www.elsevier.com/locate/epsl

Earth and Planetary Science Le

Increasing vegetation and climate gradient in Western Europe over

the Last Glacial Inception (122–110 ka): data-model comparison

M.F. Sanchez Gonia,*, M.F. Loutreb,1, M. Crucifixc,2, O. Peyrond,3, L. Santose,4,

J. Dupratf,5, B. Malaizef,6, J.-L. Turonf,7, J.-P. Peypouqueta,8

aEPHE, DGO, UMR-CNRS 5805, Universite Bordeaux 1, Avenue des Facultes, 33405 Talence, FrancebUniversite catholique de Louvain, Chemin du Cyclotron, 2, B-1348 Louvain-la-Neuve Belgium

cHadley Centre for Climate Prediction and Research, Met Office, FitzRoy Road, Exeter, EXI 3PB, U.K.dLaboratoire de Chrono-Ecologie, UMR-CNRS 6565, Universite de Besancon, 16 rte de Gray, 25030 Besancon, France

eUniversidade da Coruna, Facultade de Ciencias, Campus da Zapateira s/n15071 A Coruna, SpainfDGO, UMR-CNRS 5805, Universite Bordeaux 1, Avenue des Facultes, 33405 Talence, France

Received 18 June 2004; received in revised form 10 December 2004; accepted 16 December 2004

Editor: E. Bard

Abstract

High-resolution terrestrial (pollen) and marine (planktic and benthic isotopes, coarse fraction, and N. pachyderma (s))

analyses have been performed in the Marine Isotope Stage (MIS) 5 interval of IMAGES core MD99-2331 retrieved in the

northwestern Iberian margin. This study shows the occurrence of a Zeifen Interstadial/Stadial succession on land and in the

ocean during the first part of MIS 5e. In northwestern Iberia, the Eemian is marked from 126 to 122 ka by the development of

deciduous Quercus forest at the same time as Mediterranean forest colonised southern Iberia, and deciduous Quercus-Corylus

forest occupied northernmost European regions. From 121 to 115 ka Carpinus betulus forest developed in NW Iberia indicating

a winter cooling by 2 8C on land and an increase in annual precipitation by 100–200 mm along with a Sea Surface Temperature

0012-821X/$ - s

doi:10.1016/j.ep

* Correspon

E-mail addr

michel.crucifix@

j.duprat@epoc.u

jp.peypouquet@1 Tel.: +322 Tel.: +443 Tel.: +334 Tel.: +345 Tel.: +336 Tel.: +337 Tel.: +338 Tel.: +33

tters 231 (2005) 111–130

ee front matter D 2004 Elsevier B.V. All rights reserved.

sl.2004.12.010

ding author. Tel.: +33 540002628; fax: + 33 556488408.

esses: mf.sanchezgoni@epoc.u-bordeaux1.fr (M.F. Sanchez Goni)8 loutre@astr.ucl.ac.be (M.F. Loutre)8

metoffice.gov.uk (M. Crucifix)8 Odile.Peyron@Univ-FComte.FR (O. Peyron)8 xesantos@udc.es (L. Santos)8

-bordeaux1.fr (J. Duprat)8 b.malaize@epoc.u-bordeaux1.fr (B. Malaize)8 jl.turon@epoc.u-bordeaux1.fr (J.-L. Turon)8

epoc.u-bordeaux1.fr (J.-P. Peypouquet).

10473299; fax: +32 10474722.

1392884105; fax: +44 1392885681.

381666669; fax: +33 381666568.

981167000; fax: +34 981167065.

540008865; fax: + 33 556488408.

540008794; fax: + 33 556488408.

540008384; fax: +33 556488408.

540008857; fax: +33 556488408.

M.F. Sanchez Goni et al. / Earth and Planetary Science Letters 231 (2005) 111–130112

(SST) decreasing trend off Iberia. A similar cooling has been documented at the same time in northern Germany (528 N) by the

replacement of deciduous forest by coniferous (Abies-Picea) formations, implying a southward displacement of the deciduous

tree line between ~608 N and 508 N as early as 120 ka. The southward migration of the tree line between 728 N and 588 Nsimulated by the Earth Model of Intermediate Complexity MoBidiC from 122 and 120 ka and considered as a major process to

initiate the last glaciation is, therefore, compatible with data. Between 115 and 110 ka, the substantial ice accumulation in

northern high latitudes (MIS 5e/5d transition) was synchronous with successive drops, C26 and C25, in northeastern Atlantic

SST. In northwestern Iberia Abies-Pinus trees developed at the expense of Quercus-Carpinus forest. A tundra-like environment

occupied northern Germany, marking the end of the interglacial in northwestern Europe at 115 ka, and boreal forest likely

colonised northeastern France. The first displacement of the vegetation belts at 121 ka was enhanced at 115 ka indicating an

amplification of the vegetation and climate gradients in northeastern Atlantic and European borderlands probably related with

the well-developed ice caps at that time. The comparison between the general trend in the estimated and simulated MoBidiC

winter and summer temperatures for latitudes between 35 and 458 N, shows that both follow quite straightforwardly the

precession signal although the simulated and reconstructed temperatures agree better in the South than North of 408 N. Annualprecipitation is exhibiting opposite trend in the data and in the model. This contradiction is likely the fact that the zonal climate

simulated by the model may not accurately represent the regional climate features, as reconstructed from the pollen.

D 2004 Elsevier B.V. All rights reserved.

Keywords: Eemian; MIS 5; Last Glacial Inception; western Europe; northwestern Iberia; EMIC; climate simulation; pollen; vegetation; climate

reconstruction

1. Introduction

The last interglacial/glacial transition (122 to 110

ka BP) has been in the past the focus of a number

of studies with models of different complexity [1,2].

These studies have put forward the role of land

processes, and specifically changes in vegetation,

for entering in the last glaciation. The replacement

of taiga by tundra acts as positive feedback on

accumulation of perennial snow because snow

covered tundra has a larger albedo than snow

covered taiga. In particular, the Earth Model of

Intermediate Complexity (EMIC), MoBidiC, has

simulated between 122 and 120 ka a southward

migration of the tree line by 148 in latitude, from

728 to 588 N, and decrease in summer temperatures

by 2–38C in the midlatitudes (30–508 N) of Eurasia

[3]. Detailed comparison between these simulations

and palaeoenvironmental data is required to assess

the ability of such models to simulate the observed

climatic variability and to discuss the mechanisms

responsible for it. However, this comparison is

generally impaired by the lack of a reliable

chronology of terrestrial records for that time period

[4]. Pollen sequences from the continental realm

covering the last interglacial period have often a time

scale dependent on the marine stratigraphy [5] which

is based either on the assumption that the whole MIS

5e interval coincides with the Eemian period on land

[6,7] or that Eemian not only encompasses MIS 5e but

much of MIS 5d as well [8] giving different time

spans, from 12,000 to 23,000 yr, for the duration of

the Eemian.

The Eemian Interglacial is recognised by a

succession of tree-dominant pollen assemblage bio-

zones, bracketed between open vegetations of the

previous and succeeded glacial periods, which have a

remarkable consistency right across temperate areas of

western and northern Europe [9,10]. Further, palyno-

logical investigations of the maar lakes in central Italy,

of the Greek long sequences of Ioannina and Tenagi–

Philippon and of the southwestern Iberian margin core

MD95-2042 show that the Eemian in the Mediterra-

nean region can be brought into the same perspective

as northwestern European sequences [11,12]. It has

been shown recently, on the basis of the direct

correlation between pollen and marine proxy records

from deep sea core MD95-2042 [13,14] that the onset

of the Eemian lagged the beginning of MIS 5e by

6000 yr and the accumulation of substantial ice

volume (MIS 5e/5d, ~115–110 ka) was synchronous

with a 5000 yr long lasting deciduous forest in

southern Iberia. The U/Th-derived chronology of this

Portuguese core and the varve-dated German sequen-

ces has allowed the reassessment of the chronology of

the vegetational succession over the Eemian for these

M.F. Sanchez Goni et al. / Earth and Planetary Science Letters 231 (2005) 111–130 113

two western European regions [15–18]. These works

suggest that the Eemian in southern Iberia lasted

16,000 yr, from 126 ka to 110 ka, while it was shorter

north of the Alps and Pyrenees, i.e. between 11,000

and 13,000 yr, from 126 ka to ~115 ka. Tundra

vegetation colonised latitudes north of 508 N between

115 and 110 ka. However, between southern (378 N)and northern (528 N) western European regions the

timing of the vegetational succession is a subject of

debate [16,19]. Consequently, the spatial distribution

of the vegetation belts in western Europe related with

ice growth and north eastern Atlantic SST changes

remains poorly known. For this reason, it is important

to examine additional MIS 5 records off western

Europe.

The purpose of this paper is to compare the western

European vegetation as inferred from a north–south

Fig. 1. Location of the palaeoclimatic sequences referred to in the text. (1)

Flachere, (5) Lathuile, (6) Les Echets, (7) Meikirch, (8) Beerenmfsli, (9Samerberg, (14) Zeifen, (15) Imbramowice, (16) Glowczyn, (17) Gr

distribution of vegetational belts after [54].

transect of pollen records over the last interglacial/

glacial transition (122–110 ka) and the pollen-derived

climate reconstruction, with MoBidiC simulations to

verify the role of vegetation changes as a feedback

mechanism to the glacial inception. For this, we

present new data from another Iberian margin deep

sea core, MD99-2331 (42809 N; 09841 W) retrieved

off the Galician coast (northwestern Spain), 58 far

north from core MD95-2042. This core provides an

integrated image of the vegetation at a sub-continental

scale and, therefore, the climate of a large region,

which can be more appropriate to compare with the

model outputs. On the other hand, it is strategically

located in the southern limit of the Eurosiberian

region providing a vegetational succession during

MIS 5 sharing botanical features with northern Euro-

pean and Mediterranean Eemian sequences allowing

MD95-2042, (2) MD99-2331 and SU8132, (3) Le Bouchet, (4) La

) Grande Pile, (10) Jammertal, (11) Fqramoos, (12) Eurach, (13)

fbern, (18) Bispingen, (19) Valkenweg-Amsterdam. Present-day

M.F. Sanchez Goni et al. / Earth and Planetary Science Letters 231 (2005) 111–130114

us to improve the north–south correlation of the major

chronostratigraphic units of the last interglacial com-

plex (Fig. 1). By doing this and through a detailed

direct correlation between marine and terrestrial

(pollen) stratigraphies from core MD99-2331 we will

give a consistent description of the timing and nature

of the vegetation and climate changes in western

Europe and in the middle latitudes of the eastern

North Atlantic directly related with ice volume growth

during the last interglacial/glacial transition. Further,

the floristic diversity of marine pollen spectra from

both MD99-2331 and MD95-2042 sequences has

made possible a first quantitative reconstruction of the

temperature and precipitation of western Iberia for

that period.

2. Present-day environmental setting and pollen

deposition

Core MD99-2331 was retrieved 100–200 km off

the Cape Finisterre on the slope of the continental

shelf at 2120 m water depth, close to the site of core

SU 8132 [6] (Fig. 1). This location is at present

influenced by the North Atlantic Deep Water [20,21].

The upper 500 m is driven by the Canary Current, the

Portuguese branch of the North Atlantic Drift. These

surface waters are submitted in summer to an

upwelling due to intense and persistent winds from

the north [22]. The influence of dominant Atlantic

winds over the year results in a temperate and humid

climate in northwestern Iberia with mean annual

temperature of 12.5 8C (Mean Temperature of the

Coldest Month, MTCO=6 8C; Mean Temperature of

the Warmest Month, MTWA=21 8C over the period

1956–1985) and mean annual precipitation between

1000 and 2000 mm [23,24]. This region, incised by

the Rias Baixas valleys (Galician coast basin) and the

Mino-Sil and Douro rivers (Sil and Douro basins)

[25], is at present dominated by deciduous oak

woodlands (Quercus robur, Q. pyrenaica, and Q.

petraea), heaths (Ericaceae including Calluna),

brooms (Genista) and gorses (Ulex) [26].

Experimental studies in marine sediments located

close to river mouths, as it is the case of core MD99-

2331, show that the pollen preserved in these sedi-

ments is mainly of fluvial origin [27]. This appears

particularly true in the Galician margin where the

dominant winds come from the northwestern Atlantic

Ocean [23]. Pollen grains belong to the fine-particle

fraction and consequently it is expected that pollen

and fine sediments behave similarly during the

sedimentary processes [28]. Studies on the present-

day sedimentary processes and budgets indicate that

sediments on the northwestern Iberian margin are

mainly provided by the northwestern Iberian rivers

Douro and Mino [29] and in small amounts by the

Rias. In fact, our work in progress on the present-day

pollen signature of marine sediments off northwestern

Iberia shows that the pollen content in the top

sediment samples from the Ria de Vigo, the Douro

estuary, and the north Portuguese continental shelf is

similar to that found in the top sample of MD99-2331

deep-sea core. Therefore, this suggests that the pollen

preserved in this core mainly comes from the Galician

and Douro fluvial basins. It represents the vegetation

and, therefore, the climate of northwestern Iberia.

3. Material and methods

Deep sea core MD99-2331 was retrieved by the

Marion Dufresne oceanographic ship using the

CALYPSO corer. This core is mainly composed by

hemipelagic clay. It is 37 m long and covers MIS 8 to

1 without any corebreak. Sedimentological descrip-

tion and SCOPIX radioscopic analysis of this core

indicate that no turbidite layers are present in this

sequence. Further, the interval covering the end of

Termination II does not show any bioturbated level.

MIS 5 is clearly identified in the planktic and benthic

oxygen isotopic records of the upper 22 m [30].

3.1. Chronology

The most straightforward approach to obtain a

chronological model of MD99-2331 would be to

match the benthic isotopic curve of this core with that

of the astronomically-independent record of MD95-

2042 [10]. However, the benthic data from MD99-

2331 core are too noisy when compared to those from

MD95-2042 to establish reliable stratigraphic pin

points. For this reason, we consider an alternative

age model for MD99-2331, based on the correlation

of the major climatic zones detected from its pollen

record with those dated in core MD95-2042. This

Table 1

Control points for the establishment of the age model of the

palaeoclimatic record from MD99-2331

Control points MD99-2331 MD95-2042

(Shackleton et al., 2003)

Top stadial II 1495 cm 65 ka

Top Ognon Is 1545 cm 73 ka

Top stadial I 1565 cm 74 ka

Top St Germain II 1585 cm 75.5 ka

Top Melisey II 1625 cm 81.5 ka

Top St Germain Ic 1665 cm 87 ka

Top Montaigu 1732 cm 101.5 ka

Top St Germain Ia 1742 cm 104 ka

Top Melisey I 1765 cm 107 ka

Top Eemian 1777 cm 109.7 ka

Base Eemian 1885 cm 126.1 ka

Base Stadial 1905 cm 127.5 ka

M.F. Sanchez Goni et al. / Earth and Planetary Science Letters 231 (2005) 111–130 115

assumes synchroneity of major climatic phases in

Iberia (Table 1). For example, the warming allowing

the re-expansion of temperate forest at the onset of the

Eemian must have started at the same time in the two

regions. Further, pollen, dinocysts, and planktic

foraminifera assemblages from both cores detect a

short cooling just before the Eemian. Such an abrupt

cold event was certainly correlated to southwards

incursions of polar waters in the North Atlantic Ocean

and must have also occurred simultaneously in the

two records. This pollen-based age model transferred

from the y18O MD95-2042 chronology assigns the

onset of the Eemian in core MD99-2331, 126 ka, at

level 1885 cm, and its end, 109.7 ka, at level 1777 cm.

Since the temporal resolution between samples

oscillates between 0.4 and 2 ka, we may expect such

a range in the dating uncertainty over the sequence.

3.2. Marine and terrestrial proxy analyses

Planktic (Globigerina bulloides) and benthic

isotopic measurements were performed in core

MD99-2331 over the interval from the 6/5 transition

to the present. The interval encompassing MIS 5 was

analysed every 5 cm or less [30]. Cibicides

wuellestorfi was picked up for benthic analysis.

However, the scarcity of this species in the glacial

substages of MIS 5 and its absence over the

penultimate deglaciation, obliged to analyse the

y18O on Melonis barleeanum and Bulimina exilis

species. Correcting factors used to make a continu-

ous stack are presented in [30]. The measurements,

done against NBS19 standard and then calibrated to

PDB, were made at the Department of Geology and

Oceanography of Bordeaux 1 University using an

Optima Micromass mass spectrometer.

N. pachyderma (s) foraminifers and coarse grain

fraction were counted at every 10 cm on the sand-

size fraction (N150 Am). In some key intervals such

as the deglaciation period the analysis was performed

every 5 cm.

Most of the MIS 5 interval was analysed for pollen

analysis every 10 cm. However, a 5 cm-sampling was

performed from 1860 to 1730 cm encompassing the

Last Glacial Inception interval (MIS 5e/5d transition).

The pollen preparation follows the same technique

explained in previous works for core MD95-2042

(e.g. [12]). After chemical treatment (cold HCl and

cold HF) the samples were sieved through 10 Amnylon mesh screens. The final residue for pollen

analysis was mounted unstained in double distilled

glycerine. Pollen grains were counted using a Zeiss

Axioskop light microscope at �400 and �1000 (oil

immersion) magnifications. A minimum of 100 pollen

grains excluding Pinus, due to its overrepresentation

in marine sediments [31], was counted at each of the

55 samples analysed. We estimate to twenty the

number of taxa needed in each level to have an

accurate image of the vegetation composition [32].

However, one third of the samples have a number of

taxa between 12 and 15. These poor-pollen diversified

levels, mainly concentrated in the upper part of the

MIS 5 interval, are not the focus of this paper. The

main pollen sums, including Pinus, range from 222 to

3301 per sample and a total of 73 taxa (69 pollen and

4 spore morphotypes) were identified. The pollen

percentages for the terrestrial taxa are based on a main

pollen sum which excludes Pinus, aquatics, spores,

indeterminables, and unknowns. Percentages of Iso-

etes fern have been calculated, to keep the same

criteria used for MD95-2042 record, based on the total

sum (Pollen (including Pinus) + spores + indetermin-

ables + unkowns).

3.3. Pollen-derived quantitative climatic

reconstruction

The Modern Analogue Technique (MAT) devel-

oped for continental pollen sequences [33–36], has

M.F. Sanchez Goni et al. / Earth and Planetary Science Letters 231 (2005) 111–130116

been used first on core MD95-2042 to estimate annual

precipitations and winter and summer temperatures in

southwestern Iberia during the last interglacial period.

This technique is based on a modern pollen database

including 1328 modern pollen spectra from Europe,

Eurasia, and the Mediterranean basin [37]. The

climatic parameters – mean temperature of the coldest

and the warmest months, annual precipitation – have

been calculated at each modern pollen site from

meteorological data, using an interpolation based on

an Artificial Neural Network technique (ANN). The

bioclimatic parameters selected seem to play a

determinant role on the distribution of the vegetation

and related pollen assemblages [38]. Since Pinus is

overrepresented in marine sediments, this pollen type

has been removed from the marine pollen counting as

well as from the continental pollen database. In the

standard modern analogues technique [34], similarity

between fossil and modern pollen assemblages is

evaluated by a chord distance. Usually, 5–10 modern

spectra which have the smallest chord distance are

considered as the best modern analogues of the given

pollen spectrum, and used for the reconstruction.

Their climatic parameters estimated are averaged by a

weighting inverse to the chord distance. The error is

computed as lower and upper limits of positive and

negative deviation of extreme analogues compared to

the mean value. Such method does not imply a direct

analogy between modern and fossil assemblages,

although the quality of the results depends on the

size and diversification of the modern data set. The

application of the MAT to the northernmost core

MD99-2331, out of the Mediterranean region, recon-

structs cold temperatures that are too cold in the

middle of the Eemian in contrast to the temperate

forest biome recorded by the pollen spectra. For this

reason, and to reduce uncertainties such as human

disturbance or the lack of perfect modern analogues in

the reconstruction, the MAT has been constrained by

the standardized and objective biomisation procedure

[35]. Based on the bplant functional typesQ concept,the biomisation technique attributes to each of the

modern and fossil pollen spectra a type of biome such

as tundra, taiga, temperate deciduous forest, warm

mixed forest, or steppes. The MAT cannot, therefore,

choose all the modern pollen analogues close to our

fossil spectra but only those similar corresponding to

the same biome. To reduce the bias induced by human

disturbance, we have compared the biomes obtained

from the modern pollen samples with the potential

vegetation simulated by the global vegetation model

BIOME1 [39]. Those pollen samples for which there

is no exact match between the calculated and the

potential biomes are not reliable and have been

excluded from the modern database. As the result,

the biomisation of the MD99-2331 pollen record is

based on a very recently improved modern pollen

database from Europe, Eurasia, Scandes Mountains,

Tibetan Plateau, and northern Africa composed by

868 pollen spectra [40]. The final climate estimates

are based on the modern analogues assigned to the

same biome as the fossil assemblage.

4. Vegetation and climate evolution in and off

northwestern Iberia during MIS 5

In core MD99-2331, fifteen pollen zones have

been recognised over MIS 5 (Fig. 2). Pollen zones

dominated by the percentages of tree taxa such as

deciduous Quercus, Corylus, and Carpinus betulus

reflect warm-temperate conditions in northwestern

Iberia. Among these tree-dominant zones, zones

MD31-4, MD31-8, MD31-10, and MD31-13, are also

characterised by the maximum values of Mediterra-

nean trees and shrubs, mainly evergreen Quercus.

Inversely, pollen zones marked by the development of

herbaceous taxa such as Ericaceae, Poaceae, Artemi-

sia, Chenopodiaceae, Ephedra, Taraxacum-type, and

Caryophyllaceae indicate cold conditions in this

region. Pollen zones MD31-1 and MD31-15, com-

posed by percentages of arboreal pollen between 20

and 40%, indicate open forest developed during

interstadial periods. Comparison of the MD99-2331

pollen sequence with those from western Europe [41–

43] and that off Portugal (MD95-2042) [12] allows

correlation with the major chronostratigrahic units of

the last interglacial complex: Eemian, St-Germain Ia,

St-Germain Ic, and St-Germain II warm periods as

well as the Zeifen and Ognon Interstadials. These

events alternated with the short post-Zeifen Stadial,

Melisey I, Montaigu, and Melisey II cold events, as

well as with the stadial I and II cold phases of the

Lanterne Glacial period. The Montaigu event in

northwestern Iberia appears, as everywhere in western

Europe, dominated by steppe and Pinus formations.

Fig. 2. Pollen diagram with selected taxa only from the MIS 5 interval of deep sea core MD99-2331. Grey intervals indicate cold events.

M.F.Sanchez

Goniet

al./Earth

andPlaneta

ryScien

ceLetters

231(2005)111

–130

117

M.F. Sanchez Goni et al. / Earth and Planetary Science Letters 231 (2005) 111–130118

The comparison of the MD99-2331 pollen record with

that off Portugal (MD95-2042) reflects the same

climatic variability in south and north Iberia over

MIS 5. However, a different vegetational succession is

observed between both regions. Semi-desert vegeta-

tion dominated by Artemisia-Chenopodiaceae-Ephe-

dra alternated with Mediterranean forest in the south

while in northwestern Iberia more humid open

vegetation with Ericaceae, Poaceae, and other steppe

elements alternated with deciduous oak (deciduous

Quercus)-hornbeam (C. betulus) forest. This vegeta-

tional succession in northern Iberia is rather similar to

that documented in northernmost western European

sequences.

Fig. 3. Multiproxy record for the MIS 5 interval of deep sea

In core MD99-2331, cold phases on land are

associated with heavy values of planktic y18O, likelyreflecting cold events in the ocean rather than

increases in salinity. Further, the post-Zeifen Stadial

event, the Melisey I, the Melisey II, and to a lesser

extent the Montaigu event, all characterised by the

simultaneous decrease in pioneer and temperate trees,

are contemporaneous with peaks in the polar fora-

minifera N. pachyderma (s) (Fig. 3). At the bottom of

our record cold sea surface temperatures, indicated by

the strong percentages of this foraminifera and

maximum coarse fraction concentrations, are followed

by sea surface warming marked by the decrease in N.

pachyderma (s) values. This warming is contempora-

core MD99-2331. Grey intervals indicate cold events.

M.F. Sanchez Goni et al. / Earth and Planetary Science Letters 231 (2005) 111–130 119

neous with the Zeifen interstadial which precedes the

Stadial event. Within the MIS 5, only the Melisey I

cold event and, paradoxically, the temperate Zeifen

Interstadial are associated with peaks in the coarse

grain fraction. The peak in Melisey I event is, as

previously shown by core MD95-2042 [13], related

with iceberg discharges at the time of marine cold

event C24 [30,44]. The following terrestrial cooling,

Montaigu event, is synchronous with cold event C23

[30,44]. The MD99-2331 palaeoclimatic record

detects, therefore, a climatic variability in and off

northwestern Iberia very similar to that previously

shown by European and North Atlantic records and by

the southern core MD95-2042 [12,45]. A warm-cold

succession on land and ocean preceding the Eemian,

the Zeifen Interstadial, and Stadial events, occurred at

the beginning of MIS 5e. This climatic succession has

been previously identified in core MD95-2042. The

substantial accumulation of ice in northern high

latitudes (MIS 5e/5d transition, 115–110 ka) is

contemporaneous with terrestrial and marine cooling.

In northwestern Iberia this cooling is marked by the

decline of deciduous oak-hornbeam forest. The direct

correlation between terrestrial and marine proxies

shows that the Eemian in Iberia does not correspond

with the entire MIS 5e, confirming the previous

conclusion reached from the study of the southwestern

Iberian margin core [12,14].

5. A focus on the Eemian and the Last Glacial

Inception in and off western Europe

5.1. Vegetation and sea surface temperatures

The vegetational succession during the Eemian is

similar to that already recorded from the low

resolution analysis of the twin core SU 8132 [6].

At the onset of this interglacial both cores reveal the

development of pioneer trees and, in particular,

Betula in northwestern Iberia from 126 to 125 ka

associated with the increase of deciduous Quercus.

Pioneer tree expansion at the beginning of the

Eemian is also detected in French, Alpine, German,

Dutch, and Poland sequences (pollen zones E1 and

E2) [36,41,46–48]. The varve-dated German pollen

sequences estimate the duration of this interval to

1000 yr [13] (Table 2). Southern Iberia experienced

the abrupt increase of Mediterranean plants at that

time (Fig. 4). According to our record, Mediterra-

nean plants developed ~1000 yr (subject to dating

uncertainties) later in northwestern Iberia than in the

south, i.e., roughly at the same time as the temperate

forest expanded in Europe further north and east

(pollen zones E3, E4 and first part of E5). This

scenario is compatible with the distance of the

Mediterranean refugia. The varve-dated German

sequences suggest a 2000 yr-long lasting climatic

optimum from 125 to 123 ka. SSTs from the Iberian

margin [49] and other north Atlantic cores [50,51]

exhibit the highest values during this interval related

with minimum ice volume. At ~123 ka, a first

winter cooling by 2–3 8C is documented for

northern Germany on the basis of the development

of Carpinus betulus [18]. In Iberia, the first decrease

in winter temperatures appears to be later as

revealed by the C. betulus expansion from 121 to

115 ka detected in the north and at the latitude of

core MD95-2042. This terrestrial cooling is also

recorded in northern Germany (528 N), on the basis

of the expansion of Abies-Picea coniferous forest

(second part of E5 and E6, 121–118 ka) followed by

a Pinus-Betula taiga formation (E7, 117–115 ka)

(Fig. 5). Off Iberia and in other northernmost

Atlantic regions a decrease in SST is suggested by

the increase in the planktic y18O values from ~118

ka to 108 ka although alkenone derived SSTs from

core MD95-2042 show the beginning of the

decrease earlier, at around 123 ka (Fig. 4). Chapman

and Shackleton pointed out from core NEAP18K

(52846VN, 30821VW) [52] that the most prominent

cooling of surface waters occurred at the latitudes of

Britain at around 115 ka. This event, C26, charac-

terised by temperatures as cold as those of the C24

event may imply the end of the interglacial in this

region. At similar latitudes the German pollen

sequence of Grfbern reflects the replacement of

taiga by tundra at 115 ka and confirms the end of

the interglacial in northwestern European regions at

that time (Fig. 5). Later on, Carpinus-deciduous

Quercus forest declined in northwestern Iberia

leading to the development of Abies and Pinus

between 112 and 110 ka at the same time as SST

continue to decrease off Iberia. This ocean cooling

recorded by MD99-2331 is certainly related with the

C25 cold event detected in central North Atlantic

Table 2

Vegetational succession of the Eemian interglacial at different sites located on a north–southwestern European transect

Age ka Grobern, 52ºN; 12ºE * (Kühl & Litt, 2003)

111112113 Tundra114115116117 E7 Pinus-Betula expansion118119120 E6 Picea-Abies forest with Carpinus121122 E5 Carpinus-forest123124 E4 Quercus-Corylus 125 E3 Quercetum mixtum

MD95-2042, 37ºN; 10ºW (Sanchez Goñi et al., 1999; Shackleton et al. 2003)

Deciduous Quercus-Carpinus forest

Deciduous and evergreen Quercus and Mediterranean trees with Carpinus

Deciduous and evergreen Quercus and Mediterranean trees126 E1/E2 Pioneer trees (Betula, Pinus)

MD99-2331, 42ºN; 9ºW (this work)

Pinus-Abies expansion

Carpinus-deciduous Quercus forest

Quercus-Corylus forest with Carpinus

deciduous Quercus-Corylus

Betula-deciduous Quercus

Pinus-Picea-Betula expansion

Picea-Abies forest with Carpinus

Carpinus-forest

Quercus-CorylusQuercetum mixtumPioneer trees (Betula, Pinus)

52ºN-42ºN ** (Cheddadi et al., 1998; Turner, 2002; Klotz et al., 2003; Müller et al. 2003)

Shaded interval corresponds to the timing of the first major southward migration of vegetation belts. * This scheme can be applied to Dutch sequences [47]. ** All the sequences

located between 528 and 428 N are indicated in Fig. 1.

M.F.Sanchez

Goniet

al./Earth

andPlaneta

ryScien

ceLetters

231(2005)111

–130

120

Fig. 4. Comparison of palaeoclimatic records from the interval 136–100 ka of MD95-2042 (southwestern Iberian margin) and MD99-2331

(northwestern Iberian margin) deep sea cores. Grey intervals indicate cold events.

M.F. Sanchez Goni et al. / Earth and Planetary Science Letters 231 (2005) 111–130 121

[52,44] and that off south Portugal dated at ~112 ka

[13]. The C25 cooling appears to be more subtle in

the eastern North Atlantic than far north and western

subpolar Atlantic [51]. Nevertheless, it could have

contributed to the replacement of hornbeam forest

by conifers in northwestern Iberia from 112 to 110

ka and, this is more so in the Vosges region

considering that it is 58 further north. Therefore,

the replacement of Carpinus-forest by boreal for-

mations in the latter region, recorded by Grande Pile

pollen sequence at the end of the Lure interglacial

(zone E7) [41], is likely to have occurred before 112

ka. This contradicts the hypothesis by Kukla and

collaborators [19] who propose, on the basis of

counting the putative annual laminations over the

Lure interglacial, the persistence of a well-developed

Carpinus-forest between 115 and 110 ka and the

following coniferous forest expansion leading to a

20,000-year-long interglacial. The integrity of these

laminated sequences has been questioned by Turner

[17] and, therefore, casts doubts about Kukla’s

timing of the Eemian vegetational succession in

northeastern France.

The late development of Abies in northwestern

Iberia relative to the early development in high

altitudes of the Alps (pollen zone E4) and into the

central European lowlands (pollen zone E5) would be

related to the late low summer temperature threshold

and higher precipitation [47] reached in northwestern

Iberia favouring the development of this tree. This

hypothesis is compatible with the decrease in summer

temperatures and increase of precipitation estimated

from the same pollen assemblages (Fig. 6). The record

of Abies in the southern core is later than in

northwestern Iberia and coincides with the onset of

the Melisey I cold period [12]. In this case,

quantitative estimates indicate, however, a strong

seasonality with increasing temperatures during sum-

Fig. 5. Distribution of vegetation belts at ~125 ka in western Europe, ~121 ka, ~115 ka, and ~109 ka in western Europe. Extension of

Scandinavian ice-sheets from Mangerud [64]. Shaded intervals in Iberia represent the main pollen source area of cores MD99-2331 and

MD95-2042.

M.F. Sanchez Goni et al. / Earth and Planetary Science Letters 231 (2005) 111–130122

mer (Fig. 7). Climate can influence the tree taxa not

only directly but also indirectly by changing inter-

specific composition [53]. Oak forest decline at the

beginning of Melisey I could lead to the late Abies

development in southwestern Iberia. The analysis of

cores located at different geographical location shows,

therefore, a time-transgressive development of tem-

perate tree species following directly or indirectly the

climate gradient.

In summary, from ~126 to 121 ka, the whole of

western Europe was covered by deciduous forest

concomitant with ice volume minimum (Fig. 5). A

first winter cooling in northern Germany is docu-

mented at around 123 ka. By 121–120 ka,

Fig. 6. From top to bottom: mid-June Insolation curve at 658 N from 126 to 100 ka; simulated ice volume evolution; y18O benthic and planktic

isotopic curves from core MD99-2331; simulated intensity of the NADW formation (dotted line); reconstructed mean temperatures of the

warmest and coldest months (MTWA and MTCO) and annual precipitations from pollen assemblages of core MD99-2331; and comparison with

the values simulated by MoBidiC (dotted lines) for the latitudinal band 40–458 N. Grey intervals indicate cold events. Thin solid lines indicate

the lower and upper errors for each MTWA, MTCO and Pann estimate.

M.F. Sanchez Goni et al. / Earth and Planetary Science Letters 231 (2005) 111–130 123

deciduous forest persisted in Iberia as far north as

428 N but coniferous forest began to expand in

northernmost regions (528 N). These data suggest,

therefore, that a southward displacement of the

deciduous tree-line from ~608 N, which is approx-

imately its present-day position [54], to ~508 N

occurred as early as 120 ka increasing the vegeta-

tion gradient between north and south European

latitudes. This gradient appears to be enhanced at

115 ka when substantial ice accumulated in northern

high latitudes and tundra vegetation, occupying

today regions north of 708 N, colonised northern

Germany. An oak-hornbeam forest mixed with fir

(Abies) and pine still survived in northwestern Iberia

(428 N). At this time coniferous forest likely

dominated the Vosges area (478 N).

5.2. Temperature and precipitation estimates

The application of the MAT to the interval between

136 and 100 ka of core MD95-2042 distinguishes two

climatic episodes within the Eemian (Fig. 7) in

Fig. 7. From top to bottom: mid-June Insolation curve at 658 N from 126 to 100 ka; simulated ice volume evolution; y18O benthic and planktic

isotopic curves from core MD95-2042; simulated intensity of the NADW formation (dotted line); reconstructed mean temperatures of the

warmest and coldest months (MTWA and MTCO) and annual precipitations from pollen assemblages of core MD95-2042; and comparison with

the values simulated by MoBidiC (dotted lines) for the latitudinal band 35–408 N. Grey intervals indicate cold events. Thin solid lines indicate

the lower and upper errors for each MTWA, MTCO and Pann estimate.

M.F. Sanchez Goni et al. / Earth and Planetary Science Letters 231 (2005) 111–130124

contrast with the four climatic phases suggested by the

previous qualitative climatic interpretation of the

pollen assemblage succession [12,45]. The warmest

and driest phase occurred between 126 and 117 ka

with annual precipitation around 800 mm and mean

temperature of the coldest month (MTCO) around 5

8C. These figures are similar to those observed at

present day in southwestern Iberia. In the second

phase, from 117 to 110 ka, we observe an increase of

annual precipitation by 100–200 mm and a decrease

of mean temperatures of the coldest and warmest

month (MTCO and MTWA) by 2–3 8C relative to the

early phase of the Eemian.

When we apply the same technique to the northern

core MD99-2331, winter and summer temperatures of

the middle of the Eemian (not shown) are too low,

�15 8C and �10 8C, respectively, to be associated

with the development of the C. betulus-Quercus

temperate forest biome [39] reflected by the pollen

assemblage. As explained before, we have constrained

M.F. Sanchez Goni et al. / Earth and Planetary Science Letters 231 (2005) 111–130 125

the MAT by biomisation. The new reconstruction

shows (Fig. 6) that, as in the southern core, the

warmest and driest period occurred in the first part of

the Eemian, between 126 and 118 ka. The first

cooling starts around 118 ka and is associated with

a gradual increase in annual precipitation. A 2–3000

yr lag exists between the timing of the cooling, at

~121 ka, suggested by vegetation changes in south

and north Iberia, and that estimated with the MAT.

This can be explained by the lack of sensitivity of this

technique to reconstruct climatic changes from small

shifts in vegetation composition. Between 118 and

110 ka, the annual precipitation is 200 mm higher than

that of the first part of the Eemian and the MTCO 2–3

8C lower as showed before by core MD95-2042. In

contrast, summer temperatures slightly decrease in

northwestern Iberia while an increasing trend is

suggested for the south. Both cores revealed, as

already pointed out by several authors for northern

and easternmost European regions [33,47], that

climate was drier and more continental during the

early part of the Eemian than in its latter part. A

summer cooling associated with a precipitation

increase is physically sound because it implies a

southward shift of the atmospheric fronts.

6. Data-model comparison for the western

European vegetation and climate at the Last

Glacial Inception

The 2D Earth Model of Intermediate Complexity

(EMIC), MoBidiC, is used to simulate the climate

changes between 126 and 100 ka. The model is first

initialised by performing a 15,000-yr long equili-

brium experiment, where orbital parameters are

computed for 126 ka and CO2 concentration is fixed

to 261 ppmv. The topography of the Antarctic ice

sheet is as today and Greenland is reduced to half its

present day volume. Then the transient simulation is

performed by forcing the model by the transient

variations in the seasonal and latitudinal distribution

of insolation [55] and in CO2 concentration [56]

from 126 to 100 ka. MoBidiC links a zonally

averaged atmosphere to a sectorial representation of

the surface. Each zonal band, with a 58 resolution in

latitude, is divided into sectors representing the main

continents (Eurasia and America) and the oceans

(Atlantic, Pacific, and Indian). The ocean component

is described through a primitive-equation ocean

model in which the variables are sectorially averaged

over each of the three oceans. A simple thermody-

namical–dynamical sea-ice component is coupled to

the ocean model. The dynamical terrestrial vegeta-

tion model, VECODE [57] provides the distributions

of forest, grass and desert over each continent.

MoBidiC also includes a vertically integrated model

of the three main Northern Hemisphere ice sheets

(Greenland, North America, and Eurasia) [58] and

Antarctica. This model is fully described in [59]

except that in addition, it is here synchronously

coupled to a model of the ice sheets and underlying

bedrock [60]. MoBidiC simulates the time evolution

of several variables representative of the long-term

evolution of climate, e.g. continental ice volume,

temperature, precipitation, sea-ice and snow-field

extent, energy balance at the surface, and latitudinal

heat transport. MoBidiC simulates full interglacial

conditions with no continental ice in the Northern

Hemisphere from 124 to 119 ka. Then, ice sheets

grow until 109 ka (14�106 km3). It is followed by a

partial melting of 6�106 km3 over the next 5000 yr.

The Northern Hemisphere annual mean temperature

starts to decrease 2000 yr before the re-growth of the

ice sheets. The minimum summer temperature

(14.28C at 116 ka) and maximum Arctic sea-ice

extent (117 ka) are reached well before the con-

tinental ice volume maximum. The amplitude of the

subsequent warming is smaller. At mid-latitudes,

winter temperature is almost out of phase with

summer temperature: it is maximum at 117 ka and

minimum at 106 ka. North of 758 N, the response of

the system is characterised by a cooling trend both in

winter and in summer by 4 8C. The cooling trend

characterises the whole interval but it is the more

intense before 115 ka, such that the winter temper-

ature gradient between mid- and high latitudes is

gradually enhanced during the early phase of

continental ice sheet build up (Figs. 6 and 7).

As regards the vegetation, MoBidiC simulates a

southward migration of the tree line by 178 in latitude

(from 70 to 538 N) between 126 and 116 ka. It results

from the drop in the GDD0 (growing degree days)

index north of 608 N, which is explained by the

summer cooling and a 1-month delay of the start of

the growing season at 116 ka [3]. Most of the

M.F. Sanchez Goni et al. / Earth and Planetary Science Letters 231 (2005) 111–130126

replacement of taiga by tundra northward of 608 N

takes place between 122 and 120 ka. It is associated

with the rapid increase of sea ice and land snow cover.

Southward shift of the tree line continues at the

beginning of the build-up of the ice sheets. However,

forest in the high northern latitudes (60–908 N)

recovers as soon as continental temperature starts to

increase at ~116 ka. Previous sensitivity experiments

with the same model showed that these environmental

changes were mainly controlled by precession and

that they cause an increase of surface albedo which

nearly quadruples the direct effect of astronomical

forcing [3].

In the following tentative comparison between

model and data it must be taken into account that

climate and vegetation in western Europe, as

reconstructed from the data, may not be representa-

tive for the mean climate and mean vegetation over

the whole zonal belt, as simulated by the model.

Botanical data document a major ~108 southward

displacement of vegetation belts in western Europe

from 121 ka to 115 ka rather synchronous with the

replacement of taiga by tundra simulated by MoBi-

diC from 122 to 120 ka for regions northern than

Fig. 8. MoBidiC simulation over the period 126–100 ka: (A) January Atlan

over Eurasia between 50 and 558 N. (C) Eurasia tree fraction between 50

608 N. Consequently, data support the hypothesis

that early Eemian vegetation changes in northern

high latitudes played a major role for the glacial

inception. This is in contrast with the hypothesis that

the increased SST gradient between polar and

equatorial zones and related feedbacks at 115 ka

are the major triggers for ice nucleation, as put

forward by the 3D-IPSL-CM2 model [61]. However,

MoBidiC likely underestimates the feedbacks asso-

ciated to SST cooling. After 115 ka, a gradual

opening of the oak-hornbeam forest occurred in

northwestern Iberia. At the same time, tundra

environments settled in northern Germany (Fig. 5)

agreeing with the forest decline and the decrease in

January temperatures on land simulated by MoBidiC

for mid-latitudes between 508 N and 558 N (Fig. 8).

However, this model does not simulate the drop in

SST documented for this latitudinal band.

The general trend in the estimated and simulated

winter and summer temperatures for latitudes between

35 and 408 N, follows quite straightforwardly the

precession signal. Summer temperature decreases

from 126 ka until 117 ka and then increases up to a

maximum value at 105 ka. Winter temperature is

tic SST between 50 and 558 N. (B) January land surface temperatures

and 558 N.

M.F. Sanchez Goni et al. / Earth and Planetary Science Letters 231 (2005) 111–130 127

experiencing the reverse general trend. Although the

amplitude of the signal is smaller in the data than in

the model, the good agreement between model and

data in terms of phase relationships certainly argues in

favour of the stratigraphic scale of the record. By

contrast, annual precipitation is exhibiting opposite

trend in the data and in the model.

Data from core MD99-2331 show that the warmest

winter and driest period also occurred in northwestern

Iberia between 126 and 118 ka. Temperature and

precipitation variability from the northern core are

very small in comparison with that observed from the

southern core but this may be an artefact linked to the

biomisation procedure. The agreement between

model and data for the northern core is less good

probably because it is more under regional Atlantic

influence, an aspect for which MoBidiC is not

suitably designed. Further north, from 118 to 115

ka, the January cooling derived from the Grfbernsequence (528 N) is 8–10 8C [18]. Even if ill-dated,

European pollen sequences located between 448 N

and 528 N reveal a winter cooling by 5 to10 8Crelated with the expansion of Carpinus forest in the

second part of the Eemian [36,62]. This winter

cooling in the north, associated with an increase in

precipitation, is more than twice the winter cooling,

2–3 8C, experienced in Iberia at the same time. These

temperature amplitudes, estimated from different kind

of transfer functions, indicate an enhancement of the

climate gradient between north and southwestern

European regions during the ice growth period,

which is also in agreement with the model. In

MoBidiC, this enhanced temperature gradient is

essentially due to a combination of increased winter

insolation at mid-latitudes, and increased albedo at

higher latitudes (development of ice sheets and

southward shift of tree line). In the real world, more

complex dynamical phenomena may have played a

role. For example, Felis et al. [63] recently emphas-

ised the relationship between the NAO/AO and the

insolation seasonal cycle. Their results can be

interpreted in the sense that the gradual reduction in

insolation seasonal cycle during the period 126–115

ka could have caused a gradual shift from a NAO+

like mean state towards a NAO� like mean state,

thereby explaining the gradual increase in precipita-

tion inferred from the data, but not reproduced with

MoBidiC.

Annual precipitation over land and ocean surface

temperature show several rapid shifts (110–100 ka)

that are not reproduced by MoBidiC (Figs. 6 and 7).

This concerns the succession of Melisey I (semi-desert

biome, cold SST), St. Germain Ia (forest biome, warm

SST), and Montaigu (semi-desert biome, cold SST)

events detected by terrestrial and oceanic proxies

between 110 and 100 ka. However, MoBidiC is able

to simulate rapid climate changes related to rapid

decrease in the NADW strength although the timing

of these events is not well constrained. The first

climatic change is simulated at 106 ka and, therefore

3000 yr later than the C24 cooling. After the model,

the increase in summer insolation caused the melting

of the ice sheets that grew up during isotopic stage 5d,

and consequently a change in the density of the water

column and an abrupt decrease in the strength of the

North Atlantic overturning cell, associated with shifts

in convection sites. This has little consequences on

temperatures around 408 N but it suggests that the

cold events C24 and C25 may reasonably be

explained by temporary disruptions of the convective

activity in the North Atlantic.

7. Conclusions

We have compared palaeoclimatic records from

deep sea cores MD99-2331 and MD95-2042 with

western European pollen sequences to have a better

picture of the timing and nature of vegetation changes

along a north–south transect over the last interglacial/

glacial transition. The inferred qualitative and quanti-

tative reconstructions have been compared with

MoBidiC simulations in spite of the approximations

of the climate model, uncertainties in pollen-based

climatic estimations, and the problem of establishing a

common chronology between fossil archives and

models. Remarkably, MoBidiC and data display the

maximum ice volume at the same time (110–108 ka).

Besides, quantitative climatic reconstruction (MTCO,

MTWA, and annual precipitation) for Iberia between

136 and 100 ka have been compared with those

simulated by MoBidiC for the latitudinal bands 35–

408 N and 40–458 N. This comparison shows that

simulated and reconstructed temperatures agree better

in the South than North of 408 N likely due to the fact

that northern regions are more under regional Atlantic

M.F. Sanchez Goni et al. / Earth and Planetary Science Letters 231 (2005) 111–130128

influence, an aspect for which MoBidiC is not

suitably designed. Both data and MoBidiC simula-

tions show a ~108 southward displacement of

vegetation belts in western Europe as early as ~121

ka and support the crucial role of vegetation shift in

the northernmost latitudes for triggering the last

glaciation. Further, the enhancement of vegetation

and climate gradients in western Europe related with

substantial ice growth was also synchronous with the

reconstructed and simulated settlement of perennial

snow at northern high latitudes, from ~122 to 108 ka.

Acknowledgements

IPEV and PNEDC French programs provided

financial support. We thank logistics and coring teams

on board of the R/V Marion Dufresne II during the

Geosciences oceanographic cruise and Marie-Helene

Castera, Karine Charlier, Olivier Ther, and Francoise

Vincon for invaluable technical assistance. We are

also grateful to Elsa Cortijo, Chronis Tzedakis, and an

anonymous referee for valuable suggestions which

improved the manuscript as well as to Nick Shackle-

ton, Jean-Claude Duplessy, and Sebastian Zaragosi for

discussions. Modelling results were obtained under

BIOCLIM EU project (FIKW-CT-2000-00024). We

thank the coordinator (ANDRA), the technical secre-

tary (Quantisci), and all the BIOCLIM participants.

This paper is Bordeaux 1 University, UMR-CNRS

5805 EPOC Contribution no. 1544.

References

[1] M. Yoshimori, M.C. Reader, A.J. Weaver, N.A. McFarlane,

On the causes of glacial inception at 116 ka BP, Clim. Dyn. 18

(2002) 383–402.

[2] G. Vetoretti, W.R. Peltier, Post-Eemian glacial inception, Part

I: Impact Summer Seasonal Temp. Bias J. Clim. 16 (2003)

889–911.

[3] M. Crucifix, M.F. Loutre, Transient simulations over the last

interglacial period (126–115 kyr BP): feedback and forcing

analysis, Clim. Dyn. 19 (2002) 417–433.

[4] LIGA members, Report of 1st discussion group: the last

interglacial in high latitudes of the northern hemisphere:

terrestrial and marine evidence, Quat. Int. 10–12 (1991) 9–28.

[5] J. Guiot, Back at the last interglacial, Nature 388 (1997)

25–27.

[6] J.-L. Turon, Direct land/sea correlations in the last interglacial

complex, Nature 309 (1984) 673–676.

[7] J.-L. de Beaulieu, M. Reille, The last climatic cycle at La

Grande Pile (Vosges, France). A new pollen profile, Quat. Sci.

Rev. 11 (1992) 431–438.

[8] G. Kukla, J.F. McManus, D.-D. Rousseau, I. Chuine, How

long and how stable was the last interglacial? Quat. Sci. Rev.

16 (1997) 605–612.

[9] C. Turner, R.G. West, The subdivision and zonation of

interglacial periods, Eiszeitalt. Ggw. 19 (1968) 93–101.

[10] C. Turner, Formal status and vegetational development of the

Eemian interglacial in northwestern and southern Europe,

Quat. Res. 58 (2002) 41–44.

[11] C. Turner, Volcanic maars, long quaternary sequences and the

work of the INQUA subcommission on European Quaternary

stratigraphy, Quat. Int. 47/48 (1998) 41–49.

[12] M.F. Sanchez Goni, J.-L. Turon, F. Eynaud, N.J. Shackleton,

High resolution palynological record off the Iberian margin:

direct land–sea correlation for the Last Interglacial complex,

Earth Planet. Sci. Lett. 171 (1999) 123–137.

[13] N.J. Shackleton, M. Chapman, M.F. Sanchez-Goni, D. Pailler,

Y. Lancelot, The classic marine isotope substage 5e, Quat.

Res. 58 (2002) 14–16.

[14] N.J. Shackleton, M.F. Sanchez Goni, D. Pailler, Y. Lancelot,

Marine isotope substage 5e and the Eemian interglacial, Glob.

Planet. Change 757 (2003) 1–5.

[15] G.J. Kukla, M.L. Bender, J.-L. de Beaulieu, G. Bond, W.S.

Broecker, P. Cleveringa, J.E. Gavin, T.D. Herbert, J. Imbrie, J.

Jouzel, L.D. Keigwin, K.-L. Knudsen, J.F. McManus, J.

Merkt, D.R. Muhs, H. Mqller, R.Z. Poore, S.C. Porter, G.Seret, N.J. Shackleton, C. Turner, P.C. Tzedakis, I.S. Wino-

grad, Last interglacial climates, Quat. Res. 58 (2002) 3–13.

[16] P.C. Tzedakis, Timing and duration of last interglacial

conditions in Europe: a chronicle of a changing chronology,

Quat. Sci. Rev. 22 (2003) 763–768.

[17] C. Turner, Problems of the duration of the Eemian interglacial

in Europe north of the Alps, Quat. Res. 58 (2002) 45–48.

[18] N. Kqhl, T. Litt, Quantitative time series reconstruction of

Eemian temperature at three European sites using pollen data,

Veg. Hist. Archaeobot. 13 (2003) 205–214.

[19] G.J. Kukla, J.-L. de Beaulieu, H. Svobodova, V. Andrieu-Ponel,

N. Thouveny, H. Stockhausen, Tentative correlation of pollen

records of the Last Interglacial at Grande Pile and Ribains with

Marien isotope stages, Quat. Res. 58 (2002) 32–35.

[20] H.M. Van Aken, The hydrography of the mid-latitude north-

east Atlantic Ocean: I. The deep water masses, Deep-Sea Res.

I 47 (2000) 757–788.

[21] H.M. Van Aken, The hydrography of the mid-latitudes

Northeast Atlantic Ocean: II. The intermediate water masses,

Deep-Sea Res. I 47 (2000) 789–824.

[22] W.S. Wooster, A. Bakun, D.R. McLain, The seasonal upwell-

ing cycle along the eastern boundary of the North Atlantic, J.

Mar. Res. 34 (1976) 131–141.

[23] Atlas Nacional de Espana, Climatologıa, Madrid, Ministerio

de Obras Publicas y Transportes, Direccion General del

Instituto Geografico Nacional, 1992, 24 pp.

[24] J.J. Loureiro, M.L. Machado, M.E. Macedo, M.N. Nunes, O.F.

Botelho, M.L. Sousa, M.C. Almeida, J.C. Martins, Monogra-

fias hidrologicas dos principais cursos de agua de Portugal

M.F. Sanchez Goni et al. / Earth and Planetary Science Letters 231 (2005) 111–130 129

continental. Lisboa, Direccao Geral dos Servicos Hidraulicos,

1986, 569 pp.

[25] Atlas Nacional de Espana, Hidrologıa, Madrid, Ministerio de

Obras Publicas, Transportes y Medio Ambiente, Direccion

General del Instituto Geografico Nacional, 1993, 32 pp.

[26] J. Izco, Galicia, in: M. Peinado Lorca, S. Rivas Martinez

(Eds.), La vegetacion de Espana, Universidad de Alcala de

Henares, Alcala de Henares, 1987, pp. 385–419.

[27] L. Heusser, Quaternary palynology of marine sediments in the

northeast Pacific, northwest Atlantic, and Gulf of Mexico,

Pollen records of Late-Quaternary North American sediments,

AASP Foundation, 1985, pp. 385–403.

[28] G.L. Chmura, A. Smirnov, I.D. Campbell, Pollen transport

through distributaries and depositional patterns in coastal

waters, Palaeogeogr. Palaeoclimatol. Palaeoecol. 149 (1999)

257–270.

[29] J.M.A. Dias, R. Gonzalez, C. Garcia, V. Diaz del Rio,

Sediment distribution patterns on the Galicia-Minho continen-

tal shelf, Prog. Oceanogr. 52 (2002) 215–231.

[30] A. Gouzy, B. Malaize, C. Pujol, K. Charlier, Climatic bpauseQduring termination II identified in shallow and intermediate

waters off the Iberian margin, Quat. Sci. Rev. 23 (2004)

1523–1528.

[31] J.-L. Turon, Pollen et spores dans la sedimentation actuelle le

long de la ride Rekjanes et dans la fracture de Gibbs. Interet

dans la determination des apports sedimentaires, Comptes

Rendus-Academie Des Sciences., Paris D291 (1980) 453–456.

[32] J.H. McAndrew, J.E. King, Pollen of the North American

Quaternary: the top twenty, Geosci. Man 15 (1976) 41–49.

[33] J. Guiot, A. Pons, J.L. de Beaulieu, M. Reille, A 140.000 years

continental climate reconstruction from two European pollen

records, Nature 338 (1989) 309–313.

[34] J. Guiot, Methodology of the last climatic cycle reconstruction

from pollen data, Palaeogeogr. Palaeoclimatol. Palaeoecol. 80

(1990) 49–69.

[35] J. Guiot, J.-L. de Beaulieu, R. Cheddadi, F. David, P. Ponel,

M. Reille, The climate in Western Europe during the last

glacial/interglacial cycle derived from pollen and insect

remains, Palaeogeogr. Palaeoclimatol. Palaeoecol. 103

(1993) 73–93.

[36] R. Cheddadi, K. Mamakowa, J. Guiot, J.-L. de Beaulieu, M.

Reille, V. Andrieu, W. Granoszewski, O. Peyron, Was the

climate of the Eemian stable? A quantitative climate recon-

struction from seven European pollen records, Palaeogeogr.

Palaeoclimatol. Palaeoecol. 143 (1998) 73–85.

[37] O. Peyron, J. Guiot, R. Cheddadi, P. Tarasov, M. Reille, J.-L.

de Beaulieu, S. Bottema, V. Andrieu, Climatic reconstruction

in Europe for 18,000 yr BP from pollen data, Quat. Res. 49

(1998) 183–196.

[38] I.C. Prentice, M.T. Sykes, M. Lautenschlager, S.P. Harrison,

O. Denissenko, P. Bartlein, Modelling global vegetation

patterns and terrestrial carbon storage at the last glacial

maximum, Glob. Ecol. Biogeogr. 3 (1993) 67–76.

[39] I.C. Prentice, W. Cramer, S.P. Harrison, R. Leemans, R.A.

Monserud, A.M. Solomon, A global biome model based on

plant physiology and dominance, soil properties and climate, J.

Biogeogr. 19 (1992) 117–134.

[40] O. Peyron, C. Begeot, S. Brewer, O. Heiri, M. Magny,

L. Millet, P. Ruffaldi, E. Van Campo, G. Yu, Lateglacial

climatic changes in Eastern France (Lake Lautrey) from

pollen, lake-levels, and chironomids, Quat. Res. (in

press).

[41] G.M. Woillard, Grande Pile Peat Bog: a continuous pollen

record for the last 140.000 years, Quat. Res. 9 (1978) 1–21.

[42] A. Pons, J.L. Guiot, J.L. de Beaulieu, M. Reille, Recent

contribution to the climatology of the last glacial–interglacial

cycle based on French pollen sequences, Quat. Sci. Rev. 11

(1992) 439–448.

[43] J.L. de Beaulieu, M. Reille, A long upper Pleistocene pollen

record from Les Echets, near Lyon, France, Boreas 13 (1984)

111–132.

[44] J.F. McManus, G.C. Bond, W.S. Broecker, S. Johnsen, L.

Labeyrie, S. Higgins, High-resolution climate records from the

North Atlantic during the last interglacial, Nature 371 (1994)

326–329.

[45] M.F. Sanchez Goni, F. Eynaud, J.-L. Turon, N.J. Shackleton,

O. Cayre, Direct land–sea correlation for the Eemian and its

comparison with the Holocene: a high resolution palyno-

logical record off the Iberian margin, Neth. J. Geosci. 79

(2000) 345–354.

[46] J.-L. de Beaulieu, V. Andrieu-Ponel, M. Reille, E. Grqger, P.C.Tzedakis, H. Svobodova, An attempt at correlation between

the Velay pollen sequence and the middle Pleistocene

stratigraphy from central Europe, Quat. Sci. Rev. 20 (2002)

1592–1603.

[47] W.H. Zagwijn, An analysis of Eemian climate in western and

central Europe, Quat. Sci. Rev. 15 (1996) 451–469.

[48] U.C. Mqller, J. Pross, E. Bibus, Vegetation response to rapid

climate change in central Europe during the past 140,000 yr

based on evidence from the Fqramos pollen record, Quat. Res.

59 (2003) 235–245.

[49] D. Pailler, E. Bard, High frequency palaeoceanographic

changes during the past 140 000 yr recorded by the organic

matter in sediments of the Iberian margin, Palaeogeogr.

Palaeoclimatol. Palaeoecol. 181 (2002) 431–452.

[50] E. Cortijo, J.-C. Duplessy, L. Labeyrie, H. Leclaire, J. Duprat,

T.C.E. van Weering, Eemian cooling in the Norwegian Sea and

North Atlantic Ocean preceding continental ice-sheet growth,

Nature 372 (1994) 446–449.

[51] J.F. McManus, D.W. Oppo, J.L. Cullen, G.C. Bond, Thermo-

haline circulation and prolonged interglacial warmth in the

north Atlantic, Quat. Res. 58 (2002) 17–21.

[52] M.R. Chapman, N.J. Shackleton, Global ice-volume fluctua-

tions, North Atlantic ice-rafted events, and deep-ocean

circulation changes between 130 and 70 ka, Geology 27

(1999) 795–798.

[53] H. Lischke, A.F. Lotter, A. Fischlin, Untangling a Holocene

pollen record with forest model simulations and independent

climate data, Ecol. Model. 150 (2002) 1–21.

[54] P. Ozenda, Les vegetaux dans la biosphere, Doin, Paris, 1982,

431 pp.

[55] A. Berger, Long-term variations of daily insolation and

Quaternary climatic changes, J. Atmos. Sci. 35 (1978)

2362–2367.

M.F. Sanchez Goni et al. / Earth and Planetary Science Letters 231 (2005) 111–130130

[56] J.R. Petit, J. Jouzel, D. Raynaud, N.I. Barkov, J.-M. Barnola,

I. Basile, M. Bender, J. Chappellaz, M. Davis, G. Delaygue,

M. Delmotte, V.M. Kotlyakov, M. Legrand, V.Y. Lipenkov,

C. Lorius, L. Pepin, C. Ritz, E. Saltzman, M. Stievenard,

Climate and atmospheric history of the past 420,000 years

from the Vostok ice core, Antarctica, Nature 399 (1999)

429–436.

[57] V. Brovkin, A. Ganopolski, Y. Svirezhev, A continuous

climate-vegetation classification for use in climate-biosphere

studies, Ecol. Model. 101 (1997) 251–261.

[58] H. Gallee, J.P. van Ypersele, T. Fichefet, I. Marsiat, C. Tricot,

A. Berger, Simulation of the last glacial cycle by a coupled,

sectorially averaged climate-ice sheet model: Part II. Response

to insolation and CO2 variation, J. Geophys. Res. 97 (1992)

15713–15739.

[59] M. Crucifix, M.F. Loutre, P. Tulkens, T. Fichefet, A. Berger,

Climate evolution during the Holocene: a study with an earth

system model of intermediate complexity, Clim. Dyn. 19

(2002) 43–60.

[60] M. Crucifix, A. Berger, Simulation of ocean–ice sheet

interactions during the last deglaciation, Paleoceanography

17 (2002) 1054.

[61] M. Khodri, Y. Leclainche, G. Ramstein, P. Braconnot, O.

Marti, E. Cortijo, Simulating the amplification of orbital

forcing by ocean feedbacks in the last glaciation, Nature 410

(2001) 570–574.

[62] S. Klotz, J. Guiot, V. Mosbrugger, Continental European

Eemian and early Wqrmian climate evolution: comparing

signals using different quantitative reconstruction approaches

based on pollen, Glob. Planet. Change 789 (2003) 1–18.

[63] T. Felis, G. Lohmann, H. Kunhert, S.J. Lorenz, D. Scholz, J.

P7tzold, S.A. Al-Rousan, S.M. Al-Moghrabi, Increased

seasonality in Middle East temperatures during the last

interglacial period, Nature 429 (2004) 164–168.

[64] J. Mangerud, Ice sheet limits on Norway and the Norwegian

continental shelf., in: J.G. Ehlers, P. Gibbard (Eds.), Quater-

nary glaciations—extent and chronology, Elsevier, Amster-

dam, 2004, pp. 271–294.