619.12.7428.11.04 
V4b5.000.1720.02.7928.21.04 
V4b6.000.3119.42.8725.91.00 
V4b7.000.7117.62.8726.31.01 
V4b8.000.7317.52.7625.71.01 
V4b9.001.7812.92.9724.70.99 
V4b10.002.0511.82.8623.40.98 
V4b10.50   25.41.12 
V4b11.001.8112.92.7824.51.16 
V4b11.50   22.51.15 
V4b12.002.3810.52.7721.91.03 
V4b13.002.1911.32.8021.81.12 
V4b14.001.7513.12.8623.31.15 
V4b15.001.2115.42.9125.81.25 
V4b15.251.2515.22.73   
V4b15.701.1615.62.95   
V4b16.001.0316.12.6027.11.13 
V4b16.301.3814.62.76   
V4b16.800.8017.12.73   
V4b17.001.0116.32.7326.31.06 
V4b18.001.5813.82.9025.11.03 
V4b19.002.4110.42.9821.11.03 
V4b20.002.0012.03.02   
V4b20.50   21.31.04 
V4b21.002.1511.42.87   
V4b21.50   19.71.07 
V4b22.001.3314.82.8421.30.96 
V4b23.001.3814.62.7623.31.03 
V4b24.000.1919.92.5726.71.06 
V4b25.000.0220.72.4627.01.19 
V4b26.000.3719.12.2624.61.11 
V4b26.50   23.61.15 
V4b27.001.6313.62.4222.91.14 
V4b27.50   22.71.18 
V4b28.001.4714.32.41   
V4b28.501.8912.52.5321.51.16 
V4b29.002.2111.22.5023.01.17 
V4b30.001.3914.62.2723.41.32 
V4b31.00−0.2922.11.9026.11.23 
V4b32.000.8417.02.3826.11.27 
V4b33.000.4618.72.2124.91.14 
V4b34.000.7017.62.14   
V4b34.501.2615.22.1523.11.24 
V4b35.001.5913.82.5222.61.31 
V4b35.50   23.41.07 
V4b36.002.0711.71.99   
V4b36.50   21.91.15 
V4b37.002.1111.62.0822.01.16 
V4b37.00   21.21.15 
V4b37.501.9012.52.2621.01.09 
V4b38.000.6917.61.5124.21.28 
V4b39.000.1320.21.6926.81.21 
V4b40.00−0.2521.91.4028.31.27 

2.4. Mg/Ca and Sr/Ca Analysis

[11] Following sample drilling, 0.5 to 2.0 mg of powder was weighed into clean 0.5 mL centrifuge tubes. Samples were dissolved using 0.5 mL Aristar (BOH-Merck) grade 0.1 M HNO3 and then centrifuged to settle any undissolved material. 0.4 mL of solution was pipetted into clean 12 mL centrifuge tubes and all solutions then were diluted to ∼60 ppm Ca concentration prior to element/Ca ratio determination, assuming all the powder material to be composed of CaCO3. Samples were kept in the refrigerator until analysis by ICP-AES.

[12] Mg/Ca and Sr/Ca ratios were analysed on a radially viewed Perkin Elmer 3300RL ICP-AES at the NERC ICP-AES facility situated at Royal Holloway University of London. The wavelengths of the element lines used were 315 nm for Ca, 279.5 nm for Mg and 407 nm for Sr. Calibration was performed via an established intensity-ratio method [de Villiers et al., 2002], using solutions in the range 0–30 mmol/mol for Mg/Ca and 0–5 mmol/mol for Sr/Ca. Calibration solutions were prepared from single-element ICP-MS grade standard solutions, using Aristar 0.1 M HNO3 and 18.2 MΩ Milli-Q water. Our preliminary assessment of instrument performance confirms previous work that indicates minimal matrix effects due to variable Ca concentrations, as well as linearity of elemental-ratio calibration lines, for this ICP-AES configuration [Wara et al., 2003]. Instrumental drift was monitored by running one of the intermediate calibration standards every 5 to 10 samples and data then were corrected accordingly. On the three days of analysis, analytical precision was 0.56%, 0.86% and 1.13% for Mg/Ca (mean = 0.9%) and 1.39%, 1.86% and 1.22% for Sr/Ca (mean = 1.5%), based on analyses of an internal consistency solution run concurrently with all samples. The average precision for Mg/Ca and Sr/Ca of 6 shell samples run in triplicate was 1.3% and 1.5%, respectively.

[13] In order to assess the accuracy of our analytical procedures for Mg/Ca and Sr/Ca ratio determinations, we also analysed a set of solutions prepared by the Elderfield group at the University of Cambridge (M. Greaves, personal communication, 2003) [cf. de Villiers et al., 2002]. An aliquot of each solution was diluted to obtain a Ca concentration of 60 ppm and run concurrently with our samples. The differences between expected and measured values for these solutions are presented in Table 2. Our measured values for the Cambridge solutions generally are less than 2% different from expected values, with the exception of solution 2 for which our Mg/Ca ratio was 7.2% greater. The linearity of our intensity-ratio calibration lines, combined with the independent confirmation of the accuracy of our analytical procedure, confirms the veracity of the high P. nobilis Mg/Ca ratios observed in this study.

Table 2. Comparison of Expected to Measured Values for the Cambridge Reference Solutionsa
Cambridge Solution NumberExpected ValueMeasured Value% Difference
Mg/CaSr/CaMg/CaSr/CaMg/CaSr/Ca
20.5060.5060.5420.510+7.2+0.8
42.3741.0632.4071.055+1.4−0.8
15.1302.0885.2162.093+1.7+0.2
59.1623.5549.2093.609−0.8+1.5
618.4307.33718.3747.439−0.3+1.4

2.5. Estimated Temperature Calculations

[14] In an approach similar to Elderfield and Ganssen [2000], P. nobilis shell calcification temperatures were calculated from measured shell δ18O of calcite using the equation of O'Neil et al. [1969]:

equation image

where δc is the δ18O (VPDB) of shell calcite and δw is δ18Ow (VPDB), calculated as δ18Ow (VSMOW) − 0.27‰ [Hut, 1987]. Due to the extremely low precipitation in the region of the Mediterranean of our study salinity is mainly a function of evaporation, with δ18Ow mirroring salinity changes [Pierre, 1999]. In our shell calcification temperature calculations we assume δ18Ow to be constant, with a value of 1.13‰ as reported for Mediterranean Water from 50 m depth with a salinity of 38.8 [Epstein and Mayeda, 1953]. This assumption is justified since the salinity for this region has an annual average value of 36.85 (2σ = 0.46) [Picco, 1990], corresponding to a variation of only 0.14‰ δ18Ow.

2.6. Statistical Analysis

[15] Two-sample T tests were used to determine statistically whether significant differences existed between the measured geochemical proxies and calculated calcification temperatures. Herein, probability levels less than 5% (p < 0.05) are considered to indicate a significant difference.

3. Results

3.1. Shell δ18O and δ13C Profiles

[16] Both specimens of Pinna nobilis were live collected in September 1995. The δ18O records produced for Ag4a and V4b in this study display 10.5 and 4.5 cycles, respectively. Calcification temperatures predicted from equilibrium isotope fractionation between water and shell calcite have already been used in Ag4a to show that the δ18O record reflects cycles in the annual seawater temperature in southern Spain [Kennedy et al., 2001], corresponding to yearly growth patterns from 1985 to 1995. In V4b the growth period was from 1991 to 1995. Given that milling could not commence directly at the umbo the shell record could probably extend a few months earlier. The δ18O values display similar ranges in both shells, from 2.4‰ to −0.2‰ in Ag4a and from 2.4‰ to −0.3‰ in V4b (Figures 3a and 3b), indicating that both individuals recorded similar minimum and maximum calcification temperatures during their lives. P. nobilis shell accretion is rapid during early growth, i.e., close to the umbo, leading to broad δ18O cycles, but growth rates diminish with age [Richardson et al., 1999], leading to a narrowing of the annual cycles and also an attenuation of the amplitude of the δ18O cycles [Kennedy et al., 2001]. A significant reduction in the amplitude of the δ18O yearly cycles was observed with increasing age over the entire Ag4a shell record (r2 = 0.69, p = 0.002). This amplitude reduction can be attributed primarily to a reduction in the recorded maximum δ18O values (r2 = 0.61, p = 0.005), because the recorded minimum δ18O values are not significantly correlated with age (r2 = 0.27, p > 0.05). No correlation was evident between the amplitude of δ18O ratio cycles and age in Vb4 (r2 = 0.28, p > 0.05). Shell calcification temperatures estimated from the measured shell calcite δ18O values, assuming a constant δ18Ow and using the O'Neil et al. [1969] paleotemperature equation (Figures 3a and 3b), range from 10.5 to 21.9°C in Ag4a and 10.4 to 22.1°C in V4b.

Figure 3.

The δ18O, δ13C, Mg/Ca, and Sr/Ca ratio and calcification temperature records versus distance from the umbo toward the shell margin for shells (a) Ag4a and (b) V4b. Note that the δ18O scale has been inverted to correspond with calcification temperatures increasing upward. Calendar years counted back from shell margin and live sampling date of 1995.

[17] In shells Ag4a and V4b δ13C ratios range from 0.7 to 3.1‰ and 1.4 to 3.0‰, respectively (Figures 3a and 3b). Our current analysis of shell Ag4a shows the previously described pattern of cyclic variation in δ13C ratios, with a trend toward lighter values with increasing distance from umbo and therefore greater age [Kennedy et al., 2001]. The annual cyclicity of δ13C ratios is much less clear in shell V4b, with δ13C values remaining relatively stable in the first 2 years of the shell record, followed by a trend toward lighter values with increasing age over approximately the last 2 years of growth of V4b (Figure 3b). Annual cyclicity in Ag4a δ13C values and the trend toward lighter values with increasing age have been interpreted previously as reflecting a continuous increase in the proportion of metabolic carbon incorporated into the shell, modulated either by seasonal environmental changes or seasonal changes in the extent of metabolic carbon incorporation [Kennedy et al., 2001].

3.2. Shell Mg/Ca and Sr/Ca Ratio Profiles

[18] Mg/Ca ratios range between 20.3 to 29.5 mmol/mol and 20.3 to 28.3 mmol/mol for P. nobilis shells Ag4a and V4b, respectively. Clear seasonal oscillations are evident in the Mg/Ca ratio profiles of both shells, with these trends following the patterns evident in the δ18O records (Figures 3a and 3b). Shell Mg/Ca ratios exhibit a significant inverse correlation to shell δ18O values in both specimens (Ag4a, r2 = 0.67, p < 0.001; V4b, r2 = 0.58, p < 0.001). Mg/Ca ratios in both P. nobilis shells also show a weak, but significant, inverse correlation with distance from the shell umbo (Ag4a, r2 = 0.09, p = 0.039; V4b, r2 = 0.14, p = 0.006), suggesting some degree of ontogenetic control on Mg incorporation in their shells in addition to an apparent relationship to calcification temperatures.

[19] In shells Ag4a and V4b Sr/Ca ratios range between 1.0 to 1.4 mmol/mol and 0.8 to 1.3 mmol/mol, respectively (Figures 3a and 3b). No consistent interannual oscillation in the Sr/Ca records is evident in either specimen throughout its lifetime. Shell Sr/Ca ratios are not significantly related to either shell δ18O or Mg/Ca ratios, which themselves exhibit an apparent relationship to seasonal temperature changes. Increasing Sr/Ca ratios in V4b do show a weak, but significant positive relationship with increased distance from umbo (r2 = 0.46, p < 0.001), an observation also seen in the first 4.5 recorded years of specimen Ag4a (r2 = 0.31, p < 0.001). Such a trend is not evident in the older part of the Sr/Ca record obtained from Ag4a, i.e., the last 4.5 years of growth (r2 = 0.08, p > 0.05).

4. Discussion

4.1. Shell Sr/Ca Ratio Profiles

[20] In both Ag4a and V4b, the Sr/Ca ratio was not significantly correlated to calcification temperature (r2 = 0.05, p > 0.05 and r2 = 0.04, p > 0.05, respectively). In contrast, bivalves with aragonite shells can display a clear seasonality in shell Sr/Ca ratios [Stecher et al., 1996; Toland et al., 2000], consistent with the relationship between Sr/Ca ratios and temperature being under thermodynamic control in aragonite [Speer, 1983]. Sr incorporation also has been shown to increase with higher precipitation rates in inorganic calcite [Morse and Bender, 1990], as well as in some biogenic calcites, in which Sr incorporation was proposed to be under kinetic control [Klein et al., 1996b; Lea et al., 1999]. Therefore the lack of a significant relationship between the Sr/Ca ratios and calcification temperature in the calcite shells of P. nobilis either may be due to the lack of a fundamental thermodynamic control on Sr incorporation into calcite or is confounded by metabolic and possibly kinetic influences.

[21] Sr/Ca ratios in both shells, in the first 4.5 years of recorded growth, show a significant increase with distance from the umbo to the outer margin. In a fossil bivalve Venericardia planiscosta with an aragonite shell, a similar ontogenetic trend of increasing Sr/Ca ratio was observed and attributed to the influence of metabolic effects [Purton et al., 1999]. Using shell δ13C ratios as a proxy shown to be under at least partial metabolic control [Kennedy et al., 2001], the significant inverse correlation between Sr/Ca and δ13C in both Ag4a (r2 = 0.34, p < 0.001) and V4b (r2 = 0.48, p < 0.001), suggests a common mechanism or mechanisms that control both proxies during the first 4.5 years of recorded growth in specimen Ag4a. It is apparent that any such control broke down afterward in Ag4a, indicating that any ontogenetic influence on shell Sr/Ca ratios is inconsistent through the life of the organism. By comparison, Sr/Ca and δ13C ratio profiles from M. trossulus shells display a direct, rather than inverse, relationship [Klein et al., 1996b], suggesting that bivalve shell Sr/Ca ratios cannot be assumed to be controlled solely by metabolic factors.

4.2. Ontogenetic Variability in P. nobilis Shell δ18O and Mg/Ca Ratio Records

[22] P. nobilis shell δ18O and Mg/Ca ratios exhibit notable differences in their variability with increased distance from the umbo (=age) in the longer-lived specimen Ag4a. Therefore the mechanism(s) that cause ontogenetic change in the shell geochemistry appear not to be the same for δ18O and Mg/Ca ratios. First, Mg/Ca is weakly, but significantly, inversely related to increased distance from the shell umbo, while δ18O is not. Second, the reduction in the amplitude of the interannual Mg/Ca ratio cycles with increasing age results mainly from a decrease in the magnitude of the peak Mg/Ca values (r2 = 0.50, p = 0.015), that represent the highest temperatures. Minimum Mg/Ca values are not significantly related to shell age. By comparison, the reduction in the amplitude of the interannual δ18O cycles with increasing age results from a decrease in δ18O maximum values (r2 = 0.59, p = 0.005), i.e., an increase in the minimum shell calcification temperatures recorded. Minimum δ18O values are not significantly correlated to shell age. Thirdly, for the overlapping shell records for the period 1991–1995, i.e., the last 4.5 years of the Ag4a record and the whole of the V4b record, there is a significant difference in the behaviour of the two geochemical proxies. Comparisons between the two independent shell records indicate that for this time interval (1) the mean annual δ18O is not significantly different between specimens (p > 0.05), suggesting minimal δ18O variability between individual shells, but also that differences in mean annual temperature and salinity between the two locations, Aguamarga and Villaricos, are most likely negligible, and (2) the mean annual Mg/Ca ratio is significantly different (p = 0.048), being lower in the last 4.5 years growth of Ag4a, suggesting a significant difference between individual shells that most likely derives from ontogentic changes not related to shell growth rates, which should influence both δ18O and Mg/Ca records.

[23] All of the evidence presented above supports the presence of a variable ontogenetic influence on the absolute Mg/Ca and δ18O values recorded in P. nobilis shells. However, the strength of the ontogenetic influence is far larger during later than earlier growth. Therefore the Mg/Ca ratio versus calcification temperature calibration presented herein is derived solely from geochemical data corresponding to the early growth years of both shells, i.e., 4.5 years of shell accretion, since this is the maximum age of specimen V4b (Table 3 and Figure 4). Additional geochemical records are, nevertheless, required from other long-lived P. nobilis fan mussel specimens in order to confirm and further explain the ontogenetic controls on shell chemistry.

Figure 4.

Comparison of derived Mg/Ca ratio to temperature relationships in biogenic carbonate calcite in bivalves P. nobilis and Mytilus trossulus, as well as planktonic and benthonic foraminifera and inorganic equilibrium calcite. The 95% confidence intervals to the curve fits are shown for bivalve data sets. An exponential calibration was applied to the data from Klein et al. [1996a] although a linear fit was used originally by the authors. Solid circle, P. nobilis (this study); plus sign, M. trossulus [Klein et al., 1996a]; benthonic foraminifera, Cibicidoides spp. [Lear et al., 2002]; planktonic foraminifera, mixed species (gray curve) [Elderfield and Ganssen, 2000] and for G. bulloides (red curve) [Mashiotta et al., 1999]; inorganic precipitation [Oomori et al., 1987]. Note that several estimates for DMg for inorganic calcite are available in the literature, and we use the work of Oomori et al. [1987], which provides an extensive data set with a good temperature control. Calibration equations are included in Table 4.

Table 3. Regression Parameters for Shell Mg/Ca Ratios Versus Calcification Temperature in P. nobilis Using Exponential Fits, With 95% Confidence Intervals
Shell CalibrationMg/Ca = a * exp(b*T)T = a + b * ln (Mg/Ca)nr2SE of Estimate Mg/CaaSE of Estimate °C
abab
  • a

    Standard error of estimate Mg/Ca is equal to square root of mean square error (MSE) in units of ln (Mg/Ca).

Ag4a < 4.5 years16.278 (±1.11)0.0236 (±0.006)−71.32 (±23.05)27.65 (±7.24)320.65±0.056±1.9
Ag4a last 4.5 years18.171 (±1.21)0.0153 (±0.012)−34.74 (±39.77)16.08 (±12.66)210.25±0.060±2.0
V4b17.427 (±1.08)0.0217 (±0.005)−82.51 (±22.68)30.69 (±7.09)380.67±0.053±2.0
V4b + Ag4a < 4.5 years17.159 (±1.95)0.0216 (±0.004)−74.82 (±17.06)28.50 (±5.34)700.62±0.057±2.1

4.3. Relationship Between P. nobilis Shell Mg/Ca Ratios and Calcification Temperatures

[24] Shell Mg/Ca ratios in the first 4.5 years of recorded growth of P. nobilis specimens Ag4a and V4b are significantly correlated to calcification temperatures (r2 = 0.64, p < 0.001 and r2 = 0.67, p < 0.001, respectively). All the individual data values obtained from this growth interval in both specimens have been used and correlated using an exponential fit of the form: Mg/Ca (mmol/mol) = a * e(b*T), where a is a pre-exponential constant, b is an exponential constant and T is the calcification temperature in °C. However, there is no significant difference (p > 0.05) in using an exponential fit over a linear fit to the data. The exponential fits to the data for each of the two specimens (Table 3) are not significantly different from each other (p > 0.05), which indicates minimal variation in the relationship between Mg/Ca and temperature for the first 4.5 years of recorded growth in each organism. Data from both shells were then combined to determine a single exponential fit of shell Mg/Ca ratios to calcification temperature, including 95% confidence intervals on the derived relationship (Table 3, Figure 4). The error in temperature estimates associated with the exponential fit to the P. nobilis data is ±2.1°C (Table 3), while the error associated to an analytical uncertainty in Mg/Ca determinations of 0.9% is ±0.3°C.

4.4. Comparison to Other Mg/Ca–Temperature Relationships

[25] The thermodynamic temperature dependence of Mg substitution into calcite is predicted to be exponential [Lea et al., 1999], and such exponential fits have been used in inorganic precipitation experiments [e.g., Mucci, 1987; Oomori et al., 1987] and for planktonic and benthonic foraminifera [e.g., Lea et al., 1999; Lear et al., 2002; Martin et al., 2002; Rosenthal et al., 1997]. Most work on the validation and application of Mg/Ca ratios in biogenic carbonates as a means of reconstructing past temperatures has focused on extant and fossil foraminifera. Our work extends such calibrations to a new marine bivalve species, the fan mussel Pinna nobilis, and provides evidence for significant interspecies difference in the shell Mg/Ca ratio to temperature relationship for marine bivalves.

[26] Partition coefficients (Dx) are a means of quantifying the partition of an element between solution and solid phases, i.e., DMg = Mg/Cacalcite/Mg/Caseawater. Reported values for DMg, the partition coefficient for Mg incorporation into inorganic calcite precipitated from seawater under chemical equilibrium at 25°C and 1 atm show a wide range: 0.0123 (±0.008) [Mucci and Morse, 1983], 0.0172 (±0.0022) [Mucci, 1987], and 0.019 (±0.001) [Oomori et al., 1987]. Oomori et al. [1987] provide one of the best and most extensive data sets with a good temperature control and as such we use there values in our study. Assuming a modern seawater Mg/Ca molar ratio of 5.13 [Morse and Bender, 1990], the range of partition coefficients for Mg uptake into our P. nobilis shells, across the range of calcification temperatures determined, is between 0.004 and 0.006 (1σ = 0.0004). These values are lower than reported DMg for inorganic calcite but are, however, higher than one calculated in a similar manner for another bivalve species M. trossulus (DMg = 0.0012; 1σ = 0.0003), using shell Mg/Ca data presented by Klein et al. [1996a]. Biogenic carbonates, including most foraminifera (DMg = 0.001 to 0.002 [e.g., Lea et al., 1999]) with the exception of some neritic shallow-water benthonic foraminifera that have high shell Mg contents [Toyofuku et al., 2000], have substantially lower DMg than measured values for equilibrium inorganic-calcite precipitation. Clearly, such organisms have evolved biological mechanisms that regulate the incorporation of Mg into their calcite shells. It is interesting to recognise, however, that the DMg calculated for P. nobilis is the closest, of all the marine organisms for which well-constrained shell Mg/Ca ratio data now are available, to that measured for inorganic calcite (Figure 4). This observation likely reflects the physiology and specific calcification mechanism of P. nobilis.

[27] In bivalves, shell deposition occurs from the fluid (extra-pallial fluid or EPF) present in the extra-pallial space (EPS), an enclosed environment situated between the mantle, a thin organ that in bivalves completely encloses the other soft-parts of the animal, and the inner shell surface [e.g., Wilbur and Saleuddin, 1983]. In most bivalves, e.g., the mussels M. trossulus and M. edulis, the mantle is constantly attached to the shell edge through an organic layer, the periostracum, sealing the EPF from the external environment [e.g., Watabe, 1988]. A thick periostracum has been described in mytilids, the family which includes M. trossulus and M. edulis [Harper, 1997]. Consequently, the inorganic composition and pH of the EPF of M. edulis has been found to be different from ambient seawater and is controlled by valve opening [Crenshaw, 1972, 1980]. Similarly, in foraminifera calcite deposition occurs under a strong biological control, at a site of mineralization that is significantly different in composition from ambient seawater [Elderfield et al., 1996; Wolf-Gladrow et al., 1999; Zeebe et al., 1999]. In bivalve species with intricate and ornamented shell morphology, e.g., oysters, scallops and P. nobilis, the periostracum is thin (<1 μm) and in some species even perforated [Harper, 1997], thereby allowing ions to diffuse passively and freely through the periostracum so that the chemical composition of the EPF is more similar to seawater. Hickson et al. [1999] proposed such an open transport model to explain oxygen-isotope equilibrium in the shell of the queen scallop Aequipecten opercularis.

[28] In P. nobilis the mantle is most often retracted from the shell edge, where deposition of new shell occurs, into the interior of the pallial cavity, suggesting the EPS to be a non-permanent feature. As such, we speculate that in P. nobilis the EPF has a chemical composition more similar to seawater than that for other calcifying organisms. This factor is a probable reason for DMg values in P. nobilis being closer to inorganic calcite, than those for other bivalve species (i.e., M. trossulus) and foraminifera. Furthermore, different calcification mechanisms used by bivalves and foraminifera might explain the lower Mg/Ca ratio to temperature sensitivity in bivalves (P. nobilis, 2.2% per °C for and M. trossulus, 4.9% per °C), which are closer to the value for inorganic calcite (3.3% per °C), than foraminifera (∼6 to 11% per °C) (Table 4).

Table 4. Comparison of Published Mg/Ca–Temperature Calibrations for Inorganic Calcite Precipitated in Chemical Equilibrium as Well as for Biogenic Calcitesa
 Mg/Ca Range (mmol/mol)Temperature Range (°C)Calibration Mg/Ca = a * exp(b*T)Commentb
ab
Inorganic calcite45 to 12010 to 5044.50.033(1)
 
Foraminifera
Benthic<1 to ∼30.8 to 18.40.788 to 1.0080.061 to 0.119mixed species (2)
 1 to ∼70.8 to 18.40.867(±0.049)0.109(±0.007)Cibicidoides spp (3)
 92 to 15010 to 2594.40.0175neritic, high Mg (4)
Planktonic∼1 to ∼70 to 300.28 to 0.550.083 to 0.102(5)
 0.5 to ∼40 to 190.520.1calcification temperature e
 2 to 810 to 250.4740.107(6)
 
Bivalves
P. nobilis19.7 to 29.510.4 to 2217.159(±1.95)0.022(±0.004)calcification temperature (7)
M. trossulus2.8 to 10.86.5 to 233.106(±1.13)0.049(±0.008)roasted at 380°C; salinity: 25 to 28.9 (8)

[29] Strong taxon- and species-specific controls on DMg values and resultant shell Mg/Ca ratios have profound implications for the application of biogenic carbonate and specifically (sub-) fossil mollusc shell element/Ca ratio records to the reconstruction of past temperatures. It is now apparent that an understanding, or at least an appreciation, of an organism's physiology and calcification mechanism is paramount to enable the use of an appropriate shell Mg/Ca ratio to calcification temperature relationship otherwise the veracity of reconstructed temperatures will be in doubt.

5. Conclusions

[30] 1. New geochemical data for the fan mussel Pinna nobilis show that shell Sr/Ca ratios are not a valid temperature proxy, since they most likely are predominantly controlled by a combination of metabolic and kinetic factors.

[31] 2. Shell Mg/Ca ratios in P. nobilis are a valid temperature proxy during the first 4 to 5 years of growth, after which ontogenetic effects decouple significantly the relationship between shell Mg/Ca ratios and temperature.

[32] 3. The Mg/Ca–temperature best-fit relationship during the early growth years is described by the equation Mg/Ca = 17.16 ± 1.95 * exp(0.022 ± 0.004 * T).

[33] 4. Mg/Ca ratios in P. nobilis shells are higher than in another bivalve species (M. trossulus) and foraminifera across the same range of calcification temperatures. We suggest these offsets are due to a higher degree of similarity between seawater and calcification fluid composition in P. nobilis than in other bivalves and foraminifera.

[34] 5. The sensitivity of shell Mg/Ca ratios to temperature in P. nobilis shells is lower than that for inorganic calcite, and also 2 to 5 times lower than the response of other bivalve species or foraminifera.

[35] 6. Our findings support the view that Mg/Ca–temperature relationships in bivalves are species-specific, such that an assessment of the physiology and calcification mechanisms of biogenic carbonate archives is required before robust paleotemperature reconstructions can be made from extinct even extant organisms.

Acknowledgments

[36] The authors thank Paul Kennedy at the School of Ocean Sciences for his help and technical support for stable-isotope-ratio analyses. Access to the U.K. Natural Environment Research Council ICP-AES Facility at the Royal Holloway University of London, contract ICP/196/1201, as well as the technical assistance of Sarah James and Jacqui Duffet, is acknowledged gratefully. This research was partially funded by Fundação para a Ciência e Tecnologia (FCT), Portugal, through a scholarship to Pedro Freitas, contract SFRH/BD/10370/2002.

Ancillary