Solar influence on climate variability and human development during the Neolithic: evidence from a high-resolution multi-proxy record from Templevanny Lough, County Sligo, Ireland

Quaternary Science Reviews 67 (2013) 138e159

Contents lists available at SciVerse ScienceDirect

Quaternary Science Reviews journal homepage: www.elsevier.com/locate/quascirev

Solar influence on climate variability and human development during the Neolithic: evidence from a high-resolution multi-proxy record from Templevanny Lough, County Sligo, Ireland Susann Stolze a, *, Raimund Muscheler b, Walter Dörfler c, Oliver Nelle a a b c

Institute for Ecosystem Research, Christian-Albrechts University, D-24118 Kiel, Germany Department of Geology, Lund University, SE-22362 Lund, Sweden Institute of Prehistory and Ancient History, Christian-Albrechts University, D-24118 Kiel, Germany

a r t i c l e i n f o

a b s t r a c t

Article history: Received 30 March 2012 Received in revised form 17 December 2012 Accepted 16 January 2013 Available online

The relationship between climatic variations, vegetation dynamics and early human activity between c. 4150e2860 BC was reconstructed from a high-resolution pollen and geochemical record obtained from a small lake located in County Sligo, Ireland. The proxy record suggests the existence of a woodland with a largely closed canopy at the start of the fourth millennium BC. Only minor human disturbance is recorded. Following an episode of increased rainfall at c. 3990 BC, a decrease in the elm population occurred between c. 3970 and 3820 BC. This coincided with a period of warming and drying climatic conditions and an initial increase in anthropogenic activities. A second episode of high precipitation between c. 3830e3800 BC was followed by a steep increase in human impact on the landscape, which became most pronounced between c. 3740 and 3630 BC. At this time, the lake level of Templevanny Lough was at its lowest during the Neolithic. The onset of wetter and cooler conditions after c. 3670 BC, representing the transition from the Early to the Middle Neolithic, coincided with a period of woodland recovery. The Middle Neolithic was characterised by pronounced climatic oscillations including periods of substantial rainfall between c. 3600 and 3500 BC and between c. 3500 and 3460 BC. A nearly century-long climatic amelioration between c. 3460e3370 BC facilitated a revival of human activity on a small scale around the lake. Abandonment of the area and full woodland recovery occurred after a period of particularly wet and cool conditions ranging from c. 3360e3290 BC. The pollen and geochemistry data suggest that the Late Neolithic was marked by a period of ameliorated conditions between c. 3110e3050 BC that was followed by two episodes of high rainfall at c. 3060e3030 BC and c. 2940e2900 BC. The timing of the climatic shifts inferred from the Templevanny Lough record is in agreement with those of moisture/precipitation and temperature reconstructions from northern and western Europe and the Alps, suggesting that the studied period was characterised by a high-frequency climate variability. The results of the present study imply that human development during the Irish Neolithic was influenced by climatic variations. These climatic shifts correspond to variations in solar activity, suggesting a solar forcing on climate. Ó 2013 Elsevier Ltd. All rights reserved.

Keywords: Neolithic Ireland Human impact Climatic variability Solar activity

1. Introduction Over the past decades, the Neolithic in Ireland has been the focus of a number of palaeoenvironmental and archaeological studies as this time period marks a rapid increase in human impact on the landscape (cf. O’Connell and Molloy, 2001; Cooney, 2007).

* Corresponding author. Present address: Institute of Arctic and Alpine Research (INSTAAR), University of Colorado, 1560 30th Street, Boulder, CO 80309, USA. E-mail address: [email protected] (S. Stolze). 0277-3791/$ e see front matter Ó 2013 Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.quascirev.2013.01.013

The introduction of Neolithic practices including the construction of houses and other occupation sites, the erection of megalithic structures and the adoption of arable farming during the first quarter of the fourth millennium BC (Cooney, 2000; Cooney et al., 2011 and references therein) resulted in the widespread clearance of the primeval forest (O’Connell and Molloy, 2001). Recent investigations suggest that climatic fluctuations had a major control on the intensity of human impact and societal developments during prehistoric and historic times (Magny, 2004; Mayewski et al., 2004; Turney et al., 2006; Büntgen et al., 2011). Palynological records from Ireland suggest that the notable

S. Stolze et al. / Quaternary Science Reviews 67 (2013) 138e159

increase in human activity during the Early Neolithic (4000e3600 BC; cf. Sheridan, 1995) broadly coincided with a period of climatic amelioration (Tipping, 2010; Stolze et al., 2012). Human impact was less intense during the Middle Neolithic (3600e3100 BC; cf. Sheridan, 1995) and declined noticeably during the Late Neolithic (3100e2500 BC; cf. Sheridan, 1995), facilitating woodland recovery in many areas (O’Connell and Molloy, 2001). This change was likely associated with a period of climatic deterioration (Caseldine et al., 2005; Verrill and Tipping, 2010; Stolze et al., 2012). Based on the high-resolution multi-proxy record from Templevanny Lough in County Sligo, the present study confirms that climatic variability influenced human development in the region during the Neolithic. To test whether the observed variations correspond to large-scale climate patterns, the Templevanny Lough record is compared to high-resolution temperature and precipitation/moisture records from northern and Western Europe and the Alps. Following previous studies that established a link between climatic changes and fluctuations in the solar output during the Holocene (Bond et al., 2001; Mauquoy et al., 2002, 2008; Speranza et al., 2002; Blaauw et al., 2004; Magny, 2004), the established large-scale climate patterns are correlated to time-dependent variations in proxies of solar activity. Based on this comparison, it is argued that the climate during the Irish Neolithic was influenced by multi-decadal to centennial scale fluctuations in solar activity.

139

2. Study area The study site is located in the southeastern Ballymote Lowlands in County Sligo (Fig. 1). This low-lying land is covered by glacial sand and gravel deposited on Carboniferous limestone (MacDermot et al., 1996). The Bricklieve Mountains, hills of cherty limestone, and the nearby hill of Keshcorran are prominent upland areas that rise to over 300 m above sea level (asl). To the south, the Ballymote Lowlands are bordered by the Curlew Mountains, a ridge of Devonian sandstone that reaches over 250 m asl (MacDermot et al., 1996). A series of some twenty passage tombs, referred to as the CarrowkeeleKeshcorran megalithic complex, is located in the largely blanket bog covered Bricklieve Mountains and on Keshcorran (McGloin and Moore, 1996). A cluster of 153 hut sites was recognised on the limestone plateau of Mullaghfarna in the northeastern Bricklieve Mountains (Fig. 1) dating from between 3200 and 1100 BC (S. Bergh, pers. comm.). At present, mild and moist climatic conditions with an average annual precipitation of 1150 mm and a mean annual temperature of 9
C prevail. The predominant wind direction is west to southwest (online data of Met Éireann, 2012). The study site, Templevanny Lough (54
02.100 N, 08
24.350 W) represents a glacially formed lake of oval shape (Fig. 2). Its small size of 2 ha allows the reconstruction of changes in the vegetation

Fig. 1. Regional map of southwestern County Sligo, Ireland, showing the location of Templevanny Lough with respect to the locations of the passage tombs of the Carrowkeele Keshcorran megalithic complex and hut sites at Mullaghfarna. Also shown are the previously studied sites of Lough Availe and Loughmeenaghan (Stolze, 2012a,b; Stolze et al., 2012).

140

S. Stolze et al. / Quaternary Science Reviews 67 (2013) 138e159

Fig. 2. Map of Templevanny Lough showing the lake bathymetry and the coring location. Vegetation units surrounding the study site are given.

in a radius of c. 1000e3000 m from the coring location (cf. Davis, 2000; Hellman et al., 2009). The lake is located 1500 m west of the Bricklieve Mountains and 2300 m south of Keshcorran, at an elevation of 84 m asl. It has a number of inflows and a stream that drains to the southwest. An artificial island (crannóg) that may have been used as a lake dwelling is located in the western part of the lake. The lake is surrounded by intensely used pastoral land enclosed by hedgerows (Fig. 2). A boggy area covered with ericaceous shrubs (Calluna vulgaris and Erica tetralix), mosses, grasses and sedges stretches along the western part of the lake. The shallow water of the lake is inhabited by Hippuris vulgaris, Sparganium, Equisetum cf. fluviatile and Nuphar lutea. Characteristic taxa growing on the lake shore include Phalaris arundinacea, Phragmites australis, Glyceria fluitans, Poa palustris, Potentilla palustris, Caltha palustris, Lychnis flos-cuculi, Iris pseudacorus, Valeriana officinalis, Ranunculus, Fabaceae and Cyperaceae. Plant identification and nomenclature followed Webb et al. (1996). 3. Methods 3.1. Field investigations The morphology of Templevanny Lough was determined through a bathymetric survey, employing a single-beam acoustic depth sounding system along with a GPS unit for position control.

Coring was conducted in the centre of the bowl-shaped lake that has a maximum depth of 6 m. Duplicate sediment cores were retrieved using a rod-operated Usinger piston corer (Mingram et al., 2007), yielding a continuous sediment sequence of 13 m in length. Coring terminated in basal clay. The fresh sediment cores were cut in half longitudinally and photographed in the field. Following transfer to the laboratory, the cores were stored in the dark at 4
C. 3.2. Pollen and non-pollen palynomorph analysis Prior to sampling, the core interval corresponding to the Neolithic period was identified by quick examination of the pollen content of the fresh sediments following the procedure described by Stolze et al. (2012). The sediments in this interval (1000e 870 cm) consist of brown organic fine detritus gyttja. Samples were taken at 1-cm increments and processed following standard methods (Faegri and Iversen, 1989). This included treatment with HCl and KOH, sieving using a 125 mm screen, processing with HF and acetolysis. Tablets of Lycopodium clavatum spores were added prior to chemical treatment to permit the calculation of pollen concentrations (Stockmarr, 1971). The pollen residues were mounted in glycerine. Routine counting was carried out at 400-fold magnification. Critical identifications were made under oil immersion at 1000-fold magnification, applying both bright field and phase contrast illumination.

S. Stolze et al. / Quaternary Science Reviews 67 (2013) 138e159

Identification of the observed pollen types followed Moore et al. (1991) and Beug (2004). Non pollen-palynomorphs were determined after Pals et al. (1980), van Geel et al. (1989) and Komárek and Jankovská (2001). A total terrestrial pollen and spore sum of c. 1000 grains was counted. Data of all palynomorphs are presented as relative abundance of the total terrestrial pollen and spore sum. Pollen diagrams were constructed with the program TGView 2.0.2 (Grimm, 2004). Pollen assemblage zones were established by stratigraphically constrained cluster analysis using the program PAST (Hammer et al., 2001), including all pollen types with frequencies above 2%. 3.3. Geochemical analyses Loss on ignition (LOI) analysis was performed on sediment samples taken from the same depths as the pollen samples. The samples were dried at 105
C for 12 h. Determination of the organic and inorganic carbon contents was then carried out after combustion of the samples in a muffle furnace at 550
C for 4 h, and at 950
C for 2 h (Heiri et al., 2001). The chemical composition of the sediments was determined by Xray fluorescence (XRF) analysis using an Avaatech XRF core scanner. The resolution of the point sensor was set to 0.5 cm. Measurement conditions of 0.25 mA and 10 kV with a count time of 30 s were applied. Signal intensities were determined for selected elements, including Al, Si, K, Ca, Ti, Mn and Fe. Analyses of georeference materials showed that the intensity measurements of these elements could be repeated at a precision of <5% relative standard deviation. Stratigraphically constrained cluster analysis was performed on the geochemical data using the program PAST (Hammer et al., 2001). 3.4. Radiocarbon dating Sixteen samples were collected for radiocarbon dating to constrain the absolute timing of significant changes in the pollen spectra. Slices of 2-cm thickness were taken from the sediment core, soaked in 10% HCl to remove the carbonate fraction and wetsieved through a 500 mm screen. Remains of terrestrial plants were picked from the residues under a binocular microscope. As no terrestrial plant remains were found at 965.5e963.5 cm, a pollen concentrate sample was prepared. Chemical treatment

141

followed Brown et al. (1992) and Kilian et al. (2002). Particles outside the size range of most pollen grains were removed through sieving using 10 and 63 mm screens. Further pollen enrichment involved density separation with lithium metatungstate solutions. The highest pollen yield was obtained within the specific gravity range of 1.15e1.2 g cm3. Following physical removal of contaminants and acid-alkaliacid pre-treatment of the samples, radiocarbon dating was conducted by accelerator mass spectrometry (AMS) using the protocol of Nadeau et al. (1997). Calculation of the conventional 14C ages followed Stuiver and Polach (1977) with a d13C correction for isotopic fractionation. The measured radiocarbon ages were calibrated using the program OxCal 4.1 (Bronk Ramsey, 1995, 2001, 2009a) applying the INTCAL09 curve (Reimer et al., 2009). Bayesian age-depth modelling was performed using a Poisson depositional model (Bronk Ramsey, 2008) at 0.5-cm increments. The calculation was based on a number of events per unit length of 10 cm1. To acknowledge changes in the deposition rate, boundaries were set at 975.5 cm and 897.5 cm. Outlier analysis of the radiocarbon dates followed Bronk Ramsey (2009b). Ages obtained by Bayesian analysis are expressed as the mean of the posterior probability distributions and rounded to the nearest 10 years. The 95.4% highest probability density (HPD) intervals are used as a measure of uncertainty of this age estimate. 4. Results 4.1. Radiocarbon dating Radiocarbon dating of eleven samples produced reliable ages, while four samples were identified as outliers (Table 1). The radiocarbon ages of samples KIA40925 and KIA40928 are regarded to be too old as the samples were taken close to inwash layers. The radiocarbon age of the pollen concentrate KIA43111 is considered to be too old, possibly due to the presence of aquatic plant material such as algal remains that were not completely removed during sample preparation, resulting in a hard-water effect. KIA40922 yielded a radiocarbon date that is regarded as too young, possibly suggesting contamination with younger carbon. Due to the low carbon yield, sample KIA40931 produced no useable age.

Table 1 AMS dates from Templevanny Lough, County Sligo. Lab code

Depth cm

Material dated

Sample size mg C

Radiocarbon age BP  1s

d13C &

Calibrated age BC (2s)

KIA40931 KIA43365

880.5e878.5 891.5e889.5

0.1 0.4

No usable age 4497  65

27.0a 36.53  0.13b

N/A 3368e2936

KIA40930

897.5e895.5

0.6

4316  55

27.0a

3097e2764

KIA40929

910.5e908.5

Rubus fruticosus seed Alnus seed, deciduous bud scale, leaf fragments (6), leaf bud Betula seeds (2), Carex nutlets (2), leaf fragment, deciduous bud scale Betula seed, Carex nutlets, deciduous bud scale, wood fragment Leaf fragment, wood fragment Alnus seed (1), leaf fragments (20), bark fragment (1) Deciduous bud scale, leaf fragment Leaf fragments (10) Leaf fragment, Carex nutlet, deciduous bud scales (2) Pollen concentrate Wood fragment, leaf fragment, deciduous bud scale Deciduous bud scale, leaf fragment, stalk Wood fragments (2), leaf fragment Carex nutlets (2), deciduous bud scales (2), leaf fragment Carex nutlets (3), deciduous bud scale Deciduous bud scale, leaf fragments (11)

0.5

4545  58

25.74  0.29

3498e3028

KIA40928 KIA43364

924.5e922.5 927.5e925.5

KIA40932 KIA43363 KIA40927 KIA43111 KIA40926 KIA40925 KIA40924 KIA40923 KIA40922 KIA43362

942.5e941.5 950.5e948.5 957.5e955.5 965.5e963.5 974.5e972.5 980.5e978.5 985.5e983.5 989.5e987.5 996.5e994.5 999.5e997.5

a b

0.6 2.1

4912  55 4510  32

0.4 0.3 1.1 0.5 0.6 0.5 1.1 0.6 1.0 1.0

4533 4851 4721 5335 5057 5436 5000 5220 4967 5273

         

70 79 30 56 48 58 31 46 40 41

a

27.0 28.66  0.18

3907e3538 3354e3097

23.77 35.48 24.52 30.07 27.0a 28.58 26.54 27.0a 26.94 28.58

3500e2944 3894e3377 3633e3376 4328e4004 3963e3714 4336e4262 3939e3702 4229e3957 3926e3652 4233e3987

   

0.20 0.20b 0.18 0.44

 0.34  0.24  0.27  0.46

Because no stable d13C value was obtained during measurement, a canonical value of 27& was used for the calculation of the conventional Due to the low carbon yield of the sample, the d13C value may be anomalously low (M.-J. Nadeau, pers. comm.).

14

C age.

142

S. Stolze et al. / Quaternary Science Reviews 67 (2013) 138e159

Bayesian age-depth modelling yielded probability densities that span comparably narrower age ranges than the calibrated radiocarbon dates (Fig. 3). The 95.4% HPD intervals span from 25 to 300 years, with a median value of 45 years (Fig. 3). The analysed sequence covers a period from c. 4150e2860 BC. 4.2. Pollen analysis 4.2.1. The terrestrial pollen record The terrestrial pollen diagram is divided into four pollen zones and eight subzones (Figs. 4 and 5). TVL-P1 (1000e984.5 cm; c. 4150e3920 BC): The pollen record opens with high frequencies of total arboreal pollen (92e95%). Corylus (30e42%), Ulmus (20%), Alnus (20%) and Quercus (13%) are the dominant arboreal pollen (AP) types. Clumps of immature pollen grains of these types are present. Ilex type, Viburnum opulus type and Hedera helix pollen are common. Pollen and spores of disturbance indicators such as Pteridium aquilinum and Plantago lanceolata type are observed. The first reduction in percentage and concentration data of Ulmus pollen occurs at 988 cm (c. 3970 BC), which is not accompanied by changes in other pollen types. The total pollen concentration shows a low at 988 cm. TVL-P2 (984.5e944.5 cm; c. 3920e3500 BC): Decreased AP data and an abundance of anthropogenic pollen indicators characterise this zone. Four subzones are distinguished. TVL-P2a (984.5e976.5 cm; c. 3920e3810 BC): The decrease in Ulmus pollen continues, leading to very low percentage (1.4%) and concentration values in the uppermost sample of this subzone. The absolute and relative values of Corylus and Quercus pollen decline towards the upper boundary. Pollen grains of light-tolerant trees and shrubs including Sorbus group and Ulex type are more frequent. The AP decline is accompanied by an increase in pollen types typical of grassland taxa including Poaceae: wild grass group, Trifolium type, Oxyria type and Ranunculus type pollen. Initiation of the Plantago lanceolata type pollen curve is recorded. A general decline in the total pollen concentration is observed with a low at 979 cm (c. 3840 BC). TVL-P2b (976.5e969.5 cm; c. 3810e3740 BC): This subzone is marked by declined AP abundances (75%) and concentrations. High values of Plantago lanceolata type pollen open this subzone. This pollen type achieves its highest percentages (8%) and concentrations within the studied core interval. Single occurrences of Polygonum aviculare type pollen and Thelypteris palustris spores are recorded. TVL-P2c (969.5e958.5 cm; c. 3740e3630 BC): Markedly low AP representations of 67e73% and a scarcity of clumped AP grains mark this subzone. Alnus pollen shows a notable drop in its absolute and relative values between 967 and 965 cm (c. 3710e3690 BC). The non-arboreal pollen (NAP) record is dominated by Poaceae: wild grass group pollen (14e19%), while Plantago lanceolata type pollen shows a decline in absolute and relative values. The shift in the NAP record is accompanied by increased representations of Rumex acetosella type, Ranunculus type and Lactuceae pollen. A drop in the total pollen concentration is recorded above 961 cm (c. 3650 BC). TVL-P2d (958.5e944.5 cm; c. 3630e3500 BC): The AP frequencies recover to 88%, with Alnus (30%) and Corylus (40%) pollen being the main contributors. An increase in the relative abundance of Ulmus (8%), Sorbus group and Fraxinus pollen is recorded. Clumped AP pollen grains are more frequent. This rise is accompanied by decreasing concentrations and percentages of Poaceae: wild grass group and Plantago lanceolata type pollen. Trifolium type, Lactuceae pollen and Pteridium aquilinum spores become scarce. The total pollen concentration shows a general increase. A low is noted at 947 cm (c. 3520 BC).

Noticeable are the slightly elevated abundances of Hordeum group pollen in zone TVL-P2. As its relative and absolute values parallel those of Poaceae: wild grass group pollen, the presence of Hordeum group pollen possibly reflects the occurrence of Glyceria in the lake fringe rather than the cultivation of Hordeum plants (cf. Stolze et al., 2012). TVL-P3 (944.5e899.5 cm; c. 3500e3060 BC): An increase in AP proportions from 88 to 96% and the resurgence of Hedera helix pollen values typify this zone. TVL-P3 is divided into two subzones. TVL-P3a (944.5e924.5 cm; c. 3500e3300 BC): While the Alnus pollen show increased values (34%), other AP values decline or remain at the previously attained levels. The recovery of various NAP types such as Poaceae: wild grass group, Lactuceae, Plantago lanceolata type and Trifolium type is documented. The total pollen concentration shows lows at 933 cm (c. 3380 BC) and 928 cm (c. 3340 BC). TVL-P3b (924.5e899.5 cm; c. 3300e3060 BC): This subzone opens with a continuous Fraxinus pollen curve. The AP values rise to high proportions of 95% at 901e900 cm (c. 3080e3070 BC), with Alnus and Corylus pollen contributing 36% and 34%, respectively. The representations of pollen and spores indicating disturbance such as Ulex type, Plantago lanceolata type and Pteridium aquilinum decrease notably. TVL-P4 (899.5e870 cm; c. 3060e2860 BC): High AP representations of c. 95% are recorded. This zone is divided into two subzones to acknowledge changes in the AP composition. TVL-P4a (899.5e879.5 cm; c. 3060e2920 BC): Increased percentage data of Quercus and Fraxinus pollen and declined abundances of Corylus pollen are recorded. Clumped grains of Corylus pollen are absent. At 883 cm (c. 2950 BC), Ulmus pollen attains 15%, being immediately followed by a decline. The initiation of the Taxus pollen curve is observed at 888 cm (c. 2980 BC). Clumped AP grains occur. A steep drop in the total pollen concentration is noted above 897 cm (c. 3040 BC). TVL-P4b (879.5e870.0 cm; 2920e2860 BC): Corylus, Fraxinus and Taxus pollen values increase, whereas Ulmus and Alnus pollen are present at lower levels. Disturbance indicators including Plantago lanceolata type pollen and Pteridium aquilinum spores recur. Lows in the total pollen concentration occur at 879 (c. 2920 BC) and 876 cm (c. 2900 BC). 4.2.2. The aquatic pollen and non-pollen palynomorph record Important changes in the down-core distribution of the suberised basal cells of mucilaginous hairs of Nymphaeaceae (Pals et al., 1980) are summarised in Table 2. Other important aquatic pollen and non-pollen palynomorph (NPP) types include Sparganium type pollen and Ceratophyllum leaf spines (Fig. 5). A single occurrence of both Sparganium type pollen and Ceratophyllum leaf spines is recorded in TVL-P2a. Ceratophyllum remains are further observed in TVL-P2b and TVL-P2d, while Sparganium type pollen is frequent in TVL-P2c concurrent with the maximum abundance of Nymphaeaceae remains. In TVL-P3a, Sparganium type pollen is registered at 939 and 934 cm (c. 3440 and 3390 BC). Above 924 cm (c. 3300 BC), aquatic pollen and NPP types diminish from the Templevanny Lough record. 4.3. Geochemistry The geochemical record is divided into four zones and six subzones (Fig. 6). TVL-C1 (1000e984.5 cm; c. 4150e3920 BC): A high organic carbon content (43%) opens the record, but declines above 989 cm (c. 3990 BC). The decrease in this parameter is coupled with an increase in the non-combustible mineral fraction (46e53%) and

S. Stolze et al. / Quaternary Science Reviews 67 (2013) 138e159

143

Fig. 3. Bayesian age-depth model for the sediment sequence from Templevanny Lough. The plot shows the probability distributions for the single calibrated dates (light grey) and the posterior probability distributions taking the depth model into account (dark grey). The depth model curve envelops the 95.4% highest probability density intervals that represent a measure of uncertainty of the age estimates. Samples considered as outliers are marked with an asterisk.

S. Stolze et al. / Quaternary Science Reviews 67 (2013) 138e159

Fig. 4. Percentage diagram of the main terrestrial pollen and spore types from Templevanny Lough. A twentyfold exaggeration is applied for low percentages. Concentrations are given for selected pollen types.

144

somewhat elevated count rates of Si, Ti and Mn. An inwash layer at 985 cm marks the end of this zone (Table 2, Fig. 6). TVL-C2 (984.5e921.5 cm; c. 3920e3270 BC): A high inorganic carbon content and pronounced Ca intensities define this zone. Four subzones are distinguished. TVL-C2a (984.5e974.5 cm; c. 3920e3790 BC): Ca and inorganic carbon reach high values at 980e979 cm (c. 3860e3840 BC). This is succeeded by an increased non-combustible mineral fraction and higher readings of Al, Si, K, Ti, Mn and Fe at 977e976 cm (c. 3820e 3800 BC), coinciding with a second inwash layer (Table 2, Fig. 6). TVL-C2b (974.5e962.5 cm; c. 3790e3670 BC): The inorganic carbon content, Ca intensity and Ca/Si ratio attain maximum values at 970 cm (c. 3740 BC). The low intensities of the terrigenous elements Al, Ti and Mn recover at 965 cm (c. 3690 BC). TVL-C2c (962.5e945.5 cm; c. 3670e3510 BC): The terrigenous elements Al, Si, K and Ti show a rise in their intensities between 954 and 947 cm (c. 3590e3520 BC). TVL-C2d (945.5e921.5 cm, c. 3510e3270 BC): The values of inorganic carbon and Ca are lower than in the previous subzones and show a notable drop at 926 cm (c. 3320 BC). Distinct increases in the Al, Mn and Fe intensities are observed at 945 cm (c. 3500 BC). An increase of Al, Si, Ti, Mn and Fe occurs between 929 and 923 cm (c. 3350e3290 BC). TVL-C3 (921.5e896.5 cm, c. 3270e3030 BC): Organic carbon steadily increases and shows a plateau of 43% at 901e898 cm (c. 3080e3050 BC). A reduction in the intensities of Al, Si, Ti and Mn is observed at 905e898 cm (c. 3110e3050 BC), which is followed by an increase of these values. TVL-C4 (896.5e870 cm, c. 3030e2860 BC): This zone is marked by a declined proportion of organic carbon and increased amounts of terrigenous minerals. TVL-C4 is divided into two subzones. TVL-C4a (896.5e882.5 cm, c. 3030e2940 BC): Markedly increased values of the non-combustible fraction, Al and Si open this subzone. TVL-C4b (882.5e870 cm, c. 2940e2860 BC): A renewed increase in the non-combustible fraction and the Al, Si, Ti, Mn and Fe readings is recorded between 882 and 877 cm (c. 2940e2900 BC). A rise in inorganic carbon and Ca intensities is observed above 877 cm. 5. Discussion 5.1. Woodland dynamics, human impact and climatic variability at Templevanny Lough during the fourth millennium BC Coupled with the detailed chronology, the high-resolution proxy record from Templevanny Lough documents changes in vegetation, human development and climate during the fourth millennium BC. As lake systems are sensitive to variations in climate, the variations in the observed geochemical and biological parameters at Templevanny Lough register changes in climate and integrate information about climate-driven changes in the catchment area (cf. Adrian et al., 2009). Through the combination of multiple geochemical and biological proxies, the present study reveals that climatic stress influenced the intensity of human activity in the area. 5.1.1. Pre- and early elm decline environment between c. 4150e 3920 BC (Meso-/Neolithic transition) The pollen assemblage of TVL-P1 suggests that the landscape around Templevanny Lough was largely forested, with Ulmus, Corylus, Quercus and Alnus being the predominant taxa. Clumps of immature pollen grains indicate that these trees occurred in proximity to the lake (cf. Janssen, 1986). Viburnum opulus, Ilex aquifolium and Hedera helix, which commonly thrive under mild

S. Stolze et al. / Quaternary Science Reviews 67 (2013) 138e159 Fig. 5. Continued percentage diagram of the main terrestrial pollen and spore types from Templevanny Lough. Pollen and micro-remains of aquatic taxa are also displayed. A twentyfold exaggeration is applied for low percentages. Concentrations are given for selected pollen types.

145

146

S. Stolze et al. / Quaternary Science Reviews 67 (2013) 138e159

Table 2 Correlation between the abundance of Nymphaeaceae remains, geochemical data and sediment characteristics in the studied core sequence from Templevanny Lough. Depth (Age): Relative abundance of Nymphaeaceae remains

Depth (Age): Geochemistry

Depth (Age): Sediments

989e984 cm (c. 3990e3910 BC): Low (c. 1.5%)

989e985 cm (c. 3990e3930 BC): Increase in terrigenous elements 984 cm (c. 3910 BC): Rise in Ca, LOI 950
C 980e979 cm (c. 3860e3840 BC): High Ca, LOI 950
C 977e976 cm (c. 3820e3800 BC): Increased readings of terrigenous elements 970e965 cm (c. 3740e3690 BC): High Ca, LOI 950
C, low of terrigenous elements

985 cm (c. 3930 BC): Thin light grey inwash layer (1 mm)

984 cm (c. 3910 BC): Beginning rise 983e980 cm (c. 3900e3860 BC): 3.4e4.7% 978e977 cm (c. 3830e3820 BC): Low (c. 2%) 971e967 cm (c. 3750e3710 BC): Steep rise 967e963 cm (c. 3710e3670 BC): Maximum abundance (8.5%) 963e958 cm (c. 3670e3620 BC): Steep decline (8.5e3.3%) 955-947 cm (c. 3600e3520 BC): Decline (4.8e2.4%) 944-941 cm (c. 3490e3460 BC): Low (c. 2%) 941e931 cm (c. 3460e3370 BC): 4e5% 930e923 cm (c. 3360e3290 BC): Decline (2e0.7%) 923e908 cm (c. 3290e3140 BC): Recovery (2.5%) 905e900 cm (c. 3110e3070 BC): Recovery (4%) 899e896 cm (c. 3060e3030 BC): Low (2.2%) 880e877 cm (c. 2920e2900 BC): Minimum abundance (0.2%) 877 cm (c. 2900 BC): Beginning rise

965 cm (c. 3690 BC): Beginning increase of terrigenous elements 954e947 cm (c. 3590e3520 BC): Sudden rise in terrigenous elements 945 cm (c. 3500 BC): Distinct rise in terrigenous elements 941e929 cm (c. 3460e3350 BC): No further increase in terrigenous elements 929e923 cm (c. 3360e3290 BC): Increase in terrigenous elements 926 cm (c. 3320 BC): Drop in Ca 921e898 cm (c. 3270e3050 BC): LOI 550
C increase 905e898 cm (c. 3110e3050 BC): Low in terrigenous elements 898 cm (c. 3050 BC): Increase in terrigenous elements 882e877 cm (c. 2940e2900 BC): Increase in terrigenous elements 877 cm (c. 2940 BC): Rise in LOI 950
C and Ca

and humid climate conditions, were components of the woodland community. Minor woodland disturbance is suggested by the occurrence of Pteridium aquilinum spore and Plantago lanceolata type pollen. Early human activity in the region is supported by radiocarbon dates from the causewayed enclosure at Magheraboy (Danaher, 2007). Pre-elm decline woodland perturbation has also been noted elsewhere in Ireland (Hirons and Edwards, 1986; Molloy and O’Connell, 1987; Plunkett et al., 2008; Stolze et al., 2012). A notable change in the forest community at Templevanny Lough took place at c. 3970 BC with the onset of the elm decline. This age closely corresponds to the range of dates recorded for the onset of the elm decline in other Irish pollen records (cf. Parker et al., 2002). The initial decline in Ulmus pollen may be attributed to an elm-specific disease as palynological changes reflecting a reduction of other woodland components or an expansion of open-land taxa are not observed (cf. O’Connell and Molloy, 2001; Ghilardi and O’Connell, 2012; Stolze et al., 2012). The aquatic NPP and geochemical data in TVL-P1 and TVL-C1 suggest that particularly wet conditions prevailed during the early stage of the elm decline. The reduced abundance of Nymphaeaceae remains indicates that the floating-leaf community grew distant from the lake centre between c. 3990e3910 BC. In conjunction with increased count rates of terrigenous elements (Si, Ti and Mn), this suggests a period of higher rainfall leading to a rise in lake level and the deposition of an inwash layer at c. 3930 BC. The total pollen concentration shows a concomitant low, most likely indicating an increase in the sediment accumulation rate rather than a low pollen input. The Nymphaeaceae remains in the Templevanny Lough record may originate from Nuphar and/or Nymphaea, the two genera of Nymphaeaceae occurring in Ireland today (Webb et al., 1996). As their distribution is largely limited to slow-moving and shallow waters of up to 3 m depth (Heslop-Harrison, 1955a,b), Nymphaeaceae remains can be used as a measure for lake-level oscillations throughout the considered time interval (Table 2; see also Stolze et al., 2012). In most instances, low values of Nymphaeaceae

976.5 cm (c. 3810 BC): Thin light grey inwash layer (5 mm)

First occurrence of small gastropod shell fragments at 960 cm (c. 3640 BC)

remains are accompanied by increased values of terrigenous minerals, supporting the assumption of a higher lake level due to increased rainfall. In contrast, high values of Nymphaeaceae remains are associated with a low influx of terrigenous material (Figs. 5 and 6). The bathymetry of Templevanny Lough (Fig. 2) indicates that a lowered lake level would increase the proportion of shallow water areas, which in turn would facilitate the expansion of the floating-leaf vegetation. 5.1.2. Elm decline and early human activity between c. 3920e3810 BC (Early Neolithic) The decline in the local elm population at Templevanny Lough spanned a period of c. 150 years (c. 3970e3820 BC), as reflected in the falling Ulmus pollen curve (20e1.4%) in TVL-P1 and TVL-P2a. Comparable reductions in Ulmus pollen from about 20 to <5% have also been noted in other limestone areas of Ireland where Ulmus contributed significantly to the woodland cover (Brown et al., 2005; Ghilardi and O’Connell, 2012; Stolze et al., 2012). The longevity of this decline suggests that it was possibly caused by a combination of environmental stress and human impact (cf. Peglar and Birks, 1993), although its initiation may be attributed to an elm-specific disease. The decline in other arboreal pollen and the growing presence of Plantago lanceolata type pollen in TVL-P2a indicate that anthropogenic disturbance intensified around Templevanny Lough during this time. The thinning of the woodland cover favoured the spread of light-tolerant trees and shrubs including Sorbus and Ulex and the expansion of grassland taxa in the landscape. The total pollen concentration shows a concomitant decline, possibly reflecting the increased abundance of herbaceous taxa and the associated change in pollen productivity and dispersal. It has been shown that deforestation can lead to the replacement of arboreal taxa with a high pollen representation by non-arboreal taxa with a low pollen production (Odgaard, 1999). The Nymphaeaceae remains in TVL-P2a provide evidence for a beginning decline in lake level at c. 3910 BC. The presence of Ceratophyllum remains in the lake sediments may suggest an

S. Stolze et al. / Quaternary Science Reviews 67 (2013) 138e159

Fig. 6. Down-core distribution of loss on ignition data and element count rates on the studied sediment sequence from Templevanny Lough.

147

148

S. Stolze et al. / Quaternary Science Reviews 67 (2013) 138e159

increase in water temperature. Wilkinson (1963) noted a critical minimum temperature of 20
C for the growth of Ceratophyllum. The establishment of somewhat drier and warmer conditions is supported by the geochemical data in TVL-C2a. Despite the reduction in woodland cover, no further increase of allochthonous material in the lake sediment is recorded. Increased Ca intensities and inorganic carbon values point to notable authigenic carbonate formation between c. 3860e3840 BC, possibly reflecting increased photosynthetic activity and/or elevated spring and summer temperatures (cf. Wetzel, 2001). A subsequent interval of elevated rainfall resulting in lake-level rise and the deposition of a second inwash layer occurred between c. 3830e3800 BC, as indicated by a low in the Nymphaeaceae remains and elevated values of terrigenous elements (Table 2). This hydrological change coincided with the lowest abundance of elm around the lake. 5.1.3. Human impact between c. 3810e3630 BC (Early Neolithic) Pollen zone TVL-2b suggests that Plantago lanceolata was a dominant component of the herbaceous community on the mineral soils around the lake between c. 3810e3740 BC. The rise in Nymphaeaceae remains, coupled with the occurrence of Thelypteris palustris spores, suggests a drop in lake level and progressive terrestrialisation. The geochemical data in TVL-C2b imply that authigenic carbonate formation was particularly high during this period. Despite the increased availability of open terrain, the influx of allochthonous material into the lake was low. This line of evidence points to the establishment of drier and warmer conditions. The pollen data in TVL-2c indicate a subsequent phase of marked woodland clearance between c. 3740e3630 BC, reflecting the most intense human activity at Templevanny Lough during the Neolithic. During this interval, Plantago lanceolata was progressively replaced by grasses, which was also accompanied by a shift in the composition of other herbaceous taxa. A similar succession was also observed at nearby Loughmeenaghan, where wheat cultivation was practiced during the Plantago lanceolata phase (Stolze et al., 2012). The vegetational change has been interpreted to reflect a shift from arable to pastoral farming. A notable reduction in the Alnus population, typically growing on waterlogged soils, occurred at the time. The highest abundance of Nymphaeaceae remains in TVL-P2c, the high Ca and inorganic carbon values and low intensities of terrigenous elements in TVLC2b collectively suggest that particular dry and warm conditions prevailed between c. 3710e3670 BC, leading to the lowest lake level at Templevanny Lough during the Neolithic period. The presence of Sparganium type pollen and Thelypteris spores suggest further terrestrialisation. The dry conditions appear to have ended at c. 3670 BC. Increased intensities of terrigenous elements in TVL-C2b associated with a steep drop in the Nymphaeaceae remains and in the total pollen concentration in TVL-P2c (Table 2) suggest an increased inwash from the lake’s catchment area and a rise in lake level in response to higher rainfall prevailing through c. 3620 BC. The changing environmental conditions were accompanied by the occurrence of freshwater molluscs in the lake centre (Table 2). 5.1.4. Woodland recovery between c. 3630e3500 BC (Middle Neolithic) Coinciding with the climatic deterioration, a first decline in human activity occurred in the lake’s catchment area as suggested by the recovery of arboreal pollen and a decline in Plantago lanceolata type pollen in TVL-P2d. The disappearance of Pteridium aquilinum spores from the pollen record provides supporting evidence that clearance activities were limited. Grazed or mown grassland declined notably in the landscape.

Abandonment of the area facilitated expansion and recuperation of Corylus, Ulmus, Fraxinus and Sorbus. The vegetational change in the pollen source area is also reflected in the rising total pollen concentrations (cf. Odgaard, 1999). Along the lake shore, the local Alnus population regenerated, reflecting an increase in waterlogged soils. Clumps of immature pollen grains suggest that various tree taxa grew in proximity to the lake. Although inwash of pollen clumps cannot be ruled out, analysis of surface sediment samples showed that pollen deposition in the central part of the lake is largely influenced by the vegetation around the lake (Stolze, unpubl. data). Despite the decline in Nymphaeaceae remains in TVL-P2d, reflecting a rise in lake level, the Ca intensities and inorganic carbon fraction still remain relatively high. This suggests that increased spring/summer temperatures continued to prevail. The presence of Ceratophyllum remains may support the occurrence of increased water temperatures, at least until c. 3600 BC. A period of increased rainfall occurred shortly thereafter between c. 3600e3520 BC. 5.1.5. Small-scale human impact between c. 3500e3300 BC (Middle Neolithic) Following the initial decline in human impact, the pollen record in TVL-P3a suggests that human activity continued on a smaller scale between c. 3500e3300 BC. The recurrence of Trifolium type and Lactuceae pollen suggests that open areas were likely maintained for pastoral farming. Alnus grew abundantly around the lake, implying an increase in waterlogged soils around the lake. The change in soil hydrology appears to have resulted in the nearby decline in Corylus, Ulmus and Quercus as suggested by the absence of clumped pollen. Notable is the increase in Hedera helix pollen. The reduced human impact and woodland recovery appears to have positively affected the growth of Hedera helix around the lake, as this taxon is sensitive to cutting and grazing (Metcalfe, 2005). The NPP and geochemical data of TVL-P3a and TVL-C2d support the presence of wetter conditions at c. 3500 BC (Table 2). The subsequent increase in Nymphaeaceae remains and continued influx of terrigenous material suggest no further deterioration of the climatic conditions between c. 3460e3370 BC. The simultaneous occurrence of Sparganium type pollen and Thelypteris palustris spores indicates that the declined water table allowed the expansion of marshy terrain during the time. A return to increased wetness between c. 3360e3290 BC is inferred from the reduced abundance of Nymphaeaceae remains and the increase in terrigenous elements (Table 2). Concomitant lows in the total pollen concentration point to an increase in the sediment accumulation rate. The drop in Ca and inorganic carbon possibly reflects the onset of cooler spring/summer temperatures at c. 3320 BC, marking the end of a relatively warm interval that begun at c. 3910 BC. 5.1.6. Woodland recovery between c. 3300e3060 BC (Middle/Late Neolithic) Human impact on the landscape began to further diminish at c. 3300 BC, as reflected in the decline of disturbance indicators in TVLP3b. This led to an almost complete woodland recovery between c. 3080e3070 BC. While Alnus further expanded around the lake, the previously underrepresented Fraxinus possibly became part of the wetland community. Declining human impact associated with full woodland regeneration at this time was also observed elsewhere in Ireland (Hirons and Edwards, 1986; O’Connell and Molloy, 2001; Verrill and Tipping, 2010; Ghilardi and O’Connell, 2012). The geochemistry data in TVL-C3 suggest that a lower influx of allochthonous material into the lake occurred, possibly in response to increasing soil stability linked to woodland closure. Along with the recovery of the Nymphaeaceae values in TVL-P3b, this suggests

S. Stolze et al. / Quaternary Science Reviews 67 (2013) 138e159

that the climatic conditions ameliorated after c. 3290 BC (Table 2), with warmest and driest conditions during this period occurring between c. 3110e3060 BC (Table 2). 5.1.7. Changing woodland composition and recurring human impact between c. 3060e2860 BC (Late Neolithic) The low human interference facilitated the expansion of Ulmus, Quercus and Fraxinus in the local woodland community, replacing Corylus as suggested by the declined pollen abundance and the disappearance of clumped pollen grains in TVL-P4a. The woodland experienced a noticeable change in its composition after c. 2950 BC, associated with a decline in the local Ulmus population. TVL-P4b indicates another change in woodland composition after c. 2920 BC. Corylus and Taxus expanded in the landscape, while Alnus declined notably. The Taxus expansion is in agreement with the established population dynamics for western and southwestern Ireland, where a possibly climate-related spread occurred at c. 2900 BC (O’Connell and Molloy, 2001). Anthropogenic activity in the study area revived slightly. A similar increase in human impact was also observed at other sites in western Ireland at the beginning of the third millennium BC, preceding the Bronze Age (O’Connell and Molloy, 2001). The NPP and geochemical data of TVL-P4 and TVL-C4 indicate that the influx of allochthonous material increased markedly at Templevanny Lough at c. 3060 BC. Along with the high AP representations of 94%, lows in the total pollen concentration and low abundances of Nymphaeaceae remains, this suggests that periods of intense rainfall occurred between c. 3060e3030 BC and between c. 2940e2900 BC. The expansion of Taxus coincided with the latter episode, corroborating that its spread was climate-related. The increase in Ca and inorganic carbon in the sediments along with a beginning rise in Nymphaeaceae remains provide evidence for climatic amelioration for the period after c. 2900 BC.

5.2. A climate record for the CarrowkeeleKeshcorran area between c. 4100e2600 BC Peatlands and lake systems located in the same region respond similarly to regional driving forces (Anderson, 1998; Livingstone, 2008). In lakes, the degree of coherence is greatest for parameters affected by climate forcing (Livingstone, 2008). Although modulation of large-scale climate signals by local factors can affect the strength of the correlation between climate forcing and the lake response, a coherent response of lake systems to large-scale climate signals can be expected (Livingstone et al., 2010). In addition to Templevanny Lough, the nearby lake of Loughmeenaghan and fen site of Lough Availe (Fig. 1) have been previously studied (Stolze et al., 2012; Stolze, 2012a, b). Based on the above findings, comparison of the same geochemical and biological parameters in the three records from the CarrowkeeleKeshcorran area permits the reconstruction of temporally and spatially coherent climate signals during most of the Neolithic period between c. 4100e2600 BC (Fig. 7). The timing of the hydroclimatic changes described below is largely based on the Templevanny Lough archive, as it represents the most continuous record between c. 4150e2860 BC. The Loughmeenaghan record provides important constraints for the period between c. 2860e2620 BC. Due to a different local hydrology, the Lough Availe record is more complex than those of the two lake sites. 5.2.1. The Early Neolithic (4000e3600 BC) The wet conditions centring on c. 3950 BC inferred from the Templevanny Lough record are supported by the Lough Availe archive. An elevated abundance of aquatic pollen provides evidence

149

for increased wetness during the time, given the uncertainties of radiocarbon dating (Fig. 7, event A; Stolze, 2012a). The onset of drying conditions after the elm decline is recorded at all three sites in the CarrowkeeleKeshcorran area (Fig. 7, event B). An interval of very slow deposition and/or a depositional hiatus starting at c. 3920 BC and persisting for c. 660 years was observed at Lough Availe (Stolze, 2012a, b). This date is in agreement with the beginning decline in lake level at Templevanny Lough at c. 3910 BC, with lowest conditions occurring between c. 3710e3670 BC (Fig. 7, event D). Lowering of the lake level at Loughmeenaghan began at c. 3720 BC, coinciding with the intensification of human impact at this site. The lowest water level prevailed between c. 3600e3570 BC (Fig. 7, event D; Stolze et al., 2012). Compared to the Templevanny Lough and Lough Availe records, the Loughmeenaghan chronology appears to be slightly biased towards younger ages during the Early Neolithic (cf. Stolze, 2012a). However, the similar succession of changes in the biological and geochemical proxies in both the Templevanny Lough and Loughmeenaghan records points to a synchrony of events. This is supported by the occurrence of short-lived intervals of increased precipitation at the end of the elm decline and during the lowest lake-level phase (Fig. 7, events C, E) at both sites as well as the similar length of the lowest lake-level phase of c. 30 and 40 years, respectively (Fig. 7, event D). 5.2.2. The Middle Neolithic (3600e3100 BC) The Templevanny Lough record shows that the warm and dry interval ended at c. 3670 BC (Fig. 7, event F). This was followed by a still relatively warm period with two wet periods between c. 3600e3520 BC and between c. 3500e3460 BC (Fig. 7, events G, H). A drier interval is recorded between c. 3460e3370 BC (Fig. 7, event I). Wet conditions associated with a noticeable drop in spring/ summer temperatures prevailed between c. 3360e3290 BC (Fig. 7, event J). The Loughmeenaghan proxy record corroborates a relatively sudden shift from drier to wetter conditions, which dates to c. 3570 BC (Fig. 7, event F). Multiple climatic shifts synchronous within dating uncertainties with those from Templevanny Lough were inferred for the Middle Neolithic. Wet shifts were noted at c. 3550, 3500 and 3330 BC (Fig. 7, events G, H, J). In agreement with the Templevanny Lough data, the interval between c. 3330e3280 BC (Fig. 7, event J) led to a notable rise in lake level and an expansion of the local Alnus population around Loughmeenaghan. Similar to the Templevanny Lough data, an intermediate period of somewhat warmer and drier conditions centring on c. 3400 BC (Fig. 7, event I) was also noted during which human activity revived (Stolze et al., 2012). The Lough Availe record, which extends to c. 3080 BC, indicates that the moisture availability must have been comparatively low prior to c. 3260 BC. This suggests that temperatures were high enough to compensate for increased rainfall at this site during the relatively warm period as inferred from Templevanny Lough. A steep rise in the deposition rate and the formation of a lake environment on the previously dry grounds occurred just prior to c. 3260 BC, implying that the wet and cool period centring on c. 3300 BC, as noted at the two other sites, led to a steep increase in moisture availability (Fig. 7, event J; Stolze, 2012a,b). 5.2.3. The Late Neolithic (3100e2500 BC) The Templevanny Lough record suggests a climatic amelioration between c. 3110e3060 BC (Fig. 7, event K) that was succeeded by pronounced wet shifts between 3060e3030 BC and between c. 2940e2900 BC (Fig. 7, events L, M). A shift towards drier conditions is registered at c. 2900 BC (Fig. 7, event N). The Loughmeenaghan proxy record, which extends further to c. 2620 BC, reveals a period

150 S. Stolze et al. / Quaternary Science Reviews 67 (2013) 138e159 Fig. 7. Comparison of important geochemical and biological variables from Templevanny Lough, Loughmeenaghan and Lough Availe in the CarrowkeeleKeshcorran area. Alnus pollen, suberised basal cells of mucilaginous hairs of Nymphaeaceae and loss on ignition data are key proxies used in the reconstruction of the combined climate record. Changes in woodland cover and the intensity of human impact during the Neolithic are reflected by the total tree pollen, Ulmus and Plantago lanceolata type pollen. The following key events were identified: (A) wet; (B) drying; (C) wet; (D) dry; (E) wet; (F) wet; (G) wet; (H) wet; (I) drier; (J) wet/cool; (K) drier; (L) wet; (M) wet; (N) drier; (O) wet. See text for details of the events and discussion on the chronologies.

S. Stolze et al. / Quaternary Science Reviews 67 (2013) 138e159

of particularly high rainfall between c. 3060e2900 BC (Fig. 7, events L, M), supporting the evidence from Templevanny Lough. In addition, a drier period is registered between c. 2890e2770 BC and a period of wetter conditions between c. 2770e2680 BC (Fig. 7, events N, O). 5.3. Climate variability in Ireland during the Neolithic (4000e2500 BC) Today’s climate of Ireland is strongly influenced by the North Atlantic Oscillation and the effect of the North Atlantic Drift (Hulme and Barrow, 1997). A coherent response of lake variables to variations in these atmospheric and ocean circulation patterns has been detected in lake systems in Ireland on a regional scale (George et al., 2010). To examine the temporal and spatial validity of the inferred climatic fluctuations, the established palaeoclimate reconstruction from the CarrowkeeleKeshcorran area is compared with hydroclimatic records from Ireland (Fig. 8). It can be shown that the palaeoclimatic shifts established for the CarrowkeeleKeshcorran area occurred more or less synchronously, within the limits of radiocarbon dating, across the island. 5.3.1. The Early Neolithic (4000e3600 BC) The combined climate record from the CarrowkeeleKeshcorran area indicates that a period of increasing spring/summer temperatures and declining precipitation occurred between c. 3910e3670 BC, with the warmest and driest conditions of the Neolithic prevailing between c. 3710e3670 BC. An intermediate wet event occurred at c. 3830 BC. Evidence for low moisture availability during the Early Neolithic has also been noted elsewhere in Ireland. Stable-isotope stratigraphy from Lough Gur, County Limerick, in Western Ireland suggests the occurrence of a low lake level at c. 3920 BC, coinciding with the first evidence for human activity in the area (Ahlberg et al., 2001). This date corresponds to the onset of slow deposition at Lough Availe and the initial lake-level decline at Templevanny Lough. A shift towards drier conditions at c. 3850 BC resulted in higher humification levels in peat sequences at Achill Island in County Mayo (Caseldine et al., 2005) and drier bog surface conditions at Derragh Bog in central Ireland (Langdon et al., 2012). The speleothem d18O record from Crag Cave in County Kerry, southwest Ireland, indicates a rise in temperature after c. 3870 BC (Fig. 8; McDermott et al., 2001). The tree-ring chronology of Irish bog oaks indicates that a period of warmer and drier conditions occurred between 3900 and 3450 BC (Briffa and Atkinson, 1997). Irish bog and lake-edge tree populations were abundant at around 3650 BC (Fig. 8), suggesting that water tables were notably lower than today (Turney et al., 2006). This age corresponds, within dating uncertainties, to the interval of c. 3710e3670 BC obtained for the driest conditions at Templevanny Lough. 5.3.2. The Middle Neolithic (3600e3100 BC) The CarrowkeeleKeshcorran climate record indicates that the warm and dry phase was followed by a still relatively warm interval marked by several pronounced hydroclimatic shifts. The observation that temperatures remained relatively high during the Early and parts of the Middle Neolithic is supported by the slow peat accumulation recorded on Achill Island between c. 3850e3250 BC (Caseldine et al., 2005). This is in agreement with the observed slow deposition at Lough Availe between c. 3920e3260 BC and the elevated Ca intensities and inorganic carbon data from Templevanny Lough between c. 3910e3320 BC. In the CarrowkeeleKeshcorran area, a wet shift is recorded at c. 3670 BC. The multi-proxy record from An Loch Mór on the Aran

151

Islands indicated acceleration in lake-level rise at c. 3650 BC (Holmes et al., 2007). At around the same time, bog surface wetness began to increase at Derragh Bog (Fig. 8; Langdon et al., 2012), supporting the CarrowkeeleKeshcorran evidence. Multi-proxy analysis of sediments from Lough Mask in County Mayo revealed a period of increased rainfall starting at c. 3550 BC (Murnaghan et al., 2012), which possibly correlates to the wet shifts at c. 3600 BC and 3500 BC observed in the present study. The speleothem d18O record from Crag Cave implies a low in temperatures at c. 3600 BC followed by notable temperature fluctuations at a multi-decadal timescale with cooling occurring after c. 3520 BC (Fig. 8; McDermott et al., 2001). The CarrowkeeleKeshcorran archives suggest that this wet period was followed by a dry shift at c. 3460 BC. Increasing mire surface dryness was also recorded from Derragh Bog between c. 3500e3350 BC (Langdon et al., 2012) and from Stanshiel Rig after c. 3450 BC (Cayless and Tipping, 2002). Elevated temperatures appear to have prevailed at c. 3460, 3260 and 3390 BC (Fig. 8; McDermott et al., 2001). A number of Irish proxy records support, within dating uncertainties, the observation of the present study that a pronounced wet and cool phase prevailed between c. 3360e3290 BC. The Derragh Bog data imply a shift towards increased wetness at c. 3350 BC (Fig. 8; Langdon et al., 2012). The peat sequences from Achill Island record an increase in the amount of clastic material, suggesting increased storminess and heavy rain sometime between c. 3250e 3150 BC (Caseldine et al., 2005). At the same time, a cessation of varve formation was recorded at An Loch Mór (Holmes et al., 2007). Caseldine et al. (2005) suggested that this extreme climatic event was unique for the majority of the Holocene period. Similar to the deposition model of Lough Availe, the age-depth diagram of a core from the Glenulra basin, County Mayo, reveals a steep increase in peat accumulation centring on c. 3200 BC (Molloy and O’Connell, 1995). The high-resolution speleothem d18O record from Crag Cave registered a climatic cooling after c. 3390 BC with particular low temperatures occurring at c. 3360 BC and 3290 BC (Fig. 8; McDermott et al., 2001). It has been suggested that this cooling was possibly linked to a weaker North Atlantic thermohaline overturning at 3260 BC (McDermott et al., 2001). The CarrowkeeleKeshcorran evidence of a subsequent climatic amelioration is supported by the spread of pine on bog surfaces on Achill Island (Caseldine et al., 2005) and elsewhere in Ireland (Caulfield et al., 1998; O’Connell and Molloy, 2001). The speleothem d18O record from Crag Cave suggests an increase in temperatures (Fig. 8; McDermott et al., 2001). 5.3.3. The Late Neolithic (3100e2500 BC) The palaeoclimate reconstruction from CarrowkeeleKeshcorran suggests that frequent climatic oscillations took place during the Late Neolithic (Fig. 8). The establishment of drier/warmer conditions centring on c. 3080 BC is supported by the abundant occurrence of Irish bog and lake-edge tree populations at c. 3050 BC (Turney et al., 2006), suggesting a lower water table than present. The inferred wet phase between c. 3060e2900 BC has also been registered in the peat sequences from Belderrig Beg where highest surface wetness associated with rapid peat accumulation occurred at c. 2990 BC (Verrill and Tipping, 2010). The onset of wetter climatic conditions at c. 2850 BC is reflected by an acceleration of lake-level rise at An Loch Mór (Holmes et al., 2007) and increased peat accumulation on Achill Island (Caseldine et al., 2005). A notable temperature decline after c. 2830 BC is implied by the speleothem d18O record from Crag Cave (Fig. 8; McDermott et al., 2001). The wet period between c. 2770e2680 BC observed in the CarrowkeeleKeshcorran area correlates, within the uncertainties of radiocarbon dating, to an increase in peat accumulation and bog

152 S. Stolze et al. / Quaternary Science Reviews 67 (2013) 138e159 Fig. 8. Comparison of the combined precipitation and temperature reconstruction from the CarrowkeeleKeshcorran with moisture/precipitation and temperature reconstructions from northern and western Europe and the Alps for the period between 4000 and 2600 BC. Grey shading denotes periods of warmer and drier conditions inferred from the CarrowkeeleKeshcorran archives. References: (1) McDermott et al. (2001, data are given as five-point moving average of d18O); (2) Turney et al. (2006); (3) Langdon et al. (2012); (4) Hughes et al. (2000); (5) Langdon et al. (2003); (6) Anderson et al. (1998); (7) Karlén and Kuylenstierna (1996); (8) Gunnarson (2008); (9) Helama et al. (2002); (10) Grudd et al. (2002); (11) de Jong et al. (2006); (12) Magny et al. (2006); (13) Wanner et al. (2011).

S. Stolze et al. / Quaternary Science Reviews 67 (2013) 138e159

surface wetness at Céide Fields at c. 2700 BC (Molloy and O’Connell, 1995). At Lough Mask, a phase of increased rainfall was observed between c. 2750e2550 BC (Murnaghan et al., 2012). Increased bog surface wetness also occurred at Derragh Bog between c. 2900e 2650 BC (Langdon et al., 2012). Although the Irish proxy records registered climatic shifts during the Late Neolithic, the timing of the events appears to be less clear than during previous phases, possibly suggesting that the climatic oscillations were less pronounced in magnitude. 5.4. Evidence for climatic variability in northern and Western Europe and the Alps between 4000 and 2600 BC Large-scale circulations of the atmosphere and ocean, such as the North Atlantic Oscillation, influence the climate in Europe (Woolings, 2010) and have an impact on marine, terrestrial and freshwater ecosystems across the region (cf. Hurrell et al., 2003). For wetland and lake ecosystems, a spatial and temporal coherence in the response of physical, chemical and biological variables to climatic oscillations has been documented in Europe on regional to supraregional scales (e.g., Hughes et al., 2000; Blenckner et al., 2007). To better constrain the timing of climatic variations during the Irish Neolithic, the climate record from the Carrowkeele Keshcorran area and the compiled data from Ireland are compared to temperature and moisture/precipitation reconstructions from northern and western Europe and the Alps. The climate variability in these regions is influenced by similar atmospheric circulation patterns (cf. Blenckner et al., 2007). To facilitate comparison with the published records, all radiometric data are reported in calendar years BC. Although the different proxy records vary in their sensitivity to hydrological variability, sample resolution and uncertainties associated with radiometric measurements (cf. Charman et al., 2006), several major climate shifts occurring across large parts of Europe can be identified (Table 3). 5.4.1. The Early Neolithic (4000e3600 BC) The wet climatic conditions centring on c. 3950 BC established for the CarrowkeeleKeshcorran area fall within a period of cool and wet conditions in northern and Western Europe (Fig. 8) and overlap with the second of six cool relapses during the Holocene between c. 4550e3950 BC (Wanner et al., 2011). Analysis of peat sequences revealed wetter conditions at Walton Moss in England for the period between c. 4050e3950 BC (Hughes et al., 2000), at Temple Hill Moss in southeast Scotland between c. 4300e3850 BC (Langdon et al., 2003), at Meerstalblok in the Netherlands at c. 4010e3910 BC (Blaauw et al., 2004) and at two bog sites in northern Norway at c. 3940 BC (Vorren et al., 2012). Lake-level reconstructions from central Europe indicate highstands between c. 4400e3950 BC (Magny, 2004). Evidence for depressions in summer temperature during this time is provided by Scandinavian tree-ring chronologies (Fig. 8). The 7400-year Scots pine chronology from northern Sweden suggests that summer temperatures were about 2
C lower than present between c. 4150e3950 BC (Grudd et al., 2002). Particularly cold conditions between 3950 and 3921 BC were inferred from the 7500 year-long Scots pine chronology from northern Finland (Helama et al., 2002). In agreement with the climate reconstruction from the CarrowkeeleKeshcorran area, climatic amelioration after c. 3910 BC has also been observed elsewhere in northern and western Europe and the Alps (Fig. 8). Drier conditions were noted in northeastern Scotland between c. 3900e3150 BC (Dubois and Ferguson, 1985) and in northwestern Scotland between c. 4050e3550 BC, with the driest conditions occurring at c. 3550 BC (Anderson et al., 1998). The peat record of

153

Glen Carron reflects a dry episode between c. 3920 and 3780 BC (Anderson, 1998). At Stanshiel Rig in southern Scotland, drier peat surface conditions were inferred for the interval between c. 3850e 3650 BC (Cayless and Tipping, 2002). A shift towards drier conditions was recorded in southeastern Scotland (Langdon et al., 2003) and in England (Hughes et al., 2000) at c. 3900 BC, persisting through to c. 3550 BC. Lake-level reconstructions in west-central Sweden using subfossil wood of Pinus sylvestris L. preserved in bogs and lakes provide evidence for low water levels between c. 3800e3600 BC (Gunnarson, 2008). Temperature reconstructions based on tree-line fluctuations and tree-ring chronologies from Scandinavia reveal higher summer temperatures between 3850 and 3600 BC (Fig. 8). In northern Sweden and Norway, the pine tree-line was about 200 m above the present limit at c. 3650 BC, which corresponds to 1.2
C higher summer temperatures (Karlén and Kuylenstierna, 1996). This date is in accordance with the timing of lower water stands in Ireland. The summer temperature reconstruction from northern Sweden (Grudd et al., 2002) shows a steep increase to modern levels at c. 3900 BC, which prevailed until c. 3600 BC (Fig. 8). A century-long warm period between 3841 and 3742 BC has been inferred from the tree-ring chronology from northern Finland (Helama et al., 2002). This falls within the particular warm and dry period established for the CarrowkeeleKeshcorran area. Coinciding with the warm and dry conditions in northern Europe, glacier retreats and lower lake levels were recorded in central Europe. A recession of the Tschierva Glacier in the eastern Swiss Alps was identified between c. 4250e3700 BC (Joerin et al., 2008). Temperature modelling yielded a rise in summer temperature of 1.8
C, at constant precipitation (Joerin et al., 2008). Climatic amelioration resulted in lower lake levels at Lake Morat in Switzerland and Lake Annecy in eastern France, which facilitated the construction of Neolithic lake shore villages (Magny et al., 2005). Remains of the villages were tree-ring dated at 3840 and 3780 BC. The palaeoenvironmental data from northern and western Europe and the Alps collectively suggest that the onset of the warmer and drier conditions occurred sometime between c. 3950e 3850 BC. The end of this phase, however, is less well constrained and appears to have occurred between c. 3750e3250 BC. This uncertainty is probably best explained by the fact that the different proxies used show variable sensitivities to hydrological changes. 5.4.2. The Middle Neolithic (3600e3100 BC) Multiple climatic shifts similar in timing to those inferred from the CarrowkeeleKeshcorran archives were noted elsewhere in Europe during the Middle Neolithic (Table 3, Fig. 8). In the CarrowkeeleKeshcorran area, climatic deterioration commenced at c. 3670 BC. This was followed by pronounced hydroclimatic shifts during the Middle Neolithic. At Stanshiel Rig in southern Scotland, a shift towards increased mire surface wetness occurred between c. 3750e3450 BC (Cayless and Tipping, 2002). Humification analysis on peat profiles in Scotland points to increased wetness at c. 3550 BC (Anderson et al.,1998). Wet conditions were also recorded in Norway between c. 3780e 3690 BC and between c. 3520e3480 BC (Vorren et al., 2007). The latter coincides with a lower abundance of subfossil pine wood in bogs and on lake edges in west-central Sweden c. 3600e3400 BC (Gunnarson, 2008). At Meerstalblok in the Netherlands, a shift towards wet conditions occurred at c. 3635 BC, which were particularly pronounced at c. 3535 BC (Blaauw et al., 2004). Increased deposition of aeolian material on raised bogs in southwestern Sweden suggests that the climatic deterioration was accompanied by higher storminess between c. 3750e3500 BC (Björck and Clemmensen, 2004). In the eastern Swiss Alps, the climatic shift resulted in a centennial-scale advance of the Tschierva Glacier at c. 3700 BC

154

S. Stolze et al. / Quaternary Science Reviews 67 (2013) 138e159

Table 3 Comparison between the inferred climatic conditions in northern and western Europe, the solar activity as inferred from the 14C production rate (Muscheler et al., 2005) and periods of human development and archaeological presence during the Irish Neolithic (see text for references). All dates refer to calibrated ages. Climatic conditions

Solar activity

c. 4050e3950 BC

Wet/cool

c. 3950e3700 BC

Dominantly dry/warm

c. 3950e3850 BC

Drying/warming

c. 3850 BC

Wetter/cooler

c. 3700e3250 BC

c. 3700 BC c. 3650 BC c. 3550 BC c. 3400 BC c. 3350e3250 BC c. 3250e2950 BC

c. 2950e2850 BC

c. 2850e2600 BC

Periods of human development and archaeological presence in Ireland

4040e3940 BC

Generally low

4015e3940 BC 3940e3680 BC

Decline Generally higher

3940e3860 BC 3860e3830 BC 3850 BC

Increase High Small low

Dominantly wet/cool

3700e3260 BC

Marked multidecadal fluctuations

Wet/cool Wet/cool Wet/cool Warm/dry Very wet/cool Dominantly dry/warm

3720e3700 3680e3630 3550e3490 3490e3400 3360e3260 3260e2950

Decline Decline Decline Increase Low Generally higher

Dominantly wet/cool

Less pronounced hydroclimatic shifts

BC BC BC BC BC BC

3210 BC 3150 BC 3060e3040 BC 2980e2950 BC 2950e2830 BC

High High High High Generally lower

2950 BC 2875e2830 BC 2830e2600 BC

Steep decline Low Generally lower amplitude than during fourth millennium BC Increase Decline Decline

2830e2805 BC 2750e2720 BC 2700 BC

(Joerin et al., 2008), which corresponds in timing to the wet shift recorded at Templevanny Lough. Extended snow cover and increased catchment erosion at c. 3550 BC have been inferred from a high mountain lake in the central Austrian Alps (Schmidt et al., 2002). The mid-European lake-level reconstruction suggests the onset of a phase of higher water tables at c. 3700 BC (Magny, 2004). A lake-level rise in response to abrupt climatic changes was also recorded at Lake Constance (Magny et al., 2006). The increased wetness in northern and western Europe and the Alps overlapped with notable depressions in summer temperature (Fig. 8). This is reflected in the lowering of the pine tree-line limit and glacial advances in Scandinavia after c. 3650 BC (Karlén and Kuylenstierna, 1996). The summer temperature reconstruction from northern Sweden suggests that temperatures were c. 2
C lower than today at c. 3500 BC (Grudd et al., 2002). Cooler periods have also been identified from the tree-ring chronology from northern Finland between 3710 and 3681 BC and between 3453 and 3443 BC (Helama et al., 2002). The CarrowkeeleKeshcorran record reveals a nearly centurylong interval of warmer and drier conditions between c. 3460e 3370 BC. Lower lake levels were also recorded in west-central Sweden between c. 3400e3250 BC (Gunnarson, 2008) and coincided with increasing summer temperatures in northern Sweden at c. 3300 BC (Grudd et al., 2002). The onset of drier conditions at c. 3410 BC has also been inferred from Meerstalblok in the Netherland (Blaauw et al., 2004). At Lake Constance, a settlement phase associated with house construction between 3384 and

Meso-/Neolithic transition Causewayed enclosure at Magheraboy (c. 4150e3995 BC) Early Neolithic Palynological evidence for arable farming (c. 3950e3700 BC); causewayed enclosure at Donegore (c. 3855e3665 BC; 2s)

Archaeological evidence for cereal cultivation and use (c. 3800e3600 BC); onset of rectangular house building tradition (c. 3800 BC); early megaliths at Carrowmore, Co. Sligo (c. 3750 BC) Middle Neolithic More mobile lifestyle; circular, less substantial houses (Grogan, 2002); declining human activity in NW Ireland

Wooden trackway construction at Corlea, Co. Longford (c. 3580 BC) Archaeological evidence for cereals in eastern Ireland (c. 3500e3450 BC) Local site abandonment in NW Ireland including Céide Fields, Co. Mayo Middle/Late Neolithic Late megalith construction phase at Carrowmore, Co. Sligo (c. 3000 BC); tomb construction in the Boyne Valley, Co. Meath (c. 3345e2900 BC)

Late Neolithic Woodland regeneration, only weak evidence for farming activities in Ireland; end of the initial passage tomb use in the second half of the 29th century BC

3370 BC overlapped with a low stand in the lake level (Magny et al., 2006). The present study identified a particularly wet and cool phase between c. 3360e3290 BC. The bog surface wetness reconstructions from Temple Hill Moss (Langdon et al., 2003) and Walton Moss (Hughes et al., 2000) also indicate a major wet shift at around 3350 BC (Fig. 8). Wetter conditions were also recorded from several sites in Scotland at c. 3300 and 3180 BC (Tipping, 1995; Anderson, 1998) and from Sellevollmyra between c. 3320e3150 BC (Vorren et al., 2007). The reconstruction of winter precipitation in western Norway indicates a shift towards maximum values during the mid-Holocene at c. 3250 BC (Bjune et al., 2005). The midEuropean lake levels showed a maximum stand between c. 3350e3250 BC (Magny, 2004). A rapid rise in lake level was also noted at Lake Constance at c. 3370 BC (Magny and Haas, 2004). A contemporary temperature depression at c. 3250 BC was inferred from tree-line limit and glacier reconstructions in Scandinavia (Karlén and Kuylenstierna, 1996). The summer temperature reconstruction from Sweden shows a drop to c. 1
C below present at c. 3300 BC (Grudd et al., 2002). This is supported by the summer temperature reconstruction from northern Finland, which reveals the coldest 100-year long interval during the fourth millennium BC between 3250 and 3151 BC (Helama et al., 2002). The presence of increased amounts of aeolian material in peat from Scandinavia between c. 3250e3150 BC corroborates that increased storminess accompanied this wet and cool phase (Björck and Clemmensen, 2004; de Jong et al., 2006).

S. Stolze et al. / Quaternary Science Reviews 67 (2013) 138e159

5.4.3. The Late Neolithic (3100e2500 BC) A succession of multi-decadal dry and wet phases is inferred from the CarrowkeeleKeshcorran archives and supported by the Irish evidence. Similar to the inferred drier conditions in the Carrowkeele Keshcorran area between c. 3110e3050 BC, decreased moisture availability facilitated the expansion of pine trees on bog surfaces in northern Scotland between c. 3200e3000 BC (Moir et al., 2010). The pine trees showed two periods of particular high growth rates between 3110 and 3090 BC and between 3040 and 3020 BC (Moir et al., 2010). The combined humification record from three bogs in northern Scotland suggests the establishment of drier/warmer conditions after c. 3050 BC (Anderson et al., 1998). A shift towards a lower water table at c. 3000 BC was inferred from the bog surface moisture reconstruction at Temple Hill Moss (Langdon et al., 2003). A hiatus has been noted in the peat sequence of Meerstalblok between c. 3160e2910 BC, but has been attributed to a major wet erosional event (Blaauw et al., 2004). The temperature reconstructions from Scandinavia reveal a short-lived increase in summer temperatures (Fig. 8). This is reflected in the higher tree line in Sweden and Norway at c. 3150 BC (Karlén and Kuylenstierna, 1996) and in the tree-ring chronology from northern Finland between 3132 and 3123 BC (Helama et al., 2002). The fifth warmest summer during the past 7500 years occurred at 3123 BC (Helama et al., 2002). Although the climate reconstruction from the Carrowkeele Keshcorran area clearly shows a succession of climatic shifts after c. 3000 BC, a lower number of proxy records from northern and western Europe and the Alps registered these changes, confirming the assumption that the amplitude of these climatic oscillations was lower than during the fourth millennium BC. The occurrence of wet and cool conditions recorded in the CarrowkeeleKeshcorran area at c. 2940e2900 BC has also been observed elsewhere in Europe (Fig. 8). Proxy records from Scandinavia indicate increased storminess at c. 2950 BC (Björck and Clemmensen, 2004; de Jong et al., 2006) and at c. 2850 BC (de Jong et al., 2006, Fig. 8). Periods of low summer temperatures occurred between 2990 and 2961 BC and between 2900 and 2801 BC (Helama et al., 2002). Vorren et al. (2007) reported wet shifts from peat sequences in northern Norway between c. 2940e2920 BC and c. 2890e2870 BC. A phase of higher lake levels has also been inferred from mid-European lake-level reconstructions between c. 2900e2850 BC (Magny, 2004). The CarrowkeeleKeshcorran climate reconstruction suggests an interval of drier/warmer conditions between c. 2890e2770 BC. A rise in the altitude of the glacier equilibrium line to higher elevations than present has been reconstructed for two glaciers in western Norway at c. 2850 BC (Bjune et al., 2005). Analysis of subfossil pine wood from lakes and bogs in west-central Sweden indicates that lake levels were lower between c. 2900e2600 BC (Gunnarson, 2008). The wet phase from the CarrowkeeleKeshcorran area between c. 2770e2680 BC is supported by a number of records that point to a climatic deterioration between c. 2850e2650 BC. At Lough Farlary in northern Scotland, dying of pine growing on peat surfaces has been observed between c. 2850e2650 BC (Tipping et al., 2007), suggesting an increase in mire surface wetness. Increased storminess was recorded for southwest Sweden at c. 2725 BC (Björck and Clemmensen, 2004). In west-central Sweden, particularly high winter precipitation was recorded at c. 2750 BC (Bjune et al., 2005). The climate data from the CarrowkeeleKeshcorran area suggest that a dry shift followed this wet/cool period at c. 2600 BC. The establishment of warmer conditions at that time is supported by a concomitant increase in July temperatures in northern Finland (Helama et al., 2002).

155

5.5. Solar forcing on climate between 4000 and 2600 BC The observed synchrony in climatic shifts is in agreement with previous studies comparing the timing of palaeohydrologic shifts between European and global records, suggesting that major changes in temperature and moisture/precipitation were contemporaneous throughout Europe and the world during the Holocene (Charman et al., 2006; Wanner et al., 2011). Previous authors have postulated that major climatic changes during the Holocene were linked to solar variability (Magny, 2004; Charman et al., 2006; Wanner et al., 2011). To test whether the identified multi-decadal climate shifts can be linked to solar variations, the CarrowkeeleKeshcorran data and the reviewed European climate records are compared to solar proxy records. 5.5.1. Cosmogenic radionuclides and solar activity The time-dependent production rates of the cosmogenic radionuclides 14C and 10Be can be used to reconstruct solar activity. Both radionuclides are produced in the Earth’s atmosphere in cascades of nuclear reactions initiated by high-energy cosmic ray particles. The intensity of cosmic rays reaching the Earth’s atmosphere is modulated by solar activity. A high solar activity causes a higher diversion of cosmic rays, leading to a lower production of the two radionuclides (Lal and Peters, 1967). In the present study, the time-dependent 14C production rate based on tree-ring D14C data is used (Muscheler et al., 2005). The most detailed reconstruction of the variations in the 10Be production rate for the Holocene has been developed from the GRIP ice core (Vonmoos et al., 2006). It has been shown that the 14C and 10Be production rates are in good agreement despite the complexity and partly different pathways of formation (Knudsen et al., 2009). Thus, both cosmogenic radionuclide records can be used as independent measures of solar activity. 5.5.2. Solar output and climate variability The timing of the recorded climate shifts in northern and western Europe and the Alps is in good agreement with changes in the production rates of the solar proxies 14C and 10Be, indicating that the climate was largely influenced by variations in solar irradiance. Periods of increased summer temperatures and lower rainfall overlap with declining and low recordings of the production rates of both 14C and 10Be, indicating an increasing and higher solar activity. Lower summer temperatures, increased precipitation and storminess coincide with intervals of higher radionuclide production rates (Table 3, Figs. 7 and 8). It is remarkable that the succession of wet and dry shifts across Europe mirrors the sequence of low and high solar activity during the studied time interval (Figs. 8 and 9). The particularly wet conditions prevailing in northern and western Europe at c. 3950 BC coincided with increased amounts of ice-rafted debris in a marine sediment core off the west coast of Ireland (Bond et al., 1997). Based on the 14C and 10Be production rates, which reflect a low solar activity between 4015 and 3940 BC (Fig. 9), this event has been linked to perturbations in the Sun’s energy output (Bond et al., 2001). The subsequent warming/drying phase across northern and western Europe and the Alps between c. 3950 and 3850 BC overlaps with a global decline in the number of cold events (Wanner et al., 2011) and an increase in solar activity between 3940 and 3855 BC (Table 3). The established warm and dry conditions between 3850 and 3700 BC overlap with a low number of cold events at the global scale (Fig. 8; Wanner et al., 2011). This phase is marked by a generally high solar activity (Fig. 9). The 14C and 10Be production rates suggest small-scale oscillations in the solar activity during this time, possibly connected to a short-lived wet shift at Templevanny Lough and elsewhere in Europe at c. 3850 (Table 3, Fig. 9).

156

S. Stolze et al. / Quaternary Science Reviews 67 (2013) 138e159

Fig. 9. Precipitation and temperature reconstruction from the CarrowkeeleKeshcorran area between 4000 and 2600 BC compared to variations in the production rate of the cosmogenic isotopes 14C and 10Be and periods of human development and archaeological presence in Ireland. Grey shading denotes periods of warmer and drier conditions inferred from the CarrowkeeleKeshcorran archives. References: (1) Muscheler et al. (2005); (2) Vonmoos et al. (2006); (3) references in Stolze et al. (2012); (4) Turney et al. (2006); (5) Sheridan (1995).

The termination of the warm and dry interval coincided largely with a steep decline in solar activity between 3680 and 3630 BC (Table 3, Fig. 8). The following period prevailing through 3250 BC was characterised by multi-decadal climatic oscillations coinciding with frequent and large fluctuations in solar activity. These solar oscillations showed the highest amplitude during the studied period, resembling the solar minimum pattern during the Little Ice Age. The wet shift at c. 3550 BC overlapping with a global increase in cold events at the time (Wanner et al., 2011) can be linked to a drop in solar activity starting at 3550 BC (Fig. 9). The particularly wet and cool conditions observed between c. 3350e3250 BC coincided with a notable increase in global cold events (Fig. 8; Wanner et al., 2011). These conditions occurred during an interval of reduced solar activity that marked the longest period of low solar activity and possibly most unfavourable climatic conditions between 4000 and 2600 BC.

A period of warmer/drier conditions followed between c. 3250e 2950 BC. Despite smaller short-lived fluctuations, the solar proxy data indicate that the solar output was generally high, in particular at 3210, 3150, between 3060 and 3040 and between 2980 and 2960 BC (Table 3). The summer temperature reconstruction from northern Finland indicates that the fifth warmest summer during the past 7500 years occurred at 3123 BC (Helama et al., 2002). The climate deteriorated dramatically between c. 2950e2850 BC, as reflected in heavier rainfall and lower summer temperatures. This interval falls closely within a period of an increasing number of globally registered cold events (Fig. 8; Wanner et al., 2011) and declining solar irradiance between 2950 and 2830 BC. Based on the observation that the reviewed proxy records show a less clear pattern of climatic fluctuations after c. 2850 BC, it is suggested that climate oscillations must have become less intense. Variations in

S. Stolze et al. / Quaternary Science Reviews 67 (2013) 138e159

solar activity became less pronounced during the same time, possibly hampering the identification of simultaneous changes in both climate and solar activity. The wet shift at c. 2770 BC observed in the CarrowkeeleKeshcorran archives overlaps with a low in solar activity at 2750 BC, while the onset of warmer/drier conditions at c. 2600 BC may reflect the marked increase in solar activity at 2565 BC. 5.5.3. Solar cyclicity between 4000 and 2600 BC The high-frequency climatic variability in northern and western Europe and the Alps recorded between 4000 and 2600 BC overlapped with a period of large fluctuations in solar activity. Spectral analysis of the Holocene D14C and 10Be records suggests that solar cycles were particularly pronounced between 4050 and 2550 BC, with periodicities of 88, 150, 220 and 400 years (Knudsen et al., 2009). Variations of solar activity during the Holocene were only equally marked between 1050 and 50 BC (Knudsen et al., 2009). The reconstruction of sunspot numbers since 9500 BC highlights the occurrence of five Maunder-type minima between 4000 and 2600 BC, representing 18% of all grand minima over the last c. 11,000 years (Usoskin et al., 2007). 5.6. Implications for human development during the Irish Neolithic Review of the archaeological and palynological evidence for human activity during the Irish Neolithic indicates that variations in the intensity of human activity were largely contemporaneous with climatic changes and fluctuations in solar activity (Table 3, Fig. 9). The introduction of Neolithic practices and traditions occurred during the first centuries of the fourth millennium BC (Cooney et al., 2011). Palynological and archaeological evidence for cereal cultivation and processing was recorded between c. 3950e3600 BC (cf. Stolze et al., 2012) and overlapped, within the dating uncertainties, with a period of ameliorated climatic conditions and increasing/increased solar activity between 3940 and 3680 BC (Table 3). The establishment of the rectangular field system of the Céide Fields (O’Connell and Molloy, 2001), the construction and use of rectangular timber buildings and the erection of megalithic monuments also fell within this period of favourable climatic conditions (cf. Cooney et al., 2011). The settlement probability established by Turney et al. (2006) corroborates that pronounced human activity prevailed during the Early Neolithic (Fig. 8). At c. 3600 BC, a change in Neolithic traditions and practices and a decline in human impact have been inferred from palynological and archaeological archives across Ireland (Stolze et al., 2012). These changes overlapped with a period of notable climatic variability and oscillations in the solar output (Table 3, Figs. 8 and 9). A shift towards pastoral farming and the end of the rectangular house-building tradition coincided with the establishment of wetter and somewhat cooler conditions and the decline in solar activity at the transition from the Early to Middle Neolithic (Stolze et al., 2012 and references therein). A rise in the water table resulted in the construction of a wooden trackway across a boggy area at Corlea in County Longford at 3580 BC (Caseldine and Hatton, 1996). Pollen records suggest that the decline in human impact was more pronounced in northwest Ireland than in the northeastern part of the island (Stolze et al., 2012). Areas in the northwest including the Céide Fields were locally abandoned, in particular following the wet and cool period between c. 3350e3250 BC (e.g., O’Connell and Molloy, 2001; Ghilardi and O’Connell, 2012), which was likely linked to the century-long interval of low solar activity. At Carrowmore in County Sligo, the most intense use of megalithic monuments appears to have ended at that time (Bergh and Hensey, in press).

157

In the northeastern part of the island, arable farming revived at around 3400 BC and after 3250 BC (Eogan, 1963; Waddell, 2000; Plunkett et al., 2008). This coincided with increased solar activity and warmer and drier climatic conditions. The timing of the warmer and drier period after 3250 BC overlaps with passage grave construction in the eastern part of the island including the Boyne Valley (Cooney, 2000; Cooney et al., 2011). The observation that previously inhabited areas in northwestern Ireland showed only low human interference during these periods of climatic amelioration suggests that the population density must have been low during this time. 6. Conclusions The present study illustrates that the complementary analysis of biological and geochemical proxies at high-resolution permits the reconstruction of vegetational and environmental changes that took place on decadal to centennial time scales. The comprehensive analysis of multiple independent proxies provides a powerful tool to identify variations in single proxies that could be otherwise interpreted as analytical artefacts. Bayesian agedepth modelling using radiocarbon dates obtained on closelyspaced samples produces posterior probability distributions that are comparably tight, even if closely neighbouring dates fall in plateaus of the calibration curve. The present study shows that climatic shifts inferred from the CarrowkeeleKeshcorran area between 4100 and 2600 BC were synchronous with similar changes throughout northern and western Europe and the Alps, suggesting supra-regional changes in climate during this time. Based on the correlation with Holocene variations in solar activity inferred from cosmogenic radionuclide records, it is demonstrated that the reconstructed climate changes largely coincide with variations in the solar activity. Comparison of the inferred climatic shifts with the Neolithic archaeological record of Ireland suggests that the climate represented an important control on the early human development during the fourth millennium BC. Early arable farming and a more sedentary lifestyle coincided with climatic amelioration and increased solar activity. In contrast, declining human activity and a shift towards a more mobile lifestyle occurred during periods of climate deterioration and lower solar activity in particular at c. 3350 BC. The occurrence of rapid and unusually pronounced shifts in climate conditions in response to variations in solar activity makes the Neolithic a unique period in human development. Acknowledgements We thank M. O’Connell, I. Feeser and P. O’Rafferty for assistance during coring at Templevanny Lough. T. Monecke is thanked for field assistance and constructive discussions. XRF analysis was kindly supported by R. Tjallingii, E. Mollier-Vogel, V. Rohde-Krossa and D. Garbe-Schönberg. We acknowledge P. Grootes and M.-J. Nadeau for discussion of the radiocarbon dates. S. Bergh is thanked for helpful comments on Irish archaeology. F. McDermott kindly provided the d18O speleothem data from Crag Cave. The manuscript benefited from the critical review by two anonymous referees. This study was financially supported by the German Research Foundation (NE 970/2-1, 2-2). S. Stolze acknowledges a research award from the Canadian Association of Palynologists. References Adrian, R., O’Reilly, C.M., Zagarese, H., Baines, S.B., Hessen, D.O., Keller, W., Livingstone, D.M., Sommaruga, R., Straile, D., Van Donk, E., Weyhenmeyer, G.A.,

158

S. Stolze et al. / Quaternary Science Reviews 67 (2013) 138e159

Winder, M., 2009. Lakes as sentinels of climate change. Limnol. Oceanogr. 54, 2283e2297. Ahlberg, K., Almgren, E., Wright Jr., H.E., Ito, E., 2001. Holocene stable-isotope stratigraphy at Lough Gur, County Limerick, Western Ireland. Holocene 11, 367e372. Anderson, D.E., 1998. A reconstruction of Holocene climatic changes from peat bogs in northewest Scotland. Boreas 27, 208e224. Anderson, D.E., Binney, H.A., Smith, M.A., 1998. Evidence for abrupt climatic change in northern Scotland between 3900 and 3500 calendar years BP. Holocene 8, 97e103. Bergh, S., Hensey, R. Unpicking the chronology of Carrowmore. Oxford J. Archaeol., in press. Beug, H.-J., 2004. Leitfaden der Pollenbestimmung für Mitteleuropa und angrenzende Gebiete. Pfeil, München. Björck, S., Clemmensen, L.B., 2004. Aeolian sediment in raised bog deposits, Halland, SW Sweden: a new proxy record of Holocene winter storminess variation in southern Scandinavia? Holocene 14, 677e688. Bjune, A.E., Bakke, J., Nesje, A., Birks, H.J.B., 2005. Holocene mean July temperature and winter precipitation in Western Norway inferred from palynological and glaciological lake-sediment proxies. Holocene 15, 177e189. Blaauw, M., van Geel, B., van der Plicht, J., 2004. Solar forcing of climatic change during the mid-Holocene: indications from raised bogs in the Netherlands. Holocene 14, 35e44. Blenckner, T., Adrian, R., Livingstone, D.M., Jennings, E., Weyhenmeyer, G.A., George, D.G., Jankowski, T., Järvinen, M., Aonghusa, C.N., Nõges, T., Straile, D., Teubner, K., 2007. Large-scale climatic signatures in lakes across Europe: a meta-analysis. Glob. Change Biol. 13, 1314e1326. Bond, G., Showers, W., Cheseby, M., Lotti, R., Almasi, P., deMenocal, P., Priore, P., Cullen, H., Hajdas, I., Bonani, G., 1997. Pervasive millennial-scale cycle in North Atlantic Holocene and glacial climates. Science 278, 1257e1266. Bond, G., Kromer, B., Beer, J., Muscheler, R., Evans, M.N., Showers, W., Hoffmann, S., Lotti-Bond, R., Hajdas, I., Bonani, G., 2001. Persistent solar influence on North Atlantic climate during the Holocene. Science 294, 2130e2136. Briffa, K.R., Atkinson, T., 1997. Reconstructing Late-glacial and Holocene climates. In: Hulme, M., Barrow, E. (Eds.), Climates of the British Isles: Present, Past and Future. Routledge, London, pp. 84e111. Bronk Ramsey, C., 1995. Radiocarbon calibration and analysis of stratigraphy: the OxCal program. Radiocarbon 37, 425e430. Bronk Ramsey, C., 2001. Development of the radiocarbon calibration program. Radiocarbon 43, 355e363. Bronk Ramsey, C., 2008. Deposition models for chronological records. Quat. Sci. Rev. 27, 42e60. Bronk Ramsey, C., 2009a. Bayesian analysis of radiocarbon dates. Radiocarbon 51, 337e360. Bronk Ramsey, C., 2009b. Dealing with outliers and offsets in radiocarbon dating. Radiocarbon 51, 1023e1045. Brown, T.A., Farwell, G.W., Grootes, P.M., Schmidt, F.H., 1992. Radiocarbon AMS dating of pollen extracted from peat samples. Radiocarbon 34, 550e556. Brown, A.G., Hatton, J., O’Brien, C.E., Selby, K.A., Langdon, P.G., Stuijts, I., Caseldine, C.J., 2005. Vegetation, landscape and human activity in Midland Ireland: mire and lake records from the Lough Kinale-Derragh Lough area, Central Ireland. Veg. Hist. Archaeobot. 14, 81e98. Büntgen, U., Tegel, W., Nicolussi, K., McCormick, M., Frank, D., Trouet, V., Kaplan, J.O., Herzig, F., Heussner, K.-U., Wanner, H., Luterbacher, J., Esper, J., 2011. 2500 years of European climate variability and human susceptibility. Science 331, 578e582. Caseldine, C., Hatton, J., 1996. Early land clearance and wooden trackway construction in the third and fourth millennia BC at Corlea, Co. Longford. Biol. Environ. 96B, 11e19. Caseldine, C.J., Thompson, G., Langdon, C., Hendon, D., 2005. Evidence for an extreme climatic event on Achill Island, Co. Mayo, Ireland around 5200e 5100 cal. yr BP. J. Quaternary Sci. 20, 169e178. Caulfield, S., O’Donnell, R.G., Mitchell, P.I., 1998. 14C dating of a Neolithic field system at Céide Fields, County Mayo, Ireland. Radiocarbon 2, 629e640. Cayless, S.M., Tipping, R.M., 2002. Data on mid-Holocene climatic, vegetation and anthropogenic interactions at Stanshiel Rig, southern Scotland. Veg. Hist. Archaeobot. 11, 201e210. Charman, D.J., Blundell, A., Chiverrell, R.C., Hendon, D., Langdon, P.G., 2006. Compilation of non-annually resolved Holocene proxy climate records: stacked Holocene peatland palaeo-water table reconstructions from northern Britain. Quat. Sci. Rev. 25, 336e350. Cooney, G., 2000. Landscapes of Neolithic Ireland. Routledge, London. Cooney, G., 2007. Parallel worlds or multi-stranded identities? Considering the process of ‘going over’ in Ireland and the Irish Sea zone. Proc. Brit. Acad. 144, 543e566. Cooney, G., Bayliss, A., Healy, F., Whittle, A., Danaher, E., Cagney, L., Mallory, J., Smyth, J., Kador, T., O’Sullivan, M., 2011. Chapter 12: Ireland. In: Whittle, A., Healy, F., Bayliss, A. (Eds.), Gathering Time: Dating the Early Neolithic Enclosures of Southern Britain and Ireland. Oxbow, Oxford, pp. 562e669. Danaher, E., 2007. Monumental Beginnings: the Archaeology of the N4 Sligo Inner Relief Road. In: NRA Scheme Monographs 1. Wordwell, Dublin. Davis, M.B., 2000. Palynology after Y2K e understanding the source area of pollen in sediments. Annu. Rev. Earth Planet. Sci. 28, 1e18. de Jong, R., Svante, B., Björckman, L., Clemmensen, L.B., 2006. Storminess variation during the last 6500 years as reconstructed from an ombrotrophic peat bog in Halland, southwest Sweden. J. Quaternary Sci. 21, 905e919.

Dubois, A.D., Ferguson, D.K., 1985. The climatic history of pine in the Cairngorms based on radiocarbon dates and stable isotope analysis, with an account of the events leading up to its colonization. Rev. Palaeobot. Palynol. 46, 55e80. Eogan, G., 1963. A Neolithic habitation site and megalithic tomb in Townleyhall Townland, Co. Louth. J. Roy. Soc. Antiqu. Scotl. 93, 37e81. Faegri, K., Iversen, J., 1989. In: Faegri, K., Kaland, P.E., Krzywinski, K. (Eds.), Textbook of Pollen Analysis, fourth ed. John Wiley & Sons, New York. George, G., Jennings, E., Allott, N., 2010. The impact of climate change on lakes in Britain and Ireland. In: George, D. (Ed.), The Impact of Climate Change on European Lakes. Aquatic Ecology Series 4. Springer, pp. 359e386. Ghilardi, B., O’Connell, M., 2012. Fine-resolution pollen-analytical study of Holocene woodland dynamics and land use in North Sligo, Ireland. Boreas. http:// dx.doi.org/10.1111/j.1502-3885.2012.00292.x. Grimm, E.C., 2004. Tilia Graph v. 2.0.2, Software. Illinois State Museum, Research and Collection Centers, Springfield. Grogan, E., 2002. Neolithic houses in Ireland: a broader perspective. Antiquity 76, 517e525. Grudd, H., Briffa, K.R., Karlén, W., Bartholin, T.S., Jones, P.D., Kromer, B., 2002. A 7400-year tree-ring chronology in northern Swedish Lapland: natural climatic variability expressed on annual to millennial timescales. Holocene 12, 657e665. Gunnarson, E.B., 2008. Temporal distribution pattern of subfossil pines in central Sweden: perspective on Holocene humidity fluctuations. Holocene 18, 569e577. Hammer, Ø., Harper, D.A.T., Ryan, P.D., 2001. PAST: paleontological statistics software package for education and data analysis. Palaeontol. Electron. 4, 9. Heiri, O., Lotter, A.F., Lemcke, G., 2001. Loss on ignition as a method for estimating organic and carbonate content in sediments: reproducibility and comparability of results. J. Paleolimnol. 25, 101e110. Helama, S., Lindholm, M., Timonen, M., Meriläinen, J., Eronen, M., 2002. The supralong Scots pine tree-ring record for Finnish Lapland: part 2, interannual to centennial variability in summer temperatures for 7500 years. Holocene 12, 681e687. Hellman, S., Gaillard, M.-J., Bunting, J.M., Mazier, F., 2009. Estimating the relevant source area of pollen in the past cultural landscapes of southern Sweden e a forward modelling approach. Rev. Palaeobot. Palyno. 153, 259e271. Heslop-Harrison, Y., 1955a. Nuphar Sm. J. Ecol. 43, 342e364. Heslop-Harrison, Y., 1955b. Nymphaea L. em. Sm. (nom. conserve.). J. Ecol. 43, 719e 734. Hirons, K.R., Edwards, K.J., 1986. Events at and around the first and second Ulmus declines: palaeoecological investigations in Co. Tyrone, Northern Ireland. New Phytol. 104, 131e153. Holmes, J., Jones, R., Haas, J.N., McDermott, F., Molloy, K., O’Connell, M., 2007. Multiproxy evidence for Holocene lake-level and salinity changes at a Loch Mór, a coastal lake on the Aran Islands, Western Ireland. Quat. Sci. Rev. 26, 2438e 2462. Hughes, P.D.M., Mauquoy, D., Barber, K.E., Langdon, P.G., 2000. Mire-development pathways and palaeoclimatic records from a full Holocene peat archive at Walton Moss, Cumbria, England. Holocene 10, 465e479. Hulme, M., Barrow, E., 1997. Climates of the British Isles. Routledge, London. Hurrell, J.W., Kushnir, Y., Ottersen, G., Visbeck, M., 2003. An overview of the North Atlantic Oscillation. In: Hurrell, J.W., Kushnir, Y., Ottersen, G., Visbeck, M. (Eds.), The North Atlantic Oscillation: Climatic Significance and Environmental Impact. Geophys. Monogr. 134. AGU, Washington, pp. 1e35. Janssen, C.R., 1986. The use of local pollen indicators and of the contrast between regional and local pollen values in the assessment of the human impact on vegetation. In: Behre, K.-E. (Ed.), Anthropogenic Indicators in Pollen Diagrams. Balkema, Rotterdam, pp. 203e208. Joerin, U.E., Nicolussi, K., Fischer, A., Stocker, T.F., Schlüchter, C., 2008. Holocene optimum events inferred from subglacial sediments at Tschierva Glacier, Eastern Swiss Alps. Quat. Sci. Rev. 27, 337e350. Karlén, W., Kuylenstierna, J., 1996. On solar forcing of Holocene climate: evidence from Scandinavia. Holocene 6, 359e365. Kilian, M.R., van der Plicht, J., van Geel, B., Goslar, T., 2002. Problematic 14C-AMS dates of pollen concentrates from Lake Gosciaz (Poland). Quat. Int. 88, 21e26. Knudsen, M.F., Riisager, P., Jacobsen, B.H., Muscheler, R., Snowball, I., Seidenkrantz, M.-S., 2009. Taking the pulse of the Sun during the Holocene by joint analysis of 14C and 10Be. Geophys. Res. Lett. 36, L16701. Komárek, J., Jankovská, V., 2001. Review of the Green Algal Genus Pediastrum; Implication for Pollen Analytical Research. Bibliotheca Phycologica 108. Schweizerbart, Stuttgart. Lal, D., Peters, B., 1967. Cosmic ray produced radioactivity on the Earth. In: Flügge, S. (Ed.), Handbuch der Physik, vol. 46/2. Springer, Berlin, pp. 551e612. Langdon, P.G., Barber, K.E., Hughes, P.D.M., 2003. A 7500-year peat-based palaeoclimatic reconstruction and evidence for an 1100-year cyclicity in bog surface wetness from Temple Hill Moss, Pentland Hills, southeast Scotland. Quat. Sci. Rev. 22, 259e274. Langdon, P.G., Brown, A.G., Caseldine, C.J., Blockley, S.P.E., Stuijts, I., 2012. Regional climate change from peat stratigraphy for the mid- to late Holocene in central Ireland. Quat. Int. 268, 145e155. Livingstone, D.M., 2008. A change of climate provokes a change of paradigm: taking leave from two tacit assumptions about physical lake forcing. Int. Rev. Hydrobiol. 93, 404e414. Livingstone, D.M., Adrian, R., Arvola, L., Blenckner, T., Dokulil, M.T., Hari, R.E., George, G., Jankowski, T., Järvinen, Jennings, E., Nõges, P., Nõges, T., Straile, D., Weyhenmaeyer, G.A., 2010. Regional and supra-regional coherence in

S. Stolze et al. / Quaternary Science Reviews 67 (2013) 138e159 limnological variables. In: George, D. (Ed.), The Impact of Climate Change on European Lakes. Aquatic Ecology Series 4. Springer, pp. 311e337. MacDermot, C.V., Long, C.B., Harney, S.J., 1996. A Geological Description of Sligo, Leitrim, and Adjoining Parts of Cavan, Fermanagh, Mayo and Roscommon, to Accompany the Bedrock Geology 1:100,000 Scale Map Series, Sheet 7, SligoLeitrim, with Contributions by Claringbold, K., Daly, D., Meehan, R. and Stanley, G. Geological Survey of Ireland, Dublin. Magny, M., 2004. Holocene climate variability as reflected by mid-European lakelevel fluctuations and its probable impact on prehistoric human settlements. Quat. Int. 113, 65e79. Magny, M., Haas, J.N., 2004. A major widespread climatic change around 5300 cal. yr BP at the time of the Alpine Iceman. J. Quat. Sci. 19, 423e430. Magny, M., Peyron, O., Bégeot, C., Guiot, J., 2005. Quantitative reconstruction of midHolocene climatic variations in the northern Alpine foreland based on Lake Morat (Swiss Plateau) and Lake Annecy (French Pre-Alps) data. Boreas 34, 434e444. Magny, M., Leuzinger, U., Bortenschlager, S., Haas, J.N., 2006. Tripartite climate reversal in Central Europe 5600e5300 years ago. Quat Res. 65, 3e19. Mauquoy, D., van Geel, B., Blaauw, M., van der Plicht, J., 2002. Evidence from northwest European bogs shows ‘Little Ice Age’ climatic changes driven by variations in solar activity. Holocene 12, 1e6. Mauquoy, D., Yeloff, D., van Geel, B., Charman, D.J., Blundell, A., 2008. Two decadally resolved records from northewest European peat bogs show rapid climate changes associated with solar variability during the mid-late Holocene. J. Quaternary Sci. 23, 745e763. Mayewski, P.A., Rohling, E.E., Stager, J.C., Karlén, W., Maasch, K.A., Meeker, L.D., Meyerson, E.A., Gasse, F., van Kreveld, S., Holmgrend, K., Lee-Thorp, J., Rosqvist, G., Rack, F., Staubwasser, M., Schneider, R.R., Steig, E.J., 2004. Holocene climate variability. Quat. Res. 62, 243e255. McDermott, F., Mattey, D.P., Hawkesworth, C., 2001. Centennial-scale Holocene climate variability revealed by a high-resolution speleothem d18O record from SW Ireland. Science 294, 1328e1331. McGloin, A., Moore, S., 1996. A Guide to the Carrowkeel/Keshcorran Megalithic Complex. Arrow Community Enterprise Ltd, Castlebaldwin. Met Éireann, 2012. http://www.met.ie/(accessed 01.03.12.). Metcalfe, D.J., 2005. Hedera helix L. J. Ecol. 93, 632e648. Mingram, J., Negendank, J.F.W., Brauer, A., Berger, D., Hendrich, A., Köhler, M., Usinger, H., 2007. Long cores from small lakes-recovering up to 100 m-long lake sediment sequences with a high-precision rod-operated piston corer (Usingercorer). J. Paleolimnol. 37, 517e528. Moir, A.K., Leroy, S.A.G., Brown, D., Collins, P.E.F., 2010. Dendro-chronological evidence for a lower water-table on peatland around 3200e3000 BC from subfossil pine in northern Scotland. Holocene 20, 931e942. Molloy, K., O’Connell, M., 1987. The nature of the vegetational changes at about 5000 BP with particular reference to the elm decline: fresh evidence from Connemara, western Ireland. New Phytol. 106, 203e220. Molloy, K., O’Connell, M., 1995. Palaeoecological investigations towards the reconstruction of environment and land-use changes during prehistory at Céide Fields, western Ireland. Probl. Küstenforsch. Südl. Nordseegeb. 23, 187e225. Moore, P.D., Webb, J.A., Collinson, M.E., 1991. Pollen Analysis. Blackwell, Oxford. Murnaghan, S., Taylor, D., Jennings, E., Dalton, C., Olaya-Bosch, K., O’Dwyer, B., 2012. Middle to late Holocene environmental changes in Western Ireland inferred from fluctuations in preservation of biological variables in lake sediment. J. Paleolimnol. 48, 433e448. Muscheler, R., Beer, J., Kubik, P.W., Synal, H.-A., 2005. Geomagnetic field intensity during the last 60,000 years based on 10Be and 36Cl from the Summit ice cores and 14C. Quat. Sci. Rev. 24, 1849e1860. Nadeau, M.-J., Schleicher, M., Grootes, P.M., Erlenkeuser, H., Gottdang, A., Mous, D.J.W., Sarnthein, J.M., Willkomm, H., 1997. The Leibniz-Labor AMS facility at the Christian-Albrechts University, Kiel, Germany. Nucl. Instrum. Methods B123, 22e30. O’Connell, M., Molloy, K., 2001. Farming and woodland dynamics in Ireland during the Neolithic. Biol. Environ. 101B, 99e128. Odgaard, B.V., 1999. Fossil pollen as a record of past biodiversity. J. Biogeogr. 26, 7e17. Pals, J.P., van Geel, B., Delfos, A., 1980. Paleoecological studies in the Klokkeweel bog near Hoogkarspel (prov. of Noord-Holland). Rev. Palaeobot. Palyno. 30, 371e418. Parker, A.G., Goudie, A.S., Anderson, D.E., Robinson, M.A., Bonsall, C., 2002. A review of the mid-Holocene elm decline in the British Isles. Prog. Phys. Geog. 26, 1e45. Peglar, S.M., Birks, H.J.B., 1993. The mid-Holocene Ulmus fall at Diss Mere, Southe East England e disease and human impact? Veg. Hist. Archaeobot. 2, 61e68. Plunkett, G., Carroll, F., Hartwell, B., Whitehouse, N.J., Reimer, P.J., 2008. Vegetation history at the multi-period prehistoric complex at Ballynahatty, Co. Down, Northern Ireland. J. Archaeol. Sci. 35, 181e190.

159

Reimer, P.J., Baillie, M.G.L., Bard, E., Bayliss, A., Beck, J.W., Blackwell, P.G., Bronk Ramsey, C., Buck, C.E., Burr, G.S., Edwards, R.L., Friedrich, M., Grootes, P.M., Guilderson, T.P., Hajdas, I., Heaton, T.J., Hogg, A.G., Hughen, K.A., Kaiser, K.F., Kromer, B., McCormac, F.G., Manning, S.W., Reimer, R.W., Richards, D.A., Southon, J.R., Talamo, S., Turney, C.S.M., van der Plicht, J., Weyhenmeyer, C.E., 2009. IntCal09 and Marine09 radiocarbon age calibration curves, 0e50,000 years cal BP. Radiocarbon 51, 1111e1150. Schmidt, R., Kamenik, C., Tessadri, R., Koinig, K.A., 2002. Climatic changes from 12,000 to 4,000 years ago in the Austrian Central Alps tracked by sedimentological and biological proxies of a lake sediment core. J. Paleolimnol. 35, 491e 505. Sheridan, A., 1995. Irish Neolithic pottery: the story in 1995. In: Kinnes, I.A., Varndell, G. (Eds.), Unbaked Urns of Rudely Shape. Oxbow Monographs 55, Oxford, pp. 3e22. Speranza, A., van Geel, B., van der Plicht, J., 2002. Evidence for solar forcing of climate change at ca. 850 calBC from a Czech peat sequence. Glob. Planet. Change 35, 51e65. Stockmarr, J., 1971. Tablets with spores used in absolute pollen analysis. Pollen Spores 13, 615e621. Stolze, S., 2012a. Human Impact and Environmental Changes During the Neolithic in the CarrowkeeleKeshcorran Area, County Sligo, Ireland. Ph.D. thesis, Christian-Albrechts University, Germany. Stolze, S., 2012b. Environmental change and human impact during the Neolithic in the CarrowkeeleKeshcorran area: evidence from Lough Availe. In: Stefanini, B.S., McGlynn, G. (Eds.), Roscommon and South Sligo. IQUA Field Guide No. 30. Irish Quaternary Association, Dublin, pp. 83e90. Stolze, S., Dörfler, W., Monecke, T., Nelle, O., 2012. Evidence for climatic variability and its impact on human development during the Neolithic from Loughmeenaghan, County Sligo, Ireland. J. Quat. Sci. 27, 393e403. Stuiver, M., Polach, H.A., 1977. Discussion: reporting of 14C data. Radiocarbon 19, 355e363. Tipping, R., 1995. Holocene evolution of a lowland Scottish landscape: Kirkpatrick Fleming. Part I, peat- and pollen-stratigraphic evidence for raised moss development and climatic change. Holocene 5, 69e81. Tipping, R., 2010. The case for climatic stress forcing choice in the adoption of agriculture in the British Isles. In: Finlayson, B., Warren, G. (Eds.), Landscapes in Transition. Oxbow, Oxford, pp. 66e76. Tipping, R., Ashmore, P., Davies, A., Haggart, A., Moir, A., Newton, A., Sands, R., Skinner, T., Tisdall, E., 2007. Peat, pine stumps and people: interactions behind climate, vegetation change and human activity in wetland archaeology at Loch Farlary, northern Scotland. In: Barber, J., Clark, C., Cressey, M., Crone, A., Hale, A., Henderson, J., Housley, R., Sands, R., Sheridan, A. (Eds.), Archaeology from the Wetlands: Recent Perspectives: Proceedings of the 11th WARP Conference. Society of Antiquaries of Scotland, Edinburgh, pp. 157e164. Turney, C.S.M., Baillie, M., Palmer, J., Brown, D., 2006. Holocene climatic change and past Irish societal response. J. Archaeol. Sci. 33, 34e38. Usoskin, I.G., Solanki, S.K., Kovaltsov, G.A., 2007. Grand minima and maxima of solar activity: new observational constraint. Astron. Astrophys. 471, 301e309. van Geel, B., Coope, G.R., van der Hammen, T., 1989. Palaeoecology and stratigraphy of the Lateglacial section at Usselo (The Netherlands). Rev. Palaeobot. Palyno. 60, 25e129. Verrill, L., Tipping, R., 2010. Use and abandonment of a Neolithic field system at Belderrig, Co. Mayo, Ireland: evidence for economic marginality. Holocene 20, 1011e1021. Vonmoos, M., Beer, J., Muscheler, R., 2006. Large variations in Holocene solar activity: constraints from 10Be in the Greenland Ice Core Project ice core. J. Geophys. Res. 111, A10105. Vorren, K.-D., Blaauw, M., Wastegård, S., van der Plicht, J., Jensen, C., 2007. Highresolution stratigraphy of the northernmost concentric raised bog in Europe: Sellevollmyra, Andøya, northern Norway. Boreas 36, 253e277. Vorren, K.-D., Jenxen, C.E., Nilssen, E., 2012. Climate changes during the last c. 7500 years as recorded by the degree of peat humification in the Lofoten region, Norway. Boreas 41, 13e30. Waddell, J., 2000. The Prehistoric Archaeology of Ireland. Wordwell, Bray. Wanner, H., Solomina, O., Grosjean, M., Ritz, S.P., Jetel, M., 2011. Structure and origin of Holocene cold events. Quat. Sci. Rev. 30, 3109e3123. Webb, D.A., Parnell, J., Doogue, D., 1996. An Irish Flora. Dundalgan Press (W. Tempest) Ltd, Dundalk. Wetzel, R.G., 2001. Limnology: Lake and River Ecosystems. Academic Press, London. Wilkinson, R.E., 1963. Effects of light intensity and temperature on the growth of waterstargrass, coontail, and duckweed. Weeds 11, 287e290. Woolings, T., 2010. Dynamical influences on European climate: an uncertain future. Phil. Trans. R. Soc. A 368, 3733e3756.