Element / Calcium ratios in middle Eocene samples of Oridorsalis umbonatus from Ocean Drilling Program Site 1209

Introduction Conclusions References


Conclusions References
Tables Figures

Back Close
Full 3 ] and temperature evolution at Site 1209 using multi-element ratios and global minimization solutions.

Materials and methods
Element/calcium ratios were measured on specimens of Oridorsalis umbonatus picked from the >150 µm size fraction.Samples were processed and cleaned for contaminating phases using the standard Cambridge oxidative procedure that is based on Barker et al. (2003).All down-core B/Ca and Li/Ca ratios and some Mg/Ca and Sr/Ca ratios were determined from matrix-matched intensity ratios on the Q-ICP-MS at the University of Cambridge following the procedure of Yu et al. (2005).Additional down-core Mg/Ca and Sr/Ca measurements were made using the Varin Vista ICP-AES following the method of de Villiers et al. (2002).These two methods yield consistent Sr/Ca and Mg/Ca results based on long-term dual measurements (M.Greaves, personal communication, 2011) All data were screened for contaminating phases using ancillary element ratios (i.e.Al/Ca, Fe/Ca, Si/Ca, Mn/Ca).Any samples that fell outside the range reported by Barker et al. (2003) were excluded.We routinely achieved the analytical accuracy and precision on the X /Ca ratios reported by Yu et al. (2005).The relative standard deviation (% rsd) on replicate benthic foraminifera samples (core top specimens, n = 3) is better than 1.2 % for Li/Ca, 1 % for B/Ca, 0.2 % for Mg/Ca and 0.07 % for Sr/Ca.Ocean Drilling Site 1209 ( 32• 39.108 N, 158 • 30.3564E) was located in the northern subtropical Pacific Ocean throughout the middle Eocene, with an estimated paleodepth of ∼1.9-2.5 km (Dutton et al., 2005;Bohaty et al., 2009).The intermediate water depth of Site 1209 was above the Eocene CCD (Tripati et al., 2005), although carbonate preservation may have been influenced by the relatively shallow Pacific lysocline at this time (Hancock and Dickens, 2005).Introduction

Conclusions References
Tables Figures

Back Close
Full Chronostratigraphic datums for Site 1209 are based on biostratigraphic markers and indicate that the studied interval (140-164 revised meters composite depth, rmcd) is middle Eocene in age (Bralower, 2005;Petrizzo et al., 2005).We use the "best-guess" age model for Site 1209 reported in Dawber and Tripati (2011) but note that due to high planktonic fragmentation in parts of the studied interval and the regional diachrony of middle Eocene calcareous nannofossil datums (Petrizzo et al., 2005;Bohaty et al., 2009), there may be some uncertainty associated with the age model.The age model may be subject to reinterpretation if biostratigraphical datums are later revised.
Specimens of O. umbonatus become increasingly fragmented in the upper part of the studied interval.Care was taken to select the best-preserved specimens for analysis (i.e.intact and non-chalky).Foraminiferal dissolution indices are from Dawber and Tripati (2011).The weight percentage of organic carbon was calculated using the "loss on ignition" principal (e.g.Heiri et al., 2001).O. umbonatus δ 13 C measurements were made on a homogenized sub-sample of the specimens used to determine Mg/Ca and Sr/Ca via ICP-AES (but separate to X /Ca analyses via Q-ICP-MS).Carbon isotope ratios were determined on two gas source mass spectrometers in the Department of Earth Sciences, University of Cambridge.Long term analytical precision based on replicate analyses of an in-house standard is 0.06 ‰. same increasing trend between ∼39 and 37 Ma, but Mg/Ca initially decrease and then increase ca.37.5 Ma.

]
The empirical core top relationships established for O. umbonatus Li/Ca, B/Ca, Mg/Ca and Sr/Ca and bottom water ∆[CO 2− 3 ] (Table 1, Dawber and Tripati, 2011) are applied to the middle Eocene X /Ca records from Site 1209.The residence time of Li, B, Mg, Sr and Ca ranges between 10 5 -10 7 yr (Broecker and Peng, 1982), therefore it is likely that seawater X /Ca were different during the Eocene.Although there are proxy-and modelbased constraints for some of the element ratios (e.g.Mg/Ca, Wilkinson and Algeo, 1989;Horita et al., 2002;Coggon et al., 2010), there are notable differences between published estimates.We therefore consider relative changes in middle Eocene intermediate water ∆[CO

2−
3 ] implied by each of the X /Ca records (Fig. 1).X /Ca-estimated changes in ∆[CO 3 ] established from core tops is not appropriate for the Eocene or that secondary parameters contribute to, or may indeed be the dominant influence on O. umbonatus X /Ca.Yu and Elderfield (2007) report changes in North Atlantic Ocean intermediate water and the Holocene based on a record of Cibicidoides mundulus B/Ca.For comparison, Sr/Ca data for the same core (from C. wuellerstorfi, Lear et al., 2003), when normalised to O. umbonatus using the species offsets of Lear et al. (2003) and converted into ∆[CO 2− 3 ] using the core top calibration from Dawber and Tripati (2011), yield a LGM-Holocene change of ∼70 µmol kg −1 .Although this comparison of LGM-Holocene ∆[CO 2− 3 ] estimates may be overly simple and biased by inaccuracies in the species corrections, core top calibrations, and the analytical uncertainty associated with measurements on different instruments, it illustrates that there are also discrepancies be- 3 ] reconstructions during the Pleistocene.For Li, B, Mg and Sr, seawater composition across the LGM-Holocene transition should be similar to modern, minimizing any potential influence on the X /Ca sensitivity to ∆[CO Mg/Ca ratios in core top specimens of O. umbonatus exhibit a strong correlation with bottom water temperature (Lear et al., 2002(Lear et al., , 2004;;Rathmann et al., 2004;Healey et al., 2008;Dawber and Tripati, 2011), although the systematics of this relationship are poorly understood.A number of studies have used the Mg/Ca ratio of O. umbonatus to Introduction

Conclusions References
Tables Figures

Back Close
Full 3 ] and other parameters is minor/negligible (e.g.Lear et al., 2000Lear et al., , 2004;;Billups and Schrag, 2003;Dutton et al., 2005;Tripati and Elderfield, 2005;Tripati et al., 2005;Sosdian and Rosenthal, 2009;Dawber and Tripati, 2011).Multiple linear regression of new and published core top O. umbonatus Mg/Ca against bottom water temperature and ∆[CO 2− 3 ] yields a more significant relationship than simple linear regression on either parameter (Table 2, significance assessed by regression r 2 value, p-values of parameter coefficients and the Akaike information criterion).We consider several subsets of the combined Mg/Ca data as uncorrected anthropogenic DIC data is not available for all localities, and some studies have questioned the reliability of some of the published Mg/Ca data due to potential issues arising from contamination or diagenetic alteration (i.e.Marchitto et al., 2007).Regardless of the data subset used in the multiple linear regression, the Mg/Ca sensitivity to temperature is an order of magnitude greater than the sensitivity to ∆[CO 2).Multiple linear regression is not however, an infallible test of the relative influence of temperature and ∆[CO 2− 3 ] on O. umbonatus Mg/Ca as these two parameters exhibit a high degree of covariation in the oceans leading to issues of spatial autocorrelation in the regression.As similar approach to that adopted by Yu and Elderfield (2008), which examines depth dependent variations in benthic Mg/Ca across the last Glacial Maximum-Holocene transition at a site with well constrained bottom water temperature and ∆[CO Unfortunately, measurements of core top O. umbonatus Li/Ca, B/Ca and Sr/Ca that span a significant temperature range are not available, so it is not possible to rigorously assess the temperature influence on these element ratios.However, we note that a weak, negative relationship between Li/Ca and temperature has been reported for Introduction

Conclusions References
Tables Figures

Seawater composition
Foraminifera are thought to calcify through the vacuolization of seawater (Erez et al., 1994;Erez, 2003), so foraminifera X /Ca may reflect changes in the magnitude and/or X /Ca of cation fluxes into and out of the ocean.Sequence stratigraphic and geochemical proxy reconstructions support highly variable sea level during the middle Eocene (Browning et al., 1996;Miller et al., 2005;Tripati et al., 2005;Dawber and Tripati, 2011;Dawber et al., 2011).Increases in the Ca flux from continental weathering when unaccompanied by carbonate compensation is a mechanism of lowering seawater X /Ca, and has been proposed as one way to account for the apparent increase in seawater [Ca] during the middle Miocene glacial expansion (Griffith et al., 2008).During the Neogene the magnitude of seawater [Ca] variations on time scales of 2-4 Myr is ∼20 % (Fantle and DePaolo, 2005), which is similar to the long-term variability in the middle Eocene Sr/Ca and Mg/Ca records at Site 1209.However, it is difficult to conceive that seawater [Ca] could be decoupled over million year periods from calcite compensation.Griffith et al. (2008) suggest that changes in the rate of dolomitization and or the Ca:HCO − 3 of riverine inputs are possible processes to decouple the Ca and C cycle, but additional work is required to evaluate these hypotheses.Short-term (<1 Myr) variations in the X /Ca at Site 1209, if solely due to changes in seawater [Ca], would require extremely large and unrealistic changes over a geologically rapid time (∼15 % in [Ca] in <400 kyr).
The release of Sr into the ocean via the erosion and/or recrystallisation of continental shelf aragonite may be an important control on seawater Sr/Ca over a range of time scales, and is hypothesized to drive Pleistocene glacial-interglacial variations of up to ∼12 % (Stoll andSchrag, 1998, Stoll et al., 1999).Switching the locus of carbonate deposition between the shelf and deep sea may amplify changes in seawater Sr/Ca since most aragonite calcifying species live on the continental shelf (Martin et al., 1999).The Introduction

Conclusions References
Tables Figures

Back Close
Full effect of aragonite weathering fluxes on seawater Li-and B concentrations is not well constrained because partition coefficients for biogenic aragonite species are largely undetermined.
During parts of the middle Eocene, the Site 1209 B/Ca record is positively correlated with the seawater δ 18 O reconstruction for this site (Fig. 3, Dawber and Tripati, 2011).
A notable exception is between ∼38.7 and 38 Ma when seawater δ 18 O decreases, but B/Ca increases (Fig. 3).In contrast, the O. umbonatus Sr/Ca and Li/Ca records are negatively correlated with seawater δ 18 O throughout the middle Eocene (Fig. 3).Be-  3 ] and temperature using the Nelder-Mead algorithm, we compute the values of these parameters and element ratio regression coefficients that minimize the sum of the squared difference between the actual and predicted element ratios, i.e. a global minimum solution.The technique is sensitive to outlying data points therefore we restrict the dataset to remove large amplitude, high frequency variability that is unsupported by two or more data points.
We find that there are several solutions to the Site 1209 data set that provide estimates of middle Eocene temperature and ∆[CO 2− 3 ] that are consistent with existing proxy constraints.No single solution consistently and accurately replicated all four of the measured X /Ca ratios.In some solutions, the Li/Ca and B/Ca values were replicated very well, but the B/Ca ∆[CO 2− 3 ] regression coefficient and the Mg/Ca temperature coefficients were negative, which is the opposite relationship to existing core top data for O. umbonatus and other benthic species (e.g. this study, Lear et al., 2000Lear et al., , 2002;;Healey et al., 2008;Yu and Elderfield, 2007).Another solution, provided consistent estimates for Li/Ca, Mg/Ca and Sr/Ca across a range of values, although the accuracy was poor and B/Ca ratios were poorly replicated.
The lack of a congruent solution likely reflects the limitation of our assumptions that X /Ca in O. umbonatus are influenced by only two parameters and/or the X /Ca temperature and ∆[CO 2− 3 ] regression coefficients are constant throughout the middle Eocene.Secondary diagenesis and changes in seawater composition are additional parameters that may influence X /Ca in O. umbonatus and may need to be factored into the regression models.Global minimization of foraminifera multi-element ratio data is still a potentially powerful paleoproxy tool, but additional studies examining down core data with better constrained hydrographic and carbonate chemistry histories are need to access the utility of this method.Introduction

Conclusions References
Tables Figures

Back Close
Full

Influence of pore water chemistry?
A potential uncertainty of investigating O. umbonatus X /Ca ratios as proxies for bottom water ∆[CO

2−
3 ] is its reported infaunal habitat, in the upper 0-4 cm in the sediment (Corliss, 1985;Rathburn and Corliss, 1994).Models predict that pore water ∆[CO 2− 3 ] may be substantially different (up to 25 %) from bottom waters and vary considerably within the upper few centimeters of the sediment (Martin and Sayles, 1996) as a result of pore water processes (e.g.respiratory oxidation of organic matter in sediments) and acid neutralization reactions (Emerson and Bender, 1981;Archer 1991).In saturated bottom waters where pore water ∆[CO 2 3 ] is predicted to decrease rapidly relative to bottom waters (Martin and Sayles, 1996), or in environments with high rates of respiratory driven dissolution, the reported sensitivity of benthic foraminiferal X /Ca ratios to bottom water ∆[CO 2− 3 ] may be underestimated for infaunal taxa.However, it would be difficult to quantifying this uncertainty.In addition, infaunal foraminifera may migrate within the sediment in response to food availability, temperature, and carbonate saturation (Gross, 2000), which may introduce additional inaccuracies into the core-top calibration.
Pore water borate concentrations may also become decoupled from bottom water levels, as the speciation of dissolved boron is pH-dependent (e.g.Hemming and Hanson, 1992).Borate is the species thought to be incorporated into calcium carbonate (Hemming and Hanson, 1992), therefore a shift in the speciation of the pore water boron reservoir as a result of respiratory driven dissolution and acid neutralization reactions may introduce some bias into B/Ca-based reconstructions of ∆[CO

Possible implications for middle Eocene climate
If the seawater δ 18 O reconstruction for Site 1209 accurately reflects changes in continental ice storage (Dawber and Tripati, 2011) (Lyle et al., 2005;Tripati et al., 2005), and a compilation of carbonate content for other tropical Pacific sites have been interpreted to record increased deep water carbonate preservation associated with glacial expansion (ca.41.5 Ma), as a result of a deepening of the saturation horizon (ca.41.5 Ma; Lyle et al., 2005;Tripati et al., 2005).The apparent discrepancy between records from Site 1209 and other Pacific sites may reflect local differences in carbonate saturation.It is also possible that the lysocline and CCD were decoupled during the middle Eocene (Site 1209 would have been closest to the paleo-lysocline).Alternately O. umbonatus Sr/Ca ratios may not primarily reflect changes in bottom water ∆[CO 2− 3 ].

CPD Introduction Conclusions References
Tables Figures

Back Close
Full except for B/Ca and Li/Ca -except for B/Ca and Li/Caµmol mol Full  (2006).CAE is an abbreviation for the carbonate-accumulation events documented by Lyle et al. (2006).The calcite compensation depth (CCD) reconstruction is from Tripati et al. (2005).Introduction

Conclusions References
Tables Figures

Back Close
Full Screen / Esc Printer-friendly Version Interactive Discussion Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | a seawater oxygen isotope reconstruction and (4) by calculating down core estimates of bottom water ∆[CO 2− Screen / Esc Printer-friendly Version Interactive Discussion Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | X/Ca ratios Records of O. umbonatus Li/Ca, B/Ca, Mg/Ca and Sr/Ca (Fig. 1) display large amplitude variations on short-and long-time scales during the middle Eocene (up to 25 % in ∆Sr/Ca, 30 % in ∆Li/Ca, 40 % in ∆Mg/Ca and 56 % in ∆B/Ca).On short time scales (<1 million years), B/Ca ratios do not exhibit any statistically significant correlation with the Li/Ca, Mg/Ca or Sr/Ca records (Fig. 1).Li/Ca and Sr/Ca exhibit the Discussion Paper | Discussion Paper | Discussion Paper | inferred from the smoothed function fitted to the raw data.Estimates of intermediate water ∆[CO 2− 3 ] based on Sr/Ca and Mg/Ca span ∼37-45 Ma, with additional estimates from Li/Ca and B/Ca between ∼41-37 Ma (measurements of Li/Ca and B/Ca not made on samples prior to 41 Ma).Between ∼38.7 and 37 Ma, ∆[CO 2− 3 ] estimates based on Sr/Ca and Li/Ca increase (∼60 µmol kg −1 and ∼85 µmol/respectively), while estimates based on B/Ca and Mg/Ca show no net change.Although the variability of ∆[CO 2− 3 ] based on B/Ca and Mg/Ca is broadly comparable (∼15 µmol kg −1 and 20 µmol kg −1 respectively), the structure of the B/Ca and Mg/Ca records is notably different.The discrepancy in the change in ∆[CO 2− 3 ] based on the Sr/Ca and Li/Ca records (∼25 µmol kg −1 ) may in part reflect the large amount of high frequency variability (∼20 %) in the Sr/Ca record around 39-38.2 Ma, which is compensated for by the smoothing function.If the change in ∆[CO 2− 3 ] between 38.7 and 37 Ma is based on the lower values of the Sr/Ca raw data (∼0.69mmol mol −1 , n > 15 samples), rather than the value of the smoothed function, the value (80 µmol kg −1 ) is comparable to the estimate based on the Li/Ca record. Discussion Paper | Discussion Paper | Discussion Paper | at ∼40.5 Ma to at least 38 Ma.The O. umbonatus δ 13 C record for Site 1209 also shows a small decrease (0.3-0.4 ‰) during CAE-4, which may reflect a decrease in carbon isotope composition of pore waters as a result of organic carbon remineralization.The apparent discrepancy between the Site 1209, 1218 and 1219 weight percent CaCO 3 records, the CCD reconstruction, the Site 1209 Sr/Ca and Li/Ca estimates of ∆[CO 2− 3 ] and the Site 1209 foraminifera dissolution indices during CAE-4 is somewhat of a paradox.Taken at face value, these trends may indicate a decoupling of the CCD and lysocline and/or bottom water and pore water ∆[CO 2− 3 ] at Site 1209.During earlier carbonate accumulation events in the middle Eocene, foraminifera dissolution indices are consistent with the weight percent CaCO 3 records for Site 1209, 1218 and 1219 (Fig. 2).The Site 1209 Sr/Ca record also exhibits small increases, equivalent to ∼20 µmol kg −1 , during CAE-3 at ∼41 Ma, and CAE-2 at ∼44.5 Ma.The Mg/Ca record shows the opposite response to the Sr/Ca record, the foraminifera dissolution indices and CCD reconstruction during CAE-4, CAE-3 and CAE-2, suggesting that ∆[CO 2− 3 ] is not the dominant influence on Mg/Ca.During CAE-3 there is no change in the Site 1209 O. umbonatus δ 13 C and weight percent organic carbon records, or in the O. umbonatus δ 13 C record across CAE-2.O. umbonatus δ 13 C and weight percent organic carbon are influenced by additional processes, e.g.CaCO 3 :C organic rain rate, which may be obscuring changes in these proxies due to bottom water and/or pore water ∆Sr/Ca and Li/Ca based estimates for the change in ∆[CO 2− 3 ] at Site 1209 are consistent with the reconstruction of the Pacific CCD on million year timescales.However, quantitatively the Sr/Ca and Li/Ca based estimates of ∆[CO 2− 3 ] across CAE-4 are inconsistent both with each other and with the estimated change in the depth of the CCD, assuming a similar scaling to the Oi-1 glaciation (i.e.1.2 km deepening of the CCD and an estimated change of 54 µmol kg −1 in Pacific deepwater Discussion Paper | Discussion Paper | Discussion Paper | the depth of the CCD, would propagate to intermediate water depths.Nonetheless, the discrepancies in the structure and amplitude of the X /Ca records at Site 1209 indicate that either the sensitivity to ∆[CO 2− Therefore discrepancies between the different X /Ca-∆[CO 2− 3 ] reconstructions likely indicates the influence of secondary parameters.4.1 Influence of secondary parameters on middle Eocene X/Ca records 4.1.1Temperature Screen / Esc Printer-friendly Version Interactive Discussion Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | reconstruct past temperature variations on the assumption that the sensitivity of Mg/Ca to bottom water ∆[CO 2− , would provide a secondary assessment of the relative influence of these parameters.Nonetheless, the structure of the Mg/Ca record for Site 1209 is consistently different to the Li/Ca, B/Ca and Sr/Ca records during the middle Eocene, further suggesting that it is temperature, rather than ∆[CO 2− 3 ], that is the dominant influence on O. umbonatus Mg/Ca.
Screen / Esc Printer-friendly Version Interactive Discussion Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper |

4. 2
Estimates of ∆[CO 2− 3 ] and temperature based on multi-element ratios The comparisons of the Site 1209 middle Eocene X /Ca records with each other, independent carbon cycle proxies and a seawater δ 18 O reconstruction indicate that element ratios in O. umbonatus are not controlled solely by bottom water ∆[CO 2− 3 ].Recently, Gaetani and Cohen (2006) and Gaetani et al. (2011) introduced an application of the global minimization technique to solve variations in multiple parameters based on multi-element ratios in corals.Global minimization of multi-element ratio data in foraminifera is potentially a powerful paleoproxy tool for investigating the relative sensitivity of X /Ca to several environmental parameters and a means of estimating down Discussion Paper | Discussion Paper | Discussion Paper | core variations in these parameters.As an exercise, we assume that Li/Ca, B/Ca, Mg/Ca and Sr/Ca in O. umbonatus can be defined by multiple linear functions of both bottom water ∆[CO 2− 3 ] and temperature.By iteratively changing initial estimates of bottom water ∆[CO 2− Screen / Esc Printer-friendly Version Interactive Discussion Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Screen / Esc Printer-friendly Version Interactive Discussion Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | as determined from multiple linear regression.Additional culturing work and down core multi-element studies records are required to further assess the potential of O. umbonatus X /Ca ratios as proxy archives for bottom water ∆[CO 2Discussion Paper | Discussion Paper | Discussion Paper | tion state, Earth Planet.Sc.Lett., 258, 73-86, 2007.Yu, J. and Elderfield, H.: Mg/Ca in the benthic foraminifera Cibicidoides wuellerstorfi and Cibicidoides mundulus: Temperature versus carbonate ion saturation, Earth Planet.Sc.Lett., 276, 129-139, 2008.Yu, J., Day, J., Greaves, M., and Elderfield, H.: Determination of multiple element/Discussion Paper | Discussion Paper | Discussion Paper |
tween 41 and 40.5 Ma, Li/Ca appear to be decoupled from Sr/Ca and seawater δ 18 O, although the significance of this trend is unclear as it is supported by only a few data points.Sr/Ca values of O. umbonatus decrease by ∼21 % across the ∼1.1 ‰ positive shift in seawater δ18 O commencing at ∼41 Ma, which is interpreted as a major episode of glacial expansion associated with a glacioeustatic lowering(Dawber and  Tripati, 2011).If aragonite weathering was primarily responsible for seawater Sr/Ca variations, the O. umbonatus Sr/Ca record should be positively correlated with the seawater δ 18 O reconstruction.Our observations suggest that glacially mediated variations in the continental and shelf flux of Sr to the ocean are not the primary influence on O. umbonatus Sr/Ca.Li and B partition coefficients for biogenic shelf aragonite are needed to evaluate this effect on Li/Ca and B/Ca ratios in O. umbonatus.
, the observed variations in O. umbonatus δ 13 C, Li/Ca, B/Ca and Sr/Ca records support the hypothesis that middle Eocene glaciation was closely coupled to the carbon cycle.Although, the lack of a consistent correlation between all four records throughout the middle Eocene likely indicates that additional parameters are influencing the Li/Ca, B/Ca and Sr/Ca records and that the processes linking the records are complex.The O. umbonatus Sr/Ca record exhibits the most striking correlation with the seawater δ 18 O reconstruction, but as discussed, the relationship is opposite to what might be expected if O. umbonatus Sr/Ca was controlled by seawater composition related to shelf aragonite recrystallisation.If O. umbonatus Sr/Ca ratios are primarily governed by bottom water ∆[CO 2− 3 ], the reconstructions imply that during middle Eocene glacial intervals, carbonate saturation was reduced at Site 1209.Detailed records of CaCO 3 MARs for Sites 1218 and 1219 Detailed records of O. umbonatus Li/Ca, B/Ca, Mg/Ca and Sr/Ca from ODP Site 1209 exhibit well-defined and large amplitude shifts on a number of time-scales during the middle Eocene.Bottom water ∆[CO 2− 3 ] may be an important influence on X /Ca, however discrepancies in the nature and magnitude of downcore X /Ca records at Site 1209 suggest that either empirically derived coretop regression models are not appropriate for the early Cenozoic and/or X /Ca are influenced by secondary parameters.The O. umbonatus Mg/Ca record for Site 1209 is consistently different to the Li/Ca, B/Ca and Sr/Ca records suggesting that bottom water ∆[CO 2− 3 ] is not the dominant control on Mg/Ca ratios for this species.This hypothesis is supported by the order of magnitude difference in the relative sensitivity of O. umbonatus Mg/Ca to temperature and ∆[CO

Table 1 .
Linear least square regression models fitted through X /Ca and ∆[CO