Sediment instability on the Portuguese continental margin under abrupt glacial climate changes (last 60 kyr)

lable at ScienceDirect

Quaternary Science Reviews 28 (2009) 3211–3223

Contents lists avai

Quaternary Science Reviews

journal homepage: www.elsevier .com/locate/quascirev

Sediment instability on the Portuguese continental margin under abrupt glacialclimate changes (last 60 kyr)

S.M. Lebreiro a,*, A.H.L. Voelker a, A. Vizcaino b, F.G. Abrantes a, U. Alt-Epping c,S. Jung d,1, N. Thouveny e, E. Gracia b

a LNEG (National Laboratory for Energy and Geology; ex-INETI), Marine Geology Research Unit, Estrada da Portela, 2721-866 Alfragide, Portugalb Unitat de Tecnologia Marina – CSIC, Centre Mediterrani d’Investigacions Marines i Ambientals, 08003 Barcelona, Spainc University Bremen, RCOM, 28334-Bremen, Germanyd VU University Amsterdam, Institute for Earth Sciences, 1081 HI-Amsterdam, The Netherlandse CEREGE, Europole Mediterraneen de l’ARBOIS, BP 80, 13545 Aix-en-Provence, cedex 04n, France

a r t i c l e i n f o

Article history:Received 19 March 2008Received in revised form9 August 2009Accepted 10 August 2009

* Corresponding author. Present address: Spanishtute, Dept. of Geosciences Research and Prediction –23, 28003-Madrid, Spain.

E-mail addresses: [email protected] (S.Mlneg.pt (A.H.L. Voelker), [email protected] (A. Vizca(F.G. Abrantes), [email protected] (U. Alt-E(S. Jung), [email protected] (N. Thouveny), egracia@

1 Present address: School of GeoSciences, The GranEdinburgh, The King’s Buildings, Edinburgh EH9 3JW

0277-3791/$ – see front matter � 2009 Published bydoi:10.1016/j.quascirev.2009.08.007

a b s t r a c t

It is well established that orbital scale sea-level changes generated larger transport of sediments into thedeep-sea during the last glacial maximum than the Holocene. However, the response of sedimentaryprocesses to abrupt millennial-scale climate variability is rather unknown. Frequency of distal turbiditesand amounts of advected detrital carbonate are estimated off the Lisbon–Setubal canyons (core MD03-2698, at 4602 mwd), within a chronostratigraphy based on radiometric ages, oxygen isotopes andpaleomagnetic key global anomalies. We found that: 1) higher frequency of turbidites concurred withNorthern Hemisphere coldest temperatures (Greenland Stadials [GS], including Heinrich [H] events). Butmore than that, an escalating frequency of turbidites starts with the onset of global sea-level rising (andwarming in Antarctica) and culminates during H events, at the time when rising is still in its early-midstage, and the Atlantic Meridional Overturning Circulation (AMOC) is re-starting. This short time spancoincides with maximum gradients of ocean surface and bottom temperatures between GS and Antarcticwarmings (Antarctic Isotope Maximum; AIM 17, 14, 12, 8, 4, 2) and rapid sea-level rises. 2) Trigger ofturbidity currents is not the only sedimentary process responding to millennial variability; land-detritalcarbonate (with a very negative bulk d18O signature) enters the deep-sea by density-driven slope lateraladvection, accordingly during GS. 3) Possible mechanisms to create slope instability on the Portuguesecontinental margin are sea-level variations as small as 20 m, and slope friction by rapid deep andintermediate re-accommodation of water masses circulation. 4) Common forcing mechanisms appear todrive slope instability at both millennial and orbital scales.

� 2009 Published by Elsevier Ltd.

1. Introduction

Millennial-to-centennial scale abrupt climate variabilityis widely recognised on the Portuguese continental margin inpaleooceanographic changes (e.g. Lebreiro et al., 1996; Zahn et al.,

Geological and Mining Insti-Global Change, c/Rıos Rosas,

. Lebreiro), antje.voelker@ino), [email protected]), [email protected] (E. Gracia).t Institute, The University of

, United Kingdom.

Elsevier Ltd.

1997; Bard et al., 2000; Shackleton et al., 2000; Thouveny et al.,2000; de Abreu et al., 2003; Skinner and Elderfield, 2007). Duringthe last glacial period (w60–10 ka BP), Greenland Stadial (GS)–Interstadial (GI) cycles (also called Daansgard–Oeschger, D–O) areassociated with severely reduced surface water temperatures(w7 �C in decades; de Abreu et al., 2003) and a periodicity of 1.5 ka(Bond et al., 1997; Grootes and Stuiver, 1997). A number of GS–GIcycles terminated in massive ice-discharges from the NorthernHemisphere ice sheets, every 7–10 ka (Bond cycles; Bond and Lotti,1995), known as Heinrich (H) events. H events reflect even coolertemperatures with gradients of w10 �C in decades (de Abreu et al.,2003). During H events, ice-rafted debris was delivered when thepolar front reached the Iberian Margin. Although the forcingmechanisms behind GS–GI are not yet fully explained (e.g. Schulzet al., 2002), certainly ocean temperature changes have influenced

S.M. Lebreiro et al. / Quaternary Science Reviews 28 (2009) 3211–32233212

all the North Atlantic ocean (Bard et al., 2000), southwards to theIberian Margin, the Gulf of Cadiz (Voelker et al., 2006) and theAlboran Sea (Cacho et al., 1999). Greenland D–O long-lasting events(H) have their counterpart in the Antarctic warm events (AntarcticIsotopic Maximum, AIM) (Blunier and Brook, 2001; EPICACommunity Members, 2006). At the Portuguese margin, AIM 17–1equivalent events are easily identifiable in the benthic oxygenisotope record of Shackleton et al. (2000), with temperatureamplitudes of 1–2 �C. This bipolar seesaw concept (Broecker, 1998)explains the inter-hemispheric balance between heat and salt inthe ocean. The glacial climate variability of the North and SouthHemispheres (NH, SH) is coupled through the Atlantic MeridionalOverturning Circulation (AMOC). Collapse of the AMOC (at glacialsand GS) cools the water masses in the NH and warms the SH basin(at interglacials and AIM), because heat is no longer exportednorthwards. South warms during stadial conditions in the NH, andstarts to cool when D–O begins to warm (Stocker and Johnsen,2003).

Set off of GI–GS appears to be connected with a continental ice-volume equivalent sea-level change of intermediate range (�80 m)like during Marine Isotope Stage (MIS) 3 (Schulz, 2002). Recentstudies performed relatively accurate reconstructions of sea-levelchanges, at the sub-orbital scale (Yokoyama et al., 2001; Chappell,2002; Siddall et al., 2003, 2008; Thompson and Goldstein, 2005;Arz et al., 2007). The impact of the MIS 3 strong temperatureoscillations and subsequent sea-level and AMOC changes on thesedimentary processes is poorly known.

Achieving a precise chronology is the critical point to consis-tently link atmospheric, oceanic and sedimentary processesoccurring at the high frequency of multi-centennial to millennialscale. The SFCP2004 time scale for core MD95-2042 on the Portu-guese margin (Shackleton et al., 2004) unified different age scales(GISP2, GRIP, SS09sea), which are now accurate in a few hundredyears.

Assuming our chronology is precise, we show here that highfrequency of turbidites and land exported detrital carbonate arecoeval with rapid millennial-scale low sea-level stands. Wehypothesise that the increasing trigger of turbidity currents anddensity-driven detrital-sediment slope advection is caused byinstability of continental margins. Sediment instability is attributedto two mechanisms: rapid sea-level rising and ocean circulationand slope interaction during rapid re-accommodation of watermasses, driven by abrupt temperature and density changes, andhence associated with the strengthening of thermohaline circula-tion. We recommend oxygen isotope ratios of bulk carbonate as anefficient proxy for potential varying contributions of eroded/landtransported carbonate rocks. To our knowledge, our is the firstrecord showing the vulnerability of continental stability due tosudden climate changes in sea-level and deep- and intermediate-water circulation in short time spans for the last 63 ka.

2. Material and methods

Calypso core MD03-2698 (38�14.370N, 10�23.420W and 4602 mwater depth) was retrieved with the Marion Dufresne researchvessel during the MD134-PICABIA cruise, in July 2003. Core siteMD03-2698 is located to the northern border of a huge levee grownbetween the Tagus–Sado canyons system and the Cascais canyon(Fig. 1). Its length is 35.43 m, stored in 24 sections of about 1.5 m atLNEG, Research Unit of Marine Geology (LNEG – UGM). Depth inmetres below sea floor has been corrected for core section lengthsand voids, as well as for all data sets.

The lithology of core MD03-2698 was described by colour,texture, structures, presence of biogenic shells and detritic grains,bioturbation and chemical alteration. Silty-layers were identified in

detail to the millimetre. The total core was photographed (digitalimages taken onboard, Supplementary Fig. 1) and characterisedgeochemically (planktonic and benthic oxygen and carbon stableisotopes, bulk sediment oxygen isotope ratios, thirteen elementsX-ray fluorescence and CaCO3) and with magnetic parameters(magnetic susceptibility, and natural and anhysteretic remanentmagnetizations).

To date the lithological sequence, twenty-one AMS 14C dateswere obtained from planktonic foraminifera, mainly Globorotaliainflata, occasionally completed with Globigerinoides sacculifer, Glo-bigerinoides ruber-white, Globorotalia hirsuta, Globorotalia trunca-tulinoides, Orbulina universa, Globigerina bulloides, Globigerinellaaequilateralis, and Globigerina calida (Table 1). G. inflata is preferredbecause it calcifies in the well mixed winter layer, assuring goodexchange with the atmosphere, and eluding a potentially largerreservoir age due to upwelling of an aged water mass. All sampleswere measured in the Leibniz laboratory for Radiocarbon Datingand Isotope Research in Kiel (Germany). Samples with KIA numbers29691–29696 were measured in specific wheels for older sampleswith a run time of 12 days instead of the regular 3 days (M.J.Nadeau, personal communication). The longer run time results ina higher precision and consequently smaller error bars for therespective age. The conventional ages with �1s error bars werecorrected for a 400 years reservoir effect (Abrantes et al., 2005). Asit is impossible to judge glacial reservoir changes on the Portuguesemargin the same reservoir correction was applied to all samples.Corrected 14C ages up to 26 ka were calibrated with the INTCAL04(Reimer et al., 2004) data set. Older ages were calibrated followingHughen et al. (2004).

For chronostratigraphic purposes (d18O) and identification ofwater masses (d13C), stable isotope measurements were performedon foraminifera: 5–26 specimens of G. bulloides (planktonic) and1–6 specimens of Cibicidoides sp. (benthic) picked from the fraction>250 mm in hemi-pelagic samples. Stable isotope ratios in theforaminifera shells were measured in a Finnigan MAT 252 massspectrometer at Marum (University Bremen, Germany). The massspectrometer is coupled to an automated Kiel-carbonate prepara-tion system. The long-term precision is �0.07& for d18O and�0.05& for d13C based on repeated analyses of internal andexternal (NBS-19) carbonate standards.

Bulk d18O isotopes were used as a proxy for terrigenouscarbonate input. Hemi-pelagic samples taken along the core wereanalysed in two different laboratories: Marum-University of Bre-men, and VU University of Amsterdam. At Marum, approximately50 mg of crashed sediment was treated with 100% phosphoric acidin a Kiel-carbonate device type and loaded into a Finnigan MAT 251mass spectrometer. The precision of the equipment is �0.02& andof �0.07& for repeated measurements. In Amsterdam, approxi-mately 2 cc of the sediment was powdered and sub-samples ofroughly 20 mg were analysed in a Finnigan MAT 252 mass spec-trometer coupled to an automated CarboPrep unit (Bremen-type)following standard procedures. The long-term reproducibility is�0.08&.

To characterise the element composition of deep-sea and land-derived sediments, Ca and K, Ti, Fe were measured, in counts perarea per 30 s (cts), within a series of thirteen chemical elements in aX-Ray Fluorescence (XRF) AVAATECH core-scanner at BCR-Bremen.This allows non-destructive, rapid and efficient quantitativeanalyses (Jansen et al., 1998), done at a 2 cm interval resolution,increased to 1 cm or less in the presence of silty-layers.

To calibrate the high-resolution Ca record (in cts), calciumcarbonate content was measured in a number of samples in twodifferent laboratories: 255 samples from the interval 8.58–35.30 m(corrected depth) at LNEG – UGM and 101 samples from theinterval 0–8.58 m at RCOM – Bremen. At LNEG – UGM, CaCO3 was

Fig. 1. Bathymetry of the Portuguese margin (A) and 3.5 kHz profile taken onboard the Marion Dufresne research vessel, during the MD134-PICABIA cruise, in July 2003 (B) with thelocation of site MD03-2698 (38�14.370N; 10�23.420W) on the levee between the Cascais and the Lisbon–Setubal submarine canyons. Core MD95-2042 is shown for reference.

S.M. Lebreiro et al. / Quaternary Science Reviews 28 (2009) 3211–3223 3213

measured from an initial volume of 2 mg of dried, ground andhomogenised total sediment, using a CHNS-932 Leco elementanalyser. Inorganic Carbon (Cinorg) was calculated by the differ-ence between Total (TC) and Organic Carbon (Corg) in weightpercentage (wt%). Corg was combusted in a furnace at 400 �Ctemperature (4 steps of 100 �C/h plus 3 h at 400 �C). The precisionof 3 duplicates of each sample (TC and Cinorg) is 0.04 wt%. AtRCOM – Bremen, samples were freeze-dried, ground and homo-genised. Cinorg was calculated, instead, by the difference betweencombusted TC and Corg. Cinorg was digested with 1 M HCl andoven-dried at 60 �C; samples were not washed after treatment withHCl to prevent the loss of suspended material (Schubert and Niel-sen, 2000). All samples were run in an Element Vario-EL3 elementAnalyser, with an analytical error of 2%. Precision was ensured bycontinuous control measurements of lab internal standards. Threesamples reproduced in both laboratories show a precision of0.035 wt%. CaCO3 is calculated by multiplying Cinorg (assuming it ispredominantly carbonate and aragonite) by 8.332 and given inpercentage.

For chronostratigraphic purposes and also to approach the flowspeed of deep-water masses, magnetic property measurementswere performed at CEREGE (Aix en Provence, France). The core wassub-sampled with U-channels (length¼ 1.5 m; section¼ 2� 2 cm)pushed into the half split sediment core. U-channels were stored ina cold room for few weeks and then in a zero-field storage for oneweek prior to measurement of the natural remanent magnetization(NRM). The low-field magnetic susceptibility K (expressed here as10�5 SI unit) was measured at 2 cm resolution on U-channelsamples using a Bartington MS2B. Natural and anhysteretic

remanent magnetizations (NRM and ARM) were measured each2 cm on a three axis direct current SQUID magnetometer (2G model760R). Response curves of respectively w4 cm (for the X and Y axes)and w6 cm (for the Z axis) of the SQUID sensor coils imposea smoothing without significantly affecting the resolution of thesignals measured in such high sedimentation rate environments.NRM and ARM intensities and the ARM/K ratio (magnetic grainsize) have the dimension of volume magnetization (A m�1) whilethe ratio NRM/ARM is dimensionless. Alternating field (AF)demagnetizations were progressively imposed at 5, 10, 15, 20, 30,40, 60 and 80 milliTesla (mT) and the residual NRM. The ARMimparted in a 100 mT AF and a 0.1 mT steady field, was measuredand treated like the NRM. Isothermal Remanent Magnetizationswere imparted in a 1 Tesla direct field (IRM1T) and a 0.3 T backfield(IRM–0.3T).

All proxy data are stored in the PANGAEA data bank http://www.pangaea.de.

3. Results

3.1. Age model and sedimentation rate

Twenty-one 14C radiometric dates were obtained in planktonicforaminifera (Fig. 2A and Table 1) which indicate continuoussedimentation in core MD03-2698. For the chronostratigraphy,however, only 14C ages younger than 29.7 14C ka BP were used. Forthe older part of core MD03-2698, G. bulloides d18O isotopes werecorrelated with G. bulloides d18O isotopes of core MD95-2042

Table 1AMS 14C measurements in core MD03-2698.

AMS LabCode

Original depth(cm)

Correcteddepth (cm)

Sample material 14C age �400 yr reservoircorrected (yr BP)

�1 sigmaerror

Calibratedage (yr BP)

Age SFCP2004(yr)

KIA 27687 13–14 13–14 Mixed planktonics >250 mm 390 25 470 250KIA 29278 125–127 125–127 G. inflata > 250 mm 4430 35 5020 7550KIA 29279 155–156 156–157 Mixed planktonics >250 mm 5535 40 6340 9180KIA 29280 190–192 189–191 G. inflata >315 mm 7420 40 8260 10 640KIA 29281 261–262 260–261 G. inflata >250 mm 10 140 50 11 800 12 520KIA 27894 340–341 340–341 G. bulloides, G. inflata, O.

universa >250 mm11 335 55 13 210 13 700

KIA 29282 400–401 400–401 Mixed planktonics >150 mm 11 840 70 13 700 14 600KIA 29283 459–460 456.5–458.5 Mixed planktonics >150 mm 12 495 55 14 610 15 390KIA 29284 643–645 631–633 G. bulloides >150 mm 13 400 70 15 920 17 170KIA 29285 Sect. 9, 12–14 1184.2–1186.2 Mixed planktonics >250 mm 15 440 70 18 780 18 890KIA 29286 Sect. 9, 120–122 1292.2–1297.2 G. inflata, G. aequilateralis,

G. calida >315 mm16 500 110 19 670 19 320

KIA 29287 Sect. 10, 134–136 1456.2–1458.2 G. inflata >250 mm 18 260 130 21 730 20 290KIA 29288 Sect. 11, 85–86 1557.2–1558.2 Mixed planktonics >250 mm 19 180 140 22 750 21 120KIA 29289 Sect. 14, 37–40 1959.2–1961.2 G. inflata >250 mm 22 480 210 26 280 26 790KIA 29691 Sect. 15, 138–140 2211.2–2213.2 G. inflata >315 mm 26 790 29 890 32 340KIA 29692 Sect. 16, 113–115 2335.2–2337.2 G. inflata >315 mm 29 700 34 100 34 320KIA 29693 Sect. 17, 40–42 2412.7–2414.7 G. inflata, O. universa,

G. bulloides >250 mm31 260 35 360 36 330

KIA 29694 Sect. 19, 4–6 2682.7–2684.7 G. inflata >250 mm 36 080 40 880 41 110KIA 27686 Sect. 20, 75–77 2906.2–2911.2 G. inflata >250 mm 42 880 þ700/�645 44 680 a 47 110KIA 29695 Sect. 21, 52–54 3043.9–3045.9 G. inflata, O. universa,

G. bulloides >315 mm50 060 49 900

KIA 29696 Sect. 21, 110–112 3101.9–3103.9 G. Inflata, O. universa >315 mm 54 330 62 800 b 53 100

a Hughen et al. (2004) and Voelker et al. (2000).b Assumed to be GI 18 and 62.80 ky on the SFCP2004 time scale.

S.M. Lebreiro et al. / Quaternary Science Reviews 28 (2009) 3211–32233214

(Shackleton et al., 2004), located at the Portuguese margin, no morethan 54 km apart and 1456 m shallower than MD03-2698 (Fig. 1).

A precise stratigraphy exists for core MD95-2042 (37�480N;10�100W; 3146 m water depth), known as the SFCP2004 time scale(Shackleton et al., 2004) (Fig. 2B). In this one, glaciological modelsof Greenland layer-counting and Antarctica ice cores, radiocarbon14C dates in marine sediment cores and speleothems, 230Th dates incorals, the paleomagnetic Laschamp collapse of the Earth’smagnetic field, and 10Be concentrations were accommodated,seeking to reduce age uncertainties to correlate and understandplanetary abrupt climate changes. To ensure the interaction ofsedimentary processes with rapid climatic oscillations, a wellconstrained global time scale is necessary. Hence, core MD03-2698was timed into the SFCP2004 scale (Fig. 2).

During MIS 3, GI and GS oscillations of both curves match quitewell. GI 4 and GI 6 are given in calibrated (cal) ages of30.43� 0.29 ka BP and 34.72� 0.36 ka BP on the SFCP2004 scale(Shackleton et al., 2004), compared to 28.90� 0.45 cal ka BP and33.74� 0.61 cal ka BP in the new Greenland Ice Core Chronology(GICC05; Andersen et al., 2006) applied to the NGRIP and GISP2cores, and with interpolated 28.26 cal ka BP and point measured34.10 cal ka in MD03-2698, respectively. The age differences arewithin the uncertainty of the dating methods. In MIS 2, though, it isdifficult to identify common features to correlate the d18O isotopecurves between the two cores (Fig. 2A,B). Instead, we must rely on14C radiometric ages. The calibrated age of 23.65� 0.21 ka BP to GI 2(Shackleton et al., 2004) fits the 23.34� 0.30 cal ka BP given inGICC05, and, reasonably, the interpolated 24.33 cal ka BP in MD03-2698. The same range of discrepancy is evident along termination I,where GI 1 is aged 14.68� 0.09 cal ka in the GICC05 and SFCP2004time scales (Svensson et al., 2006) but 13.70� 0.07 cal ka BP inMD03-2698. The lack of coherency between MD95-2042 andMD03-2698 could indicate that the MD03-2698 sample wascontaminated with younger 14C, due to sediment bioturbation ortap water during washing. As a result then, the chronology of

MD03-2698 is constructed based primarily on sixteen 14C radio-metric ages for the period covering the Holocene to GI 6, tied up atthe common age of 34 cal ka BP between the two cores, and thencorrelated with core MD95-2042 on the SFCP2004 time scalebackwards to GI 18.

Our independent global paleomagnetic signatures obtained forcore MD03-2698 validate the previous chronology (Fig. 2C). TheNRM/ARM curve (Fig. 2C) can be correlated at the first order withother relative paleointensity curves gathered for the North Atlanticand European region such as GLOPIS (Fig. 2D; Laj et al., 2004) andthe Portuguese margin stack (Thouveny et al., 2004), being stronglysupported by a cosmonuclide production record (authigenic10Be/9Be obtained on the same core MD95-2042; Carcaillet et al.,2004; Muscheler et al., 2005). The RPI (Relative Paleointensity) ofcore MD03-2698 presents the typical 20–60 ka shape of geomag-netic paleointensity variations. The most characteristic RPI low atw41 ka BP is probably the trace of the Laschamp excursion. Thisexcursion has been dated by 39Ar/40Ar at 40.4� 2 ka BP on the Lavaflows of Laschamp (Guillou et al., 2004). The associated RPI low ofcore MD03-2698 and the directional signature occur prior to H 4 asin the GLOPIS record (Laj et al., 2004), and the Portuguese margincores (Carcaillet et al., 2004; Thouveny et al., 2004). Its cosmonu-clide signatures in the Greenland ice cores match the onset of GS 10(Muscheler et al., 2005). A minor pattern is correlated toa secondary feature, often observed in high-resolution RPI recordsat w34 ka BP (Benson et al., 2003) and frequently, but not firmly,associated with the Mono lake excursion.

Altogether, the paleodeclination and paleoinclination records ofcore MD03-2698 present coherent secular variation patterns,comparable to other Portuguese margin cores, although no furtherexcursions are documented by swings of larger amplitude.

In core MD03-2698 the hemi-pelagic glacial sedimentation rateis distinctly higher (1.08 m ka�1 for MIS 2 and 0.54 m ka�1 for MIS3) than the Holocene (0.23 m ka�1) (Fig. 3). Turbidite sedimentationrepresents only 1% of the whole core length, an increase in 0.051 m

‰

‰δ

HOLOCENE MIS2 MIS3

42

A

B

C

D

Fig. 2. Age model of Iberian deep marine core MD03-2698. Planktic foraminifera d18O isotopes in G. bulloides of core MD03-2698, where circles indicate depths of 14C samples. AMS14C calibrated ages in kyr (as in Table 1) are pointed by vertical lines (A); d18O in G. bulloides of core MD95-2042 (Shackleton et al., 2004; ncdc.noaa.gov), stars duplicate MD03-2698data from (A) for easy comparison of absolute values in both cores (B); Global paleomagnetic intensity anomalies as the Laschamp and possibly Mono Lake excursions, identified inintensity of natural and anhysteretic remanent magnetization ratio (30 mT) (C), compared to the GLOPIS-75 stack obtained from 24 individual worldwide relative paleointensityrecords (arbitrary units) (Laj et al., 2004) (D). All curves are put on the SFCP2004 time scale (Shackleton et al., 2004). Vertical stripes place Heinrich (H) events 6–1, the YoungerDryas (YD), and Greenland Stadials. Small even numbers point out Greenland Interstadials.

S.M. Lebreiro et al. / Quaternary Science Reviews 28 (2009) 3211–3223 3215

for MIS 3 and 0.448 m for MIS 2, thus negligible when comparingtotal and hemi-pelagic sedimentation rates. Moreover, the glacialsedimentation rate of core MD03-2698 doubles that of nearby,open slope and shallower core MD95-2042 (Fig. 3).

3.2. Distal turbidites and frequency

The majority of the 35.43 m long core MD03-2698 lithologyconsists of (light) olive–grey, slightly bioturbated, non-disturbed,biogenic hemi-pelagic silty-clay. However, the hemi-pelagic sedi-ment is frequently interlaid by dark greenish-grey, silt or silty-sandlayers of up to 3 mm in thickness, except for 4 layers which have2–4 cm (Supplementary Table 1, circles in Supplementary Fig. 1). Allthese layers show basal net contacts. However, given their thinness,grain size grading, laminations and structures are not easy todiscriminate. We considered all layers as fine distal turbidites in thesense of Stow and Shanmugam (1980). A total of 276 turbiditeswere counted along the whole core length. Eight of the 276 turbi-dites (identified as ‘‘th’’, for thick turbidites, in Supplementary

Table 1; Fig. 4A) thicker than 3 mm show laminations, but thelaminations could be mistaken by individual mm-turbidites. Thefinal number of turbidites would then be higher than 276, if weconsidered that laminations in turbidites ‘‘th’’ were rather indi-vidual turbidites. In any case, the total number is likely under-estimated because: only the largest turbidity flows triggeredreached deep site MD03-2698, very fine grain size distal turbiditesare not always detectable, and some turbidity currents might haveeroded previous deposited turbidites. We assume however thatthin distal turbidites had negligible erosive capacity, and they didnot hamper the continuous hemi-pelagic sedimentation. TheMD03-2698 hemi-pelagic record looks like other deep coresreporting the last glacial and Holocene North Atlantic deeppaleocirculation.

The number of turbidites is distributed in 500 yr time intervalsto obtain frequency of emplacement (Fig. 4A). The interval of 500 yris considered the appropriate for the average time resolution of thewhole core. In core MD03-2698, most frequent turbidites (5–10turbidites/500 years) coincide with millennial-scale H 6, 5, 4, 2, and

Cal

enda

r Age

(kyr

BP)

60

40

20

00 10 20 30

Depth (m)

MD03 - 2698

MD95 - 2042

0.54 m /kyr

1.08 m/kyr

0.23 m/kyr

Fig. 3. Comparison of sedimentation rates between core MD03-2698 and MD95-2042.

S.M. Lebreiro et al. / Quaternary Science Reviews 28 (2009) 3211–32233216

1 (Fig. 4A), but slightly lower frequent turbidites (3–8 turbidites/500 years) also happen at the sub-millennial GS 14, 11, and 10.Turbidites extinguish just after H 1, being then absent during theYounger Dryas (YD; often seen as H 0). At the larger orbital scale,the density of high frequency of turbidites during MIS 2 is greaterthan that of MIS 3, and very much contrasts with the bare Holocene(Fig. 4A; Supplementary Fig. 1).

3.3. Composition of continental derived sediments

Potassium, Ti and Fe are considered typical land-derivedelements, indicative of more sediment transport from land to thedeep-sea. Abundances of K/Ca, Ti/Ca and Fe/Ca ratios and magneticsusceptibility contents of core MD03-2698 follow very similartrends (Fig. 4B–D). Average values are superior along MIS 3 thanMIS 2, and much lower during the Holocene (Fig. 4B–D). Magneticsusceptibility reproduces the element distributions, although theglacial and Holocene magnetic background is more identical(Fig. 4E). Overlying the general trend, prominent peaks coincidewith H events 6–1 and GS 14, 11, 10, and 7. Higher K, Ti and Fe, andmagnetic susceptibility match frequent turbidites and interlaidhemi-pelagic fine carbonates. Thus, curves B–D of Fig. 4 illustratethe integrated chemical composition of turbidites and fine hemi-pelagic material. Given the large number and thinness of theturbidites it is impracticable to represent all the individual beds ina graph, and to differentiate the element signal of turbidites fromthat of fine hemi-pelagic sediment.

3.4. Continental detrital carbonate inputs

The bulk d18O record of hemi-pelagic sediment shows a mode ofgreat variability with many oscillations. The amplitude is of 5&

during MIS 3, 3& during the MIS 2 and 1& during the Holocene(Fig. 5A). Two aspects stand out during MIS 3. Firstly, if put in a G.bulloides d18O isotope reversed scale, that is, lighter bulk values inconcert with GS, it strongly resembles the NH’s GI–GS climatevariability (Fig. 5A). Even the ‘‘mid-Heinrich stadial warmingevents’’ of Skinner and Elderfield (2007) are resolved for H 5 and 4.We should emphasise that this curve was not used for chro-nostratigraphic tuning. Secondly, there is a clear step of about 1&

at around 42 kyr between the lightest values in the first (�4&) andsecond (�3&) halves of MIS 3, emulating the sea-level shapecurves of the MIS 3 (Siddall et al., 2003, 2008); amplitude also

decreased from 4 to 2& between the first and second parts of MIS 3(Fig. 5A).

3.5. Deep-water mass composition and speed

Core MD03-2698 presents Ca distribution (cts) encompassingthe millennial-scale oscillations-type of MIS 3 (Fig. 5B). Values varybetween a maximum of 3500 cts at the GI 14 and minimum of1500 cts at H 6, 5, 4, 3, 2, and 1. Relative maxima also occur at otherwarm GI, and relative minima at cold GS 14, 11, 10, 7, 5, and the YD.Although prominent oscillations further continue along the MIS 2and Holocene, the Ca averages are high (3500 cts) for the Holocene,low (2000 cts) for MIS 2, and intermediate (2500 cts) for MIS 3. Thehighest abundances happen at the beginning of the Holocene,together with GI 14.

The high-resolution XRF-Ca in cts was calibrated at lowerresolution with % CaCO3 (Fig. 5B). Results show 10% CaCO3 duringGS and around 20% during GI along MIS 3; average 15% over MIS 2,and 30–40% during the Holocene (Fig. 5B). The concentration ofCaCO3 in the deep-sea sediments measures the effect of a water-mass type bathing the North Atlantic deep-sea basin (more disso-lution during presence of Antarctic Bottom Water, AABW, andbetter preservation within North Atlantic Deep Water, NADW).

The presence of AABW/NADW water masses in the deep NorthAtlantic basin is also identified by the range values of d13C inforaminifera shells, reflecting the ambient conditions at theirgrowth time (Kroopnick, 1985). At site MD03-2698, average valuesof Cibicidoides sp. d13C are 0.5& for the Holocene, �0.4& for MIS 2and 0& for MIS 3, showing a strong influence of AABW, approxi-mately 0.5& lighter than values of shallower core MD95-2042(Fig. 5C).

The ARM/K ratio reflects magnetic grain size and is a proxy fordeep flow speed. Background values are around 0.6; above it majorpeaks rise at the base of GI 17/16, 14, 12, and the Holocene ClimaticOptimum, and minor peaks at the base of GI 8, 7, 4. The mostprominent increases during MIS 3 are coherent with warmings ofAIM 14, 12, and 8 (Figs. 2A,B and 5D,E).

4. Discussion

4.1. Intensification of sedimentary processes at millennial-scalesea-level changes

Variations in two sedimentary processes are investigated in thelight of short time scale climate variability of the order of thousandand hundreds of years. We examined variations in turbiditefrequency and continental detrital carbonate export in relation toshort and small sea-level changes. Few recent sea-level curvesfocus on the millennial-scale oscillations of MIS 3, studied at high-resolution. Although they follow indistinctly the temperaturepatterns of the SH (Chappell, 2002; Siddall et al., 2003) or NH(Arz et al., 2007), all curves have in common four major sea-levelfluctuations within MIS 3, with magnitudes of 20–30 m. A differ-ence between �60 m sea-level in the first 60–45 ka of the MIS 3and �80 m in the second 45–30 ka half is also persistent (forcompilation see Fig. 7 in Siddall et al., 2008).

The sea-level curves of Chappell (2002) and Siddall et al. (2003),based on coral series of Papua New Guinea and oxygen isotopetemperatures of the Red Sea, respectively, go along with the airtemperature changes of Antarctica. Besides, it is well establishedthat warming in Antarctica preceded warming in Greenland byseveral millennia (Blunier and Brook, 2001; Stocker and Johnsen,2003). And also that high stands lasted 1–2 ka after H events andwere followed by longer falling periods, synchronous with AIM(17, 14, 12, 8, 4) (Shackleton et al., 2000; Chappell, 2002; Siddall

(SI)

1110 14

7

B

C

D

E

A

Fig. 4. Frequency of turbidites and composition of land-derived sediments in core MD03-2698. Number of turbidites in time intervals of 500 kyr; open bars show the difference inthe frequency of turbidites if laminations of turbidites ‘‘th’’ are counted as individual turbidites (see text for detail) (A). Terrigenous element indicators in counts per area per 30 s(cts): K (B), Ti (C) and Fe (D); U-channel magnetic susceptibility K in 10�5 SI (E). Vertical stripes place Heinrich events (H) 1–6, the Younger Dryas (YD), and Greenland Stadials (smallnumbers).

S.M. Lebreiro et al. / Quaternary Science Reviews 28 (2009) 3211–3223 3217

et al., 2003; Rohling et al., 2009) (Fig. 5E,F). Does it imply that Hextreme cold (D10 �C) short events in the North Atlantic Ocean arecaught in the middle of global sea-level rises up to 35�12 mwithout any disruption (Fig. 5E,F)? It might be that either ampli-tude sea-level changes lower than the Bond cycles are globallynegligible (Knorr and Lohmann, 2003; Stocker and Johnsen, 2003;Knutti et al., 2004), or the reef archive is not able to resolve thehigher required resolution to detect GI–GS cycles in Chappell(2002). In Siddall et al.’s (2003) curve, GS 11 and 10 seem to beresolved, but it is only in 2007 when Arz and co-authors claimedthe NH coupling with sub-millennial sea-level changes. Siddallet al. (2008) tried to reconcile the magnitude and NH–SH phasing ofglobal eustatic ice-volume fluctuations within MIS 3, regardless theunsolved debate of North or South Poles leading global climatechange. Within this still indefinite global sea-level scenario, wematched the intensification of turbidity activity registered at siteMD03-2698 with Siddall et al.’s (2003, 2008) millennial sea-levelchanges of 35�12 m.

On one hand, the benthic foraminifera oxygen isotope ratios ofcore MD03-2698 are comparable to MD95-2042 (Shackleton et al.,2004) on the Portuguese margin (Fig. 5E, adjusted to the SFCP04scale), and fit the Antarctic Dome C (EDC; EPICA CommunityMembers, 2006) ice core temperatures. Global ice volume (hence,sea-level) and deep-sea temperature changes are the only influ-ences on the deep-sea d18O record of MD95-2042. Our results showhigh coherence between rapid sea-level rising and warming

towards the AIM 17, 14, 12, 8, 4 events (Shackleton et al., 2000;Siddall et al., 2003, 2008) and most frequent turbidites (5–10turbidites/500 years) at GS 17 (H 6), GS 12 (H 5), GS 8 (H 4)(Fig. 5G,F). On the other hand, relatively lower frequency of turbi-dites (3–8 turbidites/500 years) happens as well at shorter low sea-levels, coinciding with GS 14, 11, 10 and H 2 (Fig. 5G,F), and coupledwith NH temperature oscillations (Fig. 2A,B). Looked in detail,turbidite frequency starts to increase with SH deep-sea warmingand sea-level rising, but suddenly escalates just at the middle ofNH’s very cold (H 6, 5, 4, 2, 1) events (Fig. 5G,F,E). This fact points toa causal relationship between sea-level rise, abrupt oceantemperature changes between Greenland and Antarctica, andintensification of turbidity currents. From these evidences, weconfidently state that abrupt, small sea-level changes of the orderof 20 m suffice to create instability on the Portuguese continentalmargin, generating turbiditic deposits, coupled with both NH andSH temperature changes.

Apart from the turbiditic sediments derived from the continent,we discuss next the source of the hemi-pelagic sediment depositedat site MD03-2698.

If the d18O of bulk sediment is tied with the d18O of planktonicforaminifera (corresponding to surface water temperatures;Shackleton et al., 2000), an intriguing relationship is found:although the curve strongly resembles the NH climate variability ofGI–GS, the bulk d18O scale is reversed compared to the G. bulloidesd18O data (Figs. 2A,B and 5A). In the bulk isotope scale, the light

Fig. 5. Increasing frequency of turbidites and continental detrital carbonate at site MD03-2698 timed with abrupt sea-level, temperature and deep-water circulation climate changes ofNorth and South Hemispheres. Amount of detrital land-derived sediment (light d18O bulk isotopes) (A); deep-water circulation of NADW or AABW shown by concentration of Ca (XRFin counts per area per 30 s, solid line) and % CaCO3 (dotted line) (B); benthic foraminifera d13C of MD03-2698 (Cibicidoides sp., dotted line) and MD95-2042 (Planulina wuellestorfi,solid line; Shackleton et al., 2000; ncdc.noaa.gov) (C); current strength measured by the ratio anhysteretic remanent magnetisation/magnetic susceptibility in A m�1 (D); benthicforaminifera d18O of MD03-2698 (Cibicidoides sp., dotted line) and MD95-2042 (P. wuellestorfi, solid line; Shackleton et al., 2000; ncdc.noaa.gov) reflecting deep-water temperaturechanges (Shackleton et al., 2000), where AIM stands for warm Antarctic Isotope Maximum events (EPICA Community Members, 2006) (E); sea-level rises (Siddall et al., 2003; Thompsonand Goldstein, 2005) (F); increase in frequency of turbidites (number layers per 500 kyr time intervals) (G); amount of detrital carbonate delivered to the deep-sea, given as d18Odifference between bulk sediment and G. bulloides (H). All data sets are put into the SFPC2004 time scale (Shackleton et al., 2004), except the curve of Thompson and Goldstein(2005). Vertical stripes place Heinrich events (H) 6–1, the Younger Dryas (YD), and Greenland Stadials (GS). Small numbers point out Greenland Interstadials and Stadials (e.g. GS 7precedes GI 7).

S.M. Lebreiro et al. / Quaternary Science Reviews 28 (2009) 3211–32233218

S.M. Lebreiro et al. / Quaternary Science Reviews 28 (2009) 3211–3223 3219

values appear in concert with cold GS. Two questions arise: why arecold bulk d18O values light, and why are they so light (to �4&)?

The contribution to bulk oxygen isotopes of specific biogeniccomponents, mainly coccolithophores and foraminifera, is dis-cussed by Steinmetz (1994). It is stated there that calcareousnannoplankton isotopes although not as widely used as forami-nifera isotopes are a useful counterpart to the foraminifera, alsoacting as paleotemperature and paleoclimate recorders of sea-surface waters. As coccolithophores monospecific samples aredifficult to obtain, the oxygen isotope discrete value of a samplewould, ultimately, be proportional to the species most abundant inthe assemblage. Off Portugal at present-day, this species is Emilia-nia huxleyi (Moita, 2001). The vital effect correction factor assumingE. huxleyi as main driver of the d18O bulk carbonate would accountfor �1.6& (Dudley and Nelson, 1989) of our total deviation of 6&

from the d18O of G. bulloides (Fig. 5H). Thus, in core MD03-2698 thecoccolithophore component of d18O bulk carbonate does neitherexplain alone the decrease to �4& (Fig. 5H), nor reflect surfaceocean properties (SST and ice-volume).

Given the insufficient biogenic-related effect to explain thenegative enrichment of bulk d18O, we rather attribute the lightervalues down to �4& to a relatively higher detrital than biogenicfraction of carbonate in the isotopic ratio, as has recently been doneby Hodell and Curtis (2008). However, they explained the decreasein bulk d18O within H (difference of maximum 7& between Neo-globoquadrina pachyderma and bulk carbonate) found in the NorthAtlantic at 50� latitude (IODP site U1308) by the input of Canadiandolomite limestones, and possibly other sources for H 3 and H 5. Onthe Portuguese margin, we find a slightly less negative excursion ofmaximum 6&, which would be expected by longer distance toCanada if we wanted to invoke the common carbonate origin, as wepreviously did, in fact, for the origin of Heinrich events at 39�N inthe Tore seamount (Lebreiro et al., 1996). However, in the light ofthe new evidence presented above, where turbidites wereemplaced in concert with H events controlled to a certain extent byshort low sea-levels, we cannot exclude a share of the lightningbulk isotopes with local terrigenous sources of sediments. In thiscontext, it seems reasonable to accept that other shelf-slope sedi-mentary processes, like those associated with density-drivenlateral advection of hemi-pelagic sediment, would be induced, orintensified, supplying detrital carbonate from a proximal area.

In core MD03-2698, if the temperature/ice-volume component(G. bulloides) of d18O is subtracted from the bulk d18O, the detritalcomponent is enhanced (Fig. 5H). The diagram highlights thehighest detrital carbonate proportion (�1.5 to �1&) coherentduring H 6, 4, 3, 1, and GS 14, and comparatively lower (�0.5&)during other GS like 11, 10, 7 and 6. The process is magnified duringGS, but occurs also during MIS 2 (�0.5&). The Holocene (0.5&) andGI (0.5–2.5&) quite insignificant departure from the referencespecies-specific record (G. bulloides), means scarce contribution ofcontinental detrital carbonate.

Additionally, we cannot ignore the coincident stepwise reducedmagnitude of the oscillations in both the bulk d18O and sea-level forthe first and second halves of MIS 3. This similarity suggestsa proportional effect of lower sea-level stands and maximumcontribution of continental detrital carbonated sediment(Fig. 5A,F).

Finally, higher glacial sedimentation rates of core MD03-2698,compared to nearby core MD95-2042 (Fig. 3), further substantiatethe argument of increasing fine detrital carbonate flux exportedmainly during the first half of MIS 3 (older than 35 kyr), but alsoalong the 35–15 ka period. The Holocene and late deglaciation areinstead undersupplied.

The prime origin of the continental sediments transporteddirectly to the shelf or into the canyons is the Tagus river drainage

basin where eroded material is collected from outcrops of Creta-ceous and Cenozoic carbonate rocks located on both sides of theTagus River. Fine sediments would then be injected to the deep-seadown- and off-slope, most probably through intermediate andbottom nepheloid layers (Hall and McCave, 2000; de Stigter et al.,2007). Overall, a combination of detrital carbonate sources (Canadaand Iberia) seems a more realistic contribution for the very nega-tive bulk oxygen isotope ratios, although it is out of the scope of thiswork to recognize all their possible different origins.

To summarise thus, the intensification of continental sedimentsupply to the deep-sea during MIS 3 is directly controlled byAntarctic short rises in sea-level, synchronised with warm deep-ocean temperature (AIM), as well as Greenland cold surface oceantemperature (H and GS).

4.2. Instability of slopes regardless of sea-level magnitude(onset of orbital and millennial scales)

It is well accepted that retreat of ice sheets and low stands ofsea-level implied exposure of larger continental shelves (e.g. Clarkand Mix, 2000) and intensive erosion and large amounts ofterrigenous sediments transported to the deep-ocean basins.

At the glacial–interglacial orbital scale, the straight relationshipbetween generation of turbidity currents through the slope and/orcanyons and low sea-level stands is also consensual (e.g. Heezenand Drake, 1964; Vail et al., 1977; Shanmugam and Moiola, 1982;Pilkey and Cleary, 1986; Lebreiro et al., 1997; Rothwell et al., 2000).In core MD03-2698, the density of more frequent turbidites duringMIS 2 is relatively higher than during MIS 3, and differs from thebarren Holocene (Fig. 5G,F; Supplementary Fig. 1). The glacial–interglacial contrast is probably real, although it cannot be dis-regarded the possibility that the Holocene contains fine-grainedturbidites not visually recognizable from over- and underlyinghemi-pelagic sediment. Indeed, there are records of turbiditeemplacement during the Holocene on the Portuguese margin, butrelated to mechanisms other than global sea-level changes. In thenearby Tagus and Horseshoe abyssal plains (Lebreiro, 1995; Leb-reiro et al., 1997) turbidites are rare but not absent during theHolocene, and explained as the result of past seismicity or localisedtectonic processes (Thomson and Weaver, 1994; Lebreiro et al.,1997; Gutscher, 2004; Garcıa-Orellana et al., 2006; Vizcaino et al.,2006). Also not too far south, Holocene turbiditic activity incanyons off NW Africa concurred with cyclic strengthening of windregimes (Hanebuth and Henrich, 2008). Even considering thatsome turbidites might be undetected, our data confirm that at theorbital scale, large amounts of sediments are transported from thecontinental shelf to the deep-sea, preferably during the low sea-level stand of the last glacial period than during the Holocene.

Moreover, focussing on the last deglaciation, and using assupport the sea-level curve of Thompson and Goldstein (2005)which covers 130 m of sea-level rising (Fairbanks, 1989; Clark andMix, 2000), we observe that the turbiditic flows are circumscribedto the onset, first 20 m, of rising (Fig. 5G,F). The two-step turbiditemaximum, at 17.5–17 ka BP and 16.5–16 ka BP, could be a conse-quence of the MIS 2 abrupt termination with a rapid sea-level riseof 10–15 m associated with the 19-ka melt-water pulse (Clark et al.,2004). Turbidite frequency started to increase at 18.5 ka, although itpeaked at 17.5–17 ka BP. Turbidites unexpectedly extinguish at15.9 ka BP, after H 1 (Fig. 5G).

This interruption of turbiditic activity at the early sea-levelrising differs from previous works, where full sea-level rises andfalls of interglacial and glacial transitions were considered to causeinstability of shelves and slopes and trigger and deliver gravityflows downwards (Weaver and Kuijpers, 1983; Lebreiro et al., 1997).The higher resolution records of core MD03-2698 seems to reveal

S.M. Lebreiro et al. / Quaternary Science Reviews 28 (2009) 3211–32233220

that instability in the continental margins is restricted to theinception of sea-level rises.

At the millennial time scale, a larger number of turbiditesconcentrated at H events during MIS 3 coincident with low sea-levels, but mostly with the onset-to-mid of sea-level rises (pre-sented in Section 4.1), suggests an analogy between the orbital andmillennial time scales.

Thus, from the detailed described pattern of intervening turbi-dites in core MD03-2698, it is proposed here the onset-to-mid sea-level rising, no matter its magnitude, as the preferential time ofinstability of shelves/slopes at either the orbital and (sub)millennialscales.

4.3. Does deep-water circulation impinge millennial-scaleinstability on the continental margins?

Oxygen isotope measurements in Antarctica’s ice came up withwell-individualised gradual warmings in the SH (Blunier et al.,1998; EPICA Community Members, 2006). The onset of these SHwarmings preceded in 1.5–3 ka the onset of coolest events (H) inthe NH (Blunier and Brook, 2001). However, when SH air temper-ature reached the peak warmth, the deep-ocean temperature in thesubtropical NH off Iberia (Shackleton et al., 2000; Skinner et al.,2003; Skinner and Elderfield, 2007), was in phase with the onset ofGI (Shackleton et al., 2004) (Figs. 2A,B and 5E), or already warmestGI (Schmittner et al., 2003; EPICA Community Members, 2006;Steig, 2006).

Shutdown of the AMOC during GS, prevents NADW to flow tothe South Pole. Subsequently, Southern Ocean Water ventilates thedeep North Atlantic Ocean, driven by mass balance, shoaling up to2.5 km depth in the subtropical North Atlantic (Sarnthein et al.,2001). As a dynamic effect to maintain ocean salt and fresh-waterequilibrium, fresh Antarctic Intermediate Water (AAIW) alsooccupies the intermediate depths above salty NADW up to theAtlantic high latitudes during GS (Rickaby and Elderfield, 2005).Additionally, SH-sourced warm surface water extends to a latitudeof 40�N (Knorr and Lohmann, 2003). Altogether, the interplay ofdifferent water masses induce rapid changes in circulation, erodestratification of the upper ocean, destabilize the water column, andforce the AMOC to restart, therefore turning stadial into interstadialconditions.

The presence of AABW in the deep Atlantic basin at thesubtropical Iberian latitude during glacial/stadial periods whenNADW production decreased/ceased (conveyor belt concept;Broecker, 1991) was established by carbon isotope ratios of benthicforaminifera ranging between 1& (Holocene) and 0& (glacial)(Sarnthein et al., 2001; Curry and Oppo, 2005) (Fig. 5G). Off Iberia,deeper site MD03-2698 is more strongly ventilated by AABW, asseen in the approximately 0.5& lighter d13C signal of Cibicidoidessp. (Fig. 5C) if compared to core MD95-2042 (Shackleton et al.,2000).

Moreover, the concentration of calcium carbonate in sea-floorsediments reflects the varying chemistry of the deep-water massescirculation (Crowley, 1983). Glacial periods are dominated by 20–30% carbonate-rich clayey silts whereas interglacial periods are 40–70% carbonate-enriched in well preserved foraminifera oozes in thenearby Tore seamount slope at 3585 mwd (Lebreiro et al., 1996).Core MD03-2698 adds a clear record of that variability although atmillennial-scale. Being deeper, core MD03-2698 contains only 10%CaCO3 during GS and near 20% during GI; comparatively theHolocene reaches 40% (Fig. 5B). Furthermore, Shackleton et al.(2000) discovered that on the Portuguese margin, d18O benthicforaminifera copies the temperature over Antarctica and d18Oplanktonic foraminifera that of Greenland. This said, the match ind18O benthic (Fig. 5E) and planktonic (Fig. 2A,B) foraminifera

between the MD03-2698 and MD95-2042 cores, with the Caabundance and % CaCO3 in our core (Fig. 5B), seems to measure therelative presence of AABW/NADW at the Portuguese margin.Higher influence of corrosive AABW (lighter benthic d13C) goesalong with colder bottom waters (heavier benthic d18O) and lowerCaCO3 concentrations; the opposite is true when NADW dominates.At 4600 mwd, the imprint of AABW and NADW seems to dilute andneglect any effect of changing productivity in surface waters,shown in maximum 1 wt% organic Carbon measured along all core(data not shown). At site MD03-2698, we find AABW during the ‘‘offmodes’’ (Stocker, 2003) of the AMOC, GS including the YD; NADWduring ‘‘warm modes’’ like the Holocene, the Bølling/Allerød, andGI (McManus et al., 2004); and moderate AABW during MIS 2(Broecker, 1998; Sarnthein et al., 2001; Seidov and Maslin, 2001;Elliot et al., 2002; Weaver et al., 2003). Even though the high-resolution Ca (XRF) record, calibrated at lower resolution withCaCO3 (Fig. 5B), closely resembles the MIS 3 millennial-scaleoscillations, it was not used to tune events while constructing theage model, to guarantee the independence of the proxy. Thus,certainly the replacement of AABW and NADW occurs at millen-nial-scale.

In the same way, correlation with independent rock magneticparameter ratio anhysteretic remanent magnetisation/magneticsusceptibility (ARM/K) (Fig. 5D) reinforces the identification ofNADW (Moreno et al., 2002). The ARM/K traces not only the Nordicbasaltic province magnetic content of the sediments and its decayalong the NADW path, but also measures the strength of a deepcurrent (Kissel et al., 1999), irrespective of its type (ex. NADWduring GS 14 and AABW during YD; Figs. 5D and 2A,B). It wasreported that stronger flow occurred during warmer sub-orbitalperiods on the Portuguese margin (Hall and McCave, 2000). At themillennial scale, we are now able to resolve increasing deep flowspeed at the base of GI 17/16, 14, 12, 8, concomitant with transitionsto AIM 14, 12, 8, (7, 6), 1 (SFPC2004 time scale) (Figs. 2A,B and 5D,E),or heights of AIM (EDML time scale; EPICA Community Members,2006). This happens thus while the AMOC is recovering (or is fullyrecovered, EDML) at the millennial-scale oscillations. A similarprocess occurs at the Holocene Climatic Optimum with a propor-tionally higher flow speed (Fig. 5D).

We have seen a gradual increase in the frequency of deep-seaturbidites at site MD03-2698 culminating at the coldest GS andearly-to-mid sea-level rising (Fig. 6). We have also seen hemi-pelagic sediment being transported off-slope preferentially duringcoldest GS. However, once maximum ocean temperature is attainedand sea-level reached the highest stand, hardly any turbidites areemplaced during tranquil GI (best examples in GI 14 or AIM 14, GI12 or AIM 12, GI 8 or AIM 8, and H 1). But that calm is the time whenthe deep flow speed strengthens at the base of (or full) GI, and AIMevents, that is at the time of maximum temperatures at the twopolar hemispheres (GS and AIM events) (Figs. 2A,B and 5E). Weinfer that a sort of accumulated strain is gradually released toa threshold when all sediments become so unstable that largevolumes are flushed down through canyons and/or continentalslopes.

We hypothesise that a rapid replacement of deep-water masses,with compelled re-organization of the intermediate water masses(Rickaby and Elderfield, 2005), in the NH or SH, bearing unlike flowspeed, encroach different friction on the geostrophic flow along theslopes of the continental margins. The abrupt temperature, salt anddensity changes of deep-water circulation (e.g. Alley and Clark,1999; Stocker, 2003; Broecker, 2006), the rapid shut down (glacialor stadial mode), turn on (interglacial or interstadial mode) orHeinrich modes of the bipolar seesaw mechanism (Broecker, 1998,2000; Stocker, 1998), affect intermediate waters enhancingproduction and northwards heat transport (deMenocal et al., 1992;

sea levelsea

bottom

sea surface

HEINRICH

TIMEbo

ttom w

ater m

ass

temperature

temperature

spee

d

land

det

rital

car

bona

te

ANTARCTIC EVENT

+

-

EDML

EDC

turbiditefrequency

Fig. 6. Sedimentary-climatic model as a function of time. A higher proportion ofdetrital land-eroded carbonate concurs with Greenland Stadials (including Heinrichevents, H). An increasing frequency of turbidites coincides with the onset of sea-levelrise, and increasing bottom ocean temperature, but escalates and peaks at the middleof the sea-level rise and bottom ocean temperature warming, when surface oceantemperature is the coldest (H events). After this peak, turbidity activity ceasescompletely while bottom and surface temperatures converge at its maximum(Antarctic warm events, AIM, and Greenland Interstadials, GI). The time of maximumtemperature conversion between AIM (EDC, EDML or Byrd ice core records) and GI(NGRIP), and between events in the same core, is variable (see Fig. 2 in EPICACommunity Members, 2006); and represented here by alternative lines for sea bottomtemperature (bold and dashed lines). Among the ice cores, the SFPC2004 time scale tiethe best with EDC, that is convergent onsets of GI and AIM. Maximum speed of bottomwater masses coincides with inception of warming, or warmest, bottom ocean andhighest sea-level stand.

S.M. Lebreiro et al. / Quaternary Science Reviews 28 (2009) 3211–3223 3221

Haddad and Droxler, 1996; Pahnke and Zahn, 2005). These changesare particularly strongly felt off Portugal perceived by highertemperature amplitudes between H and minor GS (Ganopolski andRahmstorf, 2001). The deep subtropical North Atlantic basin seemstherefore submitted to a fast dynamics due to the NADW/AABWreplacement, and subsequent intermediate and surface water-massbalance. Added to previous, the ECBILT-CLIO model predicteda 0.3–0.4 m sea-level risen dynamic topography when the AMOCcollapses on the Portuguese margin (Flueckiger et al., 2006).Thoroughly, they transmit instability on to mid and upper depthcontinental slopes, the most probable source area for turbiditycurrents release, as well as for detrital fine-sediment lateraladvection.

Remarkably, apart from the persistent high frequency of turbi-dites triggered during relatively low(est) stands of sea-level duringGS (H) (Fig. 5F,G), it emerges that global maximum temperaturedifference between H and AIM events (which restarts the AMOC,rises sea-level and extends the water column), effectively impactson the continental margins, destabilises slopes, and intensifiescontinental sedimentary processes.

5. Conclusions

A high-resolution chronostratigraphy reveals a close connectionbetween abrupt climate change and strengthening of sedimentaryprocesses in the deep Portuguese margin for the last 63 ka.

Short sea-level changes of the order of 20 m, suffice to generateturbidity currents and increase off-slope nephleloid flow-transport.A pervasive pattern of high frequency of turbidites results frommillennial-scale low sea-levels, independently of its magnitude,both at H events and less prominent GS. A higher amount ofadvected hemi-pelagic sediment (shown in very negative decreasesto�4& of bulk d18O isotopes, meaning higher proportion of detritalland eroded to biogenic carbonates), also concurs at GS. Bothsedimentary processes reflect sub-millennial scale induced insta-bility of continental slopes.

Ocean temperature gradients between the North and SouthPoles: Heinrich stadials (surface ocean temperatures governed byGreenland) and Antarctic warming events (bottom ocean temper-atures governed by Antarctica), caused density gradients and,hence, a critical and abrupt re-accommodation of deep and inter-mediate water masses while invigorating thermohaline circulation.Abrupt variation in ocean circulation impacts on the continentalmargins, inducing slope instability, and subsequent intensificationof sedimentary processes, like the trigger of an escalating numberof turbidity currents (during H 6, 5, 4, 2, 1) and slope density-cascading of detrital carbonate (during all GS), released to the deepsea within centuries.

A novel concept of major increasing instability of slopes at thevery onset-to-mid sea-level rises at the millennial oscillations scale,reinforced by the AMOC re-start, and reflected in the intensificationof sedimentary processes, can be transferred to the orbital scale, tothe onset of deglaciations (as shown here at the transition from MIS2 to the Holocene).

As a corollary, opposed hypotheses concerning timing versusmajor instability of slopes are elucidated, and might reside solely inthe possible time-resolution of records.

Finally, understanding the precise occurrence and frequency ofpast sedimentary processes, assists prediction of geological hazardscaused by mass movements (including turbidity currents) in thepotential scenario of future abrupt climate changes.

Acknowledgements

The Commission Human Potential Programme Access toResearch Infrastructures Activity (ARI) (through the Climate Vari-ability and Abrupt Climate Changes off Portugal application) tosupport coring of Calypso MD03-2698 during the PICABIA cruise(F.G.A.); the use of XRF Core-Scanner from the University of Bremen(Ocean Drilling Program Repository) also through an ARI Paleo-studies proposal (S.M.L.). The funding from ESF-Euromarginsprojects: SEDPORT – Sedimentation Processes on the PortugueseMargin: The Role of Continental Climate, Ocean Circulation, Sea-level, and Neo-tectonics (Fundaçao para a Ciencia e TecnologiaPDCTM/40017/2003) and SWIM (01-LEG-EMA09F and REN2002-11234E-MAR); and GRICES-CSIC (Cooperaçao Cientıfica e TecnicaLuso-Espanhola). Funding support was also provided by LNEG.A.H.V. acknowledges the postdoctoral fellowship (SFRH/BPD/21691/2005). We also thank the laboratory technical support team of LNEG– UGM (C. Monteiro and D. Ferreira for LECO analyses; M.J. Custodio,A.M. Silva and P. Conceiçao for sampling; and A.L. Rodrigues forisotope samples washing; A. Rebotim picked the foraminiferaisotope samples). Three anonymous referees helped to clarify andimprove the initial manuscript. Charles Darwin, for inspiration.

References

Abrantes, F.G., Lebreiro, S.M., Rodrigues, T., Gil, I., Bartels-Jonsdottir, H., Oliveira, P.,Kissel, C., Grimalt, J.O., 2005. Shallow marine sediment cores record climatevariability and earthquake activity off Lisbon (Portugal) for the last 2000 years.Quat. Sci. Rev. 24, 2477–2494.

S.M. Lebreiro et al. / Quaternary Science Reviews 28 (2009) 3211–32233222

Alley, R.B., Clark, P.U., 1999. The deglaciation of the northern hemisphere: a globalperspective. Annu. Rev. Earth Planet. Sci. 27, 149–182.

Andersen, K.K., Svensson, A., Johnsen, S.J., Rasmussen, S.O., Bigler, M.,Rothlisberger, R., Ruth, U., Siggaard-Andersen, M.-L., Steffensen, J.P., Dahl-Jensen, D., Vinther, B.M., Clausen, H.B., 2006. The Greenland Ice Core Chro-nology 2005, 15–42 ka. Part 1: constructing the time scale. Quat. Sci. Rev. 25,3246–3257.

Arz, H., Lamy, F., Ganopolski, A., Nowaczyk, N., Patzold, J., 2007. Dominant northernhemisphere climate control over millennial-scale glacial sea-level variability.Quat. Sci. Rev. 26, 312–321.

Bard, E., Rostek, F., Turon, J.L., Gendreau, S., 2000. Hydrological impact of Heinrichevents in the subtropical Northeast Atlantic. Science 289, 1321–1324.

Benson, L., Liddicoat, J., Smoot, J., Sarna-Wojcicki, R., Negrini, S., Lund, S., 2003. Ageof the Mono Lake excursion and associated tephra. Quat. Sci. Rev. 22, 135–140.

Blunier, T., Brook, E.J., 2001. Timing of millennial-scale climate change in Antarcticaand Greenland during the last glacial period. Nature 291, 109–112.

Blunier, T., Chapellaz, J., Schwander, J., Dallenbach, A., Stauffer, B., Stocker, T.F.,Raynaud, D., Jouzel, J., Claussen, H.B., Hammer, C.U., Johnsen, S.J., 1998. Asyn-chrony of Antarctic and Greenland climate change during the last glacial period.Nature 394, 739–743.

Bond, G., Lotti, R., 1995. Iceberg discharge into the North Atlantic on millennial timescales during the last deglaciation. Science 267, 1005–1010.

Bond, G., Showers, W., Cheseby, M., Lotti, R., Hajdas, I., Bonani, G., 1997. A pervasivemillennial-scale cycle in North Atlantic Holocene and glacial climates. Science278, 1257–1266.

Broecker, W.S., 1991. The greater conveyor belt. Oceanography 4, 79–89.Broecker, W.S., 1998. Paleocean circulation during the last deglaciation: a bipolar

seesaw? Paleoceanography 13, 119–121.Broecker, W.S., 2000. Abrupt climate change: causal constraints provided by pale-

oclimate record. Earth-Sci. Rev. 51, 137–154.Broecker, W.S., 2006. Abrupt climate change revisited. Global Planet. Change 54,

211–215.Cacho, I., Grimalt, J.O., Pelejero, C., Canals, M., Sierro, F.J., Flores, J.A., Shackleton, I.N.,

1999. Dansgaard–Oeschger and Heinrich event imprints in Alboran Sea paleo-temperatures. Paleoceanography 14, 698–705.

Carcaillet, J., Bourles, D.L., Thouveny, N., Arnold, M., 2004. A high resolutionauthigenic 10Be/9Be record of geomagnetic moment variations over the last300 ka from sedimentary cores of the Portuguese margin. Earth Planet. Sci. Lett.219, 397–412.

Chappell, J., 2002. Sea level changes forced ice breakouts in the Last Glacial cycle:new results from coral terraces. Quat. Sci. Rev. 21, 1229–1240.

Clark, P., Marshall McCabe, A., Mix, A.C., Weaver, A.J., 2004. Rapid rise of sea level19 000 years ago and its global implications. Science 304, 1141–1144.

Clark, P.U., Mix, A.C., 2000. Ice sheets by volume. Nature 406, 689–690.Crowley, T.J., 1983. Calcium-carbonate preservation patterns in the central North

Atlantic during the last 150 000 years. Mar. Geol. 51, 1–14.Curry, W.B., Oppo, D.W., 2005. Glacial water mass geometry and the distribution of

d13C of SCO2 in the western Atlantic Ocean. Paleoceanography 20, PA1017.doi:10.1029/2004PA001021.

de Abreu, L., Shackleton, N.J., Schoenfeld, J., Hall, M., Chapman, M., 2003. Millennial-scale oceanic climate variability off the Western Iberian margin during the lasttwo glacial periods. Mar. Geol. 196, 1–20.

deMenocal, P.B., Oppo, D.W., Fairbanks, R.G., Prell, W.L., 1992. Pleistocene d13Cvariability of North Atlantic Intermediate Water. Paleoceanography 7, 229–250.

de Stigter, H.C., Boer, W., Mendes, P.A.J., Jesus, C.C., Thomsen, L., van den Bergh, G.D.,van Weering, T.C.E., 2007. Recent sediment transport and deposition in theNazare Canyon, Portuguese continental margin. Mar. Geol. 246, 144–164.

Dudley, W.C., Nelson, C.S., 1989. Quaternary surface-water stable isotope signalfrom calcareous nannofossils at DSDP Site 593, Southern Tasman Sea. Mar.Micropaleontol. 13, 353–373.

Elliot, M., Labeyrie, L., Duplessy, J.C., 2002. Changes in North Atlantic deep-waterformation associated with the Daansgard–Oeschger temperature oscillations(60–10 ka). Quat. Sci. Rev. 21, 1153–1165.

EPICA Community Members, 2006. One-to-one coupling of glacial climate vari-ability in Greenland and Antarctica. Nature 444, 195–198.

Fairbanks, R.G., 1989. A 17 000-year glacio-eustatic sea-level record influence ofglacial melting dates on the Younger Dryas event and deep ocean circulation.Nature 342, 637–642.

Flueckiger, J., Knutti, R., White, J.W.C., 2006. Oceanic processes as potential triggerand amplifying mechanisms for Heinrich events. Paleoceanography 21,2041–2052.

Ganopolski, A., Rahmstorf, S., 2001. Rapid changes of glacial climate simulated ina coupled climate model. Nature 409, 153–158.

Garcıa-Orellana, J., Gracia, E., Vizcaino, A., Masque, P., Olid, C., Martınez-Ruiz, F.,Pinero, E., Sanchez-Cabeza, J.-A., Danobeitia, J., 2006. Identifying instrumentaland historical earthquake records in the SW Iberian margin using 210Pb turbi-dite chronology. Geophys. Res. Lett. 33, L24601–24607.

Grootes, P.M., Stuiver, M., 1997. 18O/16O variability in Greenland snow and ice with10�3–105 year time resolution. J. Geophys. Res. 102, 26455–26470.

Guillou, H., Singer, B.S., Laj, C., Kissel, C., Scaillet, S., Jicha, B.R., 2004. On the age ofthe Laschamp geomagnetic excursion. Earth Planet. Sci. Lett. 227, 331–343.

Gutscher, M.-A., 2004. What caused the Great Lisbon Earthquake? Science 23,1247–1248.

Haddad, G.A., Droxler, A.W., 1996. Metastable CaCO3 dissolution at intermediatewater depths of the Caribbean and western North Atlantic: implications for

intermediate water circulation during the past 200 000 years. Paleoceanog-raphy 11, 701–716.

Hall, I.R., McCave, I.N., 2000. Paleocurrent reconstruction, sediment and Thoriumfocussing on the Iberian margin over the last 140 ka. Earth Planet. Sci. Lett. 178,151–164.

Hanebuth, T.J.J., Henrich, R., 2008. Recurrent decadal-scale dust events over Holo-cene western Africa and their control on canyon turbidite activity (Mauritania).Quat. Sci. Rev. 28, 261–270.

Heezen, B.C., Drake, C.L., 1964. Grand banks slump. Bull. Am. Assoc. Petrol. Geol. 48,221–233.

Hodell, D.A., Curtis, J.H., 2008. Oxygen and carbon isotopes of detrital carbonate inNorth Atlantic Heinrich Events. Mar. Geol. 256, 30–35.

Hughen, K., Lehman, S., Southon, J., Overpeck, J., Marchal, O., Herring, C., Turnbull, J.,2004. 14C activity and global carbon cycle changes over the past 50 000 years.Science 303, 202–207.

Jansen, J.H.F., Van der Gaast, S.J., Koster, B., Vaars, A.J., 1998. CORTEX, a shipboardXRF-scanner for element analyses in split sediment cores. Mar. Geol. 151,143–153.

Kissel, C., Laj, C., Labeyrie, L., Dokken, T., Voelker, A., Blamart, D., 1999. Rapid climaticvariations during Marine Isotopic Stage 3: magnetic analyses of sediments fromNordic Seas and North Atlantic. Earth Planet. Sci. Lett. 171, 489–502.

Knorr, G., Lohmann, G., 2003. Southern Ocean origin for the resumption of Atlanticthermohaline circulation during deglaciation. Nature 424, 532–536.

Knutti, R., Flueckiger, J., Stocker, T.F., Timmermann, A., 2004. Strong hemisphericcoupling of glacial climate through freshwater discharge and ocean circulation.Nature 430, 851–856.

Kroopnick, P.M., 1985. The distribution of d13C of SCO2 in the world oceans. DeepSea Res.: Part A 32, 57–84.

Laj, C., Kissel, C., Beer, J., 2004. High resolution global paleointensity stack since75 kyr (GLOPIS-75) calibrated to absolute values. In: Channell, J.E.T., et al. (Eds.),Geophys. Monogr. Ser., 145. AGU, Washington, D.C.

Lebreiro, S.M., 1995. Sedimentation History off Iberia: Tore Seamount, Tagus andHorseshoe Abyssal Plains. PhD, University of Cambridge, 192 pp.

Lebreiro, S.M., McCave, I.N., Weaver, P.P.E., 1997. Late Quaternary emplacement ofturbidites on the Horseshoe Abyssal Plain (Iberian Margin). J. Sediment. Res. 67,856–870.

Lebreiro, S.M., Moreno, J.C., McCave, I.N., Weaver, P.P.E., 1996. Evidence for Heinrichlayers off Portugal (Tore Seamount: 39�N, 12�W). Mar. Geol. 131, 47–56.

McManus, J.F., François, R., Gherardi, J.-M., Keigwin, L.D., Brown-Leger, S., 2004.Collapse and rapid resumption of Atlantic meridional circulation linked todeglacial climate changes. Nature 428, 834–837.

Moita, M.T., 2001. Estrutura, Variabilidade e Dinamica do Fitoplancton na Costa dePortugal Continental. PhD, Universidade de Lisboa, 272 pp.

Moreno, E., Thouveny, N., Delanghe, D., McCave, I.N., Shackleton, N.J., 2002. Climaticand oceanographic changes in the Northeast Atlantic reflected by magneticproperties of sediments deposited on the Portuguese margin during the last340 ka. Earth Planet. Sci. Lett. 202, 465–480.

Muscheler, R., Beer, P., Kubik, P.W., Synal, H.-A., 2005. Geomagnetic field intensityduring the last 60 000 years based on 10Be and 36Cl from the Summit ice coresand 14C. Quat. Sci. Rev. 24, 1849–1860.

Pahnke, K., Zahn, R., 2005. Southern Hemisphere water mass conversion linkedwith North Atlantic climate variability. Science 18, 1741–1746.

Pilkey, O.H., Cleary, W.J., 1986. Turbidite sedimentation in the Northwestern AtlanticOcean Basin, The Geology of North America: the western North Atlantic region.Geol. Soc. Am., 437–450.

Reimer, P.J., Baillie, M.G.L., Bard, E., Bayliss, A., Beck, J.W., Bertrand, C.J.H.,Blackwell, P.G., Buck, C.E., Burr, G.S., Cutler, K.B., Damon, P.E., Edwards, R.L.,Fairbanks, R.G., Friedrich, M., Guilderson, T.P., Hogg, A.G., Hughen, K.A.,Kromer, B., McCormac, G., Manning, S., Ramsey, C.B., Reimer, R.W., Remmele, S.,Southon, J.R., Stuiver, M., Talamo, S., Taylor, F.W., van der Plicht, J.,Weyhenmeyer, C.E., 2004. INTCAL04 Terrestrial Radiocarbon Age calibration,0–26 cal kyr BP. Radiocarbon 46, 1029–1058.

Rickaby, R.E.M., Elderfield, H., 2005. Evidence form the high-latitude NorthAtlantic for variations in Antarctic intermediate water flow during the lastdeglaciation. Geochem. Geophys. Geosyst. 6, Q05001. doi:10.1029/2004GC000858.

Rohling, E.J., Grant, K., Bolshaw, M., Roberts, A.P., Siddall, M., Hemleben, Ch,Kucera, M., 2009. Antarctic temperature and global sea level closely coupledover the past five glacial cycles. Nature-Geosci. 2, 500–504.

Rothwell, R.G., Reeder, M.S., Anastasakis, G., Stow, D.A.V., Thomson, J., Kahler, G.,2000. Low sea-level stand emplacement of megaturbidites in the western andeastern Mediterranean Sea. Sediment. Geol. 135, 75–88.

Sarnthein, M., Stattegger, K., Gdreger, D., Erlenkeuser, H., Grootes, P., Haupt, B.J.,Jung, S., Kiefer, T., Kuhnt, W., Pflaumann, U., Schafer-Neth, C., Schulz, H.,Schulz, M., Seidov, D., Simstich, J., van Kreveld, S., Vogelsand, E., Voelker, A.,Weinelt, M., 2001. Fundamental modes and abrupt changes in the NorthAtlantic circulation and climate over the last 60 kyr – concepts, reconstructionand numerical modelling. In: Schafer, P., Ritzrau, W., Schlutter, M., Thiede, J.(Eds.), The Northern North Atlantic: a Changing Environment. Springer, Berlin,pp. 365–410.

Schmittner, A., Saenko, O.A., Weaver, A.J., 2003. Coupling of the hemispheres inobservations and simulations of glacial climate change. Quat. Sci. Rev. 22,659–671.

Schubert, C.J., Nielsen, B., 2000. Effects of decarbonation treatments on d13C valuesin marine sediments. Mar. Chem. 72, 55–59.

S.M. Lebreiro et al. / Quaternary Science Reviews 28 (2009) 3211–3223 3223

Schulz, M., 2002. The tempo of climate change during Daansgard–Oeschger inter-stadials and its potential to affect the manifestation of the 1470-year climatecycle. Geophys. Res. Lett. 29 2–1, 2–4.

Schulz, M., Paul, A., Timmermann, A., 2002. Relaxation oscillators in concert:a framework for climate change at millennial timescales during the late Pleis-tocene. Geophys. Res. Lett. 29, 2139–2142.

Seidov, D., Maslin, M., 2001. Atlantic Ocean heat piracy and the bipolar climate see-saw during Heinrich and Dansgaard–Oeschger events. J. Quat. Sci. 16, 321–328.

Shackleton, N.J., Fairbanks, R.G., Chiu, T.-C., Parrenin, F., 2004. Absolute calibrationof the Greenland time scale: implications for Antarctic time scales and for D14C.Quat. Sci. Rev. 23, 1513–1522.

Shackleton, N.J., Hall, M., Vincent, E., 2000. Phase relationship between millennial-scale events 64 000–24 000 years ago. Paleoceanography 15, 565–569.

Shanmugam, G., Moiola, R.J., 1982. Eustatic control of turbidites and winnowedturbidites. Geology 10, 231–235.

Siddall, M., Rohling, E.J., Almogi-Labin, A., Hemleben, C., Meischner, D., Schmeizer, I.,Smeed, D.A., 2003. Sea-level fluctuations during the last glacial cycle. Nature423, 853–858.

Siddall, M., Rohling, E.J., Thompson, W.G., Waelbroeck, C., 2008. Marine IsotopeStage 3 sea level fluctuations: data synthesis and new outlook. Rev. Geophys.46, RG4003. doi:10.1029/2007RG000226.

Skinner, L.C., Elderfield, H., 2007. Rapid fluctuations in the deep North Atlantic heatbudget during the last glacial period. Paleoceanography 22, PA1205.

Skinner, L.C., Shackleton, N.J., Elderfield, H., 2003. Millennial-scale variability ofdeep-water temperature and d18Odw indicating deep-water source variations inthe Northeast Atlantic, 0–34 cal. ka BP. Geochem. Geophys. Geosyst. 4, 1–17.

Steig, E.J., 2006. The south–north connection. Nature 444, 152–153.Steinmetz, J.C., 1994. Stable isotopes in modern coccolithophores. In: Winter, A.,

Wiesser, W.G. (Eds.), Coccolithophores. Cambridge University Press, Cambridge,pp. 219–228.

Stocker, T.F., 1998. The seesaw effect. Science 282, 61–62.Stocker, T.F., 2003. South dials north. Nature 424, 496–499.Stocker, T.F., Johnsen, S.J., 2003. A minimum thermodynamic model for the bipolar

seesaw. Paleoceanography 18 11-11 to 11-19.Stow, D.A.V., Shanmugam, G., 1980. Sequence of structures in fine-grained turbi-

dites: comparison of recent deep-sea and ancient flysch sediments. Sediment.Geol. 25, 23–42.

Svensson, A., Andersen, K.K., Bigler, M., Clausen, H.B., Dahl-Jensen, D., Davies, S.M.,Johnsen, S.J., Muscheler, R., Rasmussen, S.O., Roethlisberger, R., Steffensen, J.P.,Vinther, B.M., 2006. The Greenland Ice Core Chronology 2005, 15–42 ka. Part 2:comparison to other records. Quat. Sci. Rev. 25, 3258–3267.

Thompson, W.G., Goldstein, S.L., 2005. Open-system coral ages reveal persistentsuborbital sea-level cycles. Science 3008, 401–404.

Thomson, J., Weaver, P.P.E., 1994. An AMS radiocarbon method to determine theemplacement time of recent deep-sea turbidites. Sediment. Geol. 89, 1–7.

Thouveny, N., Carcaillet, J., Moreno, E., Leduc, G., Nerini, D., 2004. Geomagneticmoment variation and paleomagnetic excursions since 400 kyr BP: a stackedrecord from sedimentary sequences of the Portuguese margin. Earth Planet. Sci.Lett. 219, 377–396.

Thouveny, N., Moreno, E., Delanghe, D., Candon, L., Lancelot, Y., Shackleton, N.J.,2000. Rock magnetic detection of distal ice-rafted debris: clue for the identi-fication of Heinrich layers on the Portuguese margin. Earth Planet. Sci. Lett. 180,61–75.

Vail, P.R., Mitchum, R.M.J., Todd, R.G., Widmier, J.M., Thompson, S.I., Sangree, J.B.,Bubb, J.N., Hatelid, W.G., 1977. Seismic stratigraphy and global changes of sealevel. In: Payton, C.E. (Ed.), Seismic Stratigraphy – Applications to HydrocarbonExploration. AAPG Mem., vol. 26. Am. Assoc. of Petrol. Geol., Tulsa, Ok, pp.49–212.

Vizcaino, A., Gracia, E., Pallas, R., Garcıa-Orellana, J., Escutia, C., Casas, D.,Willmont, V., Diez, S., Asioli, A., Danobeitia, J., 2006. Sedimentology, phys-ical properties and age of mass transport deposits associated with theMarques de Pombal Fault, Southwest Portuguese margin. Norwegian J. Geol.86, 177–186.

Voelker, A.H.L., Grootes, P.M., Nadeau, M.-J., Sarnthein, M., 2000. Radiocarbon levelsin the Iceland Sea from 25–53 kyr and their link to the Earth’s magnetic fieldintensity. Radiocarbon 42, 437–452.

Voelker, A.H.L., Lebreiro, S.M., Schoenfeld, J., Cacho, I., Erlenkeuser, H.,Abrantes, F.F.G., 2006. Mediterranean outflow strengthening during NorthernHemisphere Coolings: a salt source for the glacial Atlantic? Earth Planet. Sci.Lett. 245, 39–55.

Weaver, P.P.E., Kuijpers, A., 1983. Climatic control of turbidite deposition on theMadeira Abyssal Plain. Nature 306, 360–363.

Weaver, A.J., Saenko, O.A., Clark, P.U., Mitrovica, J.X., 2003. Meltwater pulse 1 fromAntarctica as a trigger of the Bolling–Allerod warm interval. Science 299, 1709–1713.

Yokoyama, Y., Esat, T.M., Lambeck, K., 2001. Coupled climate and sea-level changesdeduced from Huon Peninsula coral terraces of the last ice age. Earth Planet. Sci.Lett. 193, 579–587.

Zahn, R., Schoenfeld, J., Kudrass, H.R., Park, M.H., Erlenkeuser, H., Grootes, P., 1997.Thermohaline instability in the North Atlantic during meltwater events: stableisotopes and ice-rafted detritus records from Core SO75-26KL. Paleoceanog-raphy 12, 696–710.