Lateglacial to Holocene relative sea‐level changes in the Stykkishólmur area, northern Snæfellsnes, Iceland

Comparatively little research has been undertaken on relative sea‐level (RSL) change in western Iceland. This paper presents the results of diatom, tephrochronological and radiocarbon analyses on six isolation basins and two coastal lowland sediment cores from the Stykkishólmur area, northern Snæfellsnes, western Iceland. The analyses provide a reconstruction of Lateglacial to mid‐Holocene RSL changes in the region. The marine limit is measured to 65–69 m above sea level (asl), with formation being estimated at 13.5 cal ka BP. RSL fall initially occurred rapidly following marine limit formation, until ca. 12.6 cal ka BP, when the rate of RSL fall decreased. RSL fell below present in the Stykkishólmur area during the early Holocene (by ca. 10 cal ka BP). The rates of RSL change noted in the Stykkishólmur area demonstrate lesser ice thicknesses in Snæfellsnes than Vestfirðir during the Younger Dryas, when viewed in the regional context. Consequently, the data provide an insight into patterns of glacio‐isostatic adjustment surrounding Breiðafjörður, a hypothesized major ice stream at the Last Glacial Maximum.


Introduction
During the Last Glacial Maximum (LGM), Iceland was covered by a considerable ice mass (Ing olfsson et al., 2010). However, in comparison with the UK and Fennoscandia, the extent and configuration of the LGM Icelandic Ice Sheet (IIS) is still relatively poorly understood, particularly in the northwest (Andrews and Helgad ottir, 2003). Determining the scale of the LGM IIS is important, as freshwater input to the North Atlantic during deglaciation could have affected global thermohaline circulation (Hubbard et al., 2006). Consequently, various techniques have been used to reconstruct the LGM IIS, including submerged feature mapping (e.g. Olafsd ottir, 1975;Spagnolo and Clark, 2009), glacial striation identification (e.g. Keith and Jones, 1935;Hoppe, 1982), sedimentology (e.g. Andrews et al., 2000), ice sheet modelling (Hubbard et al., 2006) and ice stream investigation (e.g. Bourgeois et al., 2000). In addition, relative sea-level (RSL) studies have been used to determine the patterns of deglaciation in Iceland, through both marine limit and geomorphological mapping (e.g. Ing olfsson, 1991;Norðdahl and P etursson, 2005) and isolation basin study (e.g. Rundgren et al., 1997). Despite this range of approaches, there remains uncertainty surrounding the vertical and lateral extent of the LGM IIS (Rundgren and Ing olfsson, 1999) and the associated style and pattern of deglaciation (Norðdahl et al., 2008).
One important area for LGM IIS investigations is Breiðafj€ orður, a large fjord separating Vestfirðir and Snaefellsnes (Fig. 1), as it was potentially the location of a major ice stream (Bourgeois et al., 2000;Hubbard et al., 2006). Previous research has investigated several features within the region, highlighting the potential maximum ice sheet extent at the LGM (e.g. Olafsd ottir, 1975), former ice flow dynamics (Bourgeois et al., 2000) and marine limit elevations (e.g. Norðdahl and P etursson, 2005; Fig. 1). The study of RSL on the northern coastline of Breiðafj€ orður has also provided an accurate age for deglaciation, alongside the rates of associated glacio-isostatic adjustment (GIA) over the Lateglacial and Holocene (Lloyd et al., 2009;Fig. 1).
In this paper, we present the results of RSL study from Snaefellsnes, on the southern coast of Breiðafj€ orður, with the aim of reconstructing postglacial RSL change to investigate deglaciation and ice loading patterns, through comparison with existing datasets. Six isolation basins and two coastal lowland cores were investigated, alongside mapping of the local marine limit, to produce a record of postglacial RSL change. This record provides an opportunity to determine rates of GIA for comparison with Bjarkarlundur, situated 60 km north-northeast on the northern coast of Breiðafj€ orður (Lloyd et al., 2009).

Previous Research
Because of limited empirical evidence of LGM IIS extent, icesheet modelling studies have relied on a small number of dated raised shoreline and marine limit sites and a suite of undated marine limit elevations (e.g. Le Breton et al., 2010). The marine limit elevation, or highest point reached by postglacial sea level (Andrews, 1970), is linked to the timing of deglaciation at particular locations. It is therefore important to establish the age of marine limit formation, providing an opportunity to predict ice thickness and extent, when integrated into GIA models. Unfortunately, marine limit and raised shoreline sites often suffer from a lack of dateable material (e.g. Principato, 2008) and poor spatial coverage, as well as having difficulties in relating the feature formation to mean sea level (Lloyd et al., 2009). Despite this, several existing RSL and deglacial records have been developed using geomorphological evidence (Norðdahl and P etursson, 2005;Norðdahl et al., 2008).
Isolation basin sediments have been widely used to construct comprehensive postglacial RSL records for several locations, including Norway (Kjemperud, 1986), Russia (Corner et al., 2001), the UK (Shennan et al., 2000) and Greenland (Long et al., 1999). Isolation basin studies have also been previously undertaken in Iceland (e.g. Rundgren et al., 1997;Lloyd et al., 2009). These RSL records can act as important tests for models of GIA, providing an opportunity to investigate rates of isostatic uplift or subsidence over time. The first isolation basin study in Iceland was undertaken in northernmost Skagi (Fig. 1), demonstrating a total fall in RSL of 45 m at an average rate of À15.5 mm cal a À1 between ca. 13.2 and ca. 10.3k cal a BP, during which two transgressions of 5-m amplitude occurred (Rundgren et al., 1997). These transgressions have been linked to expansions of the IIS during the Younger Dryas and Preboreal (Rundgren et al., 1997;Norðdahl and P etursson, 2005).
In Bjarkarlundur, Vestfirðir, a fall in RSL of $80 m since deglaciation has been demonstrated (Lloyd et al., 2009). The research highlights two periods of rapid RSL change, with a rate of À38 mm cal a À1 during the Bølling-Allerød and À16 mm cal a À1 during the early Holocene (Lloyd et al., 2009). It was noted that the Younger Dryas reduced the rate of RSL fall in southern Vestfirðir to À4 mm cal a À1 (Lloyd et al., 2009). In Bjarkarlundur, RSL fell below present around 9 cal ka BP, which correlates with the isolation basin record from Skagi (Rundgren et al., 1997) alongside records generated by other sources (Norðdahl and P etursson, 2005). This fall below present at Bjarkarlundur was followed by a transgression during the late Holocene from which RSL is assumed to have fallen to present (Lloyd et al., 2009).
Furthermore, RSL study in western Iceland has demonstrated a series of raised shoreline features from the marine limit to close to present sea level (Ing olfsson, 1988;Ing olfsson and Norðdahl, 2001;Norðdahl and P etursson, 2005;Norðdahl et al., 2008;Ing olfsson et al., 2010). Ing olfsson and Norðdahl (2001) report marine limit shorelines at 105 and 148 m asl in Akrafjall and St ori-Sandh oll, Borgarfj€ orður, western Iceland, with St ori-Sandh oll dating to ca. 15.45 AE 0.245k cal a BP (uncorrected). A second extensive Younger Dryas shoreline is reported at 60-70 m asl (Ing olfsson and Norðdahl, 2001). The high marine limit elevations in the Borgarfj€ orður region are taken to be a consequence of rapid deglaciation in western Iceland (Ing olfsson and Norðdahl, 2001).

Research location
Snaefellsnes is a large peninsula in western Iceland, forming the southern coastline of Breiðafj€ orður (Figs 1 and 2). Snaefellsnes is dominated by the ice-capped Snaefellsj€ okull volcanic system, which is situated to the extreme west of the peninsula. The geology of the area is characterized by the Snaefellsnes Volcanic Belt which overlies Tertiary basalts and trends from east to west, contrasting with the typical northsouth trend of the central volcanic zones (Sigurdsson, 1970). The region is classified as tectonically inactive, despite several small magnitude seismic events between 1912 and 1962 (Sigurdsson, 1970).

Methodology
This research uses a combination of bio-and lithostratigraphic data to investigate RSL changes using a range of field and laboratory methods. The key technique employed in this research is the isolation basin methodology (Kjemperud, 1986). Isolation basins are rock depressions, which have over time been inundated by and isolated from marine conditions (Lloyd et al., 2009), denoted by an isolation contact within the sediments. Kjemperud (1986) notes three isolation contacts within isolation basins: sedimentological, hydrological and diatomological. An accurate record of environmental change is preserved due to the impervious rock 'sill' (Lloyd and Evans, 2002), the altitude of which must be accurately measured, as it is the point which controls the dominant environmental conditions (marine, brackish or freshwater) at each site.
Potential isolation basin sites for analysis were identified in the field using the characteristics outlined by Long et al. (2011). Initial site stratigraphies were determined by coring transects using a gouge corer, with samples for analysis retrieved using a Russian or Livingstone corer depending on the site stratigraphy. Sediments were extracted from an infilled section of the basin or from the rear of a small boat where the lake covered most of the basin. Sediments were described using the Troëls-Smith (1955) classification scheme.
Samples were prepared for diatom analysis using the standard techniques outlined by Palmer and Abbott (1986). A minimum of 250 diatoms were enumerated per sample, using a range of taxonomic guides for identification (e.g. Brun, 1965;Foged, 1972Foged, 1973Foged, , 1977Hartley, 1996). Following identification, diatoms were grouped into five classes: marine, brackish, salt tolerant, freshwater and salt intolerant (Hustedt, 1957). Within this paper, summary diagrams are provided for each site, with full diatom counts available as supplementary information (Table S1).
To determine the elevation of former RSL, the indicative meaning should be subtracted from the sill altitude (see Shennan et al., 1999). The indicative meaning is defined as the relationship between the dated sea-level index point and the tidal frame (Van Der Plassche, 1986), which was determined by combined bio-and lithostratigraphic analysis at each location to identify the hydro-and diatomological isolation contacts (Kjemperud, 1986). Mean high water spring tide (MHWST) is frequently used for the diatomological isolation contact (Lloyd et al., 2009;Long et al., 2011), which is characterized by predominantly freshwater conditions with a minor brackish element (Shennan et al., 1994). The hydrological isolation contact represents entirely freshwater conditions and equates to highest astronomical tide (HAT; Shennan et al., 1994).
Tephrochronology and radiocarbon dating were used to provide chronological control for the diatomological isolation contacts. The major element composition of tephra layers was determined by wavelength dispersive analyses on a fivespectrometer Cameca SX100 electron microprobe at the Tephra Analytical Unit, University of Edinburgh. Tephra samples were subjected to an acid digestion before analysis (Persson, 1971) and were subsequently sieved at 63 and 125 mm (e.g. Larsen et al., 1999). Tephra results were compared with previously published compositions using Tephrabase to identify individual horizons (Newton, 1996;Newton et al., 2007). Radiocarbon samples were generated as accelerator mass spectrometry bulk dates from organic sediments close to the diatomological isolation contact, following standard preparation techniques. Radiocarbon ages were subsequently calibrated using CALIB 7.0 (Stuiver et al., 2005) and the IntCal13 calibration dataset (Reimer et al., 2013).

Results
Sites were selected at a range of altitudes from present sea level to the marine limit to investigate RSL changes throughout the Lateglacial and Holocene. Samples were collected from a small geographical area to minimize the impacts of differential rates of isostatic rebound (Fig. 2). Radiocarbon analyses at each site are summarized in Table 1 as 2s ranges, with tephra geochemistries presented in Table S2.
Borgarland 10 (65˚2 0 40.78N, 22˚43 0 35.51W) core elevation: þ1.7 m asl Borgarland is an extensive coastal lowland area grading into a modern salt marsh to the north of Helgafell on the Thorsnes peninsula (Fig. 2). The lowland contains a former natural channel and later man-made drainage system. A total of 18 diatom samples were analysed, showing a transition from predominantly brackish to mainly freshwater conditions, with a minor brackish component towards the top of the core ( Fig. 3A; Appendix). The diatomological isolation contact at 50 cm yields a radiocarbon age for the decrease in marine influence of 7156-7252 cal a BP.
Borgarland 11 (65˚2 0 40.78N, 22˚43 0 35.51W) core elevation: þ3.3 m asl A second core, Borgarland 11, was taken close to the location of Borgarland 10 ( Fig. 2; Appendix). Four diatom samples were counted to a minimum of 250 diatoms, but preservation was generally poor (Fig. 3B). There were no visible tephra deposits within the sediment core. The diatom flora indicate a brackish-marine signal from the base of the core up to 104 cm (Fig. 3B). A radiocarbon date from 104 cm dates the transition from brackish to fully freshwater conditions to 9915-10 097 cal a BP.
Skjaldarvatn (65˚2 0 50.67N, 22˚47 0 11.35W) sill altitude: þ4.4 m asl Skjaldarvatn is located to the south-west of the Thorsnes peninsula (Fig. 2). The basin is elongated, 700 m long and 180 m wide and has an infilled section to the north-east of   Skjaldarvatn lake. Higher ground is found to the north, south and east, with a stream draining the lake to the south-west. The bedrock sill was located within the stream bed, lying at þ4.37 m asl. The sample for analysis was retrieved from the infilled section to the north-east (Appendix). A tephra layer at 546 cm was analysed, but could not be geochemically matched to any previously published record. Diatom analysis demonstrates a transition from marine to brackish dominance, followed by freshwater and salt-intolerant species dominance (Fig. 3C). The shift in diatom assemblage from marine-brackish to freshwater dominance at 540 cm represents the isolation contact and is radiocarbon dated to 11 253-11 619 cal a BP (Fig. 3C).
Þingvallavatn (65˚3 0 33.30N, 22˚42 0 43.16W) sill altitude: þ5.3 m asl Þingvallavatn is a basin situated to the north of the Thorsnes peninsula (TH1, Fig. 2). The basin contains a lake approximately 300 m in diameter, with the basin sill being covered by peat, situated to the north-east. A grid of cores led to the sill being identified at þ5.34 m asl. Two transects were cored using a boat, with a sample for analysis being collected near the centre of the lake (Appendix). The tephra layers at 456 and 479 cm were analysed, providing chemical signatures that could not be matched to published data. Diatom analyses (Fig. 3D) revealed a basal marine-brackish zone, moving to a transitional brackish phase at 505-498 cm which is then replaced by freshwater conditions. The diatomological isolation contact is at 507 cm, providing a date for isolation of 11 089-11 219 cal a BP.
Saurar 1 (65˚1 0 4.53N, 22˚41 0 47.11W) sill altitude: þ9.0 m asl The Saurar 1 basin is approximately 7 km south of Stykkish olmur (SA1, Fig. 2). Saurar 1 is situated within an area of low, undulating topography and is dissected by a forest road. The basin measures approximately 350 m long by 120 m wide, with an outlet stream draining the site to the north-west. The basin sill was identified based on a grid of cores and a sample was retrieved from the centre of the site (Appendix). Tephra analysis was undertaken on one sample from 334 cm, which demonstrated a mixed geochemical signature. Diatom analysis identified a predominantly freshwater flora throughout the core, but with a significant marine-brackish component to the flora towards the base of the core. The diatomological isolation contact is at 332 cm, which was radiocarbon dated to 12 135-12 544 cal a BP (Fig. 3E).
Helgafellsvatn (65˚2 0 18.87N, 22˚44 0 23.25W) sill altitude: þ12.8 m asl Helgafellsvatn is situated below Helgafell, a small hill of 60 m, which dominates the local landscape (HE1, Fig. 2). The basin consists of a lake approximately 750 m long and 400 m wide, as well as an extensive infilled section to the northeast. The basin sill was located close to the church at Helgafell. A transect of cores were taken from the infilled section at the lake edge, with a sample for analysis being extracted close to the centre of the transect (Appendix). The tephra layers at 578 and 615 cm are geochemically similar in composition but could not be identified and therefore were unable to assist in the constraint of isolation. Diatom analysis shows a lower marine-brackish zone (660-635 cm), an intermediate transitional zone (635-616 cm) and an upper freshwater zone (616-560 cm). The diatomological isolation contact at 616 cm was radiocarbon dated, providing an age of 11 224-11 408 cal a BP (Fig. 3F).
Saurar 3 (65˚0 0 18.79N, 22˚43 0 6.19W) sill altitude: þ16.2 m asl The basin at Saurar 3 is located along the same forest road as Saurar 1 (Fig. 2). The basin is made up of a large lake and bog, with the basin measuring 570 m by 180 m. Higher ground is found to the north, west and east, with a lower lying area to the south. The basin sill was identified through a coring grid to the south of the lake. A transect of cores was investigated to the north of the lake. A sample for analysis was collected using a Livingstone corer towards the centre of the core transect (Appendix). Geochemical analysis was undertaken on the two tephra layers. The upper tephra layer at 755 cm had a distinct Veiðiv€ otn-type signature, with the lower deposit at 790 cm probably being sourced from Grímsv€ otn. Diatom analysis identified a transition from a basal marine-brackish zone (845-825 cm), to a brackish zone (825-810 cm) then an upper freshwater zone from 810 cm. A radiocarbon date from the diatomological isolation contact at 808 cm provided a calibrated age of 12 558-12 646 cal a BP (Fig. 3G).
Ytra-B arvatn (64˚59 0 2.97N, 23˚11 0 39.34W) sill altitude: þ57.6 m asl Ytra-B arvatn is located on the Setberg peninsula approximately 20 km west of Stykkish olmur, and consists of a large lake with a diameter of approximately 220 m (YBR1, Fig. 2). The lake is drained to the west by a small stream, with the sill being identified within the stream bed. Higher ground is found to the north and south, with the basin situated within a small valley. The basin is situated below the highest raised shoreline identified in the area and thus provides a minimum age for highest postglacial sea level. Two perpendicular transects were cored within the basin to establish site stratigraphy, with a sample being taken from the northern part of the lake, which yielded thicker sediments (Appendix). Within the core sample, two distinct tephra horizons at 210 and 354 cm underwent geochemical analysis. The upper tephra deposit did not have a homogeneous geochemical signature, being of mixed origin. However, the lower tephra deposit demonstrates a correlation with the Saksunarvatn tephra, providing an age of 10 210 AE 35 cal a BP (Lohne et al., 2014;Fig. 4). A radiocarbon date from the diatomological isolation contact at 468 cm returned a calibrated age of 20 074-20 536 cal a BP; Fig. 3H. A second date at 440 cm returned an age of $24 000 cal a BP, suggesting a problem with contamination by old carbon at the site.

Marine limit formation in northern Snaefellsnes
In the Stykkish olmur area, the highest raised shorelines were measured between 65 m (BR1) and 69 m (OS1) asl (Fig. 2). These local marine limit values are lower than postglacial marine limits identified on the northern shore of Breiðafj€ orður at 78-98 m (Lloyd et al., 2009) and those identified in inner Breiðafj€ orður at 90-110 m (Norðdahl and P etursson, 2005). However, the marine limit elevation identified in the Stykkish olmur area is similar to the altitude of the postglacial marine limit from Skagi at 65 m (Rundgren et al., 1997) and elsewhere in western Iceland at 60-70 m (Ing olfsson, 1991), which formed during the B€ olling Period (Norðdahl and P etursson, 2005). In addition, the marine limit lies at a similar elevation to the Younger Dryas marine limit identified in innermost Breiðafj€ orður (Norðdahl and Asbj€ ornsd ottir, 1995). Consequently, if the Stykkish olmur area marine limit is to be used to assess GIA in the region, detailed chronological control of the marine limit is needed, as marine limits of similar elevation appear not to have all formed simultaneously.
Isolation basin studies provide an opportunity to determine the age of the marine limit at a given location and also the minimum age for deglaciation where sites with elevations close to the marine limit are available (e.g. Lloyd et al., 2009). Ytra-B arvatn, the highest basin investigated in this study, at 57.6 AE 0.6 m asl could therefore provide a minimum date for deglaciation in northern Snaefellsnes. The diatom assemblage of Ytra-B arvatn demonstrates a weak brackish signal towards the base of the core sample (Fig. 3H). This probably indicates a weak marine signal, as the basin is close to isolation and therefore only inundated for part of the tidal cycle.
Three radiocarbon dates were obtained at Ytra-B arvatn to establish the timing of the decrease in marine influence at the site and establish the age of tephra deposits found in the sediment sample retrieved. The lowest bulk sediment sample at 468 cm provided an age of 16 841 AE 76 14 C a BP (20 074-20 536 cal a BP), which is considerably older than previously reported isolation basin records (e.g. Rundgren et al., 1997;Lloyd et al., 2009). This sample was organic-poor and it is therefore likely that the age generated is misrepresentative of the basin isolation, being affected by the in-washing of carbon from the surrounding landscape (Bj€ orck and Wohlfarth, 2001) or reservoir effects. Bj€ orck and Wohlfarth (2001) have demonstrated that similar organic-poor sediments have produced dates which are thousands of years older than organic-rich sediments in similar settings. Previous research has highlighted that plant macrofossils provide the most reliable dating constraints from lake sediments (Abbott and Stafford, 1996), but the lack of identifiable plant macrofossils within the isolation basin sediments led to the use of bulk accelerator mass spectrometry dates in this study. Similar limitations have been highlighted in other isolation basin records in Iceland (e.g. Lloyd et al., 2009).
Consequently, the date produced from the commencement of organic sedimentation at the site may be more representative of basin isolation, once the position within the tidal frame represented has been established. However, a second date at 440 cm provided an even older age (Table 1), suggesting a problem with old carbon contamination at the site. The upper tephra deposits within the Ytra-B arvatn sediment sample were of mixed geochemistry and therefore probably represent the reworking of tephra from the surrounding landscape. The geochemical analysis of the lower tephra deposit at YBR1, however, demonstrates a good correlation to the Saksunarvatn tephra (Fig. 4). Using this as a chronological marker (10 210 AE 35 cal a BP, Lohne et al., 2014) and assuming constant sedimentation throughout the core sample, it is possible to estimate the timing of isolation at the site, if all radiocarbon ages are discounted (due to potential contamination). If these two assumptions are accepted, basin isolation can be estimated to ca. 13 500 cal a BP (Fig. 5). Indeed, this is likely to be a minimum age, as the rate of sedimentation is likely to be relatively lower initially after isolation of the basin (relatively organic-poor accumulation), rising slightly as organic content increases.
Extrapolation using the sedimentation rate between the upper radiocarbon date at 210 cm and Saksunarvatn tephra at 354 cm provides a younger age of isolation of ca. 12.2 cal a BP, which would correlate well with the Younger Dryas shorelines in inner Breiðafj€ orður (Norðdahl and Asbj€ ornsd ottir, 1995). The full range of possible ages for isolation of Ytra-B arvatn are therefore plotted on Fig. 5. However, the younger extrapolated isolation age is younger than isolation ages of lower basins, such as SA1 and SA3. For this scenario to be accepted, both lower basins would need to be affected by contamination from older carbon within the landscape. While this is possible, our preferred interpretation is for an age for the isolation of YBR1 of ca. 13.5 cal ka BP, which correlates well with the results of Lloyd et al. (2009).
Accordingly, we propose that the marine limit identified in the Stykkish olmur area formed before the Younger Dryas, supported by the radiocarbon ages retrieved from lower elevation sites. Previous studies in the Bjarkarlundur area (Lloyd et al., 2009), northernmost Skagi (Rundgren et al., 1997) and lower Borgarfj€ orður (Ing olfsson and Norðdahl, 2001) have determined ages of 14 125 AE 340, >14 000 and 15 450 AE 245 cal a BP, respectively, for the marine limit shorelines in these areas. The estimated age of the marine limit in the Stykkish olmur area based on the isolation of Ytra-B arvatn (>13 500 cal a BP), therefore, appears reasonable. Tephra geochemistry for Ytra-B arvatn, northern Snaefellsnes. Saksunarvatn data from Tephrabase (www.tephrabase. org;Newton, 1996). Analytical conditions: beam voltage, 15 kV; beam current, 2 nA for Si, K, Al, Fe, Mg, Na and Ca and 80 nA for Ti, Mn and P; beam size, 8 mm. A mixture of minerals, metals and artificial glasses were used to calibrate the instruments and analytical stability was checked using a BCR2G standard.

Lateglacial to mid-Holocene RSL change
The six isolation basin and two coastal lowland sites investigated in northern Snaefellsnes constrain the RSL change during the Lateglacial to mid-Holocene (Table 1; Fig. 6). Based on our preferred interpretation a new RSL curve for the Stykkish olmur area has been constructed using the isolation basin and coastal lowland sea-level index points (Fig. 6), with chronological control being provided through radiocarbon dates and tephra analysis.
Following deglaciation, RSL fell rapidly over the course of the Lateglacial and early Holocene in northern Snaefellsnes. Following the isolation of Ytra-B arvatn (YBR1, 57.6 m asl) RSL fall occurred rapidly until the isolation of Saurar 3 (SA3, 16.2 m asl) at 12 544-12 722 cal a BP (Fig. 6). This RSL fall demonstrates the rapid rate of isostatic rebound, which is comparable to regional trends (Lloyd et al., 2009). This correlation is likely given the relative proximity of the two study areas ($60 km) (Fig. 1). The rates of initial RSL fall and isostatic rebound reported in Skagi (Rundgren et al., 1997) are notably lower than those reported for the Stykkish olmur and Bjarkarlundur areas (Lloyd et al., 2009).
Following the isolation of Saurar 3 (SA3, 16.2 m asl), the rate of RSL change decreased to À24.5 mm cal a À1 until the isolation of Saurar 1 (SA1, 9.0 m asl) at 12 135-12 544 cal a BP. This reduced rate of RSL change corresponds to a decrease in the rate of isostatic rebound to þ46 mm cal a À1 , when eustatic sea-level changes over the period are taken into consideration (Fairbanks, 1989). Previous studies have demonstrated a similar reduction in the rate of RSL change during this period, associated with the expansion of Younger Dryas ice caps and glaciers (Lloyd et al., 2009).
From the RSL record generated in this study, there is only evidence for a limited Younger Dryas ice readvance, as there is little indication of a notable transgression during this period in the Stykkish olmur area. The RSL curve generated for the Stykkish olmur area suggests an exponential decrease in the rate of glacio-isostatic rebound since deglaciation and thus minimal influence from Younger Dryas ice regrowth. It is also evident that RSL was higher on the northern coast of Breiðafj€ orður during this period (Lloyd et al., 2009) (Fig. 7). This difference in RSL could be attributed to a larger Younger Dryas ice cap in Vestfirðir than on the Snaefellsnes peninsula (Norðdahl and P etursson, 2005).
Following the isolation of Saurar 1 (9.0 m asl), RSL fell to between 2.6 and 3.2 m asl, with a decrease in marine influence noted at Skjaldarvatn (SK1) between 11 253 and 11 619 cal a BP and the isolation of Þingvallavatn (TH1) at 11 066-11 241 cal a BP (Fig. 6). These decreases in marine influence probably occurred in quick succession, as the error associated with the determination of site elevation and decrease in marine influence mean that these events could have occurred concurrently. Both Skjaldarvatn and Þingvallavatn have visible tephra deposits close to the point of decreased marine influence.
Þingvallavatn shows a clear shift from marine-brackish to freshwater dominance within the diatom assemblage (Fig. 3D) and contains two tephra layers with geochemical signatures that are unable to be matched to previous records. The geochemical signature attained from analysis of the Skjaldarvatn deposits shows a mixed assemblage, with tephra shards demonstrating Veiðiv€ otn-, Grímsv€ otn-, Katla-and Snaefellsnes Volcanic Belt-type geochemical compositions (Tephrabase, 2012). It is possible that the geochemical signal highlighted in the Þingvallavatn core is masked at Skjaldarvatn by the in-washing of additional material from the surrounding landscape. A mixed tephra composition is probably the result of an unvegetated landscape, where tephra can move freely without being trapped by plant material or sediment.
At Skjaldarvatn, the transition between marine-brackish and freshwater dominance is very sharp (Fig. 3C). This transition within the diatom assemblage is probably affected by the tephra deposit at 546 cm, although basin isolation had commenced before this point in the diatom assemblage. As a result, the age of isolation of Skjaldarvatn is older than anticipated due to the radiocarbon age being sought at the point of freshwater dominance rather than the completion of isolation within the diatom assemblage. However, there is relatively little influence on the RSL curve for the region.
It is clear that RSL was between 2.6 and 3.2 m asl between 11 066-11 241 and 11 253-11 619 cal a BP when corrected for the indicative meaning. As a result, the timing of isolation at Helgafellsvatn appears to be anomalous (Figs 5 and 6). The diatom assemblage for the site demonstrates a clear transition between marine-brackish and freshwater dominance, suggesting a relatively rapid isolation of the basin (Fig. 3). However, the timing of isolation appears anomalously young compared with the other sites presented here Figure 6. RSL curve for northern Snaefellsnes, with sea-level index points plotted against calibrated ages. Sea-level index point at HE1 is not used to constrain the curve due to a presumed incorrect age of isolation. Black crosses represent sea-level index points, with grey crosses showing limiting ages. Figure 7. RSL curves from north-west Iceland, demonstrating regional trends and variability between sites investigated in Snaefellsnes (plus (þ)), Bjarkarlundur, Vestfirðir (diagonal crosses (x), Lloyd et al., 2009) and Skagi (open circle (o), Rundgren et al., 1997). For Snaefellsnes, black plus symbol (þ) represent sea-level index points, with grey showing limiting ages.
(11 224-11 408 cal a BP). Based on our preferred interpretation, the date from Helgafellsvatn must therefore be anomalous; hence, it is not used to constrain RSL. An alternative interpretation accepting the age of the isolation contact at Helgafellsvatn would require a rise in RSL of approximately 5 m following the isolation of SA1 at 12 135-12 544 cal a BP, or a larger rise of approximately 8 m following the isolation of SK1 at 11 253-11 619 cal a BP. An additional alternative interpretation is that the isolation contacts for sites SA3, SA1, SK1 and (to a lesser degree) TH1 are too old and the rate of RSL fall during the Lateglacial is rather more gradual than our preferred interpretation suggests.

Mid-to Late Holocene RSL change
Following the isolation of Þingvallavatn (TH1; 5.3 m asl) at 11 066-11 241 cal a BP, a sea-level index point at the coastal marsh site at Borgarland (BO11) records RSL falling below 0.9 m at 9915-10 097 cal a BP (Fig. 6). Following this, RSL probably fell below present; the index point from BO10 records RSL below present (at À0.2 m) by 7148-7259 cal a BP (Fig. 6). This fall below present sea level is later than existing records, which demonstrate a fall below present sea level at ca. 8.2-10.6 ka cal BP (Rundgren et al., 1997;Norðdahl and P etursson, 2005;Lloyd et al., 2009). The RSL index point from BO10 suggests that RSL probably fell lower than this point during the mid-Holocene. Following this, there are two scenarios: a transgression during the mid-Holocene or a gradual rise to present. The altitude of transgression during the mid-Holocene is limited by the lack of a second marine incursion recorded in the diatom assemblage from Skjaldarvatn or Þingvallavatn. As a result, RSL did not reach higher than 2.6 AE 0.7 m asl during the mid-to late Holocene in the Stykkish olmur area. There is, however, a short-lived recurrence of mesohalobous species within the Borgarland sequence (BO10, Fig. 3A). Previous studies have noted the potential for Holocene transgressions in Iceland, including in the south-west at 2-6 m asl at 2.9 and 2.3 cal ka BP (Símonarson and Leifsd ottir, 2002) and Hvammsfj€ orður, western Iceland at 3 and 5 m asl after 4.5 cal ka BP (Norðdahl and P etursson, 2005;Ing olfsson et al., 2010). To further develop the late Holocene RSL record for Snaefellsnes, additional sites at low elevations are required. However, such sites were not identified in the Stykkish olmur area and sites on the southern coast of the Snaefellsnes peninsula were determined to be unsuitable for isolation basin research due to the nature of the basin sills.

Conclusions
The study of RSL change in the Stykkish olmur area offers an opportunity to investigate GIA on the south coast of Breiðafj€ orður, when coupled with existing isolation basin and marine limit records from the region. The RSL record from the Stykkish olmur area highlights a difference in patterns of GIA in and around Breiðafjorður since deglaciation. In the Stykkish olmur area, the postglacial marine limit is identified at 63-69 m asl. Previous research has highlighted higher marine limit elevations in northern and innermost Breiðafjorður than determined in this study. This indicates that ice thicknesses were probably greater in northern and innermost Breiðafj€ orður than in the Stykkish olmur area before deglaciation, or that deglaciation occurred later in the Stykkish olmur area.
RSL fell rapidly following deglaciation on both the northern (Lloyd et al., 2009) and the southern coastlines of Breiðafj€ orður. The rate of RSL fall in the Stykkish olmur area correlates well with Lloyd et al. (2009) when the hypothesized marine limit age of 13 500 cal a BP is employed. The new RSL record from the Stykkish olmur area also provides an opportunity to explore the consequences of Younger Dryas ice regrowth. There is no evidence for a transgression during the Younger Dryas based on the isolation basin data from the Stykkish olmur area. Instead, the RSL curve shows a decrease in the rate of isostatic rebound from the LGM. There is little evidence for the influence of Younger Dryas ice regrowth in the Stykkish olmur area in the RSL record. This differs from the results on the northern coastline of Breiðafj€ orður (Lloyd et al., 2009), suggesting variation in ice conditions between the north and south coast of Breiðafj€ orður during the Younger Dryas, and that the influence of any glacial loading was relatively localized and did not significantly affect the Stykkish olmur area.
During the mid-Holocene, RSL fell below present at 7148-7259 cal a BP following the decrease in marine influence at Borgarland 11 at 9915-10 097 cal a BP. This appears later than noted in previous studies in western Iceland, which have noted an RSL fall below present at $8.2-10.6 cal ka BP (e.g. Lloyd et al., 2009). There is evidence of a subsequent minor transgression during the Late Holocene reaching less than 2.6 m asl (based on the indicative meaning correction of the sill elevation at Skjaldarvatn). Following this, RSL regressed to present although additional RSL data are required to better constrain these recent RSL changes.

Supplementary information
Additional supporting information can be found in the online version of this article: Table S1. Full diatom counts from isolation basin and coastal lowland sediment core samples studied in the Stykkish olmur area, northern Snaefellsnes, Iceland. Table S2. Tephra geochemical data for isolation basin and coastal lowland sites analysed in the Stykkish olmur area, northern Snaefellsnes, Iceland.
Appendix. Lithostratigraphic data from the Stykkish olmur area.