A icle h ps://doi.o g/10.1038/s41467-023-38337-4
Mois and wa m condi ions in Eu asia du ing
helas glacialo heMiddlePleis ocene
T ansi ion
Ma ía Fe nanda Sánchez Goñi
1,2
, Thomas Ex ie
2
,
Josué M. Polanco-Ma ínez
3,4
, Co alie Zo zi
2
, Te esa Rod igues
5
&
And é Bah
6
The end o he Middle Pleis ocene T ansi ion (MPT, ~ 800-670 housand yea s
be o e p esen , ka) was cha ac e ised by heeme genceo la geglacialice-
shee s associa ed wi h anomalously wa m No h A lan ic sea su ace em-
pe a u es enhancing mois u e p oduc ion. S ill, he di ec ion and in ensi y o
mois u e anspo ac oss Eu asia owa ds po en ial ice-shee s is poo ly con-
s ained. To econs uc la e MPT mois u e p oduc ion and dispe sal, we
combine eco ds o uppe ocean empe a u e and pollen-based Medi e a-
nean o es co e , a ace o wes e lies and p ecipi a ion, om a sub opical
d ill-co e collec ed o Sou h-Wes Ibe ia, wi h eco ds o Eas Asia summe
monsoon(EASM)s eng handWes Pacific su ace empe a u es, and model
simula ions. He e we show ha sou h-wes e n Eu opean win e p ecipi a ion
and EASM s eng h eached high le els du ing he Ma ine Iso ope S age 18
glacial. This anomalous si ua ion was caused by nea ly-con inuous mois u e
supply om bo h oceans and i s anspo o highe la i udes h ough he
wes e lies, likely uelling he accele a ed expansion o no he n hemisphe e
ice-shee s du ing he la e MPT.
The s ong non-linea beha iou o he clima e sys em in esponse o
he Milanko i ch as onomical o cing is pa icula ly well exemplified
du ing he Middle Pleis ocene T ansi ion (MPT, ~1.2–0.67 Ma) when
100,000-yea ice age cycles eme ged wi hou any significan change in
o bi al cyclici ies domina ed by he combina ion o 23,000 and
41,000-yea cycles1.Imb iee al.’s concep ual model2shows ha he
g ow h and decay o ice-shee s ampli y, h ough albedo, winds and
ocean mass p ope ies, he sys em’s ini ialmodes esponses o o bi al
o cing by changing he anspo o hea o bo eal la i udes and
a mosphe ic CO
2
concen a ion. A he end o he MPT (Ma ine Iso-
ope S age-MIS 19 o 17, ~800–670 ka), which is ma ked by he mani-
es a ion o la ge con inen al ice-shee s, pa adoxical esponses a e
obse ed wi h wa m sea su ace empe a u es (SST) in he No h
A lan ic mid-la i udes gene a ing excess mois u e ha uelled ice-
shee g ow hinEu opeand u he no h
3–5. This clima ic e olu ion
culmina ed in he maximum sou he n ex en o he Eu asian inland ice-
shee s du ing MIS 16 a ~650 ka6con empo aneously wi h a shi o
g ea e ice accumula ion in No h Ame ica7. Wi hin he end o he MPT,
MIS 18 (cen ed a ~730 ka) is a wa m glacial in he No h A lan ic
Ocean3,8and ep esen s an anomaly du ing his pe iod. Howe e , he
cause o he mild exp ession o MIS 18 emains unclea ye . I s glacial
limi s a e unknown and geomo phological da a sugges s ha he gla-
cia ion was limi ed o he high la i udes, especially in he Sou he n
Hemisphe e9. Fu he mo e, li le is known abou he Eu asian en i -
onmen s and a mosphe ic clima e du ing he final s ages o he MPT.
We hypo hesise ha he combina ion o unusual wa ming in he mid/
Recei ed: 21 No embe 2022
Accep ed: 26 Ap il 2023
Check o upda es
1
Ecole P a ique des Hau es E udes (EPHE, PSL Uni e si y), Pa is, F ance.
2
Uni . Bo deaux, CNRS, Bo deaux INP, EPOC, UMR 5805, 33600 Pessac, F ance.
3
Uni
o Excellence GECOS, IME, Uni e si y o Salamanca, 37007 Salamanca, Spain.
4
Basque Cen e o Clima e Change (BC3), 48940 Leioa, Spain.
5
Di isão de
Geologia e Geo ecu sos Ma inhos, Ins i u o Po uguês do Ma e da A mos e a, Rua Al edo Magalhães Ramalho, 6, 1495-006 Lisboa, Po ugal.
6
Ins i u e o
Ea h Sciences, Heidelbe g Uni e si y, Im Neuenheime Feld, 234, 69120 Heidelbe g, Ge many. e-mail: ma ia.sanchez-goni@u-bo deaux.
Na u e Communica ions | (2023) 14:2700 1
1234567890():,;
1234567890():,;
high-la i ude No h A lan ic, s ong SST g adien in he opical Wes
No h Pacific Ocean and in e media e glacial backg ound condi ions
should ha e enhanced long- e m mois u e p oduc ion leading o
in e glacial clima es in he sub opical la i udes o Eu asia du ing MIS
18, and hus ice accumula ion in he sou he n sec o s o Eu asia.
He e, we documen he Eu asian en i onmen s and clima es
du ing he final pa o he MPT using high- esolu ion pollen-based
ege a ion da a, a p oxy o p ecipi a ion, and alkenone-based SST
om IODP Si e U1385 loca ed in he Sou h-Wes Ibe ian ma gin
(37°34.285’N, 10°7.562’W, 2578 m dep h) combined wi h published sea
subsu ace empe a u e eco ds om he same si e3(Fig. 1). These
eco ds a e compa ed wi h geochemical10 and g ain size11,12 loess
eco ds accoun ing o he e olu ion o he Eas Asian monsoon in
No h-Eas China, and alkenone-based SST o he opical and sub-
pola Wes No h Pacific Ocean13,14 (Fig. 1). Robus s a is ical analysis
echniques applied o he paleoclima ic eco ds combined wi h iLO-
VECLIM model simula ions we e used o iden i y he main a mosphe ic
and oceanog aphic p ocesses unde lying he wa m ocean-maximum
ice g ow h pa adox a he end o he MPT (Me hods).
Resul s and discussion
In e glacial clima e in Sou h-Wes Eu ope du ing MIS 18 glacial
We quali a i ely es ima e o es co e changes using a bo eal e sus
non-a bo eal pollen pe cen ages. This simple app oach ha acknowl-
edges he non-linea ela ionship be ween plan abundances and pol-
len pe cen ages15, is suppo ed by a wide a ay o s udies on he
mode n ela ionship be ween pollen assemblages and ege a ion (e.g.,
e s. 16,17)and,specifically, by he ecen wo k o he pollen ep e-
sen a ion o he ege a ion in he Tagus basin (Sou h-Wes Ibe ia)18,
he mos impo an sou ce o pollen p ese ed a IODP Si e U1385. We
in e he s ong de elopmen o he Medi e anean o es co e ,
mainly composed o deciduous Que cus (~40–60 pollen %) and scle -
ophyllous ees and sh ubs (e e g een Que cus, Olea, Philly ea, Pis acia
and Cis us, ~10 pollen %), du ing he MIS 18d-b in e al, 740 o 725 ka,
wi h a b ie o es se back a a ound 733 ka (MIS 18c) (Fig. 2,Supple-
men a y Fig. S1). This in e al is b acke ed by he dominance o semi-
dese landscapes and hea hlands indica ing cold-d y and cold-humid
clima es du ing MIS 18e and MIS 18a, espec i ely. The o es pollen
pe cen age eached 60 % a c. 730 ka, a simila alue cha ac e ising
o he in e glacials in Sou h-Wes Ibe ia such as MIS 13 (~500 ka19), he
las in e glacial (MIS 5, ~130 ka20) and he Holocene21.Mo eo e ,ou
da a eco d all he phases cha ac e ising he succession o ege a ion
du ing in e glacial pe iods in he Medi e anean egion (Junipe us and
Be ula pionee woodlands/deciduous and e e g een Que cus-Olea-
Philly ea-Pis acia o es /deciduous and e e g een Que cus woodlands)
(Fig. S1)22. Si e U1385 indica es ha he expansion o he deciduous
oaks pa allels he de elopmen o he Medi e anean scle ophyllous
axa, wi h a synch onous maximum expansion. The pollen assem-
blages eco ded a his si e du ing he wa m phases a e e y simila o
he mode n ones in e ed om samples collec ed in he deciduous oak
woods o he Tagus basin cha ac e ised by ~10–20% o scle ophyllous
pollen axa om he he momedi e anean bel and ~40–60% o
deciduous oak pollen axa om he mesomedi e anean bel 18.No
a bo eal ees ex inc a p esen in Eu ope a e eco ded du ing his
in e al a Si e U1385. The ege a ion in e ed om he pollen
assemblages a Si e U1385 is, he e o e, e y close o he p esen -day
Medi e anean o es o Sou h-Wes Ibe ia, domina ed by b oadlea
ees, such as he oak and mixed scle ophyll o es s23.
Based on ansien model simula ions wi h ime- a ying insola ion
and a mosphe ic CO
2
concen a ions21, he s ong Medi e anean o -
es co e de elopmen eflec s high amoun o win e p ecipi a ion,
and hus ela i ely zonal and weak wes e lies di ec ed owa ds
sou he n Eu ope. This in e p e a ion is suppo ed by he high s a is-
ical co ela ion be ween he NAO index, con olling he in ensi y and
di ec ion o he wes e lies, i.e., win e p ecipi a ion, and he p esen -
day Medi e anean o es co e changes24. The mode a e alues (~10%)
o scle ophyllous plan s (e e g een Que cus, Olea, Pis acia, Philly ea
and Cis us) eflec u he mo e a weak seasonal clima e cha ac e ised
by ela i ely cold and we summe s, compa ed o o he in e glacials
and, pa icula ly, o MIS 5, ma ked by scle ophyllous plan alues
peaking a 20%. Fu he mo e, he eco d o a ew pollen g ains o he
summe -d ough in ole an ee Cas anea25 a c. 725 ka highligh s he
empe a e and humid clima e du ing MIS 18, wi h he wes e lies
a ec ing sou he n Ibe ia all o e he yea . SST a he Sou h-Wes
Ibe ian ma gin also eached he highes alues, ~18 °C, (Fig. 2)andno
subs an ial eshwa e fluxes a ec ed his ma gin du ing his in e al
as indica ed by he %C
37:4
eco d (Fig. 2). Uppe ocean wa ming o
Sou h-Wes Ibe ia du ing MIS 18 was caused by enhanced subsidence
Fig. 1 | Loca ion o he si es discussed in he ex . IODP Si e U1385 ( his s udy),
ODP Si es 980, 981, 983, 984, U1313, U1314 and DSDP 607 ( e . 4), Lake Oh id40,
Tenaghi Philippon39, Chinese Loess Pla eau selec ed sequences: Gulang10 and Pin-
gliang, Jingshuan, Baoji, Ling ai, Puxian11,12,Jingyuan
41, ODP Si e 882 ( e . 14), ODP Si e
1143 ( e . 13). NAC No h A lan ic Cu en , STG Sub opical Gy e, SPG: Subpola Gy e,
ENACW: Eas e n No h A lan ic Cen al Wa e s, KC Ku oshio Cu en , EASM Eas Asia
summe monsoon, EAWM Eas Asia win e monsoon, SH Sibe ian High, a e 78.G een
a ow: p esen -day wes e lies du ing a low index o he No h A lan ic Oscilla ion
(cen ed a ~40°−45°N49), eaching mo e sou he n la i udes du ing he glacial pe i-
ods as shown by da a and model simula ions o he Las Glacial Maximum79.
A icle h ps://doi.o g/10.1038/s41467-023-38337-4
Na u e Communica ions | (2023) 14:2700 2
o wa m Eas No h A lan ic Cen al Wa e s (ENACW) esul ing in he
gene a ion o ela i ely wa m glacial he mocline wa e s o Ibe ia3
(Fig. 2). The esul an ex ensi e uppe ocean hea ese oi p o ided a
subs an ial sou ce o hea and mois u e a ha ime o Ibe ia ha
likely a ou ed he s ong de elopmen o he egional o es co e
du ing he MIS 18 glacial.
A wa m and we inc easing end in Sou h-Wes Eu ope om
MIS 19 o MIS 17
The en i e pe iod om MIS 19 o MIS 17 was cha ac e ised by an
enhanced subduc ion o wa m mid-la i ude su ace wa e s o he
he mocline o he Ibe ian ma gin gi ing ise o sus ained high SST a
Si e U1385 ( e . 3,Fig.3). The mid-la i ude sou cing o he mocline
wa e s a Si e U1385 is illus a ed by he g adien be ween he mocline
empe a u e (T
he m
) a Si e U1385 and SST a mid-la i ude Si e U1313
(41°N) ( he ea e “T
he m
SST g adien ”), wi h a low g adien indica ing
enhanced subduc ion o wa m mid-la i ude wa e s in con as oa high
g adien indica i e o a subpola sou ce3. G adien s o ~6 °C du ing
wa m phases o MIS 19-17 a e among he lowes o he en i e in e al
MIS 44-14 ( e . 3) and p o ide e idence o he high p oduc ion o
wa m he mocline wa e s o Ibe ia. A mosphe ic mois u e p o ided
by he accumula ion o wa m sub opical wa e s in he mid-la i ude
eas e n No h A lan ic could ha e con ibu ed h ough he wes e lies
o he eco ded ex ended mild-humid and o es ed condi ions in
cen al and Sou h-Eas Eu ope om MIS 21 o 173,5. Su p isingly, he
subduc ion o wa m wa e s o he mocline le el du ing MIS 18 was
simila and e en s onge compa ed o bo h in e glacials MIS 17 and
19, espec i ely, despi e ha insola ion26, summe ene gy27,a mo-
sphe ic CO
2
concen a ions28 and ela i e sea le el29,30 we e lowe
du ing MIS 18 (Fig. 3).
Thecompa isono heglacialMIS18 o es expansionwi h hose
du ing he p e ious and succeeding in e glacials is s iking (Fig. 3). We
obse e ha he s onges Medi e anean o es de elopmen occu -
ed du ing MIS 17, cen ed a 700 ka, eflec ing a high amoun o
egional win e p ecipi a ion31.Incon as ,MIS19(~785ka),unde he
influence o bo h simila ice olume and highe a mosphe ic CO
2
Fig. 2 | Paleoclima ic eco ds o Ma ine Iso ope S age (MIS) 18 om he Ibe ian
ma gin. a insola ion a 65°N in July (black), obliqui y (blue) and p ecession ( ed)26,
summe ene gy (dashed black line), T350 defines he numbe o summe days in
which daily insola ion is abo e 350 W/m2( e . 27), bδ18O o ben hic o amini e a
om IODP Si e U1385 (blue)30, sub-s ages o MIS 18 ollow80,calkenone-based sea
su ace empe a u e (SST, ed) and pe cen ages o C
37:4
, ace o eshwa e pulses
in he Ibe ian ma gin, om IODP Si e U1385 (black, his s udy), d empe a u e
g adien be ween T he mocline om IODP Si e U1385 (37°N) and SST om IODP
Si e U1313 (41°N)3,e o al pollen concen a ions ( his s udy), semi-dese (o ange)
and hea hlands (E icaceae, blue) pollen pe cen ages, Medi e anean o es (g een)
pollen pe cen ages mainly composed o deciduous oaks and scle ophyllous
(e e g een Que cus, Pis acia, Olea, Philly ea and Cis us, yellow) ( his s udy).
A icle h ps://doi.o g/10.1038/s41467-023-38337-4
Na u e Communica ions | (2023) 14:2700 3
concen a ion, is ma ked by limi ed o es expansion32 indica ing
lowe win e p ecipi a ion compa ed o MIS 17. Mo e in e es ingly, he
MIS 18 glacial was mo e o es ed, eflec ing s onge win e ain all,
compa ed o he p eceding MIS 19 despi e ha he la e in e glacial
was cha ac e ised by highe insola ion, sea le el, a mosphe ic CO
2
concen a ions and simila wa m SST compa ed o MIS 18 (Fig. 3).
P e ious paleoceanog aphic s udies ha e shown ha p io o MIS 16
he No h A lan ic deep-wa e o ma ion, eleasing hea and humidi y
o he a mosphe e and con olling he global clima e, was loca ed
u he sou h and wes (Bo eal hea pump) compa ed o he pe iod
a e MIS 16 (No dic hea pump)33. This posi ion allowed he a i alo a
subs an ial amoun o mois u e o sou he n Eu ope du ing he in e al
MIS 19-17 bu does no explain he we e condi ions du ing he MIS 18
glacial compa ed o he MIS 19 in e glacial.
To g ound- u h he p oxy-based e idence, we u ilised simu-
la ed annual and seasonal ai empe a u es and p ecipi a ion as well
as ee ac ion om MIS 19 o 17 gene a ed wi h he iLOVECLIM
model solely o ced wi h changes in insola ion, ice-shee econ-
s uc ion and g eenhouse gas concen a ions (Figs. 3,4and S6). The
nume ical model da a in ac show high le els o win e ain all o
he h ee MISs and simila ee ac ion pe cen ages du ing MIS 18
and MIS 19 as in e ed om he pollen da a. In con as o he p oxy
da a, he simula ed ee ac ion is he weakes du ing MIS 17 (Fig. 3).
The simula ed win e ain all is he highes du ing MIS 17 bu shows
an o e all good ag eemen wi h he simula ed ee ac ion du ing
MIS 18 (Fig. 3). This misma ch be ween model and p oxy econ-
s uc ions could be explained by he di ficul y in quan i a i ely es i-
ma ing he o es co e om pollen da a15. Howe e , ou app oach
iden ifies subs an ial quali a i e o es co e and win e ain all
changes ha igh ly pa allel changes in he T
he m
SST g adien
(Fig. 3). Thus, he misma ch could be as well he esul o a eedback
p ocess ha is no well ep oduced in iLOVECLIM such as he poo
Fig. 3 | Global and Eas e n No h A lan ic paleoclima ic eco ds o he in e al
Ma ine Iso ope S age (MIS) 19-17. a insola ion a 65°N in July (black), obliqui y
(blue) and p ecession ( ed)26,bwin e La i udinal Insola ion G adien (LIG)
be ween 15°N and 65°N ( his s udy), ca mosphe ic CO
2
concen a ions28,dδ18Oo
ben hic o amini e a om IODP Si e U1385 (blue)30,a-e:sub-s ageso MIS18
80,and
ela i e sea le el changes om ODP Si e 1123 ( e . 29) compa ed o p esen -day
(black), e empe a u e g adien be ween T he mocline om IODP Si e U1385
(37°N) and sea su ace empe a u e (SST) om IODP Si e U1313 (41°N)3, alkenone-
based SST om IODP Si e U1385 ( his s udy), gsimula ed Sou h-Wes Ibe ia win e
ain all (blue), Medi e anean o es pollen pe cen ages (g een) om IODP Si e
U1385, and simula ed Sou h-Wes Ibe ia ee ac ion pe cen ages (ligh g een) ( his
s udy). Panels e,gs aigh linesindica e o dina y leas squa es fi s. Panel e hick line
indica es weigh ed a e age fi wi h a 5-sample window. G ey bands indica e he
o es ed phases ha define he e es ial in e glacials.
A icle h ps://doi.o g/10.1038/s41467-023-38337-4
Na u e Communica ions | (2023) 14:2700 4
p edic ion o he ocean he mal g adien despi e a obus SST
simula ion.
In o de o iden i y he main d i e ha may explain he inc easing
end o o es co e and win e ain all om MIS 19 o MIS 17, and
pa icula ly mo e ain all du ing MIS 18 glacial compa ed o MIS 19
in e glacial, we in es iga e ou la ge-scale econs uc ed clima e
a iables cha ac e ising he eas e n No h A lan ic (Fig. 3): (a) he
amoun o win e p ecipi a ion using he Medi e anean o es pollen
pe cen age eco d ( his s udy), (b) he posi ion and in ensi y o he
wes e lies using he win e la i udinal insola ion g adien as p oposed
by Da is and B ewe 34, (c) he posi ion o he No h A lan ic mois u e
sou ce using he T
he m
SST g adien 3, and (d) he g ow h and decay o
he no he n hemisphe e ice-shee s using he ben hic o amini e a
δ18O eco d om Si e U1385 ( e . 30) ha s ongly pa allels he LR04
δ18O s ack cu e35. Since la ge-scale econs uc ed clima e ime se ies
a e une enly spaced, we used he binned co ela ion echnique36,37 o
con e he une enly spaced paleoclima e ime se ies o e enly spaced
ime se ies (Fig. S4). To es ima e he ime-dependence o he s eng h
o associa ion be ween he la ge-scale econs uc ed ime se ies and
he Medi e anean o es pollen (MF), we used a obus Pea son co -
ela ion app oach ha akes in o accoun he se ially dependence o
he ime se ies unde s udy38 (Fig. S4).The Pea son’s co ela ion ana-
lysis shows a high indi ec co ela ion, wi h a co ela ion coe ficien
(“ ”) s a is ically significan o −0.538 [−0.735; −0.258], be ween he
Medi e anean o es and he T
he m
-SST g adien eco d o he whole
s udied in e al, MIS 17, 18, and 19. The co ela ion is also s a is ically
significan and sligh ly highe , −0586 [−0.781; −0.288] when MIS 18 is
aken indi idually. This indi ec co ela ion indica es ha he no h-
wa d displacemen end o he ENACW, weakening he T
he m
SST g adien end om MIS 19 o MIS 17, co ela es well wi h he
Fig. 4 | Paleoclima ic eco ds om Eas e n No h A lan ic ma gin and he
Chinese Loess Pla eau compa ed o subpola and opical Wes No h Pacific
sea su ace empe a u e (SST) eco ds o he in e al Ma ine Iso ope S age
(MIS) 19-17. a insola ion a 65°N in July (black), and p ecession ( ed)26,bSST eco d
om ODP Si e 882 loca ed in he subpola Wes No h Pacific Ocean14,cSST eco d
om ODP Si e 1143 loca ed in he opical Wes No h PacificOcean
13,dG ain Size
Loess (GSL) eco d om he Chinese sequences o Pingliang, Jingshuan, Baoji,
Ling ai, Puxian indica ing changes in Eas Asia win e monsoon (EAWM)11,eFe/K
eco d om he Chinese Loess Pla eau Gulang sequence indica ing changes in Eas
Asia summe monsoon (EASM) ( ed)10, and iLOVECLIM simula ed summe ain all
(g ey, his s udy— he da a ha e been smoo hed using a 10-yea unning mean),
empe a u e g adien be ween T he mocline om IODP Si e U1385 (37°N) and
SST om IODP Si e U1313 (41°N)3,gMedi e anean o es pollen pe cen ages
(g een) om IODP Si e U1385, and simula ed Sou h-Wes Ibe ia win e ain all
(blue) ( his s udy). Panels b,c, ,g: s aigh lines indica e o dina y leas squa es fi s.
Panel e, : hick lines indica e weigh ed a e age fi wi h a 23- and 5-sample windows,
espec i ely. G ey bands indica e he MIS in e glacials 19 and 17.
A icle h ps://doi.o g/10.1038/s41467-023-38337-4
Na u e Communica ions | (2023) 14:2700 5
long- e m Medi e anean o es inc ease, i.e., win e ain all, o e his
in e al. An indi ec co ela ion also exis s be ween Medi e anean
o es and ice- olume changes wi h a o –0.623 [–0.829; –0.268]. No
s a is ical significan co ela ion was ound ( =–0.178 [–0.525; 0.220])
be ween Medi e anean o es co e and win e la i udinal insola ion
g adien . The p og essi e inc ease in wa me and we e condi ions in
Sou h-Wes Ibe ia, induced by he no hwa d la i udinal shi o he
oceanic mois u e sou ce o he Ibe ian ma gin, appea s o be a dis-
inc i e ea u e o Sou h-Wes Eu ope a he end o he MPT. Simula ed
ela i e humidi y, i.e., he mois u e con en in he lowes a mosphe ic
laye (a 800 hPa), be ween Si es U1313 and U1385 shows he no h-
wa d mig a ion o he mois u e sou ce om MIS 19 o MIS 18 al hough
i does no ep oduce he no he n posi ion o he mois u e sou ce
du ing MIS 17 (Fig. S5) sugges ed om ma ine and pollen da a. Pollen
eco ds om Albania/No h Macedonia and No h-Eas G eece show
ha , while high empe a e o es co e cha ac e ised Sou h-Eas Eu -
ope du ing MIS 1839,40, no linea inc easing o es co e end is iden-
ified in his egion om MIS 19 o MIS 17 (Fig. S6).
Con empo aneous s ong mois u e p oduc ion in he No h
A lan ic and Wes No h Pacific du ing he MIS 18 glacial
In e es ingly, he long- e m end owa ds we e win e condi ions
eco ded in Sou h-Wes Ibe ia om MIS 19 o 17 (Fig. 4) pa allels he
p og essi e inc ease in he in ensi y o he Eas Asian summe mon-
soon (EASM) (Fig. 4), as in e ed om he Fe/K a io, a p oxy o
summe monsoon-induced changes, in he Gulang loess sec ion (37°
30′N, 102° 53′E, No h-Wes ma gin o he Chinese Loess Pla eau,
Fig. 1)10, bo h egions being loca ed a simila sub opical la i udes
(~35–37°N). The δ13C o he loess ca bona e om he Jingyuan
sequence (36.35°N, 104.6°E) also shows summe p ecipi a ion du ing
MIS18ashighasdu ingMIS19
41, also indica ing he in e glacial
cha ac e o MIS 18 glacial. The EASM was as s ong o s onge du ing
he weake summe insola ion and mo e glacia ed MIS 18 pe iod han
du ing he MIS 19 in e glacial, and weake compa ed o MIS 17 (Fig. 4).
The in ensi y o he Eas Asian win e monsoon (EAWM), ma ked by
cold and d y ou bu s s in he Chinese Loess Pla eau ela ed o he
de elopmen o Sibe ian Highs42, was simila o mo e in ense du ing
MIS 18 compa ed o MIS 19 as shown by he mean g ain size o qua z
pa icles (MGSQ)12 and g ain size loess (GSL) s ack eco ds om he
Chinese Loess Pla eau11 (Fig. 4and S6). Simula ed summe ain all in
Eas Asia shows an inc ease in EASM du ing MIS 18 al hough weake
compa ed o he econs uc ed loess-based EASM (Fig. 4). The amoun
o oceanic mois u e p oduc ion in he opical No h Pacificac oss he
MIS 19-17 in e al could explain he obse ed long- e m we ing end
in Eas Asia. The in ense EAWM eco ded du ing MIS 18 was associa ed
wi h SST cooling in he high la i udes14 and s ong wa ming in he
opical sec o 13 (Fig. 4and S6) causing an inc ease o he me idional
SST g adien in he Wes No h Pacific Ocean. This SST g adien
inc ease may lead o he s ong p oduc ion o wa e apou in he
Indian and Pacific opical sec o s enhancing he EASM o e No h-Eas
China43 as eflec edby heGulangandJingyuansec ions.
A p esen , a s ong SST g adien exis s o he coas o Eas Asia
du ing he ea ly summe monsoon as he esul o he p e ious win-
e ’s cold ai ou b eaks inc easing ain all o e sou he n Taiwan
(24°N)44. Du ing he midsumme , he peak wes e lies mig a e no h o
he Pla eau, he ex a opical no he lies weaken, lea ing only he
monsoon low-le el ci cula ion ha pene a es No h-Eas China,
which, coupled wi h s onge monsoonal sou he lies, leads o he
no hwa d mig a ion o he ain band44–46. The s onge EASM du ing
MIS 18 compa ed o MIS 19 in No h China would be he esul o he
pene a ion o highe amoun s o oceanic mois u e in o Eas Asia
channelled by he no hwa d mig a ion o he wes e lies. The simu-
la ed ela i e humidi y be ween ODP Si es 882 and 1143 shows indeed a
no hwa d posi ion o he mois u e sou ce du ing bo h MIS 19 and MIS
18 (Fig. S5). Du ing he MIS 18 glacial, he s ong summe p ecipi a ion
in Eas Asia was con empo aneous wi h he high mois u e p oduc ion
and win e ain all eco ded in he sub opical la i udes o he No h
A lan ic, an a mosphe ic configu a ion ha is compa ible wi h he
s ong de elopmen o he Sibe ian Highs42.
Du ing MIS 17, in e glacial condi ions associa ed wi h highe
insola ion would p oduce he in ensifica ion o he SST g adien in he
Wes No h Pacific and Indian opical sec o s and, he e o e he
inc ease o he EASM amplified by he con empo aneous inc ease o
cold ou bu s s (Fig. 4), compa ed o MIS 18. The s onge win e ain all
in Sou h-Wes Eu ope du ing MIS 17 compa ed o MIS 18 (Fig. 3and S6)
om bo h pollen and clima e simula ion would be he esul o he
nea es posi ion o he ENACW subduc ion cen e o Sou h-Wes Ibe ia
caused by he p og essi e inc ease o he no hwa d anspo o hea
and mois u e by he No h A lan ic Cu en in o he No dic Seas asso-
cia ed wi h he es ablishmen o he No dic hea pump a e ~700 ka4,33.
These esul s e eal o he fi s ime he pa adoxical in e glacial
cha ac e o MIS 18 glacial in he Eu asian sub opical la i udes, and he
ole o he oceanic mois u e p oduc ion in explaining he s ong p e-
cipi a ion du ing a glacial pe iod. No only he di ec ion and he
in ensi y o he wes e lies bu he amoun o oceanic mois u e p o-
duc ion should be aken in o accoun o explain he eco ded e olu-
ion o Sou h-Wes Eu opean p ecipi a ion and EASM. Fu he mo e,
sensi i e model expe imen s show ha du ing he in e als ma ked by
no he n ice-shee s o in e media e size, such as ou s udied in e al,
he ice g ow h a e will po en ially inc ease wi h he s eng hening o
he No h A lan ic Cu en as summe empe a u e emains a 0 °C a
he sou h o he majo ice-shee s while p ecipi a ion inc ease o e his
a ea a leas in he Eu asian sec o 4. The e o e, he nea ly con inuous
mois u e p oduc ion in he No h A lan ic and Wes No h Pacific
sub opical sec o s and i s p og essi e no hwa d anspo by he
wes e lies du ing he end o he MPT may ha e subs an ially con-
ibu ed o he s onges glacia ion o he las millions o yea s in
Eu asia and No h Ame ica du ing MIS 16, leading o he s ong
dominan 100,000-yea ice age cycles. This p ocess could also be a
wo k in p e ious mode a e glacial/in e glacial pe iods o he MPT. The
long- e m dec easing T
he m
SST g adien end in he No h A lan ic
om MIS 25 o 22 is simila o ha obse ed om MIS 19 o 173and
couldha eled o he ema kableexpansiono heNo he nHemi-
sphe e ice du ing MIS 22.
Me hods
En i onmen al se ing o IODP Si e U1385
Si e U1385 was eco e ed du ing IODP Expedi ion 339 “Medi e anean
Ou flow”. The si e is loca ed on a spu , he P omon o io dos P incipes
de A is, along he con inen al slope o he Sou h-Wes Ibe ian ma gin,
which is ele a ed abo e he abyssal plain a oiding he influence o
u bidi es30. The wa e dep h o Si e U1385 places i unde he influence
o No heas A lan ic Deep Wa e oday, al hough i was influenced by
sou he n sou ced wa e s du ing glacial pe iods47.
The Sou h-Wes Ibe ian ma gin is loca ed in he no h-eas e n
edge o he sub opical gy e, unde he influence o Eas e n No h
A lan ic cen al wa e (ENACW). The su ace wa e column is a ec ed
by he Po ugal cu en (PC), which b ings cold nu ien - ich wa e
om he no he n la i udes and o ms he ENACW o subpola o igin
(ENACWsp), and by he Azo es cu en (AC) which b ings wa m wa e
om he Azo es on gene a ing he ENACW o sub opical o igin
(ENACWs )48. The gene al dis ibu ion o wa e masses is influenced by
he seasonal mig a ion o he Azo es an icyclonic cell and i s asso-
cia ed la ge-scale wind pa e n.
Clima e in Sou h-Wes Ibe ia is di ec ly a ec ed by he in ensi y
and di ec ion o he wes e lies ha a e, in u n, con olled by he No h
A lan ic Oscilla ion (NAO)49,50. Du ing win e he No h A lan ic wes-
e lies b ing mois u e o Sou h-Wes Ibe ia, while a high-p essu e cell
de elops in he No h A lan ic du ing summe , which gene a es s ong
no he ly ade winds inducing coas al upwelling51.Thisclima e
A icle h ps://doi.o g/10.1038/s41467-023-38337-4
Na u e Communica ions | (2023) 14:2700 6
seasonali y is cha ac e ised by we and mild win e s (Tmin: 5–1°C;
Tmax: 13–8 °C) and ho and d y summe s, annual p ecipi a ion
<600 mm52, and leads o he de elopmen o a Medi e anean ege-
a ion in he adjacen landmasses domina ed by deciduous oak a
middle ele a ion, and e e g een oak, oli e ee, Pis acia, Philly ea and
ock oses (Cis us)a lowe ele a ions
18. E icaceae (hea hlands) a e
es ic ed o ainy moun ains (annual p ecipi a ion >600 mm), s ong
oceanic condi ions and loca ions wi h we soils53.
S a ig aphical amewo k
The sedimen a y sec ion eco e ed a Si e U1385 (150 m-long compo-
si e eco d) shows hemipelagic con inen al ma gin sedimen s depos-
i ed unde no mal ma ine condi ions wi h a ully oxygena ed wa e
column and a e age sedimen a ion a es o 10 cm/ky54. The s a ig aphy
o Si e U1385 was buil upon a combina ion o chemo-s a ig aphic
p oxies30. Ca/Ti a io measu ed e e y cm in all holes by co e scanning
X- ay fluo escence (XRF) was used o cons uc a composi e sec ion,
and he low esolu ion (20 cm) ben hic o amini e a oxygen iso opic
eco d (δ18O
ben h
) was co ela ed o he ma ine δ18O
ben h
s ack o LR0435
o p o ide he age model ha we p esen he e30.
Pollen-based ege a ion and clima e econs uc ion
Sedimen subsamples o 1-cm hickness and 1.5–5cm
3 olume we e
p epa ed o pollen analysis using he s anda d p o ocol o ma ine
samples (h ps://www.epoc.u-bo deaux. /index.php?lang= &page=
eq_paleo_pollens), employing coa se-sie ing a 150 μm, successi e
ea men s wi h cold HCl, cold HF a inc easing s eng h and mic o-
sie ing (10 μm mesh). Known quan i ies o Lycopodium spo es in able
o m we e added o pe mi he calcula ion o pollen concen a ions.
Slides we e p epa ed using a mobile moun ing medium o allow o a-
ion o he g ains and coun ed using a P imo S a ligh mic oscope a
400 and 1000- old magnifica ions o ou ine iden ifica ion o pollen
and spo es, espec i ely. One hund ed wen y- ou samples we e ana-
lysed e e y 4 cm (450-y a e age empo al esolu ion). Pollen coun s
oscilla e be ween 100 and 152 e es ial pollen g ains excluding Pinus
(main pollen sum), aqua ics and spo es. The main pollen sum plus
Pinus oscilla es be ween 205 and 1438. The numbe o pollen mo -
pho ypes in mos o he samples, 98 samples ou om 124, anges om
20 o 27, and om 16 o 19 mo pho ypes in he emaining samples.
Pollen pe cen ages o e es ial axa we e calcula ed agains he main
sum o e es ial g ains, while pe cen ages o Pinus we e calcula ed
agains he main sum plus Pinus. Aqua ic pollen and spo e pe cen ages
a e based on he o al sum (Pollen + spo es + inde e minables +
unknowns) (Fig. S1). The unce ain ies o he pollen pe cen ages o he
main ecological g oups a 95% ha e been calcula ed using he “exac ci”
unc ion in he P opCIs . 0.3.0 package (Package “P opCIs”). The cal-
cula iono CIsisbasedon he‘exac ’Cloppe –Pea son me hod
assuming a binomial p opo ion55,56. The a e age unce ain y o he
calcula ed pollen pe cen age o he Medi e anean o es alues in ou
analysis is less han 8% (Fig. S2). To al spo es and pollen concen a ions
oscilla e be ween 2000 and 28,000 g ains.cm−3(Fig. 2). Changes in
g ain concen a ions do no pa allel changes in pollen pe cen ages
and, he e o e, hese la e changes indica e ac ual a ia ions in o es
co e and composi ion. The in e p e a ion o he pollen diag am was
assis ed by a cons ained hie a chical clus e analysis based on Eucli-
dean dis ance be ween samples (Fig. S1). Analysis was pe o med in he
R en i onmen . 3.6.3 using he “chclus ” unc ion om he Rioja
package57.Thisanalysisiden ifies h ee main pollen zones ha we
in e p e as a s ong expansion o he o es co e b acke ed by wo
open ege a ion phases.
Sea su ace empe a u e econs uc ion and iden ifica ion o
eshwa e pulses
Sea Su ace Tempe a u e (SST) was econs uc ed using di- and i-
unsa u a ed alkenones o 37 ca bons a oms, which a e o ganic
compounds syn hesised by ma ine coccoli hopho e algae in a
empe a u e- ela ed p opo ion. Alkenones, in pa icula he e a-
unsa u a ed compounds (C
37:4
), can also be used o ack episodes o
massi e cold eshwa e inpu ( om icebe g mel ing o i e dis-
cha ges), which a e esponsible o dec easing salini ies in he su ace
wa e masses8,58. Alkenones a e pa o he o al lipid ex ac ed
(TLE) ac ion which can be ex ac ed om 2 g o sedimen by soni-
ca ion wi h dichlo ome hane and hyd olysed wi h 6% po assium
hyd oxide in me hanol. A e de i a iza ion wi h bis( ime hylsilyl) i-
fluo oace amide, he TLE was iden ified using a B uke mass spec o-
me e de ec o and quan ified wi h a Va ian gas ch oma og aph Model
3800 equipped wi h sep um p og ammable injec o and a flame
ionisa ion de ec o wi h a CPSIL-5 CB column. As a gas ca ie was used
hyd ogen a 2.5 ml.min−1. Alkenone concen a ions we e de e mined
using n-hexa iacon ane as an in e nal s anda d. Rep oducibili y es s
showed ha unce ain y in he alkenone index Uk’
37
de e mina ions
we e lowe han 0.015 ( e . 59). The Uk’
37
was calcula ed based on he
di- and i- unsa u a ed concen a ions60, and con e ed in o em-
pe a u e alues using he global co e op calib a ion o annual SST61.
The a e age empo al esolu ion be ween samples is 528 yea s and he
unce ain y o SST econs uc ion is ±0.5 °C ( e . 62).
Da a analyses
To co ela e he paleoclima e eco ds, we used fi s he binned co -
ela ion echnique36,37 o con e he une enly spaced paleoclima e
ime se ies unde analysis o e enly spaced ime se ies. This echnique
is based on a no el es ima ion app oach p oposed by e . 36 o es i-
ma ing he co ela ion be ween wo paleoclima e ime se ies wi h
di e en imescales. The idea is ha au oco ela ion means ha
memo y enables alues ob ained on di e en ime poin s o be co -
ela ed. Binned co ela ion is pe o med by esampling he une enly
spaced paleoclima e ime se ies unde analysis in o ime bins on a
egula g id, assigning he mean alues o he a iable unde sc u iny
wi hin hose bins. This me hod was ecen ly implemen ed in he R
package BINCOR63, which is eely a ailable on CRAN. To es ima e he
ime-dependence o he s eng h o associa ion be ween he la ge-
scale econs uc ed ime se ies and he Medi e anean o es pollen
(MF), we used a obus co ela ion app oach ha akes in o accoun
he se ially dependence o he ime se ies unde s udy. We ha e ol-
lowed he me hod de eloped by38 o es ima e he Pea son’s co ela-
ion coe ficien s ha akes in o accoun he se ially dependence. This
me hod also includes an es ima ion o a confidence in e al (95%)
ob ained h ough a nonpa ame ic s a iona y boo s ap echnique
wi h an a e age block leng h p opo ional o he maximum es ima ed
pe sis ence ime o he da a unde analysis. This s a is ical echnique is
implemen ed in he so wa e Pea sonT3, which is eely a ailable om
h p://www.clima e- isk-analysis.com.
iLOVECLIM model
We used he in e media e complexi y clima e model iLOVECLIM in
e sion 1.1.5 ( e . 64), which is a de elopmen b anch o he LOVECLIM
model in e sion 1.2 desc ibed in Goosse e al.65.iLOVECLIMdoesno
p esen majo di e ences in he physics o he a mosphe e and ocean
compa ed o LOVECLIM. The based componen s o his e sion a e he
a mosphe e ECBil 66, heoceanCLIO
67 and he ege a ionandland
su ace VECODE68. The modelled ee ac ion in VECODE includes
sh ub plan s such as E icaceae68.Oxygeniso opes(
18Oand16O) ha e
been implemen ed in he coupled model and e alua ed agains
obse ed da a in wa e samples and ca bona es69. An equilib ium un
was fi s pe o med o 5000 yea s unde clima ic condi ions a
800 ka. We hen an a ansien un be ween 800 and 670 ka. The
insola ion70, g eenhouse gas concen a ions28,71 and p esc ibed ice-
shee s econs uc ion72 ha ebeenupda edwi hanaccele a ion ac o
o 10 o he ansien simula ion. The model has been un a
T21 spa ial esolu ion (5.6° in longi ude and la i ude) and he ou pu
A icle h ps://doi.o g/10.1038/s41467-023-38337-4
Na u e Communica ions | (2023) 14:2700 7
a e compu ed wi h anannual imes ep and wi h a mon hly imes ep o
he p ecipi a ion only. iLOVECLIM was p e iously success ully applied
in he Asian monsoon egion o in es iga e he oxygen iso opic com-
posi ion o p ecipi a ion and calci e in associa ion wi h monsoonal
p ecipi a ion changes o e he las 150 ka73.Theδ18O in p ecipi a ion
compu ed in he model is ex ac ed o e he Eas Asia egion
(~25°N–41.5°N and ~90°E–118°E). The mean δ18O
calci e
is hen compu ed
om δ18O
p ecipi a ion
, co ec ed om ice-shee con ibu ion o he
global seawa e δ18O( e .74) and om a mosphe ic empe a u e a
2 m heigh using he ollowing Eq. (1) o syn he ic calci e unde
equilib ium condi ions75.
1000 lnαcalci e H2O
=18:03 103
T
!
32:42, ð1Þ
whe e αis he ac iona ion ac o and T he empe a u e in Kel in. The
alues a e hen con e ed om SMOW scale o VPDB scale using he
equa ion o Coplen e al. 76 (2), and smoo hed using a 10-yea unning
mean:
δ18Ocalci eVPDB =0:97002 δ18OSMOW 29:98:ð2Þ
wi h 0.97002 he iso opic ac iona ion ac o be ween VPDB and
SMOW, and 29.98 exp essed in pe mil. The ai empe a u e a 2 m
heigh , annual p ecipi a ion and ee ac ion a e also compu ed o e
Sou h-Wes Ibe ia egion (~36°N-41.5°N and ~5.6°W-11.2°W), which
includes mos o he en i e Tagus basin om whe e he majo i y o
pollen g ains a i e he o he Sou h-Wes Ibe ia ma gin sedimen s18.
Da a a ailabili y
The elemen al da a ha suppo he findings o his esea ch a e
p o ided in PANGAEA da abase (h ps://doi.o g/10.1594/PANGAEA.
957610).
Code a ailabili y
Codes o he s a is ical analyses (Binco and Pea son’sco ela ions)
a e a ailable in he Supplemen a y In o ma ion. The iLOVECLIM
sou ce code and de elopmen s a e hos ed a h p:// o ge.ipsl.jussieu.
/ludus ( e . 77) bu a e no publicly a ailable due o copy igh
es ic ions. Access can be g an ed on demand by eques o D. M.
Roche (didie . [email protected]. ) o hose who conduc esea ch in
collabo a ion wi h he iLOVECLIM use g oup.
Re e ences
1. Mudelsee,M.&S a egge ,K.Explo ing hes uc u eo hemid-
Pleis ocene e olu ion wi h ad anced me hods o ime-se ies ana-
lysis. Geol. Rundsch. 86, 499–511 (1997).
2. Imb ie, J. e al. On he s uc u e and o igin o majo glacia ion cycles
2. The 100,000-yea cycle. Paleoceanog aphy 8, 699–735 (1993).
3. Bah , A. e al. Oceanic hea pulses ueling mois u e anspo
owa ds con inen al Eu ope ac oss he mid-Pleis ocene ansi ion.
Qua . Sci. Re . 179,48–58 (2018).
4. Ba ke , S. e al. S eng hening A lan ic Inflow Ac oss he Mid-
Pleis ocene T ansi ion. Paleoceanog . Paleoclima ol. 36,
e2020PA004200 (2021).
5. Ca unda, M. C. A. e al. Subsu ace hea channel d o e sea su ace
wa ming in he high-la i ude no h a lan ic du ing he mid-
pleis ocene ansi ion. Geophys. Res. Le . 48,
e2020GL091899 (2021).
6. Ba chelo , C. L. e al. The configu a ion o No he n Hemisphe e ice
shee s h ough he Qua e na y. Na . Commun. 10,3713(2019).
7. Hodell,D.A.,Channell,J.E.T.,Cu is,J.H.,Rome o,O.E.&Röhl,U.
Onse o “Hudson S ai ”Hein ich e en s in he eas e n No h
A lan ic a he end o he middle Pleis ocene ansi ion (∼640 ka)?
Paleoceanog aphy 23, PA4218 (2008).
8. Rod igues, T. e al. A 1-Ma eco d o sea su ace empe a u e and
ex eme cooling e en s in he No h A lan ic: a pe spec i e om he
Ibe ian Ma gin. Qua . Sci. Re . 172,118–130 (2017).
9. Ehle s,J.&Gibba d,P.L.inQua e na y Encyclopaedia.1023–1031
(Else ie , 2007).
10. Sun, Y. e al. Pe sis en o bi al influence on millennial clima e
a iabili y h ough he Pleis ocene. Na . Geosci. 14,812–818 (2021).
11. Ding, Z. L. e al. S acked 2.6-Ma g ain size eco d om he Chinese
loess based on fi e sec ions and co ela ion wi h he deep-sea δ18O
eco d. Paleoceanog aphy 17,5-1–5-21 (2002).
12. Sun, Y., Clemens, S. C., An, Z. & Yu, Z. As onomical imescale
and palaeoclima ic implica ion o s acked 3.6-My monsoon
eco ds om he Chinese Loess Pla eau. Qua . Sci. Re . 25,
33–48 (2006).
13. Li, L. e al. A 4-Ma eco d o he mal e olu ion in he opical wes e n
Pacific and i s implica ions on clima e change. Ea h Plane . Sci.
Le . 309,10–20 (2011).
14. Ma ínez-Ga cia, A., Rosell-Melé, A., McClymon , E. L., Ge sonde, R.
& Haug, G. H. Subpola link o he eme gence o he mode n
equa o ial pacific cold ongue. Science 328,1550–1553 (2010).
15. Zanon, M., Da is, B. A. S., Ma que , L., B ewe , S. & Kaplan, J. O.
Eu opean Fo es Co e Du ing he Pas 12,000 Yea s: A Palynolo-
gical Recons uc ion Based on Mode n Analogs and Remo e Sen-
sing. F on . Plan Sci. 9,h ps://doi.o g/10.3389/ pls.2018.
00253 (2018).
16. Hun ley, B. & Bi ks, H. J. B. An A las o Pas and P esen Pollenmaps
o Eu ope: 0-13.000 B.P. yea s ago. (Camb idge Uni e si y
P ess, 1983).
17. P en ice, I. C. Mode n pollen spec a om lake sedimen s in Finland
and Finnma k, no h No way. Bo eas 7,131–153 (1978).
18. Mo ales-Molino, C., De aux, L., Geo ge , M., Hanquiez, V. & Sánchez
Goñi, M. F. Mode n pollen ep esen a ion o he ege a ion o he
Tagus Basin (cen al Ibe ian Peninsula). Re . Palaeobo . Palynol.
276,104193(2020).
19. Oli ei a, D. e al. Combina ion o insola ion and ice-shee o cing
d i e enhanced humidi y in no he n sub opical egions du ing
MIS 13. Qua . Sci. Re . 247,106573(2020).
20. Sánchez Goñi, M. F., Eynaud, F., Tu on, J.-L. & Shackle on, N. J. High
esolu ion palynological eco d o he Ibe ian ma gin: di ec land-
sea co ela ion o he Las In e glacial complex. Ea h Plane . Sci.
Le . 171,123–137 (1999).
21. Oli ei a, D. e al. Un a eling he o cings con olling he ege a ion
and clima e o he bes o bi al analogues o he p esen in e glacial
in SW Eu ope. Clim. Dyn. 51, 667–686 (2018).
22. ande Hammen,T.,Wijms a,T.A.&Zagwijn,W.H.inThe La e
Cenozoic Glacial Ages (ed.Tu ekian,K.K.)391–424 (Yale Uni e si y
p ess, 1971).
23. Polunin, O. & Wal e s, M. A Guide o he Vege a ion o B i ain and
Eu ope.(Ox o dUni e si yP ess,1985).
24. Gou eia,C.,T igo,R.M.,DaCama a,C.C.,Libona i,R.&Pe ei a,J.
M. C. The No h A lan ic Oscilla ion and Eu opean ege a ion
dynamics. In . J. Clima ol. 28,1835–1847 (2008).
25. Galla do-Lancho, J. F. Dis ibu ion o ches nu (Cas anea sa i a Mill.)
o es s in Spain: possible ecological c i e ia o quali y and man-
agemen ( ocusing on imbe coppices). Fo . Snow Landsc. Res. 76,
477–481 (2001).
26. Laska , J. e al. A long- e m nume ical solu ion o he insola ion
quan i ies o he Ea h. AA 428,261–285 (2004).
27. Huybe s, P. Ea ly Pleis ocene glacial cycles and he in eg a ed
summe insola ion o cing. Science,h ps://doi.o g/10.1126/
science.1125249 (2006).
28. Lu hi, D. e al. High- esolu ion ca bon dioxide concen a ion eco d
650,000-800,000[ hinsp]yea s be o e p esen . Na u e 453,
379–382, h p://www.na u e.com/na u e/jou nal/ 453/n7193/
suppin o/na u e06949_S1.h ml (2008).
A icle h ps://doi.o g/10.1038/s41467-023-38337-4
Na u e Communica ions | (2023) 14:2700 8
29. Elde field, H. e al. E olu ion o ocean empe a u e and ice olume
h ough he mid-Pleis ocene clima e ansi ion. Science 337,
704–709 (2012).
30. Hodell, D. A. e al. A e e ence ime scale o si e U1385 (Shackle on
Si e)on heIbe ianMa gin.Glob. Plane . Change 133,49–64 (2015).
31. Sánchez Goñi, M. F. e al. P onounced no hwa d shi o he wes-
e liesdu ingMIS17leading o he s ong 100-ky ice age cycles.
Ea h Plane . Sci. Le . 511, 117–129 (2019).
32. Sánchez Goñi, M. F. e al. T opically-d i en clima e shi s in sou h-
wes e n Eu ope du ing MIS 19, a low eccen ici y in e glacial. Ea h
Plane . Sci. Le . 448,81–93 (2016).
33. W igh , A. K. & Flowe , B. P. Su ace and deep ocean ci cula ion in
he subpola No h A lan ic du ing he mid-Pleis ocene e olu ion.
Paleoceanog aphy 17,20-21–20-16 (2002).
34. Da is, B. A. S. & B ewe , S. O bi al o cing and ole o he la i udinal
insola ion/ empe a u e g adien . Clim. Dyn. 32,143–165 (2009).
35. Lisiecki, L. & Raymo, M. E. A Pliocene-Pleis ocene s ack o 57
globally dis ibu ed ben hic δ18 O eco ds. Paleoceanog aphy 20,
PA1003 (2005).
36. Mudelsee, M. Clima e Time Se ies Analysis: Classical S a is ical and
Boo s ap Me hods. Fi s edi ion edn, (Sp inge -Ve lag, 2010).
37. Mudelsee, M. Clima e Time Se ies Analysis. Classical S a is ical and
Boo s ap Me hods. Second edi ion edn, (Sp inge Cham, 2014).
38. Mudelsee, M. Es ima ing Pea son’sco ela ioncoe ficien wi h
boo s ap confidence in e al om se ially dependen ime se ies.
Ma h. Geol. 35,651–665 (2003).
39. Tzedakis, P. C., Hooghiems a, H. & Pälike, H. The las 1.35 million
yea s a Tenaghi Philippon: e ised ch onos a ig aphy and long-
e m ege a ion ends. Qua .Sci.Re .25,3416–3430 (2006).
40. Wagne , B. e al. Medi e anean win e ain all in phase wi h A ican
monsoons du ing he pas 1.36 million yea s. Na u e 573,
256–260 (2019).
41. Sun, Y. e al. Di e se mani es a ions o he mid-Pleis ocene clima e
ansi ion. Na . Commun. 10, 352 (2019).
42. Wu, B. & Wang, J. Win e a c ic oscilla ion, Sibe ian High and Eas
Asian win e monsoon. Geophys. Res. Le . 29,3-1–3-4 (2002).
43. Chen, F. e al. Eas Asian summe monsoon p ecipi a ion a iabili y
since he las deglacia ion. Sci. Rep. 5, 11186 (2015).
44. Johnson, R. H. & Toy, M. D. in The Global Monsoon Sys em: Resea ch
and Fo ecas (edC.P.Change al.)3–12 (Wo ld Scien ificPublishing
Co., 2017).
45. Chiang, J. C. H., Kong, W., Wu, C. H. & Ba is i, D. S. O igins o Eas
Asian summe monsoon seasonali y. J. Clim. 33,7945–7965 (2020).
46. Yihui, D. & Chan, J. C. L. The Eas Asian summe monsoon: an
o e iew. Me eo ol.A mos.Phys.89, 117–142 (2005).
47. Shackle on, N. J., Hall, M. A. & Vincen , E. Phase ela ionships
be ween millenial scale e en s 64,000-24,000 yea s ago. Paleo-
ceanog aphy 15,565–569 (2000).
48. Ríos,A.F.,Pé ez,F.F.&F aga,F.Wa e massesin heuppe and
middle No h A lan ic Ocean eas o he Azo es. Deep Sea Res. Pa
A. Oceanog aphic Res. Pap. 39, 645–658 (1992).
49. Hu ell, J. W. Decadal ends in he No h A lan ic Oscilla ion:
egional empe a u es and p ecipi a ion. Science 269,
676–679 (1995).
50. T igo,R.M.e al.No hA lan icoscilla ioninfluence on p ecipi a-
ion, i e flow and wa e esou ces in he Ibe ian Peninsula. In . J.
Clima ol. 24,925–944 (2004).
51. Fiúza,A.F.D.G.,Macedo,M.E.D.&Gue ei o,M.R.Clima ological
space and ime a ia ion o he Po uguese coas al upwelling.
Oceanologica Ac a 5,31–40 (1982).
52. Peinado Lo ca, M. & Ma ínez-Pa as, J. M. in La ege ación de
España (eds. M. Peinado Lo ca & S. Ri as Ma ínez) 163–196 (Uni-
e sidad de Alcalá de Hena es, 1987).
53. Loidi,J.inLa ege ación en España (eds. Peinado Lo ca, M. & Ri as
Ma ínez, S.) 49–75 (Uni e sidad de Alcalá de Hena es, 1987).
54. S ow, D. A. V., He nández-Molina, F. J., Al a ez Za ikian, C. A. & and,
. E. S. in P oceedings o he IODP (P oceedings IODP, 339,Tokyo
(In eg a ed Ocean D illing P og am Managemen In e na ional,
Inc.), 2013).
55. Cloppe , C. J. & Pea son, E. S. The use o confidence o fiducial
limi s illus a ed in he case o he binomial. Biome ika 26,
404–413 (1934).
56. Sche e , R. P opCIs: Va ious Confidence In e al Me hods o P o-
po ions, R package e sion 0.3-0. h ps://CRAN.R-p ojec .o g/
package=P opCIs.(2018).
57. Juggins, S. Package “ ioja”-Analysis o Qua e na y Science Da a,
The Comp ehensi e R A chi e Ne wo k (2009).
58. Ma a , B. e al. Fou cycles o ecu ing deep and su ace wa e
des abiliza ions on he Ibe ian ma gin. Science 317,
502–507 (2007).
59. Villanue a, J., Peleje o, C. & G imal , J. O. Clean-up p ocedu es o
he unbiased es ima ion o C37-C39 alkenones sea su ace em-
pe a u es and e igenous n-alkane inpu s in paleoceanog aphy. J.
Ch oma og . 757,145–151 (1997).
60. P ahl, F. G. & Wakeham, S. G. Calib a ion o unsa u a ion pa e ns in
long-chain ke one composi ions o palaeo empe a u e assess-
men . Na u e 330,367–369 (1987).
61. Mülle , P. J., Ki s , G., Ruhland, G., on S o ch, I. & Rosell-Melé, A.
Calib a ion o he alkenone paleo empe a u e index U37K′based on
co e- ops om he eas e n Sou h A lan ic and he global ocean
(60°N-60°S). Geochim. Cosmochim. Ac a 62,1757–1772 (1998).
62. G imal , J. O., Cal o, E. & Peleje o, C. Sea su aces paleo-
empe a u es e o s in Uk'
37
es ima ion due o alkenone measu e-
men s nea he limi o de ec ion. Paleoceanog aphy 16,
226–232 (2001).
63. Polanco-Ma ínez, J., Medina-Elizalde, M., Sánchez Goñi, M. F. &
Mudelsee,M.BINCOR:AnRpackage o es ima ing heco ela ion
be ween wo une enly spaced ime se ies. R. J. 11,170–184 (2019).
64. Roche,D.M.,Dumas,C.,Bügelmaye ,M.,Cha bi ,S.&Ri z,C.
Adding a dynamical c yosphe e o iLOVECLIM ( e sion 1.0): cou-
pling wi h he GRISLI ice-shee model. Geosci. Model De . 7,
1377–1394 (2014).
65. Goosse, H. e al. Desc ip ion o he Ea h sys em model o in e -
media e complexi y LOVECLIM e sion 1.2. Geosci. Model De . 3,
603–633 (2010).
66. Ops eegh,J.D.,Haa sma,R.J.,Sel en,F.M.&Ka enbe g,A.
ECBILT: a dynamic al e na i e o mixed bounda y condi ions in
ocean models. Tellus A 50,348–367 (1998).
67. Goosse, H. & Fiche e , T. Impo ance o ice-ocean in e ac ions o
he global ocean ci cula ion: a model s udy. J. Geophys. Res. 104,
23337–23355 (1999).
68. B o kin, V., Ganopolski, A., Claussen, M., Kuba zki, C. & Pe oukho ,
V. Modelling clima e esponse o his o ical land co e change.
Glob. Ecol. Biogeog . 8,509–517, h p://www.js o .o g/s able/
2997870 (1999).
69. Roche,D.M.&Caley,T.δ18O wa e iso ope in he iLOVECLIM
model ( e sion 1.0) –Pa 2: E alua ion o model esul s agains
obse ed δ18O in wa e samples. Geosci. Model De . 6,
1493–1504 (2013).
70. Be ge , A. Long- e m a ia ions o daily insola ion and Qua e na y
clima ic changes. J. A mos. Sci. 35,2362–2367 (1978).
71. Loule gue, L. e al. O bi al and millennial-scale ea u es o a mo-
sphe ic CH4 o e he pas 800,000[ hinsp]yea s. Na u e 453,
383–386, h p://www.na u e.com/na u e/jou nal/ 453/n7193/
suppin o/na u e06950_S1.h ml (2008).
72. Ganopolski, A. & Calo , R. The ole o o bi al o cing, ca bon dioxide
and egoli h in 100 ky glacial cycles. Clima e 7,1415–1425 (2011).
73. Caley, T., Roche, D. M. & Renssen, H. O bi al Asian summe mon-
soon dynamics e ealed using an iso ope- enabled global clima e
model. Na . Commun. 5,5371(2014).
A icle h ps://doi.o g/10.1038/s41467-023-38337-4
Na u e Communica ions | (2023) 14:2700 9
