Research Article

Synchronized TerrestrialAtmospheric Deglacial Records Around the North Atlantic

See allHide authors and affiliations

Science  15 Nov 1996:
Vol. 274, Issue 5290, pp. 1155-1160
DOI: 10.1126/science.274.5290.1155


  • Fig. 1.

    (A) Map of northwest central Europe with the three Swedish lakes [Torreberga (ALT), Lake Madtjärn (MA), and Lake Mjällsjön (LM)], Lake Gosciaz (3) in central Poland, the ENAM 93-21 core (50), and the main find area for the YD-PB pine stumps (5, 6, 7) in southern Germany. The Scandinavian ice sheet margin in the late YD is also indicated. (B) The paleogeographic situation in south Scandinavia with the extent of the Baltic Ice Lake just before the YD-PB boundary (44), when the 25-m up-dammed lake still drained through the Danish-Swedish Strait. The pathway of the subsequent catastrophic drainage in south central Sweden is indicated by an arrow.

  • Fig. 2.

    (A) Δ14C as derived from the IntCal 93 (51) German oak and pine chronology (5) and additional oak data (HD). The IntCal 93 and German oak series are corrected for a 41-year gap. The German pine is matched to the absolutely dated oak chronology by aligning the long-term Δ14C trend. The uncertainty of this link is difficult to quantify, but the long-term slope of the 14C concentration allows for a range of ±80 years. (B) Standardized radiocarbon dates (◊) from the three lakes smoothed against the new 14C calibration set. Standardization was based on the known thickness of the YD in each lake, the calibrated age of the youngest 14C-dated levels, and the following assumptions: LC corresponds to the YD-PB boundary; sedimentation rates are more or less continuous during the YD and change at LC, where they are continuous up to the youngest PB dates; the YD is 1150 years long (3, 10, 11, 12); and sedimentation rates are standardized to 1 mm year−1. The solid line is the fast Fourier transform (FFT)-smoothed curve through the AMS dates versus standardized depth (cut-off length of 24 cm, equivalent to a low-pass filter of 240 years). Note the two 14C age plateaus at 10,000 and 9600 14C years B.P. and the rapid transition between the two plateaus.

  • Fig. 3.

    Pollen diagram from Lake Madtjärn with the most diagnostic AL-PB pollen types, 14C dates (Table 1), a sediment log, and the positions of LC, Allerød (AL), the YD, and the PBO. The outer pollen curve is a 10× exaggeration of the inner solid curve. The values at the bottom are percentages of total pollen from terriphytic spermatophytes (that is, the pollen sum at the right). Concentration represents the amount of terriphytic spermatophyte pollen in grains per cubic centimeter. The sediments are as follows: 900 to 835 cm, blackish blue-gray silty clay (marine) with some FeS coloring, upper boundary (UB) is very gradual; 835 to 832.5 cm, gray to beige gyttja clay, UB is rather gradual; 832.5 to 829.5 cm, greenish-gray clay gyttja with moss remains, UB is rather sharp; 829.5 to 821.5 cm, brown clay gyttja, rich in mosses, UB is rather sharp; 821.5 to 800 cm, brownish-gray clay gyttja with occasional mosses, UB is very gradual; 800 to 793 cm, brown clayey fine detritus gyttja, UB is very gradual; 793 to 737 cm, dark brown fine detritus gyttja.

  • Fig. 4.

    By a tentative 75-year reduction of the GRIP chronology (1950 A.D.) and the addition of 161 years to the dendrochronology, a close fit is achieved between the 5-year mean δ18O record from GRIP (upper curve) and the 5-year running average of the ring-width record of the German pines (lower curve) (6) at the YD-PB boundary in both records.

  • Fig. 6.

    Organic C content (reflecting aquatic production) in the early Preboreal lacustrine sediments (MA, Lake Madtjärn; LM, Lake Mjällsjön; ALT, Torreberga), as well as the content of detrital carbonates from Torreberga (dashed line), indicating soil erosion and allochthonous input related to depth below lake surface. Detrital carbonate content (DC) was obtained by a mass balance calculation (52). All values are related to percentage of dry weight. DC exhibits values of 70 to 80% in the YD before it rapidly decreases to 0% at LC (341 cm). The stratigraphic position of the PBO is marked by shading.

  • Fig. 5.

    A synchronized graph of different records between 13,000 and 10,800 years B.P. The δ18O values (per mil relative to SMOW) are 5-year mean values. Air temperatures are estimated from 50-year periods (12). Ring widths are the mean of all pines with a 15-year running average. Principal components analysis (PCA) scores represent a 15-year running average of the results from the first PCA axis (covering 97.3% of total variance). The analyzed data set consists of the measured 13C and 2H values together with the mean ring widths of the chemically analyzed tree rings without weighting or data transformation. Centering was done by variables on a covariance matrix. The Δ14C curve is based on the 14C-dated dendrochronology with the additional 161 years (Fig. 5); the data were low-pass filtered by FFT smoothing. The Δ14C values from Lake Madtjärn were calculated by aligning calendar ages for the AMS-dated levels 796 to 835 cm (Fig. 3 and Table 1); the age of LC was fixed at 11,450 calendar years B.P., and two constant sedimentation rates [0.28 mm year−1 (late AL-YD) and 0.3 mm year−1 (PB)] were used on the basis of the 1150-calendar year length of the YD (3, 10, 11, 12) and the calibrated calendar ages of the youngest 14C dates. The CO2 values [from (33)] were related to our chronology by assuming a constant sedimentation rate through the YD (33). The older values illustrate the general drop in the late Allerød.

  • Fig. 7.

    A box-diffusion model (31) on Δ14C variations (lower graph) and resulting 14C ages (upper graph) caused by changes in deep-ocean mixing and in the atmosphere-mixed layer exchange rate (upper trace), in the atmosphere (middle trace), and in the mixed layer (lower trace). The maximum reduction in the model parameter is 50% (from 4200 m2 year−1 and 1/6.9 years, respectively). Kdiff denotes the coefficient of vertical diffusive exchange, and KAM is the coefficient of exchange between the atmosphere and the mixed layer. The long-term trend in Δ14C, mainly attributed to an increase in the geomagnetic field strength, was ignored, and we thus only modeled the residual over the long-term trend.

  • Fig. 8.

    Five different curves showing the oscillating development between 15,000 and 11,000 calendar years B.P. The chronology was fixed by setting the YD-PB boundary to 11,450 years B.P. in all records. The Δ14C record only represents relative changes; the dashed part of the line indicates periods of uncertain changes. The δ18O record (per mil relative to SMOW) from the GRIP ice core (10, 11) is based on 100-year running mean values. The planktonic δ18O (per mil) record (solid line) and percentages of the polar Neogloboquadrina pachyderma sinistral (s.) foraminifera and IRD counts (dashed line) derive from core ENAM 93-21 (50) in the Faeroe-Shetland Channel (Fig. 1A) at a water depth of 1020 m in the gateway to the Nordic Seas. The Gerzensee/Killarney oscillation is seen in all the marine records and the AL-YD transition is characterized by a distinct δ18O depletion, interpreted as a freshwater spike, together with increased occurrence of N. pachyderma (s.) and a distinct IRD peak followed by relatively high IRD values throughout the YD. The onset of the PB is shown by abruptly decreasing N. pachyderma (s.) frequencies and another δ18O depletion, perhaps caused by a meltwater signal. The marine chronology is based on two 14C dates and the positions and ages of the Vedde and Saksunarvatn tephras (13, 32). The top curve depicts the strength of the North Atlantic conveyor on the basis of the combined terrestrial, ice-core, and marine records presented in this article. Across the top, the five most prominent deglacial cooling periods in the North Atlantic region between 15,000 and 11,000 years B.P. are shown.


  • Table 1.

    AMS dates (excluding nine dates with <1 mg C) and other records from the three lake sites. Abbreviations for macrofossils: B, Betula sp.; D, Dryas octopetala; Di, Distichium sp.; E, Empetrum nigrum; M, Menyanthes trifoliata; Ny, Nymphaea sp.; P, Polygonum viviparum; Pi, Pinus sylvestris; Po, Populus tremula; Pol, Polytrichum sp.; S, Salix sp.; Sc, Scirpus lacustris; So, Saxifraga oppositifolia; and W, undetermined wood. Calibrated ages [(51) with additional 161 years] are shown [without the often high σ values on the 10,000- to 9900-14C year plateau (5)] for dates around LC when the age is covered by the dendrochronology-based calibration. The AL-YD boundary in the Lake Madtjärn record is at 832.5 cm. The sedimentation rate in Torreberga is three times that in the other lakes. Climate proxy changes at LC are as follows: For Lake Madtjärn, relative increase of organic C, 28%; mean temperature rise in July and January [according to mutual climatic range analysis of beetles (53)], 9.3° and 18.2°C, respectively. For Lake Mjällsjön, relative increase of organic C, 25%. For Torreberga, relative increase of organic C and carbonate C, 108% and 116%, respectively; absolute changes in δ13C and δ18O of Candona ostracodes, −2.3 and +6.1 per mil, respectively.

    Depth (cm) for AMS dateDated macrofossils14C age B.P. (±σ)Calibrated age B.P.13C (per mil) of dated materialAMS sample number
    Lake Madtjärn (LC depth, 800 cm)
    773 to 777B + S9,670 ± 195−29.30Ua-10258
    777 to 781B + S9,490 ± 100−28.50Ua-10259
    781 to 785B + S + D9,490 ± 90−28.70Ua-10260
    785 to 789B + D + E + S9,550 ± 85−28.44Ua-10223
    789 to 791B + D9,395 ± 115−28.21Ua-10261
    791 to 793B + D + S9,935 ± 16511,190−28.02Ua-10262
    793 to 794B9,930 ± 13511,160−28.12Ua-10263
    794 to 796B + D + S9,995 ± 10011,330−29.22Ua-10264
    796 to 798B + D + S10,010 ± 11011,365−28.31Ua-10265
    798 to 800D + S9,990 ± 10511,330−27.81Ua-10266
    800 to 802.5B + D + S10,060 ± 10011,490−27.90Ua-10267
    802.5 to 806D + S10,185 ± 11512,080−28.02Ua-10268
    806 to 810D + S10,330 ± 125−28.38Ua-10269
    810 to 813.5D + S10,155 ± 8512,000−28.29Ua-10222
    813.5 to 814.5S10,455 ± 115−28.48Ua-10270
    814.5 to 818D + P + S10,375 ± 70−28.35Ua-10221
    818 to 821.5D + P + S10,425 ± 80−28.07Ua-10220
    821.5 to 824S10,340 ± 70−27.82Ua-10219
    824 to 827D + S10,440 ± 65−27.95Ua-10218
    827 to 829.5S10,400 ± 80−28.59Ua-10217
    829.5 to 831D + S + So10,625 ± 75−28.49Ua-10216
    831 to 832.5D + E + O + P + S + So10,625 ± 70−29.68Ua-10215
    832.5 to 835D + E + S + So10,995 ± 75−28.96Ua-10214
    835 to 845S10,820 ± 85−27.85Ua-10272
    835 to 845D + E + P11,065 ± 150−27.00Ua-10271
    845 to 855E + S10,750 ± 100−26.85Ua-10274
    875 to 885S11,630 ± 190−28.56Ua-10281
    885 to 895S11,470 ± 135−29.53Ua-10352
    Lake Mjällsjön (LC depth, 435 cm)
    413 to 416Pi9,375 ± 120−27.24Ua-4643
    416 to 418.5B + Pi9,565 ± 125−28.22Ua-4644
    421.5 to 424.5B9,910 ± 16511,160−27.99Ua-4646
    424.5 to 427B + Pi9,680 ± 13511,090−28.92Ua-4647
    427 to 429.5B + Po9,840 ± 16511,150−27.44Ua-4648
    431.5 to 433.5B + E + Po + S9,900 ± 11511,150−28.09Ua-4650
    437 to 439B + Pol + S10,085 ± 11511,640−28.70Ua-4654
    439 to 442Pol + S10,110 ± 13511,860−27.78Ua-4656
    442 to 445B + D + Di + Pol10,040 ± 16011,465−26.10Ua-4657
    Torreberga (LC depth, 341 cm)
    215 to 220M9,160 ± 80−26.58Ua-4468
    220 to 223M9,305 ± 85−26.70Ua-4467
    223 to 235M9,350 ± 90−26.96Ua-4466
    255 to 265B + M + Po9,210 ± 105−27.13Ua-4465
    274.5 to 280Ny9,525 ± 95−25.64Ua-4464
    290 to 300Sc9,805 ± 12511,150−27.78Ua-4463
    320 to 330.5B + Po + Sc9,890 ± 8511,150−29.24Ua-4462
    330.5 to 341B + S9,875 ± 11011,150−31.32Ua-4461
    341 to 360W10,145 ± 8511,970−29.77Ua-4460
    360 to 380B + D10,310 ± 180−30.26Ua-4458
    395 to 410B + D + S10,095 ± 9011,820−30.63Ua-4457
    410 to 425B + D10,365 ± 125−30.27Ua-4456
    425 to 450B + D + S10,510 ± 235−31.29Ua-4455

Stay Connected to Science