Earth and Planetary Science Letters 258 (2007) 283–292 www.elsevier.com/locate/epsl

H4 abrupt event and late presence in Iberia ⁎ Pierre Sepulchre a, , Gilles Ramstein a, Masa Kageyama a, Marian Vanhaeren b, Gerhard Krinner c, María-Fernanda Sánchez-Goñi d, Francesco d'Errico e,f

a Laboratoire des Sciences du Climat et de l'Environnement/IPSL, UMR CNRS-CEA-UVSQ 1572, CE Saclay, L'Orme des Merisiers Bât. 701, 91191 Gif sur Yvette Cedex, France b Ethnologie Préhistorique, UMR 7041 CNRS-ArScAn, 92023 Nanterre, France c Laboratoire de Glaciologie et Géophysique de l'Environnement, CNRS-UJF Grenoble, BP 96, F-38402, Saint Martin d'Hères, France d Ecole Pratique des Hautes Etudes, Département de Géologie et Océanographie, UMR 5805 CNRS-EPOC, Université Bordeaux1, 33405 Talence, France e Institut de Préhistoire et de Géologie du Quaternaire, UMR 5199 CNRS-PACEA, Université Bordeaux1, 33405 Talence, France f Department of Anthropology, George Washington University, 2110 G Street, NW Washington, DC 20052, United States Received 15 November 2006; received in revised form 23 March 2007; accepted 23 March 2007 Available online 31 March 2007 Editor: M.L. Delaney

Abstract

Heinrich event 4 (H4) is well documented in the North Atlantic Ocean and the adjacent continents as a cooling event 39,000 yr before present (BP). To quantify the impact of this event with respect to climate and vegetation over the Iberian Peninsula, we perform numerical experiments using a high-resolution general circulation model forced by sea surface temperatures before and during H4. Our model simulates an expansion of aridity over the peninsula during H4, a desertification of the south, and a replacement of arboreal by herbaceous plants in the north, all of which are in agreement with contemporaneous sequences from marine cores located off the Iberian Peninsula. Our simulations demonstrate that the H4 marine event imprinted drastic changes over Southern Iberia, which would not have favoured its occupation by Anatomically Modern Humans, therefore providing a plausible explanation for the delayed extinction of in this region inferred from the archaeological record. © 2007 Elsevier B.V. All rights reserved.

Keywords: Neanderthals; paleoclimate; Heinrich; abrupt event; Quaternary; Aurignacian; Mousterian

1. Introduction Several hypotheses based on ecological factors have been put forward to explain the apparent late survival of The dynamics of the colonisation of by Neanderthal populations in the southern and western Anatomically Modern Humans (AMH) and Neanderthal regions of the Iberian Peninsula. The first of these extinction are the subjects of lively debate [1–6]. hypotheses suggests that AMH populations bearing an Climatic variability has been repeatedly invoked as a Aurignacian technology may not have expanded south potential cause for such crucial population events [2], of the Ebro River during Interstadial 9 which occurred occurring at ca. 41–32 cal ky BP (37–29 14C ky BP) [7]. just prior to Heinrich Event 4 [8–10] (H4, 39,000 calendar yr BP) [11]. This lack of expansion to the south ⁎ Corresponding author. Tel.: +33 1 69 08 65 49; fax: +33 1 1 69 08 77 16. may have been due to the fact that the regions to the E-mail address: [email protected] (P. Sepulchre). south were significantly more wooded [12] and thus less

0012-821X/$ - see front matter © 2007 Elsevier B.V. All rights reserved. doi:10.1016/j.epsl.2007.03.041 284 P. Sepulchre et al. / Earth and Planetary Science Letters 258 (2007) 283–292 favourable since large mammals, on which their subsis- duration of the meltwater injection in the North Atlantic tence systems were focused, preferred open landscapes. corresponding to this event has been recently updated to According to this hypothesis, AMH populations would 250±150 yr [20]. The duration of the cooling due to this have moved southwards only when the cold conditions perturbation is longer, as the associated with H4 allowed for the southern expansion needs another 400 yr to adjust and restart after the end of of the prey species that formed the basis of their the meltwater pulse. In terms of continental response, subsistence systems. Other authors [13] have proposed considering these ranges and regarding vegetation dy- that the climatic changes of the Oxygen Isotope Stage 3 namics, it seems that the H4 could have perturbed the (OIS 3, between ca. 60 and 25 ky BP) could have fatally biosphere for at least one millennium (see discussion). perturbed the Neanderthal seasonal activity cycle even Previous modelling studies [3] have been conducted on before the arrival of modern competitors in these south- the OIS 3 palaeoclimatic [21] and palaeovegetation [22] ern regions. A third scenario [2], based on comparisons variations, but have not specifically focused on the of archaeological and pollen data, suggests that the late impact of an abrupt cold climatic shift on neighbouring survival of Neanderthals in Southern Iberia is the continental areas. Kageyama et al. [23] investigated the consequence of the expansion of semi-desert environ- impact of H1 over western Europe in terms of temper- ments over this region due to the climatic conditions ature and water cycle, but the interest of H4 is that this associated with H4, which limited the advance of AMH climate shift may have occurred contemporaneously groups, dependant on large ungulate species, into with important changes in human populations. Yet no regions south of the Ebro. Such a scenario would have spatial reconstruction of the H4 impact on Iberian delayed the replacement and eventual extinction of vegetation is available. Well-dated indications of envi- Neanderthals. The aim of this paper is to use appropriate ronmental changes are recorded in pollen-rich marine models to evaluate these hypotheses in terms of climatic cores located off the Iberian peninsula [24,25] and allow and vegetation changes associated with H4. for comparisons with our model outputs. Here we The small number of human remains associated with present two simulations made with a Global Circulation archaeological material dated to this period represents a Model (GCM) quantifying the climatic and vegetation major problem for testing hypotheses regarding the response before (BH4) and during (DH4) Heinrich 4 nature, extent, and chronology of relationships between event. AMH and Neanderthals. It is generally acknowledged, however, that the Aurignacian culture is a reliable 2. Methods for the dispersal of AMH, that these new populations arrived in regions north to the Iberian Peninsula between To quantify climate changes produced by H4 over 42 and 41 ky cal BP (37–36 14CkyBP)[6,7], and that this region, we use the atmospheric general circulation their colonisation of southwestern Iberia occurred much model (AGCM) called LMDz developed at Laboratoire later, at around 35,200 cal BP (33 14CkyBP)[14].We de Météorologie Dynamique [26] with a high regional argue that the impact of H4, which is centered around ca. resolution over western Europe, providing a 60 km 39 ky BP, and its imprint on vegetation for a period of resolution over Iberia. The Dynamic Global Vegetation more than 1000 yr, played an important role in this Model (DGVM) ORCHIDEE [27] is forced by climate process. This phenomenon may explain, in part, the delay variables produced by LMDz in order to quantify the in Modern Humans progression southward and the longer spatial distribution of vegetation changes on the Iberian survival of Neanderthal populations in the south of Iberia. peninsula. In order to evaluate the environmental impact Data from deep-sea sediments [8], continental se- of the H4 event over Iberia, we run two numerical quences [15] and polar ice cores [9] show that, during experiments with the AGCM. The first experiment OIS 3, several abrupt climatic shifts occurred over the (BH4 for “Before H4”) aims at estimating the climate Atlantic Ocean and Europe. High levels of lithic ice- just before H4. It is constrained with appropriate 39 ky rafted debris (IRD) and low foraminiferal concentrations cal BP orbital parameters [28] and CO2 concentration define several Heinrich events of massive iceberg dis- set at 209 ppmv (partial pressure in parts per million, charges in the North Atlantic Ocean [10,16,17]. Among from the Vostok [29]). Since no global dataset them, H4 was associated with strong decreases in North describing the SSTs before the H4 exists, the SSTs used Atlantic Sea Surface Temperatures (SSTs) [18]. Its exact for this first simulation are those given by the CLIMAP date is still a matter of debate (e.g. see Bard et al. [11] [30] Last Glacial Maximum global reconstructions. and Shackleton et al. [19]) but Roche et al. [20] suggest Similarly, no global ice-sheet reconstruction is available 38.9 and 39.3 ky BP as the most plausible dates. The for this period, but the sea-level drop at that time is P. Sepulchre et al. / Earth and Planetary Science Letters 258 (2007) 283–292 285 estimated to be 60 m lower than present [31]. We have nique applied to planktonic assemblages therefore chosen to use the 14 ky BP ice-sheets recon- [18]. These authors reconstruct SSTs at several points in struction from Peltier [32] (ICE4G) which corresponds the North Atlantic Ocean. Error bars associated with to the same sea-level drop, and which is the closest these reconstructions are rather weak compared to the reconstruction to the ice sheets which existed before and anomaly. We use these reconstructions to create zonal during H4. This reconstruction is consistent to what temperature anomalies that we apply to the CLIMAP other authors used for the same period [33]. Thus our SST dataset (Fig. 1). LGM SSTs are reduced by 4 °C first experiment is a “glacial” state, with appropriate between 40°N and 50°N, 3 °C between 50 and 54°N, boundary conditions representing the pre-H4 conditions. 2 °C between 54 and 60°N, and 1 °C towards the North In the second experiment (DH4 for “During H4”)we Pole. Southward from 40°N, the imposed anomaly keep the same boundary conditions (insolation and CO2 follows a linear decrease to reach 0 °C at 32°N. Sea ice concentration vary very little between before and after is imposed where SST is lower than −1.8 °C. By apply- H4) except for the SSTs. We cool the North Atlantic ing these anomalies to CLIMAP SSTs, we simulate the using H4 SST anomalies consistent with reconstructions cooling effect of the H4 over the North Atlantic. by Cortijo et al. based on the modern analogue tech- Reconstructions from Gulf of Cadiz [34] and off western

Fig. 1. SST anomalies (grey scale bands) applied to the CLIMAP dataset for the DH4 experiment. Between 40°N and 60°N, top and bottom numbers indicate absolute values of summer SSTs as reconstructed from Cortijo et al. [18] and prescribed in the DH4 run, respectively. Top and bottom numbers below 40°N indicate mean SSTs reconstructed from [34,35] and prescribed in the DH4 run, respectively. 286 P. Sepulchre et al. / Earth and Planetary Science Letters 258 (2007) 283–292

Africa [35] also suggest a cooling at lower latitude only considered sites with AMS ages, and conventional during H4. However, CLIMAP SSTs are consistent 14C ages made on charcoal. Such underestimations have with the reconstructed values for H4 at these latitudes been shown to be common with conventional 14Cages, (Fig. 1), so we did not add a supplementary negative particularly when made on collagen [4,7], and attribut- anomaly. Further modelling studies may have to assess ed to ineffective sample treatments for samples at the the potential impact of such low-latitudes anomalies to temporal limits of the radiocarbon method. We also see if it can affect atmospheric circulation and excluded sites of controversial cultural attribution [4,14]. precipitation patterns over Iberia. Both experiments In order to attribute archaeological sites to specific climate have been run for 11 yr and provided outputs that have events we used the 14C temporal frame proposed by Elliot been averaged over the last 10 yr to define the BH4 and et al. [38]. DH4 climates. The vegetation response to these climatic changes is 3. Results computed with the ORCHIDEE model in stand-alone mode, i.e. forced by outputs from the climate simula- The annual rainfall over the Iberian Peninsula fol- tion. ORCHIDEE is forced by diurnal cycle variation, lows a NW-SE gradient for both simulations (Fig. 2a, b). daily rainfall, monthly means of air temperature, relative Rainfall occurs primarily from November to March and humidity, 10-meter wind speed and cloud cover, all is related to winter storms, with the strongest storms these variables coming from daily outputs of the two along the Iberian NW coast. The BH4 experiment shows climatic experiments. To be consistent with the high rainfall (N800 mm/yr) over Northwestern Iberia, CLIMAP SSTs used for the BH4 experiment, the initial while the southeast is very dry (b100 mm/yr, Fig. 2a). vegetation cover comes from the equilibrium results of This pattern changes in DH4, with the decreased SSTs ORCHIDEE forced by the climatic outputs of a classical leading to an extension of arid zones from the south to run of the Last Glacial Maximum (LGM) [36]. This the center-west of Iberia (the 200 mm/yr isohyet moves glacial vegetation was used as initial condition for our northwestwards), and the reduction of the northwestern two simulations that were run for 200 yr to reach precipitation. On the Atlantic coast, the 200 mm isohyet equilibrium. Additional climate simulations, forced by shifts northward by 2°. This change is not associated the resulting vegetation, have also been performed to with a reduction in storminess. Rather, the storms check that the vegetation feedback does not have a coming from the Atlantic carry colder and therefore strong impact on the regional climate. The equilibrium drier air towards the Iberian region. On the other hand, vegetation is described in terms of Plant Functional the cooling over the North Atlantic produces a relatively Types (PFTs) [37]. We compare our results to pollen weak decrease of continental annual temperatures data from four deep-sea cores taken off the Iberian (0.5 to 1.5 °C), which mainly occurs on the Atlantic margin. High-resolution pollen records from IMAGES coast (Fig. 2c, d). MD95-2042 and MD95-2039 cores reflect vegetation Precipitation estimates [24,25] inferred by applying changes that occurred during OIS 3 in western Iberia, to the pollen record the modern analogue technique whereas those from MD95-2043 and Ocean Drilling [39,40] indicate a precipitation decrease by 400 mm on Program 976 (ODP 976), retrieved in the Alboran Sea, the Atlantic and by 500 mm on the Mediterranean side cover the floristic changes that took place in western during H4. Our simulations are consistent with this Mediterranean borderlands [24,25]. These cores contain decrease on the Atlantic coast (Fig. 2b). However, some unambiguous signatures of H4, in terms of percentage of discrepancies exist on the Mediterranean side, which is polar foraminifera as well as variations of the pollen already very dry in our BH4 experiment (b50 mm/yr) assemblages. and prevents the model from simulating an anomaly as The archaeological data used in this study were com- large as 500 mm/yr. However the essential result is that piled from a variety of sources by two of us (FdE and MV) the DH4 experiment reproduces the rainfall gradient in the framework of a project funded by the ECLIPSE occurring during the event between the two sides of program of the Centre National de la Recherche Scienti- Iberia satisfactorily. fique. This database contains the geographic coordinates, The cooling and drying of H4 induce large vegetation recorded stratigraphic levels, associated cultural affilia- changes. The maximum vegetation fraction (i.e. the max- tions, and ca. 6000 radiocarbon ages from ca. 1300 imum value of total vegetation cover over a pixel) is archaeological sites in Europe. In an effort to minimize the already very low in BH4 (b50%, Fig. 3a). This fraction possibility of incorporating sites associated with radio- shifts to even lower values (b25%) in DH4, with a north- carbon ages that are underestimates of their true ages, we ward expansion of desert-like environments, especially P. Sepulchre et al. / Earth and Planetary Science Letters 258 (2007) 283–292 287

Fig. 2. Climatic changes induced by the simulated Heinrich 4 (H4) event over Iberia. (a, b) Absolute annual amount of rainfall (mm) before H4 (BH4, left) and the anomaly between both simulations (DH4–BH4, right), showing a strong northwest southeast gradient of decrease. (c, d) BH4 annual surface temperature mean (°C, left) and difference of annual average of surface temperature between both simulations (DH4–BH4, right). The response to the SST decrease occurs mainly on the western coast, while it is very weak in the inland. north of the Ebro valley (Fig. 3b). Simultaneously, the woody PFTs present in Iberia. Boreal and Temperate already weak arboreal percentage (b10%) decreases in Broadleaved Summergreen trees occupying the center of inland Iberia, but remains constantly close to the south- the peninsula and southwest Portugal in BH4 undergo a western coast, with a mean fraction of 15% (Fig. 4a, b). strong reduction during H4. Temperate Needleleaf trees, The model simulates significant changes in the three initially present over a large part of the peninsula, are 288 P. Sepulchre et al. / Earth and Planetary Science Letters 258 (2007) 283–292

Fig. 3. Maximum percentage of vegetation cover simulated over Iberia and geographic distribution of radiocarbon dated Mousterian and Aurignacian sites before (a) and during (b) the Heinrich 4 event. (c) Geographic distribution of radiocarbon dated Mousterian and Aurignacian sites after the Heinrich 4 event. essentially confined to the Atlantic coast during this cold data [24,25,41,42], which show abrupt decrease in open event. Grasses remain on the Cantabrian and Mediterra- Mediterranean forest (deciduous and evergreen Quercus, nean coasts as well as in the Ebro basin (Fig. 4c, d). Pollen Phillyrea, Olea, Pistacia, associated with Ericaceae) and P. Sepulchre et al. / Earth and Planetary Science Letters 258 (2007) 283–292 289

Fig. 4. Maximum simulated fraction of arboreal and herbaceous coverage when equilibrium is reached. (a, b) Trees, which are present from the Atlantic coast to the center of the peninsula before H4 (left), are confined to the coast during H4 (right). This fraction is calculated as the sum of six arboreal Plant Functional Types. (c, d) Grasses, which are the major part of the vegetal cover before H4 (left), disappear almost totally from the south during H4 (right). This fraction corresponds to the simulated C3 grasses PFT.

increase in steppe-to-semi-desert vegetation (Artemisia, delay between the maximum ocean perturbation and the Chenopodiaceae, Ephedra) are in good agreement with onset of aridity over the continent. our model outputs. In the North, H4 triggers a reduction in The geographic distribution of archeological sites deciduous woodland and forest, and an expansion of before H4 (Fig. 2) identifies numerous Mousterian and grassland. The MD95-2042 core (Fig. 5) gives important Aurignacian occupations in Northern Iberia and a few information concerning the timing of the vegetation Mousterian coastal sites in the South. During H4, the response to H4. It shows that the maximum of semi-desert Aurignacian remains confined to the North and the plants occurs ca. 1000 yr after the δ18O perturbation and number of sites decreases in Northern Iberia. Two sites, the first Ice Rafted Debris peak. Moreover, high Foradada and Beneito, have 14C ages that attribute their percentages of semi-desert plants last longer than the occupation to the H4 event. However, the cultural IRD/δ18O event, and another millennium is necessary for attribution of the layers that have yielded the dated the vegetation to recover from H4. Thus there is a clear samples is problematic. After H4, Aurignacian sites are 290 P. Sepulchre et al. / Earth and Planetary Science Letters 258 (2007) 283–292

Fig. 5. Detail of Heinrich event 4 (H4) from the pollen diagram of core MD95-2042. The gray interval indicates H4. H4 is dominated by semi-desert vegetation (Artemisia, Chenopodiaceae and Ephedra). Arrows indicate the lag between the first IRD peak and the onset of semi-desert plants, which lasts ca. 1000 yr. Moreover, it needs another millennium to observe paleoenvironmental settings similar to before H4 (onset of Mediterranean forest and decrease of semi-desert plants). recorded in the North and the South of the Iberian pen- a decline of land biomass (as simulated by our DGVM) insula while Mousterian sites only persist in the south- the duration of which, as attested by the pollen record, west and in the center. could have led to a drop in the biomass of ungulates and consequent drop in Neanderthal population density. 4. Discussion and conclusions Development of semi-desert landscape in Iberian inland may also have resulted in delaying the Aurignacian The timing of Neanderthal extinction is a controver- advance. According to these hypotheses, competition sial issue [1–6,43–45]. However, previous research between the two populations would have been tempo- [2,4,46,47] and results presented here support the rarily prevented as environmental changes created a scenario of a late survival of Mousterian Neanderthals Southern refugium for Mousterian Neanderthals, which in Southern Iberia. The archaeological and radiocarbon delayed their replacement by Aurignacian groups. evidence indicates that populations bearing Aurignacian This study is based on a comprehensive simulation technologies were already present in Northern Iberia at of the climate and continental responses, in terms of the onset of H4 but colonised the South only after this vegetation changes, to Heinrich event 4, using the most climatic event. The large aridification of Central and appropriate boundary conditions available. Our model Southern Iberia depicted in our DH4 simulation implies results are in good agreement with pollen from marine P. Sepulchre et al. / Earth and Planetary Science Letters 258 (2007) 283–292 291 cores and challenge the hypothesis that a stationary [6] J. Zilhao, F. d'Errico, The chronology of the Aurignacian and Mediterranean forest persisted almost unchanged in Transitional technocomplexes. Where do we stand? in: J. Zilhao, F. d'Errico (Eds.), The Chronology of the Aurignacian and of the Southern Iberia coastal areas during OIS 3 and was not Transitional Technocomplexes: Dating, Stratigraphies, Cultural affected by D-O millennial scale climatic variability Implications, Lisbon, 2003. [1,13,48–50]. The added value from the modelling ap- [7] P. Mellars, A new radiocarbon revolution and the dispersal of proach is the spatialisation of the climate and vegetation modern humans in Eurasia, Nature 439 (2006) 931–935. response over Iberia. While modelling quantifies climate [8] S.R. Hemming, Heinrich events: massive late detritus layers of the North Atlantic and their global climate imprint, Rev. and vegetation responses spatially, paleoclimatic data Geophys. 42 (2004) 1–43. provide relevant information concerning the timing and [9] W. Dansgaard, et al., Evidence for general instability of past the duration of the latter's response to the cooling. These climate from a 250-kyr ice-core record, Nature 364 (1993) (pp.). results may underscore the role played by sub-millennial [10] H. Heinrich, Origin and consequences of cyclic in the scale climatic variability on population dynamics during northeast Atlantic Ocean during the past 130,000 years, Quat. Res. 29 (1988) 142–152. the Middle and the Upper Palaeolithic. Further studies [11] E. Bard, et al., Present status of radiocarbon calibration and com- will surely benefit from more and more accurate data, parison records based on Polynesian corals and Iberian Margin notably Sea Surface Temperatures reconstructions. sediments, Radiocarbon 46 (2004) 1189–1202. Future modelling studies will have to use these new [12] J. Zilhao, Le passage du Paléolithique moyen au Paléolithique data (from Gulf of Cadiz [34], off western [35] supérieur dans le Portugal, in: Cabrera (Ed.), El Origen del Hombre Moderno en el Suroeste de Europa, Universidad Nacional de but also from Mediterranean Sea) to better constrain Educacion a Distancia, Madrid, 1993, pp. 127–145. boundary conditions for H4 as well as for the other [13] J.C. Finlayson, Neanderthals and Modern Humans. An Ecological abrupt climatic events of the last 30,000 yr. and Evolutionary Perspective, Cambridge University Press, 2004. [14] J. Zilhao, Chronostratigraphy of the Middle-to-Upper Paleolithic – Acknowledgements Transition in the Iberian Peninsula, Pyrenae 1 (2006) 7 84. [15] D. Genty, et al., Precise dating of Dansgaard–Oeschger climate oscillations in western Europe from stalagmite data, Nature 421 We thank Didier Roche for precious comments (2003) 833–837. concerning the timing of the H4, Edouard Bard for in- [16] W. Broecker, et al., Origin of the northern Atlantic's Heinrich formation concerning the dating of this event, E. Cortijo events, Clim. Dyn. 6 (1992) 265–273. and Joao Zilhão for providing data and helpful comments [17] F.E. Grousset, et al., Were the North Atlantic Heinrich events triggered by the behavior of the European ice sheets? Geology 28 on the manuscript, H. Valladas and N. Mercier for discus- (2000) 123–126. sion and A. Friend and W. Banks for their editorial input. [18] E. Cortijo, et al., Changes in sea surface hydrology associated We also thank an anonymous reviewer for comments. with Heinrich event 4 in the North Atlantic Ocean between 40° This work has been supported by the French Ministry of and 60°N, Earth Planet. Sci. Lett. 146 (1997) 29–45. Research (ACI Espace et territoires), the Centre National [19] N.J. Shackleton, et al., Absolute calibration of the time scale: implications for Antarctic time scales and for Delta C-14, de la Recherche Scientifique (Eclipse and OHLL pro- Quat. Sci. Rev. 23 (2004) 1513–1522. grammes), and the European Science Foundation (OMLL [20] D. Roche, et al., Constraints on the duration and freshwater and Euroclimate programmes-EC Sixth Framework release of Heinrich event 4 through isotope modelling, Nature Programme under Contract no. ERAS-CT-2003-980409). 432 (2004) 379–382. [21] E. Barron, D. Pollard, High-resolution climate simulations of Oxygen Isotope Stage 3 in Europe, Quat. Res. 58 (2002) 296–309. References [22] B. Huntley, et al., European vegetation during Marine Isotope Stage-3, Quat. Res. 59 (2003) 195–212. [1] C. Finlayson, et al., Late survival of Neanderthals at the south- [23] M. Kageyama, et al., The Last Glacial Maximum and Heinrich ernmost extreme of Europe, Nature 443 (2006) 850–853. Event 1 in terms of climate and vegetation around the Alboran [2] F. d'Errico, M.F. Sanchez Goñi, Neanderthal extinction and the Sea: a preliminary model-data comparison, C. R. Geosci. 337 millenial scale climatic variability of OIS 3, Quat. Sci. Rev. 22 (2005) 983–992. (2003) 769–788. [24] M.F. Sanchez Goñi, et al., Synchroneity between marine and [3] T.H. Van Andel, W. Davies, Neanderthals and Modern Humans terrestrial responses to millenial scale climatic variability during in the European Landscape During the Last Glaciation: Archae- the last in the Mediterranean region, Clim. Dyn. 19 ological Results of the Stage 3 Project, The McDonald Institute (2002) 95–105. for Archaeological Research, 2004 265 pp. [25] N. Combourieu Nebout, et al., Enhanced aridity and atmospheric [4] J. Zilhao, F. d'Errico, The chronology and taphonomy of the high-pressure stability over the western Mediterranean during the earliest Aurignacian and its implications for the understanding of North Atlantic cold events of the past 50 k.y. Geology 30 (2002) Neanderthal extinction, J. World Prehist. 13 (1999) 1–66. 863–866. [5] J.N. Conard, M. Bolus, Radiocarbon dating the appearance of [26] F. Hourdin, et al., The LMDZ4 general circulation model: climate modern humans and timing of cultural innovations in Europe; new performance and sensitivity to parameterized physics with results and new challenges, J. Hum. Evol. 44 (2003) 331–371. emphasis on tropical convection, Clim. Dyn. 27 (2006) 787–813. 292 P. Sepulchre et al. / Earth and Planetary Science Letters 258 (2007) 283–292

[27] G. Krinner, et al., A dynamic global vegetation model for studies [40] J. Guiot, Methodology of the last climatic cycle reconstruction of the coupled atmosphere–biosphere system, Glob. Biogeo- from pollen data, Palaeogeogr. Palaeoclimatol. Palaeoecol. 80 chem. Cycles 19 (2005). (1990). [28] A. Berger, M.F. Loutre, Insolation values for the climate of the [41] K.H. Roucoux, et al., Combined marine proxy and pollen last 10 million years, Quat. Sci. Rev. 10 (1991) 297–317. analyses reveal rapid Iberian vegetation response to North [29] J.R. Petit, et al., Climate and atmospheric history of the past Atlantic millenial-scale climate oscillations, Quat. Res. 56 (2001) 420,000 years from the Vostok ice core. Antarctica, Nature 399 128–132. (1999) 429–436. [42] M.F. Sanchez Goñi, et al., European climatic response to [30] CLIMAP Members, Seasonal Reconstructions of the Earth's millenial-scale changes in the atmosphere–ocean system during Surface at the Last Glacial Maximum, 1981. the Last Glacial Period, Quat. Res. 54 (2000) 394–403. [31] C. Waelbroeck, et al., Sea-level and deep water temperature [43] T. Higham, et al., Revised direct radiocarbon dating of the changes derived from benthic foraminifera isotopic records, Vindija G1 Upper Paleolithic Neanderthals, Proc. Natl. Acad. Quat. Sci. Rev. 21 (2002) 295–305. Sci. U. S. A. 103 (2006) 553–557. [32] W.R. Peltier, paleotopography, Science 265 (1994) 195–201. [44] J. Zilhao, et al., Analysis of Aurignacian interstratification at the [33] N.S. Arnold, et al., Extent and dynamics of the Scandinavian Ice Chatelperronian-type site and implications for the behavioral Sheet during Oxygen Isotope Stage 3 (65,000–25,000 yr B.P.), modernity of Neanderthals, Proc. Natl. Acad. Sci. U. S. A. 103 Quat. Res. 57 (2002) 38–48. (2006) 12643–12648. [34] A.H.L. Voelker, et al., Mediterranean outflow strengthening [45] J. Zilhao, P.B. Pettitt, On the new dates for Gorham's Cave and during northern hemisphere coolings: a salt source for the glacial the late survival of Iberian Neanderthals, Before Farming 3 Atlantic? Earth Planet. Sci. Lett. 245 (2006) 39–55. (2006) 1–9. [35] M. Zhao, et al., Molecular stratigraphy of cores off northwest [46] P. Mellars, Neanderthals and the modern human colonization of Africa: history over the last 80 ka, Europe, Nature 432 (2004) 461–465. Paleoceanography 10 (1995) 661–676. [47] L.G. Vega, El transito del Paleolitico medio al Paleolitico [36] A. Jost, et al., High-resolution simulations of the last glacial superior en el Sur de la Peninsula Iberica, El Origen del Hombre maximum climate over Europe: a solution to discrepancies with Moderno en el Suroeste de Europ, Universidad Nacional de continental palaeoclimatic reconstructions? Clim. Dyn. 24 (2005) Educacion a Distancia, Madrid, 1993, pp. 147–170. 577–590. [48] J. Carrion, The use of two pollen records from deep sea cores to [37] W. Cramer, Using Plant Functional Types in a Global Vegetation frame adaptive evolutionary change for humans: a comment on Model, Plant Functional Types Their Relevance to Ecosystem “Neanderthal extinction and the millennial scale climate Properties and Global Change, Cambridge University Press, variability of OIS 3” by F. d'Errico and MF Sanchez Goni, Cambridge, 1997. Quat. Sci. Rev. (2004). [38] M. Elliot, et al., Changes in North Atlantic deep-water formation [49] C. Finlayson, et al., Did the moderns kill off the Neanderthals? associated with the Dansgaard–Oeschger temperature oscilla- A reply to F. d'Errico and Sanchez Goni, Quat. Sci. Rev. (2004). tions (60–10 ka), Quat. Sci. Rev. 21 (2002) 1153–1165. [50] C. Finlayson, J.S. Carrion, Rapid ecological turnover its impact [39] R. Cheddadi, et al., Holocene climatic change in Morocco: a on Neanderthal and other human populations, Trends Ecol. Evol. quantitative reconstruction from pollen data, Clim. Dyn. 14 (1998). 20 (2007) 457–463.