Initiation of the last glaciation in northern europe

@umrr~ury Science Keview, Vol. 16, pp. 1 197-I 215. 1997. Pergamon I% 199X Published hy Elwvier Science Ltd. ,411rights reserved. Printed in Great Britain. PII: SO277-3791(97)00028-O INITIATION OF’ THE LAST GLACIATION LARS FORSSTR6M* *Department f Department and MIKKO 0277-379 IN NORTHERN PUNKARIJ-$ University of 0~1~1, Linnanmaa, FIN-90570 Oulu. Finland of Geology and Geophysics, University of Edinburgh, King ‘s Buildings, Edinburgh EH9 3JW, U.K. (E-mail: Punkari@mpurzk.l,p..fi) INTRODUCTION Terminologically, the Weichselian/Wisconsin tion follows the EemianKangamonian interglacial, glaciacorre- sponding to marine Oxygen Isotope (01) Substage 5e (130-l 17 ka BP) and reachedits maximum extent during 01s 2 about 20 ka BP. The Early Weichselian includes (01s) Substages5d-5a. However. the time of the initial growth of the last ice sheet has been debated both in Europe and North America. Three main alternative periods have been proposed: (a) soon after the interglacial. SubstageOIS 5e; (b) during the transition of 01s 5/4, and (c) during the transition of 01s 3/2 (see Mangerud, 1991a; Imbrie et al., 1992; Clark it (11.. 1993; Baumann ef al., 1995). Some schemesenvisage several periodsof glacial growth with intervening ice-free periods. According to Ljungner (1949), the build-up of the last Scandinavian ice sheet began in the Norwegian mountains, where small mountain glaciers expandedand joined to form a uniform ice sheet.Basedon glacial morphology FIN-02400 author. Present address: Humaljirventie Kirkkonumml, bland. EUROPE of Geology, Abstract - The bio- and chronostratigraphy of the Eemian interglacial (marine isotope substage 5e) and an Early Weichselian glaciation (5d-a) established from representative and detailed sequences can be correlated with the deep-sea oxygen isotope stratigraphy. ice-core data, sea-level fluctuations and coupled ice sheet-climate models. Biostratigraphic sequences from Fennoscandian key sections are correlated with reference sequences from Estonia and from sections located near 01 beyond the margins of the last glaciation. Organic sediments previously attributed to Early and Middle Weichselian interstadial periods in Finland are argued to be redeposited and mixed older (last interglacial) material. Pollen and diatom spectra of the undisturbed materials suggest that the Eemian climatic optimum was followed by a continuously cooling climate and a regressive marine level. If only undisturbed sequences are considered. the major climatic tluctuations of the Early Weichselian, apparent in Central and Western Europe, are not apparent in the sequences from the central part of the glaciated terrain. Instead, some sequences are truncated by sediments indicating approaching ice sheets soon after the interglacial. This may imply that the ice sheet grew over Finland during the first Early Weichselian stadial. The prcscrvation of the interglacial beds and the lack of younger non-glacial sediments support the interpretation that the area remained ice-covered until the final deglaciation. During the Early Weichselian. the Norwegian coast was probably occasionally ice free, similar to the coastal zone of Greenland today. The authors’ interpretation of the Fennoscandian organic deposits of the last glaciation may also explain I;imilar observations from the central parts of the Laurentide ice sheet. 0 19% Published by Elsevier Science Ltd. All rights reserved $Corresponding II97 $32.00 22. 1197 QSR and striations, he argued that the Weichselian glaciation comprised two stadials, separated by an interstadial during which the ice sheet melted partly or completely. This model gained support in Sweden, where Lundqvist ( 1967) interpreted waterlain, till-covered sediments in JLmtland to representan interstadial horizon dividing the Weichselian glaciation into two parts. Lundqvist correlated this Jlmtland interstadial with the warmest Early Weichselian interstadial, Briirup in Denmark (Andersen, 1961). In northern Finland, Korpela (1969) similarly proposedthat till-covered organic bedscontaining pollen and plant macrofossils indicative of cool climates belonged to the so-called Perapohjola interstadial, which was correlated with the Jsmtland interstadial. He suggestedthe period to have occurred at about 35 ka BP basedon a single finite radiocarbon date. One of the present authors (ForsstrGm, 1982, Forsstriim, 1984) compiled European stratigraphic data and correlated the two main Early Weichselian interstadials (Br8rup in Denmark, Andersen. 1961; and Odderade in northern Germany, Averdieck. 1967)with 01s 5c (100 ka BP) and 5a (80 ka BP). He suggestedthat the JCmtland and Pertipohjola sedimentscould representeither of these interstadials. In Sweden, b~~rneorganic beds were also correlated with the Odderade (Morner. 198 I ; Lagerback and Robertsson, 1988; Lundqvist and Mook. 1981; Lundqvist and Miller, 1992; Robertsson and Ambrosiani, 1992) and similarly in northern Norway, two Early Weichselian organic horizons were proposed (Olsen. 1988). Mangerud ( 199 I a, b) and Lundqvist ( 1992) argued that the Scandinavian ice sheet melted almost entirely during these two Early Weichselian interstadials. Tanner ( I91 5) observed that in northern Fennoscandia. northerly glacial flow lineations representing an early stage of the glaciation are overprinted by younger ones indicating more western ice flow during deglaciation. Korpela (1969) suggested that these two flow directions might represent two different glaciations. The two were later correlated with the Early and Late Weichselian stadiais (ice sheet growth phases) in Finland. based on till stratigraphy and the dating of organic fragments (Hirvas et rrl., 1977; Hirvas. 1991). Aario and Forsstriim (1979) and Punkari (1984, 1991) criticised the model of two superimposed stadial flow systems because it neglected the shift of the ice divide during the last glaciation. and because the last glacial-deglacial cycle was apparently represented in the stratigraphy by only a thin cover till. These workers instead proposed that different ice flow directions were a consequence of shifting ice stream configurations during the last glaciation or deglaciation. Thi5 new concept raised the question of whether there is any real evidence for multiple Weichselian stadials in central Fennoscandia. In this paper a new reconstruction of the chronology and environments of the last glacial inception in northern Europe is proposed. Biostratigraphic information on the Eemian interglacial and Early Weichselian glacial is compiled from representative and detailed sequences from Estonia and different sites in Fennoscandia. The sequences are then correlated and these are compared with each other and with undisturbed sequences located near or beyond the margins of the last glaciation. The latter indicate the development of vegetation from the climatic optimum of the Eemian interglacial towards a periglacial environment. followed by the spreading of ice sheets. We then estimate the timing of glacial growth in the central parts of the Scandinavian ice sheet. We argue that previous chronologies based on absolute dating methods are in error. Instead. we correlate our results with other data sets, which reflect continental ice volume variations, such as deep-sea oxygen isotope stratigraphy. records of ice-rafted detritus (IRD) and sea-IevJel data. Coupled ice sheet-climate models also yield information about the possible spatial distribution and temporal fluctuations of ice volutne. METHODS OF DATING NON-GLACIAL EVENTS Previous reconstructions of Early Weichselian events in northern Fennoscandia are based mainly on absolute dating methods. The limitations and possible errors of such methods are considered next. The first attempts to develop a chronology used radiocarbon dates, but these were later recognised to be infinite within this time frame. Subsequent schemes have relied on thermoluminescence (TL), optically stimulated luminescence (OSL). amino acid and uranium/thorium (U/Th) dating, but the methods are still of low resolution or experimental and subject to limitations (Dreimanis, 1992: Clark (of LII., IY93). For example, several TL and OSL dates of Finnish sub-till samples range from 80000 to 100000 years BP and have therefore been correlated with the Early Weichselian (Jungnet-, 1987; Hiitt et (11.. 1993). However, if the ages are corrected by about 355 40% to account for the effect of shallow traps in the crystals (Me,jdahl. 1992), they suggest interglacial or older ages. OI- indicate incomplete bleaching of the crystals (Punkari and Forsstrijm, 1995). In Lithuania, OSL dating of the Tilirr zone of the Eemian interglacial gave an age of 833~8 ka, the expected age being about 125 ka (Gaigalas, 1993). In North America, TL estimates of 69-77 ka have been obtained for sediments also regarded as last interglacial (Clark et d., 1993). Recently. younger OSL ages of 36.6-70 ka have been obtained on sub-till sands from Finland (Nenonen, IYYS). The authors believe that these scattered ages are not related to ice-free stages. because other data sets for the same period indicate complete ice cover over Fennoscandia. as discussed later. Successful U/Th dating of peat requires thick and impermeable deposits to have created a closed system and prevented the introduction of external thorium (Dreimanis. 1992: Heijnis, 1992). The UlTh dated sandy organic beds at Oulainen (95125 ka) and thin peat layers at Tervola (80-IO0 ka), both in Finland. and thin peat layers at Pilgrimstad in Sweden (100 ka) (Heijnis, 1992) and at Tisjo in Svveden (75 ka) (Lundqvist and Miller. 1992) do not meet these requirements, and the Early Weichselian ages obtained should be regarded as minima for those deposits. Similarly, a wood sample from peat in Canada gave a UiTh age of 67.7 ka, although lithostratigraphic and palaeoecologic data indicate its interglacial (01 Substage Se) age (Causse and Vincent, 1989). The amino acid dating method has been applied to molluscs and foraminifera along the Norwegian coast as the basic method to correlate the sediments with marine oxygen isotope stages (Miller et al.. 1983). Amino acid dating is based on the time dependency of the rate at which the isoleucine epimerization isomer ratio (aIle/Ile) changes. The method is used successfully as a relative dating technique, using the same species to correlate adjacent sites with similar environmental histories, However, this ratio is also temperature dependent. approximately doubling for every 4°C increase in temperature (Miller and Hare, 1980). In addition, the ratio is dependent on several other factors which have not been considered in earlier datings (Kaufman and Sejrup. 19YS). The method has thus not given reliable or only low resolution numerical ages and these may vat-y within the same stratigraphic horizon (Funder et al.. 199 I ; Kaufman and Sejrup. I9YS). L. Forsstriim et ~1.: Initiation of the Last Glaciation in Northern Europe Biostratigraphy of terrestrial sequences indicate that the Eemian interglacial was the only forested period during 01s 5 in Estonia (Liivrand, 1991) and the Early Weichselian interstadials were significantly cooler than the Eemian in Central and Western Europe (Andersen, 1965; Behre, 1989; Guiot er al., 1992). The beds containing organic matter in Finland, described later, are within or beneath a relatively thin till, and thus it is unlikely that they could represent older interglacials than the Eemian. Thus the authors correlate such interglacial deposits with the well established Eemian interglacial in Central and Western Europe. One of the authors (Punkari, 1984, Punkari, 1991) investigated the sedimentology of sub-till organic beds in Finland and argued that they are mostly reworked and so cannot be used to date Early Weichselian stadial or flow events. Such deposits have been shown to contain interglacial-type pollen (warm indicators) or pollen indicative of cooler phases pre- and post-dating the interglacial. Previous biostratigraphic interpretations were mostly based on a concept that any sub-till organic sequence, including evidence for cool climates, represented an interstadial. The possibility of redeposition, and so of incomplete coverage, was neglected for such pollen sequences. However, some fragmentary sequences in Fennoscandia can be correlated with undisturbed sequences elsewhere, for example with that at Prangli. Estonia, which contains all the Eemian phases and a subsequent cold period covering the beginning of the Early Weichselian (Liivrand, 1991). In this paper, biostratigraphic sequences are correlated in order to make chronostratigraphic inferences. The reference sections and observations made in Fennoscandia are described next. REFERENCE SECTIONS The succession of vegetation during the last interglacial is well known according to several representative pollen sequences from Denmark (Andersen. 196 I). northern Germany (Behre. 1989), Poland (Krupinski, 1995). Russia (Grichuk, 1961; Liivrand, 1991), France (Woillard, 1978: Guiot rt al., 1992) and Great Britain (West. 1957). The pollen spectra of these sequences indicate a significant warming correlated with the Eetnian interglacial and a subsequent lengthy cold phase showing less pronounced climate fluctuations. The two interstadials correlated with the 01 sub-stages 5c and 5a are represented by relatively short and moderately warm phases followed by a periglacial climate. A largely complete interglacial section is known from Prangli, Estonia (Liivrand, 1991), close to Finland (Fig. 1). The Central and Western European sequences and especially the Prangli sequence can be used to correlate the fragmentary sequences from Fennoscandia. Prangli Liivrand (1991) presented analyses of cores through marine interglacial beds which extended for several 1199 kilometres across the island of Prangli (Fig. 1). The Prangli pollen succession seems to represent an almost complete record of the last interglacial and the subsequent cooling of the climate to an Early Weichselian periglacial environment (Fig. 2). An upward shift from marine finegrained sediments to sands reflects the lowering of regional sea levels associated with the build-up of the last continental ice sheets (e.g. Zagwijn, 1983). This is supported by the varved clays above the sands, which would have been deposited in a proglacial basin. The overlying till was most likely deposited by the same ice advance. After the Eemian Picrrr maximum (Ccrr-pi77u.s :otze, E6), thermophilous trees such as QUC~~UJ. Ulrrzir.s and Tiliu rapidly diminish, wjith the exception of Ccwpinus. Pinus dominates after the decline of Picea in the last zone of the interglacial (E8), indicating cooling. During the Early Weichselian (zones EW I. EW2 and EW3). Berultr rises and dominates at the expense of Pirzus. Tundra-like periglaciat species including Betdo ~mzc~. Poaceae. Cyperaceae. Artrmisitr, Chenopodiaceae and El-iur1r.v prevail. This transition lies below the change of sediments from marine clays and silts into sands 1.5 m thick at 65.85 tn depth. The sands, which are devoid of pollen, are overlain by varved clays 0.6 tn thick, in turn overlain by a till layer 20-60 tn thick. The pollen sequences appear continuous from the underlying marine silty clay to overlying clays. The sands have similar thicknesses in five adjacent cores and may represent only a short period of deposition. Thus the gap in the pollen sequence caused by the sandsis probably insignificant. Liivrand (1991) divided the Early Weichselian part of the sequenceinto three pollen zones. the middle of which (EW2) was correlated with the Brorup interstadial (Andersen, 1961). Although a slight warming is evident, the authors suggestthat the differences between the zones are not significant. Zones EW I and EW2 do not record a long interval of climate changesin vegetation such as of SubstagesSd and SCelsewherein Europe. The sequence records only one significant change in vegetation. the disappearanceof pine and its replacement by birch and NAP species and only little more than one metre of sedimentswas deposited (Fig. 2). These changesare not comparable with the records of the Early Weichselian in northern Germany, for example (see Behre, 1989). Thus the zones EWI and EW2 at Prangli probably correspond to SubstageSd and together cover a period of lessthan t 0 000 years. After the interglacial, silts and clays in places up to several tens of metres thick were deposited in buried valleys of the Estonian mainland (Liivrand, 1991). Pollen analysis of the sediments indicates periglacial flora and various amounts of redeposited interglacial palynomorphs. The pollen stratigraphy reveals no significant changes which could be correlated with the Early Weichselian stadial and interstadial cycles, although this could be due to the mixing of material by periglacial processes(Liivrand, 1991. Fig. 57). Liivrand ( I99 I) suggestedthat the ice sheet did not cover this 1200 I Norwegian ‘Sea ------, INLAND’: fJ I ‘\ 24’ \ I I 1 400km ’ \ 32O \ FIG. 1. Locations of reference sections. the maximum extent of the Weichselian glaciation and the marginal position during the Younger Dryas stadial of the deglaciation period. area until OIS 4, when the global ice sheetswere more extensive. The lack of indications of the Early Weichselian climate fluctuations and the glaciomarine varved clays overlying the interglacial beds at Prangli suggestthat the ice sheet would have covered southern Finland for most of Early Weichselian time. BIOSTRATIGRAPHIC DATA FROM FINLAND Several sections containing pre-Late Weichselian organic matter are known in northern Finland, summarised by Hirvas (1991). According to our observations, most are incomplete and disturbed, limiting their utility for palaeobotanic and climatic reconstructions of ice-free periods. There are no sequences showing a real interstadial succession, although some pollen spectra may indicate cool or cold climate. Organic matter previously correlated with interstadials has been interpreted to lie between late-glacial deposits. which means that they actually are in a secondary position (Aario and Forsstrijm, 1979; Punkari? 1984, Punkari, 1991; Punkari and Forsstrhm, 1995; Forsstrdm, 1995). Sections are also known in central and southern Finland. closer to the reference sequence of Prangli (Liivrand. I991 ) and the Central and Western European sequences.Here we describe key sections in Finland and show that these sequencescan be correlated with interglacial reference sequencescovering periods from the early Eemian onwards (seeFig. I for locations). Evijiirvi The section (60 m a.s.1.)containshorizons of gyttja and silt between till layers. Pollen, including abundant L. ForsstrGm et 01.: Initiation of the Last Glaciation in Northern Europe PRANGLI sand FIG. 2. Lithostratigraphy and simplified pollen diagram from Prangli. Estonia (Fig. I) covering the Late Saalian, Eemian and Early Weichselian. Proglacial sediments and till in the upper part of the sequence record the first Weichselian glaciation Quercus and Picea, and marine diatoms show that the gyttja and silt were originally deposits of the Corylus, Eemian interglacial sea and indicate that the area was the sequencecontains mostly freshwater diatoms, Griinlund (1991) concluded that the interglacial inception in the area corresponded to a freshwater phase. submergeduntil late interglacial time (Eriksson et al.. 1980; Eriksson, 1993). Haapavesi Interglacial marine silt and gyttja layers are covered by till. Diatom and pollen represent an earlier phaseof the Eemian interglacial sea than the Evijlrvi sequence. However, the uppermost comminuted organic matter was originally deposited synchronously with the Evij4rvi deposits,but in a terrestrial environment. This sequence suggeststhat the Eemian sea covered the area at the beginning of the interglacial and reached about 100 m a.s.1. during the invasion of spruce (ForsstCm et al.. 1988). Norinkylii The sedimentsof Norinkyll have been correlated with those of Haapavesi and Evijirvi, but are also considered to include the very beginning of the interglacial (Eriksson. 1993). Becausethe clay in the lower part of Oulainen Forsstriim (1982). Forsstriim ( 1984) proposed that depositscontaining organic matter in a till-covered esker at Oulainen (Fig. 1) could record the beginning of the Eemian interglacial. This was based on the observation that birch forest pollen give way to pine forest pollen, which is a natural successionin the beginning of a warm period. However, the matter had depositedwithin a freshwater environment, and other data indicate that, at that time, the area was submergedby the sea(ForsstrGmet al.. 1988). Donner (1983) correlated the deposits with the Early Weichselian Briirup interstadial. However. the vegetation suggestsa much warmer climate than expected for that interstadial. The organic fragments could have been lifted into the ice and mixed, so that the vertical succession is not representative of climate change (Punkari. 1991). This interpretation makes it possible that the sediments could represent the end of the interglacial (Forsstriim. 1989. Forsstriim, 199I ). Vimpeli The sequence of well-preserved organic fragments within a till bed at Vimpeli resembles the one at Oulainen. in that it also suggests a terrestrial environment and an upwards warming trend (Aalto et trl.. 198.1. 1989). The interglacial nature of the microfossil content is more pronounced. Some species represent a climate warmet than the present: one example is 7’huli(~trunt Iuc~c/u/~~, which today grows south of Finland (Aalto rt trl.. 3983). The organic matter at Vimpeli intrude\ into till from beneath and is thus obviously redeposited and mixed (Punkari. 1991 ). The microfossil content and sediments of terrestrial origin indicate that this bed can be correlated with the upper sequence at Prangli, representing the end of the interglacial, similar to the Oulainen aequencc (Forsstriim, 1991). Harrinkangas At Harrinkangas. well-preserved, folded moss layers occur within fine-sand esker sediments which are covered by a thin till bed (Punkari. 1988). These organic beds have been interpreted to be ill situ (Gibbard ot trl., 1989). but were later argued to be redeposited (Punkari and Forsstrdm, 1995). Pollen analyses of the sediments indicate a colder climate than the present. but the macroscopic remains of pine (P ifz~s .s$~~sr~i.s) beneath the moss layers indicate an interglacial origin. These facts suggest that the beds can be correlated with the upper part of the Prangli sequence reflecting the end of the Eemian interglacial. INTERGLACIAL VEGETATION AND INITIATION OF GLACIATION IN WESTERN FINLAND We argue that the sequences previously interpreted to indicate Early (and Middle) Weichselian interstadials in Finland actually comprise redeposited and mixed interglacial material. A reconstruction of the interglacial vegetation history using such materials is not straightforward, but is possible by comparison with reference sections. The interpretable pollen successions. the degree of disintegration and the internal structure of the organic beds selected for the reconstruction, all suggest that these sediments could be slices of more extensive beds, transported and redeposited by the ice sheet. The authors assume that transport distances were in the order of kilometres. similar to most till materials (Salonen, 1986). Thus the sequences can be used to infer former climatic conditions and elevations of sea-level within a certain area of flat terrain and thin surficial deposits. A hypothetical compilation of the biostratigraphy of western Finland, based on the sections described earlier. is presented (Fig. 3: cf. Eriksson. 1993). Pollen assemblage zones of the last interglacial have been constructed by combining data from different sections in western Finland. The data of zones I. II, 111and IV derive from NorinkylC section (NOR), zones I and IV originating from one core (NOR1 ). and zones II and III from another (NOR2). Zone V originates from the Vimpeli section (VIM) (Aalto et (il.. 19X9), and upper zones VI-VIII from Oulainen (OUL) in a generalised form (Forsstriim, 1982. 1984). I, Birch period. Erikason ( 1993) defined the BetulLt pollen assemblage Lone ax the first interglacial zone in Finland. where birch is the dominant tree pollen. Griinlund ( IYY I ) concluded that the clay in these sediments was deposited during a freshwater stage prevailing in the Baltic Sea basin. However. the Prangli section in Estonia shows that saline water invaded the Baltic Sea during an early stage of the interglacial (Liivrand, I99 1). The freshwater clays may instead represent proglacial sediments mixed and redeposited by the ice sheet. Therefore the authors question whether these observations of the earliest Betula zone are correct. even if it is most likely that the interglacial started with birch-dominating forests (Fig. 3). II, Bird-pittr-otrk pet.iod. Abundant Quely~~ pollen (about 5%) show that the vegetation is clearly interglacial in nature. Marine deposits representing thix zone occur at altitudes of more than 100 m a.s.1. (Griinlund. 199 I), showing the minimum extent of the \ea in western Finland. III. Birc,h-rrltlrr-htr,-rl priod. Alms pollen increase to the same proportion as Brtttlrr. Cor?;/us pollen rise to their maximum. exceeding 10%. From the thermophilous trees, QUCUXS is still abundant, U/m~u is frequent and Emrn.rittu.\ and Tiliu are rare. Marine diatom flora indicate that sea-level remained at a high level. IV, Rin.h--LildCI.--.sl:j)t.tl(.(l-lift,el period. Picea pollen increase to a maximum of 30%. Small amounts of Curpirtt(.y occur. reaching a maximum of 2% corresponding with the Piwct maximum. Cor$u.s remains abundant, but Qtwrwtttttt misturn species decrease significantly. Sea-level has dropped during the Piceu maximum to a level of about 100 m a.s.1. (ForsstrGm et trl., I Y88; Eriksson. 1YY3). V, Pi/?~-hi~-~lt-,s~)t~t~~~~, pc,riotl. Pinus pollen predominate, but Betulcr and Piwtr remain abundant. Alms and thermophilous trees disappear almost completely. Deposits are terrestrial, indicating lowered relative sea-level (Forsstriim et (11.. 1988). VI, Pine period. Pinrr.s predominates in pollen spectra (80%). but Brttrltr is also common (IO-I 5%). The proportion of non-arborcal pollen (NAP) is low in the lower part of this zone. but increases upwards. The deposits representing this Lone at Oulainen (Forsstrbm. I Y82) and Vimpeli (Aalto et cl/., 1983) are terrestrial. so 1hat sea-level i\ at a low level. VII, Pirtp--l?jy-(.lt--/t1t~. pcriotl. Pirtus and Brttkt dominate the arboreal pollen ( AP). The percentage of NAP increases in this Lone. Poaceae being most common NAP. The deposits are terrestrjal at both Oulainen (Forsstrijm, I Y82) and Harrinkangas (Punkari and Forsstriim, 1995 ). VIII. Ho>, period. The amount of NAP increases more L. Forsstriim ut ml.: Initiation of the Last Glaciation in Northern Europe OSTROBOTHNIA I I Er-3 clay L-zj Silt m Gyttja silt w FIG. 3. Schematic pollen diagram of western Finland Oulainen (OUL), Vimpeli (VIM). EvijLrvi (EVI) sequences can be correlated with the interglacial Gyttja covering by the reduction in thermophilous wzz’,‘&d[ the Eemian and Norinky trees. The section representing zone V derives from Vimpeli. zone VI from Oulainen 130 km farther north (Fig. I). Excluding this geographical difference. zone VI may reflect a cooling climate which did not favour the growth of spruce forests. The vegetation would correspond to the present pine forest area of northern Finland, north of the occurrence of spruce forests. where the mean annual temperature is about - I “C and the mean July temperature +14”C (Atlas of Finland, 1987). ’ 1 A jS’$zz” and Early ; Weichselian, ] Sand B compiled ‘il. Pinus i III, BetulaAlnusGOryIllS ~~,$r~s&d from peat sections at key See text for (NOR]-?). The compilation showz that the Finnish reference sequences references. in relation to AP and Poaceae is the predominant NAP. The deposits remain terrestrial at both Oulainen and Harrinkangas. The pollen succession compiled in Fig. 3 corresponds well with those of the Eemian interglacial from Denmark (Andersen. 1965), Estonia and north Germany, especially considering the northern location of Finland. The inferred development of sedimentary environments from a proglacial basin or marine conditions to later terrestrial conditions corresponds to expectations of glacio-isostatic land-uplift during an interglacial. A distinct cooling of the climate between zones IV and V is suggested m I i horn Estonia and Ccnlral Europe. Pine declines as a consequence of further cooling. and open pine-birch woodland is indicated by NAP which dominates in zone VII. Vegetation resembles present day areas close to tundra in northernmost Finland and northern Norway, where Brtulu tortuosa forms the forest limit (Vaarama and Valanne. 1973; Hyvarinen. 1976). The uppermost Poaceae Lone probably represents tundra vegetation dominated by grasses. The proposed pollen succession (Fig. 3) is unlikely to include all changes in interglacial vegetation development. For example, no clear birch forest phase can be seen at the end of the interglacial. although this has been recorded at Leveaniemi in Sweden (Robertsson. 197 I ) and at Prangli in Estonia (Liivrand. 1991). This retlects the mixing of the organic matter in the Finnish sequences during glacial transportation. Mixing can be seen in the sequence al Oulainen. which represents the end of the interglacial in the compiled diagram. and at Vimpeli. At both locations a birch zone predates a pine zone (Forsstriim. 1982. Forsstriim, 19X4: Aalto et cl/., 1983). Such a vegetation succession is not known elsewhere from the end of an interglacial, and the deposits could not 1204 Q~rurrrnary Sciencr Reviews: Volume represent the beginning of the ice-free period, as argued earlier. An inverse stratigraphy could have been generated as a result of folding of glacially transported material. or following a mechanismwhere material from a land surface is picked up by a glacier and is then deposited elsewhere before the deposition of material eroded later from subsurfacelayers of the sameoriginal bed (Punkari and Forsstriim, 1995). The replacement of forest vegetation by tundra in western Finland indicates significant cooling, in association with the expansionof ice sheets.However, there is no indication in the Finnish stratigraphic data when the Scandinavian ice sheet extended to that area. The abundance of well-preserved and non-disintegrated redeposited organic matter representing the interglacial, preserved within Late Weichselian sediments, suggests that the glacier covered western Finland only once during the Weichselian glaciation (ForsstrBm, I99 I ; Punkari and ForsstrGm. 1995). Harsh conditions during a long-lasting subsequentice-free periglacial period. or repeatedglacial advances, would have destroyed fragile interglacial organic beds. However, they would have remained intact beneath stagnant cold-basedice, particularly in englacial position. BIOSTRATIGRAPHIC DATA FROM OTHER PARTS OF FENNOSCANDIA Sweden Laeiiniemi Pollen analysis of this section in northern Sweden (Fig. 1) suggests that it represents most of the last interglacial (Robertsson. 1971). Here only the upper pollen zones are described, recording a continuous cooling of climate to glacial conditions representedby a till, which may have disturbed the top of the succession (Fig. 4). Zone c: Picea-Pinus (-Bet&-Alaus). Piceu increases dramatically during this zone to 45% of AP. This should represent quite a late stage of the interglacial becauseduring the Holocene an equivalent increaseof Picea has been dated 3085 years BP (Robertsson, 197I). Some Co~lus pollen, probably due to longdistance transport. occur through the zone. Zone d: Pinu-Beth (-Ahus). Piceu is greatly reduced and Pinus and Betulu predominate. A1nu.s decreasesand Cor$us remains present. Picea disappears entirely at the end of rhe zone indicating climate cooling. Zone e: Betulu (-Pinus). Betulu risesto more than 90% of AP showing further cooling, but decreasesto just over 50% at the end of the zone as Pinus becomes abundant. The increaseof Pinus at the end of the zone may indicate disappearance of birch forests and increasing influence of long-distance pollen transportation by wind. COT/USis sporadically present. Zone f: Pinus-Beth (-Alnus). Pinus increasesat the expense of Beth, then diminishes again, Alms rises but does not exceed 10%. The in .vitu nature of the 16 zone is questionable. The increase of both Pinus and may indicate warming, or the upper layer may be mixed and redepositedby glacial flow (Robertsson, 1971). Thermophilous pollen in the upper part of the sequencemay also be due to increasing long-distance transportation. Altzus Pilgrimstud Kulling (1945) describedthis section in central Sweden (Fig. 1) showing till-covered gravel bedscontaining gyttja and silt, and also a few mammoth and deer bones. Lundqvist (1967) correlated the organic beds with the Early Weichselian Brarup interstadial and usedthe site as a key section to divide the Weichselian glaciation into two parts. Robertsson( 1988) discussedwhether two Early Weichselian interstadials may be representedor only the younger (Odderade). One of the present authors (Forsstrdm, 1989, Forsstriim. 1991) reinterpreted the sequence, using the detailed description of the section by Kulling (1945) (Fig. 6). and concluded that the organic beds are redeposited and could represent the end of the last interglacial. The bones, and also a varved clay layer containing abundant macroscopic remains of plants (Dyes, Suliw, Brrulu IZCII~U, etc.), were interlayered with glaciofluvial gravels, indicating redeposition after englacial transportation. Norway On the southwesterncoast of Norway, biostratigraphic analysesand amino acid dating methods have been used to correlate the sedimentsof Fjasanger and BQ (Fig. 1) and to correlate these with Weichselian oxygen isotope stages(Mangerud et al., 198la; Miller et al., 1983). The results have been used to construct time-distance diagrams of the Scandinavian ice sheet during the Early Weichselian (Mangerud. 199I b). Mangerud et al. ( 198la) interpreted this section of marine deposits, comprising gravel, sand and silt, to represent a continuous Eemian-Early Weichselian succession(Fig. 5). Pollen spectrarepresentingthe beginning of the interglacial contain pioneer vegetation (Betulu. Pinus. Ju~Qenr.c). followed by thermophilous species (Quercus, Car-plus, Alrzus). and later by a Piceudominated (up to 40%) coniferous forest flora assemblage at the end of the interglacial. At the beginning of the Early Weichselian cold period, the proportion of herbs and shrubsrises with the decline of Piceu reflecting the disappearanceof forests. The marine and glaciomarine sedimentsof this phasecontain abundant Alnus and Cot$u.v pollen which must be redeposited. Mangerud et (I/. ( 198la) recognised this, but proposeda Poaceae-Ericaleszone as a characteristic pollen assemblageof the Early Weichselian. The pqllen spectra suggestconstant climatic conditions, likely also due to redeposition or rapid deposition. Nonetheless, L. Forsstriim et al.: Initiation of the Last Glaciation in Northern Europe 1205 LEVEiiNIEMI /5iZi TREES(ApI U Setula -C-Alnus -e * Pinus Picea -e QM ~Trees~ShrubstjHerbs Drift m Kdy m Sand Gyttja peat a Till FIG. 4. Lithostratigraphy and simplified pollen diagram from LeveCniemi, northern Sweden (from Robertsson, 1971). Mangerud et al. (1981a) differentiated the Fana interstadial based on 0.6 m of gravel containing an abundant molluscan fauna and foraminifera. They concluded that during 01 substage 5d the ice margin was as extensive as during the Younger Dryas stadial (Fig. 1). but receded during the Fana interstadial from the fjord and its catchment area. The Fana gravels are interlayered with glaciomarine silts recording colder climates. The overlying till was deposited immediately after the deposition of the silt by an advancing glacier as shown by Mangerud ( 199 1a) by several arguments. Mangerud et a/. (I 98 la) argued that the Fana interstadial was cooler than the warm Brorup interstadial in Denmark and correlated it with the first Weichselian interstadial in Denmark, Rodebaek (Andersen, 196 1). Thus the entire Early Weichselian succession would have been deposited during 01 Substage 5d. Subsequently. amino acid age estimates obtained on molluscs and foraminifera in the Fana gravels of about 100 ka BP led to its correlation with 01 Substage 5c (Miller et al., 1983: Sejrup, 1987; Mangerud. 199 I a. Mangerud, 1991 b; Baumann et al., 1995). Isoleucine epimerization (aIle/Ile) ratios were measured in the pelecypod mollusc Mya truncatula and benthic foraminifer Chides lobatulus at Fjosanger. The method gave an age of 93 ka for molluscs in the Fana gravels. This age corresponds to the Brorup interstadial (01 Substage 5c) (Mangerud, 199la). The temperature model used and the assumption that the change of isomer ratio was linear through Weichselian time may be among reasons for unreliable (too young) ages for these samples (Miller et al.. 1983; Kaufman and Sejrup, 1995). Ratios for the foraminifera were similar to those of interglacial samples, suggesting that at least the foraminifera were redeposited (Miller et al., 1983). The authors agree with the earlier interpretation by Mangerud et ~1. ( 198 1a) in that the Fana interstadial seems to represent such a minor and short-lasting climatic warming that it could be related to local factors. to the oscillation of mountain glaciers, for 1206 FJQSANGE Picea Silt FIG. 5. Lithostratigraphy and pollen diagram from F$anger, et al.. example, and not to the main Early Weichselian climate fluctuations. Andersen et al. (1983) described the B@ section where layers of marine sand are interbedded with diamicton (Fig. 6). A lower sand horizon with shells, the Avaldsnes Sand (El), contains abundant Picerr (>50%) and can be correlated with the end of the interglacial at Fjasanger (Fig. 5). The overlying 0.4 m thick stratified Torvastad Sand (Dl) was deposited in a sheltered. brackish covering IYHIa). the Eemian and Early Weichselian (from Mangerud environment, possibly a lagoon. Pollen spectra indicate a cool climate, possibly lacking trees during the latest phase. Molluscan fauna include species which tolerate cold water, similar to marine conditions of the middle to low arctic region today. The lower Avaldsnes Sand. representingthe end of the interglacial is overlain by the Torvastad Sand, and no hiatus is seen between the deposits (Andersen et al., 1983). However, based on young amino acid ages (7184 ka) for the Torvastad Sand, Andersen IX (11.(1983) regarded it much younger than the Avaldsnes Sand and Mangerud ( 1991a). Mangerud ( 199lb) correlated it with 1.. Forsstriim er rrl.: Initiation of’ the Last Glaciation BO ON KARMBY in Northern ISOTOPE -0RMATION -___ I Haugesund Diamicton 1207 Europe -.-.. STAGES c --___ Sa? I I 2? Deformed sediments ! 5b I 60 5C Sand Karmoy Diamicton FIG. 6. Sketch of the stratigraphy at Bo in southwestern Norway (from chronostratigraphy. the younger of two Early Weichselian interstadials (01 Substage 5a). This interpretation can be disputed by amino acid dating of the foraminifer Cibides Iohatulus, found in both the Avaldsnes Sand and the Torvastad Sand (Miller et al., 1983) yielding average aIle/lle ratios of 0.168 and 0.123, respectively. The variation of only O.&l5 is very small - a similar difference in the ratio is observed at Fjosanger during the last zone (I in Fig. 5) of the Eemian interglacial. The Torvastad Sand could thus represent an initial Early Weichselian stadial (01 Substage 5d), with no need for a hiatus between the Avaldsnes Sand and the Torvastad Sand. The Torvastad Sand is overlain by the Karmoy Diamicton (2.5 m thick), the lower part of which contains folded sandy beds attesting to a glacial origin. The till could also have been deposited during 01 SubstageSd when the ice sheet covered the area. The overlying Bo Sand is a laminated silt and fine sand with scattered dropstones and shells deposited during glacial retreat after deposition of the Karmoy Diamicton, as proposed by Andersen et ~1. (1983). Pollen spectra indicate an open cold climate vegetation: Artemisia 4%. 5d T”&Ydad + Avaldsnes Sand ~ Andersen it ccl., 1X$3), 5e and reinterpretation of’ the Sa1i.r 2%, Juniprrus 3%. Brtulu 20-300/o.and high NAP (30-550/o). The Bs Sand could represent a single interstadial (01 Substage5c), prior to a new advance. The last phase(b’ in Fig. 6) comprisesglaciomarine stratified silty clay with abundant dropstones and a molluscan fauna. including the high-arctic species Po~tl~~rrlia urcticu (Andersen et ui., 1983). This interstadial was most likely the first warming period (01 Substage SC) after the end of the Eemian interglacial. The Bo Sand is overlain by a glacial Haugesund Diamicton (5.5 m thick). The lower part of the diamicton (A21 consistsof folded beds of sand, silt and clay. They may representproglacial material depositedin front of an expanding ice sheet during the stadial (01 Substage Sb), and were subsequentlytectonisedbeneaththe ice sheet.A continuous IRD deposition in the Norwegian Sea from ratio Subctage Sb onwards tidewater (Baumann suggests et nl., 19%; that the ice Goldschmidt, margin 19%). was so that the area was glaciated until 13- I2 ka BP. In conclusion regarding the Early Weichselian in Norway. the ice sheet first expanded to the Bo region during 01 Substage Sd and deposited the Karmoy ‘e Reviews: Volume Diamicton (Fig. 6). During 01 Substage SC the region was again ice-free, but the Bo Sand records such a cold climate that the ice margin probably remained nearby. A readvance during 01 Substage 5b is recorded by the upper part of the Bo Sand. These observations support an interpretation that during the Early Weichselian stadials in Norway, the extent of the Scandinavian ice sheet exceeded that of the deglacial Younger Dryas stadial. INITIATION OF THE SCANDINAVIAN GLACIATION COMPARED WITH THE LAURENTIDE GLACIATION The Early Weichsehan/Wisconsinanhistories in Europe and North America can be comparedby applying the approach of this study of the Scandinavian ice sheet to explain observations made in the Laurentide ice sheet. Finland is located close to the centre of the Scandinavian ice sheet, comparable to the Hudson Bay Lowland in relation to the Laurentide ice sheet. These areasmay be expected to have similarities in their glacial record. According to reconstructions by Vincent and Prest (1987) and Clark ef al. (1993) the build-up of the Laurentide ice sheet may have taken place during 01s 5. The ice sheet grew to cover Keewatin. Labrador 16 and Baffin Island. but did not reach the near-maximum extent until 01s 4 (Fig. 7). The Hudson Bay Lowland is located at the centre of the former ice sheet and is also adjacent to the initial centres of ice sheetgrowth. Skinner ( 1973) proposed that organic and minerogenic sediments southwest of James Bay, the Missinaibi Formation, represent the entire last interglacial cycle in the area. The Lake Ontario basin and the St. Lawrence Lowland are critical areaswith regard to the extent of the ice sheet during the last glaciation (Fig. 7). Advance of the ice sheetinto the lowland generatedhigher lake levels in the Lake Ontario basin. The Scarborough Sandsindicate high delta surfaces during 01 Substage 5b (St-Onge, 1987; Eyles and Williams, 1993).at a time when ice blocked the St. Lawrence valley and deposited the Lewrard Till (Clark et CII., 1993). Clark (1992) suggestedthat if TL dates from sections of the Hudson Bay Lowland, which indicate an interval of deglaciation, marine incursion and isostatic recovery after the last interglacial, were too young, then initiation of glaciation would be recorded by the tills overlying the last interglacial Missinaibi beds (e.g. Long Spruce, Sachico, and Adam tills). The lowland could have remained icecovered throughout the last glaciation subsequentto 01 Substage Sd. as previously proposed by Andrews and Barry (1978). This is supported by ice flow patterns -- FIG. 7. Maximum extent of the Laurentide ice sheet (ca. 20 ka BP) and marginal positions during the Wisconsin glaciation at 9 ka BP (Dyke and Prest, 1987) and 50 ka BP (Dredge and Thorleifson, 1987: Hypothesis I). The conceptual model by Vincent and Prest (1987) suggests that the ice sheet initially grew as Labrador (L), Keewatin (K) and Baffin Island (B) domes. A later marginal position of the growth (G. after 13 ka of growth) is also shown. An ice-covered area during Early Wisconsinan is approximately delineated by the 9 ka BP isochrone, hut the configuration of the growing ice sheet may have resembled the one indicated by the line G. L. Forsstriim ef al.: Initiation of the Last Glaciation in Northern Europe interpreted to record a continuous growth-decay cycle around JamesBay (Veillette, 1995). In the Hudson Bay Lowland, sands and organic material occur between Wisconsinan tills, and basedon their stratigraphic position and on TL and amino acid dates most have been regarded as younger than the last interglacial (Mott and DiLabio. 1990; Wyatt, 1990: Dredge et al.. 1990; Thorleifson et al., 1992; Clark et al., 1993). According to some hypotheses (see Lamothe et (11.. 1992; Clark et al., 1993; Occhietti et cll., 1996). the ice sheet receded from the St. Lawrence Lowland during interstadial Substage Sa and deposited the St. Pierre sediments,a non-glacial fluvial and lacustrine unit. These problematic sediments include peats, the micro- and macrofossil composition of which indicate climates almost as mild as the present (Terasmae, 1958). The question of whether such a warm climate was possible after the last interglacial remains open. The authorssuggestthat an alternative model, similar to our Fennoscandianone (Punkari and Forsstriim. 1995). could explain the problematic sequencesin the Hudson Bay Lowland. Organic beds containing interglacial-type vegetation within Wisconsinansedimentscould constitute redeposited and partly mixed interglacial material (including the cool end of the interglacial). This could also apply to someinterglacial bedsregarded as older than the Sangamonian interglacial in the St. Lawrence Lowland (Dredge et al.. 1990). They could actually represent Sangamonian beds in situ, while beds lying in higher stratigraphic positions representmaterial of the sameage redeposited during glaciation but previously regarded as being ilz situ. Processesof redepositionof organic material in the Great Lakes region have been describedby Hicock and Dreimanis (1992b). Peripheral parts of the Hudson Bay Lowland could have been deglaciated later in 01 5 (Vincent and Prest, 1987). but subsequentlythe lowland probably was covered by ice until the final deglaciation. OTHER EVIDENCE FOR INITIATION GLACIATION OF Several other sourcesof information (Fig. 8) indicate growth of the ice sheet soon after the last interglacial. Deep-seaoxygen isotopevalues suggestincreasingglobal ice volumes since 01 Substage5e, and moderate values during the Early Weichselian interstadials (Martinson et ul., 1987: Imbrie et ~1.. 1992). The Summit (GRIP) ice core data indicate a general trend of progressively growing ice sheets(or decreasing temperatures) with intervening minor fluctuations (Fig. 8f). A drastic drop of s’s0 occurred at 1IS111 ka BP (01 Substage 5d); 01 Substage SC is represented by moderate values; the period 103-90 ka BP (01 Substage5b) shows a progressive decrease;and 01 Substage5a rises to moderate values similar to 01 Substage5c (Johnsenet al., 1995). Compilation of relative sea-levelrecords (Fig. 8a) from different sites by Lundberg and Ford (1994) shows a 1209 lowering of about 15 m at the end of the Eemian (Seidenkrantz and Knudsen, 1994), and a drop to about 70 m by 112 ka BP (SubstageSd). The records show that a major glacial build-up, corresponding to half of the maximum ice volume of later 01s 2, occurred soon after the interglacial. Subsequently, ice volumes decreased during the warm interstadials SC and Sa to l/7 of the maximum ice volume. If these values are simply applied to the Scandinavian ice sheet. most of Scandinavia, Finland and the Kola Peninsulawould have remained ice covered during the Early Weichselian fluctuations. However. these global records do not explain the geographical distribution of ice volume. Ice-rafted cletritusrecorded in the Norwegian Sea (Figs 8d, e) showsa minimum at the last interglacial Substage Se, risesto a moderatevalue at 01 Substage5b, is slightly lower at 80-70 ka BP. reachesits maximum at 5.5ka BP, decreasesduring OIS 3 and increasesagain at the last glacial maximum (Baumann et rrl., 1995; Goldschmidt. 1995). The major IRD peaks most likely represent significant changesin glacial dynamics, such asadvances and collapsesof marine-basedice margins. The amount of IRD was probably considerably lower while highvelocity ice streams. subject to effective basal melting and release of their englacial debris, remained in the Norwegian fjords. A similar situation explains the low IRD input from the present day ice sheetsof Greenland, West Antarctic (e.g. Alley et ol.. 1989) and Spitsbergen (Dowdeswell and Dowdeswell, 1989). The IRD data can be interpreted to show a general growing trend of the Scandinavian ice sheet during OIS 5, but the ice margin remainedin the coastal zone. During the transition of 01s Sa/4 the ice sheet expanded to the continental shelf causing an increase in IRD sedimentation. Lower IRD values during 01s 3 in the easternNorwegian-Greenland Sea may record locally stable or lessextensive ice sheets. The first Heinrich event (H6) of the Laurentide ice sheet occurred at the beginning of 01s 4 (66 ka BP), although minor events appear at OIS Sd and 5b (McManus et ml.. 1994). An extensive ice sheet reaching to the continental shelf is required to generate a collapse of vast ice massesto the sea(McAyeal, 1993). The first comparable IRD spike of the Scandinavian ice sheettook place later, at 54.3 ka BP (Baumannet [II., 1995). and can be related to the main glacial expansion to the continental shelf. However, a moderate IRD peak at 01 SubstageSb (93 ka BP) suggeststhat the first glacial advance beyond the coastal zone occurred then. These records are concluded to be consistent with an early inception of both the Scandinavian and Laurentide glaciations. Foraminiferal studies in northern Denmark show progressively decreasingtemperatures of the North Sea from Substage01 Sd to 01s 4 and a final change to full glacial conditions during the transition of 01s S/4 (Seidenkrantz and Knudsen, 1994). These results are consistent with terrestrial records indicating glaciation at about 75 ka BP in Denmark (Houmark-Nielsen, 1994). A compilation of biostratigraphic records from Central and Western Europe suggestsa temperature decreaseof l214 C from 01 SubstageSe to 5d. and 01 Substage SC 1210 Extent of the ice sheet (km) summit c IRI) Sea level (m) 30 w - 4o 50 - 60 A 70 Fg 80 o- 90 FIG. 8. Sea-level diagrams (a) by Lundbergand Ford (I 994) based on coral reefs and (b) and (c) by Shackleton (1987) based on deep-sea oxygen isotope variations and on coral reefs; records of ice rafted detritus (IRD) in the Norwegian Sea [(d) accumulationrates,&/cm’ ka, and(e) grain % of 125-500urn fraction] accordingto Baumann ef trl. ( 1995); (1‘) smoothed curve of oxygen isotope variations in the Summit ice core based on data by Johnsen et cl/. (19YS): (g) time-distance diagram showing the extent of the Scandinavian ice sheet trlong a transect from the initial ice-divide in northern Sweden via Estonia to Poland (this study): (h) the extent of a melted-bed /one beneath the outer 300 km of the margin. affected by subglacial erosion and dclhrmation of unconsolidated deposits (Punkari, I995 ). (Brorup) was 24°C colder than present (Lebret et (11.. 1994). Climate reconstructions made from La Grande Pile in France indicate relatively mild conditions during the Early Weichselian interstadials compared with OK 4-2 (Guiot et al.. 1992). However. the disappearance of forests during 01s 4 is probably caused by the local climate in front of the approaching ice sheets and mountain glaciers. Therefore the severity of the global climate deterioration of the Early Weichselian cannot easily be realised in this reconstruction. Several 2D and 3D climate-coupled models of the Scandinavian ice sheet predict fast spreadingof thin ice sheets about 118-l I5 ka BP (Marsiat, 1994). Summer temperature lowering of 6’C from the present values is neededto create ice sheetsin the Norwegian mountains. and a drop of 7°C causesexpansion of the Eurasian ice sheet to its maximum known limits (Sanberg and Oerlemans, 1983). A steady state is attained with a build-up time of 80-100 ka (Huybrechts and T’siobbel. 1995). Some models suggest that the minimum time required to create a full-grown Scandinavian ice sheet is 30-60 ka, without extreme forcing (Oerlemans, I98 1; Sanberg and Oerlemans, 1983; Lindstrom, 1990). However, calving and periodic collapses at marine margins may have dissipatedice more effectively than considered in some numerical models and so slowed glacial growth (Deblonde and Peltier, 199I; McAyeal, 1993). A major ice sheet can survive in a moderately warm climate because high surface elevations of the main ice body keepsthe ablation zone narrow. In most Scandinavian ice sheet models, initial fast growth takes place in the arctic regions, and especially in northwestern Russia. This may explain the drastic sealevel drop during 01 Substage Sd. Otherwise such a lowering would indicate ice cover south of the Gulf of Finland. Extensive glaciations in the arctic regions during Early Weichselian time have been suggestedfrom the northernmost Laurentide ice sheet (Clark et al., 1993) Greenland (Funder et (I/.. I99 I ) and Russia (Astakhov, 1992). DISCUSSION The biostratigraphic analysesof pre-Late Weichselian organic beds in northern Europe indicate that Early Weichselian interstadials (01 Substages 5c and 5a) recorded in the pollen spectra of Central and Western Europe are not seenin the terrestrial record from central parts of the glaciated area. The authors argue that earlier interpretations of such interstadial depositsare basedon L. ForsstrGm et al.: Initiation of the Last Glaciation absolute dating methods which have given too young ages for the interglacial samples. In addition, some pollen sequences recording colder climates than the present are incomplete, and their stratigraphic position suggeststhem to be slices of more extensive interglacial beds transported by ice. It should be kept in mind that palynology is not a dating method and the interpretation of vegetational sequencesbecomes more difficult the less species are involved, and especially if the sequencesare incomplete. Many redeposited interglacial beds are so well preserved (non-disintegrated surface peats, for example) that it is difficult to imagine that they could have been overridden by one (Early Weichselian) glacial advance, and then picked up and redeposited by another (Late Weichselian) ice sheet. The preservation of interglacial organic matter and its redeposition in Late Weichselian deposits suggest that the material remained in an englacial (or frozen subglacial) position after the first ice advance soon after the interglacial. This could mean that subglacial erosion and deformation were absent during the time between the growth and final deglaciation and that the area was covered by ice during almost the entire Weichselian. The inferred configuration of the Scandinavian ice sheet during growth (G in Fig. 9) is based on the 1211 in Northern Europe evaluation of older ice flow directions (Tanner, 1915: Punkari, 1995).Our resultssuggestthat an areadelineated approximately by the ice marginal position of the Younger Dryas (I 1 in Fig. 9) was covered by ice during the Early Weichselian interstadials, and remained glaciated until the last deglaciation period. This would be an area of ahnost twice the size for the Early Weichselian stadials suggestedby Andersen and Mangerud (1989). The lack of interglacial organic beds. either in sits or redeposited. in a zone a few hundred kilometres within and beyond the Younger Dryas moraines in Finland and Sweden may indicate not only destructive glacial activity during the deglaciation. but also glacial erosionduring the Early Weichselian marginal fluctuations in that area. Our reinterpretation of the data, both from Fennoscandia and North America, suggests that the last glacial histories of these areaswere quite similar. After the last interglacial. ice sheetscovered the FennoscandianShield and the Hudson Bay Lowland during 01 Substage Sd (Fig. 7). Ice sheetsprobably recededconsiderably during the warmer intervals of 01s 5 as revealed by the sea-level data (Shackleton, 1987; Lundberg and Ford, 39943,but the lack of in situ younger non-glacial horizons indicate that most of Finland and the Hudson Bay Lowland remained ice-covered (Fig. 8). Barents Sea FIG. 9. Maximum extent of the Scandinavian ice sheet and the marginal position during the Younger Dryas ( I I ka BP). The ice sheet initially grew in the northern (N) and southern (S) Scandinavian domes. A later marginal position inferred to represent the growth CC). based on older flow directions, is also shown. The distribution of preserved interglacial deposits and some stratigraphic sequences indicate that the ice margin fluctuated in the zone delineated by linex I I and G during the Early Weichselian inters,tadials. OIS 4 was characterised by a large expansion of the ice sheets, especially in the southern sector of the Scandinavian and Laurentide ice sheets. In North America, the ice sheet extended to Lake Erie (Hicock and Dreimanis, 1092a) or at least to the Toronto area (Eyles and Williams, 1992). In Europe, the ice sheet reached Denmark (Houmark-Nielsen, 1994) and northern Poland (Mojski, 1992). During the Early Weichselian. the Norwegian coast was probably occasionally ice free. similar to the coastal mountains of Greenland today. However, input of icerafted detritus (Baumann it al., 1995: Goldschmidt, 1995) shows that the ice sheet had entered the open sea during the Early Weichselian stadials, and during 01s 4, 3 and 2. Early Weichselian fluctuations would have been insignificant relative to the area of the Scandinavian ice sheet remaining on land, but may have been equivalent to the collapses (Heinrich events) of the marine margins of the Laurentide ice sheet (Bond et al., 1993; McAyeal, 1993). The situation in Norway may have been similar to the Cordilleran ice sheet. where mountain glaciers, which initiated the growth of the ice sheet had to be reduced to near their present dimensions before significant recession took place in the core area of the ice sheet (rulton, 199 I ). Mangerud et al. (1981b) defined the Alesund interstadial (40-30 ka BP) in the outermost part of western Norway, but an ice-free period of similar age has also been proposed from continental area>. Radiocarbon dating of sub-till organic fragments (including interglacial-type matter) in central Fennoscandia frequently gives finite Middle Weichselian and even younger ages (Andersen et LII., 1983; Hirvab, 1991; Olsen, 1988; Nenonen, 1995). Acceptance of these ages implies that almost the whole of Fennoscandia was ice-free only a few thousands of years before the last glacial maximum. This model should be regarded with caution and may be largely due to erroneous dating methods. It should be kept in mind that even when modern radiocarbon dating technologies are used, upper limits for successful dating are 35 ka for wood and 20 ka for shell samples (Dredge and Thorleifson. 1987; Vincent and Prest, 1987: Clark et al., 1993). prior to the glacial advance. and thus they contain cold climate microfossils even if the ice-free period in question was an interglacial (Mott and DiLabio, 1990; Punkari and Forsstriim, 1995 ). Many redeposited interglacial beds are so well preserved that the authors regard it as unlikely that they could have been overridden by more than one glacial advance. The preservation of interglacial organic matter and its redeposition in Late Weichselian deposits suggest that the material remained in an englacial (or frozen subglacial) position after the first ice advance soon after the interglacial. This could also mean that the central areas of the ice sheet did not reach the pressure melting point associated with destructive subglacial erosion and deformation during the time between its growth and final deglaciation (see Fig. Xh). and that the area was covered by ice during almost the entire Weichselian. The climate fluctuations of Early Weichselian time (01 Substages Sd, 5c, Sb and 5a) recorded in the deep-sea oxygen isotope stratigraphy and pollen spectra in Central and Western Europe are not seen in the terrestrial record in the central parts of the glaciated area. Our conclusion is that this is due to the expansion of the ice sheet into these areas soon after the Eemian interglacial. This ice mass probably attained much of its volume during the cool Early Weichselian interstadials and did not decay until it became unstable after the last glacial maximum. Further appreciation should be given to the possibility that this model may also explain some problematic sections described in the area of the Laurentide ice sheet. ACKNOWLEDGEMENTS We thank comments, manuscript. G. S. Boulron, P. li. Clark and D. Praep for discussions and IWO ;ulonymou\ reviewers for critically reading and the REFERENCES Aalto, M., interglacial Finland. Dormer, J.. Hirxas, beaver dam H. and Niemeli, J. (1989) at Vimpeli. Ostrobothnia. deposit Bulletirz of‘ the Grok)cgicul Sunv; of Finlmd An 348, I- 34. CONCLUSIONS Most existing reconstructions explaining the initiation of the last glaciation in Fennoscandia are affected by unreliable absolute dating methods (“C. TL, OSL. amino acid). Some sequences in Finland containing old organic matter, including pollen recording colder climates than the present, are incomplete and their stratigraphic position suggests them to be slices of more extensive beds transported by ice. Most organic beds beneath or within Weichselian deposits are interglacial (01 Substage Se). representing the climatic optimum or the cool period following it (earliest Early Weichselian time, 01 Substage 5d). Stiff, non-disintegrated peat blocks discovered in several sections are not necessarily interstadial deposits; they may as well be interpreted as deposits formed just Aalto. M., eroded Finland. Donner, interglacial Bulktirl J.. Nierneli. deposit J. and Tynni, at Vimpeli. of tkr Geological R. South (1983) An Bothnia, Survey of‘ Finlund 324. 133. Aario. R. and Forsstriim, Koillismaa and North Alley. R. B., Blankenship. R. (1989) Sedimentation L. (1979) Glacial stratigraphy of Kainuu, Finland. Fennia 157, 149. D. D.. Rooney, S. T. and Bentley. C. beneath ice shelves the view t‘rom ice stream B. Murirw G’~dogy Andersen, S. T. (1961) Vegetation Denmark in the Early Weichselian 85, and 101-t 20. its environment Glacial (last glacial). in Dan~nctrkv Gro1ogi.h Urtdc~r.wkrlsr 380, 189-X I Andersen. S. T. ( 1965) Interglaciarer ag interstadialer i Danmarks kvartaer. Mc,t/rlrl~~/.trr- ,fi-n I)c~~.sk Geologisk Fiirenittg 1.5. 4X6-506. Andersen,B. G. andMangerud.J. (1989)The last interglacialglacial 21-29. cycle in Fennoscandia. Qunrernav Intrmatiotzal 3/4. Andersen,B. G.. Sejrup.H. P. and Kirkhus, 0. (1983)Eemian andWeichselian deposits at Bc$ on Karmoy. SW Norway: a L. Forsstrom et a/.: Initiation of the Last Glaciation preliminary report. Norges Geologiske Under.sokelse 380, 189-201. Andrews, J. T. and Barry, R. G. (1978) Glacial inception and disintegration during the last glaciation. Annual Revie~+~sof Earth and Planetary Sciences 6, 205228. Astakhov, V. (1992) The last glaciation in West Siberia. Sveriges Geolagiska Under~sfikning Ser. Ca 81. 2 l-30. Atlas of Finland. 1987. Folio 131. Climate. Averdieck, F. R. (1967) Die Vegetationsentwicklung des EemInterglazials und der Frtihwiirn-Interstadiale von Odderude/ Schleswig-Holstein. Fundamenta 2, 101-125. Baumann, K.-H., Lackschewitz, K. S., Mangerud. J.. Spielhagen, R. F., Wolf-Welling, T. C. W., Henrich. R. and Kassens. H. (1995) Reflection of Scandinavian ice sheet fluctuations in Norwegian Sea sediments during the past lSO.000 years. Qtcaternary Researcl~ 43, I&- 197. Behre, K.-E. (1989) Biostratigraphy of the last glacial period in Europe. Quaternatys Science Reviews 8, 25544. Bond, G., Broecker, W., Johnsen, S., McManus. J., Labeyrie. L.. Jouzel, .I. and Bonani. G. (1993) Correlations between climate records from North Atlantic sediments and Greenland ice. Nature 365, 143-147. Causse, C. and Vincent. J.-S. (1989) Th-U disequilibrium dating of Middle and Late Pleistocene wood and shells from Banks and Victoria Islands, Arctic Canada. Canczdicm Jawm~l of Earth Sciencr.s 26. 27 18-2723. Clark, P. U. (19Y2) The last interglacial-glacial transition in North America: Introduction. In The Lust IntergiuciuGlacial Transition in North America, eds P. U. Clark and P. D. Lea. Geological Society cfAmrrica Special Paper 270. l-11. Clark, P. U., Clague, J. J., Curry. B. B., Dreimanix. A.. Hicock, S. R., Miller, G. H., Berger, G. W., Eyles, Eyles. Lamothe. M., Miller. B. B.. Mott. R. J., Oldale, R. N., Stea. R. R.. Szabo, J. P., Thorleifson, L. H. and Vincent. J.-S. (1993) Initiation and development of the Laurentide and Cordilleran ice sheets following the last interglaciation. Quaternur~s Sciences Reviews 12, 79-I 14. Deblonde, G. and Peltier, W. R. (1991) Simulations of continental ice sheet growth over the last glacial-interglacial cycle: experiments with a one-level seasonal energy balance model including realistic geography. Jotanal u/ Groph~sicul Research 96(DS), 9 189-92 15. Dormer, J. J. (1983) The identification of Eemian interglacial and Weichxelian interstadial deposits in Finland. A1zrrtrle.s Academine Scientiarum Fennictrr A III 136. l-38. Dowdeswell, J. A. and Dowdeswell, E. K. (1989) Debris in icebergs and rates of glaci-marine sedimentation: observation from Spitsbergen and a simple model. Jotlrnul of’Grolo,yy 97, 221-231. Dredge, L. A. and Thorleifson. L. H. (1987) The Middle Wisconsinan history of the Laurentide ice sheet. Glographir physique rf Quatemaire 41, 2 15-235. Dredge, L. A.. Morgan, A. V. and Nielsen, E. ( 1990) Sangamon and pre-Sangamon interglaciations in the Hudson Bay Lowlands of Manitoba. Geographic physique et Quuternaire 44, 319-336. Dreimanis, A. (1992) Transition from the Sangamon Interglaciation to the Wisconsin Glaciation along the southeastern margin of the Laurentide ice sheet. North America. In Start oj a Glacial. eds. G. J. Kukla and E. Went. NATO ASI Series 13, 225-25 I. Eriksson, B. (1993) The Eemian pollen stratigraphy and vegetational history of Ostrobothnia, Finland. Balletin of the Geological Survey of Finlund 372, 1-36. Eriksson. B., Griinlund. T. and Kujansuu. R. (1980) Interglasiaalikerrostuma Evijarvelki. Pohjanmaalla. Summary: An interglacial deposit at Evijarvi in the Pohjanmaa region, Finland. Geologi 32, 65-7 I. Eyles, N. and Williams. N. E. (1992) The sedimentary and biological record of the last interglacial-glacial cycle at Toronto, Canada. In The Last Interglncial-Gltrci(II Transition in Northern Europe 1213 in North America, eds P. U. Clark and P. D. Lea. Geological SocieQ qf America Special Paper 270, 119-137. Forxstriim, L. (1982) The Oulainen interglacial in Ostrobothnia, western Finland. Acta Uniwrsitatis Uuluensis, A 136, Geologica, 4, l-l 16. Forsstrom. L. (I 984) Eemian and Weichselian correlation problems in Finland. Bo~cd.s 13. 301-318. Forsxtriim, L. (1989) The Fennoscandian ice sheet in the last half million years. Nordia 23. 91-103. Forsxtrom, L. ( 1991) The Early Weichselian climate in Finland. Norsk Geologisk Tid.skrijt 71. 133--l 36. Forsstrom, L. ( 1995) The last glacial cycle (Weichselian) at the centre of the Fennoscandian glaciated area, evidence from Finland. Res Terrae, Ser. AII, Publications of the Department of Geology, University of Oulu. pp. l-34. Forsstrlim, L.. Aalto, M.. Eronen, M. and Gronlund, T. (1988) Stratigraphic evidence Ibr Eemian crustal movements and relative sea-level changes in eastern Fcnnoscandia. Palueogeography, Palaeoc~lin~utol~)g~, Pulaeorcolog~ 68, 3 17-335. Fulton, R. J. (I 99 I) A conceptual model for growth and decay of the Cordilleran ice sheet, Geogruphie physique et Qnaternaire 45, X1-286. Funder, S.. Hjort. C. and Kelly. M. ( 1991) Isotope stage 5 (13074 ka) in Greenland. a review. Quaternaty b~ternational lo12, 107-l 22. Gaigalas. A. ( 1994) On palaeogeography of Late Pleistocene in the Lithuania. Acta Uni\er.sitati.s Nicolai Copernici, Geografia XXVII, Nauki Matematyczno-Przyrodnicze, Zeszyt 92, 183-194. Gibhard. P., Forman. S.. Salomaa, R.. Alhonen, P., Jungner, H.. Peglar, S., Suksi, J. and Vuorinen. A. ( 1989) Late Pleistocene stratigraphy at Harrinkangas, Kauhajoki, western Finland. Anna1e.s Acudemiae Scirntiarnm Fennicae A III 150, I-36. Goldschmidt, P. M. (1995) Accumulation rates of coarsegrained terrigenous sediment in the Norwegian-Greenland Sea: signals of continental glaciation. Murine Geology 128. 137-151. Grichuk. V. P. (1961) Fossil floras as a paleontological basis for the stratigraphy of Quatemary deposits in the north-western part of the Russian Plain. Relief and stratigraphy of the Quaternary deposits in the north-western Russian Plain (in Russian). Nauka, Moscow. pp. 25-7 1. Gronlund, T. (1991) The diatom stratigraphy of the Eemian Baltic Sea on the basis of sediment discoveries in Ostrobothnia, Finland. Genlogicul Snrvex of Finland, Report a/ Investigation 102, I-26. Guiot, J., Reille, M., de Beaulieu. J. L. and Pons, A. (1992) Calibration of the climatic signal in a new pollen sequence from La Grande Pile. Climute Dwamics 6, 259-264. Heijnis. H. ( 1992) Uranium/Thorium dating of Late Pleistocene peat deposits in N. W. Europe. Dissertation, Rijksuniversiteit Groningen. Hicock. S. R. and Dreimanis. A. (1992a) Sunnybrook Drift in the Toronto area. Canada: Re-investigation and re-interpretation. In The Last Interglacial-Glacial Transition in North America, eds P. U. Clark and P. D. Lea. Geological Societ!, of Americu Special Paper 270. 139-I 6 I Hicock, S. R. and Dreimanis, A. (1992) Deformation till in the Great Lakes region: implications for rapid flow along the south-central margin of the Laurentide ice sheet. Canadian Juwnul of’ Eurth Sciences 29, 1565- 1579. Hirvas. H.. Alfthan, A.. Pulkkinen, E.. Puranen. R. and Tynni, R. (1977) Raportti malminetsinttia palvelevasta maaperltutkimuksesta Pohjois-Suomessa 1972-l 976. Summary: A report on glacial drift investigations for ore prospecting purposes in northern Finland 1972-l 976. Geologinen tutkimnsmitos, Tntkimnsraportti 19. l-53. Hirvas, H. (1991) Pleistocene stratigraphy in Finnish Lapland. Bnlletin of the Geological Snrve,y of Finland 354, I-123. Houmark-Nielsen. M. (19Y4) Late Pleistocene stratigraphy, glaciation chronology and Middle Weichselian environmental history from Klintholm, Mon. Denmark. Bdetin o/ the Geological Soriq of Denmark 41, 18 I-202. Huybrechts, P. and T’siobbel, S. (1995) Thermomechanical modelling of Northern Hemisphere ice sheets with a twolevel mass-balance parameterization. Attmls of’ G/wio/qq 21. 111-116. Hiitt. G.. Jungner. H.. Kujansuu, R. and Saarnisto. M. (1993) OSL and TL dating of buried podsols and overlying sands in Ostrobothnia, western Finland. .IoI/~~zNI of Q~~ut~rt~u~:\ S&we 8, 125-I 32. Hyviirinen. H. (1976) Flandrian pollen deposition rates and treeline history in northern Fennoscandia. Boreas 5. i63- 175. Johnsen, S. J.. Clausen, H. B.. Dansgaard. W.. Gundestrup. N., Hammer, C. U. and Tauber, H. ( 1995) The Eem stable isotope record along the GRIP ice core and its interpretation. Quufernury Research 43. I 17-l 24. Jungner, H. (1987) Thermoluminescence dating of sediments from Oulainen and Vimpeli. Oxtrobothnia. Finland. Rorcjtrv 16, 23 l-236. Kaufman, D. S. and Sejrup, H. P. (1YYS) Isoleucine epimeriration in the high-molecular-weight fraction of Pleistocene Ardca. Quarertray ~~oc~i~~ottoio~~~ (Quutc~rnut-y .%irwe.r Reviews) 14, 337-350. Korpela, K. (1969) Die Weichsel-Eiszeil und ihr Interstadial in Perlpohjola (n8rdliches Nordfinnland) im Licht van submor6nen Sedimenten. An/~&s Acudetttiue .S~~irtzfiarunt Fottnicue A III 99. I-108. Krupinski, K. M. ( 1995) Analiza pylkowa osadow interglacjalu eemskiego z Proniewicz na Podlasiu. P~~,p/nr/ Genlogicy? 43(7), 581-585. Kulling. 0. (1945) Om fynd av mammut vid Pilgrimstad i Jltntland. Svrri~gr.s Geologiska Undursijktting C 473, I-61 Lagerbtick, R. and Robertsson, A.-M. (1088) Kettle holes ~ stratigraphical archives for Weichsclian geology and palacoenvironment in northernmost Sweden. Borcrrs 17, 43Y-468. Lamothe. M., Parent, M. and Shilts, W. W. ( 1992) Sangamonian and early Wisconsinan events in the St. Lawrence Lowland and Appalachians of southern Quebec. Canada. In T/w Ltrxr l~~t~r,~i~~c.ial-Gl~t~i~t/ Tran.rifion in Norrh Amct~ic~o. eds P. II. Clark and P. D. Lea. &o/ogictr/ Soc.ic~r\‘(!f’ Atttet-icrt Spc~~itri Paper 270, 17 l- 184. Lebret. P., Dupas. A., Clet. M.. Coutard, J.-P.. Lautridou. J.-P.. Courbouleix, S., Garcin. M., Levy. Levy and Van VlietLanoe, B. (1994) Modelling of permafrost thickness during the late glacial stage in France: preliminary results. Cutwciitrt~ Jourt~u/ of Earrh Scirtwc.\31, 9SY-Y68. Liivrand. E. ( 199 I ) Biostt-ctriXrn/,h~ of’ rhc Plei.sroc~enr Drpt.sitv in Estmiu and C‘orrdariott.s it7 rlze Btr//ic Regiot~. Department of Quaternary Research. University 01‘Stockholtn. Report, 19. pp. l-114. Lindstrom. D. R. (1990) The Eurasian ice sheet I’ormation and collapse resulting from nalural atmospheric CO2 concentration variations. Puleoc,rclnoRru/~h~ 5. 207.-227. Ljungner, E. (1949) East-west balance of the Quaternary ice caps in Patagonia and Scandinavia. Btriierin of’ Geo/o,+c,tr/ ImriturcJ, Uppsulu 33. I l-96. Lundberg, J. and Ford, D. C. ( 1994) Late Pleistocene sea level change in the Bahamas from mass spectrometric U-series dating of submerged speleothem. Quctterttrrr~~ Scicttce ReIGrrz~.s13. l-14. Lundqvist, J. (1967) Submor5na sediment i Jiimtlanda kin. .Sl~eri,y.s Grologi,ska Uttder.siiknin~~ C 618. I-267. Lundqvist, J. (1992) Glacial stratigraphy in Sweden, Gcwiogic~rtl Sur\,eJ of Finlard, SpYal Pqwr 5. 43-59. Lundqvist, J. and Miller, U. (1992) Weich\elian stratigraphy and glaciations in the Tasj(i-Hoting area. Central Sweden, S\,erijy.v Groingiska Lttrrlet-.siikr,in,g C 826. I-31. Lundqvist, J. and Mook. W. G. (1981 ) Finite date of the Jiitntland Interstadial. Bowus 10. 133-I 35. Mangerud. J.. Sonstegaard, E., Sejrup, H.-P. and Haldorsen. S. (198 I) A continuous Eemian-Early Weichselian sequence containing pollen and marine fossils at Fjasanger, western Norway. Bowav 10, 137-208. Mangerud, J., Gulliksen. S.. Larsen. E.. Longva. 0.. Miller, G. H.. Sejrup. H.-P. and Senstegaard, E. (198 I) A Middle Weichselian ice-free period in Western Norway: the ,&esund Interstadial. Botwts 10, 447462. Mangerud, J. (1991 a) The last ice age in Scandinavia. In LCrrp Qtt(t~vrmtty Strutigraph!~ itt the Nordic Court tries 150 0(X/S 000 B.P., eds B. G. Andersen and L. -K. Kiinigsson. Srriar 34. I s-30. Mangerud. J. (199 I h) The Scandinavian ice sheet through the last interglacial/glacial cycle. In KiirnLiR~.sc,llic,itrlichr Prohlrr?zc~&r lecrrt~ 130,000 .Itrhre,ed. B. Frenzel. pp. 307-330. G. Fischer. Stuttgart. Marsiat, I. ( IYY4) Simulation of the Northern Hemisphere continental ice sheets over the last glacial-interglacial cycle: experimenls with ;I latitttde-longitude \,ertically integrated ice sheet model coupled to a ronally averaged climate model. Ptr/aroc/it~~trrc~.s 1. SY-YX. Martinson. D. G., Pisias, N. G.. Hays. J. D.. Imbrie. J., Moore, T. C. and Shackleton. N. J. ( 1987) Age dating and orbital theory of the ice age\: development of a high resolution 0 to 300,000-year chronostratigraphy. Quarrtnrrr~~ Ro.search 27. I-2Y. McAyeal, D. R. ( 1993) Binge/purge oscillations of the Laurentide ice sheet a\ a cause of the North Atlantic’s Heinrich events. Pa/c~ot~rtr,~~~g~~ri~/7~ 8. 775-783. McManus. J. F.. Bond. G. C.. Broecker, W. S., Johnsen. S.. Laheyrie, L. and Higgins, S. ( IY94) High-resolution climate records from the North Atlantic during the last interglacial. Nutuw 371. 336. Mejdahl, V. ( 1992) Thermoluminescence dating of samples from Jameson Land, East Greenland. In Wcichselian uttd Holowue Gltrcitrl uttrl Murirw Hisroy of’ Eust Svalhard. Pwlitninuu\ Report ott the PONAM Fielrl,~wX itt 1991, eds. P. Miiller, C. Hjort and 0. Ing6lfsson. LUNDQUA Reporr 35, Lund. pp. 21 I-214. Miller. G. H. and Hare, P. E. ( 1980) Amino acid geochronology; integrity of the carbonate matrix of mollusk fossils. In Biocitemi.srty~ ofAtt7it70 Ac.il1.c.eds. P. E. Hare. T. C. Hoering and K. King, pp. 315434. John Wiley, New York. Miller. G. H.. Sejrup. H. P., Mangerud, J. and Andersen, B. G. ( 1983) Amino acid rulios in Quaternary molluscs and foraminifera iiom western Norway: Correlation. geochronoogy and palaeotemperature estimates. Bortws 12. 107-12-t. Mqjski, J. E. (1992) Vistulian stratigraphy and TL dates in Poland. Slx~rt’gcs Gw/ogi.rktr Uttrkeriikning Ser. Ca 81, I YS200. Mott, R. J. and DiLabio. N. W. I IYYO) Paleoecology of organic deposits of probable last interglacial age in northern Ontario. GBo,yrqd7ic, pii~.s~cfut~ cl ~uc4temrrire 44, 3OY- 3 IX. Miirner. N.-A. ( 19X1) Weichselian chronostratigraphy and correlations. Borrus 10, 46.?p470. Nenonen. K. i 1995) Pleistocene stratigraphy and reference sections in sourhern and western Finland. Academic dissertation. Geological Survey o(‘ Finland. Regional office for MidFinland, l-94. Occhietti, S.. Balescu. S.. Lamothe, M., Clet, M., Cronin, T., Ferland. P. and Pichel, P. ( IYY6) Late stage 5 glacio-isostatic sea in the St. Lawrence Valley, Canada and United States. Q~~t~w7q~ Rctetrtd7 45. 12% 137. Oerlernans, J. (IYXI) Modelling of Pleistocene European ice sheets: some experiments with simple mass-balance parameterizations. Qttuternar\. Rcwttrch 15. 77-85. Olsen. L. (1988) Stadials and interstadials during the Weichsel glaciation on Finnmark\vidda. northern Norway. Borrus 17. 517-539. Punkari, M. (1984) The relations between glacial dynamics and tills in the eastern part of the Baltic Shield. Strittr 20. 49-S4. Punkari. M. ( 19X8) Vanhaa orgaanista ainesta Harrinkankaan hrjussa. Suttmttrr~: Old organic matter in the esker of L. Forsstriim er rrl.: lnitiation of the Last Glaciation and. northern Ontario. G&graphic) physiqw et Quatrrnrriw 44, 389-393. Punkari, M. (1991) Old organic matter inside glacial deposits in Finland: sedimentation models. In L&r Quuterrruv Strntigraph? in the Nordic Countries 150000-15 000 B. P.. cds. B. G. Andersen and L.-K. Kiinigsxon. Striw 34, 77-83. Punkari, M. (1995) Glacial tlow systems in the zone of confluence between the Scandinavian and Novaya Zcmlya ice sheets. Quatunwr;~ Science Re\~ien~.s14. 589-603. Punkari, M. and Forsstriim. L. (1995) Organic remains in Finnish subglacial sediments. Qutmrrw~ Rr.tctrrch 43. 4 I& 42s. Robertsson. A.-M. (197 I) Pollen analytical investigation of the Levelniemi sediments. In The /ntc~r,y/acict/ lk/msit.s rlt the Lrl~iiniemi Miw, S~‘~//,/“~\‘“LIIW, Snwiish Ltr/d~mcl. cd. J. LundyvJist. Swriges Gvologidtr Uder.siikr?ir~g 658. X2- 10.3. Robertsson, A.-M. (198X) A Rei/~llr.stiRcltio/t t!/’ the Hin.ttrtrtigmphy of the Till-Cm~rrrrl Sedirwnt.s rrt Pilgrirnstrrtl md ojc>. Cerurd .Su~&/r. University of Stockholm, Department of Quaternary Research. Report 10, pp. I-53. Robertsson. A.-M. and Ambrosiani. K. G. (1992) The Pleistocene in Sweden - a review of research, 1960-1990. S,,erigr.s Geologkktr Umkrsiikrting Ser. Ca 81. 299-306. Salonen. V.-P. ( 1986) Glacial transport distance distributions of surface boulders in Finland. Bulletin of‘ the Gco/o,~ic~r/ Sur\‘cy of Finlrd 338, I-51. Sanberg, J. A. M. and Oerlemans, .I. (1983) Modelling of Pleistocene European ice sheets: the effect of upslopc precipitation. Geo/o,yir en Mijdmr~~~ 62, 267-173. Seidenkrantz, M.-S. and Knudsen, K. L. (1993) Marine high resolution records of the last interglacial in northwest Europe: a review. G?ogruphir ph~xiyur ct Quutenrtrirc~ 48. 157-l 68. Sejrup. H. P. (1987) Molluscan and foraminiferal biostratigraphy of an Eemian-Early Weichselian section on Karmoy. southwestern Norway. Bowas 16, 2742. Shackleton, N. .I. (1987) Oxygen isotopes, ice volume and sea level. QuatPrtmy Scimw Rwie~s 6. 183-l 90. in Northern Europe 1115 Skinner. R. G. (1973) Quatcrnary xtratigraphy of the Moose River basin, Ontario. Hdktin of' the Geologic~cd SLI~YJ~ oj Cunadu 225. I-17. St-Onge. D. A. (1987) The Sangamonian Stage and the Laurentide ice sheet. C;Bo,qrophi(, ph~.riyuc~ et Quuter~wir~ 41. I X9-198. Tanner. V. ( I91 5) Studier iifver hv,artarsystemet i Fennoskandias nordliga delar. Frrrt~irr 36. 1-X IS. Terasmac, J. ( 1958) Contributions to Canadian palynology. Huktir~ of'tlw Grolo~ic~rr/ S'~~~IYJJ~ of' Cruxrrlu 46. l-28. Thorleifson. L. H.. Wyatt. P. H.. Shilts. W. W. and Nielsen, E. ( I992) Hudson Bay Lowland Quaternary xtratigraphy: Ev*idencc for Early Wisconsinan glaciation centered in Quebec. In The Lust Ifltc,,;S/trr.itrl-(;/rr(,itr/ Trmsitiort in North .Arwric~tr. cds P. U. Clarh and P. D. Lea. Gcn/o,gicrr/ .Soc~ic~t~ of Amrrit~rr Sprc,irrl Prrper 270. 207-23 I. Vaarama. A. and Valanne. T. ( I973 I On the taxonomy, biology and origin of Brtultr tor’tuow I.ebed. R(,ports (!f Kp~w Srrhtrr(~tic~Resetrrch Sfntiml 10. 70-X4. Vrillette. J. J. (1995) New evidence for northwestvvard glacial ice flow. James Bay region, Quchec. Geologicd Sury of C’ur~trrlo, Currettt RP.scor(,ll IYY.i-(‘. pp. 249-258. Vincent. J.-S. and Prest, V. K. ( 19x7) The early Wisconsinan history of the Laurentide ice sheet. GPogmphie /~/~~sic/rr~ et Qurrtrmtrire 41. 109-2 13. West, R. G. ( 1957) Interglacial deposits at Bobbitshole. Ipswich. P/ti/~l.so/~hic,tr/ Trtrr?srrc.tiorl s of’ t/w Roxrrl Society r~f’hurlrm R 241. I ~-3 I Woillard. G. (1978) La Grande Pile peat bog: a continuous pollen record for the last 140.000 years. Qutrtcrntr~~y Rewtrrch 9. I-21. Wyatt, P. H. ( 1990) Amino acid evidence indicating two or more ages of pre-Holocene non-glacial deposits in Hudson Bay Lowland. northern Ontario. G@o,~rtrphic ph~.sique et Quoterrltrire 44. 389-393. Zagwijn, W. H. (1983) Sea-level changes in the Netherlands during the Eemian. Grologic~ CII Mijnhou~~ 62. 437-450.