The last interglacial-glacial period on spitsbergen, Svalbard

The last interglacial-glacial period on spitsbergen, Svalbard

Quaternary Science Reviews, Vol. l 1, pp. 633-664, 1992. 0277-3791/92 $15.00 © 1992 Pergamon Press Ltd Printed in Great Britain. All rights reserved...

4MB Sizes 0 Downloads 14 Views

Quaternary Science Reviews, Vol. l 1, pp. 633-664, 1992.

0277-3791/92 $15.00 © 1992 Pergamon Press Ltd

Printed in Great Britain. All rights reserved.




Jan M a n g e r u d and John Inge Svendsen* University of Bergen, Department of Geology, Section B, Allegt. 41, N-5007 Bergen, Norway

The glaciation history of Svalbard (78°N) and the NW Barents Sea is reconstructed for the last 130 ka, based on studies of sediments exposed in coastal cliffs at the head of Isfjorden. Four different till beds separated by marine sediments are recognized. The lowest marine formation, containing Mytilus edulis, reflects warmer conditions than at present, and is correlated with the last interglacial, the Eemian of Europe and Oxygen Isotope Substage 5e in the deep sea. The post-Eemian tills are inferred to represent major glaciations around 110 ka BP, 75-50 ka BP, and 25-10 ka BP. During the intervening intervals the glaciers on Svalbard were not significantly larger than at present and the NW Barents Sea was probably ice-free. The ice-free periods, named Phantomodden and Kapp Ekholm interstadials, lasted from about 110 to 75 and from 50 to 25 ka BP respectively. The marine fauna from both these interstadials indicate seasonally ice free conditions. The ages of the recorded glaciations coincide with, or are slightly younger than, periods with insolation minima, which at this latitude is determined by a low tilt of the Earth's axis. Thus we postulate that the Quaternary glaciations of Svalbard were driven by orbital variations with the 41 ka tilt period, in contrast to the lower-latitude glaciations of Scandinavia that were partly driven by the precession cycle with a periodicity of around 23 ka.

INTRODUCTION Svalbard, reaching more than 80°N, is the northernmost land along the eastern seaboard of the North Atlantic Ocean (Fig. 1). During the Quaternary, Svalbard and the shallow Barents Sea were repeatedly covered by one of the northernmost ice sheets on earth (Denton and Hughes, 1981). Approximately 60% of the land area is currently covered by glaciers (Fig. 2). The Svalbard archipelago presently-experiences an anomalously mild climate compared to its latitude, due to northward heat transport of the Atlantic Current (Fig. 1). The growth and decay of the great Quaternary ice sheets were forced by perturbations of the Earth's orbital parameters (Berger et al., 1984). This astronomical - - or Milankovitch - - forcing varies strongly with latitude. Thus the glacial history of Svalbard, situated at such a high latitude, will be an important element in the interpretation of the Earth's response to the external forcing. The main purpose of the investigation presented in this paper was to study the glacial history of the last interglacial-glacial (Eemian-Weichselian) cycle on Svalbard. We present results from stratigraphical studies of coastal cliffs along Billefjorden, an inner branch of Isfjorden, which cut into the central part of the main island, Spitsbergen (Fig. 2). The main locality consists of a series of profiles at Kapp Ekholm, but some additional observations from Nidedalen are also presented (Fig. 3). Kapp Ekholm is the only known section in the central part of the Svalbard archipelago that has several till beds interlayered with marine sediment. *Present address: University of Bergen, Centre for Studies of Environment and Resources, HiB-ThormChlensgt. 55, N-5020 Bergen, Norway.

This section is located only 14 km from the fjord head, which is occupied by the large tide water glacier Nordenski61dbreen. This implies that during periods when Kapp Ekholm was ice free, glaciers on Spitsbergen were not much larger than at present. Indirectly, the site has also monitored major glaciations of the NW Barents Sea by means of high relative sea levels during degiaciation phases, due to the glacio-isostatic depression of the archipelago. Kapp Ekholm is therefore a key site in identifying periods when central Svalbard and the Barents Sea were deglaciated. The Kapp Ekholm section has been described previously by Lavrushin (1967, 1969), Boulton and Rhodes (1974), Boulton (1979) and Troitsky et al. (1979). Mangerud and Salvigsen (1984) pointed out that some of the earlier descriptions and dates from this section are inconsistent. In order to clarify these conflicting observations a more detailed description of the lithostratigraphy along the entire exposure is presented in this study. This reinvestigation leads us to significant new interpretations of the site. Several other sites with sediments of EemianWeichselian age have been described from Svalbard (Boulton, 1979; Troitsky et al., 1979; Miller et al., 1989; L0nne and Mangerud, 1991; Lindner and Marks, 1991; Mangerud et al., 1992; Landvik et al., 1992). It has been a common view that only one large glacial advance occurred during the Weichselian. For example, Miller et al. (1989) and Larsen et al. (1992) concluded that only one glaciation, dated to > 80 ka BP, reached the shelf off western Svalbard after the Eemian. Recently, however, Mangerud et al. (1992) and Svendsen et al. (1992) have demonstrated that an ice sheet also extended seaward of Svalbard during the Late Weichselian. A major problem associated with Quaternary studies in the high Arctic environments of Svalbard is



1. Mangerud and J.I. Svendscn

;~-.L~eenal nd__~,_~ -


~",.. Norwegian~a /



,Biernoya Barents / Sea





co S


FIG. 1. An oblique north polar map projection, showing the high latitude location of Svalbard. The two main surface currents flowing in and out of the Norwegian Sea are schematically shown. The minimum sea-ice extent (September) based on a 75% concentration limit is shaded (Hibler, 1989).

that there is no well-suited method available for accurate correlation or age assessment of sediments older than the range (40-50 ka) of radiocarbon dating. This implies that reconstructing the glacial history for the pre-Late Weichselian period by combining several localities is problematic. The advantage of the Kapp Ekholm section is that more Weichselian depositional events occur in stratigraphic superposition than at any other known site on Svalbard (Fig. 6). Thus a more or less complete glacial history for the Weichselian can be reconstructed from this single site (Fig. 22). METHOD

Location and Surveying of the Sections The Kapp Ekholm sections are situated in the bay between Kapp Ekholm and Phantomodden in Billefjorden, 1 km south of the present delta from Mathisondalen (Figs 3 and 4). The sediments are exposed in up to 30 m high coastal cliffs which are dissected by broad gullies, leaving a series of isolated sections. The studied sections are numbered I to VI from south to north (Fig. 5). We measured horizontal distances along the foot of the sections with a tape measure (Figs 4 and 5). Most observations are referred to the nearest 10 m.

For elevations we used the approximate mean tide level as the zero point. The top of the present day beach is around 1.5 m above our reference level along the sections. Among the different exposures at Kapp Ekholm, Section II was studied in most detail. This section was levelled and marked every 5 m in the outcrop. Vertical distances between marks were measured with a ruler. The other sections were less precisely measured, but in all sections at least a few points were levelled. Heights above the present sea level marked on the photos (Figs 9, 12, 13 and 14) were all levelled. Errors for elevations given in the text are in most cases less than 1 m. The Kapp Ekholm sections were investigated for five weeks during the summer of 1988. Only two days were spent studying the section at Nidedalen (Figs 3 and 19).

Radiocarbon Dating Samples were dated at two laboratories. At the Trondheim Laboratory for Radiological Dating (prefix T- on samples) dating is performed by proportional counting, using CO2 gas, under the supervision of R. Nydal and S. Gulliksen. Small samples were submitted to the T. Swedberg Laboratory, Uppsala University (prefix Ua-), for accelerator mass spectrometry (AMS) dating, supervised by G. Possnert. For samples with


Interglacial-Glacial Period on Spitsbergen 9*












30* t





300 ~'~ 200 80"



79* 0


Karls Land


78 ° Linnddalen 0'




,o ~







J 18"

I 21"

I 24*






100km l



FIG. 2. Map of Svalbard. Glaciers are white, ice free land areas are shaded. Contour lines show the equilibrium line altitude on present day glaciers, according to O. Liest¢l (cited from Kristiansen and Sollid, 1987). Figure 3 is shown by the box near the head of Isfjorden.

prefix TUa- the target was prepared in Trondheim, and the AMS measurements performed in Uppsala. All finite samples are reported as recommended by Stuiver and Polach (1977), including a correction for isotopic fractionation to - 2 5 %0 6~3C on the PDB scale (Tables 1-4), A reservoir age of 440 years is subtracted for all samples that have obtained their carbon from sea water (shells, seaweed, whalebones, etc.). This is a standardized value used by the Trondheim laboratory for the coasts of Norway including Svalbard, and is based on measurements of many preindustrial shells from these waters (Mangerud and Gulliksen, 1975). For samples with activity close to the background radiation ('non-finite old' samples) we have given some more details (as recommended by Stuiver and Polach, 1977) in order to evaluate which of the dates are finite (Tables 2 and 3). Generally, samples that are of nonfinite age on the two standard deviation criteria should be cited as such.

Thermoluminescence (TL) and Optically Stimulated Luminescence (OSL) Dating All samples are dated by V. Mejdahl at the Nordic Laboratory for Luminescence Dating, Ris~, Denmark, using sand sized feldspars (Mejdahl, 1985, 1986). The ages for TL dates are calculated by using the plateau method (Mejdahl, 1988), and assuming a water content of 14% (given as weight of water relative to weight of minerogenic matter). This was found to be a mean water content of the samples when saturated with water. As indicated for some of the samples (Table 5), the TL age would be considerably higher if one postulated that they had been completely zeroed (bleached) during deposition. Thus, the age assessments are entirely dependent of the validity of the plateau method. As mentioned above, The Nordic Laboratory use the sand fraction method, whereas Forman and Ennis (1992) in their methodological study of TL dating of sediments from Svalbard used the finegrained technique. The OSL dating (Aitken, 1992)


J. Mangerud and J.l. Svendsen




Kapp sectic ~ 0


Phantomodden 2




FIG. 3. The inner part of Billefjorden (from map sheet Bille]}orden 1 : 100 000, Norsk Polarinstitutt, Oslo, 1984). Contour intervals are 50 m, with an additional contour at 25 m a.s.l. Glaciers are shown in white. Pyramiden is a Russian coal-mining town. The location of the sections south of Kapp Ekholm and the section near Nidedalen are marked.

followed the procedures given by B~tter-Jensen et al. (1991). We collected samples for TL and OSL dating from sand lenses or sand matrix from gravel foresets. It was assumed that these sand grains had been washed back and forth on the beach, and therefore bleached by daylight, before they were deposited in the foresets. We postulated that these grains had been exposed to light for longer periods than sand particles in the sand or silt formations. However, two OSL dates from the present beach yielded ages of some few thousand years (Table 5), and two out of three Early Holocene samples yielded OSL ages that were a few thousand years too old. This shows that the collected sediments at Kapp Ekholm are not ideal for T L or OSL dating. Individual dates on samples from the same formation may vary by more than 50 ka. We therefore obtained several dates from each formation (Table 5), and, when assessing the age, outliers were omitted. For example, for Formation F six samples yielded ages between 38 + 5 and 59 _+ 3 whereas two samples that yielded ages of 104 and 80 ka were disregarded. Amino Acid Diagenesis Amino acid diagenesis measured in the protein matrix of molluscs (Miller and Brigham-Grette, 1989) was used for three purposes: (1) To correlate between

individual sections. (2) For estimating the duration of glacial ice cover and/or inundation of the site by the sea. (3) For a first order age estimate of samples of nonfinite radiocarbon age. All samples were analyzed at the Bergen laboratory (prefix BAL-) under the supervision of H. P. Sejrup, according to methods described by Miller and Mangerud (1986). In this paper we mainly use the epimerization of isoleucine to alloisoleucine for the total fraction, expressed as D/L ratios (by some labelled aIle/ lie); the D/L ratios measured for the free fraction are also given for most samples (Tables 6-11). All reported D/L ratios are from the species Mya truncata or Hiatella arctica, which have approximately similar epimerization rates (Miller, 1982). However, recently it was found that the epimerization for Mya is around 15% slower than for Hiatella in the initial phases (D/L for total fraction < 0.1) (D. Kaufmann and G. H. Miller, oral commun., 1991). This slightly different reaction rate is compatible with the ratios we obtained for the two species from samples of the same age. For paleotemperature estimates we have used a D/L ratio of 0.011 for living shells, and the Arrhenius parameters 28.1 kcal/mol for the activation energy and 16.45 for the intercept (Miller, 1985). It should be noted that the epimerization reaction is extremely slow in this cold arctic environment, and the resolution is

Interglacial-Glacial Period on Spitsbergen


FIG. 4. Vertical air photo of the study area at Kapp Ekholm. The individual sections are marked with roman numerals. The horizontal distances along the shore are given in metres, compare with Fig. 5. To the north-east of the sections are Holocene terraces, up to 90 m a.s.l. Strike and dip, that indicate they were deposited by long-shore drift, are shown for foresets in some of these terraces. The ornamented lines show karst depressions that are younger than the terraces (Salvigsen et al., 1983). The present fan-delta deposited by the river from Mathisondalen is seen to the north of the hut. Photo: Norsk Polarinstitutt $61 2958 (August, 1961).

therefore low in both estimates.

age and


S E D I M E N T FACIES The entire section at K a p p E k h o l m consists of a few sediment facies in a typical deglaciation sequence: tillmarine silt-sand-gravel. This sequence is stratigraphically repeated (Fig. 6).

Grey Diamicton (Till) These are massive grey diamictons which occur as sheet formed sediment bodies, normally 0.5-3 m thick. The lower contact of each diamicton is an erosional unconformity. The diamictons are matrix supported, non-sorted, and characterized by a high content of stones and boulders of different sizes. In contrast to the other facies present in the sections, glacially striated stones are c o m m o n . A few abraded shell fragments occur occasionally. The diamictons are compacted, and in most sections they protrude as low vertical cliffs from the m o r e gentle slopes of sediments at the angle of repose. All the grey diamicton facies look similar, except for varying amounts of inclusions of underlying sediments. We a t t e m p t e d to distinguish the different diamictons by silt, clay and carbonate content of their matrix, but without success. The CaCO3 content is very high (around 70%) in all of the diamicton units,

indicating that the local Permian limestones are the main source. These greyish diamictons are all confidently interpreted and referred to as basal tills. We are in full a g r e e m e n t with Lavrushin (1967, 1969) and Boulton (1979) on this interpretation.

Brown Diamicton On top of the grey tills at some sites, lenses, or a thin sheet (less than 30 cm) of a silty, reddish brown diamicton occurs. The texture varies along the sections. In the field descriptions we did not normally distinguish between this diamicton and a reddish brown silt that occurs at the same stratigraphic level. They were both registered as a reddish brown, transitional zone between the grey till and the grey silt above. The colour is caused by red sandstones. The brown diamicton was not studied in detail, and the genesis is not clear. At some stratigraphic levels it may form the uppermost part of the grey till, but in most levels it probably represents the base of the marine sediments. Silt This facies is c o m m o n l y massive, but in a few instances it is laminated in the lower part, just above the tills. It contains varying amounts of clay and sand, generally it becomes m o r e coarse grained upwards. Some stones of varying sizes occur, often covered by a crust of marine carbonate algae. Paired molluscs,

FIG. 5. Mosaic photo of the Kapp Ekholm sections. The valley to the left is Mathiesondalen, compare with Figs 3 and 4. Section numbers and horizontal scale (m)~ as employed in the text, are also shown on the photo.




Interglacial-Glacial Period on Spitsbergen


.c ~._~

~ o ~

P"~ ~

= O~

o u.


6~ o=

6~ E~

=¢ = o





9.7 9.99"8 11.0 ~

o o





o ,, o

~ o oo ....... ~ u l t

:~ =

~ o



38 ~ "



= - -




45.0 459 46.1





93 104--


-I- 0 . 0 0 4

76 93 1 0 7 - - 99 122 103

0.063 +0.013






4- 0 . 0 0 5

40 to 50

48.1 >45.4



Hiatella (ka)

9.5 16.00.0150.020 19.0 +0.003 4.0.002


g E~


Till .


D/L - ratios


96 9.7


G *~,=;~







Amino acid Totalfraction

TL OSL dates dates




~4C dates


76 104 114--

• !'.o0. .....


120 to 130 89

0.069 0.008


'.~_-.,~-/ diamicton -o---... oo


43.8 49.0 >48.9 >50.4




~ ~.~.~,::.~.:~:.~.:~. ~....:~ -

~ m i ; ~ l


4. 0.009


FIG. 6. Composite stratigraphy of the Kapp Ekholm sections. The thickness of formations A and B are measured in Section II, Formations C, D and E in Section IV, Formations F and G in Section V, and Formation H in Section VI. Each of the coarsening upward sequences are marked with a wedge-like symbol to the fight in the lithological column. The radiocarbon dates are given in Tables 1-3. Within each formation the dates are plotted from the oldest to the youngest obtained age. For the Holocene only some radiocarbon dates from the base of the formation are included. The TL and OSL dates are given in Table 5, and the amino acid D/L ratios in Tables 6-9. Horizontal lines between TL and OSL dates indicate that the ages were obtained from the same sample.

sometimes in living positions, and also other fossils, demonstrate a marine origin. As mentioned above, the lower part of the silt is characterized by a reddish brown colour, contrasting to the more dull greyish colour of the rest. The silt beds are generally less than 0.5 m thick and grade into sand facies upward.

blocks occur sporadically. Paired molluscs, sometimes in living positions, and also other marine fossils are more common in the sand than in the silt facies. The sand facies is interpreted as having been deposited in a marine environment, below the wave base.

Sand This facies consists of sand, generally fine in the lower part and coarse in the upper. The sand is nearly massive. However, a crude bedding occurs. Stones and

Marine Diamicton This term is here used to distinguish a sediment that consists of nearly equal amounts of sand and silt/clay, with frequent (rounded) stones. A crude, nearly horizontal bedding is visible. It contains numerous


1. Mangerud and J.l. Svendscn T A B L E 1. Radiocarbon dates from Formation H (Holocene) at Kapp Ekholm

Lab. No.




Field sample No. and description


1l.(15(t _+ 15(1


I, 110

Svalbard 1988-696. Onc shell fragment from the base of a 21) cm thick reddish brown silt overlying till. The sample was collected just above sample -695 described below. Sample TUa-69, -70, -71, and -72 were subsequently dated to test if the age of Ua-972 could be reproduced


97311 _+ 1811


I, 1 1 1 )

Svalbard 1988-695b. One shell fragment from a less than one cm thick sand layer beneath stones along the boundary between the underlying till an the reddish brown silt, described above


37,400 _+ 16011


I, 1 1 1 )

Svalbard The date has been 2500 ka

1988-695a. A thick fragment from the same sample collection a TUa-70. shows that this fragment has been redeposited from older beds. The age calculated on the basis that non-finite old calcite yields an age of 42 _% BP with the same chemical preparation


94711 +_ 1411


1, 110

Svalbard 1988-697. One gastropod from exactly the same site as sample -696


9730 _+ 120


1, 130

Svalbard 1988-235. Paired Hiatella arctica in sorted sand, 5 cm above a 10-20 cm thick bed of reddish brown silt


98111 +__ 70


I, 160

Svalbard 1988-698. Balanus balanus sitting on a boulder in a 10 cm thick lens ol reddish brown silt at the transition between thc (interglacial) gravel lk)resets anti the Holocene sand above


96911 _+ 711


11, 2811

Svalbard 1988-78. Three paired Hiatella arctica from the transition between a brownish silt and overlying fine sand. Sample collected less than 10 cm above a bed of angular pebbles that are thought to be the base of the Holocene at this locality


89311 _+ 711


II, 300

Svalbard t988-733. Paired Mytillus edulis from the lower part of the Holocenc gravel


8740 _+ 100


IV, 5 5 t )

Svalbard, 1988-317. One large Mya truncata from a boulder bed. This bed is interpreted as an erosional lag at the base of the Holocene


8610 _+ 120


IV, 550

Svalbard 1988-323. One whole, large shell of Zirphea crispata, lound on tile surfacc of slumped material


9910 +_ 130


V, 650

Svalbard 1988-447. One Lepeta from a laminated silt with reddish and greyish laminae, at the base of the Holocene sequence


8560 _+ 60


V, 7411

Svalbard 1988-438. Onc large paired Mya truncala from the base o1 thc Holocene. The underlying till is missing at this site


9760 +_ 140


VI, 870

Svalbard 1988-758. Overlying tile youngest till is 20-30 cm reddish brown silt. Small Lepeta from the base of the silt


9531/ +_ 110


V1, 8711

Sa 81-08. Paired Mya lruncuta from the basc ol the sand above the reddish brown silt described under sample 1988-758. This sample was collected by Salvigsen and Mangerud in 1981, but can easily be plotted into the section measured in 1988


908(] _+ 110


VI, 8911

Svalbard 1988-522. Two large Mya truncata in living position, from a zone rich in large Mw~, 1.5 m above the till

In "Fables 1-3, the roman numbers in the column 'Locality" relcr to the section number, and arabic numbers give metres along the horizontal scale (Fig. 7). Samples are listed from south towards north along the section.

molluscs in living positions, and it is undoubtedly of marine origin. This diamicton facies occurs only in (the interglacial) Formation B, where it forms a thick unit between the sand and gravel in the coarsening upward sequence. Gravel In general this facies consists of clast supported, rounded gravel in steeply inclined (15-25 °) foresets. The exception is Formation F where the gravels partly occur as large lenses with glaciotectonic boundaries to sand and silt. Shells and seaweed occur sporadically; most often within sand lenses between the gravel foresets. In some sections the foresets interfinger with the underlying sand and silt, showing that the gravel represents prograding sequences. Facies Relations and Interpretations Most of the sections of Kapp Ekholm are wave cut

cliffs parallel to the shore (Fig. 7). The most complete cross section occurs along the gully at 500 m (Fig. 8). The top of the sections consists of a Holocene beach terrace which is nearly horizontal in cross section. The underlying bedrock surface is part of the valley side that slopes towards the sea, and therefore the entire deposit is wedge-shaped in cross section (Fig. 8). The vertical succession of the facies is similar in most of the sections and stratigraphic levels (Fig. 6). The tills (grey diamicton) are overlain by brown diamicton or silt, followed by grey silt which grades into sand. Alternating beds of silt and sand were also found, and in Formation B the marine diamicton takes the place of the sand. The sands interfinger with the overlying gravel foresets. This succession of sediment facies, starting with till and followed by marine mud, sand and gravel is typical for emerged sequences on Svalbard (Boulton, 1979, 1990; Miller, 1984). The first order interpretation is simple: Each glaciation was succeeded

Interglacial-Glacial Period on Spitsbergen


TABLE 2. Radiocarbon dates from Formation F, at Kapp Ekholm Lab. No.

, Age

+ 1o

+ 2 o



Field sample No. and description



+2400 -1900

+5900 -3400


I[, 280

Svalbard 1988-86, -87, -88, -89, and -90. Five paired Mya truncata from gravel, collected less than 10 cm from each other and 160 cm below the boundary to the post-glacial sediments





II, 280

Svalbard 1988-18. One paired Hiatella arctica from massive sandy gravel, 1 m below boundary to Holocene sediments





IV, 530

Svalbard 1988-309. One Mya truncata from deformed gravel (Mangerud and Svendsen 1990). See further comments under next datings


> 40,000


IV, 530

Redating of the same individual as Ua-973, because of the surprisingly young age. Inner 50% of 'thick' fragment was dated

Ua-1175 T-8433

> 40,000 > 29,400

Shells Shells

IV, 530 IV, 530

As Ua-l174. Inner 50% of thinner fragment. Sameindividual as Ua-973. The reason why it gave lower nonfinite age than other samples from Trondheim is that it was measured in a very small counter. With these redatings we conclude that sample Ua-973 was contaminated during preparation, and that the real age is > 40 ka



+6610 -3560

> 45,400 Shells

IV, 570

Svalbard 1988-325, -328, -329, -331, -332, and -335. Six paired Mya truncata from silty sand, all collected less than 50 cm above a till



IV, 570


+7500 -3800 +5400 -3200



+2900 -2100 +2300 -1800


IV, 580

Svalbard 1988-360. Several individuals of Serripes grenlandica from fine sand, 2 m above sample 1988-325 Svalbard 1988-252. Large bone of Greenland whale (Balena mysticetus) from just beneath the boundary to Holocene sediments. Approximately same level as sample 1988-360, but there occur glaciotectonic thrustplanes between them



+4200 -2700


V, 650



+1800 -1500 +1800


V, 650



+1900 -1500

+4400 -2800


V, 730



+3200 -2300

+8700 -4080


VI, 870

by m a r i n e i n c u r s i o n following glacial retreat. D u e to glacio-isostatic r e b o u n d , which causes a falling relative sea level, each m a r i n e f o r m a t i o n is c o m p o s e d of a coarsening upwards sediment sequence, grading from silt to gravel. T h u s , each s e q u e n c e of t i l l - s i l t - s a n d gravel r e p r e s e n t s a m a j o r glacial cycle. C o n t i n u o u s m a p p i n g of the u n i t s is i m p o s s i b l e b e c a u s e of the gully incision b e t w e e n the sections (Fig. 7) a n d the stratigraphic r e p e t i t i v e n e s s of similar sedim e n t a r y facies m a k e s it difficult to c o r r e l a t e b e t w e e n the i n d i v i d u a l sections. I n S e c t i o n I V (Fig. 7) s e d i m e n t facies f r o m f o u r d i f f e r e n t glacial cycles o c c u r in s u p e r p o s i t i o n , which a c c o r d i n g to o u r lateral correlations is the total n u m b e r of cycles p r e s e n t (Fig. 6). H o w e v e r , r e m n a n t s of e v e n o l d e r m a r i n e s e d i m e n t s occur as a n i n c l u s i o n (lens) in the l o w e r till (Fig. 6).

STRATIGRAPHY I n this s e c t i o n we focus o n the lithostratigraphic

Svalbard 1988-716, -717, and -718. Three paired Mya truncata in a recumbent fold of sand Svaibard 1988-724, -725. From the same beds as 1988-716, etc. This sample was first burned for conventional dating. When it turned out to be too small it was precipitated as CaCo3 and prepared as a target for AMS-dating. Thus it has been through a long preparation process, and contamination might be expected Svalbard 1988-471, -472, -473, -474, -475,-476, -459, -460. All samples are paired Mya truncata in gravel directly beneath the upper till Svalbard 1988-240, -245, -247, -249. Hiatella arctica: two individuals were paired, the other two were halfs. Beneath the till is 2 m of gravel. The specimens were collected in sand, 1-1.5 m below the gravel

d e s c r i p t i o n s of the s t u d i e d sections at K a p p E k h o l m , b u t o b s e r v a t i o n s of fossils are also i n c l u d e d . W e have s u b - d i v i d e d the e n t i r e s e q u e n c e into i n f o r m a l lithostratigraphic f o r m a t i o n s d e s i g n a t e d with letters, starting with A at the base a n d e n d i n g with H for the H o l o c e n e (Figs 6 a n d 7). W e use the s a m e letter for each f o r m a t i o n in all sections, a s s u m i n g that o u r c o r r e l a t i o n s b e t w e e n the sections are correct. H o w e v e r , a f o r m a t i o n in a single s e c t i o n m a y be identified as a s e p a r a t e u n i t by c o m b i n i n g the section n u m b e r a n d the f o r m a t i o n letter; for e x a m p l e F o r m a t i o n I - B a n d F o r m a t i o n I V - B ( T a b l e s 1-9). T h e s e d i m e n t facies described a b o v e ate used to d e f i n e the lithostratigraphic units. W e have labelled each of the grey tills as a f o r m a t i o n (Fig. 6), most o f t e n r e f e r r e d to as till A , till C, etc. E a c h c o a r s e n i n g u p w a r d s s e q u e n c e of till, sand, m a r i n e d i a m i c t o n , a n d gravel facies is also d e f i n e d as o n e single f o r m a t i o n (Fig. 6). I n s o m e of the sections these f o r m a t i o n s are s u b - d i v i d e d into a lower s i l t - s a n d m e m b e r a n d a n


J. Mangerud and J.l. Svendsen T A B L E 3. Radiocarbon dates from Formation B, at Kapp Ekholm

Lab. No.


+_ t o

_+ 2 o





+ 1100 - 1000

+2300 - 1800


I, 150

Svalbard 1988-237. Piece of wood, 5 cm in diameter, and 70 cm long. It was found on the surface near the boundary between the silt and gravel members of Formation B. It could have slided downwards, but according to our interpretation there are no younger sediments above this site. except Holocene



+9700 -4300

> 49,000


II, 270

Svalbard 1988-544, -545 and -546. Three paired Mya truncata from the "Mytilus zone"



+2300 -1800

+5600 -3300


II, 280

Svalbard 1988-113. Paired, large Mya truncata, from the lowest part of the sand (Fig. 10). 5 cm above the reddish brown silt



> 53,901/> 50,400


IV. 570

Svalbard 1988-665, -667, -672. Three paired Mya truncata from the marine diamicton



> 52,900 > 48,900


IV, 570

Svalbard 1988-677. Mya truncata from the marine diamicton

> 60,900


V, 650

Svalbard 1988-682, -683, -684, and -685. Paired Mya truncata from marine diamicton, around 5 m a.s.l.


Field sample No. and description

T A B L E 4. Radiocarbon dates from the section near Nidedalen Lab. No.


+ 1 o

+ 2 o




+4000 - 2600

+ 12,000 - 4600


Svalbard 1988-6311, -633, -640. Three Hiatella arctica from sand just above the lower till





Svalbard 1988-615. Several paired Chlamys islandica from gravel foresets. 3-4 m above the upper till

Field sample No. and description

upper gravel member. The brown diamicton facies, which sporadically occurs on top of the till units, has been included in the marine silt-sand member. In the field the formations were correlated from section to section by means of sediment facies, fossils, style of glaciotectonic deformation, and stratigraphic position. The field correlations were later tested and confirmed by radiocarbon dates, TL-dates and by the amino acid D/L ratios measured on shells. The inferred ice free periods are given informal climatostratigraphical names (Nystuen, 1989). These are defined by the marine sediments deposited during the first part of the ice free periods. Kapp Ekholm interstadial is defined by Formation F, Phantomodden interstadial by Formation D, and interglacial B (Eemian) by Formation B (Fig. 6). During latter parts of the ice free periods it is assumed that the sections were sub-aerially exposed and these periods are not represented by sediments in the sections. The glaciations during which the tills A, C, etc. were deposited, are referred to as glaciation A, glaciation C, etc.

Formation A (Till) This formation was recognized in Sections I and II (Fig. 7). The following description is from Section II

where it was exposed along the base of the cliff (Figs 9 and 10). In Section I this unit was for the most part covered by slumped material (Fig. 12) and north of Section II it dips underneath the present shoreline (Fig. 7). The dip apparently reflects the descending bedrock floor. The sediments form a typical grey diamicton facies as described above. The matrix is sandy in some parts and in others more silty, the variation is apparently random. Several isolated pockets of gravel, probably derived from older shore deposits, occur in the till. The thickness of the formation varies between 1.3 and 4 m. Glacial striae occur on the underlying bedrock surface, oriented 226 ° at 300 m and 210 ° at 270 m, showing ice flow parallel with the fjord. The underlying bedrock is sharply cut by the till. At 280 m there is a 20 cm thick and 3 m long lens of a reddish brown silt/diamicton at the base of the formation which contains frequent small shell fragments, obviously from molluscs that lived during a preceding ice free period (Figs 6 and 10). The lens has a transitional boundary to the grey diamicton above, and is covered by slumped masses from the steep cliff at both ends. Its lateral extension was therefore not determined exactly. However, this unit was not seen exposed elsewhere along the section.

80 + 7


93 + 3

76 + 2

99 + 10

103 + 5

89 + 4







6.7 + 0.8

4.5 + 0.5

42 + 1




40 + 4

36 _+ 5 310--470






































Max. T L age (ka) Formation








































Field No. Svalbard

1 m above sample 1988-739. More sandy

A thin sand bed in gravel foresets. 5 m below the till. 15 m a.s.l.

Sample collected between high and low tide on present beach. The top layer was removed before sampling. T h e apparent age indicates that it is not fully zeroed

Sample from the present beach. T h e apparent age indicates that it is not fully zeroed

A s sample 1988-752

Early Holocene gravel foresets. Sand matrix sampled

Sand beneath gravel foresets. Radiocarbon age 9080 + 110 (T-8534)

Lower part of foresets, 5 cm above sand (Fig. 10). T h e OSL m e t h o d yielded three different ages

Foresets, 1.5 m below upper boundary. 19.5 m a.s.l.

Sandy gravel in the foresets, 2 m below the upper boundary. A g e plateau very short

C o m p a c t sand, 35 cm below the base of the foresets

Sand matrix in middle part of the gravel foresets

5 cm thick sand bed. 20 cm above sample 1988-509

Sand matrix in gravel. Lower part of gravel foresets

Sand in upper part of Formation F

Gravel, 1.5 m below the till G. 20 m a.s.l.

Sand in upper part of Formation F, 21.5 m a.s.I. Acceptable plateau. However, also short plateau corresponding to a T L age of about 155 ka

Massive sand, 90 cm below the boundary to the Holocene gravel. 24 m a.s.i.

Gravel at the top of Formation F; 12 m east in Fig. 15


Lab. No. refers to the Nordic Laboratory for Luminescence Dating. Plateau indicates the t e m p e r a t u r e intervals for the T L plateau. Max. age is the age obtained if one postulates that the T L signal was completely zeroed during deposition. Locality gives metres along the shore at Kapp E k h o l m (Fig. 7).


16 + 2

9.5 + 1

19 + 2



76 + 10




104 + 10


114 + 10

122 _ 10

107 + 10

53 +__7

44 _+ 5

50 + 2


38 + 5

11)4 + 10


44 + 5

59 +__ 3


Plateau (°C)

T L age (ka)


Lab. No. OSL age R (ka)

T A B L E 5. T L and OSL dates from the Kapp E k h o l m and Nidedalen Sections



O e~


J. Mangerud and J.l. Svendsen T A B L E 6. A m i n o acid D/L ratios from Formation H, the Holocene, at Kapp Ekholm


D/L-ratio Total



Field sample


1820A 1822A 1823A 1824A Mean

0.022 0.020 0,017 1~.022 0.020 +__0.002

H I- 130 235 H II-280 78 H 11-28(I 78 H 11-280 78 for Hiatella arctica from the Early Holoeene

Paired shells radiocarbon dated to 9730 _+ 120 ('1"-8322) BAL-1822, -1823, and -1824 are from 3 paired shells, radiocarbon dated to 9691) +_ 70 (T-8319)

1825 A 1829A 1830A 1831A Mean

0.018 0.014 0.017 0.012 0.015 _+ 0.003

M M M M for Mya truncata

Radiocarbon dated to 8710 _+ 100 (T-8323) BAL-1829, -1830, and -1831 are from three shells in living position, one was radiocarbon dated to 8590 + 60 (T-8326)

1V-550 317 V-740 438 V-740 438 V-740 438 from the Early Holocene

Note for Tables 6--11 of amino acid results: BAL-No. is the sample n u m b e r at the Bergen amino acid laboratory. Postscripts A and B m e a n s independent preparations and m e a s u r e m e n t s of the same specimen. W h e n more than one D/L ratio is given under the same n u m b e r and letter, a prepared sample is remeasured. Sp. m e a n s species, where M is Mya truncata and H is Hiatella arctica. The full designation for the field sample is Svalbard 1988-No. D/L ratios given in ( ) are not included in the calculated mean values and standard deviations.

Formation B (Interglacial B) This formation was studied in most detail in Section II, which is described below (Figs 9 and 10). The lower member of formation B in this section includes several beds of silt, sand, and a marine diamicton (Fig. 10). In several places a thin layer of a brown diamicton facies was observed along the transition between Formations A and B; possibly it occurs continuously along the entire Section II. The diamicton grades upwards into a 20 cm thick brownish silt which contains many shells. The gradational boundary to the marine silt above, compared with the sharp boundary to the till below, suggests that the brown diamicton was also deposited as a marine sediment. Between 260 and 285 m there is a shallow depression in the surface of the underlying till (A), presumably a channel (Fig. 10). It is filled with faintly bedded silty sand, 2.2 m thick in the centre. Paired Macoma calcarea and Mya truncata occur in this sand facies. Probably it was deposited by gravity flows from the sloping sea shore. The sand does not extend laterally, but a similar facies was exposed at the base of Section IV (at 500 m). Over the channel fill there is a 30 cm thick gravelly zone (Fig. 10) which cuts the underlying sand. In this zone we found several fragments, and also whole shells, of Mytilus edulis (Fig. 11). This is the only site and stratigraphic level where we found pre-Holocene Mytilus in the sections. Also Lavrushin (1969) has earlier reported fragments of Mytilus from this section, as far as we understand from the channel fill. Mytilus edulis requires warmer sea surface temperatures than at present, indicating that formation B is of interglacial age. The gravelly zone with Mytilus represents the lower part of a nearly 5 m thick bed of a distinctive marine diamicton facies (Fig. 10). Paired molluscs in living positions occur frequently in this diamicton, including exceptionally thick and large shells of Mya truncata.

Macoma calcarea, Serripes groenlandicus, Hiatella

arctica, and Astarte sp. are also common. The many individuals of Mya in upright, living positions, show that the bed has not been disturbed after these molluscs burrowed into the sea floor. The high content of mud in the sediments indicates quiet water environment during deposition, which requires a water depth of some 15-20 m or more. In conformity with Boulton's (1979) interpretation, we assume that most of the stones have rolled down the steep slope from the shore, others may have been dropped from sea ice. Due to the distinctive appearance of the marine diamicton as compared to the other sediment facies, Formation B could be recognized in the different sections by visual correlation (Fig. 7). A similar lateral correlation of these diamicton sections was undertaken by Boulton (1979). Above the marine diamicton there is a thick sequence of foresets consisting of well rounded gravel. In some of the sand lenses which are interbedded in the gravel foresets, seaweed and shell fragments occur. Rust staining of the gravel was seen many places, especially along the boundaries of the more finegrained sediments. We were not able to determine whether the rust staining is 'old' or Holocene in age. The sequence of gravel foresets becomes considerably thicker (12 m) in Section I (Figs 7 and 13) whereas to the north of Section II it appears to wedge out (Fig. 7). Thus, at the southern end of Section III there is only a small pocket of gravel above the marine diamicton. However, Section III is only a low ridge that was heavily covered by slumped material, and no sedimentary unit was confidently identified between the marine diamicton and Formation F at this particular exposure. In the upper part of the marine diamicton in Section IV (570 m) there is a 2.5 m thick gravel lens with a crude bedding, which is probably a remnant of the same gravel member as in Section II. There are also pockets of gravel in the till above, which may derive from Formation B. Boulton (1979) correlated a till (his unit 3) into a


Interglacial-Glacial Period on Spitsbergen

TABLE 7. A m i n o acid D/L ratios from Formation F, the Kapp Ekholm interstadial, at Kapp Ekholm D/L-ratios






Field sample

1800A B 1651A B 1652A B Mean

0.023 0.023 0.021 0.019 0.017 0.021 0.021 __. 0.002





Paired. From sand 60 cm above till

0.352 M 0.232 0.320 M 0.343 0.319 ± 0.050



Paired. From sand 30 cm above till



Paired. From sand 30 cm above till

0.183 0.208 0.268 0.241 0.232 0.243


for Mya truncata from Formation I I - F Thrusted sand, just above till Lens of gravel. Same individual yielded non-finite C-14 ages (Ua-I174/I175, T-8324) Samples 329, 331,332, and 335 are four paired shells. In sand less than 50 cm above till. Radiocarbon dated to > 45,400 (T-8324)


IV-530 IV-530

302 309


IV-570 IV-570 IV-570 IV-570

329 331 332 335



Paired. In sand one m above sample 329



As 338

1808A 1821A B 1826A 1827A 1828A 1792A B 1802A B 1793A B Mean

0.024 0.019 0.029 0.024 0.024 0.022 0.031 0.027 0.027 0.028 0.028 0.026 ± 0.003

1834A 1835A 1796A B 1832A 1833A 1797A 1794A B 1804A B 1795A B Mean

0.028 0.026 0.029 0.029 0.026 0.025 0.027 0.026 0.034 0.032 0.023 0.031 0.033 0.028 __. 0.003


V-650 V-650 V-730

716 717 460

0.146 M 0.215 M 0.136 M 0.187 M 0.228 0.215 M 0.220 (0.669) M 0.265 0.184 __. 0.049

V-730 V-730 V-730 V-765

472 475 476 402






0,184 H 0,183 0,173 H 0,162 0,177 H 0,183 0,117 ± 0,008





Samples 410-411 are paired shells in sand, 70-80 cm above the underlying silt where samples 402, 403, 406 listed above were collected



As sample 410, but not paired


0,031 0,030 0,032 0,033 0,028 0,026 0,030 + 0,002

1817A 1818A 1819A Mean

0.042 0.035 0.027 0.035 ± 0.008

0.190 H 0.215 H 0.118 H 0.174 ± 0.050

VI-870 240 Samples 240, 247, and 249 are three paired shells. From VI-870 247 sand 3 m below till. Radiocarbon dated to 46,100 (T-8533) VI-870 249 for Hiatella arctica from Formation VI-F

Mean Mean

0.026 ± 0.004 0,031 _+ 0,005

0.207 ± 0.075 0,176 ± 0,025

for Mya truncata from entire Formation F for Hiatella aretica from entire Formation F

2511 2512

0.146 0.234 M 0.206 0.208 M 0.174 0.205 __. 0.043 0.123 0.215 0.118

for Mya truncata from Fromation IV-F

Paired. From overfolded sand As 716 Samples 460, 472, 475, and 476 are four paired shells. From gravel 0.2-1 m below till. Radiocarbon dated to 42 ka (T-8327) Samples 402,403, and 406 are three paired shells. From 30 cm thick silt just above underlying till

for Mya truncata from Formation V - F

for Hiatella arctica from Formation V - F

postulated unconformity between the marine diamicton and the gravel foresets in Sections I and II. However, at the southern end of Section II (11-12 m a.s.l., Fig. 10) we discovered well-defined gravel foresets that clearly finger into the marine diamicton below. These gravel beds were mapped continuously along the cliff for more than 20 m within the upper part of the marine diamicton, and they gradually become more finegrained in the distal direction. We therefore conclude that the marine diamicton is a distal facies of the gravel foresets, and that there is no unconformity along the boundary between the two different units.

Formation C (Till) This till unit is mapped continuously along Section IV (Fig. 13), but it varies greatly in thickness and lithology. At 500 m it is only 1 m thick, and appears as a well-defined bed of compacted, homogeneous grey diamicton facies with numerous striated stones. At 550-570 m the same till unit is as much as 4-5 m thick; the upper half is massive, as at 500 m, whereas the lower part has incorporated large lenses of gravel and silt, and contains numerous shell fragments and stones that are covered with bryozoa. In some places the lower boundary is difficult to define.


J. Mangerud and J.I. Svendsen TABLE 8. Amino acid D/L ratios from Formation D, the Phantomodden interstadial, at Kapp Ekholm D/L-ratios





0.044 0.055 0.046 0.043 0.053 0.053 0.049 "4- 0.005

0.280 M 0.277 0.318 M 0.264 0.312 M 0.321 0.295 ± 0.022 0.349 0.297 0.366

1658A B Mean

0.075 0.076 0.079 0.078 0.068 0.066 0.070 0.070 0.073 + 0.005


0.063 ± 0.013

2517 2518 Mean

1657A B 1803A B



Field sample







Comments Samples 256, 257 and 259 are three paired shells. In 1.4 m thick sand just above reddish brown silt at the base of the formation

for Mya truncata for Formation D at 500 m







0.317 M 0.346 0.335 ± 0.028



for Mya truncata from Formation D at 565 m

0.314 ± 0.029

for all Mya truncata from Formation D

Samples 384, 387 and 392 are three paired shells In sand just above reddish brown silt at the base of the formation


T A B L E 9. Amino acid D/L ratios from Formation B (Eemian interglacial), at Kapp Ekholm D/L-ratios BAL-No.





Field sample

1801A B 1647A B 1648A B

0.072 0.077 0.073 0.069 0.068 0.068 0.071 0.071 ± 0.003

0.396 0.385 0.291 0.355 0.379 0.380







Samples 110, 111, and 113 are extremely large individuals, collected 5-11 cm above the reddish brown silt at the base of the formation




Radiocarbon dated to 49 ka (T-8321)

0 . 3 6 4 ± 0.035

for Mya truncata from the base of Formation l i b

0.332 0.335 0.297 0.288 0.339 0.352 0.291 0.307




Paired. Top of silt-sand member, 120 cm below gravel




As sample 127




As sample 127




Paired. Top of sand member, 80 cm below gravel

1807A B 1650A

0.073 0.076 0.055 0.056 0.078 0.076 0.056 0.062 0.052 0.061 0.062 0.062



Not paired. Otherwise as 709



Paired. 'Fop of sand member, 1 cm below gravel



0.340 M 0.352 0.301 M 0.361 0.342 0.451 0 . 3 3 4 ± 0.040


2451 2513 2514 1649A B


0 . 0 6 4 ± 0.009

1653A B

0.071 0.060

1654A B


0.077 0.071 0.070 0.073 0.082 0.081 0.073 0.076 0.078 0.077 0.074 ± 0.006


0.069 ± 0.008

1806A B 1655A 1656A B


for Mya truncata from top of the silt-sand member of Formation II-B

0.344 0.366 0.266 0.257 0.337




Paired. In silty sand 15 cm below deformed gravel at top of the formation




As 660

0.394 0.374 0.306




Samples 665,667 and 672 were paired. 2 m below gravel. Radiocarbon dated to > 50,400 (T-8335)




0.335 0.346




0.332 ± 0.044

for Mya truncata from Formation I V - B

0 . 3 4 0 ± 0.042

for Mya truncata from entire Formation B (all samples listed above)


Interglacial-Glacial Period on Spitsbergen TABLE 10. Amino acid O/L ratios from lens of reddish brown silt in lower part of till A, at Kapp Ekholm D/L-ratios BAL-No.





Field sample


0.130 0.126 0.126 0.125 0.109 0.109

0.448 0.431 0.480 0.467 0.366 0.429










0.121 --- 0 . 0 0 9

0.437 _ 0.37

2509 2510 Mean

Comments Sample 122 contains shell fragments collected 0-20 cm above bedrock

for Mya truncata p r e d a t i n g Till A

TABLE 11. Amino acid D/L ratios from the section at Nidedalen D/L-ratios BAL-No.




Field sample


0.045 0.040 0.037 0.033 0.027 0.025 0.039 0.031




Paired shells, 3 m below till



Top of sub-till sediments



Paired shells from top of sub-till sediments


0 . 0 3 4 _+ 0 . 0 0 6

0 . 2 6 6 _.+ 0 . 0 3 3

for Mya truneata f r o m the sub-till sediments

1798A B 1799A B

0.055 0.056 0.050 0.051

0.228 0.223 0.246 0.200



Paired shells from base of sub-till sediments



As sample 638


0.053 + 0.003

0 . 2 2 4 __+ 0 . 0 1 9

for two individuals o f Hiatella aretica f r o m the sub-till sediments


0.041 0.013

0.227 0.205


B 2452 2515B

0.304 0.312 0.225 0.242 0.262 0.272


Because the formation (D) above wedges out at both ends of Section IV, till C merges with the next younger till (E) in the adjacent sections (Fig. 7). Thus the lateral correlation of this till is not clear. In Section II (Figs 10 and 15) there is a till that can be bracketed between Formation B and F, and which can be either till C or E. This till unit sharply cuts the underlying gravel foresets, indicating that the lower boundary is erosive. Due to its compactness, it was easily recognized as a well-defined layer protruding from the exposures in Section II (Fig. 9). In the E - W section shown on Fig. 15, the till wedges out 5 m east of the cliff. However, it can be traced further as a horizon of striated boulders and isolated pockets of till until it re-occurs east of 15 m. Evidently the top of the till is eroded, and for stretches it is entirely removed. This indicates that the till unit in question represents Formation C and not E. Otherwise the erosional episode must have occurred after the deposition of Till E, but before Formation F, which is not very likely. In Section I the gravel foresets of Formation B are


From silt at the base of the Holocene

sharply cut by an erosional unconformity, on which there is a till of typical grey diamicton facies that can be mapped continuously from 100 to 130 m. From 130 to 180 m the till occurs as lenses or as a horizon of striated boulders, but for stretches it is completely missing. The till is a maximum 1 m thick at 100 m. It is stratigraphically bracketed between Formations B and H (Holocene) only, thus it could be correlated with any of the Tills C, E or G in the other sections (Fig. 7).

Formation D (Phantomodden Interstadial) This formation occurs in Section IV only (Figs 7 and 13). Above Till C there is a 20-60 cm thick bed of brown silt facies, which at 500 m appears as a thin layer of a brownish diamicton. There is a gradual transition from the brownish silt to sorted sand facies above. The sand contains frequent Mya truncata in living positions, especially at 560-570 m, and also paired Macoma calcarea and other molluscs. Above the sand there is a sequence of gravel foresets


.I. Mangerud and J.I. Svendsen ~i

ro°o° 4 Gravel Gravel foresets Sand ~__.~_ - Silt

Marine diamicton (in formation B)

Tm Glaciotectonic thrusts and folds L_ ] Covered by slamped material


North IV


























FIG. 7. The upper figure shows the lithostratigraphy of each section, and our correlations between the secUons. The vertical scale is exaggerated 20 ×. Most formations dip toward the sea (the viewer), and also increased in thickness in that direction. Thus elevations and thicknesses depend on how far back each section is eroded. The boundary between sand and gravel facies in the Holocene sequence was not mapped systematically, and in some places this line is drawn arbitrarily. The stippled top on Section IIl indicates a gentle slope with small sections behind the main cliff. On the lower figure radiocarbon (marked with X) and OSL/TL (marked with dot) dates are plotted in ka (thousands of years). For the Holocene one decimal is included. For samples dated with both the OSL and TL method, ages are given as OSL/TL. The OSL sample (R-902509) from Formation B (280 m) that yielded three different ages is omitted.

representing the upper member of this formation. The base of the planar foresets curves towards the lower sand. The gravel member decreases in thickness, from 10 m in the southern end of Section IV to only 2 m at the northern end (Figs 7 and 13). As seen in Fig. 8, it also becomes thinner towards the valley side. The top of the foresets are sharply cut by the overlying till (E), but they are hardly disturbed. Because this formation

was only found in Section IV we have considered the possibility that formation D might be a glaciotectonically upthrusted floe from Formation B. The obtained T L ages (Table 5) and amino acid D/L (0.063) ratios from this formation are not significantly different from the samples from Formation B (Tables 8 and 9). Thus, from the available dating methods alone, we were not able to prove that Formation D is in fact younger than

m a.s.I. West 40--




Terrace 28.5 m a . s . I .

A I l ~ v i a l ~

,~.,ctonQ,~ ~ . . . ~.,_///-.~..j[.y . . . . ~ . ~ ~ . / . x ; ~. %- .' j ~~ 20

E D ~

10 --

~ /


J ~

n, /




/ / /

r ~ - q Diamicton-basaltill ~


av,,oresets , ~~""~Gidt,"l

Beach~ O--

'l '~ 0



'2 /


I 100






Marine dlamicton



Shallow marine sediments



FIG. 8. A cross section showing the stratigraphy along the gully at the south end of Section IV (500 m in Fig. 7). The formations are marked with letters on the shore face cliff. The stratigraphic position of Till G is also marked, even though this formation is missing in this section. Note the wedge shape of the entire deposit, and that some formations wedge out towards the valley side.


Interglacial-Glacial Period on Spitsbergen

FIG. 9. Photo of Section II taken in 1981, and used for the field mapping in 1988, when the section was very similar. The horizontal scale in metres along the shore, and some levelled elevations are marked. The boundaries between the formations are marked with full lines. In Formation B the boundary between the silt-sand member and the gravel member is stippled. Some of the foresets that can be seen in the photo from Formation B are dotted.

NNE maslj




15 -~::~i~i~i~i~i~i~i~i:~:!~!~!:~:~t!:;~!:!:!:~i~i:;~t~i~i~:~ o

Formations ----~ F I


..........................i.i ..............:i:i...........i:i:::::i:i:............................~ .................................J C o r E

o °

o 70

o ° °





o o


Grain Till













. .. .

. .......









o .- .

. o .. . . .. . . . . . . o.......................

o . . . .

o . . . .

. . ..


.o ....-...~ ~ ~ ,



















S t o n y zone o " .-~!~ .'= '="J w i t h MytUus

_ . . . . . . . . . . . . . . . . . . . . . . . . . . . .


dMarine iamicton


. . . . . . . . . . . . . . . . . . . . . . . . . . . . .

0 . . . .

. -'; ~. . ~ - - . . . . . . . .. . __.o__ . 0 - - . o - - , o - -

50% . . . .


.... ~O~ZSi~2÷2.-'.2-

: _ ~ .o -~'-" ° ~o ~-._~° o- ~ - ' ~o- ~ " "o~ - o- " o" - ~ "o" ~o- ~_ o . ~ "~'-o o " ~o o . . . . . . .




o . . . .


Sand and cla~

. . . . . . .


; • 01 " •

"°}':"::""-------.° t!i.(/._. • o : '.'. . . . . ,' . '...'.:''_ _- _ - - _ - - _ - - ___

o, ..: :./


,<, -.'.'....



Ii,o' o / o .I.".'.". . ."" ""1

i s,ty sand .'~ e - I Reddish I b r o w n silt

i!::~i::i ::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::: A ::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::!i::::::]i::~ Till :::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::: ::::::::::::::::::::::::::::::::::::::::::: ::~::~::~,~:.~~ ; ~ ~i~,~~:.~:~:t

". '.'" "- ".'1 --'-~---------.



I~ ~::." L°,

----'-------~ -: o O%Oo..,-.j:.- ..[~.---

I'°'=1 : ~


Jc%~OC~ 0 280




270 Metres





FIG. 10. A sketch of the southern and lower part of the exposed sediment in Section II. The reddish brown lens with shell fragments in Till A is shown to theleft. The details of Formation B are described in the text. The grain size distribution for some samples are shown to the right. Maximum grain size included in the analysis is 16 mm. The lower sample from Till A is from a facies dominated by a sorted silt matrix, whereas the upper sample is from the typical diamicton facies.


J. Mangcrud and J.I. Svcndscn

FIG. 11. Photo of a Mytilus edulis sitting in the sediment of Formation B, Section II, sec Fig. 10. ¢)n the scale each black and white box represents I cm.

FIG. 12. Photo of Section I, which has not significantly changed since the picture was taken in 1981. Boundaries between formations are marked with full lines. In Formation B the boundary between the silt-sand m e m b e r and the overlying gravel m e m b e r is stippled. Some of the foresets are dotted.

Interglacial-Glacial Period on Spitsbergen


FIG. 13. Photo, taken in 1981, of Section IV, In 1988the lower part, between 550 and 570 m, was considerablymore covered by slumped material. The northernmost 20 m of the section is located outside the photo. Exposed boundaries between the formations are shown with full lines, dotted where covered by slumped material. The cross section shown in Fig. 8 is located from the right part and outside this picture.

FIG. 14. Photo of the northern part of Section V. The lower part was more extensivelycovered by slumped material in 1988than shown in this photo from 1981. The till beneath Formation F may be either C, E, or both.

Formation B. However, we could not find a single thrust plane or other features indicating that the t00 m long sediment sequence is dislocated. Furthermore the distinctive marine diamicton facies of Formation B is not present in Formation D. Rather there is an even transition from reddish brown silt to sand and gravel, indicating that Formation D represents a full coarsening upwards sequence which is different from Formation B. We therefore exclude the possibility that Formation D is upthrusted. The ice-free period corresponding with Formation D has been named the P h a n t o m o d d e n interstadial, after the name of the point to the south of the sections (Fig.


Formation E (Till) This consists of a hard, grey diamicton facies with frequent striated pebbles, which is interpreted as a basal till. The maximum thickness is only 1 m; yet it can be mapped continuously along the entire Section IV. As described above, this till merges with the previous till (C) towards the north and south, and it cannot therefore be confidently identified outside this section.

Formation F (Kapp Ekholm Interstadial) This formation occurs in all the sections except Section I. According to our interpretations it represents the last pre-Holocene ice free period (Mangerud et al.,


J. Mangerud and J.l. Svcndsen

m a.s.I.


W 9690 ± 70 (T - 88319)

44,000 ± 5,000 (R-892503)


.45,~o0 ± ~


Formation H (Holocene)

.37,000 ± 2000 (Ua-975)

Formation F

Angular gravel



- -



Less distinct boundary Distinct b o u n d a r y / t h r u s t

Patch of


~ ~"~











FIG. 15. Sketch of the stratigraphy in Section II, as it appears from a northfacing exposure along the gully at 280 m. Thick lines indicate distinct bedding or thrusting planes. The subhorizontal planes at 20-22 m a.s.l, are clear thrustplanes, partly with dragfolds. Formation F was deformed by an ice advance that is stratigraphicaUy placed at the unconformity between Formations F and H. T h e figure is based on a field sketch and photos. The foresets in Formation B dip towards the viewer; all the other structures are drawn as seen in this plane. T h e 44 ka date is a T L date (Table 5), the other dates are radiocarbon dates (Tables 1 and 2).

1992), which we here name the Kapp Ekholm interstadial. Formation F was studied in detail in Section II during the first part of the field work (Fig. 15), because several non-finite radiocarbon dates were previously reported from this section (Lavrushin, 1969; Troitsky et al., 1979; Mangerud and Salvigsen, 1984), and because Lavrushin (1969) and Troitsky et al. (1979) concluded that the section had not been covered by glacial ice after deposition of this formation. Also Boulton (1979) assumed that this formation had not been covered by glacial ice, but having no radiocarbon dates from this particular section (Section II), he considered this unit to be of Holocene age. After a large part of Section II was uncovered (Fig. 15), several thrust planes were observed within this formation, especially in the upper part where silty beds had been thrusted over sand and gravel. Furthermore, some dipping beds in the sea front exposure, that were previously interpreted as foresets in primary positions by Mangerud and Salvigsen (1984) and others, appeared to be low angle normal faults (Fig. 15, between 3 and 8 m on the horizontal scale). Drag folds and a recumbent fold were also recognized, showing dominant deformation directions towards W or NW (Fig. 16). At 5 m east, the uppermost bed within this formation (with the radiocarbon date 37 ka BP on Fig. 15) is a sandy massive diamicton, probably a deformation till. We obtained two radiocarbon dates (37 and 45.9 ka BP, Table 2, Fig. 15) from this formation that





55 g'2


FIG. 16. Glaciotectonic structures measured from Formation F, plotted in an equal angle (Wulffs) net. Fold axes are marked along the periphery, whereas poles to thrust planes are plotted within the net. All structures are marked with section n u m b e r s (here using arabic numbers). Most of the faults with a low angle dip towards N W are normal faults, but we assume they were caused by glacial thrusting towards NW. The arrow pointing towards W shows the direction of a keel beneath Till G in Section V. Most structures reflect a W or N W ice flow direction, which is crosswise to the fjord. Two fold axes in Section II reflect a SSW ice flow out of the qord.

Interglacial-Glacial Period on Spitsbergen confirm the 'old age' obtained by Lavrushin (1969) and Mangerud and Salvigsen (1984). The internal stratigraphy of Formation F in this particular section could not be revealed in detail because of the major overthrustings. However, the fact that silt facies occur at the base of the formation or at the base of each thrusted unit indicates that there is a coarsening up, from silt to gravel, within the formation. In Section IV, at 500 m (Fig. 7), the lower part of Formation F is a 70 cm thick bed of brown silt facies with frequent small shells (Lepeta sp., Pectinidae sp.). There are several thrust planes throughout the silt bed reflecting a push direction towards W N W (Fig. 16). Above this silt, and apparently in the correct stratigraphic position, there is a 1.5 m thick sequence of sand. At 530 m there are lenses of gravel in this sand, from which shells for several radiocarbon dates were collected. One AMS-date on a shell from one of these gravel lenses yielded the radiocarbon age 17.7 + 0.3 ka BP (Ua-973, Table 2) (Mangerud and Svendsen, 1990a), but several re-datings of this shell indicate that the correct age is greater than 40 ka BP (Table 2). One of the less disturbed sequences of Formation F was studied at 570-580 m in Section IV. Here, there is a sharp boundary between till E and a 5-20 cm thick brown diamicton, which grades into a brown silt facies above. The brownish silt gradually becomes more coarse g r a i n e d upwards and grades into sand of a greyish colour. The sand is 2-3 m thick, but thrustplanes, especially in the upper half, indicate that some of the thickness may be due to tectonic repetition. The sand contains frequent paired molluscs (dates in Table 2), such as Mya, Hiatella, Serripes, Macoma, Musculus, and several other species, including Chlamys islandica, which do not live in the most extreme high Arctic areas today. Large whale bones, identified as Greenland whale (Balaena mysticetus, Roll Lie oral communication, 1988) and dated to 44 ka BP (Table 2), were also uncovered from the upper part of this sand sequence. The sand of this formation is similar to the overlying Holocene sand, but it can be distinguished from this unit because it is slightly more compacted. From 720 m to the northern end of Section V (790 m, Figs 7 and 14) there is a gravel m e m b e r occurring above the sand described above. On the ridge at 730 m there are foresets of gravel with sand lenses that contain shells that were used for radiocarbon dating (T-8327, Table 2) and amino acid analysis. The gravel wedges out towards the valley side in the inner part of the gully at 740 m. From 750 to 790 m the gravel unit is 5-6 m thick. In this unit there are some vague structures (not measured) which were interpreted as a kind of tectonic schistosity. In spite of the glaciotectonic deformations that have apparently destroyed the primary bedding, we assume that the gravel stratigraphically overlies the silt and sand (Fig. 6). The most spectacular deformations are found at the southern end of Section V, around 660--670 m. In one of the ridges the beds are tilted to a vertical position, with a strike direction due N-S. A large overturned


fold was also recognized at this site. These deformation structures were included in Boulton's (1979) log 6 from this section. Formation F was correlated between the different exposures in the field. Subsequently, it was unambiguously distinguished from the older formations by the considerably lower D/L ratios (Fig. 6). A remaining question that can be raised is whether Formation F represents one or more depositional events. The mean amino acid D/L ratio (0.021 + 0.002) for Mya truncata from Formation F in Section II is lower than the ratio (0.028 + 0.003) obtained for the formation in Section V, which might indicate a difference in age. However, the D/L ratio (0.026 + 0.003) for Formation F in Section IV overlaps both of the former values within one standard deviation (Table 7). We consider the variation in D/L ratios between the different sections to be within the accuracy of the method. From the available evidence we conclude that Formation F represents a single marine event. This formation defines the onset of the Kapp Ekholm interstadial. As discussed below, there is evidence to suggest that this ice free interstadial lasted for a considerable period after the sediments of this formation emerged above sea level.

Formation G (Till) In Section V this formation appears as a typical grey diamicton facies, confidently interpreted as a basal till. This formation is the uppermost and youngest till which were found in the Kapp Ekholm sections. In most of Section V the till is recognized as a well-defined discordant layer above Formation F (Figs 7 and 14). From 760 to 790 m it is 1 m thick, and covered by only approximately 0.5 m of Holocene gravel (Fig. 14). From 730 to 760 m the till unit is missing, but it occurs along the entire southern part of the section. Till G is correlated with a diamictic bed in Section VI. The lower 50 cm of this diamicton has a sandy matrix similar to till G in Section V, whereas the upper 80 cm of the diamicton is more silty with a brownish grey colour. This formation dips obliquely towards the sea, and disappears below sea level shortly to the north of 900 m (Fig. 7). In Section IV the till is missing, but at places it could be traced as a lag of stones and blocks along the boundary between Formations F and H. Most of the measured glaciotectonic thrusts and folds in Formation F reflect a stress direction towards W or N W (Fig. 16), indicating a westerly ice flow direction during deposition of till G. This flow direction was crosswise to the fjord, and suggests a thick glacier that moved more or less independently of the topography (Mangerud et al., 1992). Two fold axes (Fig. 16) reflect an ice flow parallel to the fjord, but the age relation between these and the westerly directions has not been determined.

Formation H (Holocene) This is the uppermost and youngest formation, and it has been mapped more or less continuously along the


J. Mangerud and J.I. Svendsen

coastal cliff of Kapp Ekholm (Fig. 7). Formation H is thickest in the northernmost described section (Section VI) and further towards the north (Figs 4 and 6). Our field mapping of the boundary between this and older formations was confirmed by radiocarbon dates, and the lower boundary is confidently identified in all sections. At two localities there occur distinctive beds of angular gravel and stones at the base of the formation (280 m, Fig. 15 and 650 m, Fig. 17). The clasts were apparently not striated or abraded by glacial ice, but rather they resemble frost-shattered and weathered material. In Section II the overlying silt has partly filled in the pores between the gravel particles, which gives the sediment a diamicton-like appearance. This layer of sharp-edged gravel, mixed with reddish brown silt, was interpreted as a terrestrial collovium by Lavrushin (1969) and Troitsky et aL (1979). According to our interpretation the gravel beds are scree deposits that either slid from the valley side or from the ice front during deglaciation. The angular gravel is covered by a bed of reddish brown silt which is the oldest post-glacial sediment along most of the sections (Fig. 17). In general the silt bed is thin, nearly everywhere it is less than 0.5 m. At some sites it is clearly laminated, especially at the base. Most radiocarbon dates from the silt gave ages between 9700 and 9900 BP (Fig. 17, Table 1). We postulate that the silt was deposited from suspension during the high relative sea level soon after the deglaciation. There is an even transition from the silt to a sequence of sand above. The boundary is probably time transgressive, depending on water depths and distance from




the shore. In some sections there is a crude bedding parallel with the lower boundary. The sand is less silty than the marine diamicton in Formation B, but otherwise it is similar, with oversized clasts, abundant molluscs and stones with a cover of calcareous algae. Radiocarbon ages of 9500-9700 BP were obtained on shells that were sampled from the base of the sand unit. Higher up the sand is densely populated by very thick and large Mya truncata, which were radiocarbon dated between 8500 and 9100 BP. We assume that these large Mya were favoured by some local environmental factor at the site. It should be noted that at several places the lower part of the Holocene sequence is missing. At these sites the lower boundary may be recognized by distinct rows of large shells of Mya truncata (for example at 550 m and 740 m, Fig. 17). Above the sand facies there is a continuous sequence of gravel foresets, with a strike nearly 90 ° to shore, and dip towards the north (Fig. 18). The top of the gravel is represented by a distinct terrace level 25-30 m a.s.l.

Nidedalen The described section is from a distinct marine terrace (37 m a.s.1.) just to the north of Nidedalen, across the fjord from Kapp Ekholm (Fig. 3). At this site a bedrock cliff is covered by a 1.5 m thick, reddish brown, silty diamicton interpreted as a basal till (Fig. 19). The source for the reddish brown colour is probably red Devonian sandstones which outcrop on this side of the fjord. Above the till there is a 1.5 m thick sand unit overlain by a 10 m thick sequence of gravel foresets. Apparently there is an unconformity between the sand and gravel. The dip direction (SE) of

Section I i 130 m


I 670 m

740 m

650 m

550 m

280 m

160 m


O O . ::Q:. { : -.:i :~r ::'

i .:.i.-:

.':::." .:'.



2::.'.'::t 890 m li~"~.~ }'iI..: :'.:.'.' [:.i~!(:~i.!.ilt

~::i.~'.'.:::-~': :;..: 0°0 0 0 0 ~:.ii:.'!:-{.: Laminated o oo :_.~__"---~o O Zirphea:


~L._.L.A Lm!nintted I='----~--I~.9760




8560 oo



• •




at 300 m:

/ii~.i 8930 :.:....,.

L :. ::" ' 9690

e~ ....,.



Silt (Massive except where marked) Sand Gravel









."/.':'.-"i i ".:.):".. ' :z.x.z':2 O O





0 0o



compo6ite stratigraphy

110 m

-.:0:.".i 9730 -





-0 0


dl, 000 On


Angular gravel Till Very large Mya truncata in living position

FIG. 17. Sediment logs showing the lower part of Formation H (Holocene) in some of the sections. The location of each log is marked by section numbers and horizontal distance on top of each bar (compare with Fig. 7). The boundary between Formation H and the older formations (identified by letters) is used as a reference level for all columns. Both the location and age of radiocarbon samples are marked (Table 1). The column to the right shows the composite stratigraphy.


Interglacial-Glacial Period on Spitsbergen


FIG. 18. Holocene gravel foresets between 940 and 990 m (Fig. 3), deposited by longshore drift. Note that the strike direction is nearly 90° to the shore, with a dip towards north. Thus the gravel foresets were deposited towards the mouth of the valley (Mathisondalen). Beneath the gravel there is a sequence of sub-horizontallybedded sand and silt. The section is 20-25 m high (not measured).

the foresets (Fig. 19) indicates that the sediments were deposited from a brook to the north of the section. They are certainly not delta foresets derived from the larger valley Nidedalen, which is situated to the south of the section. The gravel foresets are cut by a diamicton. The lower 50 cm of this diamicton is massive, hard and sandy. The upper 30 cm is silty, with a reddish brown colour. The diamicton could be traced along the entire slope, but due to sliding gravel from the steep slope above it was only studied in detail at one site. From the field observations it is concluded that the diamicton is a basal till. Above the till there is a thin reddish brown silt, followed by gravel foresets of Holocene age all the way to the top of the terrace (Fig. 19). The terrace is a raised delta which was deposited from a local brook. Due to the similar altitude and short geographical distance between the sections we would expect that it is possible to correlate the marine sediment units in Nidedalen and Kapp Ekholm by means of their measured D/L ratios. At least the O/L ratios obtained

from Nidedalen (Table 11) demonstrate that the sub-till sediment in this section is younger than Formations B and D at Kapp Ekholm. The ratios measured for both Mya truncata (0.034) and Hiatella arctica (0.053) at Nidedalen are higher than those obtained for Formation F at Kapp Ekholm, but they overlap within two standard deviations. Thus, the recorded D/L ratios suggest that the sub-till sediments at Nidedalen are contemporaneous with Formation F at Kapp Ekholm, or, alternatively, they are older. Radioepimerization (Bonner et al., 1978) is a possible cause for the higher D/L ratios at Nidedalen, because the radioactive level (measured for T L dating) is exceptionally high in that section. The correlation with Formation F is also supported by the T L dates from Nidedalen (Fig. 19). DISCUSSION

Depositional Development Our lateral correlations reveal a complicated geometry of the formations (Fig. 7), which is not surprising when considering the complex depositional


J. Mangerud and J.l. Svcndscn


m o.s.I. \ ~o~ _

Gravelup to terrace surface, 37m a-s.I.

°°0 Q

~9180_+120 (T-8328)

"~.o~4~ 20

°o°~" AA A

,0",,9~ ",-36,000_5,000 \o~°~ ~ (R-902508) ~,~o ~ -L 40,000__.4,000 ~ oo ~j ~" (R-892509) ~ .-10

k&,&&&A& )


Bedrock I



+4 000 43,3oo-z' ~oo (T-8239)


% FIG. 19. Stratigraphy of the section just north of the mouth of Nidedalen. Radiocarbon dates (T-No.) (Table 4) and TL dates (RNo.) (Table 5) are plotted on the profile.

and erosional history of the site. Glacial erosion during the different glacial advances has removed parts of the sediments, and during the Holocene the sediment body was cut by several deep gullies. The recorded facies succession (marine silt-sandgravel) shows shallowing water depth due to glacioisostatic uplift, and probably the section was exposed to air after deposition of each of the coarsening up sequences (Fig. 6). However, we did not find any soil horizons on top of the gravels. Possibly the loose terrestrial surface was stripped off by the advancing glaciers. Rust staining occurs in all the marine units, which Boulton (1979) interpreted as proof of terrestrial conditions after emergence of each unit. This may be correct and would fit our interpretation, but we did not see unambiguous evidence that the staining is not of the Holocene age. Boulton also reported a possible ice wedge cast at the top of our Formation D. Assuming terrestrial conditions, we would expect gully formation not only during the Holocene, but also during the older ice free periods. It is apparent from Fig. 7 that the sediments of Formations C and D at Section IV, infilled a depression that was incised into Formation B. This depression may either have been formed as a result of gully incision during the later part of interglacial B and/or erosion by the glacier that deposited Till C. It is noteworthy that the base of Formation D rises towards the north, suggesting that higher land (consisting of Formations B and C?) existed in that direction.

The long and steep gravel foresets in most formations resemble foresets as they appear in Gilbert type deltas. However, all undisturbed foresets dip towards north and have strike directions at nearly 90° to the shore (Fig. 20). In some of the Holocene terraces this strike and dip direction is mapped for long distances along the fjord, and even into the mouth of the valley Mathisondalen (Fig. 4). Thus, it is clear that the main transport and depositional direction was towards the north, demonstrating that the sediments were deposited as a result of longshore drift. The present day sediment budget is completely different because now the progradation of the fan-delta from Mathisondalen dominates (Figs 3 and 4). There are two reasons for this change in depositional pattern. Firstly the higher sea levels that existed during earlier parts of the Holocene imply that the delta in Mathisondalen was confined further up the valley. More important, however, is that the river in Mathisondalen presently carries a heavy bedload from glaciers up-valley, whereas during most of the Early and Middle Holocene probably no glaciers existed in that area (Svendsen and Mangerud, 1992), and we assume that the non-glacial streams carried very little sediment to the fjord (Svendsen et al., 1989). Regarding the Holocene terraces (Fig. 4), FeylingHanssen (1955, with references to earlier work) concluded that the gravel foresets were deposited by long shore drift, on the leeside of cuspate forelands. A detailed description and sedimentological interpretation of similar foresets in Denmark is given by Nielsen et al. (1988), who demonstrated they were deposited by spit progradation. If their model also is valid at Kapp Ekholm, then topset and beach deposits should occur on top of the pre-Holocene foresets. We assume that such deposits were removed by glacial erosion, as previously postulated for terrestrial surfaces on the beach deposits. Comparison with Previous Investigations The descriptions by Lavrushin (1967, 1969), Troitsky

_a_ y c-" .0

E o LL

Valley side Shore~ N


D 13


_..1__ I






Sectionno FIG. 20. Strike and dip directions of present and former shore facies. On the right hand side are shown the strike and dip directions of the present shore in front of the sections, and also the orientation of the main valley side behind the sections (compare Fig. 3). In the matrix to the left are shown the strike and dip directions for the gravel foresets in the different formations plotted for each section. All measured foresets reflect a depositional direction indicating longshore drift of gravel towards the north.

Interglacial-Glacial Period on Spitsbergen

et al. (1979) and us of Section II can be compared in great detail. Lavrushin (1969) states the two diamictons in Section II (our Tills A and C/E, Fig. 10) are without a doubt basal tills, referring to their texture, compactness, content of striated pebbles and fabric. However, there are two major differences between our interpretations and those of Lavrushin and Troitsky: (1) Apparently they did not discover the glaciotectonic thrusts in Formation F, even though Lavrushin (1969) described disturbances that he concludes were formed by stranded icebergs. (2) They (and Punning and Troitsky, 1984) concluded that only two tills and two marine sequences (including the Holocene) are present in the sections. One reason for this discrepancy may be that they did not investigate all the sections (Fig. 7) that we did. Boulton (1979) presented seven stratigraphic logs from different sites along the sections, which he (written communications, 1991) has kindly located into our measured profile (Fig. 7). His Logs 13 and 11 are from our Section I (120 and 160 m respectively); his Logs 9, 8 and 7 are from Section IV (500, 540 and 590 m respectively); and Log 6 is from Section V (680 m). He was not able to exactly locate Log 10 in our horizontal scale; according to his map it is probably from our Section II. Generally we agree with his description of the lithostratigraphy at each individual log site, although there are significant differences in the elevations and thickness of the units. We also agree with most of his sedimentological interpretations, particularly his interpretation of all grey diamictons as basal tills. He also interpreted the brown diamictons (and probably some brown silt) as basal tills, whereas we assume that at least some of that facies is a marine sediment. It should be noted that this latter disagreement does not influence the major conclusions of this paper. However, there are some major differences between Boulton's and our own stratigraphical interpretation of the sections. In Sections II and IV (Fig. 7), Boulton did not describe the glaciotectonic thrusts above the upper till in each section, and without radiocarbon dates he erroneously ascribed Formation F to the Holocene. Thus, he postulated that only three glaciations and four marine phases (including the shells in Till A) could be inferred from the sections compared to the four glaciations and five marine phases in our composite stratigraphy (Fig. 6). Boulton correlated all gravel beds into two separate units: (1) Holocene terrace gravel and (2) an older beach gravel (his Unit 5) covered by till. In the latter unit he included the gravel members of our Formations B and D, and also Formation F where this formation is covered by till (Section V, Fig. 7). Boulton, who did not record Mytilus edulis in the sections, concluded that the derived shells in Till A (Fig. 6) are of Eemian interglacial age on the basis of the amino acid O/L ratios. In contrast, we argue that the Eemian is represented by Formation B above Till A, an interpretation based mainly on the discovery of Mytilus in this formation.


The Age and Environment of the Interglacials and Interstadials We have not undertaken detailed paleontological investigations, but rather noted easily identifiable fossils during the field work. Interglacial B (the Eemian). As mentioned earlier, Mytilus edulis requires warmer water than the present day coastal water along western Spitsbergen (FeylingHanssen, 1955). Nevertheless, the sea surface water off western Spitsbergen is warmer than anywhere else at such high latitudes, due to the heat transport into the Norwegian Sea by the North Atlantic current (Fig. 1). Mytilus edulis from NW Europe is limited to the Atlantic current proper (Mangerud, 1977; Peacock, 1989), and probably cannot live around Spitsbergen except during periods when the North Atlantic current was slightly warmer and/or stronger than today. This implies a climatic pattern where NW Europe also was as warm or even warmer than at present. Thus, Interglacial B cannot be correlated with the Brc~rup and Odderade interstadials (Isotope Stages 5c and 5a respectively (Mangerud, 1991a)), which in NW Europe were cooler than at present (Menke and Tynni, 1984; Behre, 1989; Zagwijn, 1989). Kellogg (1977) concluded that Isotope Stages 1 (the Holocene) and 5e were the warmest pulses in the northern Norwegian Sea during the last 450 ka. This alone suggests that Formation B was deposited during Isotope Stage 5e. Several subsequent studies (Belanger, 1982; Ramm, 1988; Larsen and Sejrup, 1990) have confirmed that the surface water in the Norwegian Sea was warmer during Isotope Stage 5e than during later periods. The TL and OSL dates from Formation B (Fig. 7, Table 5), yielded somewhat younger ages than would be expected (120-130 ka) if it was deposited during Isotope Stage 5e (Edwards et al., 1987) which corresponds with the Eemian interglacial in Europe (Mangerud, 1989). However, younger ages than expected have also been obtained from Eemian and Early Weichselian samples from NW Europe by the Nordic Laboratory for Luminescence Dating (Kronborg and Mejdahl, 1989). According to V. Mejdahl (oral commun., 1991) a systematic error has recently been discovered that may have caused these slightly younger ages. With the constraint that Formation B was deposited during an interglacial, the TL dates certainly suggest that this was the Eemian and not an older interglacial. The amino acid O/L ratios cannot be used directly for age estimates in this cold environment. However, as will be discussed later, they can be used to constrain the duration that the site has been covered by glacial ice and/or inundated by the sea. The mean D/L ratio of formation B suggests that the integrated duration of all subsequent glaciations lasted 50 ka or less. This indicates that Formation B cannot be from an earlier interglaciation (Isotope Stage 7 or older), which is the only feasible alternative to the proposed Eemian age.


J. Mangerud and J.l. Svendscn

We conclude that Interglacial B correlates with the Eemian interglacial of Europe and with oceanic oxygen Isotope Substage 5e, when coastal waters off western Norway were .slightly warmer than at present (Mangerud et al., 1981).

Phantomodden interstadial/Formation D. The mean amino acid D/L ratios (0.063 + 0.013) for Formation D (Table 8) overlap with the ratios (0.069 _+ 0.008) from Formation B within one standard deviation, showing that the interval that glaciers and/or sea water covered the site during the intervening period was very short. Equally, the T L dates from the two formations overlap, suggesting that the Phantomodden interstadial is only slightly younger than Interglacial B. If one accepts an Oxygen Isotope Substage 5e age for Interglacial B, then the Phantomodden interstadial should also be of Stage 5 age, probably Substage 5c, or even older. The frequent occurrence of molluscs, including large Mya truncata, Hiatella arctica, Macoma calcarea, and Lepeta sp., reflect partly open water conditions in the fjord, at least during the summer months.

Kapp Ekholm interstad&l/Formation F. Eight radiocarbon dates from Formation F yielded ages between 36 and 49 ka BP (Table 2), and one sample yielded an apparent age of 51,400, or > 45,000 (T8324) with the 2o criterion. All ages obtained from Formation F are so close to the limit of the radiocarbon method that even very minor 'young' carbon contamination of the non-finite aged shells may have caused the cited apparent ages. To test the contamination levels in shells from the section, we dated five samples of shells from Interglacial B (Table 3). Four of these samples (T8535, T-9644, T-9645, and T-9646) yielded considerably higher apparent ages than all samples from Formation F. The last sample (T-8321) yielded older age than all the samples from F, except T-8324 mentioned above. The high apparent ages of the interglacial samples show that the contamination level is extremely low, and may suggest that the finite ages of Formation F are real, and not due to contamination by 'young' carbon. The absence of contamination may be due to the sparse vegetation, so that very little carbon is introduced to the subsoil, and the fact that the shells have been sealed in permafrost during most of the time. Six, from a total of nine, T L and OSL dates yielded ages in the range 40--60 ka BP (Table 5), thus supporting the radiocarbon dates. The three T L and OSL dates from Nidedalen (Table 5) also yielded ages around 40 ka BP, supporting the assumption that Billefjorden was ice free at this time. The difference between the mean amino acid D/L ratios for Mya truncata from Interglacial B (0.069 _+ 0.008) and the Kapp Ekholm interstadial (Formation F; 0.026 _+ 0.004) corresponds to a difference in time of 50 ka by assuming a mean annual temperature of 0°C in the sediments. This is a minimum estimate of the time

gap, because the shells were exposed to much lower temperatures when the sediment was not covered by sea water or glacial ice, causing a much slower epimerization reaction during such periods (Fig. 21). Accepting that Interglacial B ended around 120 ka, this provides a maximum age of 70 ka for the onset of the Kapp Ekholm interstadial. The radiocarbon dates give a secure minimum age of 40 ka BP for Formation F. Based on the TL-, OSL-, 14C dates, and the amino acid D/L ratios, we conclude that the real age for Formation F is between 40 and 50 ka BP. Due to the low D/L ratios (0.026 -- 0.004) for shells from Formation F, we postulate that there was a long and cold terrestrial phase after deposition of this formation, which implies that the Kapp Ekholm interstadial lasted much longer than indicated by the youngest radiocarbon dates. This is indeed compatible with results from the south-western Barents Sea, where many radiocarbon dates on shell fragments incorporated in till, yielded ages between 21 and 43 ka BP (Hald et al., 1990; Elverhoi et al., submitted for publication). Formation F contains a richer fauna than Formation D. There are many stones completely covered by calcareous algae, and we also found seaweed, large bones of a Greenland whale (Balaena mysticus), and many individuals of the mollusc Chlamys islandica. The latter presently lives along the western and northern coasts of Svalbard (Feyling-Hanssen, 1955; Wiborg, 1963), and thus it can tolerate colder water than Mytilus. However, in Greenland the present and subfossil occurrence of Chlamys islandica shows only a slightly more northerly distribution than for Mytilus (Hjort and Funder, 1974; Funder and Fredskild, 1989). The fossil content of Formation F indicates that during its deposition, the northern part of the Norwegian Sea must have had open water at least during the summer months.


0 0


50,000 I






100,000 II







00 i

FIG. 21. Rates of amino acid eplmerization for Mya truncata at different temperatures according to Miller (1985). The curves start at a O/Lratio of 0.011 which is the ratio obtained for livingshells. To the left are plotted mean O/L ratios (+ one standard deviation) for Formation B and F. The intercepts of the O/Lratios with the 0°C line are shown for Formation B, and, on the horizontal scale, the corresponding ages indicated by these intercepts. This graph indicates that the shells in Formation B would have obtained the recorded D/L ratios if they had experienced 0°C for 54-72 ka.

Interglacial-Glacial Period on Spitsbergen

The Holocene. There is evidence to suggest that during the climatic optimum the glaciers on Svalbard were much smaller than at present or even absent (Svendsen arid Mangerud, 1992). Within the Holocene formation (H) there occur several mollusc species that are now absent from Svalbard because the present day sea water temperature is too cold (Feyling-Hanssen, 1955). We dated Mytilus edulis to 8930 + 70 BP (Fig. 17, Table 1), and the even more thermophilous species Zirphea crispata to 8610 + 120 (Table 1). In an attempt to determine the duration of the Holocene warm period we dated Mytilus edulis from a terrace 7 m a.s.l, at Kapp Murdoch, Sassenfjorden. The result was 4200 _+ 40 BP (T-8334). The Glacial History of Svalbard and N W Barents Sea Number of glacial cycles. At Kapp Ekholm there are four basal tills in superposition, separated by marine sediments (Fig. 6). The tills reflect the local history, namely that a glacier overran the site. A more regional glacial history can be deduced from the succession of the marine sediment facies, because they reflect a significant glaCioisostatic depression and rebound of the crust. Based mainly on the geographical pattern of uplift around Svalbard (Forman, 1990), Mangerud et al. (1992) concluded that the last glacioisostatic rebound at Kapp Ekholm was caused by the vanishing Barents Ice Sheet. In conformity with this model we postulate that the high relative sea levels succeeding each of the older glaciations were also caused by isostatic depression from similarly large Barents Ice Sheets (Landvik et al., 1992). The glacioisostatic rebound, as shown by the coarsening upwards marine sequences, suggests that most of the Barents Ice Sheet disappeared during each of the ice free periods. On the other hand, during the Phantomodden and Kapp Ekholm interstadials, the global eustatic sea level was lower than at present (Shackleton, 1987). Thus, the crust might have been glacioisostatically depressed to a certain degree, even though the relative sea level at Kapp Ekholm was lower than today. Nowhere on Svalbard have more than three glacial advances during the Weichselian been recorded. On BrCggerhalv~aya, Miller et al. (1989) recorded only one till bed above sediments of assumed Eemian age, and they concluded that the site has remained ice free since the Early Weichselian. Even though Mangerud et al. (1992) argued that BrCggerhalv~ya had been overridden by glaciers during the Late Weichselian maximum, the stratigraphy and morphology (Forman and Miller, 1984) of this area certainly indicates a very limited number of Weichselian glaciations. The three glacial cycles recorded at Kapp Ekholm are a secure minimum number of glacial cycles which have affected Svalbard and northwestern parts of the Barents Sea during the Weichselian. The small difference in amino acid D/L ratios between Formation F and the Holocene makes it unlikely that more than the Late


Weichselian glaciation occurred between the two ice fre e periods. Since Formation F monitors the isostatic depi'ession of the glaciation E, we conclude that Till E represents the last glaciation before the Late Weichselian. Likewise, it seems unlikely that another glacial cycle is missing between Interglacial B and Formation D, since the D/L ratios for these periods overlap completely. We conclude that the only stratigraphic level where a glacial cycle might possibly be missing is the time interval between Formation D and Till E (Fig.

6). Duration of the glacial events. To a certain degree the amino acid D/L ratios can be used to constrain the duration of the different glaciations in this cold region (Mangerud et al., 1992) because the epimerization rate is highly temperature dependent (Miller and BrighamGrette, 1989) (Fig. 21). During deposition of the till beds, and when inundated by the sea, the ground temperature is assumed to be close to 0°C. As seen from Fig. 21 the D/L ratio (0.069 _+ 0.008) obtained from the interglacial Formation B would have been attained within 55-70 ka, if the diagenetic temperature remained at 0°C all the time. The sediments were inundated by the sea at least four times, and a reasonable estimate for the total duration of the marine inundation is 20 ka. If correct, the integrated period of ice cover since the last interglacial was 50 ka at the most. The calculated duration of ice cover at the site is independent of the number of glaciations. Therefore, if there were more glacial advances than we postulated, the result would only be that each glaciation was shorter lived. The age calculations pre-suppose that all the epimerization took place during periods when the site was either covered by glacial ice or inundated by the sea, and thus that the annual temperature for the intervening terrestrial periods were so low (-15°C or colder, Fig. 21) that practically no epimerization took place. However, it is likely that some epimerization also occurred during the terrestrial periods. If so, the duration of the periods when the sites were covered by glacial ice could have been less than the estimated 50 ka. On the other hand, the ground temperatures were probably well below 0°C when the overriding glaciers were thin. During such stages of the glaciations the epimerization reactions would have been slower, thus the duration of periods with ice cover is underestimated. As mentioned above there are considerable uncertainties in these estimates. Obviously the simplified assumptions of the temperature history, assuming temperatures either around 0°C or extremely cold (< -15°C), produce limitations. Another, and probably more important factor, is that the slow rate of epimerization at these low temperatures means that small uncertainties in D/L ratios cause large uncertainties in calculated time. For example, the mean O/L ratios for Mya truncata in Formation F are 0.021 from Section II


J. Mangerud and J.l. Svendsen

and 0.028 from Section V (Table 7). By using the 0 ° line at Fig. 21, the latter value would give a duration for the following glaciation that is 7600 years longer than if we used the former. Miller et al. (1989) described a site at B r0ggerhalv0ya where warm water foraminiferal faunas from parts of their episode C (D/L ratios of 0.044 + 0.004 for Mya and Hiatella) are correlated with the Eemian. These should therefore be contemporaneous with Formation B at Kapp Ekholm. The lower D/L ratios indicate that Br0ggerhalv0ya was covered by ice and/or inundated by the sea for about 20 ka less than at Kapp Ekholm. This is reasonable, considering that Br0ggerhalveya is situated much farther to the west, and that some of the glacial advances recorded at Kapp Ekholm may not have reached the west coast. Furthermore, the isostatic depression must have been greater in the inner fjord areas where the ice was much thicker, and for this reason Kapp Ekholm was probably submerged for longer time intervals than sites at similar altitudes on BrOggerhalvoya.

Till A, the Saalian glaciation. We postulate that Glaciation A is of Late Saalian age, simply because it directly underlies the Interglacial B (Eemian) coarsening upward sequence; we consider these two formations to represent one glacial cycle. The only other age constraints are the D/L ratios on shells from the lens in the till (Fig. 6, Table 10), which are considerably higher (0.121 + 0.009) than those obtained from shells of formation B (0.069 + 0.008).

Till C, the Early Weichselian glaciation. The t~/L ratios for Formations B (0.069 + 0.008) and D (0.063 ___ 0.013) overlap completely (Tables 8 and 9), indicating that Glaciation C, which occurs between the two formations, was a shortlived event. The T L dates (Table 5, Fig. 6) from Formation D indicate that Glaciation C occurred soon after the Eemian. Landvik et al. (1992) correlated Till C at Kapp Ekholm with the lower tills (Formations 1-2) at Skilvika and with the Episode B till on Br0ggerhalvoya (Miller et al., 1989), both of which are located at the western coast of Spitsbergen (Fig. 2). If correct, then this glacial advance must have terminated seaward of the coast. The correlation between Kapp Ekholm and Skiivika is mainly supported by TL dates from the marine units above the mentioned tills at the two sites. It cannot be ruled out that the tills at Skilvika are of pre-Eemian age (Lycke et al., 1992). For Episode B on Broggerhalveya an alternative correlation with the Kapp Ekholm section is discussed below. Evidently, the correlation between Kapp Ekholm and these sites from the western coast of Spitsbergen is problematic and cannot at present be established with certainty. Thus we cannot determine whether Glaciation C reached the western coast, or if this glacier advance was confined within the fjords.

Till E, the Middle Weichselian glaciation. We have previously concluded that the high sea level stand during deposition of Formation F represents the penultimate deglaciation (from Glaciation E) on western Svalbard, and that the deglaciation occurred around 50 ka BP. On that basis it can be correlated with a high sea level stand on the west coast which is represented by a till-covered marine terrace located at 87 m a.s.l, in Linn6dalen, at the mouth of Isfjorden (Fig. 22) (Mangerud et al., 1992). Actually, the argument for this terrace being from the penultimate deglaciation is even stronger than at Kapp Ekholm, since the D/L ratios (0.019 --+ 0.002) measured on shells from the terrace are barely higher than those (0.015 _+ 0.003) obtained from Holocene shells. The altitude of the terrace indicates that the marine limit at that time was higher than during the last deglaciation ( < 75 m a.s.1, in Linn6dalen), suggesting that the previous glaciation was as least as extensive as the Late Weichselian glaciation, and thus that it extended to the continental shelf to the west of Svalbard (Fig. 22). During the high sea level event represented by the 87 m terrace in Linn6dalen, the lower parts of





Kapp Bar.'Its E~ Sea



3O Kapp Ekholm


87rn terrace

interstadial (formation F~

~~ ~i~'J




701 - - - -



90 5b

Phantomodden interstadial -lOO 5C






5d o,



FIG. 22. Time-distance diagram for the advancing and retreating glacier terminus. The N W Barents Sea is located s o m e 200-300 km east of Kapp E k h o l m , and is therefore plotted outside the horizontal scale. For the Kapp Ekholm locality the climatostratigraphic units and formations are marked. For the Linntdalen site we have plotted the 87 m terrace and the till that covers the terrace (Mangerud et al., 1992). To the left are plotted recorded peaks of ice rafted material ( I R D ) in deep sea cores collected to the west of Svalbard, according to Hebbeln (1991, 1992). This is a s u m m a r y index without scale. Note that the ages of three of the glacier advances are supported by this deep sea chronology. To the extreme left are _plotted periods with m i n i m u m insolation (less than 2500 w per m z) at the top of the atmosphere for the m o n t h s June, July and A u g u s t at 80°N, and periods when the tilt of the Earth's axis was less than 22.5 ° . Both indices taken from Berger and Loutre (1991).

Interglacial-Glacial Period on Spitsbergen


northern part of the Norwegian Sea are assumed to reflect the extent of the glaciers on western Svalbard. The analysis revealed higher frequencies of ice rafted detritus during Isotope Stage 2 (10-25 ka BP) and from 50 to 60 ka BP. The pronounced IRD-peak in isotope stage 2 supports the argument that the frequency of IRD really reflects glacial fluctuations because it has been demonstrated that the glacier terminus reached the shelf at this time (Mangerud et al., 1992; Svendsen et al., 1992), and the age of deglaciation, which is well dated on Svalbard (Mangerud and Svendsen, 1990; Lehman and Forman, 1992), coincides with the upper end of the IRD curve. Both the suggested age and extent of Glaciation E are supported by the concentration of ice rafted detritus around 50--60 ka BP (Fig. 22). A major inflow of meltwater to the Norwegian Sea at that time is also suggested from a dark layer and light isotope peaks of ~180 and ~)13C (Vogelsang, 1990; Henrich, 1990). The Late Saalian glaciation (Fig. 22) is also documented by concentration of ice rafted detritus by Hebbeln (1991). The fact that Hebbeln (1992) did not find a peak of ice rafted detritus that could be correlated with Till C Fig. 22), may be interpreted in two ways: (1) Glaciation C did not reach the shelf, even though the pattern of Till G, Late Weichselian glaciation. This glaciation is glacioisostatic rebound shows that it was more than a discussed in detail by Mangerud et al. (1992) and local event. As discussed above, we are not able to Svendsen et al. (1992) who demonstrated that the ice demonstrate that this glaciation reached the shelf west front during the Late Weichselian glacial maximum of Svalbard. (2) The pattern of iceberg drifting and reached the continental shelf; it was probably located melting was different during Glaciation C than during along the shelf edge (Fig. 22). The difference in D/L Glaciations E and G. We favour the first alternative ratios between Formation F (Mya; 0.026 + 0.004) and which means that the tills at Broggerhalvoya and the early Holocene (Mya; 0.015 + 0.003) suggests that Skilvika, which indicate that the ice reached the shelf, Glaciation G lasted 10-15 ka in the inner fjord areas should not be correlated with Till C at Kapp Ekholm. (Mangerud et al., 1992). Elverh¢i et al. (submitted for publication) recon- Orbital Forcing of the Glaciations struct a very similar Late Weichselian development for The supply of moisture is a critical factor for the the south-western flank of the Barents Ice Sheet, and growth of glaciers on Svalbard and in high arctic areas provide radiocarbon dates that support a re-advance in general (Svendsen and Mangerud, 1992). However, after 22 ka BP in that area. it has been demonstrated that a large ice sheet was indeed formed on Svalbard and the Barents Sea during Glaciation records in deep-sea cores off Svalbard. the coldest phase of the Weichselian, namely around 20 The chronology for the glaciation curve (Fig. 22) is ka BP (Mangerud et al., 1992; Svendsen et al., 1992; entirely based on the Kapp Ekholm section, except for ElverhOi et al., submitted for publication). Therefore the Late Weichselian glaciation. The western limit of enough snow fall for ice sheet formation was available the glacier terminus during each glaciation is based during that period, contrary to what Larsen et al. (1992) upon correlations with the more coastal sites. How- postulated. ever, as discussed above, this correlation is problemaAs appears from the glaciation curve (Fig. 22), three tic. distinct phases of glaciation occurred on Svalbard Hebbeln (1991, 1992) provided an independent way during the Weichselian, and they all occurred close to of testing both the age and partly the geographical periods with minimum insolation at 80°N (Berger and extent of the glaciations on Spitsbergen. In deep-sea Loutre, 1991) (Fig. 22, left part). At this latitude the cores collected west and north-west of Svalbard, he variation of insolation is first and foremost determined recorded ice rafted material (IRD, Fig. 22, left side) by the variation of the tilt of the Earth's axis, and the that he could partly trace back to Svalbard by means of fact that the periods of ice sheet growth seems to occur its composition. The theory is that more icebergs will simultaneously with, or shortly after, the insolation drift out into the open ocean when the ice front lies on minima indicates that the major glacial events on the shelf than when the glaciers terminate in the fjords. Svalbard follow the 41 ka tilt periodicity (Fig. 22, left Thus the variations in ice drop frequency in the part). Apparently the last glacial maximum lagged a

Br¢ggerhalvCya, which is situated only 100 km to the north of Isfjorden, must have been inundated by the sea. The youngest pre-Holocene marine sediments recorded at Br~ggerhalvcya are together with the underlying till termed Episode B (Miller et al., 1989; Larsen et al., 1992), and this probably corresponds with the 87 m terrace in Linnrdalen, and thus Formation F at Kapp Ekholm (G.H. Miller. oral commun., 1991). This would imply that Episode B on Br~ggerhalv~ya represents Glaciation E at Kapp Ekholm, and not Glaciation C as proposed by Landvik et al. (in press). There is an apparent conflict in age assignments of the two events, as Miller et al. (1989) dated Episode B on Br~ggerhalv~ya to 70 + 10 ka BP, whereas we concluded that the deglaciation of Kapp Ekholm occurred around 50 ka BP. This may partly be due to a slow deglaciation from Br~ggerhalv~ya to Kapp Ekholm, or to an erroneous age estimation at either site. There is a major step in D/L ratios between Formations D and F (Fig. 6), indicating either that Glaciation E may have lasted as long as 40 ka, or that the section was submerged for a considerable part of that period. Alternatively, there have been more glaciations than recorded during that period, with an integrated duration of around 40 ka.



Mangerud and J.l. Svcndscn

few thousand years behind the insolation minimum, but the onset of this re-advance is poorly dated. The dating control is not precise enough to determine whether the older glacial advances were simultaneous or lagged behind the insolation minima. The indications that ice sheets at high northern latitudes are forced by the tilt frequency are important in understanding the ice age climate and may explain why the tilt frequency constitutes a significant element in the global ice volume signal (Imbrie et al., 1984). The Scandinavian Ice Sheet, which was situated at lower latitudes (52-70°N), fluctuated more frequently than the Barents Ice Sheet and was apparently, to a large extent, driven by the precession cycle of around 23 ka (Mangerud, 1991a,b; Larsen and Sejrup, 1990). Because of the more southerly location, the climate in southern Scandinavia was more strongly influenced by the precession variations (Ruddiman and Mclntyre, 1981), and during the warmest interstadials additional heat was imported from even lower latitudes. In contrast to the Weichselian, the Holocene climatic changes, such as the mid-Holocene warm period and the Little Ice Age, were parallel in western Scandinavia and Svalbard. This synchroneity is due to the fact that the climate in both regions was dominated by the warm Atlantic current during the Holocene (Svendsen and Mangerud, 1992). The main reason for the decoupling of the climates of the two areas during cold ice age periods is probably that the Atlantic current was much weaker during these periods and for most of the time did not influence the northern Norwegian Sea (Mclntyre et al., 1976). In such a situation the Svalbard climate became increasingly influenced by the Arctic high pressure, and consequently direct insolation became a more important factor for the climate than at present. We have not tried to model ice sheet growth at Svalbard, and here we will only point out some possible mechanisms that may explain our observations. The simultaneous occurrence of insolation minima and ice growth suggests in itself that summer melting of snow (ablation) is the main controlling mechanism. This also takes account of the albedo feedback mechanism. The large and rapid glacier fluctuations demonstrate that Svaibard is a sensitive area for glacial fluctuations. One reason for this is that if the glacial equilibrium line altitude (ELA) was lowered by only 300 m, the entire land area of the archipelago would be above the E L A , and thus be glacierized within a short time (Fig. 2). The length of the accumulation season is also important. During the coldest periods of the Weichselian the summers were probably so cool that hardly any melting occurred, and most of the summer precipitation fell as snow, causing year-round accumulation on the glaciers (Svendsen and Mangerud, 1992). Denton and Hughes (1981) postulated that Arctic ice sheets started to grow from the surface of ice shelves in shallow water. If this occurred, the glaciers would grow faster than by only expansion from the land area. A lowering of the global sea level would benefit glacier

expansion from the land areas of Svalbard into shallow water, due to less calving, and would also enhance formation of glaciers on the shallow bank areas in the northwestern part of the Barents Sea (Elverh~i et al., submitted for publication).

CONCLUSIONS (1) Four basal tills, separated by marine sediments, occur in the sections at Kapp Ekholm (Fig. 6), which are situated in the central part of Spitsbergen. (2) The vertical succession of facies is similar above each till, starting with marine mud (mainly silt), followed by sand and gravel. This coarsening upward sequence resulted from a falling relative sea level, due to glacioisostatic rebound. Thus each sequence of tillsilt-sand-gravel represents a major glacial-deglacial cycle. (3) The lower marine unit (Formation B/Interglacial B) in the sections of Kapp Ekholm contains the mollusc Mytilus edulis, which indicates warmer sea water than at present. Based on correlation with the deep sea record, TL dates, and, to some extent, amino acid D/L ratios, we conclude that this formation is Eemian/ Oxygen Isotope Substage 5e. (4) Four major glacial-deglacial cycles were found in the Kapp Ekholm sections, of which one is of Late Saalian and three are of Weichselian age (Fig. 6). The glacio-isostatic depression and rebound reflected by the marine sediments was probably caused by a contiguous ice sheet over Svalbard and the northern Barents Sea. (5) The inner fjord areas were deglaciated at least twice during the Weichselian. The first ice free period, the Phantomodden interstadial, is of Isotope Stage 5 age, and apparently lasted for 30-40 ka. The Kapp Ekholm interstadial started around 50 ka BP, and ended sometime after 30 ka BP. (6) The substantial isostatic rebound following each glacial retreat indicates that the Barents Sea became ice free during both the Phantomodden and Kapp Ekholm interstadials (Fig. 22). (7) A correlation between the stratigraphy at Kapp Ekholm and sites on the west coast is not straightforward; for the older events different correlation alternatives are possible. However, there are no strong indications that more major glacial advances occurred during the Weichselian than the three recorded at Kapp Ekholm. (8) Glaciation history (Fig. 22) is mainly based on the Kapp Ekholm section. Three of the four major glacial advances are supported by peaks of ice rafted material in the deep sea west of Svalbard. (9) The major glacial events on Svalbard correspond to the insolation minima which at this high latitude were dominated by the 41 ka tilt periodicity. The timing of glacial advances on Svalbard are partly different from southern Scandinavia, where glaciations to a large extent were driven by the 23 ka precession cycles.

Interglacial-Glacial Period on Spitsbergen

ACKNOWLEDGEMENTS The project was funded by grants from the Norwegian Research Council for Science and the Humanities (NAVF) and Statoil. The Norwegian Polar Research Institute provided logistical support. Asbjcrn Mangerud was the field assistant. Masaoki Adachi and Jane EUingsen made the drawings. Reidar Nydal, Steinar Gulliksen and G6ran Possnert provided radiocarbon dates, which were frequently discussed with S. Gulliksen. H.P. Sejrup provided amino acid D/L ratios, and Vagn Mejdahl the TL and OSL dates. Much of this manuscript was written while Jan Mangerud spent a sabbatical year at Institute of Arctic and Alpine Research, University of Colorado, and he thanks the faculty members of that institute. An earlier version of the manuscript was critically read by Jon Landvik and Lars Ronnert. The nearly final version was critically read by Anders Elverhcfi, Svend Funder, Gifford H. Miller and Otto Salvigsen. The English language was corrected by William Austin. We thank all persons and institutions mentioned. This is a contribution to the European Science Foundation project: Polar North Atlantic Margins, Late Cenozoic Evolution (PONAM).

REFERENCES Aitken, M.J. (1992). Optical dating. Quaternary Science Reviews, 11, 127-131. Belanger, P.E. (1982). Paleo-oceanography of the Norwegian Sea during the past 130,000 years: coccolithophorid and foraminiferal data. Boreas, 11, 29-36. Behre, K.E. (1989). Biostratigraphy of the last glacial period in Europe. Quaternary Science Reviews, g, 25-44. Berger, A. and Loutre, M.F. (1991). Insolation values for the climate of the last 10 million years. Quaternary Science Reviews, 10, 297317. Berger, A., Imbrie, J., Hays, J., Kukla, G. and Saltzman, B. (eds) (1984). Milankovitch and Climate, NATO ASI Series 126 (2 volumes). D. Reidel, Dordrecht. Bonner, W.A., Blair, N.E. and Lemmon, R.M. (1978). The radioracemication of amino acids by ionizing radiation: Geochemical and cosmochemical implications. In: Hare, P.E., Hoering, T.C. and King, K. Jr (eds), Biochemistry of Amino Acids, pp. 357-374. John Wiley, New York. Boulton, G.S. (1979). Glacial history of the Spitsbergen archipelago and the problem of a Barents Shelf ice sheet. Boreas, 8, 31-57. Boulton, G.S. (1990). Sedimentary and sea level changes during glacial cycles and their control on glacimarine facies architecture. In: Dowdeswell, J.A. and Scourse, J.D. (eds), Glacimarine Environments: Processes and Sediments, Geological Society Special Publication No. 53, pp. 15-52. The Geological Society, London. Boulton, G.S. and Rhodes, M. (1974). Isostatic uplift and glacial history in northern Spitsbergen, Geological Magazine, 111,481500. BCtter-Jensen, L., Ditlefsen, C. and Mejdahl, V. (1991). Combined OSL (infrared) and TL studies of feldspars. Nuclear Tracks and Radiation Measurements, lg, 257-263. Edwards, R.L., Chen, J.H., Ku, T.-L. and Wasserburg, G.J. (1987). Precise timing of the last interglacial period from mass spectrometric determination of Thorium-230 in corals. Science, 236, 1547-1553. Elverhoi, A., Fjeldskaar, W., Solheim, A., Nyland-Berg, M. and Russwurm, L. (1992). The Barents sea ice sheet - - a model of its growth and decay during the last ice maximum (submitted for

publication). Feyling-Hanssen, R.W. (1955). Stratigraphy of the marine late Pleistocene of Billefjorden, Vestspitsberen. Norsk Polarinstitutt Skrifter, 107, 1-186. Forman, S. (1990). Post-glacial relative sea-level history of northwestern Spitsbergen, Svalbard. Geological Society of America Bulletin, 102, 1580-1590. Forman, S. and Ennis, G. (1992). Limitations of thermoluminescence to date waterlain sediments from glaciated fiord environments of western Spitsbergen, Svalbard. Quaternary Science Reviews, 11, 61-70. Forman, S. and Miller, G. (1984). Time-dependent soil morphologies and pedogenic processes on raised beaches, BrOggerhalvOya, Spitsbergen, Svalbard archipelago. Arctic and Alpine Research, 16, 381-394.


Hald, M., Saettem, J. and Nesse, E. (1990). Middle and Late Weichselian stratigraphy in shallow drillings from the southwestern Barents Sea: foraminiferal, amino acid and radiocarbon evidence. Norsk Geologisk Tidsskrift, 70, 241-257. Hebbeln, D. (1991). Sp~tquartiire Startigraphie und Paiiozeanographie in der Fram-Strasse. Berichte, Fachbereich Geowissenschaften, Universit~it Bremen, No. 22, 174 pp. Hebbeln, D. (1992). Glacial history of the Svalbard area: Correlation of the marine and the terrestrial records. Boreas, 21 (in press). Henrich, R. (1990). Cycles, Rhythms, and events in Quaternary Arctic and Antarctic glaciomarine deposits. In: Bleil, U. and Thiede, J. (eds), Geological History of the Polar Oceans; Arctic Versus Antarctic, pp. 213-244. Kluwer, Dordrecht. Hibler, W.D. III (1989). Arctic ice-ocean dynamics. In: Herman, Y. (ed.), The Arctic Seas, pp. 47-92. Van Nostrand Reinhold, New York. Hjort, C. and Funder, S. (1974). The subfossii occurrence of Mytilas edulis L. in central East Greenland. Boreas, 3, 23-33. Imbrie, J., Hays, J.D., Martinson, D.G., McIntyre, A., Mix, A.C., Morley, J.J., Pisias, N.G., Prell, W.L. and Shackleton, N.J. (1984). The orbital theory of Pleistocene climate: support from a revised chronology of the marine d180 record. In: Berger, A., Imbrie, J., Hays, J., Kukla, G. and Salzman, B. (eds), Milankovitch and Climate, NATO ASI Series C, Vol. 126, Part 1, pp. 269-305. D. Reidel, Dordrecht. Kellogg, T.B. (1977). Paleoclimatology and paleo-oceanography of the Norwegian and Greenland seas: the last 450,000 years. Marine Micropaleontology, 2, 235-249. Kristiansen, K.J. and Sollid, J.L. (1987). Svalbard, Jordartskart, 1 : 1000 000. Nasjonalatlas for Norge, Geografisk Institutt, Universitetet i Oslo. Kronborg, C. and Mejdahl, V. (1989). Thermoluminescence dating of Eemian and Early Weichselian deposits in Denmark. Quaternary International, 3/4, 93-99. Landvik, J.Y., Bolstad, M., Lycke, A.K., Mangerud, J. and Sejrup, H.P. (1992). Weichselian stratigraphy and paleoenvironments at Bellsund, western Svalbard. Boreas, 21 (in press). Larsen, E. and Sejrup, H.P. (1990). Weichselian land-sea interactions: Western Norway-Norwegian Sea. Quaternary Science Reviews, 9, 85-97. Larsen, E., Sejrup, H.P., Olsen, L. and Miller, G.H. (1992). Late Quaternary land-sea interactions: Fennoscandia and Svalbard - The Nordic Seas. Quaternary International, 10-12, 151-159. Lavrushin, J.A. (1967). Pleistocene of Northern Spitsbergen. Doklady Akademii Nauka, SSR, 176(I), 167-170 (in Russian, internal English translation, The Norwegian Polar Research Institute). Lavrushin, J.A. (1969). Quaternary Deposits of Spitsbergen. Nauka, Moscow (in Russian, internal English translation, The Norwegian Polar Research Institute). Lehman, S.J. and Forman, S.L. (1992). Late Weichselian Glacier Retreat in Kongsfjorden, West Spitsbergen, Svalbard. Quaternary Research, 37, 139-154. Lindner, L. and Marks, L. (1991). Outline of stratigraphy of the Pleistocene and Holocene in south and central Spitsbergen. Bulletin of the Polish Academy of Sciences Earth Sciences, 39, 165172. LCnne, I. and Mangerud, J. (1991). An Early or Middle Weichselian sequence of proglacial, shallow marine sediments on western Svalbard. Boreas, 20, 85-104. Lycke, A.K., Mangerud, J. and Sejrnp, H.P. (1992). Late Quaternary foraminiferal stratigraphy from western Svalbard. Boreas, 21

(in press). Mangerud, J. (1977). Late Weichselian marine sediments containing shells, foraminifera and pollen at Agotnes, western Norway. Norsk Geologisk Tidsskrift, 57, 23-54. Mangerud, J. (1989). Correlation of the Eemian and the Weichselian with deep sea oxygen isotope stratigraphy. Quaternary International, 3/4, 13-19. Mangerud, J. (1991a). The Scandinavian Ice Sheet through the last interglacial/glacial cycle. In: Frenzel, B. (ed.), Klimageschichtlige Probleme der Letzten 130 000 Jahre, Pal~ioklimaforschung 1, pp. 307-330. G. Fisher, Stuttgart. Mangerud, J. (1991b). The last interglacial/glacial cycle in northern Europe. In: Shane, L.C.K. and Cushing, E.J. (eds), Quaternary Landscapes, pp. 38-75. University of Minnesota Press, Minneapolis. Mangerud, J. and Gulliksen, S. (1975). Apparent radiocarbon ages of recent marine shell from Norway, Spitsbergen and Arctic Canada. Quaternary Reseearch, 5, 263-273.


J. Mangerud and J.1. Svendscn

Mangerud, J. and Salvigsen, O. (1984). The Kapp Ekholm section, Billefjorden, Spitsbergen: a discussion. Boreas, 13, 155-158. Mangerud, J. and Svendsen, J.I. (1990a). The last interglacial/glacial cycle on Svalbard, as seen from the Kapp Ekholm section. In: Kotlyakov, V.M. and Sokolov, V.E. (eds), Arctic Research: Advances and Prospects, Part 2, pp. 63-66. Nauka, Moscow. Mangerud, J. and Svendsen, J.I. (1990b). Deglaciation chronology inferred from marine sediments in a proglacial lake basin, western Spitsbergen, Svalbard. Boreas, 19, 249-272. Mangerud, J., S0nstegaard, E., Sejrup, H.P. and Haldorsen, S. (1981). A continuous Eemian-Early Weichselian sequence containing pollen and marine fossils at Fjcsanger, western Norway. Boreas, 10, 137-208. Mangerud, J., Bolstad, M., Elgersma, A., Helliksen, D., Landvik, J.Y., L~nne, I., Lycke, A.K., Salvigsen, O., Sandahl, T. and Svendsen, J.I. (1992). The last glacial maximum on Spitsbergen, Svalbard. Quaternary Research, 38, 1-31. Mclntyre, A. et al. (1976). Glacial North Atlantic 18,000 years ago: A CLIMAP reconstruction. Geological Bulletin of America Memoir, 145, 43-76. Mejdahl, V. Thermoluminescence dating based on feldspars. Nuclear Tracks, 10, 133-136. Mejdahl, V. (1986). Thermoluminescence dating of sediments. Radiation Protection Dosimetry, 17, 219-227. Mejdahl, V. (1988). The plateau method for dating partially bleached sediments by thermoluminescence. Quaternary Science Reviews, 7, 347-348. Menke, B. and Tynni, R. (1984). Das Eeminterglacial und das Weichselfriihglazial von Rederstali/Ditmarschen und ihre bedeutung for die mitteleuropaische Jungpleistoz~in-Gliederungen. Geologisches Jahrbuch Reihe A, 76, 3-120. Miller, G.H. (1982). Quaternary depositional episodes, western Spitsbergen Norway: aminostratigraphy and glacial history. Arctic and Alpine Research, 14, 321-340. Miller, G.H. (1985). Aminostratigraphy of Baffin Island shellbearing deposits. In: Andrews, J.T. (ed.), Quaternary Environments, pp. 394-427. Allen & Unwin, Boston. Miller, G.H. and Brigham-Grette, J. (1989). Amino acid geochronology: Resolution and precision in carbonate fossils. Quaternary International, 1, 111-128. Miller, G.H. and Mangerud, J. (1986). Aminostratigraphy of European marine interglacial deposits. Quaternary Science Reviews, 4, 215-278. Miller, G.H., Sejrup, H.P., Lehman, S.J. and Forman, S.L. (1989). Glacial history and marine environmental change during the last interglacial-glacial cycle, western Spitsbergen, Svalbard. Boreas, 18, 273-296. Nielsen, L.H., Johannessen, P.N. and Surlyk, F. (1988). A Late Pleistocene coarse-grained sequence in northern Jylland, Denmark. Sedimentology, 35, 915-937. Nystuen, J.P. (1989). Rules and recommendations for naming

geological units in Norway. Norsk Geologisk Tidsskrift Supplement 2, 69, 1-111. Peacock, J.D. (1989). Marine molluscs and Late Quaternary environmental studies with particular reference to the late-glacial period in northwest Europe: A review. Quaternary Science Reviews, 8, 179-192. Punning, J.M. and Troitsky, L. (1984). Comments (to the paper: Mangerud, J. and Salvigsen, O., 1984). Boreas, 13, 158. Ramm, M. (1989). Late Quaternary carbonate sedimentation and paleo-oceanography in the eastern Norwegian Sea. Boreas, 18, 255-272. Ruddiman, W.F. and Mclntyre, A. (1981). Oceanic mechanisms for amplification of the 23,000-year ice-volume cycle. Science, 212, 617-627. Salvigsen, O., Lauritzen, 0. and Mangerud, J. (1983). Karst and karstification in gypsiferous beds in Mathiesondalen, central Spitsbergen, Svalbard. Polar Research, 1, 83-88. Sejrup, H.P. and Larsen, E. (1992). Eemian-Early Weichselian N-S temperature gradients; North Atlantic-NW Europe. Quaternary International, 10-12, 161-166~ Shackleton, N.J. (1987). Oxygen isotopes, ice w~lume and sea level. Quaternary Science Reviews, 6. 183-190. Streeter, S.S., Bclanger, P.E., Kellogg, T.B. and Duplessy, J.C. (1982). Late Pleistocene paleo-oceanography of the NorwegianGreenland Sea: benthic foraminiferal evidence, Quaternary Research, 18, 72-90. Stuiver, M. and Polach, H.A. 11977). Reporting of ~4C data. Radiocarbon, 19, 355-363. Svendsen, J.I. and Mangerud, J. (1992). Paleoclimatic inferences from glacial fluctuations on Svalbard during the last 20,000 years. Climatic Dynamics, 6, 213-220. Svendsen, J.l., Mangerud, J. and Miller, G.H. (1989). Denudation rates in the Arctic estimated from lake sediments on Spitsbergen. Paleogeography, Paleoclimatology and Paleoecology, 72, 153-168. Svendsen, J.I., Mangerud, J., ElverhOi, A., Solheim, A. and Schiittenhelm, T.D. (1992). The Late Weichselian glacial maximum on western Spitsbergen inferred from offshore sediment cores. Marine Geology, 104, 1-17. Troitsky, L.S., Punning, J.M., Hiitt, G. and Rajam~ie, R. (1979). Pleistocene glaciation chronology of Spitsbergen. Boreas, 8,401407. Vogelsang, E. (1990). Pal~ieo-Ozeanographie des Europ~iischen Nordmeeres an Hand stabiler Kohlenstoff- und Sauerstoffisotope. Berichte Sonderforschungsbereich 313, 23, 1-136. GeologischPal/iontologisches Institut der Universit~it Kiel. Wiborg, K.F. (1963). Some observations on the Icelandic scallop Chlamys islandica (MiiUer) in Norwegian waters. Fiskeridirektorates Skrifter Serie Havundersckelser, 13(6L 38-53. Bergen. Zagwijn, W.H. (1989). Vegetation and climate during the warmcr intervals in the Late Pleistocene of western and Central Europe. Quaternary International, 3/4, 57-67.