Introduction

Although Iceland lies in a pivotal position within the North Atlantic climate system (Lamb, 1979) (Fig. 1), it is only very recently that detailed evaluations of late Quaternary marine records are becoming available (Andrews et al., 2000; Eiriksson et al., 2000a; Jennings et al., 2000). A major unresolved issue is the extent of the glaciation of Iceland during the Last Glacial Maximum (LGM), especially the extent of ice on the Northwest Peninsula (Norddahl, 1991) (Fig. 1). Our purpose is to present sedimentological, chronological, and faunal analysis of piston cores (B997-322PC, -323PC, -326PC1, and -326PC2) (Fig. 2) (henceforth 322PC) that were collected on a joint Iceland/U.S.A. research cruise (B997-) on the research vessel Bjarni Saemundsson (Helgadóttir, 1997). The basal dates from these cores gave uncorrected ages of 42,600 ± 3050 B.P. and 25,330 ± 640 B.P. (Andrews et al., 2000; Smith and Licht, 2000), suggesting that these cores might embrace ice sheet/ocean interactions during part of Marine Isotope Stage (MIS) 3 and the whole of MIS 2 and 1.

The focus of this paper is the description and interpretation of sediment sequences in these four cores, in particular the interpretation of pebbly muds (diamictons) that in 322PC and 323PC contain persistent numbers of benthic and planktonic foraminifera. Such sediments could reflect (1) open-water glacial-marine conditions, (2) reworking of glacial-marine sediments by glacial ice, or (3) subglacial tills. The first situation would indicate that the sites were not glaciated, whereas the latter two options permit the covering of the sites by grounded ice. Thus there are two important questions associated with the core stratigraphies: (1) what is the origin(s) of the diamictons? (2) is the contact between diamictons and the fine-grained muds synchronous along the trough, and does it reflect deglaciation of the shelf? In this paper dates are usually expressed in units of ka B.P. or cal ka B.P. where the dates are calibrated. We use ky to represent a rate.

Background

The glacial and volcanic history of Iceland has become of regional significance owing to the recognition of Icelandic tephras in Greenland ice cores and in North Atlantic marine and lakes cores (Bjorck et al., 1992; Grönvold et al., 1995; Haflidason et al., 2000) and because the record of far-travelled sediments of Iceland origin appear to have some nonrandom relationship with abrupt events (e.g., Heinrich events) in the North Atlantic (Bond and Lotti, 1995; Elliott et al., 1998). However, the nearshore record of the extent and history of glaciation and ice-rafting around Iceland is currently poorly documented, although significant efforts are now being directed toward this problem (Kristjansdóttir, 1999; Eiriksson et al., 2000a; Cartee Schoolfield, 2000; Jennings et al., 2000; Geirsdóttir et al., 2001, 2002).

Our paper deals with the interaction between the Iceland Ice Cap and the ocean at sites north and northeast of the Northwest Peninsula. This large peninsula is separated from the main bulk of the island by a narrow strip of land (Figs. 1 and 2). It is uncertain how the glaciation of the Northwest Peninsula is tied into the glacial history of the rest of Iceland (Norddahl, 1991) and whether there were refugia on coastal mountains (Rundgren and Ingolfsson, 1999). Was there a separate ice cap over the peninsula, or did it join with the main Iceland Ice Cap? In particular, the extent of glaciation during the LGM at roughly 22 cal ka B.P. is unknown (Ingolfsson, 1991; Eiriksson et al, 1997). Glaciological modeling is now starting to develop hypotheses of the variations in glacial extent across Iceland during the last glacial cycle. For example, Webb et al. (1999) indicate total ice coverage during the LGM. However, geomorphic reconstructions of the Iceland Ice Cap (Bourgeois et al., 2000; Stokes and Clark, 2001) suggest that the Northwest Peninsula hosted a separate ice center with an ice divide forming a broad U-shape, draining principally into Ísafjardardjúp (Fig. 1). Because of the limited accumulation area, the Northwest Peninsula ice cap may not have extended to the shelf break. Indeed, research on the sediment history in Djúpall (Andrews et al., 2002c; Geirsdóttir et al., 2002) suggested that at the LGM, the ice stream in Ísafjardardjúp extended just beyond the mouth of this large fjord complex. Glacial-marine diamictons have been recovered in Djúpall (cores B997-338 and MD99-2264, Fig. 2) that provide a continuous history of deposition between 13 and 36 cal ka B.P. (Andrews et al., 2002c; Geirsdóttir et al., 2002). If this picture is correct, and it is supported by other facts such as very low marine limits at the tip of the Northwest Peninsula (Hjort et al., 1985), then it then raises the question of how far offshore could the Northwest Peninsula ice cap have extended seaward of the Strandir coastline, and was there a separate ice stream in Reykjafjardaáll/Húnaflóaáll (Fig. 2).

There are also questions about the postulated LGM age of the Latra End Moraine, which lies at the shelf break, west of the Northwest Peninsula (Olafsdóttir, 1975; Syvitski et al., 1999). If this deposit is indeed older than the LGM (Andrews, in press) then symmetry and ice-sheet reconstructions indicate that ice may not have flowed completely along Reykjafjardaáll/Húnaflóaáll on the way to the northern shelf break. Thus, the extent of ice during the LGM in Húnaflóaáll is presently uncertain, although on the basis of the two initial basal dates in 322PC and 323PC, Andrews et al. (2000) placed an ice margin landward of B997-323PC.

Húnaflóaáll is a large trough that trends roughly south to north and thus angles away from the trend of the northeast coast of the Northwest Peninsula (Fig. 2). Where this trough meets Reykjafjardaáll, the combined feature is called Húnaflóadjúp. Húnaflóadjúp attains maximum water depths of 400 to 500 m but becomes shallower at both its landward margin and seaward (Fig. 2). For Reykjafjardaáll/Húnaflóaáll to be filled by an active outlet ice stream, there must have been an ice divide at the waist of the peninsula and a reasonable thickness of ice to promote flow to the north along Húnaflóadjúp (Bourgeois et al., 2000). There are, however, very few observations pertaining to ice extent and deglaciation of this northern area (Larusson, 1983; Hjort et al., 1985). The best estimate of the deglaciation of this part of the Northwest Peninsula is that it occurred sometime prior to 12 ka B.P. (Ashwell, 1996; Rundgren et al., 1997).

The area off northwest/north Iceland is critically situated with respect to ocean circulation (Malmberg, 1969, 1985; Stefansson, 1969a, 1969b; Lamb, 1979; Olafsson, 1999). Húnaflóaáll lies along one of the standard hydrographic transects made on a seasonal basis by the Marine Research Institute, Iceland. This transect recorded an abrupt switch from Atlantic Water to Polar Water in the 1960s associated with the Great Salinity Anomaly (Lamb, 1979; Malmberg, 1985; Dickson et al., 1988; Belkin et al., 1998). The warm Irminger Current moves up the west side of Iceland (Fig. 1); one branch splits and moves across the Denmark Strait and then southward, whereas the remainder is directed around the Northwest Peninsula and then moves eastward as the North Iceland Irminger Current (Hopkins, 1991). Farther seaward, the East Iceland Current , a branch of the East Greenland Current (Fig. 1), brings cold surface water and frequently heavy sea ice and a few icebergs to north Iceland in extreme years (Stefansson, 1969a; Sigtryggsson, 1972). In 1997 our hydrographic casts indicated the presence of a shallow, thin cold layer underlying the surface coastal water. At depths >50 m, water temperatures decreased gradually toward 0°C at site 323PC. The upper 50 m of the water column shows a rapid (2‰) increase in salinity. Below 50 m, the high salinities (∼34.9‰) suggest the presence of Atlantic Water and winter-cooled Atlantic Water at depths >200 m.

Core Sites and Seismic Stratigraphy

Cores 322PC and 323PC were collected from Húnaflóadjúp; 326PC1 and 326PC2 were collected on the landward side of the basin along Reykjafjardaáll (Fig. 2). A 3.5-kHz system was employed to obtain estimates of sediment thickness and overall seismic stratigraphy (Helgadóttir, 1997; Andrews et al., 2000) (Fig. 3). Water depths varied from 397 to 351 m. The seismic data suggest that Húnaflóadjup is occupied by a shelf sediment or drift deposit (McCave and Tucholke, 1986; Faugeres et al., 1999; Harris et al., 2001), with older sediment outcropping near the sea floor to the north and south of the drift (Fig. 3B). A well-defined seismic reflector lies deep (21 m) within the Quaternary section in the center of the basin. This reflector was cored at site MD99-2269 (Fig. 2) and shown to be the 9 ka B.P. Saksunarvatn tephra (Andrews et al., 2002a, 2002d). This reflector is near the surface at the margins of the trough, such sites of 323PC and 326PC (Fig. 3), but it is not obvious on the outer shelf near site 322PC. On the inner shelf at 329PC (Fig. 2) (Castaneda, 2001; Andrews et al., 2002d) the Saksunarvatn tephra occurs about 2.2 m below the seafloor—these relationships confirm that the main sediment drift accumulation in Húnaflóadjúp (Fig. 3) is post ∼9 ka B.P.

At 322PC the sea floor is gently rolling, with a relief of 5 to10 m. Little to no seismic penetration was achieved, suggesting a thin sediment cover over dense sediment or bedrock. At the site of 323PC, an internal reflector is close to the surface and has some limited (1 to 2 m) relief. Acoustically transparent sediment overlies the reflector, but the records are not good enough to determine much detail (Fig. 3). Immediately seaward of 326PC, over 30 m of sediment fills the basin, and one prominent reflector, the same as noted at 323PC, appears to cropout at or near the surface. Here it is clear that the reflector occurs above an additional 10 m or so of sediment that thins toward the core site (the ship drifted during the extraction of 326PC1 and 326PC2; see Table 1). During the 1999 cruise along Húnaflóaáll in the French research vessel Marion Dufresne, as part of an IMAGES V coring expedition, a 25m core (MD99-2269, Fig. 2) was recovered in the center of trough and penetrated the two main seismic reflectors. Our research indicates that the uppermost regional reflector is caused by deposition of the Saksunarvatn tephra at 9 ka B.P. (10.18 ± cal ka B.P.) (Grönvold et al., 1995; Andrews et al., 2002d).

Sediment Cores: Methods

The resolution of the 3.5-kHz system was such that we are uncertain whether the “basement” is bedrock or acoustically “hard” sediment, such as a diamicton. We used both a 10-cm gravity corer, usually with a 2- to 3-m barrel, and a 6-m piston corer (PC) with a diameter of 7 cm. Our primary goal was to achieve the longest (temporal) record, at the risk of low resolution. Cores 322PC, 323PC, and 326PC1, with lengths of ca 1.5 and 3 m (Table 1), penetrated stiff, pebbly sediments and indicate that at these sites at least, the acoustic basement is glacial or glacial-marine diamictons.

The cores were split, photographed, described, X-radiographed, and sampled for both sediments and foraminifera. We visually described the cores using a lithofacies code (Eyles et al., 1983). The primary lithofacies identified are massive, matrix-supported diamictons (Dmm) and fine-grained muds (F) (Figs. 4, 5). Sediment samples consisted of a 10.4-cc container that allowed the determination of wet and dry density. Other processes included determination of total organic carbon (TOC) and total carbonate with a Coulometer, mass magnetic susceptibility with a Bartington meter, and grain-size determinations using a long-bed Malvern laser system (Syvitski, 1991; Andrews et al., 2002b). The number of clasts >2 mm were counted in 2-cm vertical segments from X-radiographs (Grobe, 1987; Andrews et al., 1997).

The foraminiferal samples were washed through a 63-μm screen and then sieved at 105 μm and 1000 μm (Knudsen and Austen, 1996). We split the sand fraction into two using a microsplitter; we used one half for foraminiferal assemblages and numbers/g estimates and the half for sediment composition. We obtained sediment weights and expressed the foraminifera results as numbers/gram of sediment and as percentages of the total fauna. We based individual identifications on prior research and on surface collections made around Iceland (Helgadóttir, 1984; Weiner et al., 1999). We picked between 10 and 20 individuals of the polar planktonic foraminifera Neogloboquadrina pachyderma s for isotopic analyses. The samples were processed at the Woods Hole Oceanographic Institutes stable isotope laboratory supervised by R. Osterman.

We evaluated the composition of the sediment fraction in the 105 to 1000 μm fraction. Because of the high proportion of sand-size particles, we did not attempt an estimate of numbers/gram, but rather worked with percentage composition because of the potentially large error involved splitting the sample many times to achieve a 300 grain/slide coverage. We cast an even spread of material on a tray marked off in 1-cm² intervals. Samples were selected randomly for processing to avoid biases in the counting procedure. We used a binocular microscope at 40* power and counted between 200 and 300 grains, at randomly selected grid intersections. Sediment pellets were counted, but these were considered outside the sum; thus in any sample the total number of objects counted ranged between 300 and 500. We identified eight compositions, namely: bubble-wall clear ash, dark glassy or pumaceous ash, quartz, “red” grains, modified (abraded) basaltic grains, other rock fragments, foraminifera, and sediment pellets. The term “tephra” is used to describe a variety of grain shapes, colors, and compositions (Haflidason et al., 2000); we use it synonymously with “ash,” which is defined as volcanic materials <2 mm. The common element in the recognition of ash particles is that they possessed sharp, angular corners with no evidence of wear. In our usage (Figs. 4, 5), “lithics” included red grains, quartz, other, and modified basaltic grains. “Dark lithics” were principally basalts, also containing “glassy” grains, but these grains had subangular to subrounded (occasionally rounded) corners and overall shapes indicating moderate to substantial abrasion. For the first seven elements the counts were converted to percentages, whereas the sediment pellets were expressed as a percentage of the seven element count. Standard errors of the counts were computed using standard formulae (Hamilton, 1990). Because counts were converted to percentage data, with all sums hence equaling 100, in our multivariate analyses we used the log ratio transformation to avoid the closed array problem (Aitchison, 1986; Hamilton, 1990; Kovach, 1998). For our data analyses we used two main programs, Stata^TM (Stata, 1999) and Multivariate Statistical Programs (MVSP) (Kovach, 1998). In our analysis (see later) we plotted compositional ratios to discriminate between environments of deposition. For example, the ratio of lithics:lithics plus foraminifera is commonly used to designate a change between an ice-rafted debris dominated environment and more open (less or nonglacial) marine conditions (Bond and Lotti, 1995).

We picked individual tephra shards for elemental analysis. Some of these were processed at the Nordic Volcanological Institute, Iceland (Dr K. Grönvold); we also probed at the University of Colorado with a JEOL 8600 electron microprobe at 15 Kv accelerating voltage, 20 nanoAmp cup current and a 5 μm beam. The results were standardized to a US Natural History Museum basaltic glass standard A99 (#113498). Data were reduced using the Bence Albee 1968 reduction procedure. 326PC2 had discrete ash units that were visible on opening the core, but in the other three cores there were changes in the relative composition of the sand-size fraction. The geochemical results from our analyses will be available on the NOAA data base.

Core Chronologies: AMS ¹⁴C Dates and Tephras

Details on the number of foraminifera species dated and on the sample δ¹³C are given in Smith and Licht (2000). In order to obtain samples of sufficient mass we usually had to combine all species. Because these sites from relatively shallow waters (<400 m) it is unlikely that samples from near-surface and benthic species would have greatly varying ages, and δ¹³C variations are minimal.

We have not adjusted the radiocarbon dates on Table 1 and Figures 4 and 5 for an ocean reservoir correction because there is discussion as to what this correction should be at present (Håkansson, 1983; Haflidason et al., 2000), and both spatial and temporal variations are expected (Bard et al., 1994; Hagen, 1999; Voelker et al., 1998; Voelker, 1999; Eiriksson et al., 2000b). Based on a comparison between terrestrial age estimates (Haflidason et al., 2000; Wastl et al., 1999) and our marine dates on the Saksunarvatn ash at sites south and west of 326PC (Andrews et al., 2002d), it appears that for the last 9000 ¹⁴C yr B.P. the ocean reservoir correction is approximately 400 ± yr, thus somewhat less than the correction derived from sites to the east of 326PC for Preboreal times (Haflidason et al., 2000). However, we apply a 400-yr correction to our dates (Table 1), and it is these corrected ages that we report in the text . Dates in kilo years (ka B.P.) are rounded to the nearest decimal point. The 400-yr correction is probably a minimum estimate for MIS2 and 3 (Voelker et al., 1998; Hagen, 1999).

Radiocarbon dates (Table 1) from within the diamictons give ages of >25 ka B.P. (Figs. 4 and 5). The uncorrected date of 9120 ± 130 yr B.P. in the middle of 322PC is far too young—it was based on a very small sample. The dates near the base of the fined-grained lithofacies in 323PC and 326PC range between ∼12.7 and 13.4 ± ka B.P. (Figs. 4 and 5). In 322PC the date at the transition is 11.57 ± ka B.P., but the entire postglacial section in this core is only 30 cm or so; hence a date from a 2-cm interval can be only an estimate because we expect contamination from bioturbation. The dates immediately above the diamictons indicate that the net sediment accumulation rate (cm/ky) was extremely low, especially considering typical depositional rates of hundreds of centimeters per kiloyear during ice retreat off glaciated shelves (Andrews and Syvitski, 1994; Haflidason et al., 1998), and when compared with deglacial accumulation rates values off southwest Iceland (Syvitski et al., 1999; Jennings et al., 2000).

The seismic data along the track to 326PC suggested (Fig. 3) sediment loss above the Saksunarvatn reflector, and the radiocarbon dates confirmed this (Fig. 5). For example, the date above the tephra unit in 326PC2 is only 6.4 ± 0.08 ka B.P., and within a few centimeters of this date is an age of 1.57 ± 0.06 ka B.P. Similarly, at 326PC1, dates of ∼8.6 and 3.3 ka B.P. were obtained within a few centimeters of each other (Table 1).

Core Stratigraphy and Composition

The principal lithofacies we identified included massive matrix-supported diamictons, bioturbated or massive muds, mud with iceberg-rafted detritus, and sand (Figs. 4, 5). Very sharp contacts between diamictons and overlying sediments occurred in cores 322PC, 323PC, and 326PC1. We present data on sediment and foraminiferal properties and then evaluate the origins and environments of deposition.

B997-322PC: The lowermost unit is a diamicton (Fig. 8) with a dry density of 1.33 ± 0.12 g/cm³, although there is a hint from the X-radiography that this facies terminates above the base of the core (Fig. 4A). The major break in physical properties occurs at 50 cm, although another distinct change occurs at ∼38 cm. Within the upper 50 cm are several lithofacies changes including the occurrence of another diamicton. At 50 cm there is an abrupt decrease in dry sediment density and mass magnetic susceptibility (mass MS), whereas sand percentage increases. The ratio of ash:ash+lithics changes abruptly at 50 cm from ∼0.2 to 0.7, denoting a significant increase in the contribution of volcanic shards. This change is also associated with a decrease in clasts >2 mm (not shown). Dark basaltic glass rises first at 48 cm, but there is a distinct peak in clear, rhyolitic glass shards at 25 cm (Fig. 7). Total carbonate, a measure of marine productivity, remains at ca. 1.5% for much of the core, with a decrease in the upper diamicton, before rising rapidly to values of 3–4%. This pattern is paralleled by the percentages of foraminifera, with a distinct barren zone between 30 and 40 cm.

Thus the overall summary for this core is that conditions were relatively stable from near the base of the core to ca. 50 cm, when the environment changed considerably, as reflected in the physical properties within the uppermost 50 cm (Table 2). The ¹⁴C dates and tephra correlations (Figs. 4, 7) indicate that the diamictons are associated with dates 25 ka B.P. and that the transition to a thin, postglacial unit dates from 11.6 ka B.P.

B997-323PC: The sediments consists of over 2 m of diamicton (Fig. 8) overlain by a complex of lithofacies (Fig. 4B). The contact is sharp and occurs at 80 cm. Within the diamicton, sediment density averages 1.23 ± 0.12 g/cm³. The change in lithofacies at 80 cm is well seen in the density log and in the mass MS values (Fig. 4B and Table 3); however, the carbonate and total organic carbon data change closer to 40 cm (not shown), at the contact with a thin diamicton characterized by large clasts. The numbers of iceberg-rafted clasts are generally >20 although there are intervals with little evidence for such at 225, 80, and over the last 20 cm. Both the lithic and ash ratios show the major change at 80 cm (Fig. 4B), indicating that this change in lithofacies represents the major transition in the environments of deposition. Foraminifera were noted in small but consistent numbers throughout the lower diamicton but rise abruptly at 80 cm only to decrease to 0% at 40–50 cm before rising again. There is no obvious signal in the sand, which averages about 15%. Dates from within the lower diamicton give ages 25 ka B.P. whereas the overlying muds are 13 ka B.P. There is thus a temporal hiatus of ∼12 ky in the late Quaternary stratigraphic sequences in cores 322PC and 323PC.

B997-326PC1 and PC2: Cores 322PC and 323PC lie in Húnaflóadjúp, about equidistant from Húnaflói and the Strandir coast (Figs. 1, 2). Cores 326PC1 and 326PC2, however, lie in Reykjafjardaáll (Fig. 2).

Core 326PC1 has a diamicton in the lowermost meter of the core (James, 1999) (Fig. 5A). Dry sediment density averages 1.31 ± 35 g/cm², although it decreases steadily from the base of the core to the main lithofacies boundary at 195 cm. Mass MS shows no obvious decrease across this boundary, in contrast with 322PC and 323PC (Fig. 4A, B), but sand percentages do decrease in the muddy facies probably owing to an increase in airborne tephra (Table 3). The Saksunarvatn tephra is tentatively located at ∼120 cm (Fig. 7). The percentage of foraminifera in the >105-μm fraction increase from extremely low values (frequently barren) in the diamicton facies to between 20–30% in the middle section of the overlying muddy facies. The boundary between these two lithofacies is also evident in the lithics:total and ash:ash+lithics ratio graphs (Fig. 5A), which supports the argument that a major environmental change occurred at this transition. Dates above and below the facies boundary replicate the data from 322PC and 323PC (Fig. 4) and define a significant late Quaternary 12-ky depositional hiatus.

Core 326PC2 is predominantly fine-grained mud with various amounts of iceberg-rafted clasts. There is a distinct 10-cm-thick tephra unit at 70 cm which has a mixed composition (Fig. 7). The sediment above the ash has no visible iceberg-rafted component on the core X-radiographs. The radiocarbon dates and tephra stratigraphy (Figs. 5, 7) indicate limited deposition with some erosion and reworking over the last ∼12.7 ka B.P.

Foraminifera and Stable Isotopes: Cores 322PC and 323PC

Our work on the foraminifera was designed to contribute information on the origins of the sparse but pervasive foraminifera in the diamictons in 322PC and 323PC and to evaluate the environment during deglaciation. Foraminifera are present in 326PC and 326PC2 (see Fig. 5A and 5B), but species identifications have not been carried out. Four main biofacies have been delimited and are discussed further below and identified as 322PC-a, 323PC-A, etc (Fig. 9).

The presence of macro- and microfossils is a key element in debates over glacial versus glacial-marine origins of shelf diamictons (Dreimanis, 1979; Andrews and Matsch, 1983; Licht et al., 1999). Neither 322PC nor 323PC has sufficient sediment thickness for us to extract a high-resolution record of events since deglaciation 13 ka B.P.

The results of the foraminifera identification (Fig. 9A and 9B) confirm the presence of low but persistent numbers of benthic foraminifera within the diamictons, with values ranging in the few hundred individuals per100 g. Faunal zones 322PC-a and 323PC-A are dominated by Cassidulina reniforme, Elphidium excavatum, and Islandiella norcrossi. Also present in reasonable percentages are Cassidulina neoteretis and Cibicides lobatulus. This assemblage has similarities with the present benthic faunas from Djúpall (Fig. 1) (Jennings et al., in press) and from the East Greenland shelf 66–68°N (Jennings and Helgadóttir, 1994; Jennings and Weiner, 1996). Perhaps of most interest is the presence of C. neoteretis which today is linked with the presence of modified Atlantic Water.

The numbers of planktonic foraminifera per 100 g of sediment was very low within the diamicton facies, but sufficient numbers (∼20) were extracted for isotopic analyses (Fig. 10). The results from both 322PC and 323PC indicated that the surface waters were either quite fresh or that deposition occurred during an interval of moderate water temperatures. The δ¹⁸O indicate values of 3.8‰ to 3.0‰ in 322PC and between 3.5‰ and 2.5‰ in 323PC. Samples of N. pachyderma s from the present sediment/water interface at these two sites gave δ¹⁸O values of 2.79‰ and 2.38‰, respectively, indicating that near-surface waters were enriched (more glacial) in ¹⁸O during deposition of biofacies a and A (Fig. 9A, 9B). The δ¹⁸O values in 322PC and 323PC, facies a and A (Fig. 10), are similar to core-top ratios on planktonic foraminifera from the East Greenland margin between 66° and 68°N (Smith, 2001) and are generally lighter than MIS2 and MIS3 δ¹⁸O values from Blosseville Basin, north of the Denmark Strait (Hagen, 1999; Voelker, 1999).

At the transition from massive diamictons to a fine-grained, iceberg-rafted debris (Figs. 4, 9) there is a rapid increase in the numbers of both benthic and planktonic foraminifera (facies b and B). In 323PC the transition is marked by spikes in three of the four major species and the sudden appearance of Alabaminiella weddelensis, which might indicate an increase in phytodetritus (Jennings, personal commununication, 2002). However, there is no marked change in δ¹⁸O values (Fig. 10). There is a marked reduction in numbers of foraminifera in biofacies 322PC-c and 323PC-C; this interval is essentially a “barren zone” marked by an increase in iceberg-rafted accumulation (Fig. 4A, 4B). However, sufficient planktonic foraminifera were picked to record a sharp increase in δ¹⁸O to the largest values in either core, with δ¹⁸O values between 3.5‰ and 4‰ (Fig. 10). Based on the stratigraphy and bounding radiocarbon dates (Fig. 4) this cold, clast-rich interval coincides with all or part of the Younger Dryas cold interval (Ruddiman and McIntyre, 1981). East of our study area in cores HM107-04 and -05, the data of Eiriksson et al.'s (2000a, their Figs 8, 10) also show an interval with virtually no foraminifera at approximately the same time.

The last 10 ka B.P. are represented in these two cores by only ∼40 cm of sediment, whereas to the south of these sites the Húnaflóaáll sediment body (Fig. 3, core MD99-2269) reaches a thickness of 25 m over the same interval of time owing to sediment focusing (Andrews et al., 2002a). Numbers of foraminifera rise rapidly in both cores, although both also indicate a noticeable decrease in numbers in the top few centimeters (Fig. 9A, 9B). A notable feature of the Holocene is the reduced contribution of E. excavatum to the assemblages and the increase in Melonis barleeanus. At other sites on the north Iceland shelf, E. excavatum was reintroduced into the fauna in the last 1–2 ka B.P. (Eiriksson et al., 2000a; Andrews et al., 2001). The δ¹⁸O data have consistent values of between 2.5‰ and 2.8‰ (in facies d and D, Fig. 10).

The foraminifera and δ¹⁸O data indicate that the diamictons in 322PC and 323PC have consistent assemblages and isotopic ratios. The data do not show a “mixed” fauna, as has been noted in some glacial-marine/glacial settings caused by glacial reworking (Jennings et al., 1998); individual tests showed little evidence of crushing. The major distinction between the diamictons and the overlying “postglacial” sediments is in the number of foraminifera per 100 g of sediment (Fig. 9)—this transition occurs at the main lithofacies boundary in both cores.

Discussion and Conclusions

Quaternary sediments in Húnaflóadjúp consist of a basal diamicton overlain by a Holocene “drift-like” sediment body that reaches a maximum thickness of 25–30 m between 323PC and 326 (Fig. 3C) (Andrews et al., 2002a). Our sites (Fig. 2) are at the northern (322PC, 323PC) and southern (326PC) limits of the shelf sediment drift. At these sites the diamicton is overlain by only 30–200 cm of glacial-marine and marine sediments (Fig. 3A, B).

How we map the LGM extent of glacial ice onto the north Iceland shelf, off the Northwest Peninsula, clearly depends on how we interpret the origin of the diamictons in cores 322PC, 323PC, and 326PC (Figs. 4, 5). Similarly, estimates of the dates of deglaciation depend on the interpretation of the environments of deposition at the major lithofacies transition in these same cores. Radiocarbon dates at the base of the fine-grained facies are 12.5–13 ka B.P., whereas dates within the diamictons facies are 25 ka B.P. A critical question is how do we explain the 12-ky gap between the dates in the diamictons and those in the overlying fine-grained muds?

The diamictons at the three sites have similar properties (Table 3) and are distinct from the fine-grained lithofacies. There is a small but pervasive presence of foraminifera in cores 322PC and 323PC, whereas the base of 326PC1 is virtually barren (Fig. 5A). The key issue that pertains to the extent of glacial ice in the field area (Fig. 2) is whether the diamictons in 322PC, 323PC and -326PC1 (Figs. 4, 5) represent (1) basal tills; (2) glacially reworked glacial-marine sediments, possibly remolded as a deforming bed (Cofaigh and Evans, 2001); or (3) in situ glacial-marine sediments. Detailed investigations of the diamictons are in progress to better understand their depositional history (Principato, 2000). If the diamictons are tills or glacially reworked sediments, then the limit of glaciation is seaward of a core site, whereas if the sediments represent glacial-marine deposition, then the glacial margin must have lain landward. The persistent, if limited, presence of intact foraminifera in 322PC and 323PC (Fig. 9) suggests that either option 2 or 3 applies to these outer sites, whereas the lack of foraminifera in 326PC1 may indicate an origin more associated with subglacial deposition, hence probably option 1. The first two scenarios would place the limit of the LGM glaciation close to the north Iceland shelf break beyond 322PC, whereas option 3 would draw the LGM grounding line somewhere between 326PC and 323PC. The dry density data from the diamictons (Table 3) does not indicate substantial compaction; indeed, they are substantially lower than values from glacial-marine sediments in the Puget Lowlands, where densities ranged between 1.62 and 2.2 g/cm³ (Easterbrook, 1964). Data from a wide range of glacial-marine sediments on the northeast Canadian margin gave wet sediment densities of between 1.65 and 2.1 g/cm³ for ice-proximal glacial-marine sediments (Andrews et al., 1991), which compares favorably to data from the diamictons in 322PC, 323PC, and 326PC1 of between 1.6 and 2.0 g/cm³.

However, an in situ glacial-marine origin for the diamictons poses chronological and depositional problems. For example, why are there no sediments >13 and <25 ka B.P.? We suggest that the application of “Occam's razor” leads to the conclusion that the stratigraphy and chronology at 322PC and 323PC are best explained by glacial ice overrunning the sites and remolding glacial-marine sediments deposited 25 ka. The ice cap persisted in Húnaflóadjup from sometime after 25 ka until ∼13 ka. Given the evidence for a restricted ice cap on the Northwest Peninsula, with an ice stream only extending a little way beyond the mouth of Isafjordardjup (Fig. 1) (Andrews et al., 2002c; Geirsdóttir et al., 2002d), then our conclusion requires an extensive ice stream in Húnaflóaáll/Húnaflóadjup fed by ice accumulating in the highlands of west-central Iceland, possibly centered on the Langjökull massive (Bourgeois et al., 2000).

Regardless of the origin(s) of the diamictons, a major puzzle is why deposition during deglaciation (ca. 9–13 ka B.P.) was so limited (Figs. 3, 4, and 5)—this is unusual, especially when compared to the considerable thicknesses of glacial-marine sediments on the formerly glaciated continental margins of northeast Canada (Piper et al., 1991; Andrews, 2000), East Greenland (Dowdeswell et al., 1994), Norway (Vorren et al., 1983), and SW Iceland (Syvitski et al., 1999; Jennings et al., 2000). The most probable explanation for the limited thickness of glacial-marine sediments on the north Iceland shelf is that the ice margin retreated very rapidly across the shelf, giving limited time for sediment accumulation. The ice-rafted, clast-rich interval and foraminiferal barren zone in 323PC suggests a return to glacial-marine conditions during the Younger Dryas cold event. The regional interpretation of the strong Younger Dryas signal (Fig. 4) awaits integration into the evaluation of ice-rafted sediments from sites to the west (Geirsdóttir et al., 2002) and east (Kristjansdóttir, 1999; Eiriksson et al., 2000a) of our area.