The impact of abrupt deglacial climate variability on productivity and upwelling on the southwestern Iberian margin.

The impac o ab up deglacial clima e a iabili y on p oduc i i y and upwelling on he sou hwes e n Ibe ian ma gin Blanca Ausín a , b , * , Da id A. Hodell c , Anna Cu mo e d , Timo hy I. Eglin on a a Geological Ins i u e, ETH Zü ich, Zu ich 8092, Swi ze land b Depa men o Geology, Uni e si y o Salamanca, Salamanca 37008, Spain c Godwin Labo a o y o Palaeoclima e Resea ch, Depa men o Ea h Sciences, Uni e si y o Camb idge, Camb idge CB2 3EQ, Uni ed Kingdom d Depa men o Geog aphy, Uni e si y College London, London WC1E 6BT, Uni ed Kingdom a icle in o A icle his o y: Recei ed 24 June 2019 Recei ed in e ised o m 18 Decembe 2019 Accep ed 18 Decembe 2019 A ailable online 8 Janua y 2020 Keywo ds: Paleoceanog aphy No h A lan ic Holocene Deglacia ion Hein ich s adials Nanno ossils IRD S able iso opes AMOC abs ac This s udy combines high- esolu ion eco ds o nanno ossil abundances, oxygen and ca bon s able iso- opes, co e scanning X- ay fluo escence (XRF), and ice a ed deb is (IRD) o assess he paleoceanog aphic changes ha occu ed du ing he las deglacia ion on he SW Ibe ian Ma gin. Ou esul s e eal pa allel cen ennial-scale oscilla ions in coccoli hopho e p oduc i i y, nu icline dep h and upwelling phenomena no p e iously obse ed, explained by means o a i al o icebe g-mel ing wa e s, icebe g-induced u bulen condi ions, SST changes and i e ine discha ges. On millennial ime-scales, highe p ima y p oduc i i y (PP), shallowe nu icline, and upwelling occu ence/in igo a ion a e obse ed o he Las Glacial Maximum (LGM) and Bølling-Alle ød (B/A). The opposi e scena io (i.e., lowe p oduc i i y, deepe nu icline and upwelling weakening/absence) is linked o cold spells such as Hein ich S adials 2 and 1 (HS2 and HS1) and he Younge D yas (YD). Such paleop oduc i i y a ia ions a e a ibu ed o la - i udinal mig a ions o he he mal on s associa ed wi h oceanic gy es in he No h A lan ic, in pa allel o oscilla ions in he s eng h o he A lan ic Me idional O e u ning Ci cula ion (AMOC). Mode a e- o-high PP du ing he Holocene is asc ibed o he de elopmen o he mode n seasonal su ace hyd og aphy, wi h a mo e pe sis en Ibe ian Polewa d Cu en (IPC) and seasonal wind-induced upwelling. ©2020 The Au ho s. Published by Else ie L d. This is an open access a icle unde he CC BY license (h p://c ea i ecommons.o g/licenses/by/4.0/). 1. In oduc ion Ma ine sedimen a y sequences eco e ed om he SW Ibe ian ma gin ha e played a pi o al ole in un a eling he pace, du a ion and magni ude o pas clima e changes (e.g., Ab an es e al., 2010; Hodell e al., 2013;Ma a e al., 2007;Shackle on e al., 2000). These sedimen s ha e been explo ed ex ensi ely o deciphe pas en i onmen al condi ions ha occu ed du ing ab up (millennial) clima e e en s om he LGM ho ough he las deglacia ion (e.g., Ab an es e al., 2001;Boessenkool e al., 2001;Eynaud e al., 2009; Naugh on e al., 2016;Salguei o e al., 2014;S anchez-Go~ ni e al., 1999;Sch€ on eld and Zahn, 2000). Howe e , li le is known abou he impac o ab up clima e e en s and ela ed en i onmen al condi ions on he local upwelling sys em, o which high- esolu ion p oduc i i y eco ds a e lacking and conflic ing in e p e a ions exis o specific ime pe iods. Fo ins ance, du ing Hein ich S adials (HSs), high p ima y p oduc i i y (PP) and expo p oduc i i y (Pexp) ha e been in e p e ed om p oxy eco ds, asc ibed o u bulen mixing esul ing om icebe g d i , he sou hwa d pene a ion o a nu ien - ich wa e mass and/o nu ien inpu om he shel o flu ial discha ge (Eynaud e al., 2000;Leb ei o e al., 1997;Thomson e al., 2000). Con e sely, o he same pe iod, o he p oxy eco ds ha e been in e p e ed as a educ ion in PP and Pexp, as a consequence o upwelling cessa ion linked o s ong su ace wa e s a ifica ion (Inca bona e al., 2010; Paille and Ba d, 2002;Salguei o e al., 2010;Voelke e al., 2009). Addi ionally, Salguei o e al. (2014) sugges ed he p esence o on al upwelling o explain la i udinal di e ences in Pexp along he Ibe ian ma gin du ing HSs. The Las Glacial Maximum (LGM), a pe iod o mo e mode a e s adial condi ions compa ed o HS2 p eceding and HS1 p oceeding i , is cha ac e ized by highe PP a ibu ed o in ensified ade winds and subsequen upwelling (Ab an es, 1991;Palumbo e al., 2013;Voelke e al., 2009). Few PP and Pexp econs uc ions a e a ailable o he Younge D yas (YD), al hough exis ing Ibe ian Ma gin eco ds demons a e con as ing *Co esponding au ho . Geological Ins i u e, ETH Zü ich, Zu ich 8092, Swi ze land. E-mail add ess: [email p o ec ed] (B. Ausín). Con en s lis s a ailable a ScienceDi ec Qua e na y Science Re iews jou nal homepage: www.else ie .com/loca e/quasci e h ps://doi.o g/10.1016/j.quasci e .2019.106139 0277-3791/©2020 The Au ho s. Published by Else ie L d. This is an open access a icle unde he CC BY license (h p://c ea i ecommons.o g/licenses/by/4.0/). Qua e na y Science Re iews 230 (2020) 106139 es ima ions (Inca bona e al., 2010;Palumbo e al., 2013;Salguei o e al., 2014), which a e ypically no su ficien ly esol ed o confi- den ly add ess p oduc i i y a ia ions du ing i s se e al phases (Naugh on e al., 2019). Despi e he majo i y o paleop oduc i i y econs uc ions o he las deglacia ion ocus on he s adials, high PP is obse ed o he in e s adial Bølling/Alle ød (B/A) (Inca bona e al., 2010;Paille and Ba d, 2002;Palumbo e al., 2013). Inca bona e al. (2010) hypo hesized wo non-excluding scena ios o explain inc eased PP du ing glacial in e s adials: 1) eas wa d mig a ion o he Azo es Cu en (AzC) p omo ing on al upwelling and enhancing PP and 2) ein o cemen o he no he n Hemisphe e a mosphe ic ci cula ion leading o a widening o he upwelling cell nea Cape S . Vincen (Fig. 1). Finally, he Holocene ea u es dec eased (Palumbo e al., 2013), mode a e (Salguei o e al., 2010; Voelke e al., 2009), and a iable upwelling and p oduc i i y le els, he la e p edomina ely consis ing o oligo ophy du ing he Ea ly Holocene, wi h inc eased PP du ing he La e Holocene (Palumbo e al., 2013,2019;Rod igues e al., 2009). Each o hese ab up clima e changes has been associa ed wi h shi s in he s eng h o he A lan ic Me idional O e u ning Ci - cula ion (AMOC) in he No h A lan ic (McManus e al., 2004;Ng e al., 2018), including he Ibe ian ma gin (Ghe a di e al., 2005). P oxy e idence and model simula ions sugges he weakening and/ o shu down o AMOC ia eshwa e inpu in he No h A lan ic du ing HS2, HS1 and he YD led o No he n Hemisphe ic cooling, dec eased p ecipi a ion and eo ganiza ion o he ade wind ci - cula ion (Jackson e al., 2015;Men iel e al., 2008;Naugh on e al., 2016,2019). In he ocean, AMOC educ ion has been linked o Fig. 1. Map o he s udy a ea wi h SHAK06-5K co e loca ion and mode n su ace wa e ci cula ion du ing win e . Loca ion o o he sedimen co es men ioned in he ex a e ma ked by whi e do s. PC: Po ugal Cu en . IPC: Ibe ian Polewa d Cu en . AzC: Azo es Cu en . Modified om Peliz e al. (2005). B. Ausín e al. / Qua e na y Science Re iews 230 (2020) 1061392 la i udinal mig a ions o No h A lan ic oceanic gy es (Rei b ig e al., 2019;Repschl€ age e al., 2015) and a la ge educ ion in expo p oduc ion, he la e asc ibed o shoaling o he mixed laye s and he subsequen dec eased in nu ien inpu (Men iel e al., 2008; Schmi ne , 2005). As one o he main phy oplank onic g oups, coccoli hopho es can be used o deciphe upwelling s eng h and p ima y p oduc- i i y a ia ions (Baumann e al., 2005). On he mode n-day W Ibe ian ma gin, hese calci ying algae bloom du ing he final s age o he upwelling e en , when wa ming and s a ifica ion o su ace wa e s begin and coccoli hopho es ou compe e dia oms o he emaining nu ien s (Ausín e al., 2018;Sil a e al., 2008,2009). In con as , nu ien sca ci y and s ong e ical mixing du ing he downwelling egime hampe coccoli hopho e p oli e a ion. Thus, highe accumula ion a es o calca eous coccoli hopho e emains ( e e ed o as calca eous nanno ossils) can be in e p e ed in he sedimen a y eco d as pe iods o inc eased PP (Amo e e al., 2012; Palumbo e al., 2013). P e ious in es iga ions using sedimen ap, wa e column, and su ace sedimen samples ha e assessed he ecology o ex an coccoli hopho e species in he s udy a ea (Ausín e al., 2018;Cach~ ao and Moi a, 2000;Cach~ ao e al., 2000;Fe ei a and Cach~ ao, 2005;Gue ei o e al., 2013,2014;Moi a e al., 2010; Na ciso e al., 2006;Sil a e al., 2008). In o ma ion gleaned om he la e s udies acili a e he in e p e a ion o a ia ions in he abundance o key species in he ossil eco d in e ms o he specific oceanog aphic and clima ic condi ions ha p e e en ially a o o hampe hei p oli e a ion. We examined sedimen s om co e SHAK06e5K om he SW Ibe ian ma gin, whose ch onos a ig aphy p o ides accu a e and obus age con ol on he de i ed p oxy eco ds (Ausín e al., 2019a). By combining high- esolu ion nanno ossil eco ds wi h o he sedimen ological and ino ganic geochemical da a (ca bon and oxygen iso opes, XRF and IRD coun s) om he same sedimen co e, we inc eased he esolu ion o p e ious paleop oduc i i y econs uc ions and documen he impac o apid clima e ansi- ions on PP and su ace wa e dynamics on he SW Ibe ian ma gin o e he las 28 ky . 2. Hyd og aphic condi ions The SW Ibe ian ma gin is loca ed on he no h-eas e n edge o he No h A lan ic sub opical gy e (Fig. 1). Su ace wa e s in he egion a e ba hed by he Eas e n No h A lan ic Cen al Wa e (ENACW) ( an Aken, 2001). Th oughou he yea , he Po uguese Cu en (PC), which b anches om he ela i ely cold No h A lan ic Cu en , flows sou hwa d (P e ez e al., 2001). F om Oc obe o Ma ch, he Ibe ian Polewa d Cu en (IPC), which is ed by he eas e n b anch o he wa me AzC, flows no hwa d close o he coas along he W Ibe ian ma gin (Haynes and Ba on, 1990). The egion is influenced by wo ecu en he mal on al sys ems (Peliz e al., 2005): he Sub opical F on , which is associa ed o he eas e n b anch o he AzC and fluc ua es along 35e36  N, and he Wes e n Ibe ian Win e F on , which sepa a es cold wa e s along 40-38  N. Du ing imes o IPC ci cula ion, he W Ibe ian Win e F on deflec s no hwa d along he Ibe ian ma gin sepa a ing his cu en om he PC (Peliz e al., 2005). Unde lying sub-su ace wa e s (100e250 m) consis o nu ien - poo No h A lan ic Cen al Wa e o sub opical o igin (NACWs ) (P e ez e al., 2001). The subpola a ie y o his wa e (NACWsp) flows di ec ly below o 500 m dep h anspo ing colde nu ien - ich wa e s (Fiúza e al., 1998). Be ween 500 and 1200 m, he Medi e anean Ou flow Wa e (MOW) anspo s wa me sal y wa e s o he No h A lan ic (Amba , 1983). As pa o he Cana y Eas e n Bounda y Upwelling Ecosys em, one o he wo ld ocean’s ou majo coas al upwelling sys ems, he s udy a ea sus ains high p oduc i i y (A ís egui e al., 2009). Cu - en PP is d i en by seasonal changes in he a mosphe ic sys em. F om Ma ch/Ap il o Sep embe /Oc obe , wind-induced upwelling o sub-su ace wa e s domina e he hyd og aphy, e ilizing he su ace ocean and enhancing PP (Rel as e al., 2007). F om Sep embe /Oc obe o Ma ch/Ap il, p e ailing sou hwes e ly winds p omo e coas al downwelling (Peliz e al., 2005). 3. Ma e ials and me hods Kas en co e SHAK06e5K (37  34 0 N, 10  09 0 W, 2,646 m) was eco e ed du ing C uise JC089 o he RSS James Cook c uise JC089 in 2013, and he 3.33 m-long co e was subsequen ly sec ioned a 1-cm in e als (Hodell e al., 2014). 3.1. Age-dep h model The age-dep h model o SHAK06e5K is a deposi ional model (P_sequence) based on 41 AMS 14 C da es o monospecific samples o Globige ina bulloides and buil wi h he calib a ion package Oxcal (B onk Ramsey, 2009). Con en ional adioca bon ages we e cali- b a ed using he Ma ine13 cu e, which applies a ma ine ese oi co ec ion o 400 y (Reime e al., 2013), while no local ese oi was applied. Con en ional adioca bon and calib a ed ages ha e been published elsewhe e (Ausín e al., 2019a,2019b). Acco ding o Skinne e al. (2014) ma ine ese oi ages on he Ibe ian ma gin migh ha e been g ea e o HS1 and YD, bu he e, simila o he as majo i y o p e ious wo ks ocusing on he s udy egion, a s a ic co ec ion has been applied o a o compa ison wi h nea by eco ds. Acco dingly, disc epancies a e expec ed be ween he cal- enda ages es ima ed o HS1 and YD in ou eco ds and o he published wo ks whe e di e en assump ions abou he ma ine ese oi co ec ion ha e been made (i.e., Hodell e al., 2017). The age-dep h model spans he las 28 ky and esul ing sedimen a ion a es a y be ween 10 and 30 cm/ky o mos o he co e, dec easing o 3e6 cm/ky du ing he middle and la e Holocene. A 1-cm sampling esolu ion yields an a e age empo al esolu ion o 120 y o he bo om co e ( om 328 o 88 cm) and 310 y o he mos ecen sec ion ( om 88 o 0 cm). 3.2. Nanno ossil-based p oduc i i y and en i onmen al p oxies A o al o 156 samples we e s udied o calca eous nanno ossil analyses. Samples we e p epa ed ollowing he se ling echnique o Flo es and Sie o,1997. Coun s we e pe o med wi h a Zeiss Axio pola izing mic oscope wi h a phase con as de ice a 1000 magnifica ion. To ensu e a confidence le el o 99% o species wi h a p opo ion o a leas 2% o he assemblage, a minimum o 450 specimens pe sample we e coun ed and iden ified, hen used o calcula e ela i e (%) and absolu e (coccoli hs g 1 ) abundance o each axa. Rela i e abundance is exclusi ely in e p e ed as an in- dica o o assemblage composi ion a a gi en ime. Nanno ossil accumula ion a e (NAR, gi en in coccoli hs cm 2 ka 1 )was calcula ed by conside ing he absolu e abundance o each axa, sedimen a ion a e and d y bulk densi y o he sedimen . Species iden ifica ion ollows he axonomic c i e ia o Young e al. (2003). Following o he s udies (Amo e e al., 2012; Palumbo e al., 2013), o al NAR is used as indica o o coccoli ho- pho e p oduc i i y and abundance a iabili y h ough ime. The s udy o mon hly ime-se ies o wa e samples om he Ibe ian ma gin unequi ocally linked Emiliania huxleyi o he upwelling egime (Sil a e al., 2008), pa icula ly a he end o upwelling e en s, when he wa e column s a s o wa m and s a i y (Ausín e al., 2018). Acco dingly, E. huxleyi is used as an upwelling indi- ca o . G. ape a and G. e icsonii we e combined wi hin he small B. Ausín e al. / Qua e na y Science Re iews 230 (2020) 106139 3 Gephy ocapsa g oup (<3 m m). These axa we e linked o eu ophic condi ions ela ed o enhanced upwelling (Ausín e al., 2018), and a e ypically used as an indica o o nu ien - ich wa e s in he egion (Amo e e al., 2012;Colmene o-Hidalgo e al., 2004; Palumbo e al., 2013). Simila ly, Coccoli hus pelagicus ssp. b aa udii is used as a p oxy o upwelled wa e s in he s udy a ea (Amo e e al., 2012;Cach~ ao and Moi a, 2000;Pa en e e al., 2004;Sil a e al., 2008). Ausín e al. (2018) ela ed G. oceanica o nu ien con en and wa e dep hs below 50 m. This species has ypically been used as a paleop oduc i i y p oxy (Amo e e al., 2012, and e e ences he ein). Flo isphae a p o unda is a lowe pho ic zone dwelle whose abundance in wa e samples om he NW Ibe ian ma gin has been unequi ocally linked o he downwelling egime and low-p oduc i i y condi ions (Ausín e al., 2018). A ecen global calib a ion model, based on co e- op samples, allows quan i a i e es ima ion o ne PP om F. p o unda pe cen ages in ossil co es (He n andez-Almeida e al., 2019). Howe e , he op imal in e se ela ionship be ween F. p o unda and ne PP is limi ed o he la - i udinal ange be ween 30  S and 30  N, and hus, i s ela i e abundance in ou eco ds is only in e p e ed as a quali a i e indi- ca ion o nu icline dep h and PP (Beau o and Heussne , 2001; Inca bona e al., 2010, and e e ences he ein; Ma ino e al., 2014). Gephy ocapsa muelle a e is a nu ien - ich, cold-wa e indica o (Palumbo e al., 2013;Sil a e al., 2008;Wea e and Pujol, 1988), al hough i s abundance a ies ega dless o absolu e SST changes (Inca bona e al., 2010). E. huxleyi la ge (>4 m m) has been linked o bo h, cold (Colmene o-Hidalgo e al., 2002,2004) and cold and eshe (Ausín e al., 2015;Bazzicalupo e al., 2018) wa e s, whe eas Coccoli hus pelagicus ssp. pelagicus is a ypical cold wa e p oxy (Amo e e al., 2012;Colmene o-Hidalgo e al., 2004;Palumbo e al., 2013;Pa en e e al., 2004). The axa Umbilicosphae a spp., Umbel- losphae a spp., Sy acosphae a pulch a,Ooli ho us spp. and Rhab- dosphae a spp., we e g ouped wi hin he “wa m wa e g oup” (WWG) acco ding o hei common a fini y o wa e s o sub op- ical o igin (Cach~ ao e al., 2000). The N a io was calcula ed acco ding o Flo es e al. (2000): N a io ¼small placoli hs/(small placoli hs þF. p o unda) Fi s ly, i conside s he abundance o he small placoli hs (i.e. E. huxleyi and small Gephy ocapsa), which a e ypical uppe pho ic zone (UPZ) dwelle s cha ac e is ic o high-p oduc i i y condi ions. Secondly, i eflec s he abundance o he lowe pho ic zone dwelle F. p o unda ha occupies a deep nu i- he mocline posi ion. Thus, highe N a io alues imply a deepe nu icline. Specimens ob iously pe aining o olde s a ig aphic le els a e e e ed o as “ ewo ked specimens”. Thei ela i e abundance in ela ion o he au och honous assemblage is used as an indica o o pa icle ad ec ion o he s udy si e (Fe ei a e al., 2008;Inca bona e al., 2010). 3.3. XRF co e scanning Co e SHAK06-5K was subsampled along i s leng h (344 cm) wi h h ee u-channels (plas ic co e-leng h boxes wi h 2 cm 2cm c oss sec ions and maximum 150 cm leng h) onboa d ship. Each u- channel su ace was ca e ully sc aped cleaned and co e ed wi h a 4 m m hin SPEXCe iP ep Ul alene oil o a oid con amina ion and minimize desicca ion. Semi-quan i a i e elemen al da a we e ob- ained using an A aa ech X- ay fluo escence (XRF) co e scanne a he Godwin Labo a o y in Camb idge (U.K.). Each sec ion was i adia ed wi h a hodium X- ay sou ce using a ube cu en o 0.2 mA a h ee di e en ol ages: 10 kV (kV) wi h no fil e , 30 kV using a hin lead fil e , and 50 kV using a coppe fil e . XRF da a we e collec ed e e y 5 mm along he en i e leng h o each u- channel. The leng h and wid h o he i adia ed su ace was 2.5 mm (down-co e) and 12 mm (c oss-co e), espec i ely, wi h a coun ime o 60 s. The Canbe a WinAxil so wa e was used wi h s an- da d so wa e se ings and spec um-fi models o ob ain elemen in ensi ies, which depend on he elemen concen a ions. Resul s a e p esen ed in log a ios o elemen in ensi ies, which mos accu a ely eflec s changes in chemical composi ion (Wel je and Tjallingii, 2008). 3.4. IRD coun s Ice Ra ed Deb is (IRD) coun s we e pe o med using a ligh mic oscope a 2-cm esolu ion in he sedimen ac ion >300 m m om he dep h in e als encompassing HS2 (308-262 cm), HS1 (170-114 cm), and YD (96-78 cm). De i al ca bona e g ains we e coun ed as a sub-ca ego y o he o al li hic g ains. None o he samples we e spli . IRD concen a ion is exp essed in g ains pe d y ac ion weigh (g ains g 1 ;>300 m m). 3.5. Oxygen and ca bon s able iso ope eco ds High- esolu ion ca bon ( d 13 C) and oxygen ( d 18 O) s able iso ope eco ds o he plank onic o amini e G. bulloides om co e SHAK06e5K ha e been p esen ed elsewhe e along wi h de ails on sample p epa a ion and analyses (Ausín e al., 2019a). Bo h eco ds e eal cen ennial- o millennial-scale oscilla ions linked o he majo ab up clima e e en s o he las deglacia ion (eg., HS2 and HS1, B/A, and YD, espec i ely). The plank onic d 18 O a ia ions eflec changes in sea su ace empe a u e (SST), ice- olume and salini y. Plank onic o amini e a d 13 C is assumed o eflec he ca bon iso opic composi ion o dissol ed ino ganic ca bon ( d 13 C DIC ) as well as ‘ i al’e ec s, which include kine ic iso ope ac iona ion ha esul s in iso opic disequilib ium (Spe o,1992). Because d 13 C DIC depends on local (i.e. pho osyn hesis and espi a ion, changes in ad ec ion and upwelling, calcifica ion a e) and global ca bon cycle changes, d 13 C shell eflec s mul iple pa ame e s (Ra elo and Hillai e- Ma cel, 2007). A gi en loca ions whe e upwelling s ongly impac s su ace local hyd og aphy, d 13 C is o en in e p e ed as a p oxy o upwelling (P asanna e al., 2016;Voelke e al., 2009;Zhang e al., 2017). Acco dingly, we use d 13 C, along wi h o he paleop oduc i i y and nu icline dep h indica o s, o in e p e a ia ions in he up- welling dynamics. 4. Resul s 4.1. Nanno ossil abundances P ese a ion o specimens was good o mode a e acco ding o he isual c i e ia es ablished by Flo es and Ma ino (2002). Rela i e abundance and NAR o specific axa used as paleoceanog aphic p oxies in his s udy (see sec ion 3.2.1) a e shown in Figs. 2 and 3. E. huxleyi and he small Gephy ocapsa g oup a e he mos abundan axa h oughou he s udied pe iod, comp ising up o 80% and 50% o he assemblage, espec i ely (Fig. 2). These wo species a e p esen in high p opo ions du ing HS2 and HS1. Ne e heless, only G. oceanica,G. muelle ae,E. huxleyi (>4 m m), C. pelagicus ssp. pelagicus and F. p o unda inc ease hei ela i e abundance du ing ei he one o bo h cold pe iods. The pe cen age con ibu ion o F. p o unda shows a double-peak s uc u e du ing bo h HSs, comp ising up o 7% o he assemblage du ing HS2, and 20% du ing HS1 (Fig. 2). Rela i e maxima o WWG (up o 2.8%) and ewo ked coccoli hs (~13% du ing HS2 and ~19% du ing HS1) a e obse ed a 25.5, 25, 17.5 and 16.6 ka (Fig. 3). The LGM is again domina ed by he small placoli hs o E. huxleyi and small Gephy ocapsa (50% and 30% on a e age, espec i ely), bu B. Ausín e al. / Qua e na y Science Re iews 230 (2020) 1061394 Fig. 2. Nanno ossil-based p oduc i i y indica o s. Nanno ossil accumula ion a e (NAR [coccoli hs cm 2 ky 1 ]; filled cu es) and ela i e abundance ([%]; g ey lines) o coccoli- hopho e species. The d 18 O eco d is shown as e e ence o iden i ying ab up clima e e en s, indica ed by colou ed ba s. Da ke blue ba s indica e pe iods o shallowe nu icline, s onge upwelling and he a i al o icebe g-mel ing wa e du ing HSs (see sec ion 5.1.3). The dashed line sepa a es wo dis inc phases o he YD. Red diamonds ma k age con ol poin s (ky cal. BP). (Fo in e p e a ion o he e e ences o colou in his figu e legend, he eade is e e ed o he Web e sion o his a icle.) B. Ausín e al. / Qua e na y Science Re iews 230 (2020) 106139 5 Fig. 3. Nanno ossil-based cold and wa m wa e indica o s and ewo ked specimens. Nanno ossil accumula ion a e (NAR [coccoli hs cm 2 ky 1 ]; filled cu es) and ela i e abundance ([%]; g ey lines) o coccoli hopho e species. WWG: wa m wa e g oup. The d 18 O eco d is shown as e e ence o iden i ying ab up clima e e en s, indica ed by colou ed ba s. Da ke blue ba s indica e pe iods o shallowe nu icline, s onge upwelling and he a i al o icebe g-mel ing wa e du ing HSs (see sec ion 5.1.3). The dashed line sepa a es wo dis inc phases o he YD. Red diamonds ma k age con ol poin s (ky cal. BP). (Fo in e p e a ion o he e e ences o colou in his figu e legend, he eade is e e ed o he Web e sion o his a icle.) B. Ausín e al. / Qua e na y Science Re iews 230 (2020) 1061396 he assemblage is a iable h oughou his pe iod. Fo ins ance, la ge pe cen age con ibu ion o Coccoli hus pelagicus ssp. b aa - udii (~1.4%) and G. oceanica (~5%) a e obse ed om 23.5 o 21.5 ka (Fig. 2). Be ween 21.5 and 19.5 ka, only Coccoli hus pelagicus ssp. b aa udii exhibi s a clea inc ease in ela i e abundance (up o 2%). F om 19.5 o 18.5 ky , G. muelle a e,E. huxleyi (>4 m m) and G. oceanica show highe NAR and pe cen ages, whe eas smalls peaks o WWG a e obse ed h oughou he en i e LGM. Peaks up o 1.8% o Coccoli hus pelagicus ssp. b aa udii a e obse ed du ing he B/A, a pe iod ma ked by a highe ela i e abundance o small Gephy ocapsa (up o 35%) and he WWG (~4%). The fi s hal o he YD (YDa; 13.2 o 12.5 ka) is cha ac e ized by a dec ease in he ela i e abundance o mos species and a la ge inc ease in E. huxleyi, which eaches up o 70% a 12.7 ka (Fig. 2). G. oceanica and F. p o unda ea u e an inc easing end du ing he second hal o he YD (YDb; 12.5 o 11.4 ka), comp ising o 4% and 10%, espec i ely, o he assemblage a he end o he in e al. The WWG cons i u es 4% o he assemblage du ing YDb (Fig. 3). The Holocene assemblage is mainly composed o E. huxleyi (70% on a e age) wi h lesse bu significan con ibu ions o F. p o unda (20%), WWG (5% on a e age), and G. oceanica (3%). The ela i e abundance o ewo ked specimens has been p esen ed elsewhe e (Magill e al., 2018). Highe pe cen ages a e obse ed om 28 o 12.5 ka, especially du ing HS2 (~10%) and HS1 (~16%) and he fi s hal o he YD (4%), and dec ease o minimum om 12.5 ka onwa ds (Fig. 3). In gene al, o al NAR displays highe alues du ing he LGM, B/A and mos o he la e Holocene (las 3.7 ky ) (Fig. 4c). Th ee la ge peaks a e obse ed a 22.1 ka, 20.7 ka, and 19 ka o he LGM. The Holocene is ma ked by low alues un il 3.7 ka, om which high a iabili y and la ge peaks a e obse ed un il p esen . NAR o he majo i y o he axa ollow his same pa e n, wi h he ollowing excep ions: G. muelle a e and C. pelagicus ssp. pelagicus a e gene - ally mo e abundan om 28 o 12.5 ka and show li le inc ease om 3.7 ka onwa d, while F. p o unda and he WWG also show a sig- nifican inc ease in NAR along he Holocene in ela ion o he deglacia ion (Figs. 2 and 3). This ime span is cha ac e ized by he nea absence o G. muelle ae and E. huxleyi (>4 m m) (Fig. 3), and lowe NAR o small Gephy ocapsa g oup, C. pelagicus ssp. b aa udii (Fig. 2) and C. pelagicus spp. pelagicus (Fig. 3). O he axa such as E. huxleyi,G. oceanica,F. p o unda and he WWG inc ease hei NAR and a iabili y du ing he las 3.7 ky . The N a io exhibi s a double-peak dec ease du ing bo h HSs, which is mo e p onounced du ing HS1 (Fig. 4d). Maximum alues a e obse ed du ing he LGM, B/A and YDa. A gene al dec ease in N a io is obse ed om he onse o YDb ho ough he Holocene, eaching a minimum a 3 ka wi h a ela i e eco e y o high alues he ea e . 4.2. XRF and IRD coun s The log (Ca/Ti) was selec ed o iden i y a ia ions in %CaCO 3 (Hodell e al., 2013,2015), whe eas log (Ca/S ) allows he ela i e p opo ion o de i al ma e ial (dolomi e and limes one) o be dis inguished om biogenic ca bona e (Hodell and Cu is, 2008; Hodell e al., 2017). The Ca/Ti eco d shows a gene al inc easing end ac oss he las deglacia ion wi h he lowes alues occu ing in ea ly HS1, ollowed by a peak in he middle o HS1 (~17 ka), dec easing again du ing he la e pa o HS1 (Fig. 4g). Ca/Ti alues inc ease du ing he B/A, ollowed by a decline du ing he YD be o e peaking in he ea ly Holocene. In con as , Ca/S is ela i ely low h oughou he eco d wi h he excep ion o a p ominen inc ease cen e ed a 17 ka (Fig. 4h). Bo h Ca/S and Ca/Ti eco ds show a conspicuous peak du ing HS1 a 17 ka. Ca/S also shows a sligh peak du ing HS2 a 24.5 ka ha is no exp essed in he Ca/Ti signal. Li hic g ain coun s we e a ge ed in he in e als sugges i e o IRD deli e y on he basis o XRF esul s. These esul s demons a e significan amoun s o IRD om 25.2 o 24.3 ka du ing HS2 and 17.5 o 16 ka du ing HS1, eaching absolu e maximum alues a 17 ka and 25 ka, espec i ely (Fig. 4i). No IRD was ound du ing he YD e en . 5. Discussion 5.1. Hein ich s adials 2 and 1 5.1.1. Gene al condi ions HS2 and HS1 a e cha ac e ized by minimum coccoli hopho e p oduc i i y, a deepe nu icline (Fig. 4), and a gene al dec ease in Ca/Ti (Fig. 4) indica ing a la ge inpu o de i al sedimen (Hodell e al., 2013,2015). Excluding addi ional e ec s, significan in- c eases in d 13 Co G. bulloides a e in e p e ed as nu ien sca ci y in he uppe wa e column and hence dec eased upwelling o deepe wa e s (Ausín e al., 2019a;Voelke e al., 2009). Such condi ions we e likely o hinde no only coccoli hopho e p oduc i i y, bu o e all PP and Pexp, as in e ed by p e ious s udies in his egion (Inca bona e al., 2010;Salguei o e al., 2010;Voelke e al., 2009) and o he no he n loca ions o he No h A lan ic (Na e e al., 2007). I is likely ha a ia ions in he upwelling sys em we e p ima ily induced by oceanic (i.e, hyd ologic changes) a he han a mosphe ic o cing (e.g., p essu e cen e dynamics and subse- quen blowing winds). Hyd ologic changes in he physical s uc u e o he uppe wa e column, such as s ong s a ifica ion, ha e been in e ed om he a i al o cold, less-saline wa e s (Voelke e al., 2009). Mel ing icebe gs de i ed om he no he n ice-shee s deli e ed mel wa e and ice a ed deb is o he s udy egion (Ba d e al., 2000;Cay e e al., 1999;Eynaud e al., 2009;Voelke and de Ab eu, 2011). Such condi ions a e co obo a ed by he p esence o IRD coinciding wi h d 18 O dec eases o bo h HS2 and HS1 (Fig. 4). Acco ding o Fe ei a e al. (2008) and p e ious wo k unde aken in co e SHAK06e5K (Magill e al., 2018), he la ge con ibu ions o ewo ked specimens obse ed a hese imes a e he esul o inc eased la e al anspo o fine-sil size (<10 m m) sedimen s in ela ion o in e media e nepheloid laye s associa ed wi h inc eased MOW anspo du ing HS2 and HS1 (Fig. 3). 5.1.2. Di e ing impac on p oduc i i y and hyd og aphy be ween HS2 and HS1 Du ing HS1, a deepe nu icline posi ion and educed/sup- p essed upwelling is demons a ed in compa ison o HS2 (Fig. 4). These condi ions esul om he la ge impac o HS1 on he egional hyd og aphy, as indica ed by g ea e IRD abundances and a la ge d 18 O dec ease due o a mo e significan eshwa e inpu , which poin o a g ea e in ensi y and du a ion o HS1 compa ed o HS2. Simila ly, significan deli e y o de i al ca bona e da diag- nos ic ea u e o Hein ich laye s (Hodell and Cu is, 2008;Hodell e al., 2017)dis only isible du ing HS1. Such condi ions induced subs an ial di e ences in nanno ossil assemblages be ween bo h HS e en s. Fo ins ance, small Gephy ocapsa and G. muelle ae la gely con ibu e o he assemblage du ing HS2, indica ing mo e eu ophic condi ions owa ds he end o he pe iod. In con as , high pe cen ages o E. huxleyi (>4 m m) and C. pelagicus ssp. pelagicus a e only obse ed du ing HS1. Bo h species a e indica o s o sub- pola wa e and a e known o peak du ing Hein ich e en s o he Ibe ian ma gin (Na ciso e al., 2006;Pa en e e al., 2004) suppo - ing a g ea e e ec o icebe g mel ing a his ime accompanied by a la ge SST dec ease (Salguei o e al., 2014). These ea u es, along wi h a majo sou hwa d mig a ion o he pola on du ing HS1 (Eynaud e al., 2009) may ha e posed an ecological ba ie o he de elopmen o ce ain species. B. Ausín e al. / Qua e na y Science Re iews 230 (2020) 106139 7 5.1.3. Cen ennial p oduc i i y a ia ions du ing HSs Some o he p oxy- eco ds exhibi a bimodal dis ibu ion du ing bo h HSs ha dese es u he a en ion. Double peaks o F. p o unda pe cen ages led o dec eases in he N a io, coinciding wi h highe d 13 C excu sions cen e ed a 25.6 and 24.7 ka o HS2, and a 18 and 16 ka o HS1 (Figs. 2 and 4d and e), indica ing la ge cen ennial a ia ions in upwelling in ensi y and nu icline dep h. The in e ening in e als a e cha ac e ized by he a i al o icebe g and mel wa e , indica ed by IRD and XRF peaks and a sligh dec ease in d 18 O(Fig. 4 ei). The IRD and XRF peaks a 17 ka ( 14 C age Fig. 4. P oxy eco ds om co e SHAK06e5K. (a) Sedimen a ion a e. (b) Absolu e coccoli h abundance. (c) To al Nanno ossil Accumula ion Ra e. (d) N a io. (e) Ca bon and ( ) oxygen s able iso opes om Ausín e al. (2019a). (g) Ca/Ti and (h) Ca/S XRF log a ios. (i) To al (black line) and ca bona e (blue line) Ice Ra ed Deb is. Da ke blue ba s indica e pe iods o shallowe nu icline, s onge upwelling and he a i al o icebe g-mel ing wa e du ing HSs. Red do s ma k coe al cen ennial a ia ions in nu icline dep h and up- welling in ensi y du ing HSs. The dashed line sepa a es wo dis inc phases o he YD. Red diamonds ma k age con ol poin s (ky cal. BP). (Fo in e p e a ion o he e e ences o colou in his figu e legend, he eade is e e ed o he Web e sion o his a icle.) B. Ausín e al. / Qua e na y Science Re iews 230 (2020) 1061398 o 14,089 ±101) likely co espond o hose iden ified in he No h A lan ic a 16 ka cal. BP (co esponding 14 C age o 13,913 ±30) by Hodell e al. (2017, and e e ences he ein) and e med H1.1. The age disc epancy in he calenda ages esul om he di e en ese oi ages applied in each case (1200 y in Hodell e al. (2017) and a s a ic ese oi o 400 y in his s udy). Double peaks in upwelling weakening and nu icline deepening a e a ibu ed o he influence o mel wa e s igh be o e and a e maximum icebe g p esence. F eshe su ace wa e s a e expec ed o p omo e a s onge halo- cline, supp ess wind-induced upwelling, and inhibi PP and Pexp (Inca bona e al., 2010;Paille and Ba d, 2002;Salguei o e al., 2010; Voelke e al., 2009), which is consis en wi h he o e all low coccoli hopho e p oduc i i y obse ed in ou eco ds (Fig. 4c). Du ing maximum icebe g in usion (25 and 17 ka), howe e , a ela i e nu icline shoaling and upwelling in ensifica ion is obse ed (da k blue ba s in Fig. 4d and e), in con as wi h he lowes coccoli hopho e p oduc i i y obse ed in ou eco ds (Fig. 4c). Tu bulen mixing due o icebe g d i (Eynaud e al., 2000; Leb ei o e al., 1997;Thomson e al., 2000) migh accoun o such a ia ions in he nu icline dep h and upwelling in ensi y, whe eas i is likely ha o e all coccoli hopho e p oduc i i y was hampe ed by he a i al o colde and eshe wa e s, in which only E. huxleyi (>4 m m) and C. pelagicus ssp. pelagicus would find op imal g ow h condi ions (Fig. 3). Gi en ha ce ain p ima y p oduce s (e.g. di- noflagella es and dia oms) can ou compe e coccoli hopho es unde medium and high u bulen condi ions (Villama~ na e al., 2019), he p oposed scena io allows he econciling o p e ious con as ing p oduc i i y in e p e a ions o he s udy a ea: low PP and Pexp in e ed om coccoli hopho e, plank onic o amini e a, alkenone accumula ion and CaCO 3 da a (Inca bona e al., 2010;Paille and Ba d, 2002;Salguei o e al., 2010;Voelke e al., 2009), and highe PP and Pexp in e p e ed om dinoflagella e cys s and plank onic o amini e a species (Eynaud e al., 2000;Leb ei o e al., 1997;Salguei o e al., 2014). Pa allel cen ennial oscilla ions in ewo ked specimens and WWG also dese e u he a en ion. Deepe posi ion o he MOW flow co e, deduced om pe cen age inc eases in ewo ked specimens in co e SHAK06-5K (Magill e al., 2018), may ha e enabled he eas wa d displacemen o AzC (Inca bona e al., 2010, and e e ences he ein). Disc e e peaks in WWG a hese imes would hen be explained by he sho -las ing influence o oligo ophic and wa me sub opical wa e s. 5.2. The LGM Coccoli hopho e p oduc i i y inc eased significan ly du ing he LGM, linked o a shallow nu icline and an o e all in ensifica ion o upwelling (Fig. 4cee). Such condi ions sugges an inc ease in ma- ine p oduc i i y o W Ibe ia, as p e iously shown om bo h coccoli hopho e (Palumbo e al., 2013) and dia om ossil eco ds (Ab an es, 2000), as well as om Pexp es ima ed om a plank onic o amini e a-based ans e unc ion (Salguei o e al., 2010;Voelke e al., 2009). Rela i ely s able and mild SSTs p e ailed du ing his pe iod (Da euil e al., 2016), albei wi h empe a u es low enough o p omo e p oli e a ion o cold-wa e species. Coccoli hopho e p oduc i i y and upwelling in ensi y show la ge cen ennial- millennial oscilla ions ha a e supe imposed on hese gene al ends. Fo ins ance, NAR shows h ee dis inc peaks cen e ed a 22.1 ka, 20.7 ka, and 19 ka, which a e he esul o la ge changes in sedimen a ion a e along wi h concomi an inc eases in coccoli h abundance (coccoli h g 1 )(Fig. 4aec). Each o hese peaks is seen in he NAR o E. huxleyi and small Gephy ocapsa, sugges ing high p oduc i i y a hose imes. Ne e heless, changes in he assem- blage also indica e empo ally a ying en i onmen al condi ions. Small peaks o WWG along he en i e pe iod indica e he seasonal incu sion o wa me , sub opical wa e s, possibly ela ed o a Paleo- IPC (Salguei o e al., 2014;Voelke e al., 2009). Mo e eu ophic condi ions, in e ed om he la ge pe cen age con ibu ions o small Gephy ocapsa,G. oceanica, and Coccoli hus pelagicus ssp. b aa udii, a e main ained o he fi s pa o he pe iod un il 21 ka (Fig. 2). A his ime, a ansi ion o colde wa e s is in e ed om he la ge inc ease in NAR and ela i e abundance o classical cold- wa e indica o s G. muelle a e and C. pelagicus ssp. pelagicus (Fig. 3). The majo inc ease in E. huxleyi (>4 m m) abundance commencing a 19 ka indica es he influence o no only cold bu eshe condi ions (Fig. 3). Concomi an peaks o his axa in Wes e n Medi e anean co es (Ausín e al., 2015;Bazzicalupo e al., 2018;Flo es e al., 2010a) ha e been linked o majo flu ial discha ge om No h Eu ope glacie s (Bazzicalupo e al., 2018). Ou high- esolu ion e- co d suppo s massi e glacie uno om he B i ish Islands and Scandina ia in o he No h A lan ic (Bazzicalupo e al., 2018), flowing down o he s udy a ea and en e ing he Medi e anean h ough he S ai o Gib al a . 5.3. The Bølling/Alle ød The B/A was ma ked by high coccoli hopho e p oduc i i y and a shallow nu icline linked o inc eased upwelling (Fig. 4cee). Mo e eu ophic condi ions in ela ion o an e io HS1 a e suppo ed by inc eases in he ela i e abundance o small Gephy ocapsa and C. pelagicus ssp. b aa udii, in ag eemen wi h p e ious PP quali a- i e es ima ions (Inca bona e al., 2010). Con inuous SST inc ease om he Bølling o he Alle ød has been obse ed om sedimen co es along he W Ibe ian ma gin (e.g., Ma a e al., 2014; Naugh on e al., 2016;Rod igues e al., 2010). This subs an ial wa ming is accompanied by a g adual ansi ion om subpola (C. pelagicus ssp. pelagicus) o sub opical (WWG) axa (Fig. 3), likely p omo ed by he no hwa d ex ension o he sub opical gy e (Schwab e al., 2012). The subsequen no hwa d displacemen o he associa ed Azo es F on would explain he con inuous inc ease in PP and nu icline shallowing obse ed along he B/A. 5.4. The YD The YD is a majo cooling e en in he no he n Hemisphe e (Alley, 2000) ma ked by a 4  C d op in SST in he s udy a ea (Da euil e al., 2016). O e all p oduc i i y is low, accompanied by ongoing nu icline deepening and educed upwelling (Fig. 4c and d). Con e sely, Palumbo e al. (2013) obse ed a conspicuous in- c ease in NAR o he en i e YD om sedimen co e MD03-2699 (Fig. 1), in e ing inc eased PP. Simila la i udinal di e ences ha e been ecognized in he quan i a i e econs uc ion o Pexp in no he n and sou he n si es o he Po uguese coas (Ab an es e al., 2001;Salguei o e al., 2014). I has been sugges ed ha he pola on was no displaced as a sou h as ou s udy si e (Eynaud e al., 2009;Naugh on e al., 2016). Ye , he a i al o colde and less saline wa e s (Duplessy e al., 1992) a ec ed he W Ibe ian ma gin om no h o sou h. These condi ions likely p omo ed inc eased su ace s a ifica ion a he han upwelling o nu ien - ich wa e s. Ne e heless, a ecen s udy based on pollen eco ds e eal asym- me ical no h/sou h ain all pa e ns along he ma gin (Naugh on e al., 2016). I is possible ha he ex eme win e p ecipi a ion ha a ec ed he no h and cen al pa s o he ma gin led o s ong coas al nu ien inpu by flu ial discha ges (Rod igues e al., 2010), leading o inc eased p oduc i i y in hose egions despi e su ace s a ifica ion and educed/absen upwelling, con as ing wi h he gene al low p oduc i i y in SW Ibe ia. In g ea e de ail, ou eco ds enable a leas wo dis inc phases du ing he YD o be dis inguished, named he e YDa (13.4e12.5 ka) and YDb (12.5e11.4 ka). YDa is he coldes phase as indica ed by hea ie d 18 O, peaks in NAR and pe cen ages o G. muelle ae, a d op B. Ausín e al. / Qua e na y Science Re iews 230 (2020) 106139 9