SEARCH

SEARCH BY CITATION

Keywords:

  • amino acid racemization;
  • Quaternary geochronology;
  • Arctic Ocean;
  • planktonic foraminifera

Abstract

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Material and Methods
  5. 3. Existing Stratigraphic Context of AAR Samples
  6. 4. AAR Results
  7. 5. Discussion
  8. 6. Conclusions and Implications
  9. Acknowledgments
  10. References
  11. Supporting Information

[1] The long-term rate of racemization for amino acids preserved in planktonic foraminifera was determined by using independently dated sediment cores from the Arctic Ocean. The racemization rates for aspartic acid (Asp) and glutamic acid (Glu) in the common taxon, Neogloboquadrina pachyderma, were calibrated for the last 150 ka using 14C ages and the emerging Quaternary chronostratigraphy of Arctic Ocean sediments. An analysis of errors indicates realistic age uncertainties of about ±12% for Asp and ±17% for Glu. Fifty individual tests are sufficient to analyze multiple subsamples, identify outliers, and derive robust sample mean values. The new age equation can be applied to verify and refine age models for sediment cores elsewhere in the Arctic Ocean, a critical region for understanding the dynamics of global climate change.

1. Introduction

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Material and Methods
  5. 3. Existing Stratigraphic Context of AAR Samples
  6. 4. AAR Results
  7. 5. Discussion
  8. 6. Conclusions and Implications
  9. Acknowledgments
  10. References
  11. Supporting Information

[2] Climate change in the Arctic commands increasing attention as unprecedented shrinkage of sea ice has surpassed the range of historical observations and is likely heading toward a seasonally ice-free Arctic Ocean [e.g., Comiso et al., 2008]. Foreseeing the consequences of this fundamental shift in the state of the Arctic system requires an understanding of past changes in Arctic climatic and oceanic conditions. Meanwhile, acquisition and interpretation of paleoceanographic data in the Arctic face many difficulties. In addition to logistical challenges, studies of Arctic Ocean sedimentary records are hampered by generally low sedimentation rates, limited biogenic proxies due to low productivity and high dissolution, and nonanalogous conditions compared with other better investigated oceanic regions. Despite recent progress in some areas [Jakobsson et al., 2000b; Spielhagen et al., 2004; O'Regan et al., 2008], no regional stratigraphic standard for central Arctic Ocean sediments has been well established, and many aspects of their age interpretation are questionable [e.g., Backman et al., 2004; Sellen et al., 2008]. A reliable age model is necessary to place the paleoenvironmental history of the Arctic into a global context, and this requires the advent of new chronostratigraphic approaches.

[3] Considerable progress has been made in the development of age models for Quaternary sediment from the Arctic Ocean have been made since the new stratigraphic paradigm was introduced for cores from the Lomonosov Ridge [Jakobsson et al., 2000b], and has been expanded across the central Arctic Ocean [Backman et al., 2004; Spielhagen et al., 2004]. The new approach recognized the complexities of paleomagnetic records, which had been the staple of Arctic Ocean stratigraphy, and used a more realistic assessment of sedimentation rates and placement of paleoclimatic events than in earlier studies [e.g., Clark et al., 1980; Phillips and Grantz, 1997]. Recently, this new age model has been shown to be consistent with longer-term Cenozoic stratigraphy developed for the first deep-drilling record from the central Arctic Ocean [O'Regan et al., 2008]. Nevertheless, many aspects of age assignment within this general stratigraphic framework remain unclear and cannot be resolved by methods that have been applied to date [O'Regan et al., 2008; Cronin et al., 2008]. Even more problems arise with stratigraphies from other areas of the central Arctic Ocean [e.g., Sellen et al., 2008]. In this paper we advance the use of amino acid racemization in planktonic foraminifers as applied to upper Quaternary sediments from the Arctic Ocean.

[4] The extent of amino acid racemization (AAR) in fossil foraminifera has been used previously to estimate the ages of Quaternary marine sediments (Kaufman [2006] reviews previous AAR studies on foraminifera). Deep marine settings are ideal for amino acid geochronology because of the long-term thermal and geochemical stability of the depositional environment. Foraminifera inhabit most of the world's ocean and they contain high concentrations of amino acids that are well retained by their carbonate test. Sample size requirements for AAR analysis of foraminifera using reverse phase chromatography (RPC) are in the submilligram range. Preparing samples composed of a small number of tests is less time-consuming than picking larger samples such as for 14C dating or for AAR analysis by the chromatographic techniques used previously, and it can improve the accuracy of the results because the best preserved individuals can be selected, and because collections of microfossils commonly include a few individuals whose extent of racemization values fall well outside the mean of the others. These outliers might result from postdepositional reworking, postcollection contamination, or an aberrant diagenetic pathway. Regardless of the cause, by routinely analyzing multiple subsamples from a single stratigraphic level, these tests can be identified objectively and excluded from the data set.

[5] Two general approaches are used to convert the extent of AAR to a numeric timescale: In the first approach, the effects of time and temperature on the extent of racemization are determined in modern shells subjected to high-temperature laboratory experiments [e.g., Kaufman, 2006]. This relation, together with a model of racemization kinetics, is used to calculate the age of a sample if its temperature history is known. A more secure approach that does not require assumptions about temperature history is to calibrate the rate of racemization by analyzing securely dated samples of a particular taxon from a region where temperature histories are uniform [e.g., Hearty et al., 2004]. The calibrated reaction rate is then used to date samples of the same taxon of unknown age from the same region.

[6] The two previous attempts to use AAR in the Arctic Ocean [Sejrup et al., 1984; Macko and Aksu, 1986] applied the calibration approach. Both studies relied on only two independently dated samples from the North Atlantic to calibrate the rate of isoleucine epimerization in the foraminifera, Neogloboquadrina pachyderma. Because the temperature of bottom water in the North Atlantic is about 4°C higher than in the Arctic Ocean, assumptions were needed to adjust the reaction rate for the Arctic Ocean. Both studies also relied on only a single sample from each stratigraphic level; we find that replication is important for recognizing outliers. Furthermore, both studies were based on isoleucine, which reacts more slowly and therefore has lower temporal resolution than the amino acids that we use in this study. Because of these shortcomings, results of the previous AAR studies in the Arctic Ocean were difficult to interpret conclusively, and during the last 20 years, the approach was not further developed for use in the Arctic Ocean.

[7] In this study, we take advantage of recent developments in the chronostratigraphic framework of Arctic Ocean sediments to calibrate the rate of racemization using control samples from within the Arctic Ocean itself. We focus on aspartic acid, which affords enhanced age resolution where ambient temperature is low [Goodfriend et al., 1996]. We use these data to develop an AAR age equation for late Quaternary Neogloboquadrina pachyderma from the Arctic Ocean, which is applicable to samples as old as 150 ka.

2. Material and Methods

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Material and Methods
  5. 3. Existing Stratigraphic Context of AAR Samples
  6. 4. AAR Results
  7. 5. Discussion
  8. 6. Conclusions and Implications
  9. Acknowledgments
  10. References
  11. Supporting Information

[8] Samples for this study are from three long (piston) sediment cores from three major submarine ridges extending into the central Arctic Ocean: Lomonosov, Mendeleev, and Northwind ridges, plus two multicore samples from the Mendeleev Ridge (Table 1 and Figures 1 and 2) . Cores HLY0503–8JPC, −10MC and −11MC were collected on the 2005 HealyOden TransArctic Expedition (HOTRAX) [Darby et al., 2005]; the data in this paper from these HOTRAX cores have not been presented previously. Litholostratigraphic, paleomagnetic, and microfaunal data from core 88AR-P5 have been published by Poore et al. [1993, 1994]; core 96/12–1PC was studied by Jakobsson et al. [2000b, 2001] and used for correlation by O'Regan et al. [2008] as part of the development of the new stratigraphic model for the central Arctic Ocean.

image

Figure 1. Arctic Ocean with major geographic features and circulation systems, Beaufort Gyre (solid lines) and Transpolar Drift (punctured). Cores 96/12–1PC, HLY0503–8JPC, and 88AR-P5 are shown by stars on the Lomonosov, Mendeleev, and Northwind ridges (LR, MR, and NR), respectively; multicores are shown by circles. Base map from International Bathymetric Chart of the Arctic Ocean [Jakobsson et al., 2000a].

Download figure to PowerPoint

image

Figure 2. Correlation among sediment cores used in this study, showing the main lithologic features, planktonic foraminiferal abundance (black curves), and paleomagnetic inclination (gray curves) (for cores 96/12-PC and 88AR-P5, see Jakobsson et al. [2000b, 2001] and Poore et al. [1993, 1994], respectively). Note that paleomagnetic measurements are much more sparse in 88AR-P5 than in the other two cores. Solid tie line: natural remanent magnetization inclination shift. Ba and Ee (dashed tie lines): abundance maxima or unique presence events for benthic foraminifers Bulimina aculeata and Epistominella exigua, respectively. Dotted tie lines: major detrital carbonate layers. Triangles: position of samples analyzed for amino acid racemization. Assignment of marine isotopic stages (MIS) is based on the age model developed for the central Lomonosov Ridge [Jakobsson et al., 2000b; O'Regan et al., 2008]. Ages of MIS boundaries [Martinson et al., 1987] are shown in italics.

Download figure to PowerPoint

Table 1. Sediment Cores Analyzed for This Study
CoreaVessel and YearCoordinatesWater Depth (m)RegionSource of Data Used in Figure 2
  • a

    MC, multicores; PC, piston cores.

HLY0503–8JPCHealy, 200579°35.6′N 172°30.1′W2792Mendeleev RidgeThis paper
HLY0503-10MC-2Healy, 200581°13.7′N 177°13.0′W1841Mendeleev RidgeThis paper
HLY0503-11MC-2Healy, 200583°07.7′N 174°41.6′W2570Mendeleev RidgeThis paper
96/12–1PCOden, 199687°05.9′N 144°46.4′E1003Lomonosov RidgeJakobsson et al. [2000b, 2001]
88AR-P5Polar Star, 198874°37.35′N 157°53.04′W1089Northwind RidgePoore et al. [1993, 1994]

[9] U-channel samples from core HLY0503–8JPC were taken for paleomagnetic analysis. Natural remanent magnetization (NRM) was measured at 1 cm spacing after demagnetization at 14 steps in the 10–100 mT peak field range. Component magnetizations were determined using the standard 3-D least squares method [Kirschvink, 1980]. The demagnetization range used for calculation of the component directions was nonuniform down core, and depended on demagnetization behavior of each measurement position, but was usually in the 30–80 mT interval. Maximum angular deviation (MAD) values, associated with the characteristic magnetization components are generally less than 10° for the upper ∼420 cm of the core, indicating that the components are well defined, and then vary widely in the 5–30° range below that level, indicating that the uncertainty in the direction of the characteristic component is highly variable. The onset of negative inclinations at ∼420 cm bsf (cm below seafloor) is, however, well defined (Figure 2), although the control on its position (lithological versus geomagnetic) has yet to be determined.

[10] Samples for foraminifera and other analyses from core HLY0503–8JPC were taken continuously at 1 cm intervals to 500 cm bsf, freeze dried, washed, and separated by several size fractions. Foraminifers were counted in the >150 μm fraction and picked for AAR analysis from relative abundance peaks throughout the washed portion of the core and, concurrently, for 14C determinations from its upper part. Samples for 14C analyses contained 5 to 8 mg of planktonic foraminifers; no pretreatment was applied.

[11] The rate of amino acid racemization is taxon dependent and in this study we focus on monospecific samples of left-coiled Neogloboquadrina pachyderma, the predominant Arctic planktonic foraminifer. Approximately 50 tests of adult (encrusted) N. pachyderma were picked from intervals with peak foraminifer abundance. They were cleaned by sonicating repeatedly at 1 min increments until the bath water was clear, then rinsed three times in deionized water. The tests were immersed in 1 mL of 3% H2O2 for 2 hours, rinsed with purified H2O, then air dried under laminar flow. The best preserved, most thoroughly cleaned, whole tests were then picked and divided into subsamples composed of 3 to 10 individuals (average = 8), which were placed in sterilized, conical bottomed microreaction vials and dissolved in 7 μL of 6 M HCl, and sealed under N2. To recover the total hydrolysable amino acid population while minimizing the induced racemization, the solutions were hydrolyzed at 110°C for 6 hours. Subsamples were evaporated to dryness, rehydrated in 4 μL of 0.01 M HCl with 10 μM L-h Arg, an internal spike used to quantify the abundance of amino acids, and injected onto a high-performance liquid chromoatograph (HPLC).

[12] The chromatographic instrumentation and procedure used to separate amino acid enantiomers were presented by Kaufman and Manley [1998], with modifications for individual microfossils described by Kaufman [2000, 2003]. Briefly, the analytical method employed precolumn derivatization with o-phthaldialdehyde (OPA) together with the chiral thiol, N-isobutyryl-L-cysteine (IBLC), to yield fluorescent diastereomeric derivatives of chiral primary amino acids. The derivatization was performed online prior to each injection using the auto-injector of an integrated Agilent 1100 or 1200 HPLC. Separation was by a reverse-phase column packed with a C18 stationary phase (Hypersil BDS, 5 μm) using a linear gradient of aqueous sodium acetate, methanol and acetonitrile. Detection was by fluorescence.

[13] The extent of AAR was measured by the ratio of D- to L-enantiomers. The average analytical uncertainty (internal reproducibility) measured by the coefficient of variation for multiple injections of laboratory standards is typically between 2 and 5% for the D/L values of most of the amino acids. Both intralaboratory and interlaboratory comparative samples [Wehmiller, 1984] were analyzed routinely to monitor machine performance. For this study, we focused on aspartic acid (Asp) and glutamic acid (Glu), two amino acids that are among the most abundant in foraminifera protein, and are the best resolved chromatographically. More accurately, the Asp and Glu may include a small component of asparagine and glutamine, respectively, which were converted to Asp and Glu during laboratory hydrolysis. Asp and Glu elute during the first 30 min of the sample run, and span most of the range of racemization rates. Serine (Ser) also elutes during this interval and is useful index of contamination, rather than for geochronology. The procedural blank is dominated by Ser, and the blank accounts for an average of 3% and 6% of L-Asp and L-Glu, respectively. The influence of the blank is insignificant at low D/L values, but increases in older samples as the concentration of indigenous amino acids decrease. No attempt was made to adjust D/L values for the blank.

3. Existing Stratigraphic Context of AAR Samples

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Material and Methods
  5. 3. Existing Stratigraphic Context of AAR Samples
  6. 4. AAR Results
  7. 5. Discussion
  8. 6. Conclusions and Implications
  9. Acknowledgments
  10. References
  11. Supporting Information

3.1. Sediment Core Correlations

[14] The stratigraphy of the cores in this study was tied to the age model developed for the central Lomonosov Ridge [Jakobsson et al., 2000b, 2001; O'Regan et al., 2008]. For correlation we use lithological characteristics, foraminiferal abundances, NRM inclinations, and benthic foraminiferal events (Figure 2). Together with new 14C data, this stratigraphy enables the ages of each AAR sample to be estimated independently.

[15] Quaternary sediments in the central Arctic Ocean are generally composed of gray to yellow-brown, nearly abiotic mud alternating with brown, faunal-rich beds interpreted to represent a succession of glaciations and interglacial/interstadial periods, respectively [Darby et al., 2006, and references therein; O'Regan et al., 2008; Cronin et al., 2008]. This cyclic sedimentation pattern provides the basic framework for stratigraphic correlations across the Arctic Ocean; however, it cannot be used as the sole basis for correlation because of varying sedimentation rates and possible spatial differences in sedimentary environments and preservation of paleontological material. Some additional lithologic features provide more detailed correlation, notably layers of detrital carbonate that were delivered by icebergs from the Canadian Arctic Archipelago during major collapses or surges of the Laurentide ice sheet and dispersed by the Beaufort Gyre [Bischof and Darby, 1997; Phillips and Grantz, 2001]; however, these features do not occur in the Eurasia Basin and most of the Lomonosov Ridge including core 96/12–1PC.

[16] Paleomagnetic properties, particularly inclination records have been commonly used for stratigraphic correlations of Arctic Ocean sediment [e.g., Steuerwald et al., 1968; Clark et al., 1980; Backman et al., 2004; Spielhagen et al., 2004]. The paleomagnetic inclination of sediments from the central Arctic probably records not only variations in paleomagnetic field, but also changes in rock-magnetic composition, and therefore cannot be used as a simple chronostratigraphic tool [Jakobsson et al., 2000b; O'Regan et al., 2008]. Nevertheless, inclination records exhibit an apparently consistent pattern in various regions of the Arctic Ocean and therefore provide an independent means of correlation, regardless of their genesis [e.g., Spielhagen et al., 2004; Sellen et al., 2008]. In particular, the prominent change from predominantly high positive inclination values to strongly variable including negative values (solid tie line in Figure 2), previously interpreted as the Brunhes Matuyama polarity boundary, is now used as a chronostratigraphic marker in the middle of marine isotope stage (MIS) 7, based on the current age model [Jakobsson et al., 2000b; O'Regan et al., 2008].

[17] The abundance of planktonic foraminifers shows a spiky distribution with maxima in brown units in the upper part of all three cores (Figure 2). The absolute abundance of tests differs between the cores, probably because of different degrees of carbonate dissolution, but relative down core changes have a consistent pattern. Below the stratigraphic level assigned to MIS 7, sediments throughout the Arctic Ocean are mostly barren of calcareous fossils; instead, arenaceous foraminifers become common and their abundance varies with glacial-interglacial cycles [Jakobsson et al., 2001; Backman et al., 2004; Cronin et al., 2008]. This level probably marks the general weakening of carbonate dissolution in the deep Arctic Ocean and can be used as an additional stratigraphic marker.

[18] The chronostratigraphic framework is further constructed using benthic foraminiferal assemblages that contain unique components: species that are otherwise absent or sparse in other stratigraphic levels. These foraminiferal events have been described for the cores in this study or from adjacent sites [Poore et al., 1994; Ishman et al., 1996; Jakobsson et al., 2001; Polyak et al., 2004]. Among the most prominent events are the abundance maximum of Bulimina aculeata, and the occurrence of Epistominella exigua (dashed tie lines in Figure 2). Neither species lives in the central Arctic Ocean today, evidently not adapted to perennial ice conditions. Regardless of what caused their previous expansions into the High Arctic, such species occurrences are valuable stratigraphic markers. These events correspond with MIS 5a and 5e, potentially the warmest, low-ice intervals of the late Quaternary [Nørgaard-Pedersen et al., 2007].

3.2. Age Control

[19] Eleven samples from the upper 63 cm in HLY0503–8JPC and two additional samples from the northern part of the Mendeleev Ridge (HLY0503-MC10 and -MC11) were dated by AMS 14C for this study (Table 2). An additional 14C age for the top sample in core 88AR-P5 was taken from Poore et al. [1994]. Because of uncertainties in the reservoir age in the Arctic Ocean, we have not applied a ΔR to age calibration. Adult N. pachyderma in the central Arctic Ocean live mostly in the lower halocline [e.g., Volkmann, 2000], which is ventilated rapidly, on the order of 10 years in modern conditions [Ekwurzel et al., 2001]. The influence of riverine inputs from Siberia on ΔR in this water cannot be estimated conclusively; limited data from the Laptev Sea indicate a ΔR value between –100 and 460 years [Bauch et al., 2001]. In the past, especially during discharges of large volumes of glacial meltwater, reservoir ages may have been greater, so the 14C ages from these periods should be considered maximum ages. As a guideline, we assume that intervals containing planktonic foraminifera with high values of δ13C, which faithfully correspond to brown units [Poore et al., 1999; Polyak et al., 2004], were associated with high ventilation rates. Other complications in the 14C chronology might stem from bioturbation of overall thin interglacial/interstadial units and the input of old carbonaceous water during the discharges of Laurentide detrital carbonates. The latter possibility has not yet been investigated, but potentially can bias 14C ages in the Amerasia Basin, especially at sites proximal to the Canadian Arctic. Generally, 14C ages in HLY0503–8JPC show an orderly succession, consistent with ages obtained earlier from the Mendeleev Ridge [Darby et al., 1997; Polyak et al., 2004]. Some pre-Holocene ages, however, are older than expected by a few thousand years, possibly reflecting the influence of old carbonaceous water, and one sample appears too young, possibly resulting from bioturbation.

Table 2. Radiocarbon Ages of Planktonic Foraminifers Used to Calibrate Rate of Racemizationa
Lab IdentificationHLY0503 Core IdentificationDepth in Core (cm)14C Age (years B.P.)1 σ (±)Calibrated Age (years B.P.)b1 σ Range (±)
  • a

    All samples were analyzed by AMS at the NSF Radiocarbon Facility, University of Arizona.

  • b

    Holocene ages are median probability ages based on calibration using CALIB v 5.0.2 (data are available at http://calib.qub.ac.uk/calib/); older ages were calibrated using the Fairbanks0107 program [Fairbanks et al., 2005]. No ΔR was applied.

  • c

    Age rejected because it is considerably younger than others above and below it.

AA744708JPC0–1398146398472
AA744718JPC7–893147610,151103
AA744728JPC14–15894063960879
AA744738JPC24–2599175010,857127
AA778878JPC32–3336.25073041,118681
AA744758JPC34–3532,48047037,467502
AA744768JPC43–4438,66098043,218863
AA744778JPC48–4930,95040035,926c404
AA744788JPC52–53>43,300nanana
AA744798JPC56–5737,82088042,493781
AA744808JPC62–6340,500130044,8161149
AA7584510MC210–1131,76043036,731457
AA7584711MC25–626,14022030,949270

[20] Below the range of 14C dating, sample ages are based on correlation with global MIS record (the orbitally tuned, staked oxygen isotope record typically referred to as SPECMAP [Martinson et al., 1987]), as first proposed by Jakobsson et al. [2000b] for core 96/12–1PC, and further developed on correlative cores, including the IODP core from the Lomonosov Ridge [Spielhagen et al., 2004; O'Regan et al., 2008]. The general validity of this age model for the upper Quaternary is corroborated for example by correlation of units containing abundant coarse ice-rafted debris and related meltwater indicators with major ice advances along the margins of the Arctic Ocean [Spielhagen et al., 2004], and the presence of subpolar foraminifers at levels corresponding to the warmest intervals, MIS 5a and 5e [Nørgaard-Pedersen et al., 2007]. The resulting age assignment of samples is coarse, but it provides an internally consistent stratigraphic framework suitable for a first-order calibration of the rate of AAR. It is reasonable to assume that the age error for our calibration samples is typically less than one half of the duration of a marine isotope substage, that is, within ±10%.

4. AAR Results

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Material and Methods
  5. 3. Existing Stratigraphic Context of AAR Samples
  6. 4. AAR Results
  7. 5. Discussion
  8. 6. Conclusions and Implications
  9. Acknowledgments
  10. References
  11. Supporting Information

[21] A total of 47 samples of N. pachyderma were analyzed for this study. Each sample comprised 2 to 9 subsamples (average = 5; total = 249) (see auxiliary material). Some subsamples yielded D/L values that were clear outliers compared with others from the same sample (Figure 3). The cause of the variation is not clear. Outliers might reflect reworking of individual tests and are useful for understanding time averaging and age structure in the fossil record [e.g., Kosnik and Kaufman, 2008]. In this study, the goal is to derive a robust estimate of the central tendency of D/L values, and a reasonable estimate of the uncertainty of the mean D/L of a sample. Excluding outliers is therefore justified, provided that an objective and reproducible screening procedure is employed. We used a three-step screening procedure to systematically identify outliers. Rejected results are listed and the specific screening criterion applied is tracked for each subsample (auxiliary material). The criteria are:

image

Figure 3. Extent of racemization (D/L) in aspartic acid (Asp) and glutamic acid (Glu) in individual subsamples of Neogloboquadrina pachyderma from the Arctic Ocean. Subsamples rejected on the basis of three screening criteria are indicated. Data listed in auxiliary material.

Download figure to PowerPoint

[22] 1. The concentration of serine (Ser) was used to identify subsamples with aberrantly high levels of this labile amino acid, presumably resulting from contamination by modern amino acids. Subsamples with L-Ser/L-Asp > 0.8 were rejected (n = 30).

[23] 2. Next, the well-defined covariance between the D/L values of Glu and Asp was used to identify subsamples whose extent of racemization in this pair of amino acids deviated from the trend of other analyses in this study (n = 3).

[24] 3. Subsamples with D/L Asp or Glu that fell beyond ±2σ of the mean of the rest of the group, after screening by the two above criteria, were rejected, and this procedure was not iterated (n = 8). In all, 41 out of 249 subsamples were rejected, or 16% of the data.

[25] The rejected subsamples had D/L values that were generally lower than the other subsamples. Overall, the sample mean D/L values increased by an average of 0.030 for Asp and 0.012 for Glu, following screening for the 29 samples with one or more rejected subsamples. This is consistent with high L-Ser content of most of the rejected subsamples, which indicates contamination by modern amino acids. Overall, however, the abundance of Asp and Glu was not notably higher for rejected subsamples (Figure 4), suggesting that a small amount of contamination can significantly impact the D/L values. Although screening had little effect on the mean D/L values and down core trends, it significantly decreased the intrasample variability. Following data screening, the intrasample variability (CV) decreased from 21% to 5% for D/L Asp, and from 29% to 10% for D/L Glu. Overall, following screening, standard errors (σx) of mean D/L values average 2.4% for Asp and 5.0% for Glu, which can largely be attributed to analytical precision.

image

Figure 4. Amino acid abundance in subsamples of Neogloboquadrina pachyderma from the Arctic Ocean. Abundance values are the sum of aspartic acid (Asp), glutamic acid (Glu), and serine (Ser) per individual foraminifer. The mass of each test averages 8–10 μg. The abundance of amino acids decreases with increasing racemization (age) and is similar for accepted and rejected subsamples. Data listed in auxiliary material.

Download figure to PowerPoint

[26] The abundance of amino acids decreases with increasing extent of racemization (Figure 4) as expected, because amino acids degrade or are leached from the tests with time. The amino acid abundance values depend on the mass of the individual foraminifera tests, which varies among samples. On average, the mass of a single Neogloboquadrina pachyderma test analyzed in this study is between 8 and 10 μg. On this basis, the concentration of amino acids per mass of carbonate in N. pachyderma is similar to that of the larger planktonic foraminifera Pulleniatina, which as been analyzed extensively in sediment cores [Hearty et al., 2004] and laboratory heating experiments [Kaufman, 2006]. The amino acid content of N. pachyderma decreases more rapidly with increasing racemization extent than that of Pulleniatina, however, indicating a higher rate of amino acid attrition for N. pachyderma.

5. Discussion

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Material and Methods
  5. 3. Existing Stratigraphic Context of AAR Samples
  6. 4. AAR Results
  7. 5. Discussion
  8. 6. Conclusions and Implications
  9. Acknowledgments
  10. References
  11. Supporting Information

5.1. AAR Correlations Among Cores

[27] D/L values for both Asp and Glu increase regularly with depth down to about 2 m in all three long cores (Figure 5). The values then level off below this depth, corresponding to an age of about 150 ka. Some decrease in the rate of racemization with increasing age is expected as this reversible reaction approaches equilibrium. In the case of the N. pachyderma from the Arctic Ocean, however, it appears that D/L values reach a plateau at around 0.4 in Asp and 0.2 for Glu. This plateau is unlikely to have been caused by a shift from overall lower temperature prior to 150 ka to higher temperature since then, because site temperatures are currently near the freezing point. The D/L values in the two amino acids covary strongly in subsamples below this level, and none of the subsamples contain excessive Ser content (Figure 3), suggesting that the extent of racemization follows the expected diagenetic pathway. If so, then some of the samples below 2 m could be younger than we have assumed. On the other hand, when heated in the laboratory, planktonic foraminifera of the genus Pulleniatina show a sharp decrease in the rate of racemization at around the same D/L values, although D/L values continue to increase with heating time [Kaufman, 2006, Figure 4]. Determining whether the D/L values reach a plateau or whether they continue to increase with age and depth will require further analyses.

image

Figure 5. Extent of racemization (D/L) in aspartic acid (Asp) and glutamic acid (Glu) in samples of Neogloboquadrina pachyderma plotted according depth in five sediment cores from the Arctic Ocean. Tie lines connect samples from specific stratigraphic events discussed in the text and shown in Figure 2, including: inclin, paleomagnetic inclination shift; Ba and Ee, Bulimina aculeata and Epistominella exigua events, respectively; and dc, samples adjacent to the upper detrital carbonate layer. Data listed in Table 3.

Download figure to PowerPoint

[28] The two unique benthic foraminifera events and the shift in paleomagnetic characteristics (Figure 2) provide chronostratigraphic markers to compare D/L values among the three long cores (Figure 3). The D/L values for both Asp and Glu overlap within ±1σ (or ± 2σx) for the three samples from each of the marker horizons. The D/L values vary by 2 to 6% for Asp and 5 to 9% for Glu. Most of the differences in D/L values among correlative horizons in different cores can be ascribed to analytical errors and intrasample variability, mutually supporting the integrity of the stratigraphic correlations and the AAR results.

5.2. AAR Age Equation

[29] The chronostratigraphic ages and the 14C ages from all five cores were aggregated to develop a calibrated age equation that relates the extent of racemization in N. pachyderma to sample age (Table 3). Following the procedure discussed by Kosnik et al. [2008], we modeled the rate of racemization using a simple power transformation and chose the exponent that best linearized the trend in D/L versus sample age on the basis of least squares regression forced through the origin (Figure 6). Samples older than 150 ka could not be modeled using a power transformation and were excluded from the age model. Least squares linear regressions based on the remaining 38 samples yield the age equations for N. pachyderma from the Arctic Ocean for up to 150 ka. The age equation for Asp is:

  • equation image

and for Glu is

  • equation image

where t = age in ka.

image

Figure 6. Rate of racemization in (a) aspartic acid (Asp) and (b) glumatic acid (Glu) in independently dated samples of Neogloboquadrina pachyderma from the Arctic Ocean. Least squares linear regressions are based on samples <150 ka only. Samples shown in gray symbols were not included. Data listed in Table 3. Gray curve: rate of racemization in Pulleniatna from the Queensland Trough, Australia [Hearty et al., 2004].

Download figure to PowerPoint

Table 3. Extent of Racemization (D/L) in Foraminifera From Arctic Ocean Coresa
Lab Identification (UAL)Depth (cm)Aspartic AcidGlutamic AcidAgeb (ka)
nexD/L±σxD/L±σx
  • a

    n, number of subsamples included; ex, number excluded in calculation of mean and standard error; ILC, interlaboratory comparison.

  • b

    Ages with two decimals are 14C age (Table 2); other ages are estimated on the basis of stratigraphic correlations.

  • c

    Sample rejected because D/L values are considerably higher than any in the underlying 2 m of core.

  • d

    ILC samples [Wehmiller, 1984] analyzed during the course of this study.

88AR-P5
59211500.0980.0050.0500.0055.53
592055500.2480.0040.1110.00535
591889500.2880.0060.1390.00475
5922125410.3490.0120.1630.01585
5917180320.3800.0110.1690.006100
5923191410.4060.0040.2130.005105
5919233410.4040.0040.1940.003123
5916318410.3950.0110.2030.006193
5913365440.4080.0070.1940.004242
 
96/12–1PC
59260–1.9420.1540.0030.0880.00110
592718.3–20.2320.2670.0020.1180.00635
5928175.8–177.4610.3130.0080.1460.01179
5929198.4–200320.4040.0170.1940.01385
5930206.4–208.1310.4010.0140.2070.021100
5931219–220.5500.3920.0080.1840.010123
5932257.5–259410.4400.0160.2300.018193
5933275–276.6500.4630.0130.2480.018242
 
HLY0503–8JPC
61340–1410.0800.0030.0350.0003.98
61417–8400.0960.0040.0340.00310.15
614014–15210.1080.0020.0460.0049.61
613524–25310.1100.0020.0390.00110.86
613632–33400.1930.0070.0860.00541.12
614234–35600.1900.0060.0760.00637.47
613743–44510.2210.0040.1040.00643.22
614348–49410.2610.0050.1090.01135
613852–53600.2780.0030.1320.00438
614456–57410.2850.0100.1310.01442.49
613962–63600.2950.0040.1430.00444.85
638773–74410.2760.0090.1010.00752
638883–84410.2830.0060.1230.00459
638993–94200.3440.0070.1760.00870
6149105–106400.3090.0040.1540.00477
6390115–116410.3610.0080.1770.00880
6391132–133310.3550.0050.1760.00585
6392145–146200.3420.0130.1450.01389
6393c179–180400.4620.0060.2320.006100
6717180–181810.4050.0080.1970.006100
6394226–227320.4020.0070.1880.006123
6395238–239500.3950.0070.1770.007126
6396246–247320.3890.0050.1680.006129
6775282–283600.4110.0060.2010.010147
6718358–359630.4000.0100.1840.005178
6719404–405720.4190.0100.2040.008193
6720418–419810.4290.0100.2110.009196
6776428–429610.4410.0120.2190.016222
 
HLY0503-11MC-2
65105–6500.2250.0180.1060.01130.95
 
HLY0503-10MC-2
651110–11500.2790.0090.1370.00536.73
 
Interlaboratory Comparison Samplesd
ILC-Ana900.3720.0020.1840.001na
ILC-Bna900.6850.0020.4070.001na
ILC–Cna900.8890.0080.8390.001na

[30] The accuracy of ages derived from this procedure depends on an array of variables. A Monte Carlo simulation was used to quantify the overall age uncertainty in the AAR age model, and to assess its sensitivity to the input variables. The results of this simulation show that the accuracy of the AAR ages is primarily and linearly related to the uncertainty in the independent ages that are used to calibrate the rate of racemization. For each ±10% change in the uncertainty in the calibration ages, the corresponding change in the accuracy of AAR ages is about ±1.3%, assuming that the uncertainty in the D/L value (intrasample variability) is characterized by ±1σx. Using ±2σx to characterize the intrasample variability reduces the sensitivity of the AAR age accuracy to changes in the uncertainty associated with the calibration ages. At ±2σx the intrasample variability is the primary uncertainty in the overall age estimate, and the resulting accuracy of the AAR ages is about ±12% for Asp and about ±17% for Glu, regardless of whether the uncertainty in the calibration ages is 10% or 30%, for instance.

[31] Quantifying the uncertainty in the independent ages used to calibrate the rate of racemization is difficult. Instead, we rely on conventional regression statistics to estimate the uncertainty in the rate of racemization as derived from the combined data set (n = 38). The standard error of the slope of the least squares regression (rate of racemization) is ±54 ka (D/L)−2.3 for Asp, and ±136 ka (D/L)−1.6 for Glu. Combining this error in the slope with the typical ±2σx in mean D/L values determined from our samples (Asp = ±4.8%; Glu = ±10.0%) yields an overall realistic age error of about ±12% for Asp and ±17% for Glu for samples within the age range of the calibration equations (back to about 150 ka). This compares with standard error of the sample, based on the residuals, of 14 ka for Asp and 19 ka for Glu.

[32] Both the regression fit and the intrasample variability are better for Asp than for Glu, implying that Asp is a superior chronometer for these samples. For samples with D/L values higher than about 0.5, however, the rate of racemization for Asp decreases beyond that for Glu, suggesting that Glu might outperform Asp for older samples. Regardless, tracking D/L values for both amino acids is important because, although the rate of racemization in Asp and Glu are affected by the same systematic errors and are therefore not strictly independent, samples that yield similar ages for the two amino acids are viewed with greater confidence. Kosnik and Kaufman [2008] suggest that samples that yield ages for Asp and Glu that differ by more than 20% should be considered suspect. Otherwise, the ages can be averaged to generate a single age.

[33] These uncertainties are less than those estimated for the planktonic foraminifer, Pulleniatina, which are based on racemization kinetics analyzed in the laboratory (the first approach described above), with an assumed temperature uncertainty of ±0.5°C [Kaufman, 2006]. The enhanced age resolution for N. pachyderma can in part be attributed to its relatively high rate of racemization, which is about twice as high for samples from the Arctic Ocean than for Pulleniatina from the Queensland Trough, Australia [Hearty et al., 2004] (Figure 6), despite bottom water temperatures that are at least a few degrees lower in the deep Arctic Ocean (−1° to 0°C in the Arctic Ocean compared with 2°C for Queensland Trough). The higher rate of racemization in N. pachyderma seems inconsistent with its higher rate of amino acid attrition (discussed above) and suggests that, if amino acids are lost to leaching, they are not preferentially leached from the more highly racemized free amino acid fraction. We suggest that the proportion of amino acids within the intracrystalline fraction, which is dominated by free amino acids in carbonate fossils [Penkman et al., 2008] increases at a higher rate in N. pachyderma than in Pulleniatina, but this hypothesis awaits testing by analyzing the different amino acid fractions.

6. Conclusions and Implications

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Material and Methods
  5. 3. Existing Stratigraphic Context of AAR Samples
  6. 4. AAR Results
  7. 5. Discussion
  8. 6. Conclusions and Implications
  9. Acknowledgments
  10. References
  11. Supporting Information

[34] AAR values (D/L of aspartic acid and glutamic acid) analyzed in planktonic foraminifer Neogloboquadrina pachyderma show a consistent pattern of down core trends in three sediment cores from the central Arctic Ocean separated by hundreds of kilometers, which confirms the applicability of AAR geochronology to Arctic Ocean stratigraphy. Calibration of the rate of racemization by independent age assignments (14C ages and correlations with the global MIS record using the Lomonosov Ridge stratigraphy as reference) shows a convincing correlation between D/L values and age, at least for the late Quaternary. These results indicate that AAR, which can be measured on small numbers of foraminifers, provides a useful tool for stratigraphic correlation and late Quaternary geochronology in the Arctic Ocean. To confidently correlate stratigraphic sequences with reference records, such as in this paper, requires multiple stratigraphic techniques performed at relatively high sample resolution, which may not be feasible for many sites. The AAR approach offers a fast and cost efficient method for checking the age of target intervals identified by lithostratigraphic correlation across the Arctic Ocean, even in low-resolution cores. As a stratigraphic tool, the technique is unencumbered by uncertainties related to the 14C reservoir age, and is applicable beyond the range of 14C dating. Sediment cores from the Nordic seas might provide suitable material for refining and extending the AAR calibration for the Arctic Ocean because bottom water temperatures in the Nordic seas are similar to those in the Arctic Ocean (near −1°C), and the stratigraphy for the last 500 ka is well developed and correlated with North Atlantic records [e.g., Bauch and Erlenkeuser, 2003; Helmke and Bauch, 2003].

[35] The late Quaternary time interval is especially important for paleoclimatic studies as it encompasses the last interglacial, potentially the closest analog for the projected future, low-ice or even seasonally ice-free state of the Arctic [Otto-Bliesner et al., 2006]. To date, the last interglacial in the Arctic has been characterized mostly on the basis of sites along the Arctic periphery [CAPE-Last Interglacial Project Members, 2006], and without a reliable stratigraphic framework for the identification of the last interglacial in sedimentary records from the central Arctic Ocean. The AAR geochronology developed here offers a new possibility for resolving this problem with a reasonable accuracy for sediment cores that contain at least small numbers of planktonic foraminifera.

Acknowledgments

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Material and Methods
  5. 3. Existing Stratigraphic Context of AAR Samples
  6. 4. AAR Results
  7. 5. Discussion
  8. 6. Conclusions and Implications
  9. Acknowledgments
  10. References
  11. Supporting Information

[36] HOTRAX'05 coring expedition was supported by the NSF-OPP award ARC-0352359/0352395. Collection and processing of cores 96/12–1PC and 88AR-P5 was supported by the Swedish Research Council (award to J. Backman) and the U.S. Geological Survey, respectively. T Cronin and an anonymous reviewer suggested improvements on the paper. J Bright analyzed the samples in the Amino Acid Geochronology Laboratory at Northern Arizona University, which is supported by NSF grant EAR-0620455 to D.S.K. This study was also supported by NSF-OPP awards ARC-0520505 and 0612473 to LP.

References

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Material and Methods
  5. 3. Existing Stratigraphic Context of AAR Samples
  6. 4. AAR Results
  7. 5. Discussion
  8. 6. Conclusions and Implications
  9. Acknowledgments
  10. References
  11. Supporting Information

Supporting Information

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Material and Methods
  5. 3. Existing Stratigraphic Context of AAR Samples
  6. 4. AAR Results
  7. 5. Discussion
  8. 6. Conclusions and Implications
  9. Acknowledgments
  10. References
  11. Supporting Information

Auxiliary material for this article contains the primary data for each subsample of foraminifera analyzed for this study.

Auxiliary material files may require downloading to a local drive depending on platform, browser, configuration, and size. To open auxiliary materials in a browser, click on the label. To download, Right-click and select “Save Target As…” (PC) or CTRL-click and select “Download Link to Disk” (Mac).

See Plugins for a list of applications and supported file formats.

Additional file information is provided in the readme.txt.

FilenameFormatSizeDescription
palo1509-sup-0001-readme.txtplain text document2Kreadme.txt
palo1509-sup-0002-ts01.txtplain text document16KTable S1. Laboratory identification number (UAL), core identification, core depth (cm), number of individual tests (n), criterion used if the result was rejected, aspartic acid D/L, glutamic acid D/L, serine D/L, L-Ser/L-Asp, concentration of aspartic acid (pM per foraminifer), concentration of glutamic acid (pM per foraminifer), and concentration of serine (pM per foraminifer).
palo1509-sup-0003-t01.txtplain text document1KTab-delimited Table 1.
palo1509-sup-0004-t02.txtplain text document1KTab-delimited Table 2.
palo1509-sup-0005-t03.txtplain text document3KTab-delimited Table 3.

Please note: Wiley Blackwell is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.