ARTICLE IN PRESS

Quaternary Science Reviews 25 (2006) 2820–2834

Comparing proxies for the reconstruction of LGM sea-surface conditions in the northern North Atlantic A. de Vernala,, A. Rosell-Mele´b, M. Kucerac, C. Hillaire-Marcela, F. Eynaudd, M. Weinelte, T. Dokkenf, M. Kageyamag b

a GEOTOP, Universite´ du Que´bec a` Montre´al, C.P. 8888, Succursale ‘‘centre ville’’, Montre´al, QC, Canada H3C 3P8 ICREA and Institute of Environmental Science and Technology, Universitat Auto`noma de Barcelona, 08193 Bellaterra, Catalonia, Spain c Institut fu¨r Geowissenschaften Eberhard Karls Universita¨t Tu¨bingen, Sigwartstrasse 10, DE-72076 Tu¨bingen, Germany d De´partement de Ge´ologie et Oce´anographie, UMR5805-EPOC, CNRS/Universite´ Bordeaux 1, 33405 Talence Cedex, France e Institute for Geosciences, University of Kiel, Olshausen Strasse 40, D-24098 Kiel, Germany f Bjerknes Centre for Climate Research, Alle´gaten 55, 5007 Bergen, Norway g LSCE/IPSL, CE Saclay, L’Orme des Merisiers, Baˆtiment 701, 91191, Gif-sur-Yvette Cedex, France

Received 10 June 2005; accepted 28 June 2006

Abstract Within the frame of the multiproxy approach for the reconstruction of the glacial ocean (MARGO) project, sea-surface conditions of the Last Glacial Maximum (LGM, 23–19 ka) were reconstructed using different proxies, which were calibrated to a standardized modern hydrography. In the North Atlantic, the revised LGM MARGO data set provides a comprehensive coverage, including the Nordic Seas. The data set includes reconstruction based on planktonic foraminifer assemblages, dinocyst assemblages, alkenone coccolithophorid biomarkers and Mg/Ca ratios in planktonic foraminifers. Several hydrological features of the LGM North Atlantic can be identified, that are supported by all four proxies. They include an extensive perennial sea-ice cover along the eastern Canadian and Greenland margins, seasonally ice-free central North Atlantic and Norwegian Sea, and the winter sea-ice limit being located at about 551N. All proxies also suggest significantly colder than modern sea-surface conditions in the southeastern part of the North Atlantic and a more zonal temperature pattern than at present, with a steep SST gradient of the order of 10 1C between 401N and 451N. However, in the Nordic Seas large discrepancies remain, well above the level of uncertainty of the SST reconstructions. These discrepancies might be related to stratification of the upper water mass linked to large annual amplitude of temperature with contrasting winter and summer conditions, high interannual-interdecadal variability, or taphonomic processes affecting some of the proxies. The average LGM SST in the Nordic Seas cannot at present be assessed with confidence. However, the existing evidence suggests that highly variable sea-surface conditions and at least occasional advection of North Atlantic waters may have characterised the glacial Nordic Seas. r 2006 Elsevier Ltd. All rights reserved.

1. Introduction The Last Glacial Maximum (LGM) interval has long served as a common target for numerical model experiments (e.g. Manabe and Broccoli, 1985; Kutzbach and Wright, 1986; Rind, 1987). Within the frame of the paleoclimate model intercomparison project (PMIP) undertaken in the early 1990s, this interval has been selected as one of the critical time-slices used for evaluating the Corresponding author.

E-mail address: [email protected] (A. de Vernal). 0277-3791/$ - see front matter r 2006 Elsevier Ltd. All rights reserved. doi:10.1016/j.quascirev.2006.06.006

capability of climate models to simulate the climate under conditions significantly different from the present (e.g., Joussaume and Taylor, 1995; IPCC, 2001; Braconnot, 2004). The global reconstruction of LGM sea-surface temperatures (SSTs) produced by CLIMAP (1981) constituted the basis of the early phase of PMIP. The CLIMAP (1981) SST reconstruction was used as a boundary condition required for simulations using GCMs. However, the evaluation of GCMs conducted before the PMIP project (Rind and Peteet, 1985) and within the frame of PMIP (Pinot et al, 1999a, b) pointed to important discrepancies, which motivated efforts of the

ARTICLE IN PRESS A. de Vernal et al. / Quaternary Science Reviews 25 (2006) 2820–2834

paleoceanographical community to revisit the LGM in the light of new methodological developments. The second phase of PMIP deals with Atmosphere–Ocean fully coupled general circulation models independent of forcing by paleodata. Therefore, SST estimates are not used as a boundary condition to the models as in PMIP1, but rather to evaluate the performance of the atmosphere–ocean models. In turn, models can suggest mechanisms for the changes in SST reconstructed by the different proxies. SSTs need to be as robust as possible within reliable confidence interval in order to make sensitive evaluations of the coupled models. SST reconstructions produced by CLIMAP (1981) were based on transfer functions using multiple regression techniques and principally relied on planktonic foraminifer data (Imbrie and Kipp, 1971). Since this pioneering work, progress in the fields of paleoceanography and palaeoclimatology have led to the development of new proxies including biological indicators, such as dinocyst assemblages, as well as new methods for analyses of microfossil assemblage count data. These include techniques using the degree of similarity between fossil and modern assemblages (cf. for example, Guiot, 1990; Pflaumann et al., 1996; Waelbroeck et al., 1998), and artificial neural network algorithms (Malmgren et al., 2001). In addition to the above-mentioned approaches, biogeochemical analyses of organic compounds, such as alkenones produced by coccolithophorids, or the measurement of trace elements, such as Mg/Ca or Sr/Ca in biogenic calcite, yielded further insights into past temperatures in the water column (e.g., Brassell et al., 1986; Mu¨ller et al., 1998; Lea et al., 1999; Nu¨rnberg et al., 2000). Many of these recently developed methods have been used for estimating LGM sea-surface conditions. In the northern North Atlantic, for example, there are now several LGM data sets available based on planktonic foraminifera (Weinelt et al., 1996; Pflaumann et al., 2003; Sarnthein et al., 2003a, b; Kucera et al., 2005a), dinoflagellate cysts (de Vernal et al., 2000, 2002, 2005), alkenone biomarkers (Rosell-Mele´, 1997; Rosell-Mele´ and Comes, 1999; Rosell-Mele´ et al., 2004), and Mg/Ca in planktonic foraminiferal shells (Barker et al., 2005; Meland et al., 2005). With the objective of comparing to compare paleoceanographical reconstructions based on different proxies, an intercomparison exercise has been undertaken within the frame of the multiproxy approach for the reconstruction of the glacial ocean (MARGO) project. The first step was to adopt a common time frame for the LGM interval, defined by the environmental processes of the ice age: land, oceans, and glaciers (EPILOG) initiative as the time of maximum continental ice volume during the last glaciation, ranging between 23 and 19 calendar ka before present (Schneider et al., 2000; Mix et al., 2001). In the North Atlantic, the LGM thus corresponds to an interval embedded between Heinrich events H2 and H1 (Bond et al, 1992; Mix et al., 2001). A common hydrography was also adopted for the

2821

calibration of the temperature vs. proxy relationships, in order to avoid any bias that can be related to initial oceanographical data inputs. The ‘‘standardised’’ hydrography that has been selected for the MARGO exercise is the 1998 version of the World Ocean Atlas produced by the National Oceanographic Data Centre (NODC). Reconstructions based on individual proxies have been produced according to the MARGO rules and published in a special issue of Quaternary Science Reviews (cf. Kucera et al., 2005b). Here, we present some of the MARGO data for the northern North Atlantic (Fig. 1). This is an oceanic region of special importance with respect to the general thermohaline circulation because the formation of deep waters take place in the Nordic and Labrador Seas (e.g., Dickson and Brown, 1994; Dickson et al., 1996). The past intensity of the Atlantic meridional overturning (AMO) and the location of the deep water formation are of primary interest in paleoclimate modelling (e.g., Weaver et al., 1998). However, the glacial northern North Atlantic proved to be a region particularly difficult to reconstruct, as significant discrepancies emerged from the comparison of LGM SST estimates based on different proxies. The objective of this paper is to examine the communalities and discrepancies among LGM sea-surface condition reconstructions based on selected proxies, determine the possible causes of the discrepancies and make suggestions as to how these could be explained or reconciled. 2. The methods 2.1. Source of data The LGM SST reconstructions in the Northern North Atlantic (see http://www.pangaea.de/Projects/MARGO/) are based on proxies that include planktonic foraminifer assemblages (cf. Weinelt, 2004; Kucera et al., 2005a), organic-walled dinoflagellate cyst assemblages (cf. de Vernal et al., 2005), and alkenones biomarkers (cf. Lee, 2004; Rosell-Mele´ et al., 2004). For a few sites, SST estimates based on Mg/Ca measurements in surface dwelling foraminifers are also available (cf. Barker et al., 2005). The data sets are thus based on calcareous or organic biological remains, which appear to have been well preserved in LGM sediments of the North Atlantic. There is no reconstruction available from siliceous biological remains such as diatoms because of poor preservation of opaline silica in glacial sediments of this basin (Koc- et al., 1993; Lapointe, 2000). Exhaustive data sets of d18O in foraminifera are available (e.g., Duplessy et al., 1991; Weinelt et al., 1996; Waelbroeck, 2004; Meland et al., 2005) and may provide extremely important information on the structure of the upper water column (e.g., de Vernal et al., 2002; Waelbroeck et al., 2005). However, these data are not included in the present compilation because they relate to temperature, salinity and changes in isotopic composition of sea water, aside from the fact that the dominant taxon at

ARTICLE IN PRESS A. de Vernal et al. / Quaternary Science Reviews 25 (2006) 2820–2834

2822

Distribution of MARGO Last Glacial Maximum SST proxies Planktonic foraminifers Alkenones Dinocysts Mg/Ca in planktonic foraminifers Fig. 1. The location of core sites used for reconstruction of LGM conditions and presented in this study. The thin blue line corresponds to the 200 m isobath.

high latitude, Neogloboquadrina pachyderma left coiling, lives in mesopelagic waters and does not necessarily provide information about the surface layer, but about conditions

at variable depths along the pycnocline (e.g., HillaireMarcel et al., 2001a, b, 2004; Simstich et al., 2003). For the same reason, Mg/Ca data available from measurements

ARTICLE IN PRESS A. de Vernal et al. / Quaternary Science Reviews 25 (2006) 2820–2834

in N. pachyderma left coiling were not included in the compilation. The use of d18O in foraminifers to estimate temperature also requires assessment of many parameters (isotopic composition of fresh water end-members, ice meltwater, etc.) and needs independent discussion (cf. for example, Hillaire-Marcel et al., 2001b: de Vernal et al., 2002; Meland et al., 2005). 2.2. Uncertainties of sea-surface paleoreconstructions Each proxy has been calibrated against SST of the World Ocean Atlas 1998 for strandardisation of the modern hydrography within the framework of the MARGO project. The errors of prediction of SST estimates have been calculated in different ways depending upon the proxy. For biomarkers and trace elements, the error of prediction corresponds to the standard deviation around the calibration line yielding the paleotemperature equation. For microfossil data, the error of prediction is calculated by reconstructing modern SSTs from assemblages in surface sediment based on various transfer function methods (modern analogue and artificial neural network techniques; cf. Kucera et al., 2005a; de Vernal et al., 2005). On average, the error of prediction of SST is about 71 and 71.5 1C, respectively, for reconstruction based on planktonic foraminifers and alkenones. The error of prediction ranges between 71.2 and 71.7 1C, respectively, for winter and summer for reconstructions based on dinocyst assemblages that are proxies for estuarine, epicontinental and oceanic environments. For all proxies, the value of the error of SST prediction is close to the standard deviation of observed mean seasonal temperatures as reported in the World Ocean Atlas (see http://www.nodc.noaa.gov/OC5/ WOA98F/woaf_cd/search.html). This suggests that the approaches for SST reconstruction are adequate. Whilst the ability of the selected proxies to reconstruct the temperature in the modern ocean is unquestionable, some are less sensitive at extreme ends of the calibration. This is especially the case in the coldest waters, where planktonic foraminifers suffer from lack of faunal variation due to extremely low species diversity (e.g., Kucera et al., 2005a). Similarly, the dispersal of values around the calibration equations are particularly large in the cold domains for the alkenone index (e.g., Rosell-Mele´, 1998; Bendle et al., 2005), and Mg/Ca ratios (e.g., Lea et al., 1999). In addition to these intrinsic calibration problems, it is not possible to unambiguously determine which depth in the water column or which season is more accurately recorded by each proxy. Although all calibrations for SST estimations are based on surface (0–10 m) conditions, the proxy-carrying microplanktonic organisms occupy a wider range of depths in the water column depending upon their trophic behaviour and ecological affinities. Therefore, the reconstructions are based on the assumption that physical gradients (temperature or salinity) in the water column as well as the preferred habitats of individual species remain the same through time. Such an assumption may be acceptable for proxies

2823

based on autotrophic organisms, such as alkenones or many dinocysts, since they necessarily relate to production in the photic zone that is rarely deeper than the mixed layer. However, this is more critical when dealing with heterotrophic organisms, such as foraminifers, since they are mobile in the water column and may occupy mesopelagic layers. The annual amplitude of temperature or seasonal contrast of temperature is another source of uncertainty. The seasonality of the temperature signal depends upon stratification and thermal diffusion in the upper water column. In the open ocean, especially in modern offshore environments of the North Atlantic, the weak stratification fosters high thermal inertia in a thick mixed layer that results in low thermal amplitude from winter to summer. The upper water layer is therefore characterised by a strong correlation between summer and winter SSTs. For this reason, most open ocean proxies generally give a ‘‘mixed temperature’’ signal that does not allow seasonal temperatures to be easily distinguished. This is the case of planktonic foraminifer assemblages, which are calibrated against summer and winter SSTs in the open-ocean domain where the seasonal temperatures correlate together. In the hypothesis of an oceanic regime different from that of today with respect to stratification in the upper water mass, the validity of the seasonal signal from the planktonic foraminifers could be questioned. Another point to note is that the dominant species of the planktonic foraminifer assemblages in polar waters (N. pachyderma left coiled) inhabits subsurface waters (e.g., Be´ and Tolderlund, 1971) and may occur below the thermocline as it is often the case in the Arctic, subarctic seas and other stratified waters where it develops along or below the halocline (e.g., Kohfeld et al., 1996; de Vernal et al., 2002; Simstich et al., 2003; Hillaire-Marcel et al., 2004). Moreover, planktonic foraminifers, which are relatively stenohaline, are not abundant in low salinities and do not necessarily yield an accurate picture of the sea-surface conditions in stratified water masses marked by a sharp halocline. In the epicontinental low salinity (o33) environments such as shelves of Arctic and subarctic seas, the Hudson Bay, or Gulf of St-Lawrence, sediments contain extremely rare planktonic foraminifers (e.g., Bilodeau et al., 1990; Rodrigues et al., 1993; Hillaire-Marcel et al., 1994; Lubinski et al. 2001). In the Barents Sea, for example, summer SSTs reach up to 14 1C, but planktonic foraminifers belonging almost exclusively to N. pachyderma left coiled are recovered only at deeper sites, below the halocline (about 60 m) in a water mass having a temperature lower than 4 1C (cf. Lubinski et al., 2001). The dinocyst assemblages were used for simultaneous reconstruction of the winter and summer SSTs, in addition to summer salinity and sea-ice extent in number of months with sea-ice density greater than 50%. Because they include many polar and subpolar species, dinocyst assemblages may be a more accurate proxy for the reconstruction of sea-surface conditions in cold environments. Moreover,

ARTICLE IN PRESS 2824

A. de Vernal et al. / Quaternary Science Reviews 25 (2006) 2820–2834

because the modern database is representative of a wide range of environmental settings, including estuaries and epicontinental seas, dinocysts can provide indications of various hydrographic conditions with respect to surface salinity and stratification. However, weaknesses of dinocyst based SST estimates for the LGM include the relatively distant similarity of the modern analogues that are found for LGM assemblages. Another source of uncertainty is the low dinocyst productivity and fluxes, especially in the Nordic Seas, which could have resulted in overrepresentation of transported material (cf. discussion in de Vernal et al., 2005). In the case of low average productivity, an overrepresentation of populations that developed during exceptionally high productivity episodes also needs to be considered, but this applies to most proxies. The alkenone data are calibrated against mean annual SSTs (Mu¨ller et al., 1998) using the ratio of the biomarkers measured in a worldwide compilation of surface sediments. The calibration equation is statistically identical to the one derived initially from a culture of the North Pacific strain of the ubiquitous coccolithophorid Emiliania huxleyi (Prahl and Wakehman, 1987). It is inevitable to assume that the sediment signal is skewed towards the time of maximum production of alkenone, but in present-day ocean, calibrations based on annual average SST appear to yield smaller errors than those based on seasonal or monthly temperature (Mu¨ller et al., 1998). This suggests a continuous production during the year or a production during months when SST values are correlated to the annual average. Records of alkenone-based reconstructions of SSTs have been interpreted accounting for shifts in seasonality of alkenone production (e.g., Haug et al., 2005). Hence, there is the possibility that alkenone production was more concentrated in summer months during the LGM than at present in the North Atlantic, which would impinge in the interpretation of the results and perhaps the calibration. In the high-latitude regions, the time of maximum production could conceivably occur in the summer, rather then during the spring or autumn (e.g., Antoine et al., 1996). Another source of uncertainty with the alkenone approach concerns the presence of low sea-surface salinity or sea-ice on the ratio that is used for calibration equations. In such conditions, there is a particularly large error of prediction and the alkenones are characterised by a high proportion of tetra-unsaturated molecules (cf. Rosell-Mele´ and Comes, 1999; Bendle et al., 2005). Another caveat concerns the possibility of a mixed signal due to allochthonous material transported for example with icebergs. The presence of old reworked coccoliths and old reworked organic matter can be used to identify allochtonous material and could serve as an indicator of results that are likely to be biased and should therefore be discarded. In this study, a conservative approach was used and all results from LGM samples collected in the Nordic Seas were rejected because of their high proportion of tetra-unsaturated alkenones and the presence of reworked organic matter in the sediments.

The Mg/Ca measurements used in this study were made in shells of epipelagic foraminifers (Globigerina ruber or Globigerina bulloides) and calibrated against mean annual SSTs. Uncertainties in the calibration of Mg/Ca vs. temperature are large in the low temperature domain (e.g., Lea et al., 1999). Uncertainties also concern the calcification depth of species used for measurements in addition to possible species-specific vital effect (see discussion in Barker et al., 2005; Skinner and Elderfield, 2005). To avoid uncertainties as much as possible, only a few results were compiled from the MARGO intercomparison exercise: they include only sites from mid-latitudes and measurements performed on species inhabiting in the mixed layer and/or the top of the seasonal pycnocline. Here again, it should be mentioned that a conservative approach has been chosen. An LGM data set based on Mg/Ca measurements in shells of N. pachyderma is available (Meland et al., 2005). It was not used here as SST indicators because the measurements do not represent necessarily the sea-surface conditions. However, the results yield complementary information on the thermal structure of the water column and are referred to in the text (see Section 3.3). 2.3. Presentation of SST results A direct comparison of SST estimates from the different proxies is not easy because seasonal signals of reconstructed SSTs are not identical for all proxies and because the reconstructions rely on different calibration data sets with respect to the winter vs. summer temperature ranges. For example, the dinocyst data set includes data from epicontinental areas with greater and more variable seasonal contrast of temperature than is seen in the open ocean data set of planktonic foraminifers. Although biases are possible, we illustrate the mean annual LGM–SSTs and anomalies on the maps of Fig. 2 for direct comparison. We used the annual mean of SSTs estimated from planktonic foraminifers (Fig. 2a) and calculated from alkenone data (Fig. 2c). In the case of dinocysts, the mean annual SSTs were not reconstructed directly but calculated as an average of the reconstructed summer and winter SSTs (Fig. 2b). In addition, we illustrate the full seasonal range of reconstructed SSTs in Fig. 3. The seasonal range of estimated SSTs is shown for reconstructions based on planktonic foraminifers and dinocysts. A confidence interval was calculated around the mean annual temperature using one standard deviation around the average value from alkenone and Mg/Ca results. 3. LGM sea-surface temperatures 3.1. Estimates from individual proxies Despite standardisation with respect to the LGM chronological frame and the source of hydrographical

ARTICLE IN PRESS A. de Vernal et al. / Quaternary Science Reviews 25 (2006) 2820–2834

2825

Fig. 2. Map illustrating the reconstructed SSTs at LGM (left) and the LGM-Modern SST anomalies (right) in the North Atlantic based on planktonic foraminifers (Weinelt, 2004; Kucera et al., 2005a, b), alkenones (Lee, 2004), and dinocysts (de Vernal et al., 2005). The SSTs and anomalies are expressed in 1C. The left diagrams show indications of sea-ice cover: the planktonic foraminifer data (upper left) include a zone shaded in blue that corresponds to more than 99% of N. pachyderma left coiling and that is interpreted as possible perennial sea-ice, whereas the dashed line corresponds to more than 1% of subpolar taxa and delimitates the zone of seasonally ice-free conditions (Kucera et al., 2005a). The dinocyst data (middle left) include a zone shaded in grey that corresponds to an area with barren assemblages that is associated with perennial sea-ice, and the dashed line corresponds to the limit between seasonal sea-ice and year-round ice-free conditions (cf. de Vernal et al., 2005).

ARTICLE IN PRESS A. de Vernal et al. / Quaternary Science Reviews 25 (2006) 2820–2834

2826

90

80

LGM sea-surface temperatures vs. latitude in the North Atlantic: a multiproxy reconstruction

Nordic Seas

Latitude (degrees North)

70

Annual range of SSTs estimated from planktonic foraminifers

Iceland

East of 30°W West of 30°W

60

Annual range of SSTs estimated from dinocysts

50

East of 30°W West of 30°W

40

SSTs estimated from alkenones 30

Annual mean ± 1 σ (East of 30°W) Annual mean ± 1 σ (West of 30°W)

SSTs estimated from Mg/Ca

20

Annual mean ± 1 σ

10 -5

0

5

10

15

20

25

30

Sea-surface temperature (°C) Fig. 3. The latitudinal distribution of LGM SSTs in the North Atlantic based on reconstructions using foraminifers (Weinelt, 2004; Kucera et al., 2005), alkenones (Lee, 2004), dinocysts (de Vernal et al., 2005) and Mg/Ca measurements in epipelagic foraminifers (Barker et al., 2005). The annual range of estimated temperatures is shown based on planktonic foraminifer and dinocysts data. A distinction between the eastern and western sites has been made to illustrate the regional pattern that seems to characterize the North Atlantic at LGM.

data in the calibration data sets, the LGM SSTs show significant differences depending upon the proxy used for reconstruction. Planktonic foraminifer data present an excellent spatial coverage, especially for the oceanic basins east of 301W. The results indicate a zonal distribution of temperatures, with much colder than present conditions except in the northwestern Nordic Seas (Fig. 2). The new LGM SSTs estimates based on planktonic foraminifers suggest warmer conditions than those published by CLIMAP (1981). With a few exceptions, however, they still indicate colder conditions than all other proxies, with temperatures lower than 2 1C, especially north of 701N. The LGM dinocyst data present a good coverage from the west to the east, especially in areas north of 451N. The results indicate distinct regional patterns instead of zonal distribution of temperatures, with much colder conditions along the European and Canadian margins than offshore in the central area of the North Atlantic. The mean annual SST anomalies suggest colder than modern conditions at most locations, but equivalent to present in the Irminger Sea, and warmer in the Greenland and Iceland Seas. The summer and winter SST estimates also indicate very large annual amplitude of temperature ranges with pronounced seasonal contrasts (Fig. 4). In addition to SSTs, the dinocysts allow the reconstruction of sea-surface salinity, which suggests that glacial sea-surface salinity in the

northern North Atlantic has been much lower than at present by more than 1 unit at most sites (cf. de Vernal et al., 2000, 2005), in response to freshwater discharges from the surrounding Fennoscandian and Laurentide icesheets. Low sea-surface salinity is compatible with the existence of a sharp pycnocline and a low thermal inertia in the shallow surface layer, thus fostering winter cooling and summer warming of the mixed layer. Slightly lower salinity than at present may have characterised the subsurface waters occupied by N. pachyderma left coiling as calculated by Meland et al. (2005) using an approach combining Mg/ Ca and isotope measurements. The salinities estimated by Meland et al. (2005) are, however, much higher than those reconstructed based on dinocysts, which again suggests the existence of stratified upper water masses. The LGM alkenone data present a sparse coverage, with all data retained for the compilation being located south of 601N. The available results suggest SSTs slightly lower than at present in central North Atlantic, but slightly warmer in the Irminger Basin and Labrador Sea. The SST estimates from this proxy are compatible with those from dinocysts, although they suggest even warmer conditions if they really captured the signal of the mean annual temperature. It is possible that the alkenone-based SST reconstruction might be biased towards warmer than average conditions or might represent a dominantly summer signal, if the growth season of alkenone-producing

ARTICLE IN PRESS A. de Vernal et al. / Quaternary Science Reviews 25 (2006) 2820–2834

30 Calibration data (dinocyst database) SSTs estimates based on planktonic foraminifers SSTs estimates based on dinocysts

25

Winter SSTs

20

15

10

5

0

-5 -5

0

5

10

15

20

25

30

Summer SSTs Fig. 4. Graph showing the summer vs. winter LGM SSTs as reconstructed from foraminifers (blue squares) and dinocysts (green diamonds). Clearly, the dinocyst data show larger summer vs. winter SST differences than the foraminifers. This is possible since the modern dinocyst database (grey squares) includes estuarine and neritic areas, where the thermocline depth is shallow, in addition to offshore sites: it is thus representative of a wide range of seasonal contrasts. A seasonal gradient of temperature larger at LGM than at present in the offshore North Atlantic would account for a poor analogue situation for all proxies.

coccolithophores shifted towards the summer. The existence of large seasonal contrasts of temperatures during the LGM could thus account for the apparently warm annual signals as recorded by this proxy. It is noteworthy that LGM alkenone data from the Nordic Seas, north of 601N, were discarded because of a high proportion of tetraunsaturated alkenones (Rosell-Mele´ and Comes, 1999), which suggests the existence of seasonal sea-ice and low sea-surface salinity (Bendle et al., 2005). In addition, there is the possibility of contamination from allochthonous alkenones. This has been suggested after the identification of reworked, pre-Quaternary, coccoliths in glacial sediments in the North Atlantic (Weaver et al., 1999; Balestra and Ziveri, pers. Comm.), and alkenones in debris flows of fans from the North Atlantic continental margins (Kornilova, 2004). Finally, a few Mg/Ca measurements in surface dwelling or epipelagic foraminifers from the central part of the North Atlantic tend to indicate slightly colder than present conditions in the corresponding water layers. The estimated SSTs are included within the range of values yielded by other proxies, except for one site at 301N. 3.2. The consistent and reconcilable features Glacial SST reconstructions for latitudes south of 501N appear to be compatible among all four proxies, with the

2827

exception of one Mg/Ca-based estimate. Between 201N and 401N, SST estimates suggest temperatures ranging between 151 and 22 1C throughout the annual cycle (Fig. 3). Proxies also tend to indicate a steep SST gradient around 451N, which appears particularly sharp with respect to the reconstructions based on planktonic foraminifers. Between 501N and 621N, reconstructions from all four proxies indicate apparently different SSTs. However, the estimates overlap when taking into account the large annual amplitude of temperatures as reconstructed from dinocyst data (Fig. 4). Alkenone production during the coccolithophore bloom would have occurred during the warmest part of the year given that the oceanic regime at the LGM could have been substantially different owing to the presence of seasonal sea-ice cover (cf. de Vernal and Hillarie-Marcel, 2000; de Vernal et al., 2000, 2005). Thus alkenones would have blomed during the summer season and would have recorded larger summer temperature instead of annual mean. The LGM was probably marked by different seasonal contrast of temperatures than in the modern ocean. In a similar manner, low surface salinity during the warm (melt) season may have inhibited the production of surface dwelling foraminifers leading to an overrepresentation of the sub-pycnocline fauna. A large seasonal contrast of temperature together with a sharp halo-thermocline during summer could explain why planktonic foraminifers seem to have preferentially recorded a cold winter-like SST signal.

3.3. The remaining puzzling feature: the Nordic Seas The LGM reconstructions north of Iceland in the Nordic Seas are problematic. Even assuming very large seasonal contrasts of temperature and taking into account the full range of temperatures, there is no overlap between SST estimates derived from dinocysts and from planktonic foraminifers. Obviously, the different reconstructions cannot be reconciled in a simple manner (see discussion in Section 5). The interpretation of the various data from the Nordic Seas is difficult not only with respect to the discrepancies between the dinocyst and planktonic foraminifer records, but also with respect to results obtained from other proxies including coccoliths, alkenones and Mg/Ca in foraminifera. Coccolith assemblages in glacial samples from Nordic Seas are dominated by E. huxleyi and suggest the existence of relatively warm conditions (up to 14 1C) at least during the summer season (cf. de Vernal et al., 2000). The alkenone data from the Nordic Seas, which were discarded because of possible contamination with allochthonous components and a high proportion of tetra-unsaturated alkenones, correspond to production under relatively warm conditions, ranging up to 15 1C (Rosell-Mele´ and Comes, 1999). Finally, Mg/Ca measurements in N. pachyderma from the central Nordic Seas yielded ‘‘unrealistic’’ values,

ARTICLE IN PRESS 2828

A. de Vernal et al. / Quaternary Science Reviews 25 (2006) 2820–2834

corresponding to relatively high temperatures around 10 1C (Meland et al., 2005).

5. The Nordic Seas at the time of the LGM 5.1. The paradox of glacial SST reconstructions

4. Permanent and seasonal sea-ice margins The extent of the LGM sea-ice cover was reconstructed from two main micropaleontological tracers: dinocysts and planktonic foraminifers. The use of dinocyst assemblages relies on the strong relationship that has been observed between the distribution of some species and the seasonal sea-ice extent (de Vernal and Hillaire-Marcel, 2000; de Vernal et al., 2001), which results from the combined forcing of temperature and length of the ice-free season. It is believed that only a few taxa are able to live in sea-ice or to accomplish their entire biological cycle during a short growing season, whereas many species never develop in areas of seasonal sea-ice cover (Matthiessen et al., 2005). Based on modern analogue techniques, dinocyst assemblages were thus used for quantitative reconstruction of sea-ice cover, which is expressed as the number of months per year with sea-ice density greater than 50% (de Vernal et al., 2000, 2005). The other type of approach relies on planktonic foraminifers, which are used as indirect tracer for sea-ice limit based on a threshold of SST estimated from foraminifer assemblages (cf. Sarnthein et al., 2003a, b) or based on the occurrence of subpolar species indicating ice free conditions (cf. Kucera et al., 2005a). Unlike the dinocyst approach, planktonic foraminifer assemblages can only indicate which areas remained ice-free for at least some part of the year. Alkenones can also be used to infer indirectly conditions when sea-ice may have been present (Rosell-Mele´ and Comes, 1999) given that the abundances of the tetraunsaturated component indicates decreased surface salinity conditions as occurs during summer sea-ice melting (Bendle et al., 2005). The updated reconstructions of sea-ice (cf. Kucera et al., 2005a; de Vernal et al., 2005) and the distribution of tetraunsaturated alkenone (Rosell-Mele´ and Comes, 1999) are broadly compatible, suggesting that the LGM sea-ice cover was significantly more widespread than at present in the northern North Atlantic, but much less than proposed by CLIMAP (1981). The proxies indicate that extensive to permanent sea-ice (49 months/year) developed along the continental margins of eastern Canada and Greenland, whereas seasonally ice-free conditions characterised the Norwegian Sea and the central and eastern parts of the North Atlantic. The LGM winter sea-ice limit occurred at about 551N in the central and eastern North Atlantic, whereas the modern one is recorded at about 651N in the eastern North Atlantic. The LGM, however, does not correspond to the maximum extent of sea-ice during the last glaciation. Sea-ice spread further south during Heinrich events (de Vernal et al., 2000), no doubt in relation with massive iceberg discharges reaching about 401N (Ruddiman, 1977; Bond et al., 1992).

The various proxies suggest seasonally ice-free conditions, at least in the eastern Nordic Seas. However, planktonic foraminifer assemblages suggest maximum SSTs of about 4 1C (Weinelt et al., 1996; Pflaumann et al., 2003; Kucera et al., 2005a), whereas dinocyts suggest SSTs of up to 14 1C, i.e. similar or warmer than present conditions (de Vernal et al., 2000, 2005). LGM coccolith assemblages containing abundant E. huxleyi were recovered at several locations in the Nordic Seas, and led to reconstruction of relatively warm conditions using a northern North Atlantic-surbarctic modern database (Le´vesque, 1995; de Vernal et al., 2000). Similarly, despite uncertainties, alkenone data may imply a temperature within the 10–15 1C range (Rosell-Mele´ and Comes, 1999). The very large discrepancy recorded in the Nordic Seas suggests a dichotomy between proxies related to ‘‘autotrophic’’ production (coccoliths, alkenones, most dinocysts) indicating warm sea-surface conditions, and proxies related to ‘‘heterotrophic’’ production (planktonic foraminifers). This dichotomy may also concern the tolerance towards salinity: planktonic foraminifers are oceanic and relatively stenohaline, whereas dinoflagellates thrive in marine marginal settings and tolerate a wide range of salinities. Among coccolithophorids, E. huxleyi is a euryhaline species able to grow in relatively low salinities. E. huxleyi is also the main alkenone producer, and most probably accounts largely for the alkenone recorded in the North Atlantic and Nordic Seas during the LGM. Apart from primary differences in their ecology, the opposing proxies also differ in their taphonomy. Planktonic foraminiferal shells are substantially larger and heavier than dinoflagellate cysts and coccoliths and are thus much less prone to reworking and lateral transport in the water column. This issue is important to consider in view of the very low productivity that is reconstructed in the glacial northern North Atlantic based on biogenic fluxes or microfaunal assemblages (e.g., Hillaire-Marcel et al., 1994; Thomas et al., 1995; Rasmussen et al., 2003). In the Nordic Seas, concentrations of microfossils, dinocysts and alkenones are indeed very low, especially in the central part of the Greenland Sea (cf. de Vernal et al., 2000, 2005; Bendle et al., 2005). It is noteworthy that the LGM palaeogeography of the northern North Atlantic, with a low sea level and an ice margin eperon on the shelves (cf. Peltier, 1994, 2004), might have resulted in limited productivity in nearshore areas. Significantly reduced productivity in the northern North Atlantic at LGM could also be the result of decreasing nutrient supply to the photic zone due to both a southward displacement and reduction of the meridional overturning and the development of a strong halocline as simulated by Schmittner (2005) using an earth model of intermediate complexity. As a consequence of low autochthonous biogenic productivity,

ARTICLE IN PRESS A. de Vernal et al. / Quaternary Science Reviews 25 (2006) 2820–2834

lateral advection of particles may have had a strong effect on microfossil assemblages. This applies to all small size plankton, especially in the case of strong surface currents and/or advection or subduction of the surface water mass. In addition, the apparently low LGM productivity may also represent intermittent, relatively short intervals which could differ among the proxies. Considering sediment mixing by benthic organisms, two consequences could be anticipated: mixed proxies would relate to different SSTs, and reconstructed values would not be representative of the low productivity intervals, with the consequence that mean hydrographical conditions could not be reconstructed. The different ecology of the proxies and related taphonomical processes may be invoked to explain some of the discrepancies in their respective SST records. However, these differences could not explain the contradictory results obtained from the planktonic foraminifer assemblages and d18O of the N. pachyderma shells suggesting cold (4 1C or less) summer conditions and the high Mg/Ca ratio (41.1) also measured in the shells of N. pachyderma that would yield temperatures of up to 10 1C in the Nordic Seas, North of Iceland (Meland et al., 2005). These Mg/Ca values reflect temperature condititions incompatible with the usual niche for N. pachyderma and are considered ‘‘unrealistic’’ (Meland et al., 2005). Nevertheless, these results point again to the fact that LGM data

2829

in the Nordic Seas are extremely difficult to reconcile and to interpret. 5.2. The hidden record of high-frequency climate variations Another difficulty of reconstructing the glacial Nordic Seas arises from the generally low sedimentation rates (e.g., Sarnthein et al., 1995). This, together with biological mixing resulting in smoothing of the geochemical and micropaleontological signals, make it difficult or impossible to identify short term, decadal to centennial, changes in sea-surface conditions since each centimetre of sediment represents a few hundred years at least, and contains a mixed signal for all tracers, with potentially distinct mixing functions in relation to grain size distribution of the signalcarrier (e.g., Bard, 2001). This is a real issue because the few available records with high sedimentation rates demonstrate high frequency decadal to centennial variations during the LGM (e.g., Weinelt et al., 2003). An example is given here from the reconstruction of seasurface conditions based on dinocyst assemblages in the eastern Norwegian Sea core MD95-2010, in which the LGM interval is characterised by sedimentation rates of about 55 cm per thousand years (Eynaud et al., 2004; Fig. 5). The record illustrates that the southeastern Nordic Seas at LGM experienced rapid and recurrent shifts

Core MD95-2010 - Norwegian Sea ( 66.7°N-04.6°W) sea-surface conditions estimated from dinocysts 0

5

10

15°C

30 31 32 33 34

0

2 4

6

8 months/yr.

18 000 Winter

Summer

19 000

20 000

21 000

22 000

23 000

24 000

25 000 Cal. ages BP

Sea-surface temperature

Sea-surface salinity (summer)

Sea-ice cover (>50%)

Fig. 5. Diagram illustrating the high variability of LGM sea-surface conditions in the Norwegian Sea, northeastern North Atlantic (core MD95-2010; 66.71N–4.61E) during the LGM interval based on dinocyst assemblages (Eynaud et al., 2004). The thick line corresponds to the most probable value and the thin lines correspond to the maximum and minimum based on a set of 5 modern analogues in the database that includes 940 reference sites (de Vernal et al., 2005).

ARTICLE IN PRESS 2830

A. de Vernal et al. / Quaternary Science Reviews 25 (2006) 2820–2834

between two modes, from extremely cold conditions, with dense sea-ice cover, to relatively warm (up to 10–12 1C) summer conditions despite freezing winters. The alternating conditions were accompanied by large shifts in salinity, suggesting a complex ocean dynamics at a regional scale with regard to meltwater discharge (Lekens et al., 2005) and mixing with North Atlantic waters. Such variability might have characterised offshore environments of the Greenland and Iceland Seas also, but it cannot possibly be reconstructed because of too low sedimentations rates in these areas, the LGM being represented by a few tens of centimetres at most. 5.3. Working hypotheses The reconstructions based on planktonic foraminifers and dinocysts show large discrepancies, well beyond their respective level of uncertainties. Whilst a definitive resolution of the apparent inconsistencies is not at hand, three working hypotheses or any combination of them can be proposed in order to reconcile the controversial signals: (1) A large seasonal gradient of temperature and strong vertical stratification (as in estuaries and epicontinental seas) could be involved. This implies a sharp pycnocline with large ranges of temperature and salinity from the bottom to the top of the pycnocline, and from summer to winter. It also implies winter sea-ice formation, and high rate of brine formation, in addition to the export of large amounts of freshwater and sea-ice towards the North Atlantic to prevent development of permanent pack ice. Winter brine formation and summer export of freshwater and sea-ice at the surface would involve significant exchanges between the Nordic Seas and the North Atlantic. Because of low salinity, planktonic foraminifers would be restricted to the deeper mesopelagic range of the pycnocline in summer, whereas the more euryhaline dinoflagellates and coccoliths would account for conditions in the shallow and warmer true mixed layer. (2) Large amplitude interannual and interdecadal variability can be involved, with oscillation from episodically ice-free Nordic Seas as described above to quasiperennial ice-pack at the basin scale. This would imply occasional penetration of North Atlantic waters bringing thermophilous micro-organisms. In this hypothesis, the North Atlantic water advected from a southern location (with a front possibly as far south as 40–501N in the Eastern North Atlantic) would have entered into the Nordic Seas with a nutrient-depleted signature. This situation would have resulted in limited biological production in surface water because of quasi-permanent sea-ice and nutrient-poor waters during episodes of North Atlantic water penetration. The low density of microfossils in the central Nordic Seas during the LGM is an argument in favour of this scenario. It is noteworthy that rare ice rafted inputs in the North

Atlantic at LGM are also in agreement with icebergs being locked in the Nordic Seas ice pack throughout most of the interval. Indications of sporadic ice rafted events with climate oscillations during a relatively ‘‘uniform’’ LGM interval are indeed shown by records at about 601N in the northwest North Atlantic (e.g., Hillaire-Marcel et al., 2001b). This paleoceanographical scheme involving episodically ice-free Nordic Seas and significant advection of North Atlantic waters is compatible with the sedimentary and isotope records of the Fram Strait area between 761N and 801N (e.g., Nørgaard-Pedersen et al., 2003). (3) Finally, one could argue that the apparently warm signal recorded by dinocyst assemblages, coccolithophores and alkenones is a taphonomic artefact of a proportionally higher contribution of laterally advected materials during periods of extremely low productivity. In this case, long distance selective transport could have resulted in distorted assemblages with no modern equivalent and the quasi-exclusive dominance of N. pachyderma in planktonic foraminifer assemblages would reflect extremely cold conditions, close to perennial sea-ice. The nearest analogue of the Nordic Seas at the LGM would be the Arctic Ocean. This hypothesis also implies advection of North Atlantic waters to the North, possibly below the surface in an intermediate layer, to account for the transport of ‘‘warm’’ tracers (dinocysts, alkenones) into the glacial Nordic Seas.

6. Discussion: hydrological budget and ocean dynamics In spite of apparent discrepancies, the updated LGM proxy data imply seasonal sea-ice formation over the northern North Atlantic and Nordic Seas, which has to be accompanied by decreasing summer salinity at the surface and winter brine formation in the sub-surface layer. This implies strong stratification between a saline and cold subsurface layer and a low saline surface layer experiencing large amplitude temperature gradients from summer to winter, as it occurs presently in the Okhotsk Sea (e.g., Hays and Morley, 2004), in the Gulf of St. Lawrence, eastern Canada (e.g., Gilbert and Pettigrew, 1997), or in the southern Barents Sea (e.g. Lubinski et al., 2001). In the Nordic Seas, a comparison with a large estuarine system can be put forth. Under steady state conditions, the ablation rates along ice sheet margins are equivalent to the feeding rates. The LGM, which was marked by summer insolation at high latitudes near present values, corresponds to the maximum ice extension. Thus, the rate of meltwater supply along the margins was probably important (e.g., Marshall and Clarke, 1999) and large amounts of freshwater transited through the Nordic Seas and the northern North Atlantic, which were surrounded by Laurentide, Innuitian, Greenland and Fennoscandian

ARTICLE IN PRESS A. de Vernal et al. / Quaternary Science Reviews 25 (2006) 2820–2834

Meland et al., 2005), indicate the existence of a relatively stable mesopelagic layer (i.e., intermediate water mass) distinct from the surface water layer (e.g., Hillaire-Marcel et al., 2001b). The overall data related to surface and sub-surface conditions in the northern North Atlantic and Nordic Seas thus suggest stratification in the upper water column together with northward penetration of North Atlantic waters subducting below a low density surface layer. In such a scenario, subduction of North Atlantic surface water would have had to occur south of the approximate limit of sea-ice (as in the modern Barents Sea), which was located at about 50–551N in the northeast North Atlantic at LGM. Northward advection of subducted water masses at high latitude of the northern North Atlantic is a feature in ocean circulation schemes of the LGM also obtained with the most developed coupled models (e.g., Hewitt et al., 2001; Peltier and Solheim, 2004).

90 80 Nordic seas

Latitude (degree North)

70 Iceland

60 50 40 30 20 East of 30°W ± 1 σ East of 30°W ± 1 σ

10 25

30

35

2831

40

Sea-surface salinity (summer) Fig. 6. Graph showing the LGM sea-surface salinity in summer vs. latitude based on dinocyst estimates. Note that sites located west of 301W record larger range and lower salinities than those located in the eastern parts of the basins (de Vernal et al., 2002).

ice-sheets. Given the Coriolis force and the low salinity and density of the meltwater, the southward evacuation of diluted waters had to occur to the west and at the surface (Fig. 6) (de Vernal et al., 2002). Such a hydrographical pattern with southward flow of freshwater at the surface to the west implies northward penetration of more saline water to the east, with subduction possibly occurring at a relatively low latitude of the northern North Atlantic and less likely at a more northern location in the Nordic Seas, as at present. Indeed, d13C record in foraminifers (e.g., Duplessy et al., 1988) and data on 231Pa and 230Th in North Atlantic sediments between 351N and 501N (McManus et al., 2004; Gherardi et al., 2004) suggest a relatively high rate, but a shallower and more southerly AMO during the LGM. Existing data on LGM deep water formation in the Nordic Seas are sparse or ambiguous. Nevertheless, some data suggest that the production of Denmark Strait Overflow Water (DSOW) was nearly stopped during the LGM (e.g. Hillaire-Marcel et al., 2001b; Fagel et al., 2004), which is consistent with the results of regional ocean circulation model experiments (Ko¨sters et al., 2004). Significant overflow east of Iceland during the LGM was possible, but still remains to be demonstrated, and there is no clear evidence for convection in the Nordic Seas, North of Iceland, during the LGM. This is consistent with the presence of Antarctic Bottom Water (AABW) in the subpolar North Atlantic (e.g., Duplessy et al., 1988). The isotope data (d18O) of N. pachyderma foraminifer shells in the Labrador Seas, the Irminger Basin and in the Nordic Sea basins, which show relatively uniform values (e.g.,

7. Conclusion The northern North Atlantic at LGM was characterised by microfossil assemblages, which indicate sea-surface conditions very different from the present, with no perfect analogue in the modern open oceans. However, by combining information from different proxies, there are features of the LGM sea-surface conditions that can be defined with a relatively high confidence level. The data indicate perennial sea-ice cover along the eastern Canadian and Greenland margins, but seasonally ice-free conditions in the central North Atlantic and Norwegian Sea, and a winter sea-ice limit probably located at about 551N. Proxies also suggest significantly colder than modern seasurface conditions in the southeastern part of the North Atlantic, with a steep temperature gradient between 40 and 451N. In the northern part of the North Atlantic, between 401N and 601N, the different responses provided by the various proxies suggest a strong seasonality with very large annual amplitude of temperatures in surface waters, possibly related to the low thermal inertia of a relatively shallow mixed layer. In the Norwegian Sea, there is evidence for large interdecadal variability (Fig. 5) of seaice cover, sea-surface salinity and temperature. There is also indications of advection of North Atlantic waters in the Nordic Seas, perhaps episodically at the surface, but most probably in a sub-surface layer through subduction, south of the Iceland-Faroe sill, as this occurs today in the Barents Sea or in the Arctic Ocean. The different responses of the various proxies to LGM conditions in the northern North Atlantic suggest complex hydrography with respect to temporal (seasonal cycle and interannual variability) and spatial (thermocline, and halocline depth) distribution of temperature and salinity. The apparent discrepancies recorded by the proxies of temperature most probably provide complementary information on a very sensitive climate in the northern North Atlantic. Coupled model simulations of PMIP show a very

ARTICLE IN PRESS 2832

A. de Vernal et al. / Quaternary Science Reviews 25 (2006) 2820–2834

large dispersal of SST fields for the high latitude of the North Atlantic with much larger interannual variations at LGM than for the pre-industrial runs (cf. Kageyama et al., in press). During the LGM, the presence of continental ice over surrounding lands and shelves and the occurrence of sea-ice in the ocean, notably along the continental margins, certainly played a role in increasing the climate variability in the North Atlantic region. Acknowledgements We are grateful to the MARGO project members and to the PMIP community for very stimulating exchanges. This work has been supported by the Natural Sciences and Engineering Research Council (NSERC) of Canada through the Climate System History and Dynamics (CSHD) network and discovery grants to the Canadian authors. We also thank the anonymous reviewers for their care in reading the manuscript and for their critical but constructive comments. References Antoine, D., Andre´, J.M., Morel, A., 1996. Oceanic primary production. 2. Estimation at global scale from satellite (costal zone colour scanner) chlorophyll. Global Biogeochemical Cycles 10, 57–69. Bard, E., 2001. Paleoceanographic implications of the difference in deepsea sediment mixing between large and fine particles. Paleoceanography 16, 235–239. Barker, S., Cacho, I., Benway, H., Tachikawa, K., 2005. Planktonic foraminiferal Mg/Ca as a proxy for past oceanic temperatures: a methodological overview and data compilation for the Last Glacial Maximum. Quaternary Science Reviews 24, 821–834. Be´, A.W.H., Tolderlund, D.S., 1971. Distribution and ecology of living planktonic foraminifera in surface waters of the Atlantic and Indian oceans. In: Funnel, B.M., Riedel, W.R. (Eds.), The Micropaleontology of Oceans. Cambridge University Press, Cambridge, pp. 105–149. Bendle, J., Rosell-Mele´, A., Ziveri, P., 2005. Variability of unusual distributions of alkenones in the surface waters of the Nordic Seas. Paleoceanography 20, PA2001. Bilodeau, G., de Vernal, A., Hillaire-Marcel, C., Josenhans, H., 1990. Postglacial paleoenvironments of the Hudson Bay: stratigraphic, microfaunal and palynological evidences. Canadian Journal of Earth Sciences 27, 946–963. Bond, G.C., Heinrich, H., Broecker, W.S., Labeyrie, L.D., McManus, J., Andrews, J., Huon, S., Janschick, R., Clasen, S., Simet, C., Tedesko, K., Klas, M., Bonani, G., Ivy, S., 1992. Evidence for massive iceberg discharges in the North Atlantic Ocean during the last glacial period. Nature 360, 245–249. Braconnot, P., 2004. Mode´liser le dernier maximum glaciaire et l’Holoce`ne moyen. Comptes Rendus de Geosciences 336, 711–719. Brassell, S.C., Eglinton, G., et al., 1986. Molecular stratigraphy: a new tool for climatic assessment. Nature 320, 129–133. CLIMAP Project Members, 1981. Seasonal reconstructions of the earth’s surface at the last glacial maximum. Geological Society of America Map and Chart Series MC-56. de Vernal, A., Hillaire-Marcel, C., 2000. Sea-ice cover, sea-surface salinity and halo-thermocline structure of the northwest North Atlantic: modern versus full glacial conditions. Quaternary Science Reviews 19, 65–85. de Vernal, A., Hillaire-Marcel, C., Turon, J.-L., Matthiessen, J., 2000. Reconstruction of sea-surface temperature, salinity, and sea-ice cover

in the northern North Atlantic during the last glacial maximum based on dinocyst assemblages. Canadian Journal of Earth Sciences 37, 725–750. de Vernal, A., Hillaire-Marcel, C., Peltier, W.R., Weaver, A.J., 2002. The structure of the upper water column in the northwest North Atlantic: Modern vs. Last Glacial Maximum conditions. Paleoceanography 17, 1050. de Vernal, A., Eynaud, F., Henry, M., Hillaire-Marcel, C., Londeix, L., Mangin, S., Matthiessen, J., Marret, F., Radi, T., Rochon, A., Solignac, S., Turon, J.-L., 2005. Reconstruction of sea-surface conditions at middle to high latitudes of the Northern Hemisphere during the last glacial maximum (LGM) based on dinoflagellate cyst assemblages. Quaternary Science Reviews 24, 897–924. de Vernal, A., Henry, M., Matthiessen, J., Mudie, P.J., Rochon, A., Boessenkool, K., Eynaud, F., Grøsfjeld, K., Guiot, J., Hamel, D., Harland, R., Head, M.J., Kunz-Pirrung, M., Levac, E., Loucheur, V., Peyron, O., Pospelova, V., Radi, T., Turon, J.-L., Voronina, E., 2001. Dinoflagellate cyst assemblages as tracers of sea-surface conditions in the northern North Atlantic, Arctic and sub-Arctic seas: the new ‘‘n ¼ 677’’ database and application for quantitative paleoceanographical reconstruction. Journal of Quaternary Science 16, 681–699. Dickson, R.R., Brown, J., 1994. The production of North Atlantic Deep Water: sources, rates, and pathways. Journal of Geophysical Research 99 (C6), 12319–12341. Dickson, R.R., Lazier, J., Meincke, J., Rhines, P., Swift, J., 1996. Longterm coordinated changes in the convective activity of the North Atlantic. Progress in Oceanography 38, 241–295. Duplessy, J.-C., Shackleton, N.J., Fairbanks, R.G., Labeyrie, L., Oppo, D., Kallel, N., 1988. Deepwater source variations during the last climatic cycle and their impact on the global deepwater circulation. Paleoceanography 3 (3), 343–360. Duplessy, J.-C., Labeyrie, L., Juillet-Leclerc, A., Maitre, F., Duprat, J., Sarnthein, M., 1991. Surface salinity reconstruction of the North Atlantic Ocean during the Last Glacial Maximum. Oceanologica Acta 14, 311–324. Eynaud, F., Duprat, J., Turon, J.-L., Zaragosi, S., 2004. Paleohydrological evidence of a two step evolution of the Last Glacial Maximum (LGM) along the western European margins. Eigth International Conference on Paleoceanography, Biarritz, Septembre, Programme and abstracts, B1-153. Fagel, N., Hillaire-Marcel, C., Humblet, M., Brasseur, R., Weis, D., Stevenson, R., 2004. Nd and Pb isitope signatures of the clay-size fraction of Labrador Sea sediments during the Holocene: Implications for the inception of the modern deep circulation pattern. Paleoceanography 19, PA3002, doi:10.1029/2003PA000993. Gherardi J.-M., Labeyrie, L., Francois,R., McManus, J. F., Cortijo, E., Nave, S., 2004. Quantifying changes in meridional overturning circulation during the last 20 ky using radionuclides in North Atlantic sediments. Eigth International Conference on Paleoceanography, Biarritz, Septembre, Programme and abstracts, B1-158. Gilbert, D., Pettigrew, B., 1997. Interannual variability (1948–1994) of the CIL core temperature in the Gulf of St. Lawrence. Canadian Journal of Fisheries and Aquatic Sciences 54, 57–67. Guiot, J., 1990. Methods and programs of statistics for paleoclimatology and paleoecology. In: Guiot, J., Labeyrie, L. (Eds.), Quantification des changements climatiques: me´thode et programmes. Institut National des Sciences de l’Univers (INSU-France), Monographie No. 1, Paris. Hays, J.D., Morley, J.J., 2004. The sea of Okhotsk: a window on the ice age ocean. Deep-Sea Research Part I 51, 593–618. Haug, G.H., Ganopolski, A., Sigman, D.M., Rosell-Mele´, A., Swann, G.E.A., Tiedemann, R., Jaccard, S.L., Bollman, J., Maslin, M.A., Leng, M.J., Eglington, G., 2005. North Pacific seasonality and the glaciation of North America 2.7 million years ago. Nature 433, 821–825. Hewitt, C.D., Broccoli, A.J., Mitchell, J.F.B., Stouffer, R.J., 2001. A coupled model study of the last glacial maximum: was part of the North Atlantic relatively warm? Geophysical Research Letters 28, 1571–1574.

ARTICLE IN PRESS A. de Vernal et al. / Quaternary Science Reviews 25 (2006) 2820–2834 Hillaire-Marcel, C., de Vernal, A., Lucotte, M., Mucci, A., Bilodeau, G., Rochon, A., Vallie`res, S., Wu, G., 1994. Productivite´ et flux de carbone dans la mer du labrador au cours des derniers 40.000 ans. Canadian Journal of Earth Sciences 31, 139–158. Hillaire-Marcel, C., de Vernal, A., Bilodeau, G., Weaver, A., 2001a. Absence of deep water formation in the Labrador Sea during the last interglacial. Nature 410, 1073–1077. Hillaire-Marcel, C., de Vernal, A., Candon, L., Bilodeau, G., Stoner, J., 2001b. Changes of potential density gradients in the Northwestern North Atlantic During the Last climatic cycle based on a multiproxy approach. In: Seidov D., et al. (Eds.), The Oceans and Rapid Climate Changes: Past, Present and Future. Geophysical Monograph Series, vol. 126, pp. 83–100. Hillaire-Marcel, C., de Vernal, A., Polyak, L., Darby, D., 2004. Sizedependent isotopic composition of planktic foraminifers from Chukchi Sea vs. NW Atlantic sediments—implications for the Holocene paleoceanography of the western Arctic. Quaternary Science Reviews 23, 245–260. Imbrie, J., Kipp, N.G., 1971. A new micropaleontological method for quantitative paleoclimatology: application to a late Pleistocene Caribbean core. In: Turekian, K.K. (Ed.), The Late Cenozoic Glacial Ages. Yale University Press, New Haven, CT, pp. 71–181. IPCC, Climate Change, 2001. The Scientific Basis. Cambridge University Press, Cambridge. Joussaume, S., Taylor, K.E., 1995. Status of the paleoclimate modeling intercomparison project (PMIP). In: Proceedings of the First International AMIP Scientific Conference. WCRP Report, pp. 425–430. Kageyama, M., Laıˆ ne´, A., Abe-Ouchi, A., Braconnot, P., Cortijo, E., Crucifix, M., de Vernal, A., Guiot, J., Hewitt, C.D., Kitoh, A., Marti, O., Ohgaito, R., Otto-Bliesner, B., Peltier, W.R., Rosell-Mele´, A., Vettoretti, G., Weber, S.L., and MARGO Project Members, 2006. Last Glacial Maximum temperature over the North Atlantic, Europe and western Siberia: a comparison between PMIP models, MARGO sea-surface temperatures and pollen-based reconstructions. Quaternary Science Reviews 25, 2082–2102. Koc- , N., Jansen, E., Haflidason, H., 1993. Paleoceanographic reconstruction of surface ocean conditions in the Greenland, Iceland, and Norwegian seas through the last 14,000 years based on diatoms. Quaternary Science Reviews 12, 115–140. Kornilova, O., 2004. Investigation of the origin of biomarker signals associated to ice rafted debirs in North Atlantic sediments, PhD dissertation, Durham University. Kohfeld, K.E., Fairbanks, R.G., Smith, S.L., Walsh, I.D., 1996. Neogloboquadrina pachyderma (sinistral coiling) as paleoceanographic tracers in polar oceans: evidence from Northeast Water Polynya plankton tows, sediment traps, and surface sediment. Paleoceanography 11, 679–699. Ko¨sters, F., Ka¨se, R., Fleming, K., Wolf, D., 2004. Denmark Strait overflow for Last Glacial Maximum to Holocene conditions. Paleoceanography 19, PA2019. Kucera, M., Weinelt, M., Kiefer, T., Pflaumann, U., Hayes, A., Weinelt, M., Chen, M.-T., Mix, A.C., Barrows, T.T., Cortijo, E., Duprat, J., Juggins, S., Waelbroeck, C., 2005a. Reconstruction of the glacial Atlantic and Pacific sea-surface temperatures from assemblages of planktonic foraminifera: multi-technique approach based on geographically constrained calibration datasets. Quaternary Science Reviews 24, 951–998. Kucera, M., Rosell-Mele´, A., Schneider, R., Waelbroeck, C., Weinelt, M., 2005b. Multiproxy approach for the reconstruction of the glacial ocean surface (MARGO). Quaternary Science Reviews 24, 813–819. Kutzbach, J.E., Wright Jr., H.E., 1986. Simultaion of the climate of 18000 years BP: results for the North American/North Atlantic/European sector and comparison with the geologic record of North America. Quaternary Science Reviews 4, 147–187. Lapointe, M., 2000. Late Quaternary paleohydrology of the Gulf of St. Lawrence (Que´bec, Canada) based on diatom analysis. Palaeogeography, Palaeoclimatology, Palaeoecology 156, 261–276.

2833

Lea, D.W., Mashiotta, T.A., Spero, H.J., 1999. Controls on magnesium and strontium uptake in planktonic foraminifera determined by live culturing. Geochimica Cosmochimica Acta 63, 2369–2379. Lee, K.E., 2004. Compilation of alkenone LGM SST data. PANGAEA, doi:10.1594/PANGAEA.103070. Lekens, W.A.H., Sejrup, H.P., Haflidason, H., Petersen, G.Ø., Hjelstuen, B., Knorr, G., 2005. Laminated sediments preceding Heinrich event 1 in the northern North Sea and southern Norwegian Sea: origin; processes and regional linkage. Marine Geology 216, 27–50. Le´vesque, L., 1995. Distribution des assemblages de coccolithes dans les se´diments re´cents des moyennes latitudes de l’Atlantique Nord: de´veloppement de fonctions de transfert pale´oce´anographiques. M.Sc. Dissertation, Universite´ du Que´bec a` Montre´al. Lubinski, D.J., Polyak, L., Forman, S.L., 2001. Freshwater and Atlantic water inflows to the deep northern Barents and Kara seas since ca 13 14C ka: foraminifera and stable isotopes. Quaternary Science Reviews 20, 1851–1879. Malmgren, B., Kucera, M., Nyberg, J., Waelbroeck, C., 2001. Comparison of statistical and artificial neural network techniques for estimating past sea surface temperatures from planktonic foraminifer census data. Paleoceanography 16, 520–530. Manabe, S., Broccoli, A.J., 1985. The influence of continental ice-sheets on the climate of an ice age. Journal of Geophysical Research 90, 2167–2190. Marshall, S.J., Clarke, G.K.C., 1999. Modeling North American freshwater runoff through the last glacial cycle. Quaternary Research 52, 300–315. Matthiessen, J., de Vernal, A., Head, M., Okolodkov, Y., Zonneveld, K., Harland, R., 2005. Modern organic-walled dinoflagellate cysts in Arctic marine environments and their (paleo-) environmental significance. Pala¨ontologische Zeitschrift 79/1, 3–51. McManus, J.F., Franc- ois, R., Gherardi, J.-M., Keigwin, L.D., BrownLeger, S., 2004. Collapse and rapid resumption of Atlantic meridional circulation linked to deglacial climate changes. Nature 428, 834–837. Meland, M.Y., Jansen, E., Elderfield, H., 2005. Constraints on SST estimates for the northern North Atlantic/Nordic Seas during the LGM. Quaternary Science Reviews 24, 835–852. Mix, A.E., Bard, E., Schneider, R., 2001. Environmental processes of the ice age: land, ocean, glaciers (EPILOG). Quaternary Science Reviews 20, 627–657. Mu¨ller, P.J., Kirst, G., Ruhland, G., Von Storch, I., Rosell-Mele´, A., 1998. Calibration of the alkenone paleotemperature index UK’ 37 based core-tops from the eastern South Atlantic and the global ocean (601N–601S). Geochimica et Cosmochimica Acta 62, 1757–1772. Nørgaard-Pedersen, N., Spielhagen, R.F., Erlenkeuser, H., Grootes, P.M., Heinemeier, J., Knies, J., 2003. Arctic Ocean during the Last Glacial Maximum: Atlantic and polar domains of surface water mass distribution and ice cover. Paleoceanography 18 (3), 1063. Nu¨rnberg, D., Mu¨ller, A., Schneider, R.R., 2000. Paleo-sea surface temperature calculations in the equatorial east Atlantic from Mg/Ca ratios in planktic foraminifera: a comparison to sea surface 0 temperature estimates from UK 37 , oxygen isotopes, and foraminiferal transfer function. Paleoceanography 15, 124–134. Peltier, W.R., 1994. Ice age paleotopography. Science 265, 195–201. Peltier, W.R., 2004. Global glacial isostasy and the surface of the ice-age Earth: the ICE-5G (VM2) model and GRACE. Annual Review of Earth and Planetary Sciences 32, 111–149. Peltier, W.R., Solheim, L.P., 2004. The climate of the Earth at Last Glacial Maximum: statistical equilibrium state and a mode of internal variability. Quaternary Science Reviews 23, 335–357. Pflaumann, U., Duprat, J., Pujol, C., Labeyrie, L., 1996. SIMMAX: a modern analog technique to deduce Atlantic sea surface temperatures from planktonic foraminifera in deep-sea sediments. Paleoceanography 11, 15–35. Pflaumann, U., Sarnthein, M., Chapman, M., d’Abreu, L., Funnell, B., Huels, M., Kiefer, T., Maslin, M., Schulz, H., Swallow, J., van Kreveld, S., Vautravers, M., Vogelsang, E., Weinelt, M., 2003. Glacial North Atlantic: sea-surface conditions reconstructed by GLAMAP 2000. Paleoceanography 18, 1065.10.

ARTICLE IN PRESS 2834

A. de Vernal et al. / Quaternary Science Reviews 25 (2006) 2820–2834

Pinot, S., Ramstein, G., Harrison, S.P., Prentice, I.C., Guiot, J., Stute, M., Joussaume, S., 1999a. Tropical paleoclimates at the last glacial maximum: comparison of Paleoclimate Modeling Intercomparison Project (PMIP) simulations and paleodata. Climate Dynamics 15, 857–874. Pinot, S., Ramstein, G., Marsiat, I., De Vernal, A., Peyron, O., Duplessy, J.-C., Weinelt, M., 1999b. Sensitivity of the European LGM climate to North Atlantic sea-surface temperature. Geophysical Research Letters 26, 1893–1896. Prahl, F.G., Wakeham, S.G., 1987. Calibration of insaturation patterns inlong-chain ketone for paleotemperature assessment. Nature 330, 367–369. Rasmussen, T.L., Thomson, E., Troelstra, S.R., Kuijpers, A., Prins, M.A., 2003. Millennial-scale glacial variability versus Holocene stability: changes in planktic and benthic foraminiferal faunas and ocean circulation in the North Atlantic during the last 60,000 years. Marine Micropaleontology 47, 143–176. Rind, D., 1987. Components of the ice age circulation. Journal of Geophysical Research 92, 4241–4281. Rind, D., Peteet, D., 1985. Terrestrial conditions at the Last Glacial Maximum and CLIMAP sea-surface temperature estimates: are they consistent? Quaternary Research 24, 1–22. Rodrigues, C.G., Ceman, J.A., Vilks, G., 1993. Late-Quaternary paleoceanography of deep and intermediate water masses off Gaspe´ Peninsula, Gulf of St. Lawrence: Foraminiferal evidence. Canadian Journal of Earth Sciences 30, 1390–1401. Rosell-Mele´, A., 1997. Appraisal of CLIMAP temperature reconstruction in the NE Atlantic using alkenone proxies. Eos 78 (46), F28. Rosell-Mele´, A., 1998. Interhemispheric appraisal of the value of alkenone indices as temperature and salinity proxies in high latitude locations. Paleoceanography 13, 694–703. Rosell-Mele´, A., Comes, P., 1999. Evidence for a warm last glacial0 maximum in the Nordic Seas, or an example of shortcomings in UK 37 and UK 37 to estimate low sea surface temperature? Paleoceanography 14, 770–776. Rosell-Mele´, A., Bard, E., Emeis, K.C., Grieger, B., Hewitt, C., Mu¨ller, P.J., Schneider, R., 2004. Sea surface temperature anomalies in the 0 oceans at the LGM estimated from the alkenone-UK 37 index: comparison with GCMs. Geophysical Research Letters 31, L03208. Ruddiman, W.-F., 1977. Late Quaternary deposition of ice-rafted sand in the subpolar north Atlantic (lat 401 to 651N). Geological Society of America Bulletin 88, 1813–1827. Sarnthein, M., Jansen, E., Weinelt, M., Arnold, M., Duplessy, J.C., Erlenkeuser, H., Flatøy, A., Johannessen, G., Johannessen, T., Jung, S., Koc, N., Labeyrie, L., Maslin, M., Pflaumann, U., Schulz, H., 1995. Variations in Atlantic surface ocean paleoceanography, 50–801N: a time-slice record of the last 30,000 years. Paleoceanography 10, 1063–1094.

Sarnthein, M., Gersonde, R., Niebler, S., Pflaumann, U., Spielhagen, R., Thiede, J., Wefer, G., Weinelt, M., 2003a. Overview of Glacial Atlantic Ocean Mapping (GLAMAP 2000). Paleoceanography 18, PA000774. Sarnthein, M., Pfaumann, U., Vogelsang, E., Weinelt, M., 2003b. Past extent of sea-ice in the northern North Atlantic inferred from foraminiferal paleotemperature estimates. Paleoceanography 18. Schmittner, A., 2005. Decline of the marine ecosystem caused by a reduction in the Atlantic overturning circulation. Nature 434, 628–633. Schneider, R., Bard, E., Mix, A., 2000. Last Ice Age global ocean and land surface temperature: the EPILOG initiative. PAGES Newsletter 8, 19–21. Simstich, J., Sarnthein, M., Erlenkeuser, H., 2003. Paired d18O signals of Neogloboquadrina pachyderma (s) and Turborotalita quinqueloba show thermal stratification structure in Nordic Seas. Marine Micropaleontology 48, 107–125. Skinner, L.C., Elderfield, H., 2005. Constraining ecological and biological bias in planktonic foraminiferal Mg/Ca and d18Occ: a multispecies approach to proxy calibration testing. Paleoceanography 20, PA1015. Thomas, E., Booth, L., Maslin, M., Shackleton, N.J., 1995. Northeastern Atlantic benthic foraminifera during the last 45,000 years: changes in productivity seen from the bottom up. Paleoceanography 10 (3), 545–562. Weaver, A.J., Eby, M., Fanning, A.F., Wiebe, E.C., 1998. Simulated influence of carbon dioxide, orbital forcing and ice sheets on the climate of the last glacial maximum. Nature 394, 847–853. Waelbroeck, C., 2004. Compilation of d18O LGM data, PANGAEA, doi:10.1594/PANGAEA.140254. Waelbroeck, C., Labeyrie, L., Duplessy, J.-C., Guiot, J., Labracherie, M., Leclaire, H., Duprat, J., 1998. Improving paleo-SST estimates based on planktonic fossil faunas. Paleoceanography 12, 272–283. Waelbroeck, C., Mulitza, S., Spero, H., Dokken, T., Kiefer, T., Cortijo, E., 2005. A global compilation of Late Holocene planktonic foraminiferal d18O: Relationship between surface water temperature and d18O. Quaternary Science Reviews 24, 853–878. Weaver, P.P.E., Chapman, M.R., Eglington, G., Zhao, M., Rutledge, D., Read, G., 1999. Combined coccolith, foraminiferal, and biomarker reconstruction of paleoceanographic conditions over the past 120 kyr in the northern North Atlantic (591N, 231W). Paleoceanograpy 14 (3), 336–349. Weinelt, M., 2004. Compilation of global planktic foraminifera LGM SST data, PANGAEA, doi:10.1594/PANGAEA.227620 Weinelt, M., Sarnthein, M., Pflaumann, U., Schulz, H., Jung, S., Erlenkeuser, H., 1996. Ice-free Nordic seas during the last glacial maximum? Potential sites of deepwater formation. Paleoclimates 1, 283–309. Weinelt, M., Vogelsang, E., Kucera, M., Pflaumann, U., Sarnthein, M., Voelker, A., Erlenkeuser, H., Malmgren, B.A., 2003. Variability of North Atlantic heat transfer during MIS 2. Paleoceanography 18, 1071–1080.