Life on the periphery is tough: Vegetation in Northwest Iceland and its responses to early-Holocene warmth and later climate fluctuations

Field methods and stratigraphy

Four series of overlapping sediment cores were retrieved from near the centre of the palaeolake using a Livingstone piston corer with a Bolivia adaptor fitted with 75 mm diameter polycarbonate tubes. The sediments retrieved range from peat at the top of the core to silty gyttja, with minerogenic sediments at the bottom of the core.

A single depth profile was constructed by overlapping the main sequence of cores (KAGA4) with two other sequences (KAGA1 and KAGA3) to ensure continuity in the sediment column and to avoid potentially disturbed sediments at the top of the core segments. This was done by matching tephra layers, patterns in magnetic susceptibility (MS) and stratigraphic changes observed in the overlapping cores (see supplementary material, available online). The resulting sequence was 679 cm long. The cores were x-rayed for identification of tephra layers. MS measurements were performed on the split core segments at 1 cm intervals using a Bartington MS2 meter and Bartington MS2F probe (Dearing, 1994). Organic matter (OM; by loss on ignition) and dry bulk density (DBD) were measured at 1 cm contiguous intervals. OM was measured by subjecting 1.2 cm³ of sample to 550°C for 5 h (Bengtsson and Enell, 1986) and DBD by dividing the dry weight of a sample with the volume of the wet sample (Brady and Weil, 1996). The sediments analysed for this study are undisturbed lake sediments from 130 cm depth (10 cm above the Hekla 3 tephra layer; Dugmore et al., 1995) to the bottom of the core. Sediments higher in the core sequence were not analysed because of potential reworking as the lake depth decreased and peat formation began.

Pollen and macrofossil analysis

Subsamples for pollen analysis (2 cm³) were collected at 4–8 cm intervals between 133 and 675 cm depth in the core. Pollen samples were prepared using standard chemical methods of 10% HCl, 10% NaOH and acetolysis (Faegri and Iversen, 1989; Moore et al., 1991), as well as heavy-liquid separation (Björck et al., 1978; Nakagawa et al., 1998) using LST Fastfloat (a sodium heteropolytungstate solution) to remove minerals, tephra shards and organic debris. Two tablets containing spores of Lycopodium clavatum were added to each sample (batch number 1031; Stockmarr, 1971) to enable calculations of concentrations of pollen and spores in the samples, and to calculate pollen accumulation rates (PARs; grains cm⁻² yr⁻¹). The prepared samples were mounted on glass microscope slides in silicone oil of 12,500 cSt viscosity.

A minimum of 300 indigenous terrestrial pollen grains were counted for each sample, except for the three lowermost samples where pollen concentrations were extremely low. Identification of pollen grains and spores was based on Moore et al. (1991) and a pollen-type slide collection at the University of Iceland. Pollen categories and sums followed Hallsdóttir (1987) and Caseldine et al. (2006). Pollen and spore taxonomy followed Bennett (2007), with a few amendments to better reflect the Icelandic flora (Erlendsson, 2007). Percentages of fungal spores were based on the combined terrestrial pollen sum and sum of fungal spores. A record was made of non-triporate Betula pollen, high percentages of which are indicative of hybridization between Betula nana and Betula pubescens (Karlsdóttir et al., 2008). Identification of spores from coprophilous fungi relied on Van Geel et al. (2003).

The sediments were analysed for plant macrofossils at contiguous 5 cm intervals between 350 and 600 cm to establish the timing of Betula pubescens colonization of the area. Between 130 and 350 cm, and 600 and 675 cm, samples were analysed at every other 5 cm interval. Sample volume varied between 35 and 50 mL, depending on the material available. The volume of each sample was determined by displacement in water. The samples were washed through a 125 µm sieve, and vascular plant remains were picked out for identification. Identification was based on various references (e.g. Berggren, 1969; Birks, 2007; Cappers et al., 2012; Katz et al., 1965). Comparisons with reference material were made when possible. Concentrations of macrofossils in 50 mL were calculated. Betula pubescens Ehrh. (downy birch) and the subspecies Betula pubescens Ehrh. ssp. tortuosa (mountain birch) grow in Iceland. They are discussed herein as Betula pubescens (sensu lato), as the morphological variations in Icelandic birches are large and morphologies of the two may overlap (Anamthawat-Jónsson and Thórsson, 2003; Thórsson et al., 2001). Otherwise, plant taxonomy follows Kristinsson (2010). Pollen and macrofossil diagrams were constructed using TILIA (version 1.7.16) (Grimm, 2011). Pollen assemblages were divided into pollen zones (PZ) with the aid of CONISS.

Core chronology

The chronology for the core (Figure 2) was constructed using a combination of tephrochronology and radiocarbon dating. Tephra layers were identified visually when thick enough and finer layers identified by using x-ray images, spikes in MS and dips in OM measurements. Samples were taken from all suspected tephra layers found in the sediments and washed through a 90 µm sieve. After inspection in a stereomicroscope, the samples judged to consist of primarily pristine volcanic glass shards were mounted on slides, ground, polished and carbon coated. Geochemical analyses of volcanic glass were performed using Hitachi TM3000 SEM equipped with Bruker X-Flash silicon drift detector, using 15 kV beam voltage applied to a surface of 10 × 10 µm. Analyses were recorded with Bruker Espirit analysis software.

OPEN IN VIEWER

Figure 2.

Age–depth model for the Lake Kagaðarhóll core.

The tephrochronology of North Iceland suffers from a lack of dated tephra layers between the Hekla 5 tephra (7050 cal. yr BP; Thorarinsson, 1971) and the Saksunarvatn tephra (10,300 cal. yr BP; Rasmussen et al., 2006). This part of the core (362–620 cm) was systematically sampled every centimetre for macrofossils for dating. A total of nine macrofossil samples were sent for radiocarbon dating (Figure 2, Table 1). Only three samples of terrestrial macrofossils yielded enough material for dating. Because of the scarcity of terrestrial plant macrofossils in the core, six additional samples of aquatic plant remains (Potamogeton fruits and leaf fragments) were picked for dating (Table 1), as well as a sample of moss stems from minerogenic material in the lowest part of the core. The decision to opt for aquatic macrofossils over bulk sediment is based on the fact that radiocarbon dates of bulk material from Icelandic lakes are prone to errors and often do not correlate well with dated tephra layers (Caseldine et al., 2006). The reason for this discrepancy is probably erosion-derived soil organic carbon (cf. Gathorne-Hardy et al., 2009). Lakes in Iceland are devoid of inorganic carbon, and the lake is remote from geothermal activity, thus free of issues concerning these factors (e.g. Ascough et al., 2007; Axford et al., 2007). The radiocarbon dates were calibrated to calendar years BP using IntCal13 (Reimer et al., 2013) in OxCal version 4.2.2 (Bronk Ramsey and Lee, 2013). A smooth-spline model (Figure 2) was constructed in R using the package Clam (Blaauw, 2010).

OPEN IN VIEWER

Table 1.

Radiocarbon dates from the Lake Kagaðarhóll core.

Lab code	Depth (cm)	14C date	Error	δ13C	Average Calendar age	Range (cal. yr BP) 2σ error	Weight (mg)	Material
ETH-56125	385	7003	36	−20.6	7841 ± 95	7936–7746	0.83	Potamogeton leaf fragments
ETH-56126	401	7170	35	−35.0	7989 ± 53	8041–7936	0.34	Empetrum nigrum and Betula pubescens remains
ETH-56127	430	8128	37	−16.8	9120 ± 123	9242–8997	1.00	Potamogeton leaf fragments
ETH-56135	461	8209	38	−19.1	9159 ± 128	9286–9031	0.94	Potamogeton leaf fragments
ETH-56128 a	461	8132	89	−30.8	9086 ± 315	9401–8771	N/A	Miscellaneous Betula pubescens macrofossils
ETH-56129	524	8695	38	−11.7	9655 ± 109	9764–9546	0.99	Potamogeton fruits and leaf fragments
ETH-56130	583	8997	39	−11	10,092 ± 150	10,242–9942	1.00	Potamogeton fruits and leaf fragments
SUERC-47862	618	9258	35	−25 b	10,425 ± 131	10,556–10,294	1.00	Salix sp. leaf fragment
SUERC-49702	677	9695	30	−29.1	11,048 ± 158	11,206–10,890	4.50	Moss stems
SUERC-47863 c	320	5814	30	−27.8	6604 ± 75	6679–6529	32.40	Twig, Salix sp.
SUERC-47864 c	320	5788	30	−30.8	6581 ± 80	6661–6501	18.70	Twig, Betula sp.

a

Measured directly from CO₂ gas.

b

Assumed.

c

Dated from peat section BUR.

The age–depth model for the core (Figure 2) is based on the previously dated tephra layers (Appendix 1) Hekla 3 (3000 cal. yr BP), Hekla 4 (4200 cal. yr BP; Dugmore et al., 1995), Hekla Ö (6060 cal. yr BP; Gudmundsdóttir et al., 2011), Hekla 5 (7050 cal. yr BP; Thorarinsson, 1971) and Saksunarvatn (10,300 cal. yr BP; Rasmussen et al., 2006). The above tephra layers, except for the Hekla-Ö tephra, have a long and extensive record in North Iceland and the North Iceland Shelf and serve as regional marker horizons (Gudmundsdóttir et al., 2012, and references therein). One tephra layer not previously reported is used in the age–depth model (S-layer; Table 1; Appendix 1). We have observed this tephra layer in numerous peat sequences in Austur-Húnavatnssýsla and in three lake cores. Located between the Hekla-Ö tephra (Gudmundsdóttir et al., 2011) and the Hekla 5 tephra (Thorarinsson, 1971), this layer forms a distinct, dark and relatively coarse-grained stratum in organic sequences in the area. Its geochemical fingerprint is that of the Katla volcanic system (Appendix 1). One Betula spp. twig and one Salix spp. twig situated in direct contact with the tephra were extracted for radiocarbon dating from a peat sequence at the farm Stóra-Búrfell (Figure 1b; core BUR), about 4.5 km from Kagaðarhóll. The results of the two analyses provide dates of 6679–6529 and 6661–6501 cal. yr BP (Table 1), and these are fitted into the age–depth model for KAGA where the tephra was identified at 319–320 cm depth. The time frame of the Saksunarvatn tephra has not yet been established nor whether it is the product of several eruptions (Jennings et al., 2014). The layer is therefore handled as a single event in the age–depth model.

In addition to the dated tephra layers, six of the nine radio-carbon dates analysed complete the age–depth model. Two radiocarbon dates were rejected on the basis that they were too old. A Salix leaf fragment from right above the Saksunarvatn layer at 618 cm depth yielded an age of 10,425 ± 131 cal. yr BP, thus older than the tephra upon which it rests. A sample of Potamogeton leaf fragments from 430 cm depth gave an age that was too old and implies a sedimentation rate too high, considering no changes are seen in sediment characteristics based on the MS and DBD data. The third excluded ¹⁴C date, derived from a terrestrial macrofossil sample at 461 cm depth, yielded very little material for dating and was analysed as gas. Because of the large margin of error on the calibrated date, a sample of Potamogeton leaf fragments from the same depth was used in the age–depth model instead as it falls within the error of the terrestrial macrofossil sample, and is in sequence with other samples based on Potamogeton macrofossils between 461 cm and the Saksunarvatn tephra. Ages are given in calibrated years before present (cal. yr BP; with present assigned to AD 1950).

PZ 1 (c. 11,000–10,600 cal. yr BP; 680–650 cm)

Initially, the pollen assemblage (Figure 3) is characterized by a relatively high percentage of Poaceae (grass family) and pioneer taxa such as Oxyria digyna and Arenaria-type, and herbs such as Potentilla-type and Ranunculus acris–type. Occasional occurrences of Dryas octopetala and Sedum-type pollen are recorded in the lowest samples. Betula pollen percentages are relatively high in the two lowest samples. By c. 10,800 cal. yr BP, there is a peak in Empetrum nigrum pollen. Vaccinium-type pollen are found in small numbers from c. 10,700 cal. yr BP, as well occasional as Calluna vulgaris pollen. PARs (grains cm⁻² yr⁻¹) (Figure 4) are very low, with a total PAR of less than 100 grains cm⁻² yr⁻¹ during this period. Exotic pollen from Pinus and Alnus are recorded in the lowest samples. Spores from the coprophilous (dung-loving) fungi Sporormiella-type and Sordaria-type are found within this PZ (Figure 3). The few terrestrial macrofossils found (Empetrum nigrum, Juncus spp., Luzula spp., Lycopodium annotinum and Selaginella selaginoides) (Figure 5) compare well with the pollen assemblages. MS values are high within the zone, and DBD values decrease from c. 1.5 to 0.45 g cm⁻³ at the top of the zone accompanied by low OM percentages between c. 0.5% at the bottom of the zone, increasing to c. 9% at the top of the zone (Figure 6).

OPEN IN VIEWER

Figure 3.

Pollen percentage diagram from Lake Kagaðarhóll. Exaggeration curves of factor 4. Dots signify percentages below 1%. The lithology column shows dated tephra layers used in the age-depth model.

OPEN IN VIEWER

Figure 4.

Pollen accumulation rate (PAR, grains cm⁻² yr⁻¹) from Lake Kagaðarhóll for selected taxa.

OPEN IN VIEWER

Figure 5.

Plant macrofossil diagram from Lake Kagaðarhóll. Concentrations in 50 mL of sediment.

OPEN IN VIEWER

Figure 6.

Summary diagram for Lake Kagaðarhóll showing (a) Betula pollen accumulation rate (PAR) (line) and Betula pubescens macrofossil occurrence in 50 mL of sediment (bars); (b) percentage of Betula pollen; (c) percentage of non-triporate Betula pollen; (d) Juniperus communis PAR; (e) percentage Juniperus communis pollen; (f) total terrestrial PAR; (g) Renland ice-core δ¹⁸O record, 20-year average uplift corrected (Vinther et al., 2009); (h) June insolation at 60°N (Berger and Loutre, 1991); (i) magnetic susceptibility (MS; showing arbitrary values below 80); (j) dry bulk density (DBD); and (k) organic matter (OM; measured by loss on ignition).

PZ 2 (10,600–10,200 cal. yr BP; 650–600 cm)

Cyperaceae and Salix pollen increase in both relative and absolute numbers by c. 10,600 cal. yr BP (Figures 3 and 4). Although Empetrum nigrum pollen decreases after c. 10,600 cal. yr BP (Figure 3), the macrofossil record supports its continued presence with seeds and leaves (Figure 5). Pollen from pioneer herb taxa identified in PZ 1, Oxyria digyna, Arenaria-type, Potentilla-type and Ranunculus acris–type, decrease within this zone. There is a progressive decline in minerogenic content, apparent via decreasing MS values and lower DBD from c. 0.45 to 0.20 g cm⁻³. This is accompanied by an increase in OM deposition from c. 9% to 24% (Figure 6). A single Betula nana leaf is found at c. 10,500 cal. yr BP (Figure 5), and Betula pollen values are at c. 1–2% for this period (Figure 3).

The environment is disrupted by the deposition of the Saksunarvatn tephra, dated to c. 10,300 cal. yr BP (Rasmussen et al., 2006). The layer is marked by OM percentages close to zero and a steep increase in MS and DBD (Figure 6). The pollen assemblage immediately above the layer (617 cm) shows an increase in Poaceae, a relative decrease in Empetrum nigrum and Cyperaceae and some decrease in Salix pollen compared with the assemblages below the layer. Further above the tephra layer (609 cm), the pollen assemblage recovers to values similar to before its deposition, characterized by Cyperaceae, Empetrum nigrum, Salix and Ericales (Figures 3 and 4). A comparable assemblage is reflected in the macrofossil record with remains of Empetrum nigrum, Salix, Selaginella selaginoides and Carex (Figure 5). Fungal spores of Sporormiella-type and Sordaria-type occur regularly (Figure 3). Minerogenic inputs decline to pre-Saksunarvatn levels after c. 100 years according to the MS and DBD records (Figure 6).

PZ 3a (c. 10,200–9200 cal. yr BP; 600–470 cm)

Juniperus communis pollen appear in the record by c. 10,200 cal. yr BP and quickly increase to >30% and PARs > 700 grains cm⁻² yr⁻¹ by 10,100 cal. yr BP. Betula pollen increase rapidly to c. 20–30% by c. 10,000 cal. yr BP, an increase in Betula PAR from 14 grains cm⁻² yr⁻¹ at the beginning of the zone to a peak of >1000 grains cm⁻² yr⁻¹ between c. 9600 and 9400 cal. yr BP. Juniperus communis pollen increase to a peak of c. 40–60% between c. 9600 and 9200 cal. yr BP, with PARs of 1330–3070 grains cm⁻² yr⁻¹ (Figures 3 and 6). As these taxa become more abundant in the record, values of dwarf shrubs decrease, with a steady decrease in Empetrum nigrum pollen, especially after c. 9900 cal. yr BP, while Salix pollen remain at about 5–10% (Figure 3). This shift is also reflected in the macrofossil record where Betula nana fruits and catkin scales are relatively abundant in an otherwise limited record, and occasional Empetrum nigrum and Salix macrofossils are still found. One Juniperus communis macrofossil is recorded, a needle deposited at the same time as the peak in Juniperus communis PAR at c. 9300 cal. yr BP (Figure 5). One Betula pubescens fruit is found at c. 9600 cal. yr BP, but Betula pubescens catkin scales and fruits sustain a continuous record from c. 9300 cal. yr BP (Figures 5 and 6). Additionally, degraded Betula fruits that could not be identified to species level were found after c. 9600 cal. yr BP (Figure 5). Sporormiella-type and Sordaria-type spores decrease and almost disappear after c. 9000 cal. yr BP. Potamogeton pollen are very abundant within this zone but decrease abruptly at c. 9600 cal. yr BP (Figure 3) when Potamogeton natans fruits, that are abundant in the lower half of the zone, disappear. Potamogeton perfoliatus fruits are found in the record after c. 9500 cal. yr BP (Figure 5). Large fluctuations in MS and DBD values occur during this period. Spikes in MS occur between 9800 and 9300 cal. yr BP, accompanied by fluctuations in DBD between 0.17 and 0.34 g cm⁻³ with sand lenses observed in the sediments between 9700 and 9500 cal. yr BP (Figure 6). Total terrestrial PAR reaches its highest levels in the core within PZ 3a, with values of 4500–6100 grains cm⁻² yr⁻¹ between c. 9500 and 9300 cal. yr BP. The main contributor to the increase in total PAR is Juniperus communis. The proportion of non-triporate Betula pollen during this period ranges between 2% and 9% (Figure 6).

PZ 3b (c. 9200–8700 cal. yr BP; 470–436 cm)

Betula pollen percentages are >30% after c. 9200 cal. yr BP (Figure 3). This is accompanied by a decrease in Juniperus communis pollen (Figures 3 and 4). The total PAR values decrease mainly because of a drop in Juniperus communis PAR from 1460 grains cm⁻² yr⁻¹ at the bottom of the zone to 350 grains cm⁻² yr⁻¹ at the top. Betula PAR values measure between 600 and 1000 grains cm⁻² yr⁻¹ within this zone (Figures 4 and 6). PARs of Poaceae, Cyperaceae, Empetrum nigrum and herbs such as Thalictrum alpinum and Galium decrease (Figure 4). Ericales pollen disappear within this zone. Betula pubescens macrofossils are recorded continuously during this period (Figures 5 and 6). MS and DBD values are much lower and less variable than before, with DBD between 0.13 and 0.19 g cm⁻³ accompanied by elevated OM values between 29% and 43% (Figure 6).

PZ 3c (c. 8700–8200 cal. yr BP; 436–408 cm)

At c. 8700 cal. yr BP, Betula percentages drop to c. 25–30% (Figure 3), accompanied by Betula PAR drop to between c. 160 and 260 grains cm⁻² yr⁻¹ (Figures 4 and 6). Herbaceous taxa, such as Poaceae and Cyperaceae, increase (Figure 4). The decrease in Betula pollen spans about 500 years, over which period Betula pubescens macrofossils almost disappear from the record. Only a few macrofossils of Empetrum nigrum and Selaginella selaginoides are recorded during this period (Figure 5). The total PAR is extremely low, at levels similar to the early Holocene (Figure 6). OM deposition is relatively high at c. 34–48%. An unidentified tephra layer is seen in the MS, OM and DBD records at c. 8550 cal. yr BP (Figure 6), otherwise values for MS and DBD remain relatively stable within this zone, and DBD values decrease upwards from c. 0.12 to 0.14 before c. 8400 cal. yr BP and c. 0.08 to 0.10 g cm⁻³ between c. 8400 and 8200 cal. yr BP.

PZ 4a (c. 8200–6000 cal. yr BP; 408–290 cm)

Betula pubescens macrofossils establish a continuous presence from c. 8200 cal. yr BP and by c. 7900 cal. yr BP Betula pollen percentages reach c. 80% (Figure 3), with PAR values between c. 1500 and 2000 grains cm⁻² yr⁻¹ (Figures 4 and 6). As Betula pollen increase, Juniperus communis pollen become less abundant. There is also a relative decrease in herbaceous taxa such as Cyperaceae, Poaceae, Thalictrum alpinum and Galium (Figures 3 and 4). The presence of Betula pubescens is confirmed by the occurrences of its macrofossils throughout this period (Figures 5 and 6). Values of MS and DBD remain low and stable (apart from where tephra layers are present in the sediment) with DBD between c. 0.11 and 0.14 gcm⁻³. OM values are relatively high, between c. 34% and 55% (Figure 6).

PZ 4b (c. 6000–3000 cal. yr BP; 290–130 cm)

Gradually declining Betula PAR values fall below 1500 cm⁻² yr⁻¹ after c. 6000 cal. yr BP (Figures 4 and 6). This decline is not seen clearly in the pollen percentages, where there is only a slight drop in Betula at c. 6400 cal. yr BP when values become more variable than in PZ 4a (Figures 3 and 6). Betula pubescens macrofossils are found within this zone although in smaller numbers than in PZ 3a (Figure 5). Sorbus aucuparia pollen appear for the first time in the record around 6000 cal. yr BP (Figure 3).

Variations in OM, MS and DBD increase during this period. After the deposition of the Hekla 4 tephra layer at c. 4200 cal. yr BP (Dugmore et al., 1995), DBD is about 0.12–0.18 g cm⁻³ and OM is c. 30–38% after c. 4000 cal. yr BP (Figure 6). Betula PAR decreases to less than 1000 grains cm⁻² yr⁻¹ after 4200 cal. yr BP (Figures 4 and 6). Percentages of non-triporate pollen increase steadily from c. 7300 cal. yr BP, reaching the highest values in the record of 6–10% after c. 4200 cal. yr BP (Figure 6).

Early plant succession in a recently deglaciated environment

The Kagaðarhóll record extends back to c. 11,000 cal. yr BP. The pollen assemblages in the lowermost samples represent denuded, recently deglaciated landscape (Figure 3). Pollen from the pioneer taxon Oxyria digyna is prominent in the early assemblage. Pollen from plants that grow on gravel flats (Kristinsson, 2010) are found in the assemblages, for example, Ranunculus acris–type (which may originate from Ranunculus glacialis), as well as Dryas octopetala. Potentilla-type pollen that likely represent Sibbaldia procumbens, a plant typical of areas with heavy snow cover, are also prominent in the early assemblages. Arenaria-type and Sedum-type pollen likely represent taxa such as Arenaria norvegica, Sedum acre and Sedum annuum which have been found on moraines in front of the retreating glacier Skaftafellsjökull in Southeast Iceland (Persson, 1964). Low local pollen production from the sparse vegetation is evident from the low total PAR during the pioneer vegetation phase (Figure 6). The pollen record indicates rapid vegetation changes around Lake Kagaðarhóll during the earliest part of the Holocene. By c. 10,800 cal. yr BP, there is a short-term shift towards Empetrum nigrum–dominated vegetation (Figure 3) with a transition to dwarf-shrub heath at c. 10,600 cal. yr BP, as values for Salix and Cyperaceae pollen increase (Figures 3 and 4) and Calluna vulgaris pollen appear (Figure 3). Although it is difficult to infer vegetation cover from pollen data, the pollen assemblages compare favourably with present flora on moraines in front of Skaftafellsjökull, Southeast Iceland, where Empetrum nigrum and Calluna vulgaris dwarf shrubs with interspersed Salix lanata and Salix phylicifolia shrubs are prominent. Today, these species, along with mosses, account for c. 61% vegetation cover on 65-year-old moraines, and c. 67% on 120-year-old moraines (Vilmundardóttir et al., 2015). Therefore, vegetation around Lake Kagaðarhóll may have reached c. 60% cover by c. 10,300 cal. yr BP. Dwarf shrubs and herbaceous vegetation (predominantly Cyperaceae) around Lake Kagaðarhóll is similar to other vegetation reconstructions pre-dating the Saksunarvatn tephra at Tröllaskagi peninsula (Caseldine et al., 2006) and Skagi peninsula (Rundgren, 1998), North Iceland.

Specific plants may serve as temperature indicators during the earliest period of the Holocene. Indicator species found in the early-Holocene record at Lake Kagaðarhóll such as Selaginella selaginoides and Calluna vulgaris grow at a minimum July temperature of 7°C and Empetrum nigrum at 7.7°C (Kolstrup, 1979, 1980). Therefore, relatively warm summers in the area around the lake during the early Holocene may be proposed. This is in accordance with generally warm ocean conditions seen in various sediment data from the North Iceland Shelf where optimum thermal conditions were recorded from c. 10,000 cal. yr BP (e.g. Andrews and Giraudeau, 2003; Castañeda et al., 2004; Justwan et al., 2008; Knudsen et al., 2004; Rousse et al., 2006). Early-Holocene warmth is also reflected in chironomid-inferred mean July temperature (CI-T) at Tröllaskagi peninsula where July temperatures had likely reached 7°C before the deposition of the Saksunarvatn tephra (Caseldine et al., 2006). This may, however, be an underestimate of actual mean July temperatures at Tröllaskagi, as late arrival of thermophilous chironomid species during the early Holocene may skew the assemblages, and therefore the temperature reconstruction (Axford et al., 2007; Caseldine et al., 2003).

Impact of tephra deposition and aeolian processes on the terrestrial environment during the early Holocene

High fluxes of material transported by aeolian processes following deglaciation and tephra deposition from the Saksunarvatn event may have influenced the vegetation succession around Lake Kagaðarhóll as temperatures rose. The frequent recording of the Saksunarvatn tephra, often at great thickness (>10 cm) in Iceland (e.g. Caseldine et al., 2003, 2006; Larsen et al., 2012; Striberger et al., 2012; Stötter et al., 1999; Wastl et al., 2001), allows for the inference that it affected landscape stability and perhaps vegetation succession. Abrasion by wind erosion as well as burial by aeolian material may cause severe damage to vegetation (Gisladottir et al., 2005). The pollen assemblage 3 cm above the Saksunarvatn tephra indicates burial by material of thickness in excess of 10 cm. PARs and percentages of Empetrum nigrum and Cyperaceae decrease following the event, and these taxa have been shown to have an inverse relationship with sand thickness (Vilmundardóttir et al., 2009). Meanwhile the increase in Poaceae (Figure 4) may represent an increase in the species Festuca richardsonii which has a positive relationship with sand burial, with an increase in the taxon with sand thickness in excess of 10 cm (Vilmundardóttir et al., 2009). Higher above (11 cm) the Saksunarvatn tephra layer, the pollen assemblages regain similarities to those before its deposition, probably representing the re-establishment of a dwarf shrub dominated environment. The relatively fast recovery of vegetation to pre-deposition levels within about 100 years is in accordance with previous results from Skagi peninsula (Rundgren, 1998) and Tröllaskagi peninsula (Caseldine et al., 2006) where deposition of the Saksunarvatn tephra did not have a radical or permanent impact on vegetation development (Figure 1a). A gradual trend of stabilization over several centuries followed, with material left behind by retreating glaciers (Larsen et al., 2012; Striberger et al., 2012) and residual tephra being contained by expanding and developing vegetation.

The recording of spores from the coprophilous fungi Sporormiella-type and Sordaria-type (Figures 3 and 4) is of some interest. Coprophilous fungi are more or less reliant upon herbivores for spore germination (e.g. Cugny et al., 2010). Given the absence of mammal herbivores in pre-settlement Iceland, the presence of these spores may indicate the migration of birds into the post-glacial landscapes of Iceland. Sporormiella spp. and Sordaria spp. have been found on modern rock ptarmigan (Lagopus muta) faeces and that of other bird species in Iceland (Hallgrímsson and Eyjólfsdóttir, 2004).

Environmental instability, local increase in Juniperus communis and regional expansion of Betula pubescens

Change from open fell-field and tundra vegetation towards taller, more layered shrub vegetation occurs with a shift towards Juniperus communis–Betula nana–Salix shrub heath at c. 10,100 cal. yr BP. Between 9600 and 9200 cal. yr BP, there is a pronounced Juniperus communis pollen peak, with PARs up to 3070 grains cm⁻² yr⁻¹ (Figures 3, 4 and 6). The near absence of Juniperus communis macrofossils from the record is not unexpected and is probably because of the combination of lack of inflow into the lake and the relatively high weight of needles and cones preventing transport to the coring site, which is at a distance of >100 m from the shore (cf. Birks and Bjune, 2010; Dieffenbacher-Krall, 2007).

A Juniperus communis phase similar to the one at Lake Kagaðarhóll can be seen at other sites in North (Caseldine et al., 2006; Hallsdóttir, 1995; Rundgren, 1998) and Northwest (Caseldine et al., 2003) Iceland (Figure 1a). Although Juniperus communis produces large quantities of pollen, they are poorly represented in sediments and even low percentages of pollen are commonly inferred to represent a strong presence of the plant in the vicinity of sampling locations (Huntley and Birks, 1983; Schofield et al., 2007). High percentages and PARs such as those recorded after c. 10,100 cal. yr BP at Lake Kagaðarhóll can only be produced by an upright growth form of dense Juniperus communis scrub (Birks, 1973). Taller growth form of Juniperus communis indicates relatively high mean July temperature, perhaps above 10°C (Kolstrup, 1980). The presence of Myriophyllum alterniflorum in the pollen record (Figures 3 and 4) also indicates relatively high mean July temperatures, as high as c. 10°C (Kolstrup, 1980). The peak in Juniperus communis coincides with a shift from Potamogeton natans to Potamogeton perfoliatus (Figure 5) which requires relatively warm lake conditions during summers (Kristinsson, 2010). The high total PAR values, because of the substantial deposition of Betula and Juniperus communis pollen, may indicate an optimum period for pollen production for both taxa (Figure 6). This may be an indication of warm Holocene conditions in Iceland already by c. 10,100 cal. yr BP. Warm summers during the early Holocene at Lake Kagaðarhóll are in accordance with recorded warm conditions during the early Holocene in the ocean north of Iceland (e.g. Castañeda et al., 2004; Justwan et al., 2008; Knudsen et al., 2004; Rousse et al., 2006), and oxygen isotope values from the Renland ice-core record show a period of low δ¹⁸O values at this time (Vinther et al., 2009; Figure 6). This trend follows maximum solar insolation during the early Holocene (Berger and Loutre, 1991; Figure 6). Glaciers in the Icelandic highlands were probably smaller than today during the early Holocene. Larsen et al. (2012) have argued that Langjökull in the western highland was smaller than today when the Saksunarvatn tephra was deposited and Eyjabakkajökull in the eastern highlands had retreated behind its current position by 9300 cal. yr BP (Striberger et al., 2012). This is in agreement with chironomid-inferred mean July temperatures from Eyjafjörður, which also indicate that early-Holocene temperatures may have been warmer than today (Langdon et al., 2010).

The environment was still relatively unstable, indicated by the high minerogenic input to the lake seen in the OM, MS and DBD data (Figure 6). Juniperus communis is highly competitive in disturbed environs and on nutrient poor soils (García et al., 2000), but the species’ intolerance to shade (Thomas et al., 2007a) would have put it at a disadvantage as taller birch trees formed an increasingly closed canopy. The apparent delayed arrival of Betula pubescens to Lake Kagaðarhóll despite rising temperatures is likely because of limiting factors such as soil properties, or distance from seed source. Well-drained sandy soils may have given drought-tolerant Juniperus communis (Thomas et al., 2007a) an advantage over drought-intolerant Betula pubescens (Atkinson, 1992). The succession from Juniperus communis to Betula pubescens is similar to that seen in Europe during the Late Glacial, where the spread of Betula pubescens may have been hindered by factors other than temperature, such as soil moisture (Birks and Birks, 2014; Mortensen et al., 2014).

In the absence of Icelandic PAR references, the Betula PAR values may be compared with published modern Betula PAR values from Scandinavia. Betula PARs of >500 grains cm⁻² yr⁻¹ from c. 10,000 cal. yr BP at Lake Kagaðarhóll might be interpreted as sparse presence of Betula pubescens around the lake, with Betula PAR of >1000 grains cm⁻² yr⁻¹ indicating open forest (Hicks, 2001) established between c. 9550 and 9400 cal. yr BP (Figures 4 and 6). However, only one Betula pubescens fruit was found at c. 9550 cal. yr BP (Figure 6), while only Betula nana catkin scales were found (Figure 5). Further Betula fruits were found between c. 9600 and 9300 cal. yr BP, but these were too degraded to be identified to species level. As birch fruits are prominent in modern sediments from sites where birch trees are located (Jackson and Booth, 2007), the lack of macrofossils concurrent with the high Betula PARs may represent a regional increase in Betula pubescens around 10,000 cal. yr BP and a sporadic occurrence close to the lake until c. 9300 cal. yr BP. High occurrence of non-triporate pollen in the sediment provides further evidence for the presence of tree birch in the region. Non-triporate Betula pollen are recorded in modern individuals of both Betula nana and Betula pubescens in Iceland. On average, 2.4% of pollen produced by individuals of Betula nana pollen are non-triporate and 0.7% in individuals of Betula pubescens. A much higher number, 12.2%, of non-triporate pollen grains occur in hybrid individuals of the two species (Karlsdóttir et al., 2008). High proportion of non-triporate pollen therefore indicates the presence of hybrids and, in turn, the presence of both parent species in the region. The proportion of non-triporate pollen during this period ranges between 2% and 9% (Figure 6), which indicates hybridization already taking place during the early Holocene, similar to what has been observed in other early-Holocene pollen records from Iceland (Karlsdóttir, 2014; Karlsdóttir et al., 2009, 2012).

Establishment of birch woodland around Lake Kagaðarhóll

The first unambiguous evidence for the existence of Betula pubescens around the lake is seen at c. 9300 cal. yr BP, when macrofossils from the species are present continuously over several centuries (Figures 5 and 6). Further development towards Betula pubescens woodland is depicted in the pollen record from c. 9200 cal. yr BP, when Betula pollen percentages reach above 30%. Between 9100 and 8800 cal. yr BP, Betula pollen account for c. 45% of land pollen with PARs of 640 up to 1100 grains cm⁻² yr⁻¹ (Figure 3). A simultaneous decrease in Juniperus communis may reflect increased competition with Betula pubescens as Juniperus communis is intolerant of shade (Thomas et al., 2007a). The change towards a closed canopy and woodland ecosystem is also seen from the large decrease in pollen deposition from dwarf shrubs, graminoids and other herbs (Figure 4). Stabilization of the environment coincident with woodland development is reflected in lower and more stable MS values and a rise in OM deposition within the lake (Figure 6). The increase in Betula pubescens in the record between c. 9300 and 8700 cal. yr BP is indicative of relatively high summer temperatures. This period coincides with a peak in mean July CI-T at Tröllaskagi peninsula between c. 9500 and 9000 cal. yr BP (Caseldine et al., 2006). Within this period, Betula PAR reaches >1000 grains cm⁻² yr⁻¹ at Kagaðarhóll (Figures 4 and 6), which may indicate open forest according to the Scandinavian model (Hicks, 2001). This initial phase of birch woodland encroachment around the lake lasted for about 500 years.

Terrestrial ecosystem response to cooling at c. 8700 cal. yr BP

At about 8700 cal. yr BP, there is a rapid drop in Betula PAR to values of <270 grains cm⁻² yr⁻¹ (Figures 4 and 6). The low PARs indicate a decline in birch pollen deposition to values below the limit of birch presence at (c. 250 grains cm⁻² yr⁻¹; cf. Seppä and Hicks, 2006). The proportional increase in Poaceae may indicate expansion of grassland at the same time. The decrease in total PAR (Figure 6) and scarcity of macrofossils deposited (Figure 5) indicate a drop in both pollen and seed production, not only by Betula pubescens but also other plants growing around Lake Kagaðarhóll between c. 8700 and 8200 cal. yr BP. This likely represents lower spring and summer temperatures, with tritherm temperatures below 7.2°C (cf. Wöll, 2008). This is in accordance with a period of gradual cooling of mean July CI-T temperatures in Tröllaskagi peninsula between c. 9000 and 8100 cal. yr BP (Caseldine et al., 2006). Cooler summer conditions are seen in data from both Haukadalsvatn in West Iceland and Hvítárvatn in the highlands (Figure 1a) between c. 8650 and 7850 cal. yr BP (Geirsdóttir et al., 2013; Larsen et al., 2012). In Lögurinn, East Iceland (Figure 1a) a cooling takes place later, between c. 8200 and 8000 cal. yr BP (Striberger et al., 2012). A minor cooling is observed on the North Icelandic Shelf at c. 8200 cal. yr BP (Castañeda et al., 2004; Ran et al., 2006), with a small change in sea surface temperature (SST) (Andersen et al., 2004) but a more pronounced cooling is seen in cores further offshore and to the east (Eiríksson et al., 2000; Knudsen et al., 2004). The cooling seen in several terrestrial proxy records from Iceland demonstrates that the extensive impact on the terrestrial environment in Iceland occurred earlier than the short cold incursion c. 8200 cal. yr BP seen in marine data from the North Atlantic (Alley and Ágústsdóttir, 2005; Alley et al., 1997) and several Greenland ice-core records (Thomas et al., 2007b). A possible explanation for the change in vegetation around Lake Kagaðarhóll at this time is the presence of sea-ice. Air temperature in Iceland is correlated with the presence of sea-ice (Bergþórsson, 1969; Ogilvie, 1984, 1992), and an increased occurrence of sea-ice during spring and summer would have been accompanied by lower air temperatures. More frequent occurrences of sea-ice during spring and summer are likely because of episodes of weaker Irminger Current and a stronger sea-ice bearing East Greenland Current on the North Iceland Shelf between c. 9000 and 8000 cal. yr BP (Ran et al., 2006). Although most Greenland ice-core records only show a short cold period (Thomas et al., 2007b), the Renland ice-core isotope data show a trend towards higher δ¹⁸O values already at around 8700 cal. yr BP (Vinther et al., 2009; Figure 6). Colder spring and summer temperatures after 8700 cal. yr BP may have drastically hampered pollen and seed production of Betula pubescens and other plants. Cold temperatures during spring time have been linked to loss of male catkins in Betula pubescens, which in turn leads to low pollen production (Pichugina, 1972). This may be the reason for the low Betula pollen accumulation and the absence of fruits and catkin scales from the Lake Kagaðarhóll sediments. Betula pubescens trees usually have a short life span of less than 100 years (Gimingham, 1984). However, following damage, the trees can still reproduce by sprouting from basal buds (Kauppi et al., 1987), and mountain birch (Betula pubescens Erhr. ssp. tortuosa) regenerates mainly from stems (Verwijst, 1988). Therefore, the near absence of Betula pubescens from the macrofossil record and the very low Betula PAR over several centuries may indicate either a retreat of birch from the area around the lake or a shift towards a period of vegetative reproduction by trees and other plants while conditions were too harsh for sexual reproduction. The Kagaðarhóll record shares similarities with the record from Lake Vatnskotsvatn in Skagafjörður (Figure 1a). There, an initial expansion of Betula pubescens was halted at c. 8300 cal. yr BP (7500 ¹⁴C yr BP) with a regression towards a Juniperus–Salix–Betula nana heath. However, birch woodland re-established itself by c. 8000 cal. yr BP (7200 ¹⁴C yr BP; Hallsdóttir, 1995). A dip in total PAR similar to the one in the Lake Kagaðarhóll record is seen during the Betula retreat at Vatnskotsvatn. A similar regression in vegetation development is not seen in records from Eyjafjörður (Figure 1a) to the east of Skagafjörður, where birch persisted uninterrupted (Caseldine et al., 2006). Nor is it seen in data from Efstadalsvatn (Figure 1a) in the Westfjords (Caseldine et al., 2003), where dwarf-shrub/heath vegetation would have been less sensitive to cooling climate than birch woodlands of Austur-Húnavatnssýsla and Skagafjörður. Cooler conditions may have affected vegetation in Northwest Iceland worse than central North Iceland during this period, when sea-ice may have often lingered in Húnaflói and Skagafjörður during spring and early summer.

There are no indications of increased erosion despite the clear impact on birch woodland during this period. In fact, land surface seems to have become more stable, as inferred from the decrease in MS and DBD. This demonstrates the importance of a continuous vegetation cover and the ecological structure of the established woodland to inhibit erosion in the face of inhospitable conditions during this period.

Maximum extent of Holocene birch woodlands

The transition back to birch woodland occurs within a span of 300 years, with Betula pubescens macrofossils reappearing by c. 8200 cal. yr BP (Figures 5 and 6) followed by a substantial increase in Betula pollen from c. 7900 cal. yr BP (Figures 3 and 4). Betula PAR values representative of forested areas, between 1500 and 2000 grains cm⁻² yr⁻¹, are recorded after c. 7900 cal. yr BP (Figures 4 and 6). These Betula PAR values indicate a dense forest surrounding the lake according to the classification offered by Hicks (2001). The subsequent decline in Juniperus communis is likely because of the establishment of a closed canopy with heavy shade (cf. Thomas et al., 2007a). A similar decrease in herbs, such as graminoids, Galium and Thalictrum alpinum, is a further indication of a closing canopy (Figures 3 and 4). This dense birch woodland surrounded Lake Kagaðarhóll from c. 7900 to 6000 cal. yr BP. At this time, mean July temperatures at Tröllaskagi may have been higher than at present, with warm conditions representing the HTM considered to last from c. 8000 to 6700 cal. yr BP (Caseldine et al., 2006). The expansion of birch woodland during the early to mid-Holocene reflects the strong relationship between Betula pubescens growth and high summer temperatures. In Tröllaskagi, Wastl et al. (2001) found that Betula pubescens was present at elevations between 450 and 500 m a.s.l. between c. 7600 and 6800 cal. yr BP. At Vatnskotsvatn, the Betula peak occurred between c. 8000 and 6800 cal. yr BP (7200 to 6000 ¹⁴C yr BP; Hallsdóttir, 1995). In contrast, only a short peak in Betula pubescens pollen is seen in pollen data from Þistilfjörður, Northeast Iceland (Figure 1a) at c. 7200 cal. yr BP, where maximum Holocene birch woodland extent occurred later, at c. 5000 cal. yr BP (Karlsdóttir, 2014). A warm period defined as HTM is observed between c. 7850 and 5450 cal. yr BP in Hvítárvatn in the highlands (Larsen et al., 2012) and Haukadalsvatn, West Iceland (Geirsdóttir et al., 2013) and at c. 7900 to 7000 cal. yr BP in Lake Lögurinn, East Iceland (Striberger et al., 2012).

Onset of Neoglaciation and harsher environmental conditions

A decline in woodland is seen in the Betula PAR data, with values gradually falling to below 1500 grains cm⁻² yr⁻¹ from c. 6000 cal. yr BP, indicating open woodland. After c. 4200 cal. yr BP, the PAR values fall to below 1000 grains cm⁻² yr⁻¹, indicating a sparse presence of Betula pubescens (Figures 4 and 6) (Hicks, 2001). Sorbus aucuparia pollen appear after c. 6000 cal. yr BP (Figure 3) as the woodland begins to open up. Because of the entomophilous pollination of Sorbus aucuparia, it cannot become prominent in a pollen record dominated by the anemophilous Betula pubescens. Therefore, the absence of Sorbus aucuparia pollen is not necessarily proof of the plants’ absence from the Icelandic HTM woodlands (Hallsdóttir, 1995). Although the transition into Neoglaciation is seen from c. 6000 cal. yr BP at Lake Kagaðarhóll, birch is still present towards the end of the Lake Kagaðarhóll record at c. 2800 cal. yr BP, indicating that summer temperatures were still relatively high.

Increasing environmental instability is seen in the record after the deposition of the Hekla 4 tephra layer c. 4200 cal. yr BP (Dugmore et al., 1995). The Hekla 4 tephra is one of the most prominent Holocene tephra layers found in Icelandic sediments, with about 6.7 km³ of material deposited on land during the eruption (Larsen and Thorarinsson, 1977). An increase in deposition of minerogenic matter after c. 4200 cal. yr BP is seen from an increase in DBD (Figure 6). Impact of great tephra deposition and subsequent aeolian processes on vegetation are, however, not reflected in the pollen record. There is no direct indication in the pollen record of plant burial around Lake Kagaðarhóll, such as followed the deposition of the Saksunarvatn tephra. However, the increased occurrence of non-triporate Betula pollen after 4200 cal. yr BP may be because of increased hybridization in response to deteriorating environmental conditions, allowing Betula nana to invade previously forested areas. The inverse relationship between non-triporate Betula pollen percentages and Betula PAR values indicates that hybridization may increase as PAR declines because of cooling climate and harsher environmental conditions. Previous studies indicate that hybridization may occur in response to woodland expansions because of warming climate coinciding with increasing Betula PARs (Karlsdóttir, 2014; Karlsdóttir et al., 2012).

The timing of the transition to Neoglaciation at Lake Kagaðarhóll is in agreement with other terrestrial records which place the onset of cooling and more unstable conditions from c. 5500 to 4200 cal. yr BP (Geirsdóttir et al., 2013; Larsen et al., 2012; Striberger et al., 2012). On the North Iceland Shelf, optimum warmth lasted until 6000–5000 cal. yr BP, when conditions became colder and more variable (e.g. Andresen et al., 2005; Castañeda et al., 2004; Justwan et al., 2008; Knudsen et al., 2004; Ran et al., 2006; Rousse et al., 2006). Increased landscape destabilization after 4200 cal. yr BP at Lake Kagaðarhóll is comparable with other terrestrial proxy records for erosion (Geirsdóttir et al., 2013; Larsen et al., 2012; Striberger et al., 2012). The effects of deteriorating climate at the onset of Neoglaciation are undoubtedly at work here, although the consequences of the great tephra fall at c. 4200 cal. yr BP could also be of importance given that the tephra was deposited over ecosystems already under pressure from deteriorating climate. The environmental consequences of, and responses to, the large pre-historic Hekla eruptions (Larsen and Thorarinsson, 1977) are not well known (but see Caseldine and Hatton, 1994).