Magnesium/calcium (Mg/Ca) records based on N. pachyderma (dex.) and N. pachyderma (sin.) from the last 1200 years have been retrieved from high-resolution cores in the Norwegian Sea. Comparing temperatures inferred from Mg/Ca and the oxygen isotopic composition of calcite (δ18OC) with instrumental temperature records from Ocean Weather Ship Mike suggests that N. pachyderma calcifies during the summer season with an offset to oxygen isotope equilibrium of 0.6‰. In this region, summer temperatures below the thermocline are related to the winter season ventilation and breakdown of the seasonal thermocline; hence the deeper dwelling N. pachyderma (sin.) provides a winter season signal. Down-core N. pachyderma (dex.) Mg/Ca temperatures display larger variance than observed in δ18OC temperatures record derived from the same morphotype. The smaller variance in the δ18OCN. pachyderma (dex.) temperatures is probably linked to salinity changes in the upper 50 m of the water column. The Mg/Ca temperature and δ18OC records are used to reconstruct changes in the oxygen isotopic composition of water (δ18Ow) during the last 1200 years. The reconstructed δ18OW records show variation within a realistic range, allowing the influence of other water masses than those present at the site today, and suggest that the Mg/Ca signal in morphotypes of N. pachyderma in the Norwegian Sea responds to changes in climatic parameters.
The aim of this study is to test the fidelity of Mg/Ca in planktonic foraminifera as a palaeoclimatic proxy in high latitude regions. We present down-core Mg/Ca records from two planktonic foraminifera, N. pachyderma (dex.) and N. pachyderma (sin.), and compare them with paired δ18OC records of the same species generated by Andersson et al. . Two sediment cores from a high-latitude site in the Norwegian Sea covering the last 1200 years were analyzed. The site is located underneath the Norwegian Atlantic current, which brings warm, saline Atlantic water (T > 3°C, S > 35‰ [Swift, 1986]) northward toward the Arctic. Speculations about calcification habitat are constrained by comparing geochemical data from box core surface sediments with instrumental observations from Ocean Weather Station Mike (OWSM) [Østerhus et al., 1996] located in the Norwegian Sea (Figure 1). The advantage of Mg/Ca as an independent proxy for temperature is that it may be used together with oxygen isotopes in calcite (δ18OC) to reconstruct changes in the oxygen isotope composition of seawater (δ18OW), a parameter that correlates with salinity. This study shows overall consistency in the reconstructed δ18OW curves, indicating that Mg/Ca changes in N. pachyderma (both sin. and dex.) from high latitudes represent a climate signal.
2. Material and Age Control
The two cores (MD95-2011 and JM97-948/2A) were recovered from the same location on the Vøring Plateau in the Eastern Norwegian Sea (66°N, 7°E) at a water depth of 1048 m (Figure 1). Box core JM97-948/2A was taken to recover surface sediments and represents the last 564 years. MD95-2011 is 17.49 m long, however, only the upper 90 cm were examined in this study. Together with the box core this represents the last 1200 years. The age-depth models for both cores are adopted from Andersson et al. . The age-depth model for the box core was reconstructed on the basis of nine 210Pb dates from the uppermost 10 cm and two Accelerator Mass Spectrometry (AMS) 14C dates. The age-depth model for the upper 90 cm in MD 95-2011 rests on five AMS 14C dates. The 14C dates were calibrated using ΔR = 0 and converted to calendar ages using Calib 4.3 [Stuvier et al., 1998].
The site has a high sedimentation rate: 53 cm/kyr on average in the box core, whereas the piston core suggests a sedimentation rate of 147 cm/kyr on average. Piston core sedimentation rates are normally expected to be less than in the box cores due to increased compaction of overlaying sediment. However, this appeared to not be the case and the suggested age-depth model shows an increased and unrealistic sedimentation rate between 10–50 cm (AD 600–800) of 268 cm/kyr. This is caused by the coring technique which stretched the sediment in the piston core and is a persistent problem with older Marion Dufresne cores [Széréméta et al., 2004]. This is unlikely to affect the reliability of the results assuming that the sediment was not mixed. The box and piston cores were sampled every 0.5 and 1.0 cm, respectively. According to the age-depth model developed for the box core, the most recent N. pachyderma (sin.) Mg/Ca measurement is about 1975 AD. Since several data points from the uppermost 26 cm in the Mg/Ca N. pachyderma (dex.) record, including the core top, were rejected, the most recent Mg/Ca sample in this record is found at 6.5 cm and is dated to about 1937 AD. Almost all individuals in the samples were required for single analyses of both Mg/Ca and oxygen isotopes; thus our ability to analyze replicate samples was limited.
3.1. Mg/Ca Analyses
The samples were cleaned before trace element analysis to eliminate contamination by magnesium from surrounding sediments. The sample preparation for Mg/Ca analysis follows the cleaning protocol of Barker et al. , which is a modified version of the cleaning protocol for cadmium analysis in biogenic calcite of Boyle  and Boyle and Keigwin . Around 20–30 individuals were used for the Mg/Ca analyses. Tests in samples from the box core and the upper 26 cm in MD95-2011 were picked from the 150–250 μm fraction. Individuals in samples from 27–89 cm (1337-791 AD) in MD95-2011 were picked from a limited fraction range (150–212 μm) since a large range in test size results in larger Mg/Ca variability [Anand et al., 2003]. The analyzed foraminifera show no visible sign of dissolution, which is in agreement with observation of well-preserved carbonate tests found along the pathway of inflowing Atlantic water in a study by Huber et al. .
All cleaning preparation and Mg/Ca analyses were performed at the Department of Earth Sciences, University of Cambridge, UK. The samples were analyzed using a Varian Vista ICP-AES axial viewed instrument [de Villiers et al., 2002]. Measurements of a standard solution of Mg/Ca (5.13 mmol/mol) run in parallel had an analytical error of <0.011 mmol/mol (1σ). At low Ca concentrations the content of magnesium may rise to unrealistic levels due to the blank effect. Samples with Ca concentration below 30 ppm were therefore excluded as were samples enriched in Fe/Ca (above 0.1 mmol/mol) and/or Al/Ca since these elements are likely to be found in contaminants that may also contain high amounts of magnesium [Barker et al., 2003].
28 duplicate Mg/Ca N. pachyderma (sin.) measurements were performed on surface and LGM samples from the Nordic Seas [Meland et al., 2006]. A pooled standard deviation of the residuals with 56 degrees of freedom indicates reproducibility within ±0.08 mmol/mol. This equals to an error of ±0.8°C for Mg/Ca in N. pachyderma (sin.) within the concentration range used in this study according to Elderfield and Ganssen  and is adopted in this study as a general estimate of the reproducibility of each sample. The Mg/Ca temperature calibration of Elderfield and Ganssen  was used to convert Mg/Ca concentrations in N. pachyderma (dex. and sin.) to temperatures (see section 5.2).
3.2. Oxygen Isotope Measurements
Paired oxygen isotope measurements on the same samples were generated previously [Andersson et al., 2003]; therefore Mg/Ca data were not generated on the same homogenized samples used for δ18O. The measurements were performed at the Geological Mass Spectrometer (GMS) laboratory at the University of Bergen, using Finnigan MAT 251 and MAT 252 mass spectrometers both equipped with automatic preparation lines. Long-term replicate measurements of a carbonate standard gave a reproducibility of ±0.07‰ for the oxygen isotope measurements. Duplicate δ18O measurements of 39 core top samples of N. pachyderma (sin.) have been reported elsewhere and show a modified standard deviation of ±0.13‰ (0.5°C) [Meland et al., 2006]. The δ18O results are reported in ‰ versus VPDB, calibrated against NBS-19 and NBS-18. A vital effect of 0.6‰ was added to the isotopic measurements before the palaeotemperature equation of O'Neil et al. , as cited by Shackleton , was applied to calculate temperatures from the oxygen isotopic measurements (see section 5.3).
Changes in the local oxygen isotopic composition in seawater (δ18OW) were derived from δ18OC measurements (‰) and Mg/Ca temperatures (T°C) by rearranging the palaeotemperature equation of O'Neil et al.  and Shackleton  (equation (1)). A factor of 0.27 was used to convert the calcite value from the PeeDee Belemnite (PDB) to the Standard Mean Ocean Water scale (SMOW).
4. Down-Core Results
4.1. Oxygen Isotopes
The δ18OC values from right- and left-coiling N. pachyderma for the last 1200 years have averages of 1.34‰ and 2.02‰ (Figure 2, Table 1) and a similar variance of 0.23‰ and 0.28‰, respectively. The records are in general visually strongly related. A cooling of ∼2°C (∼0.5‰) from 1275–1400 AD is synchronous in the δ18OC records for both morphotypes. The heaviest isotopic values are seen between 1400 and 1900 AD, suggesting colder conditions in the period. The isotopic variability in N. pachyderma (sin.) is less in this period compared to an earlier period (800–1400 AD). A discrete cooling event at 1900 AD is observed in both records, which is more pronounced in the δ18OC record of N. pachyderma (dex).
Table 1. Summary of δ18Oc and Mg/Ca in N. pachyderma (dex.) and N. pachyderma (sin.) From JM97-948/2A and MD95-2011a
The Mg/Ca records in N. pachyderma (dex.) and N. pachyderma (sin.) have average values of 1.17 mmol/mol (1σ = 0.116) and 0.94 mmol/mol (1σ = 0.087), respectively (Figure 3, Table 1). The N. pachyderma (dex.) Mg/Ca results show two periods at ∼900–1025 and 1125–1175 with temperatures about 0.5–1°C warmer than the 20th century (Figure 4a). These warm intervals are apparent in the oxygen isotope record of N. pachyderma (dex.) (Figure 4a). Between ∼1325–1450 AD temperatures are generally warmer than the subsequent period, but are interrupted by colder spells. Cooling of ∼2°C in the Mg/Ca temperatures of N. pachyderma (dex.) at 1050 and ∼1400 AD is mirrored in the oxygen isotope-derived temperatures, but not at 1200 AD. The N. pachyderma (sin.) Mg/Ca record exhibits weaker warming and cooling trends between 920–1050 AD. At 1360 AD the Mg/Ca concentrations drop suddenly, the most prominent temperature minimum in the record, whereas the δ18OCN. pachyderma (sin.) indicate a smaller temperature decrease. Both records show a following increase in temperature. The smoothed temperatures derived from Mg/Ca of N. pachyderma (dex.) show variance twice as high as those deduced from N. pachyderma (sin.); 0.57 and 0.25, respectively. The visual correlation between the smoothed Mg/Ca curves of the two morphotypes changes after ∼1500 AD: there is a negative correlation from 850–1300 AD but a more positive correlation from 1500 and 1900 AD. A cooling event seen in both Mg/Ca records at 1900 AD and the observed subsequent warming are also synchronous with the oxygen isotope records.
5.1. Constraints on Calcification Habitat
Bauch et al.  suggests that the identification of N. pachyderma (sin.) based on morphologic distinction is not suitable when the percentage of N. pachyderma (dex.) is less than 5%. On the basis of abundance data from Andersson et al.  (40–70% N. pachyderma (dex.) and 10–30% N. pachyderma (sin.) through the last 3000 years) the N. pachyderma morphotypes in our site are distinguishable and questions regarding calcification habitat are therefore discussed separately for each of the morphotypes.
Studies suggest that N. pachyderma (sin.) calcifies within the mixed layer and sinks below the pycnocline to reproduce and gains secondary calcite [Kohfeld et al., 1996; Simstich et al., 2003]. This modifies the shell's geochemistry such that the end product reflects colder temperatures, giving an apparent calcification depth of between 100–200 m depending on the depth of the pycnocline. The depth of the pycnocline in the Norwegian Sea is within the upper 100 m [National Oceanographic Data Center, 1998]. It is generally assumed that N. pachyderma (dex.) also migrates through the water column during reproduction.
Temperatures for the last 50 years inferred from N. pachyderma (dex.) oxygen isotopes are ∼8.5°C (±0.5°C) on average after adjustment for the vital effect (section 5.3), while temperatures derived from oxygen isotopes and Mg/Ca in N. pachyderma (sin.), indicate temperatures around 6°C (±0.5°C) and 7°C (±0.8°C), respectively (Figures 4a and 4b). Calcification habitat can be further constrained when these temperatures are compared to instrumental data from OWSM (Figure 1). Water with temperatures within the range of reconstructed temperatures from N. pachyderma (dex.) is found at the surface in spring and at around 50 m during summer (Figure 5a). An apparent calcification at the sea surface is regarded as implausible since N. pachyderma (dex.) is believed to dwell in sub-surface water, not at the surface [Pak and Kennett, 2002; Berstad et al., 2003; Kuroyangi and Kawahata, 2004, von Langen et al., 2005]. Therefore we suggest that the reconstructed δ18OC temperatures indicate calcification depths at ∼50 m for N. pachyderma (dex.). Proxy temperatures derived from N. pachyderma (sin.) indicate an apparent calcification depth at 100 m or deeper. These findings are in agreement with previous studies of N. pachyderma (dex.) and N. pachyderma (sin.) in the Norwegian Sea [Berstad et al., 2003; Simstich et al., 2003] and of N. pachyderma (sin.) in the polar and northern arctic water masses [Kohfeld et al., 1996; Bauch et al., 1997]. The uncertainties in the methods result in a large range in calcification depth especially of N. pachyderma (sin.) because the temperature gradient is relatively small below the seasonal thermocline within the Atlantic water mass. Also, the season in which N. pachyderma (sin.) calcifies is difficult to determine with our data since the temperature differences between the seasons at the supposed calcification depth are smaller than the error of the methods. However, sediment trap studies reveal only one major production bloom in summer (July August-September) in the Nordic Seas [Simstich et al., 2003, and references therein], which is in agreement with the findings of maximum flux for N. pachyderma (sin.) between August and September in Arctic water [Kohfeld et al., 1996]. We therefore assume that the calcification of N. pachyderma (sin.) occurs in summer.
5.2. Reconstruction of Past Temperatures: Mg/Ca Calibration
Several studies suggest an exponential correlation between N. pachyderma Mg/Ca concentration ratio and temperature [e.g., Nürnberg, 1995; Elderfield and Ganssen, 2000, von Langen et al., 2005]. The appropriate choice of which calibration to use is crucial when reconstructing past temperatures. The selection of calibration is based on the most recent N. pachyderma (sin.) Mg/Ca samples since these can be linked to the instrumental temperature record, whereas no N. pachyderma (dex.) Mg/Ca results are within in this 50-year time window. To decide which temperature calibration to apply, the reconstructed temperatures calculated according to the multispecies calibration approach of Elderfield and Ganssen  (equation (2a)) and the species-specific calibration based on N. pachyderma (sin.) of Nürnberg  (equation (2b)) were compared to instrumental data from OWSM.
Different equations have been proposed to convert δ18O in calcite into temperatures. In this study we have used the O'Neil et al.  and Shackleton  palaeotemperature equation as it is based on benthic foraminifers dwelling in colder water rather than Bemis et al. , which are based on planktonic foraminifera dwelling in warmer water (15–25°C). The equation used is widely applied in the determination of palaeotemperatures from high-latitude planktonic species.
Modern δ18OW data collected in the upper 300 m of the Norwegian Sea (5°W–7°E, 60°–70°N, S > 35‰, T > 3°C) show an average of 0.3‰, which is used when calculating isotopic temperatures [Schmidt et al., 1999]. The δ18OC temperatures of N. pachyderma (dex.) and N. pachyderma (sin.) ranges between 8–13°C (mean = 11°C) and 6–11°C (mean = 8°C) respectively. The mean temperature derived from N. pachyderma (dex.) would suggest a calcification depth at the sea surface and is in the same range as temperatures derived from diatoms and alkenones representing sea surface conditions [Calvo et al., 2002; Jansen and Koç, 2000]. We find these temperature estimates to be highly unlikely, as explained in section 5.1. The δ18OC temperatures deviate by ∼2–3°C from the Mg/Ca temperature records of the same morphotypes (Figures 4a and 4b). Deviation between Mg/Ca and δ18OCN. pachyderma (dex.) derived temperatures has also been found in a study on sediment trap samples from the Santa Barbara Basin by Pak et al. . The weak correlation in their study was linked to the δ18OC temperature calibration equation, seasonal or interannual changes in δ18OW, inaccuracies in the Mg temperature equation, or secondary effects on either Mg/Ca or δ18OC.
Assuming Mg/Ca derived temperatures represent true values, the reason for deviation of 2–3°C between the two proxies can lie in the reconstructed δ18OC temperatures as the morphotypes of N. pachyderma build their calcite shells in disequilibrium with predicted inorganic δ18O (δ18OEq). It has been hypothesized that the disequilibrium is linked to postgametogenetic processes as planktonic foraminiferas continue to calcify when they sink through the water column [Bemis et al., 1998, and references therein]. The δ18OC may be further modified by selective dissolution since enriched gametogenetic calcite is more resistant than primary calcite [Bemis et al., 1998, and references therein].
Predicted inorganic δ18OEq are estimated from Mg/Ca temperatures inferred from six N. pachyderma (sin.) samples coincident with the instrumental record so that modern δ18OW can be used in the palaeotemperature equation of O'Neil et al.  and Shackleton  (Table 2). The median offset between predicted and measured δ18OC is 0.59‰ (n = 6). Two of the values are unrealistically low because Mg/Ca is high, despite all concentrations of potential contaminating elements being within threshold values. When these samples are excluded the average disequilibrium is 0.6‰. The δ18OCN. pachyderma (sin.) temperatures found in this study were thus estimated with an adjustment for the vital effect of 0.6‰ (Figure 4b). The same vital effect of 0.6‰ was adopted for calculating δ18OC temperatures from N. pachyderma (dex.), because it was not possible to calculate a specific δ18OEq as its Mg/Ca record does not contain samples comparable to the instrumental record. The N. pachyderma (dex.) δ18OC temperatures show an offset to the δ18O temperatures from N. pachyderma (sin.) of 2–3°C as is the difference between the reconstructed Mg/Ca temperature curves. The resulting oxygen isotope temperatures lie in the same range as the Mg/Ca temperature record from N. pachyderma (dex.) and is therefore deemed acceptable (Figure 4). Simstich et al.  suggests an offset between δ18OC in N. pachyderma (sin.) and δ18OEq of 0.6‰ in surface samples from the Nordic Seas, after including a second correction for post depositional process related to bioturbational admixture of cold sub-fossiliferous foraminifers. This is in agreement with our estimate, though we do not believe that bioturbation is the cause. Also, a plankton tow study in polar regions found a similar disequilibrium correction (between 0.5–0.7‰) for N. pachyderma (sin.) [Stangeew, 2001]. However, other works from the Okhotsk Sea and within polar and northern arctic water masses indicate a higher disequilibrium of δ18OC in N. pachyderma (sin.) of about 1.0‰ [Bauch et al., 1997, 2002; Kohfeld et al., 1996, Volkmann and Mensch, 2001], while work on surface samples in Icelandic water found a lower vital effect of 0.25‰ [Smith et al., 2005] and studies on surface and LGM samples in Arctic water found that use of vital effect of 0.6‰ (from this study) when calculating temperature from δ18OCN. pachyderma (sin.) gave an unrealistically deep calcification depth [Meland et al., 2006]. Contradicting studies indicate that this issue is not well constrained with a need for further study.
Table 2. Estimated δ18O Vital Effect Calculated by Subtracting Predicted Inorganic δ18OEq From Measured in Shell δ18OC in Six Samples of N. pachyderma (sin.)a
The predicted δ18OEq is calculated from the palaeotemperature equation of O'Neil et al.  and Shackleton  with Mg/Ca temperatures and modern δ18Ow of 0.3‰. The samples used are all within the instrumental time period.
5.4. Difference in Variance in Down-Core Temperature Reconstructions
The two proxy records have similar variance in N. pachyderma (sin.) but dissimilar variance in N. pachyderma (dex.). In an attempt to resolve this, other Mg/Ca temperature calibrations were applied [Nürnberg, 1995; Anand et al., 2003], but they did not change the pattern of higher variability in the Mg/Ca temperatures of N. pachyderma (dex.). The variances in the Mg/Ca records are controlled by the temperature sensitivity that is expressed by the exponential constant of the equation (equation (2a)). By varying the exponential constant in the Mg/Ca equation we find that a value of 0.14 is needed to achieve a variance similar to that observed in the oxygen isotope temperature records (Figure 6). However, culture work on N. pachyderma (dex.) from the Santa Barbara basin shows a Mg/Ca temperature sensitivity of 0.1 [von Langen et al., 2005] and virtually all planktonic calibrations seem to converge on an exponential value of 0.09–0.10 [Elderfield and Ganssen, 2000; Anand et al., 2003]; thus the value of 0.14 is rather high. Also, calculations according to Skinner and Elderfield  suggest that the two morphotypes require a similar Mg/Ca calibration. The explanation may more likely lie in the lower variance of the δ18OC temperatures of N. pachyderma (dex.).
The oxygen isotopic temperatures are based on biogenic δ18OC which is affected by δ18OW. Modern data show that δ18OW in the study area (60–70°N, 5°E–2°W, S > 35‰, upper 300 mwd) range from 0.25–0.4‰ [Schmidt et al., 1999]. This is equivalent to a ΔT value of 0.6°C when a fixed δ18OC value of 1.4‰ is used in the palaeotemperature equation of O'Neil et al.  and Shackleton . This temperature uncertainty plays a critical role when considering high-resolution data over a narrow temperature range. Salinity and δ18OW are correlated in the Nordic Seas, since salinity changes are caused by precipitation, river runoff or glacial meltwater [e.g., Bauch et al., 1995], whereby most sources of low salinity originate from generally cold water masses with reduced δ18O. Thus changes in δ18OC temperatures associated with movement of a thermal front may result in reduced calculated temperature levels relative to Mg/Ca temperatures if a change in δ18OW is not accounted for within the palaeotemperature equation. This indicates that the variability between the Mg/Ca and oxygen isotope temperature reconstructions of N. pachyderma (dex.) may, in part, be explained by changes in δ18OW.
5.5. N. pachyderma (sin.): A Winter Signal
Differences in patterns between species, as observed in the δ18OC and Mg/Ca records in this study, have been used by others as an argument that the two species calcify in different seasons [Berstad et al., 2003]. Figure 5b show that the summer temperatures at 0 and 100 m at Ocean Weather Ship Mike during the last 50 years display different patterns. This difference is caused by stratification of the water column during summer, so that the lower boundary of the mixed layer is less than 100 m. When atmospheric warming ceases during winter and turbulence is enhanced due to increased wind stress, the stratified structure is lost and the base of the mixed layer deepens to 250–280 m [Nilsen and Falck, 2006] causing the whole layer to have a similar temperature. At 100 m the temperature stays approximately constant throughout the year (Figure 5b). As N. pachyderma (sin.) is thought to calcify around 100 m, where the mean temperature is a remnant of the wintertime mixing, their shells would reflect the winter situation even though the calcification occurs during the summer season. Multiproxy results from the same cores (JM98-947/2A and MD95-2011) support this theory. δ18OC in N. pachyderma (dex. and sin.) show a different pattern than the SST reconstructions and have a trend that resembles the winter insolation at 65°N through the Holocene, showing a late Holocene temperature maximum [Risebrobakken et al., 2003], while surface proxies such as alkenones and diatoms are correlated with summer insolation and show an early Holocene temperature optimum [Birks and Koç, 2002; Calvo et al., 2002]. This suggests that the climate signal derived from the deeper dweller N. pachyderm (sin.) may be an indicator for wintertime conditions.
A temperature difference of ∼2°C between the two proxy records of N. pachyderma (dex.) from 1150–1350 AD may be explained by changes in δ18OW within the upper 50 m of the water column (Figure 4a, section 5.7). This is also evident in the difference between the two morphotypes (offset curve) within each proxy (ΔMg/Ca and Δδ18O, Figure 7). The offset curves show in general a coherent picture, but larger amplitude in ΔMg/Ca than Δδ18O. The reduced temperature magnitude in Δδ18O suggests influence of water mass with different δ18OW signature. However, the overall similarity of the Δδ18O with the ΔMg/Ca (°C) highlights the utility of the Mg/Ca records as a climate indicator (Figure 7).
5.7. Down-Core δ18OW Changes
Reconstructions of δ18OW can be used to validate the Mg/Ca method, if the combination of Mg/Ca temperatures and δ18OC values from the same levels give realistic δ18OW values. δ18OW correlates with salinity in seawater [Craig and Gordon, 1965]. Processes that change local salinity without affecting temperature (other than global ice volume, which is not important over the period studied) are evaporation and precipitation. Modern surface δ18OW values in Atlantic water near the core site vary between 0.25–0.4‰ [Schmidt et al., 1999]. In a geographically wider context, the range of modern δ18OW values increases due to the influence of low-salinity Polar water, containing variable amounts of δ18O-depleted water originating from rivers and glacier meltwater. In the upper 300 m of the water column in the area between 7°W and 7°E and 60° and 75°N, δ18OW varies from −1.13‰ to 0.63‰ (note that these data are not normally distributed) [Schmidt et al., 1999].
The reconstructed δ18OW values show an overall increase from ∼1250 until 1700 AD of 0.7‰ for N. pachyderma (dex.) and 0.8‰ for N. pachyderma (sin.) following a decrease of 0.9‰ and 0.7‰ from 900 to ∼1250 AD (Figure 8). The two δ18OW records display minima at 1250 AD for N. pachyderma (dex.) and 1350 AD for N. pachyderma (sin.). During the LIA δ18OW values are generally higher than earlier. The high reconstructed δ18OW values during the LIA are surprising. Colder termperatures are normally associated with lower salinity and lower δ18OW. Therefore the expected result is that colder temperatures during the LIA would be accompanied by lower δ18OW. The higher δ18OW would indicate that the surface layer was less stratified in the LIA interval than before and was less influenced by low-salinity waters. The reconstructed δ18OW curves differ in detail, but the overall consistency is encouraging for Mg/Ca in N. pachyderma (dex.) and N. pachyderma (sin.) as climate proxies.
The reconstructed δ18OW ranges between −0.2‰ and 0.7‰ (σ = 0.37‰) (Figure 8). This range is toward the upper limit of what is reasonable and would require the influence of other water masses than those present at the site today. Thus, converted to salinity and given a plausible range of salinity/oxygen isotope relationships, the reconstructed δ18OW values have an amplitude of variation that is quite high considering the location of the site within the Atlantic water mass, far from the coastal, low-salinity boundary currents on either side of the Nordic Seas. The possibility that some of the amplitude originates from imperfect knowledge of the Mg/Ca temperature calibration of these morphotypes remains an issue for further research.
However, studies have shown that neighboring Arctic water on the west side and Norwegian coastal water on the east side may at times stretch toward the site and influence the water hydrology bringing fresher water having a different δ18OW signature. Influence of Arctic water at the site has been suggested to influence the subsurface layer of Atlantic water during the Holocene [Risebrobakken et al., 2003]. Instrumental records from OWSM indicate that precipitation can indirectly influence OWSM when increased northerly winds produce an increased Ekman transport of fresh water away from the Norwegian coast and into the Norwegian Atlantic water [Nilsen and Falck, 2006]. A comparison of the lateral advection of modern North Atlantic inflow with the North Atlantic Oscillation (NAO) index showed that reduced strength of the westerlies led to a more westward position of the Atlantic waters [Orvik et al., 2001]. An increased atmospheric pressure gradient increases the strength of the westerlies, which would bring δ18OW-depleted Arctic water eastward over the site. This may explain the low δ18OW observed before the onset of LIA. However, higher δ18OW levels during LIA are in contradiction to findings by Nesje and Dahl  suggesting that rapid glacier advance in the early eighteenth century in southern Norway was caused by increased winter precipitation due to an NAO+ situation. Consequently, it remains unclear if the NAO, or other types of atmospheric circulation changes that influence the water column at the Vøring Plateau, can explain our reconstructed δ18OW.
Paired Mg/Ca and δ18OC measurements from N. pachyderma (dex.) and N. pachyderma (sin.) have been used to constrain the oxygen isotopic vital effect and calcification habitat. A vital effect in the N. pachydermaδ18OC of 0.6‰ is deduced from Mg/Ca N. pachyderma (sin.) reconstructed temperatures and the modern δ18OW of 0.3‰. The suggested vital effect is also a suitable estimate for calculating δ18OC temperatures from N. pachyderma (dex.). The data indicate that N. pachyderma (dex.) and N. pachyderma (sin.) calcify in summer at around 50 and 100 m or deeper, respectively. Signals derived from N. pachyderma (sin.) primarily reflect remnant winter conditions. The N. pachyderma (dex.) record suggests temperatures were 0.5–1°C warmer in the Medieval Warm Period than in the 20th century. A shift in the curve relation at ∼1500 AD is interpreted to represent the transition from the MWP with more stratified water mass then during LIA times. The variance in the Mg/Ca temperature record is greater for N. pachyderma (dex.) than N. pachyderma (sin.). In contrast, the δ18OC records have similar variances. This discrepancy may be due to problems with the Mg/Ca temperature calibration or, more likely, δ18OW changes. Reconstructed δ18OW records indicate that water masses with different δ18OW signature have influenced the site. Overall consistency in the morphotype offset curves derived from the Mg/Ca and δ18OC and consistency in the reconstructed δ18OW curves highlight the utility of Mg/Ca in N. pachyderma in the Norwegian Sea as a climate proxy.
We are grateful to Mervyn Greaves for his patient training and assistance during the Mg/Ca cleaning and analyses at the Department of Earth Sciences at the University of Cambridge, UK. Trond M. Dokken and Marius Meland are thanked for useful discussions. We thank Ulysses Ninnemann for helpful comments on an earlier draft of the manuscript and Cathy Jenks, who improved the English. The manuscript was improved by suggestions and comments from Thomas M. Marchitto, Pamela Martin, and two anonymous reviewers. Svein Østerhus is thanked for providing the OWSM instrumental data, and Richard Telford and Einar Hegaard are thanked for introduction to statistics and calculation of standard deviation in δ18Ow reconstruction. B.F.N. was supported by the Norwegian Research Council and the University of Bergen. This paper is a contribution from the CESOP and PACLIVA projects funded by the FP5 of the EU and the NOClim project of the Norwegian Research Council and NERC awards NE/C509531/1. This is publication A 134 from the Bjerknes Centre for Climate Research.