This paper documents the migration of the Polar Front (PF) over the Iberian margin during some of the cold climatic extremes of the last 45 ka. It is based on a compilation of robust and coherent paleohydrological proxies obtained from eleven cores distributed between 36 and 42°N. Planktonic δ18O (Globigerina bulloides), ice-rafted detritus concentrations, and the relative abundance of the polar foraminifera Neogloboquadrina pachyderma sinistral were used to track the PF position. These three data sets, compared from core to core, show a consistent evolution of the sea surface paleohydrology along the Iberian margin over the last 45 ka. We focused on five time slices representative of cold periods under distinct paleoenvironmental forcings: the 8.2 ka event and the Younger Dryas (two recent cold events occurring within high values of summer insolation), Heinrich events 1 and 4 (reflecting major episodes of massive iceberg discharges into the North Atlantic), and the Last Glacial Maximum (typifying the highest ice volume accumulated in the Northern Hemisphere). For each event, we generated schematic maps mirroring past sea surface hydrological conditions. The maps revealed that the Polar Front presence along the Iberian margin was restricted to Heinrich events. The sea surface conditions during the Last Glacial Maximum were close to those at present day, except for the northern sites which briefly experienced subarctic conditions.
 For more than 2.5 Ma, the Earth has experienced drastic climatic fluctuations among which the more pronounced occurred during the last 450 ka. Forcing factors at the origin of these major climatic seesaws are now fairly well identified throughout external (orbital) and internal parameters of the Earth's climatic system. However, phasing relationship and coupling phenomena between these parameters are still not well understood [e.g., Lisiecki et al., 2008]. In this way, climate extremes are major challenges regarding our comprehension of the Earth's climate variability and stability. The abrupt cold excursions, i.e., stadial events [Dansgaard et al., 1993] or Heinrich events [Heinrich, 1988], which occurred during the last glacial period, provide excellent archives of Pleistocene rapid climatic transitions. They, furthermore, raise major questions about the pace of the natural system. The extreme Heinrich events result from a massive calving of icebergs in relation to the sudden collapse of the periarctic ice sheets during the last glacial period [e.g., Heinrich, 1988; Bond et al., 1993; Hemming, 2004]. They took place at the end of a sawtooth shaped cooling cycle of 4 successive stadials/interstadials (e.g., Bond cycles [Bond et al., 1993]). However, the triggering of these events is still a matter of debate, as they imply complex mechanisms linking ice sheet, atmosphere and ocean dynamics. Their impact was recognized worldwide [e.g., Leuschner and Sirocko, 2000; Wang et al., 2001] and each of the Earth's surface reservoirs, i.e., the cryosphere, atmosphere, oceans, and biosphere, was affected. Some of the most severe Heinrich events impacted dramatically on the North Atlantic's overturning circulation [e.g., Zahn et al., 1997; Curry et al., 1999; Roche et al., 2004]. These events potentially constitute catastrophic analogs of modern or near-future ice sheet (Antarctic especially) collapses in the context of global warming.
 The objective of this paper is to document the impact of some of these cold extremes on the surface water hydrology of the Iberian margin. For this purpose, we focus on the reconstruction of the position of the Polar Front over the last 45 cal ka B.P. (calibrated 14C kiloyears before present; hereafter referred to as “ka”). We selected five time slices (Table 2a in section 2.2) representative of well-known cold episodes under various climatic forcing mechanisms (insolation, continental ice storage, sea level values, vegetation cover). At Present, the Polar Front (PF) closely follows the Greenland and eastern Canadian continental margins (Figure 1). In the ocean it is defined as a physical oceanographic structure with high temperature and salinity gradients, which separates two different surface water masses: the relatively warm, high-salinity Atlantic Water, and the cold low-salinity modified Polar Water carried southward by the East and West Greenland and Labrador currents [Holliday et al., 2007]. This front constitutes a strong ecological barrier in the North Atlantic Ocean and also physically constrains sea ice and iceberg drifting. The latitudinal migration of the Polar Front as south as 40°N during late Pleistocene cold events has been documented by numerous studies [e.g., McIntyre et al., 1976; Ruddiman and McIntyre, 1981; Duprat, 1983; Bard et al., 1987; Abrantes et al., 1998]. Our intention is to complement this information, since the Iberian margin offers by now an exceptional coverage of high-quality sequences, i.e., with diversified data sets, high time resolution and excellent stratigraphical control (Table 1). We choose robust paleohydrological proxies, i.e., planktonic δ18O (in Globigerina bulloides), ice-rafted detritus (IRD) concentrations and the relative abundance of the polar foraminifera Neogloboquadrina pachyderma sinistral (left coiled, N. pachyderma sinistral). These data sets were readily compared from core to core. They allowed us to create regional maps and estimate past changes in sea surface hydrology displaying the migration of the PF with a far better resolution than had been previously possible.
Table 1. Geographic Information of the Selected Cores
Water Depth (m)
Core Length (m)
2. Iberian Margin
2.1. Present-Day Hydrography
2.1.1. Dynamic Features
 The hydrography at the western Iberian margin (Figure 1) is primarily influenced by the North Atlantic subtropical gyre dynamic and a complex current system [Relvas et al., 2007]. The surface circulation regime is governed by the seasonal displacement of the Azores anticyclonic cell and its associated winds [e.g., Fiúza, 1984; Fiúza et al., 1998]. Several major features have to be considered; theyn are described in sections 22.214.171.124–126.96.36.199.
188.8.131.52. Large-Scale Currents
 These include (1) the Portugal Current (PC), flowing southward from 45°N to 30°N; (2) the Azores Current (AC) derived from the southern branch of the Gulf Stream (GS) and flowing eastward to the Gulf of Cadiz at about 38°N; and (3) the Canary Current (CC) which is fed by both the AC and the PC. Together, these currents form the Eastern Boundary Current of the subtropical gyre.
 During winter months, weak northerlies interspersed occasionally with strong south-southwest winds, favoring coastal convergence conditions. The most relevant transport mechanism is then the northward flow of a warm surface and undercurrent denominated as the Iberian Poleward Current (IPC) or Portugal Coastal Counter Current [e.g., Fiúza et al., 1998; Peliz et al., 2005] (Figure 1b). This current is recognized northward up to the Bay of Biscay, where it is known as the “Navidad Current” [Garcia-Soto et al., 2002]. The poleward flow affects both the surface and intermediate waters down to 1500 m depth [Relvas et al., 2007], adding warm and saline inputs into the study region [Haynes and Barton, 1990]. A connection with a northward recirculation of the AC has also been suggested [Peliz et al., 2005].
2.1.2. Synoptic View
 Modern sea surface conditions (Figure 1) along the margin between 37° and 41°N are characterized by mean annual temperatures ranging from 18°C to 16°C [Levitus, 1994]. These isotherms move seasonally and zonal gradients may occur during some periods of the year [Peliz et al., 2005]. During winter the Western Iberia Winter Front (WIWF), a recurrent feature between 10 and 11°W, separates the subtropical IPC waters from the cooler PC around 40°N [Peliz et al., 2005] (Figure 1b). However, major features are generated with the summer upwelling: a marked cold coastal band occurs, where the SST anomaly could reach an amplitude of −5°C (averaged values from June–September 2001–2005 [Relvas et al., 2007]).
 Intermediate circulation is dominated by the ENACW and the Mediterranean Outflow Water (MOW). Two main MOW branches are distinguishable, an upper core centered around 700–800 m, and a lower core at about 1200 m [Ambar and Howe, 1979; Iorga and Lozier, 1999]. The MOW follows a poleward route, but also propagates toward the west in the form of “meddies.” This water mass is supposed to have a dynamic impact regarding the ocean upper layers and it may notably constitute a complementary contributor to the AC [Jia, 2000].
 For this paper, we focused on 5 time slices representative of well-dated cold events in the Northern Hemisphere (Tables 2a and 2b), but which were characterized by distinct orbital configuration, atmospheric pCO2 concentration, ice sheet extension, and vegetation cover. They are as follows:
Table 2a. Stratigraphical Intervals of the Events Considered in This Papera
Table 2b. Dates Obtained After Previous Works on the Iberian Margin for Heinrich Eventsa
HE1 Along the Iberian Margin (cal ka B.P.)
HE4 Along the Iberian Margin (cal ka B.P.)
Regarding the available chronology of these events in the area (and even if error bars inherent to datings could minimize this remark), cold spells appeared earlier on the northern margin and lasted longer there.
 4. The Younger Dryas (YD) cold event, first described in palynological records from Scandinavia, is known as a true HE in the western North Atlantic (HE0 [e.g., Andrews et al., 1995]). On the European margin, the YD is marked by the installation of subpolar conditions [Baas et al., 1997] and the contraction of temperate tree populations [e.g., Naughton et al., 2007].
 The aim of this paper is to track the intrusions of polar waters over the Iberian margin and to delineate the Polar Front. Numerous high-resolution multiproxy records have been generated from sediment cores retrieved through international programs such as ENAM and IMAGES in this area. However, no synoptic synthesis between the different sequences has been carried out, yet. We have selected records adequate to both the existence of comparable proxy data among the cores and to a spatial distribution as wide as possible over the area (Table 1, see also auxiliary material), in order to tentatively draw a synthetic scenario for past sea surface hydrographical changes.
3.1. Stratigraphical Considerations
 The selected records are high-resolution records mainly from Calypso piston cores retrieved during the IMAGES I (1995) and V (1999) cruises with the R/V Marion Dufresne (IPEV). For ten of the eleven cores used, the age models were previously published (Table 3). The chronostratigraphies were based on calibrated 14C AMS datings and complemented by stratigraphical tie points from a correlation of the stable isotope data with Greenland ice core δ18O records (GRIP, GISP2, NorthGRIP). The age model of core MD95-2041 was constructed by a correlation of the planktonic oxygen isotope curve with those of the nearby sediment cores SU81-18, SO75-26KL, and MD99-2341, and with the GISP2 ice core. The lower part of the curve was tied to the stacked oxygen isotope record of Martinson et al. . For the other cores, we abstained from a further intercore fine tuning of the planktonic oxygen isotope records, and we decided to keep with the published age models, taking into account small stratigraphical discrepancies at certain intervals. These discrepancies were however at an infrascale time resolution than the one we used for our compilation, and therefore do not bias the results presented here. Further details regarding the methodology used to build the age model for each independent core are available in the papers cited in Table 3.
Table 3. Chronological Constraints on the Selected Coresa
Number of AMS 14C Dates (Last 45 ka)
Greenland Ice Core Based Time Scale?
Marine Core to Core Correlation/Comparison?
Compilation of the history of the construction of the age scales for each of the cores considered (see references). The chronology was established on the basis of 14C AMS dates but was also tied to Greenland ice cores, complemented by core to core correlation.
 The strategy to select the proxies to compile was motivated by the fact that among the existing set (see Henderson  for a review), some of the most robust and “author to author consistent” paleohydrological proxies are those based on microfossil remains. We therefore focused on data from planktonic foraminifera, which benefit from more than 50 years of paleoceanographic expertise [Kucera, 2007], and which were thoroughly tested through both geochemical (stable isotopes [e.g., Ravelo and Hillaire-Marcel, 2007]) and ecological (transfer function [e.g., Guiot and de Vernal, 2007]) investigations. In particular, we used the planktonic δ18O from G. bulloides, and the relative abundance of the polar foraminifera N. pachyderma sinistral counted in the >150μm fraction (Figures 3 and 4) . The choice to analyze this fraction was justified by the fact, that it has been conventionally adopted for large-scale paleoceanographical reconstructions [e.g., CLIMAP Project Members, 1981; Pflaumann et al., 1996; Kucera et al., 2005b]. Actually, it normally excludes morphologically immature forms (juvenile) and analysis is therefore less subjective. We combined these data with those of large lithic grain (LLG) analyses allowing us to identify ice-rafting activities (referred later in the text as LLG/IRD).
3.3. Mapping Southward Shifts of the Polar Front Throughout Time
 To support a synoptic view of the migration of the PF throughout time, we created maps reflecting the hydrological conditions over the Iberian margin for the 5 time slices. For mapping, we considered the δ18O of G. bulloides together with the relative abundance of N. pachyderma sinistral only. The LLG/IRD concentrations were analyzed from different size fractions, in particular >150 μm, >250 μm, and >355 μm, and the values were given as different units, as proportion in the respective grain size fraction or as abundance of lithic grains per gram sediment. The IRD abundance in the >150 μm fraction can be by a factor of 100 higher than those >250 μm. Regarding these inconsistencies, we decided not to map the LLG/IRD data, but to use the information to mark ice-rafting periods.
 For HE4, HE1, the YD and 8.2 ka events we adopted the approach based on the δ18O data developed by Cortijo et al. ; that is, we regarded the difference between planktonic δ18O values recorded prior to and during the event itself. During the LGM, however, such changes were not significant. We, therefore, decided to use values averaged over the whole interval. This approach was furthermore justified because of the 4 ka duration of the LGM, and also because previous LGM mapping exercises used mean values over the whole interval as well [CLIMAP Project Members, 1981; Kucera et al., 2005a]. For the δ18O mapping we decided to limit our interpretation to the plot of numerical values.
 Regarding the N. pachyderma sinistral abundances, we decided to plot the maximal abundances recorded for each event (Figure 3) in order to capture the true amplitude of the signal. Our stratigraphical constraint for each of the events is precise enough to exclude diachronism. We interpreted the plots drawing water mass limits with the intention to locate the PF. This was possible only because a good coherency exists between the modern distribution of N. pachyderma sinistral in surface sediments and the overlaying surface water masses (see Figure 4). Four major provinces were distinguished that, permit us to consider transitional settings and allow us to better qualify our observations: (1) N. pachyderma sinistral abundances of more than 90% prevail in the vicinity of the Polar Front, (2) abundances between 50 and 90% characterize a situation where modified polar water could have reached the margin and generated a front comparable to the present-day Arctic Front, (3) relative abundances between 50 and 5% indicated that subarctic waters invaded the area, and (4) relative abundances below 5% depicted sea surface conditions close to the modern situation.
4. Polar Front Signature Over the Iberian Margin
 The selected proxies showed a good coherency through time and space, and they permit to distinguish specific events which appear synchronously along the margin. HEs especially exhibit a strong paleohydrological signature, with high G. bulloidesδ18O values and high N. pachyderma sinistral percentages, together indicating low SST. The different amplitudes recorded between the cores basically reflect the latitudinal distribution of the records. N. pachyderma sinistral occurs for instance, with up to 90% in quasi-monospecific assemblages at the northern part of the margin whereas its percentages range from 10 to 15% at maximum in the south (Figure 5). By analogy to the modern distribution of this species in surface sediments of the North Atlantic Ocean (Figure 4), this pattern suggests the penetration of polar surface waters in the north and, at least subpolar surface waters in the south during HEs. LLG/IRD concentrations also peak during HEs. Even if they were low as compared to those observed within the Ruddiman belt, LLG/IRD concentrations constitute a complementary evidence of the protrusion of polar waters on this margin. Regarding LLG/IRD, HE4 occurs as one of the more drastic events, whereas no lithic was recorded during the YD cold event. The LGM depicts mild sea surface conditions as supported by the near disappearance of N. pachyderma sinistral in the fauna. No LLG/IRD pulses were recorded. High G. bulloidesδ18O values during this interval mirror the change in water δ18O related to sea level changes (global ice volume signal), rather than an invasion by cold surface waters [see Waelbroeck et al., 2005]. The 8.2 ka event appears insignificant in the selected records in spite of an accurate resolution of sampling.
 The 2-D mapping exercise (Figure 5) allows us to constrain the latitudinal range where the PF could have been positioned. According to our compilation, it becomes clear that the PF reached the area only during HEs. The compilation for HE4 and HE1 produce a very consistent picture, with a PF located close to 42°N. Arctic and subarctic waters penetrated the area as far south as 36°N. The cold water may even invaded the western Mediterranean Sea [see Cacho et al., 1999, 2006; Pérez-Folgado et al., 2003; Sierro et al., 2005]. Comparison with the δ18O values of G. bulloides shows a coherent scheme of heavier values during HE1 and HE4, here showing typical stadial conditions as was demonstrated by Voelker et al. . However, it is worth noting that the northern margin records display only an attenuate increase of 1.5 to 0.5‰. This pattern is in agreement with the proximal position of the northern records to the Ruddiman belt, where a strong depletion in planktonic δ18O values due to meltwater influx overprinted the associated signal of low sea surface temperatures (in reverse to the signal directly off Iberia).
 The YD also records drastic changes, but the equatorward protrusion of colder water masses seems to have been limited to the installation of subarctic conditions only. This is coherent with the δ18O values of G. bulloides that increased at that time, but only by 0.5 to 1‰ at maximum (Figures 3 and 5).
 The LGM and the 8.2 ka event appear to have had only minor impact on the regional SST, with no real distinctive features in the N. pachyderma sinistral records. These events display minor changes in the δ18O signal of G. bulloides (Figure 5), reaching a maximum of 2.5 to 3‰ up to 40°N and of 1.5 to 2‰ in the southernmost sites during the LGM. Although for some cores, the lower sampling resolution throughout the Holocene seems to contribute to the weak representation of the 8.2 ka event, this is not the case for the LGM which furthermore encompasses a 4 ka interval.
5.1. Impact of HEs Over the Iberian Margin
 Our compilation facilitated schematic views of the hydrological conditions prevailing along the Iberian margin during some cold events of the last 45 ka. It shows that the PF incursions were limited to episodes of major ice sheet collapses, i.e., Heinrich events. Further evidence of this drastic impact comes from other independent marine proxies, not compiled here as they were unfortunately not available for all the records considered. Alkenone data similarly revealed major decreases in SST, for instance as large as 5°C for HE4 at the southern part of the margin [e.g., Pailler and Bard, 2002; Martrat et al., 2007], and the incursion of less saline waters [Bard et al., 2000]. In the same way, other qualitative studies of plankton related groups, such as dinoflagellate cysts (dinocysts), coccolithophorids or diatoms, also registered major population changes during HEs. Dinocyst assemblages typify the migration of subpolar species over the area, in particular with the occurrence of Bitectatodinium tepikiense or Nematosphaeropis labyrinthus [Eynaud, 1999; Sánchez Goñi et al., 2000; Boessenkool et al., 2001; Turon et al., 2003]. These species are known to be linked to low winter SST and low salinities in subboreal domains today. For coccolithophorids, the small (6–9 μm) morphotype of Coccolithus pelagicus s.l. was identified during HEs, in association with N. pachyderma sinistral and LLG/IRD peaks [Colmenero-Hidalgo et al., 2004; Parente et al., 2004]. It reveals a shift in the bioprovinces following the invasion of subpolar water masses. At an orbital scale, diatom assemblages and abundances were shown to be linked to oscillations in primary productivity related to the upwelling intensity. Major changes were identified during Terminations and at a lower amplitude during HEs [Abrantes, 1992; Thomson et al., 2000]. This was coherent also with the deposition of a typical lithofacies [Baas et al., 1997], slightly diverging from the lithology of Heinrich layers in the “Ruddiman belt,” but in accordance with the more southern position of the Iberian Margin. Benthic communities revealed a drawdown of intermediate and deep-water oxygenation during HEs [Baas et al., 1998; Schönfeld et al., 2003]. These events were furthermore all synchronous with drastic changes on the adjacent Iberian continent, corresponding to the regression of the Mediterranean forest and the development of the steppic vegetation [e.g., Sánchez Goñi et al., 2000; Roucoux et al., 2001, 2005; Combourieu-Nebout et al., 2002; Daniau et al., 2007; Naughton et al., 2007, 2009].
5.1.1. Upwelling System?
 If convergent data draw the scenario of a sudden installation of extreme environments over the area, we have, however, little evidence of how the local hydrology was affected by HEs. In particular, the dynamic of the local upwelling system is under debate [see also Salgueiro, 2006; Salgueiro et al., 2008]. For instance, the total organic content in sediments off the margin during HEs was interpreted by Pailler and Bard  as indicating a decrease of the upwelling intensity, whereas Lebreiro et al.  suggested the opposite, on the basis of the assumption that high amounts of heavy minerals during HEs result in an artifact in the calculation of organic carbon proportion. This assumption was supported by planktonic foraminiferal data, as high-productivity indexed species were observed in the assemblages during HEs (cores D11957P [Lebreiro et al., 1997] and MD01-2444 [Vautravers and Shackleton, 2006]). The same was demonstrated for dinocysts (core MD95-2042 [Eynaud, 1999; Sánchez Goñi et al., 2000]). Using microcharcoal fluxes to the ocean (core MD95-2042), Daniau et al.  conversely concluded to the prevalence of a southeasterly wind component over the southern Iberian Peninsula and then to a breakdown of the upwelling during HEs and D-O stadials. Their interpretation was coherent with a scenario of a compressed subtropical gyre due to the southward progression of polar water. The Azores High maintained at low latitudes and the prevailing winds from the south could not favor upwelling conditions.
 Regarding changes in productivity linked to HEs, Lebreiro et al.  introduced an alternative solution with the idea of a system that was driven by the dynamics around icebergs rather then by the upwelling itself. Following Sancetta , they hypothesized that the presence of drifting ice during HEs, supported high production by turbulent mixing and enhanced density stratification or by the supply of major or trace nutrients. This assumption is still conceivable as enhanced production has recently been observed in the vicinity of free-drifting icebergs in the Weddell Sea [Smith et al., 2007]. It is furthermore highly consistent with our mapping, which supports the arrival of possibly nutrient-enriched polar water tongue.
 Our maps are picturing the events as time-static systems (Figure 5), it is then straightforward to distinguish features representative of a dynamic scheme in surface currents. However, strong thermal gradients during the HEs and the YD support a major sea surface hydrological reorganization in the area. The equatorward protrusion of polar or subpolar waters affected the margin in quite a similar way during the events considered. We can recurrently observe a demarcation at 40°N between the northern and southern part of the margin. This latitude has previously been suggested as the southern position of the PF during glacial times [Ruddiman and McIntyre, 1981; Duprat, 1983; Bard et al., 1987]. We demonstrate here that the PF probably oscillated along the northern margin only during HEs around this latitude, but a lens of coastal waters between Lisbon and Cape St Vincent recorded milder conditions at that time. This mirrors the scheme of an upper ocean winter circulation as synthesized by Peliz et al. [2005, see Figure 18, p. 643] with warmer waters on the southern margin linked to a AC-like northward recirculation.
 Another important point emerging from our study is that the LGM was not a cold extreme but a period of warming as compared to the surrounding HEs. Following several works, both at the regional scale off the Iberian margin [Eynaud, 1999; Bard et al., 2000; Boessenkool et al., 2001; de Abreu et al., 2003; Turon et al., 2003], as well as at the scale of the North Atlantic Ocean [Sarnthein et al., 1995, 2003; Pflaumann et al., 2003; de Vernal et al., 2000, 2005, 2006], our compilation supports the presence of warm surface water masses between the continent and 10.5°W during the LGM. Only the offshore sites off Galicia recorded a slight sea surface cooling at that time (Figure 5). This cooling is, however, of relatively low amplitude, as percentages of the polar taxa N. pachyderma sinistral do not exceed 9%. This picture of warm, nearshore waters is coherent with previous studies which have shown the expansion of thermophilous planktonic taxa along the margin (dinocyst [Eynaud, 1999] and foraminifera [de Abreu et al., 2003; Salgueiro, 2006]) and estimated LGM-SST to be cooler by only 1 to 2°C relative to present day [de Abreu et al., 2003] or resembling averaged Holocene temperatures [Sánchez Goñi et al., 2008]. Close to the continent, a lens of subtropical waters with SST very similar to the modern ones, seems then to have prevailed, in particular south of 40°N. This draws the picture of dual subpolar and subtropical-sourced water masses at that time, depending on the latitude, and suggests the penetration of subarctic water masses, but definitely not the installation of the PF on the margin. Instead, we may expect the presence of a strong frontal structure, comparable to the modern Western Iberia Winter Front [Peliz et al., 2005] (Figure 1b). The same evidence of a tongue of warm water was previously related to a “glacial Azores Current” by Parente et al.  and Rogerson et al. . A striking point regarding this “warm” LGM sea surface signature is that the period conversely appears as a cold period on land [e.g., Harrison et al., 2001]. This underlines a decoupling of atmospheric and sea surface processes, probably explained by the large volume and aerial extension of the continental ice at that time over the boreal latitudes, and moreover with a reduced vegetation cover in the vicinity of these large ice sheets. However, some recent demonstration of mitigated situations regarding pollen assemblages have been noted on the Iberian peninsula (Naughton et al., presented paper, 2007).
5.3. YD and 8.2 ka Events
 The YD experienced much more drastic conditions than the LGM, the later, regarding our mapping, resembling more the “8.2 ka event.” The recurrent feature which links these three events, is the partitioning of the margin at 40°N, with a northern part recording colder conditions. During the YD, penetration of subarctic waters was recognized offshore Galicia with a front splitting the nearshore and offshore sites around 40°N that was similar to the modern Western Iberian Winter Front (compare Figures 1 and 5). We can imagine that the milder conditions recorded along the shore, could be linked to a mixture of warmer subtropical waters from AC recirculations and potentially colder PC or upwelled waters.
 This compilation provides an integrated view of the sea surface conditions that have prevailed during the last 45 ka at the Iberian margin. Focusing on cold extremes of the late Quaternary, we have tentatively drawn schematic maps of sea surface hydrological structures, in order to visualize the migration of the PF. The maps indicated that the PF reached the Iberian margin only during HEs, prevailing in the vicinity of 40°N. The two selected HEs, HE1 and HE4, were major melting events of the Laurentian and European ice sheets. They left an imprint of very similar amplitudes regarding changes in the sea surface hydrology at the Iberian margin. Our reconstruction constrained the limits of the protrusion of polar surface waters. The maps revealed a partition of the margin into a northern area, where drastic conditions were recorded, and a southern sector where milder conditions prevailed, probably in relation to the influence of the local surface currents (Azores Current notably). The presence of the PF at that time questions the dynamic of the Portuguese upwelling.
 Our compilation reveals that the 4 ka time interval of the LGM recorded quite stable sea surface conditions close to the modern ones, at maximum 1 to 2°C cooler than today, and thus a low-amplitude cooling when compared to the HEs. More drastic paleoenvironmental changes were recorded during the YD. For the LGM, the YD and also the 8.2 ka event, the northern margin experienced the most pronounced cooling, inferring further evidence for a different response of the northern and southern Iberian margin.
 The authors would like to thank all the participants of the respective scientific cruises (IMAGES I, GEOSCIENCES, GINNA- IMAGES V, and PICABIA) along with the captains and the crews of the R/V Marion Dufresne, supported by the following French agencies: Ministère de l'Education Nationale de la Recherche et de la Technologie (MENRT), Centre National de la Recherche Scientifique (CNRS), and Institut Paul Emile Victor (IPEV). F.E. acknowledges financial support from the French contract of the ANR IDEGLACE and the ESF RESOLUTION project. L.A., E.S., and A.V. received financial support by the Portuguese Foundation for Science and Technology. J.S. acknowledges funding by the European Commission's Framework Six Programme (HERMES project, contract GOCE-CT-2005-511234-1). Comments by the reviewers, including Alessandra Asioli, significantly benefited this paper. This is UMR/EPOC CNRS 5805 contribution 1745.