Evidence for a Warm Last Glacial Maximum in the Nordic Seas or an example of shortcomings in UK37' and UK37 to estimate low sea surface temperature?

PALEOCEANOGRAPHY, VOL. 14, NO. 6, PAGES 770-776, DECEMBER 1999 Evidence for a warm Last Glacial Maximum in the Nordic seas or an example of shortcomings in U37 ' and U37 to estimate low sea surface temperature? Antoni Rosell-Mel6 • and Pau Comes Department of Fossil Fuels and Environmental Geochemistry, University of Newcastle, Newcastle upon Tyne England, United Kingdom Abstract. The occurrence of ice-free conditions in the Nordic seas during the Last Glacial Maximum has been demonstrated using microfossil and biomarker approaches. An accurate picture of the prevailing oceanographic conditions may be more difficult to ascertain. Thus, uncertainties for most proxies increase at the cold extreme of temperature calibrations and glacial sediments contain small amounts of microfossils and biomarkers. Here we discuss sea surface temperature estimates derived from alkenone indices for the Last Glacial Maximum in the Nordic seas that provide a surprising scenario, where surface temperatures were warmer than at present. We conclude that the U•:37 ' and U•:37 estimates may not be interpreted in terms of sea surface temperature as several factors may have conspired to bias the molecular signal. However, using an alternative index based on the abundance of the C37:4 alkenone, we estimate a maximum temperature of 6øC. 1. Introduction It has been assumed for some time that during the Last Glacial Maximum (LGM) a perennial sea ice cover existed on the whole of the northern North Atlantic. This scenario was convincingly described by Climate.' Long-Range Investigation, Mapping, and Prediction (CLIMAP) [1976, 1981 ], and since then climate modelers have been trying to reproduce it in their simulations [e.g., Ganapolski et al., 1998]. However, increasing evidence is being collected, from the analysis of microfossil and chemical proxies in sediment cores, that during the LGM at least seasonally open sea conditions occurred in the Nordic seas (i.e., Greenland, Iceland, and Norwegian Seas) [Hebbeln et al., 1994; Sarnthein et al., 1995; Weinelt et al., 1996; Rosell-Meld and Ko•, 1997, Rosell-Meld et al., 1998; de Vernal et al., 1999]. Using transfer functions of foraminifera, it has been argued that summer sea surface temperatures (SSTs) during the LGM in the Nordic seas were between 3 ø and 5øC, with a prevailing cyclonic surface circulation [Sarnthein et al., 1995; Weinelt et al., 1996]. However, both these results and those from CLIM,4P [1976,1981] were obtained using the same basic approach (tbraminitbral assemblages transfer functions), albeit with significant variations in the statistical analysis and the number of sediment cores employed, which have led to opposite conclusions. The variance and occurrence of microfossils can also be influenced by dissolution of their skeletons in the water column and within the sediment [Ko• et al., 1993; Samtleben et al., 1995]. In addition, the diversity of planktonic foraminifera in •Now at the Department of Geography, University of Durham, Durham, England, United Kingdom Copyright 1999 by the American Geophysical Union. Paper number 1999PA900037. 0883-8305/99/1999PA900037512.00 subarctic glacial sediments is very low, mainly composed of one species (>97% of Neogloboquadrina pachyderma) [Sarnthein et al., 1995; Weinelt et al., 1996], which restricts the sensitivity lbr reconstructing different water masses and for obtaining representative SST values [e.g., Bond et al., 1997], as the errors of the reconstruction in the Nordic seas can be as large as 3øC [Weinelt et al., 1996]. Hence it is important to have alternative methods to contrast with the evidence derived from microfossil analyses. In this paper we will discuss the use of alkenone proxies (U•:_•7 and U•:_•7 ') [Brassell et al., 1986; Prahl and Wakeham, 1987] to derive SST estimates in sediment cores from the Nordic seas during the LGM. 2. Methods 2.1. Alkenones The procedures employed to determine alkenone indices have been described in detail elsewhere [Rosell-Meli, 1994; Rose#- Meli et al.• 1995b]. Briefly, after retrieval, core samples were generally kept cool, in the dark, before being subsampled at 1-5 cm intervals. Subsamples were freeze dried, weighed (1-2 g), and solvent extracted (dichloromethane/methanol, 3:1 x 3) using a robotic workstation. After solid-phase extraction of the total extracts on silica and subsequent derivatization with bis- (trimethyl-silyl)-trifluoro-acetamide (BSTFA) the C_•? alkenones (i.e., ketones with 37 carbon atoms) were quantified by gas chromatography-mass spectrometry (GC-MS) with ammonia chemical ionization (CI). The U•:•7 and U•:37 ' indices have been calculated as [Brassell et al., 1986; Prahl et al., 1987] U3 % = 37:2-37:4 K = 37:2 37:2+37:3+37:4 U37' 37:2+37:3 ' where UK_•7 stands •br unsaturated ketones with 37 carbon atoms and 2, 3, or 4 double bonds. 770 ROSELL-MELI• AND COMES' A NORDIC SEAS WARM LAST GLACIAL MAXIMUM? 771 2..2WAnalytical Appraisal of the Data In view of the very low concentration of alkenones in the samples (<30 ng/g dry weight of sediment) we were concerned that some of the alkenone data might have suffered from a systematic shift toward warmer values. A lower threshold of 100 ng alkenones/g sediment (-10 ng per GC peak area) has been recommended for reliable palaeotemperaure determination (error <0.5øC) [l/illanueva and Grimalt, 1996]. As this threshold may vary between laboratories and methods, it has not been considered here. For instance, for the chemical ionisation mass spectrometry (CIMS) system this limit could be even lower than I ng per peak area [Rosell-Meli et al., 1995b]. Instead, mixtures of the synthetic 37:3 and 37:2 alkenones were prepared at different concentrations and analyzed at the same time as the samples to keep track of the occurrence of any preferential adsorption of the alkenones in the analytical system. In all cores analyzed a bias was observed (e.g., see Table 1), and samples likely to be affected by more than AUK37'=0.03 (- løC) were reanalyzed or reinjected in the GC. For our sample set, reliable alkenone data could only be obtained with the CIMS technique, unless a larger amount of sediment (>5g) was used, which was not usually possible because of limited sample availability. The CIMS method also allowed quantification of the 37:4 alkenone with a similar precision as the other members of the uK37 index because each peak was integrated from its mass chromatogram and a comparable snr was obtained [Rosell-Meli et al., 1995b]. In this case the relative errors of both uK37 and uK•7 ' are comparable. It may be the case that the 37:4 alkenone may increase the relative error of uK37 as compared to uK3? ' [Sikes et al., 1997], but the difference is not likely to be climatically significant. 2.3. Age Model and Selection of Time Slices Details on accelerator mass spectrometry (AMS) 14C measurements and identification of ash layers and tie points in the •80 curve to obtain the depth/age model for each core (Table 2) have been described elsewhere [Sarnthein et al., 1995; Weinelt et al., 1996]. The time slices have been selected following the approach and rationale of Sarnthein et al. [1994, 1995], and biomarker data have been averaged within the time interval corresponding to the LGM (18.5-21.2 cal kyr). Table 1. Results From the Analysis of Alkenone Standards Used As Controls During the Analysis of Core Samples Concentration, uK37 ' T, ppm øC expected value 0.38 10 0.32 0.41 11 0.64 0.42 11 1.27 0.41 11 1.27 0.38 10 Temperatures are calculated using the equation of Prahl et al. [1988]. Concentration (ppm) is equivalent to ng/mL injected in the gas chromatograph of total triunsaturated and diunsaturated alkenones. Table 2. Coordinates and Water Depth of Locations Shown in Figures 2 and 3 Core Latitude Longitude Depth, m M 17725 77ø28'N 4ø35'E 2580 M17728 76o31N 3ø57'E 2485 M 17724 76øN 8ø20'E 2354 HM94-34 73ø46'N 2ø23'W 3004 M23294 72022 'N 10ø36'W 2224 M23260 72ø08'N 11 ø27'E 2084 M23269 71ø27'N 0ø40'E 2872 M17732 71ø37'N 4ø13E 3103 M1171-1 67ø59'N 18032 ' W 850 M23068 67ø50'N 1 ø30'E 2230 M23074 66ø40'N 4ø55'E 1157 3. Results We have measured both U•:37 and •: ' U 37 in the glacial samples, although Ibr the area considered, uK37 ' shows a more equivocal correlation with SST than U•:37 [Rosell-Meli et al., 1995a; Rosell-Mel& 1998]. U•:37 in core tops t?om the Nordic seas show a general agreement with overlying water temperatures, but some important discrepancies exist (Figure 1). These may be partly due to the type of water masses in the area and their influence on the biosynthesis of alkenones [Rosell-Meli et al., 1998], lateral advection of resuspended sediments [e.g., Andruleit, 1997], and the uncertainty associated with assigning temperature and salinity values from a database to a particular location. Using sediment traps in the Nordic seas, Fl•igge [1997] has shown that the temperature of the mixed layer during the maximum production season (summer) is preserved in the U•:37 of sinking particles and transtarred to the surface sediments. The presence of alkenones in the Nordic seas during the LGM has been previously demonstrated and could be taken as an indication of ice-free marine conditions [Rosell-Meli and Ko½, 1997]. In Figure 2 both UI':3? and U•:37 ' SST estimates in the LGM time slice are mapped. uK37 temperatures are always colder than U•:37'estimates because the abundance of 37:4 is between 4 and 10% of the total abundance of C37 alkenones (Figure 3). Thus 37:4 abundance is large enough to increase significantly the ditt•rence between estimates of both indices by-2 øC, even for UK37 estimates above 10øC. This relationship between K ' U 37 and U•7 ' is unlike the one found in modern surface sediments, and the abundance of the 37:4 alkenone appears to be higher in the LGM samples compared to modern surface sediments in the Nordic seas for a given value of U•37 ' [Rosell-Meli et al., 1994; Rose/l-Me/& 1998]. The large difference in the estimates from U•37 and U•37 ' is also related to some extent to the calibration equations employed, which are different for each proxy. U•:37 values have been converted to SST using the equation derived from North Atlantic sediment core tops [Rosell-Meli et al., 1995a] corresponding to annual values at 0 m depth, whereas for U•3• ' • on have used the Prahl et al. [1988] equation based algal cultures of Emiliania huxleyi. Milllet et al. [1998] have shown that this equation, however, can also be obtained using core tops of the global ocean. For the range of values in the samples the equation of Rosell-Meli et al. [1995a] provides colder estimates, by •2øC, than if the Prahl et al. [1988] 772 ROSELL-MELI• AND COMES: A NORDIC SEAS WARM LAST GLACIAL MAXIMUM? : • •':8 .......... ......... ß -:.• ; • %•: : -- -•4:•4• • .:' .,'•. • ,..:• ....... 0" Figure 1. Map of the Nordic seas with modern surface current patterns, with black arrows corresponding to polar waters, shaded arrows corresponding to Atlantic waters, and d•hed arrows corresponding to •ctic waters. Black numerals are data on U[37 from core tops transformed into temperature values using •e calibration equation of Rosell-Mel• et al. [1995a] corresponding to annual values at 0 m depth. The sketchy isolines of temperature are drawn as an approximation of the August monthly data from the Reynolds and Smith [ 1995] climatolo•. 232.94 7.3/12.5 • 23260 / 8,3115.0 •.• ß "' 23'269 17732 • ½ :..•: ................. ', ................. • ............................... • • '• .:.-.•.:.-•,,•:::.: ...... / 1171 / • • • ' .....• / / • ' 8,•,113.4 • 23068 • X "•'•"•'••••••• '•0' 0' K Figure 2. Estimates of sea surface temperature for the Last Glacial Maximum derived .from U •7 (shaded) using the calibration ß K equation of Rosell-Meld et al. [1995a] corresponding to annual values at 0 m depth, and derived from U 37' (black) using the calibration equation of Prahl et al. [1988] from algal cultures. The corresponding core number is indicated next to each location (coordinates and water depth in Table 2). ROSELL-MELI• AND COMES: A NORDIC SEAS WARM LAST GLACIAL MAXIMUM? 773 .:•-;,... ' •:.•. ,: ..,.•.•. - .... ; lo' 6 • .................. 3% ,' sl , .' O' Figure 3. Data for each sediment core location (Table 2) corresponding to values of the relative abundance of the tetraunsaturated C37 alkenone (37:4) to the total of C37 alkenones (37:4%) for the Last Glacial Maximum time slice. The back line depicts an approximate isotherm of maximum sea surface temperature. equation had been used for uK37 . In contrast, the equation corresponding to summer temperatures at 0 m depth from the North Atlantic core-top compilation yields estimates (not shown) that are very close to those from the Prahl et al. [1988] equation. In any case the estimates from U}•37 ' obtained using any of the above equations, are still several degrees lower than the estimates from uK37 ' (Figure 2). The temperature range of the uK37 ' estimates in Figure 2 is 12.0ø-16.8øC, which is much higher than the estimates of-3 ø5øC for glacial summers derived from foraminiferal transfer functions [Sarnthein et al., 1995; Weinelt et al., 1996], and in fact they are even higher than modem SST values (see Figure 1) [Reynolds and Smith, 1995]. SSTs derived from uK37 are in the range of 4.9 ø13.1 øC, lower than those from uK37 ', although still generally warmer than modern SSTs, especially for the northernmost samples (Figure 1), and much warmer than those indicated by the data from Sarnthein et al. [ 1995] and Weinelt et al. [1996]. The alkenone data seem to be anomalously high, providing warmer glacial than interglacial SSTs. The SST gradient is also difficult to understand (Figure 2), with increasing temperatures toward the northeast, i.e., the Barents Shelf/Spitsbergen, rather than toward the southeast. The distribution of uK37/uK37 ' SST estimates (Figure 2), if considered representative of the circulation during the LGM in the Nordic seas, would thus reflect a situation very different from that in the modem oceanographic regime (see Figure 1) and certainly difficult to comprehend. In view of these results it is easy to conclude that certainly uK37 ' and probably uK37 are not suitable to estimate SST in the LGM in the area studied. However, SST estimates from dinoflagellate transfer functions are similar in magnitude to those of uK37/uK37' [de Vernal et al, 1999]. The similarity between the absolute values of SST (•-15øC for the eastern Nordic seas) from these two independent proxies is remarkable, but it could be a coincidence. Conversely, it does not seem justified to give more credibility to the estimates from foraminifi:ra transfer function approaches [Sarnthein et al., 1995; Weinelt et al., 1996] solely on the basis that as they are colder, the SSTs make more sense. 4. Discussion Reasons for possible anomalous "warm" uK37 and uK37 ' data can be diverse. 4.1. Analytical Bias Alkenone data'may be biased toward high values because components are in too low abundance (less than 30 ng/g dry sediment) to be measured accurately. Hence coelution of the alkenones with minor components during GC analysis and/or preferential adsorption of the more unsaturated components in the analytical system may have occurred [Rosell-Meld et al., 1995b; Villanueva and Grimalt, 1996]. However, as described in section 2, sample data have been contrasted with standards used for reference, and analytical errors are not larger than IøC. Moreover, the mass spectrometric method employed to quantify the alkenones precluded coelution problems [Rosell-Meld et al., 1995b]. Hence the large magnitude of uK37 and uK37 ' in Figure 1 is dependent on environmental rather than analytical factors. 4.2. Failure of the Modern Analog In cold and relatively low salinity environments, when the relative abundance of the tetraunsaturated component is enhanced, the temperature dependences of U}•37 and U}•37' is equivocal [Conte et al., 1994; Ficken and Farrimond, 1995; 774 ROSELL-MELI• AND COMES' A NORDIC SEAS WARM LAST GLACIAL MAXIMUM? Rosell-Meld, 1998]. In fact, the Nordic seas already appear as an area where the temperature dependence of the alkenones is not as evident as in other oceanic locations, for instance, in comparison to the global compilation of Miiller et al. [1998]. This situation may have been aggravated during the LGM because the type of surface water mass throughout the Nordic seas was equivalent to that of the present Arctic domain rather than the Atlantic domain in the east [Weinelt et al., 1996]. In this situation the relative abundance of 37:4 to the abundance of C37 alkenones (37:4%) is larger than 5% [Rosell-Meld, 1998], and uK37 ' does not correlate with SST. In the Nordic seas data set for the LGM the percentages of 37:4 are larger than or equal to 5% for all locations except one, which is 4% (Figure 3). Hence the relatively high abundance of 37:4 in the samples could indicate that uK37 ' and uK37 are not reliable as SST proxies. Nevertheless, the SST estimates derived from uK37 ' are much higher than those found currently in the Nordic seas for similar values of 37:4%, which suggests that perhaps additional factors might also be responsible for the high uK37 ' values in Figure 2. 4.3. Diagenesis It is sometimes argued that the alkenone ratio may be biased toward warmer values during deposition because of preferential degradation or incorporation into macromolecular organic matter of the components with more double bonds (37:4>37:3>37:2) [Freeman and Wakeham, 1992; Flagge, 1997; HoeJ3 et al., 1998]. Such diagenetic bias could thus be more important when estimating cold SST. However, the importance of the diagenetic bias may have sometimes been overemphasized and mistaken for analytical biases [Grimalt, d., C. Pelejero, E. Calvo, E. Bard, ,4. Rosell-Meld, P. Maller, and R. Schneider, Changes in uK37 ' sea surface palcotemperature due to measurements close to C37 alkenone limit of detection, submitted to Paleoceanography, 1999], and in fact, some authors have not found much cause for concern regarding diagenesis [Prahl et al., 1989; Teece et al., 1998]. Fl•igge [1997] has estimated from two incubation experiments of E. huxleyi and sediments that 37:3 degrades 2-3% faster than 37:2 and 37:4 degrades 5-20% faster than 37:2, which may explain an observed increase of uK37 with depth within the first 3-5 cm below the sediment surface in the Norwegian Sea [Fl•igge, 1997]. During these experiments, however, uK37 ' varied by •uK37 ': 0.04 and 0.07 from the start of the experiment until the point when no more alkenones were being degraded. These variations would represent a potentially significant bias in the sedimentary alkenone data of-1ø-2øC due to a relative pre/•rential removal of-25% more 37:3 than what it is left of 37:2. uK37 ' values in the glacial sediments appear more than just 2øC too warm, and in comparison to foraminiferal-based estimates or modern SST values the difference is of several degrees Celsius. To account for a diagenetic bias of 6øC or more, lbr instance, at least-56% more 37:3 than what it is left of 37:2 should be preferentially removed from the sediment, and so far, this has not been observed in any experiment or study on sediments from &positional conditions analogous to those occurring in the Nordic seas. Hence this suggests that by itself, diagenesis is not enough to explain the warm uK37 ' and uK37 signal. 4.4. Advected Signal During the LGM the Nordic seas probably experienced an extreme seasonality, with winter sea ice giving way to a brief summer melting, as southern warm waters penetrated the area [Hebbeln et al., 1994; Sarnthein et al., 1995; Weinelt et al., 1996; Rosell-Meli and Kof, 1997, Rosell-Meli et al., 1998]. Thus the possibility exists that alkenones in the Nordic seas were deposited after being advected by oceanic currents similar to the present North Atlantic drift/Norwegian Current, having been biosynthesized in warmer southern waters. However, warmer SST-UK37 ' values occur in the northeastern rather than the southeastern area of the basin, which would imply that if there was an advected input, this came principally t?om the Arctic, not the Atlantic, region. Furthermore, the major type of planktonic tbraminit•ra co-occurring in sediments with the alkenones (>97% of N. pachyderma) predominantly live in waters colder than 8øC [Wu and Hillaire-Marcel, 1994]. uK37'-SST are much warmer than this (Figure 2), so the advection mechanism discussed here cannot account by itself for the anomalously warm signal. Moreover, the uniform values of SST estimated t•om foraminiferal transfer functions, the abundance values of fbraminit•ra [Sarnthein et al., 1995; Weinelt et al., 1996], and the 37:4% over the northeast Atlantic and the Nordic seas [Rosell-Meli et al., 1998] during the LGM indicate that there was a weak latitudinal heat advection, and thus weak surthce currents may be inIbrred. 4.5. Input of Reworked Alkenones In the NE Atlantic, higher uK37'-SST estimates than those from fbraminiI•ral transfer functions, between 25 and 13 ka, cooccur with enhanced inputs of reworked Cretaceous and Paleogene nannotbssils and peak delivery of ice-rafted debris [Weaver et al., 1999]. This observation led Weaver et al. to hypothesize that ice-rafted allochthonous alkenones overprinted the local surface water alkenone signal, biasing U•:37 '. The LGM is a period characterized by the deposition of high amounts of ice-rafted debris in the sediments of the North Atlantic [Ruddiman, 1977]. The debris can also contain a significant amount of "terrigenous" (i.e., allochthonous) organic material, especially in the Nordic seas, where allochthonous glaciomarine sediment (coal and black shale) is related to high total organic contents in the sediments coming from the eastern Norwegian Sea. In there Jurassic-Cretaceous organic-rich sedimentary rocks crop out along the continental shelves of the Barents Sea, the Arctic borderlands, and the Norwegian Shelf [Wagner, 1993; Wagner and Henrich, 1994], which can leave a distinct mature biomarker signal in the sediments [Rosell-Meld and Maxwell, 1996; Rosell-Meli and Ko½, 1997]. The possibility exists therefore that pre-Quaternary C37 alkenones present in organicrich shales underlying ice caps were transported by icebergs and added to the pool of organic components present in LGM sediments, as also argued by Weaver et al. [1999]. Alkenones (as t?ee, solvent soluble lipids) have been reported even in Eocene and Cretaceous sediments [see Brassell, 1993] and occur continuously in marine sediments throughout the Quaternary and at least since the early Miocene [Boon et al., 1978; Haug, 1996]. The presence of Tertiary and Cretaceous coccoliths in glacial sediments from the North Atlantic seems to endorse this hypothesis [Rahman, 1995; Weaver et al., 1999]. Allochthonous, ice-rafted alkenones would most probably contain a warm signal associated with pre-Quaternary climates and, as found in the few organic-rich shales reported to contain alkenones [e.g., Farrimond et al., 1986], be dominated by diunsaturated ROSELL-MELI• AND COMES: A NORDIC SEAS WARM LAST GLACIAL MAXIMUM? 775 components. This could explain the warm U•:37 ' distribution in Figure 2, the product of the combination of autochthonous (with probably a cold signal, SST <10øC) and allochthonous alkenones (warm signal, SST-30øC?). The sedimentary concentration of the alkenones is low, <30 ng/g. An input of an equivalent amount of ice-rafted alkenones with uK37'--I would increase a presumed U•:37 ' autochthonous signal of uK37'=0.2 ("•5øC) to 0.6 (•16øC) and decrease the percentage of 37:4 from an expected 10 to 5%, as tbund in the samples (Figures 2 and 3). Furthermore, the north-south gradient of SST (Figure 2) may indicate the source of allochthonous alkenones as principally coming t?om the Bareats Sea continental shelf. However, neither we nor Weaver et al. [1999] have direct proof of the hypothetical presence of "ancient" allochthonous alkenones. Conversely, the presence of 37:4 alkenone in the sediments (Figure 3) shows that autochthonous C37 alkenones are present in the sediment, that is, assuming that an input of allochthonous alkenones will not contain 37:4 because the occurrence of this component in marine samples relates to a cold rather than a warm signal. There is no evidence either to date that alkenones are present in any of the organic-rich shales that may be the source of the ice-rafted debris, or that they are associated to allochthonous coccoliths, so t•rther work needs to be done to discard or prove the presence of ancient allochthonous alkenones in LGM sediments. The origin of the LGM alkenones could be found out by looking at the 14C composition of the components and determining if they are substantially older than the LGM time interval. However, the sedimentary abundance of alkenones is very low, and it may be difficult to isolate enough pure material to carry out the isotopic analysis. A simpler, less costly alternative would be to look at the range of alkenone chain lengths, as ancient alkenones appear to have a larger abundance of C39 and C40 components [Farrimond e! al., 1986] than their contemporary counterparts. If, indeed, it is shown that reworked allochthonous alkenones are mixed with the local signal, then large parts of the glacial ocean with substantial ice-rafted debris (IRD) inputs will be affected, and the reconstruction of SST estimates from U•<37' may be challenging. It will all depend, however, on the amounts of IRD in the sediments (as small contributions may not bias the signal significantly) and the sources and their relative mixing (as some IRD may be alkenone-rich and some may not contain any alkenones at all). Thus the causes for such warm SST values in Figure 1 remain unclear. From the second and fourth points above, however, there are enough reasons to suspect that the uK37 ' and uK37 data cannot not be relied upon. If we assume that allochthonous alkenones (mainly the diunsaturated component) have biased the alkenone indices, the autochthonous signal will be better represented by the 37:4 component, in this case the distribution of 37:4%, which does not show the same patterns as uK37 ' (Figures 2 and 3). Could 37:4% be used to obtain an estimate of SST? Not directly, as 37:4% might have been originally higher, as shown in Figure 2, if allochthonous alkenones diluted the marine autochthonous signal, and when 37:4% > 5%, 37:4% is not temperature-dependent [Rosell-Mel& 1998]. In fact, the surthce sediments, i.e., less 37:4; and (3) the longitudinal and latitudinal gradients of 37:4% are not so pronounced as at present. From these 37:4% data it can be inferred that Arctic domain-type waters prevailed throughout the Nordic seas during the LGM as the 37:4% values are in the range of those tbund at present under Arctic domain waters. Thus it does not seem likely, t?om the alkenone data, that warmer and more saline southern Atlantic waters (with low 37:4%) [Rosell-Meli, 1998; RosellA4el• et al., 1998] reached the area, in agreement with the conclusions of Sarnthein et al. [1995] and Weiner et al. [1996]. However, interpretation of palaeocirculation is limited as the geographical distribution of 37:4% can be due to both oceanographic conditions and hypothetical dilution of the alkenone signal by ice-rafted alkenones t?om a predominantly eastern source. Nevertheless, in the Arctic domain region of the Nordic seas, in modern sediments, when 37:4% >_ 5%, then SST is approximately-<6øC [Rosell-Mel& 1998]. Thus a maximum SST of 6øC tbr the Nordic Seas can be inferred from 37:4%, which corresponds to the highest temperature of an Arctic water mass and a value not too dissimilar from the estimates of Sarnthein et al. [1995] and Weiner et al. [1996] obtained using tbraminitbra transfer functions. 5. Conclusions Sea surface temperature estimates derived from alkenone indices (uK37 ' and uK37) provide a surprising picture of the Last Glacial Maximum in the Nordic seas, with surface temperatures warmer than at present. We conclude that the alkenone estimates cannot be taken at face value and that one or more factors may have conspired to bias or modify significantly the molecular signal. One factor is that the oceanographic conditions in the Nordic seas during the Last Glacial Maximum were equivalent to those found in modern waters belonging to the Arctic domain and hence seasonally ice free. In this case, uK37 ' and uK37 do not relate to water temperature. An additional factor may be the hypothetical presence of allochthonous, ice-rafted, ancient alkenones incorporated in the glacial sediments masking the autochthonous molecular signal [Weaver e! al., 1999]. The allochthonous alkenones might contain a warm signal that might have forced uK37'/uK37 to much higher values. Although confirmation is required, we would urge caution on the interpretation of alkenone-derived palaeotemperatures in sediments with high concentrations of ice-rafted debris. However, the presence of an autochthonous signal can be intbrred from the widespread occurrence of the tetraunsaturated alkenone. The relative abundance of 37:4 allows us to estimate a maximum temperature for the Nordic seas during the Last Glacial Maximum of, at most, 6øC, which is similar to, or within the error range of previous estimates derived from foraminiferal transtbr functions [Sarnthein et al., 1995; Weinelt et al., 1996]. Acknowledgments. This research was supported by the Environment and Climate Programme under contract ENV4-CT97-0564, a contribution to the TEMPUS project, and the Natural Environment distribution of 37:4% in the LGM samples (Figure 3) is not too Research cøuncil through the pr63isio n of a t•liøwship (ARM)i W e dissimilar to that found in the modern surface sediments [Rosellthank Paul Farrimond and Jeremy Marlow lbr comments and discussion of the manuscript. M. Sarnthein and E. Jansen are thanked tbr the Meld, 1998] except (1) the values are higher in the east; that is, provision of samples retrieved by the vessels Meteor and Hakon Mosby. there is more 37:4 relative to the total of C37 ketones in the LGM p Weaver and M. Weinelt are thanked tbr their constructive reviews of than at present; (2) 37:4% is lower in the west than in present the manuscript. 776 ROSELL-MELt• AND COMES' A NORDIC SEAS WARM LAST GLACIAL MAXIMUM? References Andrulelt, H, Coccolithophore fluxes in the Norwegian-Greenland Sea: Seasonality and assemblage alterations, Mar. Micropaleontol. 31, 45-64, 1997. Bond, G., W. Showers, M. Cheseby, R. Lotti, P. Almasl, P. deMenocal, Po Priore, H. Cullen, I. Hajdas, and G. Bonani, A pervasi millenialscale cycle in North Atlantic Holocene and glacial climates, Science, 278, 1257-1266, 1997 Boon, J J, F W. van de Meer, P. J. W. Schuyl, J. W de Leeuw, P. A. Schenck and A. L. Burlingame, Organic geochemical analyses of core samples from Site 362, Walvis Ridge, DSDP Leg 40, Initial Rep. of the Deep Sea Drill. Proj., 40, 627-637, 1978. Brassell, S. C., Applications of biomarkers for delineating marine palaeoclimatic fluctuations during the Pleistocene, in Organic Geochemtstry Principles and Applications, edited by M H. Engel and S. A. Macko, pp. 699-738, Plenum, New York, 1993. Brassell, S. C., G. Eglinton, I. T. Marlowe, U. Pflaumann, and M. Sarnthein, Molecular stratigraphy: A new tool for climatic assessment, Nature, 320, 129-133, 1986 Clnnate Long-Range Investigation, Mapping, and Planning (CLIMAP), The surface of ice-age earth, Science, 191, 1131-1137, 1976. Climate Long-Range Investigation, Mapping, and Planning (CLIMAP), Seasonal reconstructions of the Earth's surface at the Last Glacial Maximum, Geol. Soc. Am. Map Chart Ser, MC-36, 18, 1981 Conte, M H, A. Thompson, and G. Eghnton, Primary production of lipid biomarker coinpounds by Emiliania huxleyi' Results from an experimental mesocosm study in 0ords of southwestern Norway, Sarsia, 79, 319-331, 1994 de Vernal, A, C Hillaire-Marcel, J.-L. Turon, and J Matthlessen, Sea-surface conditionss in middle to high latitudes of the North Atlantic during the Last Glacial Maximum (LGM): The cold paradigm revisited, Can. J. Earth Sc•., in press, 1999 Farnmond, P, G. Eglinton, and S C. Brassell, Alkenones in Cretaceous black shales, BlakeBahama Basin, western North Atlantic, Org. Geochem., 10, 897-903, 1986. F!cken, K J, and P. Farrimond, Sed,nentary llpid geochemistry of Framvaren' Impacts of a changing environment, Mar. Chem., 51, 31-43, 1995 Flugge, A Jahreszeithche Varlab!litat von ungesattlgten C•? methylketonen !n s•nkstofffallenmaterlal der Norweglschen See und deren Abblldung !n Oberflachensedimenten, Ph D dissertation, Univ zu Kiel, Kiel, Germany, 1997 Freeman, K H, and S G. Wakeham, Variations •n the distributions and isotopic composmons of alkenones in Black Sea particles and sedllnents, Org. Geochem., 19, 277-285, 1992 Ganapolski, A, S. Rahmstorf, V Petoukhov, and M Claussen, Simulations of modern and glacial chmates with a coupled global model of intermediate complexity, Nature, 391, 351356, 1998 Haug, G H, Zur Palao-Ozeanographle und Sed,nentatlonsgeschichte im Nordwest-Pazifik wahrend der letzen 6 Millionen Jahre (ODPSite 882), Ph.D. dissertation, Univ. zu Kiel, Kiel, Germany, 1996. Hebbeln, D, T Dokken, E. S. Andersen, M Hald, and A Elverhol, Moisture supply for northern ice-sheet growth during the Last Glacial Maximum, Nature, 370, 357-360, 1994. Hoet•, M. J. L., G. J. M. Versteegh, W. I. C. R[lpstra, J W. de Leeuw, and J. S. SinningheDamstY, Postdepositional oxic degradation of alkenones: Implications for the measurement of palaeo sea surface temperatures, Paleoceanography, 13, 42-49, 1998. Ko9, N., E. Jansen, and H. Haflidason, Paleoceanographic reconstructions of surface ocean conditions in the Greenland, Iceland and Norwegian Seas through the last 14 ka based on diatoms, Quat. Sci. Rev., 12, 115-140, 1993 Muller, P J, G Klrst, G. Ruhland, I. von Storch, and A Rosell-Mel•, Calibration of the alkenone paleotemperature index U•:q? ' based on core-tops from the eastern South Atlantic and the global ocean (60øN-60øS), Geochim. Cosmochtm. Acta, 62, 1757-1772, 1998. Prahl, F. G., and S. G. Wakeham, Calibration of unsaturation patterns in long-chain ketone compos ltlOnS for palaeotemperature assessment, Nature, 320, 367-369, 1987. Prahl, F. G., L. A. Muehlhausen, and D. I. Zahnle, Further evaluation of long-chain alkenones as indicators of paleoceanographic conditions, Geochtm Cosmochtm. Acta, 52, 2303-2310, 1988 Prahl, F. G, G J de Lange, M. Lyle, and M. A. Sparrow, Post-depositional stability of longchain alkenones under contrasting redox conditions, Nature, 341,434-437, 1989. Rahman, A, Reworked nannofossils in the North Atlantic Ocean and subpolar basins: Implications for Heinrich events and oceanic c•rculatlon, Geology, 23, 487-490, 1995 Reynolds, R W., and T M. Smith, A high resolution global sea surface temperature climatology, d. Clim., 8, 1571-1583, 1995. Rosell-Mele, A, Long-chain alkenones, alkyl alkenoates and total pigment abundances as climatic proxy-indicators !n the northeastern Atlantic, Ph D dissertation, Umv of Bristol, Bristol, England, U.K., 1994 Rosell-Mele, A., Interhemispheric appraisal of the value of alkenone indices as temperature and sallmty proxies in high-latitude locations, Paleoceanography, 13, 694-703, 1998 Rosell-Mele, A., and N Ko9, Paleochmatlc significance of the stratigraphic occurrence of photosynthetic biomarker pigments in the Nordic seas, Geology, 25, 49-52, 1997 Rosell-Melb, A., and J R. Maxwell, Rapid characterization of metallo porphyrin classes in natural extracts by gel permeation chromatography/atmospheric pressure chemical lomsation/mass spectrometry, Rapid Commun. Mass Spectrom., 10, 209-213, 1996 Rosell-Mele, A, J Carter, and G. Eglinton, Survey of distributions of long-chain alkenones and alkyl alkenoates in marine surface sediments t?om the north east Atlantic, Org. Geochem., 22, 501-509, 1994 Rosell-Mele, A, G. Eglinton, U Pflaumann, and M Sarnthein, Atlantic core-top calibration of the U•:•7 index as a sea-surface palaeotemperature indicator, Geochim. Cosmochtm. Acta, 59, 3099-3107, 1995a. Rosell-Mele, A, J Carter, A Parry, and G Egllnton, Novel kprocedure for the determination of the U •? in sediment samples, Anal. Chem., 67, 1283-1289, 1995b. Rosell-Mele, A, M. Welnelt, N Koq, E. Jansen, and M Sarntheln, Variability of the Arctic front during the last climatic cycle: Application of a novel molecular proxy, Terra Nova, 10, 86-89, 1998 Ruddlman, W. F, Late Quaternary deposition of ice-rafted sand in the subpolar North Atlantic (lat 40ø-65øN), Geol. Soc. Am. Bull., 88, 18131827, 1977. Samtleben, C., P. Schafer, H. Andruleit, A. Baumann, K. H. Baumann, A. Kohly, J. Mathiesen, and A. SchrOder-Ritzrau, Plankton !n the Norweglan-Greenalnd Sea: From living communities to sediment assemblages-an actualistic approach, Geol. Rundsch., 84, 108136, 1995 Sarntheln, M K. Winn, S. J. A. Jung, J.-C. Duplessy, L. Labeyrie, H. Erlenkeuser, and G. Ganssen, Changes in east Atlantic deepwater circulation over the last 30,000 years: Eight t,ne slice reconstructions, Paleoceanography, 9, 209-267, 1994. Sarnthein, M., et al., Variations in Atlantic surface ocean paleoceanography, 50ø-80øN: A timeslice record of the last 30,000 years, Paleoceanography, 10, 1063-1094, 1995. Slkes, E. L, J. K. Volkman, L. Robertson, and J.-J. P!chon, Alkenones and alkenes in surface waters and sediments of the Southern Ocean: Implications for paleotemperature estimation in polar regions, Geochim. Cosmochim. Acta, 61, 1495-1505, 1997 Teece, M A, J. M. Gethff, J. W Leftley, R. J. Parkes, and J R. Maxwell, Microbial degradation of the marine prymnesiophyte Emthania huxleyi under oxic and anoxic conditions as a model for early diagenesis: Long chain alkadienes, alkenones and alkyl alkenoates, Org. Geochem., 29, 863-880, 1998. V!llanueva, J, and J. O Grimalt, Pitfalls in the chromatographlc determination of the alkenone UK•7 index for paleotemperature estimation, J. Chromatogr. A, 723, 285-291, 1996. Wagner, T, Orgamsches Material in pelagischen Sednnenten Glazlale/Interglazlale Variationen nn Europalschen Nordmeet, Ph.D. dissertation, Unlv zu Kiel, Kiel, Germany, 1993. Wagner, T, and R Henrich, Organofacies and l•thothcles of glacial-interglacial deposits in the Norwegian-Greenland Sea. Responses to paleoceanographic and paleoclimatic changes, Mar. Geol., 120, 335-364, 1994. Weaver, P P E, D Rutledge, M. R. Chapman, G. Eglinton, G. Read, and M Zhao, Combined coccollth, tbramimferal, and biomarker reconstruction of paleoceanographic conditions over the last 120 kyr !n the northern Atlantic (59 ø N, 23øW), Paleoceanography, 14, 336349, 1999 Weinelt, M., M Sarnthein, U. Pflaumann, H Schulz, S Jung, and H. Erlenkeuser, Ice-free Nordic seas during the Last Glacial Maximum? Potential sites of deepwater formation, Palaeochmates, 1,283-309, 1996. Wu, G, and C H!llaire-Marcel, Oxygen isotope coinposition of Slmstral Neogloboquadrina pachyderma tests !n surface sediments' North Atlantm Ocean, Geochtm. Cosmochim. Acta, 58, 1303-1312, 1994 P Coines, Department of Fossil Fuels and Environmental Geochemistry, Drummond Building, University of Newcastle, Newcastle Upon Tyne NE 1 7RU, England, U.K. (pau comes@ncl ac uk) A Rosell-Mel•, Department of Geography, University of Durham, Durham DH1 3HP, England, UK (antoni rosell@durham ac.uk) (Received January 29, 1999, revised June 14, 1999, accepted July 1, 1999 )