LETTERSPUBLISHED ONLINE: 1 SEPTEMBER 2013 | DOI: 10.1038/NGEO1924

Air–sea temperature decoupling in westernEurope during the last interglacial–glacialtransitionMaría Fernanda Sánchez Goñi1*, Edouard Bard2, Amaelle Landais3, Linda Rossignol4

and Francesco d’Errico5,6

A period of continental ice growth between about 80,000 and70,000 years ago was controlled by a decrease in summerinsolation, and was among the four largest ice expansionsof the past 250,000 years1. The moisture source for thisice sheet expansion, known as the Marine Isotope Stage(MIS) 5a/4 transition, has been proposed to be the warmsubpolar and northern subtropical Atlantic Ocean1,2. However,the mechanism by which glaciers kept growing through threesuborbital cooling events within this period, which wereassociated with iceberg discharge in the North Atlantic3,4 andcooling over Greenland5,6, is unclear. Here we reconstructparallel records of sea surface and air temperatures frommarine microfossil and pollen data, respectively, from twosediment cores collected within the northern subtropical gyre.The thermal gradient between the cold air and warmer seaincreased throughout the MIS5a/4 transition, and was markedby three intervals of even more pronounced thermal gradientsassociated with the C20, C19 and C18’ cold events. We arguethat the warm ocean surface along the western Europeanmargin provided a source of moisture that was transported,through northward-tracking storms, to feed ice sheets in colderGreenland, northern Europe and the Arctic.

Theoretical and numerical models1,2,7 predict that rapid ice sheetgrowth in the North Atlantic high latitudes is the consequence of adecrease in summer insolation, a strong thermal gradient betweenocean and landmasses, and moisture generated by persistingwarmth and salinity in the subpolar and northern subtropical At-lantic. How this process is affected by sub-orbital climatic variabilitystill needs to be elucidated with respect to the role that this vari-ability may play in changing ice–ocean–atmosphere interactions,particularly wintermoisture flux back to the ice sheets. To fine-tuneour understanding of this process we examined the MIS5a/4transition, a periodmarked by a decrease in summer insolation8 anda succession of cooling events affecting large parts of the subpolarand central North Atlantic3. We combine new high-resolutionpollen-based vegetation and foraminifera-based sea surface tem-perature (SST) data for the interval 85–50 ka MIS5a–first part ofMIS3, from coreMD04-2845 located in the Bay of Biscay (northernsubtropical gyre, 45◦ 21′N, 5◦ 13′W, 4,100m water depth) withpublished ice-rafted debris (IRD), Neogloboquadrina pachyderma(s) and benthic foraminifera δ18O records from the same core9,10.

1Ecole Pratique des Hautes Etudes, UMR-CNRS 5805 EPOC, Université Bordeaux 1, Avenue des Facultés, 33405 Talence, France, 2Aix Marseille Université,CNRS, IRD, Collège de France, UM CEREGE, Technopole de l’Arbois, BP80, 13545 Aix-en-Provence Cedex 04, France, 3IPSL / Laboratoire des Sciences duClimat et de l’Environnement CEA CNRS UVSQ UMR 8212, Bât. 712 / Orme des merisiers 91191 Gif-sur-Yvette cedex, France, 4Université Bordeaux 1,UMR-CNRS 5805 EPOC, Avenue des Facultés, 33405 Talence, France, 5CNRS UMR 5199 PACEA-PPP, Université Bordeaux 1, Avenue des Facultés,F-33405 Talence, France, 6Department of Archaeology, History, Cultural Studies and Religion, University of Bergen, 5020 Bergen, Norway.*e-mail: mf.sanchezgoni@epoc.u-bordeaux1.fr

This approach allows the identification, without chronologicalambiguity, of offsets between eastern North Atlantic Ocean surfacehydrology (temperatures and iceberg melting) and atmosphericallydriven changes in western European vegetation. The Bay of Biscaypalaeoclimatic records are compared with new (foraminifera andUK′

37-based SST) and published palaeoclimatic records from anothercore,MD99-2331, located in the northwestern Iberianmargin10 (seeFig. 1,Methods, Supplementary Information for the environmentalsetting, and Supplementary Table S1 for chronology).

The pollen record from the western French margin (Fig. 2 andSupplementary Fig. S1) shows that the long-term Atlantic foresttrends from 85 to 50 ka followed summer insolation changes.Maxima in insolation coincided with forest expansion and SSTmaxima. This is expected because present-day Atlantic forest,mainly deciduousQuercus,Carpinus andBetula, dominates westernEurope between 40◦ and 60◦N and is associated with a temperateclimate (mean winter and summer temperatures above 0 ◦C andbetween 15 and 18 ◦C, respectively) and high annual precipitation(∼600–2,000mmyr−1; ref. 11).

At millennial timescales, in contrast, we observe a marked dif-ference between, on the one hand, MIS4 and the first part of MIS3,70–50 ka and, on the other hand, the MIS5a/4 transition, 80–70 ka.Coupled increases in SST and vegetation-inferred temperatures at∼54, 59, 65 and 69 ka punctuate MIS4 and the beginning of MIS3.During MIS3, substantial SST warming events were synchronouswith climate amelioration on land as expressed by Atlantic forestexpansions. Short and weak SST increases during MIS4 coincidedwith the replacement of grasslands by more temperate heathlands.Four major SST cooling events of 5–10 ◦C on average, at∼68,∼64,∼57 and 53 ka, related to iceberg discharge in the Bay of Biscayare also observed during MIS4 and the first half of MIS3. Theycoincided with cooling in western France, as indicated successivelyby Piceawoodlands expansion, (present-day Picea forests withstandmean winter temperatures as low as ∼–7 ◦C and cool summers,<15 ◦C (ref. 11), development of semi-desert vegetation, andgrasslands expansion. Such tight coupling between the oceanic andatmospheric temperatures in the eastern North Atlantic region isgenerally considered a common feature during periods of relativelystable intermediate and low ice sheet volume10,12.

In contradiction to this rule, previously undetected decoupledSST and western European climate changes are observed during

NATURE GEOSCIENCE | ADVANCE ONLINE PUBLICATION | www.nature.com/naturegeoscience 1

© 2013 Macmillan Publishers Limited. All rights reserved

LETTERS NATURE GEOSCIENCE DOI: 10.1038/NGEO1924

70° W 60° W 50° W 40° W 30° W 20° W

20° W

10° W

10° W

80° N

70° N

60° N

50° N

40° N

30° N

30° N

40° N

50° N

0°

0°

NEAP18K

ODP 609

V29-191

MD04-2845

MD99-2331

SU90-03

MD95-2042

IRD C 19

IRD C20

Figure 1 | Map with the locations of the cores discussed in the text. Thewestern European margin cores are located in the northern subtropicalgyre. The western and central North Atlantic cores are those in which theC20 to C17 cold events were originally identified3,4. IRD (dashed blacklines) reached as far south as 42◦ N during C19. In contrast, IRD did notreach 45◦ N in the northern subtropical gyre during C20.

the MIS5a/4 transition, a period of rapid ice growth. The MIS5a/4transition is characterized, at the orbital scale, by a progressive re-duction in Atlantic forest cover concomitant with quasi-stable SST.This trend indicates an atmospheric cooling over western Francethat is not paralleled by a decrease in the Bay of Biscay SST. The ob-served decoupling is not an artefact of themodern analogue transferfunction. As repeatedly observed in our case, those functionsindicate summer SST of∼18 ◦Cwhen percentages of temperate andsubtropical foraminifera species are higher than 20% (Fig. 2). Theatmosphere–ocean decoupling is also apparent at the sub-orbitalscale: three contractions of Atlantic forest, replaced by boreal forestand indicating cooling on land, are opposed by three SST warmingevents, at ∼78, 74 and 71 ka. Conversely, at ∼76 ka Atlantic forestexpanded synchronously with strong SST cooling. Thewarmphasesdetected on land correspond to the St Germain II, Ognon I andOgnon II Interstadials, originally identified in the La Grande Pilepollen sequence13. The three cold phases, characterized by the de-velopment of boreal forest composed of Picea andAbies correspondto Stadial I, Stadial II and the Stadial after the Ognon II Interstadial,called here Stadial III, also identified at the La Grande Pile. Present-day boreal forest occupies regions above 55◦N characterized bywinter temperatures lower than −2 ◦C, more than one month ofsnow cover and less than 600mmof annual precipitation11.

The strongest ocean–atmosphere thermal gradients are observedduring Stadial I (∼77 ka) and at the onset of Stadial II (∼74 ka;Fig. 3 and Supplementary Fig. S2). A third, weaker example ofthis decoupling is observed during Stadial III (∼71 ka), when SSTincreased by ∼2 ◦C and Abies woodlands replaced the Atlanticopen forest. Multiproxy analysis of coreMD99-2331 (northwestern

Iberian margin, 42◦ 09′N, 09◦ 40′W, 2,110m water depth) confirmsstrong thermal gradients for Stadials I and III (Fig. 3 and Supple-mentary Fig. S3).Warm SSTs recorded by foraminifera assemblagesduring these cold periods in northwestern Iberia demonstrate thatthe thermal gradients recorded in the Bay of Biscay extendedas far south as 42◦N. Alkenone analyses performed in this coreindicate that sea warming affected the uppermost layers of theocean. Therefore, three phases have been identified during StadialsI, II and III when the atmosphere cooled whereas SST remainedwarm off western Europe. Further south, at 37◦N, coreMD95-2042shows instead an atmosphere–ocean temperature coupling, withSST cooling events contemporaneous with those in the central andwestern North Atlantic.

We correlate the three warm phases identified by pollen analysisin the MIS5a/4 transition to Greenland Interstadials (GI)21/NorthAtlantic Warming (W)21, GI20/W20 and GI19/W20, and the coldphases to GS21/North Atlantic Cooling (C)20, GS20/C19 andGS19/C18’ (refs 3,6 and our proposed nomenclature for C18’;Fig. 3a,c). Four reasons justify this vegetation-based attribution asopposed to correlating SST changes in coreMD04-2845 to GIs/GSs,namely: the rapidity with which vegetation in western Europeresponds to North Atlantic atmospheric processes14, the relativelyweak influence of the atmosphere on SST at the local scale due tothe large heat capacity of the ocean and very effective heat transportthrough ocean currents15, the synchronicity established in otherregions between millennial-scale induced vegetation changes, SSTand GI/GS cycles of this period16 (Fig. 3e and Supplementary Fig.S3), and incompatibility in duration between the SST warm phasesin the Bay of Biscay duringMIS5a/4 and the time span ofGIs.

The above correlation implies that warm SST in the Bay ofBiscay during the cold atmospheric conditions of Stadials I, II andIII coincided with C20 (GS21, 77 ka), a portion of C19 (GS20,74 ka), and C18’ (GS19, ∼70 ka), three periods of cooling andmoderate iceberg discharges between 60◦ and 50◦N (refs 3,4), andincreases in the ice accumulation rate17 (Fig. 3b). This correlationalso implies that the cold ocean–warm land decoupling off westernFrance during the Ognon I Interstadial coincided withW20 and thereconstructed deceleration of the ice sheet growth in the middle ofthe MIS5a/4 transition17. In sum, our results reveal three increasesand one decrease in atmosphere–ocean thermal gradient in theeastern North Atlantic associated with staggered ice growth atthe MIS5a/4 transition.

We argue that the observed pattern provides, for the firsttime, sound supporting evidence for an ice growth scenarioinvolving both orbital and sub-orbital atmosphere–ocean thermalcontrasts. As implied in the models in refs 1,7, moisturenecessary to fuel the ice sheet was produced by the strongatmosphere–ocean thermal gradient due to a decrease in summerinsolation, cooling landmasses, and enhanced Atlantic MeridionalOverturning Circulation implying warming SST. This scenario is inline with the observed high influx increase of warm Atlantic waterto the Norwegian Sea during MIS5a (ref. 18). Superimposed onthis long-term process, one observes an alternation of strong andweak thermal gradients, correlated with GS/C and GI/W phases,respectively. Strong gradients result in high rates of moistureproduction, its transport by northward-tracking storms, and itsfalling as snow due to reduced summer insolation. Weak gradientsslow down the process while still allowing ice accumulation.

During the North Atlantic and western European cold phases oftheMIS5a/4 transition the process is associatedwith a redistributionof surface currents triggered by modest iceberg melt drivingand concentrating warm waters in the eastern North Atlanticat mid-latitudes. Such a configuration favours ice accumulationover Greenland, Scandinavia and Arctic regions. Southwarddisplacement of tundra by 10◦ in latitude during cold phases C20,C19 and C18’, suggested by boreal forest colonization of western

2 NATURE GEOSCIENCE | ADVANCE ONLINE PUBLICATION | www.nature.com/naturegeoscience

© 2013 Macmillan Publishers Limited. All rights reserved

NATURE GEOSCIENCE DOI: 10.1038/NGEO1924 LETTERS

Obliquity

25

24

23

Precession

Insolation65° N June ¬0.03 530

510

490

470

450

¬0.01

0.01

0.03

240

[CO

2] (

ppm

)A

tlant

ic fo

rest

polle

n (%

)Bo

real

fore

stpo

llen

(%)

Tem

pera

te a

nd

subt

ropi

cal

fora

min

ifera

(%

)IR

D g

rain

s (g

¬1 )

Rela

tive

sea

leve

l (m

)

220

200

(W m

¬2)

80

60

40

20

0

60

40

40

30

20

10

0

¬80

¬60

¬40

MIS3 MIS4 MIS5a

¬20

2,0001,6001,200

800400

0

20

0

50 60 70 80

Age (ka)

Abies and Piceapollen (%

)H

eathlandspollen (%

)SST

jas (°C)

N. pachy. (s) (%

)δ

18O benthic

% (%

VPD

B)

Bay of Biscay (MD04-2845)

St Germain II

Stad

ial I

IIO

gnon

II

Stad

ial I

I

Stad

ial I

Ogn

on I

C17 C18 C19 C20

e

8

40

30

20

201612

40

10

10080604020

4.44.03.63.22.8

0

0

a

b

c

d

Figure 2 | Insolation and atmospheric CO2 concentration changes versus marine and terrestrial palaeoclimatic records from core MD04-2845 for theinterval 85–50 ka. a–e, Insolation, precession and obliquity curves8 (a), CO2 concentration record28 (b), pollen record of Atlantic and boreal forests, andheathlands in western France, foraminifera-based summer (JAS) SST (red line) in the Bay of Biscay, and percentages of temperate and subtropicalforaminifera assemblages (orange), IRD concentration and N. pachyderma (s) percentage records (c), reconstructed sea level changes17 (d), and benthicforaminifera δ18O curve (Cibicides wuellestorfi >150 µm) from MD04-2845 (e). Grey bands indicate warm intervals in western France. C20 to C17 indicatethe North Atlantic cooling events between 85 and 50 ka (ref. 3).

Europe, probably amplified ice growth owing to the increase insurface albedo19. Previous studies have revealed a tight couplingbetween oceanic and atmospheric temperatures to the millennial-scale cycles10,12. Our results show that this ocean–atmospherecoupling is not a pervasive feature throughout the Quaternary, andthat in the eastern North Atlantic, a contrasting situation is evidentduring periods of rapid ice growth when strong thermal gradientsdeveloped between ocean and landmasses.

Atmosphere–ocean coupling below 42◦N, observed in theMD95-2042 record, identifies the southernmost extension ofthe decoupling phenomenon. Such a limit probably reflects theatmospheric influence of the central sub-polar and western NorthAtlantic SST cooling on subtropical latitudes in a situation of rapid

ice growth. Such atmospheric teleconnection is predicted for thepresent day following an abrupt freshening of the North Atlantic20.Future research should test whether this pattern holds true forother instances of rapid ice growth such as the MIS7/6 (∼190 ka)and MIS7c/7b (230 ka) boundaries. Implications of the resultspresented here for human adaptation and demography also need tobe investigated. Strong thermal gradients in a context of long-termland cooling were probably responsible for the creation of mildcoastal environments, colonized by pines and colder inland areascovered by boreal forests (Supplementary Fig. S1). Climaticallycontrasted coastal areas have favoured in other Earth regions theemergence of hunter-gatherer adaptations characterized by highpopulation density, residential mobility and cultural diversity21.

NATURE GEOSCIENCE | ADVANCE ONLINE PUBLICATION | www.nature.com/naturegeoscience 3

© 2013 Macmillan Publishers Limited. All rights reserved

LETTERS NATURE GEOSCIENCE DOI: 10.1038/NGEO1924

18 19

19 20 21 GS

GI ¬35

¬45

¬55

T site (

°C)

T site δ

15N

20 21

¬80

¬60

¬40

80

60

40

20

20

16

12

8

40

Bay of BiscayMD04-2845

Rela

tive

sea

leve

l (m

)A

tlant

ic fo

rest

po

llen

(%)

30

20

10Atla

ntic

fore

stpo

llen

(%)

40

0

40

30

20

10

Med

iterr

anea

n fo

rest

polle

n (%

)

50

0

Age (ka)

SSTjas (

°C)

SSTjas (

°C)

24

20

16

12

8

4

SST U

K'

18

16

14

12

Stad

ial I

II

Stad

ial I

I

Stad

ial I

NW Iberian marginMD99-2331

SW Iberian marginMD95-2042

C18' C19 C20W19 W20 W21

a

b

c

d

e

37 (°C

) (

°C)

SST U

K'37

68 72 76 80 84

Figure 3 | Greenland temperature record compared with land–sea palaeoclimatic records from the three western European margin cores during theMIS5a/4 transition. a–e, Greenland temperature record29 (a), reconstructed sea-level changes17 (light blue; b), core MD04-2845 (Bay of Biscay):foraminifera-based summer SST curve (red) with the minima and maxima values found in the set of the 5 selected analogues (grey surface), Atlantic forestpollen percentage curves (green; c), MD99-2331 (northwestern Iberian margin) the same as core MD04-2845, and (UK′

37)-based SST (black; d), coreMD95-2042: Mediterranean forest pollen percentage curve (green)16 and (UK′

37)-based SST (black; e)25. The chronology of the Greenland temperaturerecords are based on the GICC05-model-ext30. Grey bands indicate warm intervals in western France.

This pattern is generally interpreted to be the consequence ofdecreases in the risk of resource failure, which is characteristic ofthose environments owing to the high productivity of their marineand terrestrial food chains throughout the year.

MethodsThe preparation technique for pollen analysis follows the protocol established at theUMR EPOC, Bordeaux 1 University. After chemical treatment (cold HCl and coldHF) the samples were sieved through 10 µm nylon mesh screens. The final residuefor pollen analysis was mounted unstained in bidistilled glycerine (according tohttp://ephe-paleoclimat.com/ephe/Pollen%20sample%20preparation.htm). Pollengrains were counted using a Zeiss Axioskop light microscope at ×400 and ×1,000(oil immersion) magnifications. A minimum of 100 pollen grains, excluding Pinus,and 20 pollenmorphotypes were counted. Both conditions are required to obtain an

accurate image of the vegetation community and its floristic diversity. The temporalresolution of the analysis is 500 years on average. A number of levels, however,mainly from the middle part of MIS4 in core MD04-2845, have pollen counts thatdo notmeet these thresholds andwere excluded from the diagram.

Foraminifer assemblages were analysed in the >150 µm fraction of the samesediment for pollen analysis, and at a similar resolution to that applied for the pollencounting for the targeted period. A minimum of 300 foraminifera was counted ineach sample. Planktonic foraminifera counts were used to quantify SST by meansof an ecological transfer function developed at the EPOC laboratory, which usesthe Modern Analogue Technique, instead of SIMMAX (Supplementary Fig. S4),and relies on an extended modern database including the North Atlantic and theMediterranean basins (1007 points) developed during the MARGO project22.Modern hydrological parameters were obtained from the WOA (1998) databaseusing the tool developed by Schaffer-Neth during the MARGO project (http://www.geo.uni-bremen.de/geomod/Sonst/Staff/csn/woasample.html). Calculations

4 NATURE GEOSCIENCE | ADVANCE ONLINE PUBLICATION | www.nature.com/naturegeoscience

© 2013 Macmillan Publishers Limited. All rights reserved

NATURE GEOSCIENCE DOI: 10.1038/NGEO1924 LETTERSof past hydrological parameters were performed with the R software using a scriptdeveloped by Guiot and Brewer (www.eccorev.fr/spip.php?article389). It relieson a weighted average of SST values from the five best modern analogues, withmaximum weight given to the closest analogue in terms of statistical distance, thatis, dissimilarity minimum22. This method permits the reconstruction of annualand seasonal SST for the 0–700m water layer with a degree of confidence (rootmean square error) of 1.1 ◦C for spring and annual SST, 1.2 ◦C for winter andautumn SST and 1.3◦ C for summer SST. The temperate and subtropical groupis composed of Globorotalia scitula, Globigerinita glutinata, Globorotalia inflata,G. hirsuta, G. truncatulinoides dextral, G. truncatulinoides sinistral, G. crassaformis,Hastigerina aequilateralis, H. pelagica, Globigerina falconensis, G. calida,O. universa,T. humilis, Globoturborotalita rubescens, Globigerinoides ruber alba, G. ruber rosea,G. conglobatus and G. tenellus.

The canonical correspondence analysis between pollen percentages and SSTreconstructions was scripted and run in the R application 2.10 GUI 1.3 (ref. 23)using the community ecology package Vegan24.

The alkenone unsaturation ratio (UK′37) was measured at CEREGE

(Aix-en-Provence, France) on freeze-dried bulk samples at 10 cm depth resolutionproviding a time resolution of about 1 ka (similar to that for pollen). Theextraction of alkenones was performed using an automated Dionex AcceleratedSolvent Extractor (ASE-200; ref. 25). The precision of our analytical proceduresfor the SST determination is about 0.5 ◦C, and has been checked against theinternational alkenone intercomparison26. SST values are calculated with thecalibration equation27 based on laboratory cultures of coccolithophorids. Thisconventional calibration follows the relationship between annual mean SST andUK′

37 measurements on core tops. In the case of the Iberian Margin, it has beenshown that themost recent UK′

37 SST estimates are indeed similar to the annual meanSST (ref. 25). However, the biological productivity in this zone depends on seasonand depth. In a first-order attempt to take this effect into account, it has beenshown that the most recent UK′

37 temperatures correspond to the modern 0–50maverage temperature weighted by primary production25.

The chronology of the three western European margin cores is based on theage model of core MD95-2042. This age model is derived from the identificationof five stratigraphic events in this core: D-O 17, D-O 14 and D-O 12 based onδ18O from planktonic foraminifera, and the identification of two sea stillstands(MIS5e/5d and MIS5a; Supplementary Table S1a). This age model provides agesfor the Mediterranean and Atlantic forest pollen percentage increases observed inthe western European margin cores that we assume are synchronous, and for theonset of Heinrich Stadial 6, which is marked by the highest IRD andN. pachyderma(s) increases (Supplementary Table S1b). Note that the shape and structure ofHeinrich Stadial 6 in the Bay of Biscay core is almost identical to that originallyidentified in ref. 3, and encompasses C17 and C18.

Data. The data for this work can be found at http://pollen.cerege.fr/fpd-epd/bibli.do and ftp://ftp.ncdc.noaa.gov/pub/data/paleo/contributions_by_author/sanchezgoni2013/.

Received 15 March 2013; accepted 22 July 2013; published online1 September 2013; corrected online 5 September 2013

References1. Ruddiman, W. F. & McIntyre, A. Oceanic mechanisms for amplification of the

23,000-year ice-volume cycle. Science 212, 617–627 (1981).2. Ruddiman,W. F. &McIntyre, A.Warmth of the subpolar North Atlantic Ocean

during northern hemisphere ice-sheet growth. Science 204, 173–175 (1979).3. McManus, J. F. et al. High-resolution climate records from the North Atlantic

during the last interglacial. Nature 371, 326–329 (1994).4. Chapman, M. R. & Shackleton, N. J. Global ice-volume fluctuations, North

Atlantic ice-rafted events, and deep-ocean circulation changes between 130and 70 ka. Geology 27, 795–798 (1999).

5. Dansgaard, W. et al. in Climate Processes and Climate Sensitivity (eds Hansen,J. E. & Takahashi, T.) 288–298 (American Geophysical Union, 1984).

6. Andersen, K. K. et al. The Greenland ice core chronology 2005, 15-42 ka. Part1: Constructing the time scale. Quat. Sci. Rev. 25, 3246–3257 (2006).

7. Risebrobakken, B. et al. Inception of the Northern European ice sheet due tocontrasting ocean and insolation forcing. Quat. Res. 67, 128–135 (2007).

8. Berger, A. Long-term variations of daily insolation and Quaternary climaticchanges. J. Atmos. Sci. 35, 2362–2367 (1978).

9. SánchezGoñi,M. F. et al. European climatic optimumand enhancedGreenlandice sheet melting during the last interglacial. Geology 40, 627–630 (2012).

10. Sanchez Goñi, M. F. et al. Contrasting impacts of Dansgaard-Oeschger eventsover a western European latitudinal transect modulated by orbital parameters.Quat. Sci. Rev. 27, 1136–1151 (2008).

11. Polunin, O. & Walters, M. A Guide to the Vegetation of Britain and Europe(Oxford Univ. Press, 1985).

12. Desprat, S. et al. in The Climate of Past Interglacials (eds Sirocko, F., Claussen,M., Sánchez Goñi, M. F. & Litt, T.) 375–386 (Elsevier, 2007).

13. Woillard, G. M. Grande pile peat bog: A continuous Pollen record for the last140.000 years. Quat. Res. 9, 1–21 (1978).

14. Gouveia, C., Trigo, R. M., DaCamara, C. C., Libonati, R. & Pereira, J. M. C.The North Atlantic Oscillation and European vegetation dynamics. Int. J. Clim.28, 1835–1847 (2008).

15. Oort, A. H. & Vander Haar, T. H. On the observed annual cycle in theocean–atmosphere heat balance over the northern hemisphere. J. Phys. Ocean.6, 781–800 (1976).

16. Sánchez Goñi, M. F. & Sirocko, in The Climate of Past Interglacials(eds Claussen, F., Sánchez Goñi, M. & Litt, M. F.) 197–205 (Elsevier, 2007).

17. Waelbroeck, C. et al. Sea-level and deep water temperature changes derivedfrom benthic foraminifera isotopic records. Quat. Sci. Rev. 21, 295–305 (2002).

18. Fronval, T. & Jansen, E. Eemian and early Weichselian (140–60 ka)paleoceanography and paleoclimate in the Nordic seas with comparisons toHolocene conditions. Paleoceanography 12, 443–462 (1997).

19. Crucifix, M. & Loutre, M. F. Transient simulations over the last interglacialperiod (126-115 kyr BP): Feedback and forcing analysis. Clim. Dyn. 19,417–433 (2002).

20. Chiang, J. C. H., Cheng, W. & Bitz, C. M. Fast teleconnections to the tropicalAtlantic sector from Atlantic thermohaline adjustment. Geophys. Res. Lett. 35,L07704 (2008).

21. Binford, L. R.Constructing Frames of Reference: An Analytical Method for TheoryBuilding Using Ethnographic and Environmental Data Sets (Univ. CaliforniaPress, 2001).

22. Kucera, M., Rosell-Melé, A., Schneider, R., Waelbroeck, C. & Weinelt, M.Multiproxy approach for the reconstruction of the glacial ocean surface(MARGO). Quat. Sci. Rev. 24, 813–819 (2005).

23. Urbanek, S. & Iacus, S. M. R: A Language and Environment for StatisticalComputing (R Foundation for Statistical Computing, 2007).

24. Oksanen, J., Kindt, R. & Legendre, P. et al. Vegan: Community Ecology Package.R package version 1. 8-8 (2007).

25. Pailler, D. & Bard, E. High frequency palaeoceanographic changes during thepast 140,000 yr recorded by the organic matter in sediments of the Iberianmargin. Palaeogeogr. Palaeoclimatol. Palaeoecol. 2799, 1–22 (2002).

26. Rosell-Melé, A. et al. Precision of the current methods to measure the alkenoneproxy U37K′ and absolute alkenone abundance in sediments: Results of aninterlaboratory comparison study. Geochem. Geophys. Geosyst. 2, 1046 (2001).

27. Prahl, F. G., Muehlhausen, L. A. & Zahnle, D. L. Further evaluation oflong-chain alkenones as indicators of paleoceanographic conditions. Geochim.Cosmochim. Acta 52, 2303–2310 (1988).

28. Ahn, J. & Brook, E. J. Atmospheric CO2 and climate on millennial time scalesduring the last glacial period. Science 322, 83–85 (2008).

29. Landais, A. et al. A continuous record of temperature evolution over a wholesequence of Dansgaard-Oeschger during Marine Isotopic Stage 4 (76 to 62 kyrBP). Geophys. Res. Lett. 31, L22211 (2004).

30. Wolff, E. W., Chappellaz, J., Blunier, T., Rasmussen, S. O. & Svensson, A. C.Millennial-scale variability during the last glacial: The ice core record.Quat. Sci. Rev. 29, 2828–2838 (2010).

AcknowledgementsWe thank the coring and logistic teams onboard the R/V Marion Dufresne duringthe IMAGES I, GINNA and ALIENOR oceanographic cruises. The work of M.F.S.G.and F.d. was supported by the ERC Advanced Grant TRACSYMBOLS no. 249587.IPEV, INSU-ECLIPSE and ANR-PICC French programmes provided financial supportto UMR-CNRS 5805 EPOC as well as the RESOLuTION project funded by the ESFprogramme EUROCORES. A.L. received support from the ANR Citronnier. E.B.acknowledges financial support from the European Commission (Project Past4Future)and the Collège de France. We thank D. Urrego for assisting us with the statisticalanalysis, E. Salgueiro for reconstructing Iberian margin SSTs from core MD95-2042using the Iberian margin database, and F. Rostek for measuring alkenones at CEREGE.We are grateful to S. Desprat, D. Urrego, M-N. Woillez, A-L. Daniau and W. Banksfor constructive discussions, M-H. Castera, M. Georget and O. Ther for their technicalassistance, and V. Hanquiez for drawing Fig. 1.

Author contributionsM.F.S.G. designed the study, performed pollen analysis, interpreted the data and wrotethe manuscript. E.B. provided the alkenone data and discussed data and interpretations.A.L. discussed the data. L.R. performed the foraminifera analysis and reconstructed seasurface temperatures. F.d. contributed to the writing of themanuscript.

Additional informationSupplementary information is available in the online version of the paper. Reprints andpermissions information is available online at www.nature.com/reprints. Correspondenceand requests for materials should be addressed toM.F.S.G.

Competing financial interestsThe authors declare no competing financial interests.

NATURE GEOSCIENCE | ADVANCE ONLINE PUBLICATION | www.nature.com/naturegeoscience 5

© 2013 Macmillan Publishers Limited. All rights reserved

In the version of this Letter originally published online, the ‘Data’ section of the Methods was missing. This has been corrected in the HTML and PDF versions of the Letter.

Air–sea temperature decoupling in western Europe during the last interglacial–glacial transitionMaría Fernanda Sánchez Goñi, Edouard Bard, Amaelle Landais, Linda Rossignol and Francesco d’Errico

Nature Geosci. http://dx.doi.org/10.1038/ngeo1924 (2013); published online 1 September 2013; corrected online 5 September 2013

ERRATUM

© 2013 Macmillan Publishers Limited. All rights reserved