4.2. Dissolution Unaffected Mg/Ca Ratios From Shallow Water Depths
Evidence for excellent carbonate preservation at shallow water depths is provided by the presence of aragonitic pteropod shells in our Caribbean core-top sediments down to ∼2700 m. Likewise, species-specific Mg/Ca ratios do not systematically decrease with increasing water depth in the depth interval <2000 m (correlation coefficients r2 = <0.1). We assume these shallow Mg/Ca to be unaffected by dissolution.
There are significant differences in Mg/Ca ratios between the species and varieties from <2000 m (Figure 2; Table 2). Mixed layer-dwelling G. ruber exhibits highest Mg/Ca of 4.20–6.02 mmol/mol. In accordance with increasing calcification depth and decreasing ambient seawater temperature, Mg/Ca successively decrease to 1.64–2.28 mmol/mol for the deepest-living species G. crassaformis (Figure 2). With respect to G. ruber, the other species show lower mean Mg/Ca in samples <2000 m water depth (Figure 3): G. sacculifer is lower in Mg/Ca by ∼15%, G. menardii by ∼34%, N. dutertrei by ∼40%, G. tumida by ∼49%, G. truncatulinoides by ∼49%, and G. crassaformis by ∼60%. G. ruber pink shows consistently lower mean Mg/Ca from <2000 m by ∼7% with respect to the white variety. The left-coiling variety of G. truncatulinoides (sinistral) is lower by ∼5% with respect to the right-coiling (dextral) variety (Figures 2 and 3). These differences between species are in general agreement to previous multispecies core-top studies of Hecht et al. , Lorens et al. , Rosenthal and Boyle , Russell et al. , Brown and Elderfield , Hastings et al. , Dekens et al. , and Anand et al.  (Figure 2).
Figure 3. Species-specific mean Mg/Ca ratios from <2000 m water depth (error bars indicate standard deviations) ranked from highest to lowest Mg/Ca versus relative calcification depth. Black dots: shallow-dwelling species; circles: thermocline-dwelling species; open square: bottom-of-photic-zone-dwelling species; black squares: deep-dwelling species. For comparison, eastern Caribbean annual mean temperatures and salinities (±standard deviations) [NODC, 2001] versus water depth are shown. Shaded area indicates the thermocline. Horizontal dashed line marks the base of the photic zone (inferred from the irradiance value of 0.1%). CW, Caribbean Water; SUW, Subtropical Underwater.
Download figure to PowerPoint
4.3. Temperature Versus Mg/Ca Ratios Unaffected by Dissolution
The accuracy of Mg/Ca versus temperature calibration curves depends on both the accuracy of Mg/Ca analyses and, more importantly, on the accuracy of the temperature assignment. The temperature assignment, however, is dependent on the foraminiferal habitat (e.g., water depth, season, life cycle, nutrient supply), which often comprises a broad range of water depths wherein calcification takes place. In Figure 4, we plotted the foraminiferal Mg/Ca ratios from <2000 m versus Caribbean annual mean temperatures [NODC, 2001]. These temperatures were converted from the specific foraminiferal habitat assignments found in the literature [Fairbanks et al., 1980, 1982; Fairbanks and Wiebe, 1980; Erez and Honjo, 1981; Hemleben et al., 1989; Sautter and Thunell, 1991a; Schweitzer and Lohmann, 1991; Ravelo and Fairbanks, 1992; Mulitza et al., 1997; Kemle-von Mücke and Oberhänsli, 1999; Schmuker and Schiebel, 2002; Anand et al., 2003; McKenna and Prell, 2004]. Considering only the habitat specifications of Schmuker and Schiebel  inferred from abundance maxima of eastern Caribbean plankton tows (April–May 1996), an exponential multispecies Mg/Ca versus temperature relationship suggests itself, which approaches the multispecies calibration curve of Anand et al.  established for the Sargasso Sea (Figure 4). The slope of such a calibration curve, however, is highly dependent on the determination of the foraminiferal habitat depths. In fact, while estimates on where foraminifers live and where calcification occurs are manifold or speculative, δ18O data of primarily shallow-dwelling species indicate that planktonic foraminifera calcify in depth zones that are significantly narrower than the overall vertical distribution of these species implies [Fairbanks et al., 1980]. From Mg/Ca of the thermocline-dwelling species G. menardii and N. dutertrei, which are lower than those of the shallow-dwelling species G. ruber and G. sacculifer (Figures 3 and 4), we hypothesize that the potential calcification depths of these species, where the Mg/Ca signal is generated, are clearly deeper than the shallow abundance maxima given by Schmuker and Schiebel .
Figure 4. Species-specific mean Mg/Ca ratios from <2000 m water depth (box heights indicate Mg/Camean ± standard deviations) versus eastern Caribbean annual temperatures [NODC, 2001]. The eastern Caribbean annual temperatures were converted from the according habitat depths compiled for the different species and their varieties (widths of open and filled boxes; see section 4.3). Considering only those habitat specifications (filled boxes) derived from eastern Caribbean plankton tows [Schmuker and Schiebel, 2002] (0–20 m (red bars); 100–200 m (orange bar); 200–300 m (blue bar)), the Mg/Ca versus temperature relationship comes close to the multispecies calibration curve of Anand et al.  (black curve).
Download figure to PowerPoint
Despite the overall correspondence between foraminiferal Mg/Ca ratios and temperature shown in Figure 4, which heavily relies on assumptions concerning the foraminiferal habitats, we note a broad intraspecific range of Mg/Ca in core-top samples from <2000 m water depth, in accordance to early observations of Nürnberg  on the high variability of Mg/Ca in foraminiferal tests. The scatter appears to be largest in shallow- and thermocline-dwelling species with a range of ∼1.4 mmol/mol in G. ruber white, ∼1.8 mmol/mol in G. ruber pink, ∼0.9 mmol/mol in G. sacculifer, ∼1.3 mmol/mol in G. menardii, and ∼1.5 mmol/mol in N. dutertrei, and a commonly smaller range in deep-dwelling species of ∼0.7 mmol/mol in G. tumida, ∼1.1 mmol/mol in G. truncatulinoides dextral, ∼0.1 mmol/mol in G. truncatulinoides sinistral, and ∼0.7 mmol/mol in G. crassaformis (Figure 2). The observed range of core-top Mg/Ca appears high, but is close to other shallow core-top and sediment-trap data from other ocean areas at similar temperatures showing a wide range of Mg/Ca of ∼1 mmol/mol [e.g., Elderfield and Ganssen, 2000; Whitko et al., 2002; Anand et al., 2003] (Figure 5).
Figure 5. Caribbean core-top Mg/Ca ratios versus temperature data of 9 planktonic foraminiferal species and varieties in comparison with published Mg/Ca-paleotemperature equations. Horizontal bars indicate entire annual temperature ranges [NODC, 2001] at the respective core locations from preferred living depths of the foraminiferal species and varieties. Symbols are placed at April-May temperatures when foraminiferal abundance maxima were observed [Schmuker and Schiebel, 2002]. Shaded areas indicate the Mg/Ca data variability around the Anand et al.  multispecies calibration curve (solid lines: Mg/Ca = 0.38 · exp(0.09T)). Average standard deviations of replicate Mg/Ca analyses (stdev) is indicated by vertical bars. (a) Deep-dwelling G. crassaformis (triangles), G. truncatulinoides dextral (dots) and sinistral (open circles), and shallow-dwelling G. sacculifer (crosses): Habitat depths of ∼230 m, ∼160 m, and ∼40 m, respectively (see section 4.3); species-specific paleotemperature calibration curves for G. truncatulinoides dextral from McKenna and Prell  (Mg/Ca = 0.355 · exp(0.098T)), and for G. sacculifer from Dekens et al.  (Mg/Ca = 0.37 · exp(0.09T)) and Nürnberg et al. [1996a, 1996b] (Mg/Ca = 0.472 · 10^(0.036T)). (b) Deep-dwelling G. tumida: Habitat depth defined to ∼180 m; species-specific paleotemperature equation for G. tumida from Rickaby and Halloran  (Mg/Ca = 0.53 · exp(0.09T)). (c) Thermocline-dwelling G. menardii and shallow-dwelling G. ruber white (open squares): Habitat depths defined to ∼150 m and ∼34 m, respectively; species-specific paleotemperature equations for G. ruber white from Lea et al.  (Mg/Ca = 0.3 · exp(0.09T)), Dekens et al.  (Mg/Ca = 0.38 · exp(0.09T)), Whitko et al.  (Mg/Ca = 0.57 · exp(0.074T)), and McConnell and Thunell  (Mg/Ca = 0.69 · exp(0.068T)). (d) Thermocline-dwelling N. dutertrei and shallow-dwelling G. ruber pink (squares): Habitat depths defined to ∼150 m and ∼37 m, respectively; species-specific paleotemperature equation for N. dutertrei from Dekens et al.  (Mg/Ca = 0.6 · exp(0.08T)).
Download figure to PowerPoint
The observed scatter in the core-top Mg/Ca ratios may be due to several reasons: (1) An annual mixed layer temperature range of ∼2–3°C at the respective core sites within the mixed layer (i.e., ∼30–50 m water depth, assumed habitat of shallow-dwelling species [Schmuker and Schiebel, 2002]) and a smaller annual temperature range of only ∼1–2°C at intermediate water depths (down to ∼230 m water depth, assumed habitat of G. crassaformis [Schmuker and Schiebel, 2002]) (Figure 5) introduces a potential variability in Mg/Ca of ∼0.1–0.4 mmol/mol when applying the Mg/Ca versus temperature equation of Anand et al. . (2) The assumed age range of ∼2–3 kyrs for the core-top sediments (see section 3.1) might have introduced a certain variance in Mg/Ca due to short-term climate variations. Indeed, a cooling of Caribbean surface water temperatures of about 2°C during the Little Ice Age [Watanabe et al., 2001] could account for a change in Mg/Ca of ∼0.7 mmol/mol. (3) A species-specific size-effect on Mg/Ca generally was avoided by selecting foraminiferal tests from narrow size ranges (see section 3.1; Table 2). Especially for the deep-dwelling species, however, we were forced to widen the size range due to insufficient numbers of specimens for analysis (Table 2). Taking the published deviations from mean Mg/Ca of 12–36% by enlargement of the foraminiferal test sizes [Elderfield et al., 2002; Anand and Elderfield, 2005], we estimate the size effect on our Mg/Ca to a maximum of ∼0.2–0.7 mmol/mol. (4) The presence of varying amounts of secondary, gametogenic calcite (see section 3.2), which is added to the foraminiferal tests at greater water depths and hence records lower temperatures, biases Mg/Ca toward lower ratios. The quantitative effect on Mg/Ca, however, remains uncertain as we have no control on the amounts of gametogenic calcite. (5) Unintended mixing of morphotypes may have led to a widened scatter in Mg/Ca. Foraminiferal species are known to form different morphotypes showing different geochemical signatures [e.g., Williams et al., 1981; Deuser and Ross, 1989]. Steinke et al.  presented a statistically significant difference in Mg/Ca for two morphotypes of G. ruber white of 0.38 ± 0.30 mmol/mol. In spite of all these factors potentially affecting Mg/Ca, the succession of foraminiferal species and varieties according to their Mg/Ca clearly reflects the expected depth habitats (Figure 3).
As we are not able to establish core-top Mg/Ca versus temperature calibrations due to the restricted temperature range covered by our Caribbean samples, we compared our (dissolution-unaffected) core-top Mg/Ca data from <2000 m water depths to published Mg/Ca versus temperature relationships. Mg/Ca ratios of the deep-dwelling species G. truncatulinoides dextral and sinistral, and G. crassaformis exhibit a clear relationship to temperature. Each species covers a regional temperature range of ∼5–6°C when applying the habitat specification of Schmuker and Schiebel  of ∼160 m and ∼230 m water depth, respectively (Figure 5a). The G. truncatulinoides dextral and sinistral, and G. crassaformis data fall onto published Mg/Ca versus temperature curves of Anand et al.  proposed for multispecies planktonic foraminifers, and of McKenna and Prell  for G. truncatulinoides dextral (Figure 5a). Also, the Mg/Ca data of Caribbean G. sacculifer agree with published calibration curves of Nürnberg et al. [1996a, 1996b], Dekens et al. , and Anand et al.  when relating water temperatures from ∼40 m water depth [Schmuker and Schiebel, 2002] to the Mg/Ca ratios (Figure 5a).
The Mg/Ca ratios of G. tumida are very close to those of G. truncatulinoides (Figure 4), suggesting akin living depths. If we assume a living depth of ∼180 m, which is in accordance to Ravelo and Fairbanks  and Chaisson and Ravelo , than the according Mg/Ca versus temperature data fall onto the Anand et al.  calibration curve (Figure 5b). Thermocline-dwelling G. menardii and N. dutertrei have less well-defined habitat specifications. In view of the low Mg/Ca ratios, the extremely shallow abundance maximum of ∼40–50 m observed by Schmuker and Schiebel  is highly unlikely to truly reflect the foraminiferal calcification depth (Figure 4). Assuming an average depth habitat of ∼150 m water depth inferred from various sources [Fairbanks et al., 1980, 1982; Fairbanks and Wiebe, 1980; Sautter and Thunell, 1991b; Ravelo and Fairbanks, 1992; Chaisson and Ravelo, 1997] results in Mg/Ca versus temperature data matching the Anand et al.  calibration curve reasonably well (Figures 5c and 5d).
In contrast to the above mentioned foraminiferal species, G. ruber is different. First, the spread of Mg/Ca ratios in dissolution-unaffected core-top samples is largest among our studied species. Second, Mg/Ca of G. ruber white and pink deviate from existing Mg/Ca versus temperature calibrations when considering depth habitats of ∼34 m and ∼37 m, respectively, which are the April-May abundance maxima in Caribbean plankton nets [Schmuker and Schiebel, 2002] (Figures 5c and 5d). Such depth estimates for G. ruber seem plausible as MOCNESS plankton tows from the western Gulf of Mexico show the depth preference for G. ruber white to be 0–45 m, for G. ruber pink to be slightly deeper (25–65 m) [Tedesco and Thunell, 2003]. Even if one would assume a very shallow habitat of <10 m being ∼0.2°C warmer than at ∼30 m, the Mg/Ca data would still deviate from available calibration curves. The reason for that is not yet clear. It is evident, however, that G. ruber definitely occupies a shallower habitat than G. sacculifer as suggested by higher Mg/Ca (Figure 4) and generally lighter δ18O values [Anand et al., 2003]. Their oxygen isotope signal commonly points to summer sea-surface conditions [Deuser, 1987], and both Flower et al.  and Anand et al.  pointed out that Mg/Ca of G. ruber white commonly represents warmer than average temperatures in the Orca Basin (Gulf of Mexico) and in the Sargasso Sea, respectively.
4.4. Correcting Deep Water Mg/Ca Ratios for the Effect of Dissolution
Dependent on the foraminiferal species and varieties, a major and systematic decrease in Mg/Ca ratios starts below species-specific critical water depths, which are situated in the depth interval between 2000–3000 m. This is far above the calcite saturation horizon (CSH) at ∼4600 m (Figure 6), which approximates the present-day top of the lysocline at a Δ[CO32−] (difference between the in situ carbonate ion concentration ([CO32−]) and the [CO32−] at saturation) of 0 μmol/kg, and is equal to a calcite solubility ratio (Ω) of 1 [Lewis and Wallace, 1998; Tyrrell and Zeebe, 2004]. In fact, below >3000 m, Mg/Ca decreases linearly with increasing water depth (r2 = 0.55–0.80; Figure 2; Table 4). This close relationship is also reflected in the linear decrease of Mg/Ca with decreasing Δ[CO32−] (r2 = 0.32–0.94; Figure 5; Table 4) below 20 μmol/kg (Figure 7; Table 4).
Figure 6. Δ[CO32−], defined as the difference between the in situ carbonate ion concentration ([CO32−]) and [CO32−] at saturation, versus water depth. Total alkalinity (TA) and TCO2 data necessary to calculate the in situ [CO32−] were obtained from the World Ocean Circulation Experiment (WOCE) Line numbers A22 and A22_2003A (Caribbean stations 1–38 and 2–37, respectively; available at http://whpo.ucsd.edu/index.htm; Figure 1). The in situ [CO32−] was calculated using the program of Lewis and Wallace  developed for CO2 System Calculations. [CO32−] at saturation was calculated after Jansen et al. . Gray crosses mark computed Δ[CO32−] data; black dots indicate Δ[CO32−] data averaged for every 20 m water depth. CSH is calcite saturation horizon (dashed line) as the approximate top of the lysocline.
Download figure to PowerPoint
Figure 7. Mg/Ca ratios versus Δ[CO32−] for each planktonic foraminiferal species and variety indicating significant Mg2+ loss below species-specific Δ[CO32−] levels (Δcritical = horizontal lines). Δcritical is defined by the intercept between the vertical lines (species-specific mean Mg/Ca calculated from samples <2000 m water depth, similar to Δ[CO32−] > 40 μmol/kg (Table 2)) and diagonal lines (regression lines from Mg/Ca at Δ[CO32−] < 20 μmol/kg). Deepest core-top sample M35024–6 from below the CSH was defined as an outlier and excluded from the calculation of the regressions.
Download figure to PowerPoint
Table 4. Species-Specific Parameters for Equations (2)–(7)a
|Species||d Correction, Equation (6)||Δ Correction, Equation (7)|
|G. ruber p.||9,403||1,381||0.79||2,631.1||−44.08||14.03||0.84||24.67|
|G. ruber w.||9,793||1,382||0.76||2,535.3||−40.93||12.61||0.71||25.28|
|G. truncatulinoides d.||6,456||1,303||0.56||3,018.8||−16.23||13.79||0.74||20.17|
|G. truncatulinoides s.||7,471||1,888||0.65||2,746.4||−27.43||21.28||0.94||25.77|
As only few Mg/Ca ratios cover the interval between 2000–3000 m (40–20 μmol/kg Δ[CO32−]), the onset of dissolution was determined by calculating the intersection between the unaffected Mg/Ca means from <2000 m (>40 μmol/kg) and the regressions through the dissolution-affected ratios from >3000 m (<20 μmol/kg). The decrease of Mg/Ca can be described as
where ΔMg/Ca is the difference between the (unknown) initial, dissolution-unaffected Mg/Ca (Mg/Cainitial) and the measured Mg/Ca (Mg/Cameasured). A reasonable approximation for Mg/Cainitial is the mean Mg/Ca (Mg/Camean) from <2000 m (Table 2). The intersection of the Mg/Camean with the regression line calculated for Mg/Ca from >3000 m water depth and <20 μmol/kg Δ[CO32−], respectively, provides the critical water depth (dcritical in m) and the critical calcite saturation state (Δcritical in μmol/kg), where Mg2+-removal due to dissolution starts:
where a is the y axis intercept, and b is the slope of the regression lines (Figures 2 and 7; Table 4). The levels of beginning preferential dissolution of Mg2+ appear to be species-specific. Δcritical spans values of ∼18–26 μmol/kg (Table 4) which correspond to Ω values of ∼1.5, accompanied by dcritical of ∼2500–3000 m (Table 4).
Different susceptibility of calcitic tests from different planktonic foraminiferal species to dissolution is a well-known feature [e.g., Berger, 1968, 1970, 1971], likewise is the heterogeneous distribution of Mg2+ in foraminiferal tests [Bender et al., 1975; Duckworth, 1977; Nürnberg, 1995; Brown and Elderfield, 1996; Nürnberg et al., 1996a; Rosenthal et al., 2000; Eggins et al., 2003; McKenna and Prell, 2004; Anand and Elderfield, 2005; Bentov and Erez, 2005, 2006]. Sadekov et al.  recently illustrated by electron microprobe mapping that low-Mg/Ca (∼1–5 mmol/mol) and high-Mg/Ca bands (∼8–11 mmol/mol) alternate within tests of planktonic foraminifera. As Mg2+ has a pronounced effect on the stability of biogenic calcite [Walter and Morse, 1985], and as ion activity products increase with increasing Mg2+, Brown and Elderfield  deduced that Mg-enriched phases are less dissolution resistant. The selective removal of Mg2+-enriched foraminiferal parts with increasing dissolution should thus cause the reduction of the intraspecific Mg/Ca scatter, which is observed in samples from >3000 m water depth in comparison to shallow samples from <2000 m. This reduction amounts to 36% for G. ruber white, 50% for G. ruber pink, 44% for G. sacculifer, 46% for G. menardii, 27% for N. dutertrei, 14% for G. tumida, 36% for G. truncatulinoides dextral, while G. crassaformis shows no discernable change.
Nonetheless, the slopes b of the regression lines indicate a relatively uniform ∼0.5–0.8 mmol/mol change in Mg/Ca ratios per kilometer water depth and ∼0.04–0.11 mmol/mol per 1 μmol/kg decrease in Δ[CO32−] with respect to Mg/Camean for all studied foraminiferal species and varieties (Table 4). In contrast to observations of Savin and Douglas , Lorens et al. , Brown and Elderfield , and Rosenthal et al. , the relatively similar decline of Mg/Ca in combination with broadly the same dcritical and Δcritical (Figures 2 and 7; Table 4) for all foraminifers studied does not support the notion that calcite of shallow-dwelling foraminifers with higher Mg/Ca dissolves preferentially with respect to deep-dwelling species bearing lower Mg/Ca.
The absolute change in Mg/Ca ratios of ∼0.5–0.8 mmol/mol per kilometer water depth is equivalent to a relative decrease of ∼11–29% per kilometer. Our results support previous core-top studies showing a decrease of Mg/Ca with increasing water depth [Savin and Douglas, 1973; Bender et al., 1975; Rosenthal and Boyle, 1993]. For G. sacculifer, Lorens et al.  showed a decrease in Mg/Ca of ∼40% in tropical East Pacific Rise and central Pacific samples over a depth range from ∼600–4000 m. Both Lorens et al.  and Brown and Elderfield  detected a decrease of 40% in Mg/Ca of G. tumida over a depth range from ∼1900–4700 m, while the decline in Mg/Ca of N. dutertrei amounts to ∼59% from ∼600–3800 m, and ∼25% for G. ruber from ∼1900–4700 m [Lorens et al., 1977], all from tropical Pacific samples. Assuming a linear Mg/Ca descent with depth, these numbers would convert to Mg/Ca reductions of ∼12% for G. sacculifer, ∼14% for G. tumida, ∼18% for N. dutertrei, and ∼9% for G. ruber per kilometer water depth. At Ontong Java Plateau, Lea et al.  showed a decrease in Mg/Ca in tests of G. ruber by ∼12% per kilometer water depth, while Dekens et al.  revealed a decrease of ∼14% per kilometer. At Sierra Leone Rise and Ceara Rise, a decrease in Mg/Ca of ∼7% and ∼5% per kilometer, respectively, was observed for G. ruber [Dekens et al., 2002]. For N. dutertrei, Dekens et al.  found a Mg/Ca decline of up to ∼23% per kilometer.
Our results on decreasing Mg/Ca ratios below species-specific Δ[CO32−] levels of ∼18–26 μmol/kg are in general accordance with Dekens et al. , who also noted the onset of Mg2+ loss below ∼20 μmol/kg Δ[CO32−] at Ceara Rise (Atlantic) and Ontong Java Plateau (Pacific). In the South China Sea, Whitko et al.  found declining Mg/Ca in core-top G. ruber below ∼2000 m water depth, implying a strong dissolution effect below ∼2000 m. This is considerably shallower than both, the present-day lysocline and the carbonate compensation depth in this ocean area at ∼3000 m and ∼3800 m, respectively [Miao et al., 1994; Feely et al., 2002, 2004]. Although we do not have Δ[CO32−] values for this ocean basin at hand, we speculate that the onset of Mg/Ca change observed by Whitko et al.  might coincide with the approximate threshold in Δ[CO32−] of ∼20 μmol/kg. It needs to be proven, though, whether this threshold is globally valid.
In order to correct the measured Mg/Ca ratios for the selective loss of Mg2+ due to dissolution, ΔMg/Ca was calculated for every measured Mg/Ca from below dcritical and Δcritical:
where d is the water depth of the core-top sample in meter (with d > dcritical) (Tables 1 and 2), and Δ is the averaged Δ[CO32−] in μmol/kg at the seafloor (with Δ < Δcritical) (Figure 6). We then combined equations (1) and (4), as well as (1) and (5) to recalculate the dissolution-unaffected Mg/Cainitial (water depth correction (d-corrected) according to equation (4) (Figure 8); and Δcritical correction (Δ-corrected) according to equation (5)):
Figure 8. Foraminiferal Mg/Ca ratios versus water depth (black dots): Mg/Ca from below dcritical were corrected for water-depth induced calcite dissolution according to equation (6). For comparison, the measured, noncorrected Mg/Ca data are included (circles). The application of the correction routine leads to a reduction of the intraspecific Mg/Ca variability by 24–64%.
Download figure to PowerPoint
Both independently developed routines for the correction of dissolution-induced Mg/Ca decline produce comparable results indicating that selective Mg2+ removal is mainly driven by the calcite saturation state (Figure 9). Application of the correction routines reduces the overall intraspecific variability of Mg/Ca by ∼24–64% (Figure 8).
Figure 9. Depth-corrected (d-corrected, equation (6)) versus Δ[CO32−]-corrected (Δ-corrected, equation (7)) Mg/Ca ratios: The independently corrected ratios from below dcritical and Δcritical highly correlate (r2 = 0.97). Related high probability values (p = 0.77) imply that the corrected Mg/Ca means of all species and varieties do not statistically deviate from each other, implying that the selective removal of Mg2+ from foraminiferal calcite is mainly a function of the calcite saturation state Δ[CO32−].
Download figure to PowerPoint
4.5. Implications for Paleorecords
We chose the Mg/Ca records of G. ruber white of two Caribbean cores [Schmidt et al., 2004] to test the downcore applicability of the proposed correction routine. Today, both records chart similar temperature conditions within the Columbian Basin (Figure 1) and hence are expected to show similar Mg/Ca ratios. The impact of dissolution on the Holocene (3–6 kyrs) mean Mg/Ca from shallower ODP Site 999A (2827 m, 4.13 mmol/mol) appears to be less than on Mg/Ca from deeper site VM22–128 (3623 m, 3.75 mmol/mol), differing by 0.38 mmol/mol. Depth correction (equation (6); parameters see Table 4) leads to higher Holocene Mg/Ca of 4.34 mmol/mol at ODP Site 999A and 4.53 mmol/mol at site VM28–122, reducing the offset to 0.19 mmol/mol. Such inter-site offset converts to a ∼0.5°C difference (calculated after Anand et al. ; see Table 2), which is in accordance with the modern temperature pattern in the Columbian Basin.