Younger Dryas Glaciolacustrine Rhythmites and Cirque Glacier Variations at Kra˚Kenes, Western Norway: Depositional Processes and Climate

Journal of Paleolimnology 31: 49–61, 2004. # 2004 Kluwer Academic Publishers. Printed in the Netherlands. 49

Younger Dryas glaciolacustrine rhythmites and cirque glacier variations at Kra˚kenes, western Norway: depositional processes and climate

Eiliv Larsen1,* and Martha K. Stalsberg1,2 1Geological Survey of Norway, N-7491 Trondheim, Norway; 2Current address: Churchills vei 8c, N-7058 Trondheim, Norway; *Author for correspondence (e-mail: [email protected])

Received 13 December 2002; accepted in revised form 15 July 2003

Key words: Western Norway, Younger Dryas, Glaciolacustrine, Rhythmite, Cirque glaciation, Climate change

Abstract

Kra˚kenes faces the open sea on the west coast of Norway. During the Younger Dryas a cirque glacier deposited a large outer and a smaller inner moraine in a cirque at the site and melt-water entered a small lake depositing glaciolacustrine sediments. The glaciolacustrine succession can be divided into three sub-units corresponding to the advance, the still-stand and the retreat phases of the glacier. The sediment succession contains both varves and other types of rhythmites, the latter being mainly deposited as turbidity underflows caused by localized slumping events. Lee-side accumulation of snow by wind and avalanching into the cirque was crucial to form and maintain the cirque glacier once summer temperatures were low enough. At maximum, the glacier likely was in equilibrium with climate. The initial retreat from the maximum position might have been triggered by fall-out of volcanic ash from Iceland, but the continued retreat was due to increased ablation season temperatures. The most rapid change in climate at the Younger Dryas/Preboreal transition occurred after the cirque glacier had melted away completely.

Introduction The Late-glacial and Holocene sediment succession in this lake has been extensively studied with The Younger Dryas cooling at the end of the respect to paleobiology, environmental and climate Pleistocene is well marked both in terrestrial and change, and chronology (Birks et al. 1996a, b, marine records in the NW Atlantic region (e.g., 2000; Gulliksen et al. 1998), but a detailed sedimen- Bjo¨rck et al. (1996)). The Scandinavian ice sheet tological study has not been performed as yet, responded to this cooling by either a halt or a read- although the Younger Dryas sequence in the lake vance (Andersen et al. 1995a, b). In many areas of was identified as glaciolacustrine and it was specul- the west coast of Norway, outside the ice sheet ated whether the rhythmites could be annual margin, cirque glaciers advanced or were formed (Larsen and Mangerud 1981; Larsen et al. 1984). during the Younger Dryas (Mangerud et al. 1979; This is important because the duration of glaciola- Larsen and Mangerud 1981; Larsen et al. 1984, custrine sedimentation is an approximation of the 1998). A small cirque glacier at Kra˚kenes formed duration of glacier occupation in the cirque above and disappeared during the Younger Dryas, a con- the lake. clusion reached by coring and dating below and In this paper we present a study of the glacio- above the corresponding glaciolacustrine sedi- lacustrine sediments found in the Kra˚kenes ments in a small lake just outside the end moraine Lake trying to distinguish between annual and (Larsen and Mangerud 1981; Larsen et al. 1984). other types of rhythmites. Variations in style of 50

Figure 1. (A) Map of Norway. Kra˚kenes is located in the outer (western) part of the framed area. (B) Southern Norway with shading showing the area covered by the Younger Dryas ice sheet (after Mangerud et al. 1979). The study site, Kra˚kenes is indicated. (C) Map of the area with cirque moraine, melt-water channel and lake at Kra˚kenes. Slightly modified from Larsen and Mangerud (1981). sedimentation to the lake during the Younger was originally 0.07 km2, 530 m long and had a Dryas are linked with cirque glacier variations maximum depth of 14 m (Larsen and Mangerud based on morphological evidence, and finally the 1981) (Figure 2). The water depth today in the climatic conditions during advance, still-stand, and threshold area is <6 m (Figure 2). Today the lake retreat of the cirque glacier are discussed. area is reduced due to artificial lowering of 2 m some 90 years ago. A melt-water channel cuts through the outer cirque moraine (Larsen and Mangerud 1981), and runs into the proximal basin Study site of the lake (Figures 1 and 2). Due to the bedrock topography, this was the only inlet to the lake Kra˚kenes is located at the outermost coast of during the Younger Dryas. We describe sediments western Norway (62 020 N; 5000 E, Figure 1). in the proximal basin, the threshold area and the The yearly mean temperature is 7.3 C and yearly southern basin (Figure 2). The southern basin precipitation is 1280 mm. The Kra˚kenes Lake is forms a branch of the lake separated from the situated 38 m above sea level, that is some 30 m main inlet–outlet flow direction and protected above the marine limit (Longva et al. 1983). The lake behind shallow thresholds (Figure 2). 51

Figure 2. Bathymetry of the Kra˚kenes Lake, showing conditions prior to lacustrine sedimentation. The mapping was based on more than 40 reconnaissance cores and the 110 mm cores used in this study referred to by numbers. The northern part of the lake has not been mapped.

The Kra˚kenes area was deglaciated from the ice Water-content, and TOC% (total organic sheet shortly before 12 300 14C yr. BP (Larsen and carbon) were measured at 2 cm intervals on core Longva 1979; Mangerud et al. 1979; Larsen et al. 502À66 (Figure 3). TOC% was determined on 1984, 1998). During Younger Dryas a small glacier acid leached sediment-samples using a Leco com- developed in the cirque above the lake and depos- bustion furnace. Grain size distribution was deter- ited a prominent marginal moraine with a maxi- mined for selected samples on the same core using mum distal height of 16 m (Larsen et al. 1984). sieves and SediGraph, Model 5100 for the <63 m Within the main moraine ridge on the eastern fraction. side of the cirque, there is a smaller end moraine A plexi-glass tray (1–1.5 m long, 6 mm deep) was (Figure 1) marking a halt or re-advance during the pressed onto a split sediment core and a thin slice of deglaciation of the cirque (Larsen and Mangerud the sediment was transferred to the tray by running 1981). The equilibrium line altitude (ELA) for the a fine wire down the length of the tray. The plexi- cirque glacier at maximum position was calculated glass trays with sediment were X-rayed, and the to 154 m a.s.l. (Larsen et al. 1984), and ELA for the photos were used for laminae-counts. Forty sedi- inner glacier position is estimated to 200 m a.s.l. ment slices (11 Â 2 Â 0.6 cm) were further extracted from the plexi-glass trays from 4 cores (51, 53, 57, and 66; Figure 2) for thin-sections studies. The Methods samples were freeze-dried and impregnated using the methods of Merkt (1971) before thin sections The sediment series was cored by a 110 mm dia- were made. Thin sections were studied under a meter modified GEONOR piston corer operated petrographic microscope using magnification from a raft on the lake and a rig on the marsh. between 1.4 and 50. Cores were split in two halves in the laboratory, The thickness of the laminae was measured on one for descriptions and sub-sampling, the other cores 51, 53, 59, and 66 (Figure 2). In order to stored as a reference. quantify the ratio of the coarse- to fine-grained 52

Figure 3. Lithology, grain size distribution, and TOC% of core 502À66. Units and sub-units used in the text are marked. Confer Figure 2 for core location. 53

Figure 4. Rates of sediment accumulation for unit F. Calculations are based on the calendar time scale of core 502À46 developed by Gulliksen et al. (1998) and Birks et al. (2000). part of a rhythmite Hart (1992) developed a rhyth- and very fine sand interpreted as a glaciolacustrine mite index (R). Hart (1992) concluded that if R < 1, sediment derived from the cirque glacier that there is no relation between the thickness of the formed and wasted away during Younger Dryas clay part and silt part of a sequence of rhythmites, (Larsen and Longva 1979; Larsen and Mangerud and hence indicates deposition as varves. If, on 1981; Larsen et al. 1984). Unit G is a brown the other hand R > 1, there is a relation between Holocene gyttja. Only unit F, related to the Younger the thickness of clay part and the silt part of Dryas cirque glaciation, is dealt with further. a sequence of rhythmites, and hence indicates The lower boundary of unit F (to unit E) is deposition as turbidites. This index was applied transitional over a few mm to cm, but erosional in on the four cores. some cores from the threshold area and the southern basin. The age of this boundary was estimated to 12 700 cal. yrs. BP (Birks et al. 2000). The upper Sediment descriptions boundary (to unit G) is sharp at the last lamina, but organic content continues to increase upwards in The study is based on eleven cores; two from the G. This boundary was dated to 11 550 cal. yrs. BP proximal basin, four from the threshold area and (Gulliksen et al. 1998). Unit F varies in thickness five from the southern basin (Figure 2). The strati- between 44 and 208 cm in the studied cores. A graphical units are designated A–G from the base fining trend from the proximal basin to the thresh- upwards (Figure 3). Correlations of the main units old area is evident. The Vedde Ash bed (Mangerud between the cores are straightforward. Unit A et al. 1984), dated to 10 310 14C yrs. BP (12 000 cal. (Figure 3) is a diamicton interpreted as basal till yrs. BP, Birks et al. 1996b), is found in the middle (Larsen et al. 1984). Unit B comprises the lowest part or a little above the middle part of the unit limnic sediment and was originally interpreted as a where it occurs as one lamina. Based on an estim- slope-wash deposit (Larsen et al. 1984), but later, ated duration of sedimentation of 1170 cal yrs. Stalsberg (1995) suggested it to be glaciolacustrine, (Birks et al. 2000), the mean sedimentation rate probably related to an ice remnant left in the cirque in the Younger Dryas was 0.9 mm yrÀ1. It varies, during regional deglaciation. Units C–D–E consist however, from 1.7 mm yrÀ1 in the deepest part of of detritus gyttja interpreted as limnic sediments the basins to 0.4 mm yrÀ1 in the threshold area. deposited in the lake after the ice melted completely Furthermore, the sedimentation rate is highest in the cirque. Unit F consists of finely laminated silt above the Vedde Ash in 7 out of 10 cores (Figure 4). 54 55

Figure 6. The number of counted couplets in sub-units F1, F2, and F3 in three cores from the proximal basin, the threshold area, and the southern basin, respectively. Black (varves) and grey (turbidites) are according to calculations of the rhythmite index following the procedure in Hart (1992). Dashed line marks the position of the Vedde Ash bed. Full lines separates the three sub-units.

Unit F may be divided into three sub-units, F1–F3 of the unit. All cores from the lake contain a graded (Figure 3). Sub-unit F1 is characterised by rhyth- fine sand lamina overlain by a thick clay lamina mites with sharp silt to clay contacts, and at the top of F3 (Stalsberg 1995; Gulliksen et al. constant laminae thickness (Figure 5c). Many of 1998). the coarse laminae (silt/very fine sand) are graded, Laterally in the lake there are systematic varia- whereas most of the fine laminae (silt/clay) are not. tions in sediment signatures in unit F as well. The In the lowermost 50 cm of F1 the clayey part of a cores from the proximal basin (Figure 2) have the rhythmite is in general thicker than the silty part largest number of rhythmites (Figure 6), the thick- (C S), but further up in the sub-unit the opposite est laminae, and in general the coarsest texture trend dominates (C S). Sub-unit F2 has more (Figure 5a). Slump structures are common in the diffuse lamination or an apparently massive tex- proximal basin. Cores located in the threshold-area ture, and only a few laminae have sharp contacts have fewer rhythmites, the texture is finer, the (Figure 5d). Where laminae can be measured in F2, laminae are thinner, and not as many rhythmites the silty part is the thicker (C < S). The Vedde Ash have sharp contacts (Figure 5b). The silt laminae is localised approximately in the middle of this are not graded and generally thicker than the clay subunit. Sub-unit F3 resembles F1, but the rhyth- laminae (C S) throughout the unit in this part of mites are not as regular, and in the proximal basin the lake. In the southern basin the rhythmite sig- sand layers with slump-related structures occur fre- nature resembles that of the proximal basin, but the quently. The silty part is generally thicker than the number of rhythmites is far less (Figure 6). The silt clayey part (C S) except in the uppermost 10 cm laminae are in general thicker than the clay laminae

Figure 5. Thin section micrographs from unit F. Depths are given in cm below lake surface. (A) Sub-unit F3, core 502À51 showing typical rhythmite signatures close to the inlet in the proximal basin. Note the distinct coarse (light) to fine laminae (dark) contacts, more diffuse fine to coarse contacts, and lack of thickness trend between coarse and fine laminae. These rhythmites are suggested to be varves. The small cracks within the laminae and vertical structure to the left are due to drying and impregnation of the sediment. (B) Sub-unit F3, core 502À57 showing the rhythmite signature (fine texture, thin silt laminae) in the threshold area. Arrow in the lower part of the photo points at load structure. The smaller arrows point at macrofossils. The white cracks are due to the drying and impregnation of the sample. (C) Sub-unit F1, core 502À51 showing the typical signature with clay (dark) to silt (light) contacts being mostly sharp. These rhythmites are interpreted as turbidites mainly formed by slumping. (D) Sub-unit F2, core 502À66 showing a relatively coarse texture, diffuse contacts, and clay laminae (dark) containing fine sand particles (white dots). The sub-horizontal fractures in the middle to upper part and vertical structure to the right are due to drying and impregnation of the sediment. 56

(C S) throughout the unit. The texture is coarse were deposited by more short-lived events than compared to the threshold-cores, but not as coarse annual. as for the proximal cores. One of these cores As discussed previously, unit F is subdivided into (no. 46, Figure 2) has an erosional contact overlain F1–F3 on the basis of sediment characteristics by 10 cm of sand just below the Vedde Ash bed. (Figure 3). The rather constant laminae thickness The sand thickens and coarsens towards the south, and sharp silt to clay contacts in F1 indicate a high in the direction of the lateral moraine (Figure 1), sedimentation rate and turbid water, and it is likely and it has been interpreted as a mass flow originat- that these rhythmites are products of tubidity ing in the lateral moraine. currents. Oribatid mites rapidly disappear in this sub-unit (Solhøy and Solhøy 2000), and diatoms almost disappear (Bradshaw et al. 2000), both Origin of the rhythmites possibly reflecting high turbidity. Thus, it is likely that conditions generally were not favourable for Rhythmic deposition is the signature of ice-contact varve formation. In F2 the laminations are rather and distal glacier-fed lake environments (Ashley diffuse with only a few contacts between coarse and et al. 1985). Some rhythmites are deposited annually fine laminae being sharp suggesting rather low as varves (DeGeer 1912), but other temporal varia- sedimentation rates (Lemmen et al. 1988), which tions in sediment influx and dispersal mechanisms further might suggest that slope instabilities were also cause rhythmicity. reduced. The latter may be supported by the indi- Kra˚kenes Lake has a relatively uneven lake bot- cations of an increase in the density of the catch- tom (Figure 2), and this is reflected in variable ment vegetation at this time (Birks et al. 2000). sediment thickness. Besides, a localized sedimenta- Although lake turbidity was reduced, we have no tion pattern near the delta front in the proximal positive sedimentological criteria for distinguishing basin (Figure 2) caused high sedimentation rate varves from other rhythmites. The laminae in F3 and a larger number of rhythmites than at more are fine-grained, thin and regular resembling clas- distal localities (Figure 6). Instabilities on the delta sical varves (Figure 5a). This may suggest annual front led to slumping and slump-generated under- cyclicity with reduced turbidity and maybe pro- flows explaining the higher number of rhythmites longed lake-ice cover promoting suspension set- in this part of the lake compared to other parts tling of fines. However, the number of couplets (cf. Ashley 1995; Figure 6). Slope instabilities varies strongly between the three sub-areas of the also caused slump structures observed in core 46 lake (Figure 6). The proximal and the southern (Figure 2) which probably triggered short-lived sub-basin have the largest number of couplets sug- flows of turbid water (Hampton 1972; Thomas gesting that pseudo-varves, maybe caused by slope 1984) causing rhythmic deposition distally to this instabilities, are present. site. Thus localized, short-lived factors tend to Applying the rhythmite index (Hart 1992) to complicate and overprint the overall sedimentation distinguish between varves and non-varve couplets pattern in the lake. gives contradictory results (Figure 6). The rhyth- In an attempt to characterize rhythmites accord- mite index in F3 in the proximal and the southern ing to lake turbidity Ashley (1975) distinguished basin suggests that the coarse and the fine part of between three types based on silt (coarse) versus most couplets were deposited as two distinct, separ- clay (fine) thickness: I (S < C) – formed in still water, ate pulses, i.e., suggesting that these are varves. II (S ¼ C) – formed in still water, III (S > C) – However, the results from the threshold in the same characterised by a high sedimentation rate/lake zone all indicate non-varve deposition. This prob- turbidity. Annual deposition will be favoured ably is due to the inability of the index to reflect under types I and II, but neither of these is diag- and distinguish between rather complex process nostic. The lake sediments at Kra˚kenes are domin- parameters. ated by the type II and III rhythmites, except in In summary, we have not been able to clearly the cores from the proximal basin where type I is distinguish between varves and non-varve couplets observed. Based solely on this one might infer that in Kra˚kenes Lake. Clearly, there are couplets that a large proportion of the rhythmites at Kra˚kenes are not varves, and it is likely that there are varves 57 as well. Kra˚kenes Lake is quite shallow and local and the retreat phase interrupted by a brief halt or factors like slope instability also were important. small re-advance. Thus, probably varve formation was rather the Late-glacial July air temperatures and annual exception than the rule. Following from this, a precipitation have been reconstructed for direct counting of rhythmites as indicated by the Kra˚kenes from pollen data (Birks et al. 2000). rhythmite index to be varves at best gives a mini- Based on the pollen-reconstructed Younger Dryas mum approximation of the duration of glaciolacus- July temperatures, ablation season temperatures at trine sedimentation or cirque glaciation (Figure 6). the ELA for the glacier at the two moraine posi- The number is well below the estimated duration of tions and for the top of the glacier, leading to 1170 years based on radiocarbon (Gulliksen et al. complete withdrawal, were calculated (Table 1). 1998; Birks et al. 2000), and even the total number This was in turn used in calculating winter of couplets in the most ‘complete’ core is below the accumulation at the same three ELA positions number of radiocarbon years. (Table 1, Figure 7). The exponential relationship between summer ablation and winter accumulation described in the caption to Table 1 is an empirical Intra Younger Dryas climate change one based on modern glaciers (O. Liestøl in Sissons 1979; Sutherland 1984; Ballantyne 1989). As it The overall stratigraphy in the Kra˚kenes Lake establish a relationship between the two most (Figure 3) represents large climatic shifts (Larsen important mass balance factors of glaciers, it is et al. 1984; Birks et al. 2000). These changes are considered to be valid for all glaciers, including represented with glacial–deglacial sediments (units paleoglaciers (e.g., Dahl et al. 1997). Comparing A and B), gyttja sediments (units C, D, and E), calculated winter accumulation with the annual glaciolacustrine sediments (unit F), and Holocene precipitation based on pollen (Figure 7), the most gyttja (unit G). The question raised here is whether striking result is that winter accumulation in there is a further sedimentolgical signal of climatic the cirque was far greater than the pollen-based change within the Younger Dryas, i.e. within unit F annual precipitation (Figure 7). Unfortunately, the (Figure 3). Younger Dryas annual precipitation reconstruc- As described earlier the Younger Dryas glacio- tion based on pollen and spores (Figure 7) is greatly lacustrine sediments are divided into three sub- underestimated due to difficulties of determining units, F1–F3. F1 is characterised by sharp silt to this parameter in an arctic and alpine plant com- clay contacts, constant laminae thickness, and munity without trees (H. Birks pers. comm. 2002). some graded coarse laminae. F2 has more diffuse To test the validity of the high reconstructed winter lamination or an apparently massive texture, and accumulation, regional winter precipitation can be only a few laminae have sharp contacts. F3 resem- estimated by further glacial geological reasoning. bles F1, but the rhythmites are not as regular. F2 is During Younger Dryas there were no ice sheet or generally more fine-grained than both F1 and F3. plateau glaciers in this area (Larsen et al. 1984). Each of the three sub-units is thick, i.e. represents This means that regional ELA was at or above the relatively long time, and they are all distributed in highest mountain, which is 645 m a.s.l. just to the entire lake. This is taken to suggest that these the northeast of Kra˚kenes. Using this elevation, differences cannot be explained by some acciden- the pollen derived temperatures and an adiabatic tal/episodic changes, but are rather related to reor- lapse rate of 0.65 C/100 m (Green and Harding ganization of the glacier and thus with climate. 1980), we calculated the maximum Younger Dryas Furthermore, there are two morphological expres- winter precipitation giving no regional glaciation in sions of glacier variations in the cirque itself to be the area (Figure 7). Comparing the latter with the considered; the outermost terminal moraine mark- winter accumulation calculated for the cirque, ing the maximum position, and a smaller moraine show that winter accumulation in the cirque was marking a halt or re-advance during retreat on average 2.5 to 3 times higher than regional (Figure 7). Thus there is a tri-partition also in winter precipitation (Figure 7). This is taken to glacier variation; the initial advance to the maxi- suggest that lee-side accumulation of snow by mum position, still-stand at the maximum position, wind and by avalanching was crucial for forming 58

Figure 7. Time–distance diagram showing inferred cirque glacier variations discussed in the text. July air temperatures and annual precipitation based on terrestrial pollen and spore data are from Birks et al. (2000). Ablation season (May 1–October 30) air temperatures and winter season (November 1–April 30) accumulation are from Table 1. Winter precipitation was obtained by assuming a regional ELA of 650 m asl. For Younger Dryas which is the lowest ELA giving no glaciers on plateaus. Accordingly, this is a calculation of maximum winter precipitation. The position of the Vedde Ash bed (Mangerud et al. 1984) is dotted. The calendar age scale is according to Gulliksen et al. (1998) and Birks et al. (2000). Numbers 1–17 are intercepts for calculations. and maintaining a glacier in the cirque. Based on after and during a long period of rather constant aspect of a large number of Younger Dryas cirques, climate conditions (Figure 7). The Vedde Ash Larsen et al. (1984) concluded that principal snow- bed caused by a volcanic eruption in Iceland bearing winds in the area was from the southwest. (Mangerud et al. 1984) is found in the middle to The Kra˚kenes cirque is situated on a northwest upper part of sub-unit F2. This large ash fall-out slope of a mountain plateau, i.e. in a favourable (Mangerud et al. 1984) covered the entire glacier, position for lee-side accumulation by winds from and the lowered albedo reduced reflection of the sector east to southwest. incoming rays probably resulting in increased melt- The initial glacier advance took place during ing of the glacier. Thus we speculate if the ash fall- sub-unit F1 (Figure 7). Obviously the initial forma- out actually triggered the initial melting of the tion of the glacier was preceded by a lowering of the glacier although we do not know if the effect was ELA in late Allerød time caused by lowering of large and lasting enough for this to happen. At the summer temperatures. Whether the advance phase F2/F3 transition laminations again become more lasted during all of sub-unit F1 is not known, but distinct. This is interpreted as more distinct melt- the uniformity (distinct laminations) throughout water pulses that likely were caused by continued leads us to suggest that the entire unit was depos- glacier retreat (Leonard 1986). This coincides ited during one type of glacier regime. The more with increasing summer temperatures, but also diffuse lamination in sub-unit F2 is taken to indic- increased winter accumulation (Figure 7). The ate a quiescence period of the glacier with less latter, however, is a maximum value based on an distinct melt-water pulses. We suggest that this ELA position at 200 m asl. Raising the ELA will corresponds to still-stand at the maximum posi- lower the accumulation estimate causing negative tion, and that the glacier obtained equilibrium. glacier mass balance. A glacier re-advance, or more The latter is suggested because still-stand occurred probably still-stand, is marked by the inner 59

Table 1. Younger Dryas temperature and precipitation/accumulation data (cf. Figure 7).

July air Ablation season T, C Winter (pollen) T, C Accumulation at 38 m a.s.l. Sea level ELA 150 m ELA 200 m ELA 250 m at ELA’s, m

17R 7.8 6.5 4.9 6.4 16R 7.6 6.3 4.7 6.0 15S 7.5 6.2 4.9 6.4 14S 7.2 5.9 4.6 5.8 13R 6.5 4.2 2.9 3.3 12R 6.4 4.1 2.8 3.2 11S 6.5 4.2 3.2 3.6 10S 6.7 4.4 3.4 3.9 9S 6.6 4.3 3.3 3.8 8S 6.5 4.2 3.2 3.6 7S 6.4 4.1 3.1 3.5 6S 6.6 4.3 3.3 3.8 5A 6.7 4.4 3.4 3.9 4A 6.7 4.4 3.4 3.9 3A 6.6 4.3 3.3 3.8 2A 6.5 4.2 3.2 3.6 1A 6.5 4.2 3.2 3.6 Younger Dryas July air temperatures are based on pollen (Birks et al. 2000), and are for the elevation of the Kra˚kenes Lake (38 m a.s.l.). Ablation season temperatures (May 1–October 30) were calculated by subtracting present ÁJuly-Ablation season temperature (1.5 C) from the Younger Dryas July air (pollen) temperatures, and adjusting for an adiabatic lapse rate of 0.65 C/100 m (Green and Harding 1980). Winter season (November 1–April 30) accumulation at the ELA was derived from an exponential relationship at the ELA between mean ablation season temperature (t) and winter accumulation (A) expressed as: A ¼ 0.915e0.339t. A is given in m water equivalent, and t in C. An ELA of 150 m a.s.l. (Larsen et al. 1984) was used for the initial advance and still-stand at maximum, an ELA of 200 m a.s.l. was used for the initial retreat and still-stand, and an ELA of 250 m asl. was used for the final retreat. Intercept points for the calculations are given in the left hand column. A ¼ advance; S ¼ Still-stand; R ¼ Retreat (cf. Figure 7). moraine position. Evidently the glacier was able to Dryas cirque basins in the Nordfjord area maintain a more positive mass balance with commonly contain a smaller end moraine within the a higher than before equilibrium line, but probably main moraine (Larsen et al. 1984). Blikra and this was only the case for a brief period of time. The Longva (1995) have evidence of frost shattering climatic causes for this halt are not obvious from along paleobeaches in western Norway occurring the climate data, although a lowered rate of tem- in two discrete periods within the Younger Dryas. perature increase and maximum accumulation Further north on the coast, in the Trondheimsfjord values is seen (Figure 7). A climatic rather than area, at least three generations of Younger Dryas local (accidental) cause is likely since Younger terminal moraines from the withdrawal of the Dryas cirque basins in the Nordfjord area com- ice sheet have been mapped (Reite et al. 1982; monly contain a smaller end moraine within the Reite 1995). Some of these may be topographically main moraine (Larsen et al. 1984). After the halt, controlled, but the overall pattern suggests regional continued temperature increase and a presumed climatic control. In a reconstruction based on somewhat lowered accumulation led to the final dinoflagellate cysts in a sediment core from retreat of the Kra˚kenes glacier. The glacier had Voldafjorden (close to Kra˚kenes), Grøsfjeld et al. wasted away completely at the F/G boundary, (1999) concluded that a milder phase with increased and it is worth noting that the most rapid tempera- sea-surface temperature and decreased sea-ice ture rise occurred after the glacier had melted cover occurred in the Younger Dryas right above completely. the Vedde Ash Bed. This corresponds to the upper In addition to the Kra˚kenes data, there are mul- part of sub-unit F2 and thus may correspond in tiple evidences for intra Younger Dryas climate time with the initial withdrawal of the Kra˚kenes change in the region. As noted above, Younger glacier from its maximum position. 60

Conclusions Sønstegaard, Jan Tveranger and Vidar Valen for participating in the field work. Mary Raste, Based on the sedimentological studies of the University of Tromsø performed sedigraph anal- Kra˚kenes Lake glaciolacustrine sediments depos- ysis. Gunvor Granaas, University of Tromsø made ited by melt-water from a cirque glacier, we were thin section photographs, and Irene Lundquist not able to distinguish clearly between varves and (NGU) drafted the figures. Climate reconstruc- other types of rhythmites. The variable number of tions were discussed with H. Birks, S.O. Dahl, rhythmites within different parts of the basin reflects Ø. Lie and A. Nesje. H. Birks enthusiastically led an uneven lake floor and local slumping events, the Kra˚kenes project (Birks et al. 1996a, b, 2000) especially from the delta front and the distal side and commented on the manuscript. J. Mangerud of the lateral moaraine. True varves are probably read an early draft of the manu-script. Thanks are present, but are ‘masked’ by other types of rhyth- also extended to the journal referees Scott mites. Probably the lake was too shallow and the Lamoreux and one anonymous for valuable sug- glacier too close to be ideal for varve formation. gestions to improvements. This is a contribution to Following from this, the duration of cirque glacia- NORPAST (Past Climates of the Norwegian tion could not be established based on varve Region), and is Kra˚kenes Project Contribution counting. number 23. A lake-wide tri-partition of the Younger Dryas glaciolacustrine sediments is linked with the initial advance, the still-stand at the maximum position, References and the retreat of the cirque glacier. Thus the main glacial regimes are reflected in the signatures of the Andersen B.G., Lundqvist J. and Saarnisto M. 1995a. The glaciolacustrine sediments. Younger Dryas margin of the Scandinavian ice sheet – An Lee-side accumulation of snow in the cirque by introduction. Quat. Internat. 28: 145–146. wind and avalanching was crucial to form and Andersen B.G., Mangerud J., Sørensen R., Reite A., Sveian H., maintain the glacier once summer temperatures Thoresen M. and Bergstrøm B. 1995b. Younger Dryas ice- were low enough. A rather long-lasting quiescence marginal deposits in Norway. Quat. Internat. 28: 147–169. Ashley G.M. 1975. Rhythmic sedimentation in glacial lake phase of the glacier at its maximum position Hitchcock, Massachusetts–Connecticut. Soc. Ec. Pal. and occurred after and during a period of stable tem- Min. Spec. Publ. 23: 304–320. perature and accumulation values, indicating that Ashley G.M., Shaw J. and Smith N.D. 1985. Glacial Sedimentary the glacier reached climatic equilibrium. Possibly Environments. In: Jopling A.V. and McDonald B.C. (eds), the initial glacier retreat from the maximum Soc. of Ec. Pal. and Min. Short Course Notes, 16: 135–205. Ashley G.M. 1995. Glaciolacustrine environments. In: Menzies J. position was triggered by fall-out of volcanic ash (ed.), Modern Glacial Environments, Processes, Dynamics on the ice surface, and continued further by rising and Sediments, Butterworth-Heinemann Ltd., Oxford, summer temperatures. A brief re-advance or halt pp. 417–444. of the glacier was probably controlled by changing Ballantyne C.K. 1989. The Loch Lomond readvance on the climate as this is a common observation in many island of Skye, Scotland. Glacier reconstruction and paleocli- matic implications. J. Quat. Sci. 4: 95–108. cirques in the area occupied by glaciers during Birks H.H., Gulliksen S., Haflidason H., Mangerud J. and the Younger Dryas. The largest climate change Possnert G. 1996b. New radiocarbon dates for the Vedde in terms of temperature at the Younger Dryas/ Ash and the Saksunarvatn Ash from western Norway. Quat. Preboreal transition occurred after the cirque gla- Res. 45: 119–127. cier had vanished and glaciolacustrine sedimenta- Birks H.H. 1996a. The Kra˚kenes late-glacial palaeoenvironmental project. J. Paleolim. 15: 281–286. tion ceased. Birks H.H., Battarbee R.W. and Birks H.J.B. 2000. The development of the aquatic ecosystem at Kra˚kenes Lake, western Acknowledgements Norway, during the late-glacial and early Holocene – a synthesis. J. Paleolim. 23: 91–114. The work was financially supported by the Bjo¨rck S., Kromer B., Johnsen S., Bennike O., Hammarlund D., Lemdahl G., Possnert G., Rasmussen T.L., Wohlfarth B., Research Council of Norway (NFR) and the Hammer C.U. and Spurk M. 1996. Synchronized terrestrial– Geological Survey of Norway (NGU). We thank atmospheric deglacial records around the North Atlantic. Hilary Birks, Josef Kusior, Jan Mangerud, Eivind Science 274: 1155–1160. 61

Blikra L.H. and Longva O. 1995. Frost-shattered debris facies Lemmen D.S., Gilbert R., Smol J.P. and Hall R.I. 1988. of Younger Dryas age in the coastal sedimentary succes- Holocene sedimentation in glacial Tasikutaaq Lake, Baffin ions in western Norway: Palaeoenvironmental implications. Island. Can. J. Earth Sci. 25: 810–823. Palaeogeogr. Palaeoclim. Palaeoecol. 118: 89–110. Leonard E.M. 1986. Varve studies at Hector Lake, Alberta, Bradshaw E.G., Jones V.J., Battarbee R.W., Birks H.J.B. and Canada, and the relationship between glacial activity and Birks H.H. 2000. Diatom responses to late-glacial and early- sedimentation. Quat. Res. 25: 199–214. Holocene environmental changes at Kra˚kenes, western Longva O., Larsen E. and Mangerud J. 1983. Beskrivelse til Norway. J. Paleolim. 23: 21–34. kvatrgeologisk kart 1019 II – M 1 : 50 000. NGU 393. Dahl S.O., Nesje A. and Øvstedal J. 1997. Cirque glaciers as Mangerud J., Larsen E., Longva O. and Sønstegaard E. 1979. morphological evidence for a thin Younger Dryas ice sheet in Glacial history of western Norway 15 000–10 000. Boreas 8: east-central Norway. Boreas 26: 161–180. 179–187. DeGeer G. 1912. A geochronology of the last 12 000 years. In: Mangerud J., Lie S.E., Furnes H., Kristiansen I.L. and Lømo L. 11th Internatl. Geol. Cong. Stockholm, 1910 Compete 1984. A Younger Dryas ash bed in western Norway, and its Rendu, pp. 241–258. possible correlations with the tephra in cores from the Green F.H.W. and Harding R.J. 1980. The altitudinal gradients Norwegian Sea and the North Atlantic. Quat. Res. 21: of temperature in southern Norway. Geogr. Ann. 62A: 29–36. 85–104. Grøsfjeld K., Larsen E., Sejrup H.P., De Vernal A., Flatebø T., Merkt J. 1971. Reliable counting of annually laminated layers of Vestbø M., Haflidasson H. and Aarseth I. 1999. lake sediments by means of long-sized thin sections. Arch. Dinoflaggelate cysts reflecting surface-water conditions in Hydrobiol. 69: 145–154. Voldafjorden, western Norway during the past 11 300 yr BP. Reite A.J. 1995. Deglaciation of the Trondheimsfjord area, Boreas 28: 403–415. central Norway. NGU Bulletin 427: 19–21. Gulliksen S., Birks H.H., Possnert G. and Mangerud J. 1998. A Reite A.J., Selnes S. and Sveian H. 1982. A proposed calendar age estimate of the Younger Dryas-Holocene bound- deglaciation chronology for the Trondheimsfjord area, ary at Kra˚kenes, western Norway. The Holocene 8: 249–259. central Norway. Norges Geologiske Undersøkelse 373: Hampton M.A. 1972. The role of subaqueous debris flow in 75–84. generating turbidity currents. J. Sed. Petrol. 42: 775–793. Sissons J.B. 1979. Palaeoclimatic inferences from former Hart J.K. 1992. Sedimentary environments associated with glaciers in Scotland and the lake District. Nature 278: Glacial Lake Trimingham, Norfolk, UK. Boreas 21: 119–136. 518–521. Larsen E. and Longva O. 1979. Jordartskartlegging, glasialgeo- Solhøy I.W. and Solhøy T. 2000. The fossil oribatid mite fauna logi og kvartrstratigrafi pa˚ Stad, Va˚gsøy, Ytre Nordfjord. (Acari: Oribatida) in late-glacial and early-Holocene sedi- Unpubl. cand. real thesis, Univ. Bergen. ments in Kra˚kenes Lake, western Norway. J. Paleolim. 23: Larsen E. and Mangerud J. 1981. Erosion rate of a Younger 35–47. Dryas cirque glacier at Kra˚kenes, Western Norway. Annals Stalsberg M.K. 1995. Seinglasial sedimentasjon i Kra˚kenes Glaciol. 2: 153–158. Lake, Va˚gsøy, Sogn og Fjordane. Unpubl. cand. scient thesis, Larsen E., Eide F., Longva O. and Mangerud J. 1984. Allerød- Univ. Tromsø. Younger Dryas climatic inferences from cirque glaciers and Sutherland D.G. 1984. Modern glacier characteristics as a basis vegetational development in the Nordfjord area, Western for inferring former climates with particular reference to the Norway. Arct. and Alp. Res. 16: 137–160. Loch Lomond Stadial. Quat. Sci. Rev. 3: 291–309. Larsen E., Attig J.W., Aa A.R. and Sønstegaard E. 1998. Late- Thomas G.S.P. 1984. A late Devensian glaciolacustrine glacial cirque glaciation in parts of western Norway. J. Quat. fan-delta at Rhosemor, Clwyd, N. Wales. Geol. J. 19: Sci. 13: 17–27. 125–141.