Quaternary Science Reviews 21 (2002) 89–101 New data for the Last Glacial Maximum in Great Britain and Ireland D.Q. Bowena,*, F.M. Phillipsb, A.M. McCabec, P.C. Knutza, G.A. Sykesa a Department of Earth Sciences, University of Cardiff, Cardiff, Wales CF10 3YE, UK b Department of Earth and Environmental Sciences, NewMexico Tech, Socorro, NM 87801, USA c School of Environmental Sciences, University of Ulster, Coleraine BT52 1SA, Northern Ireland Received 15 March 2001; accepted 28 August 2001 Abstract Understanding the history of the British and Irish Ice Sheet (BIIS) at the Last Glacial Maximum (LGM) has been advanced by new approaches, in particular, by cosmogenic nuclide surface-exposure dating, aminostratigraphy of ‘shelly’ glacial deposits, AMS radiocarbon dating, and the evidence from continental margin marine cores, all of which supersede the previously weak geochronologic control. It was formerly believed that Great Britain and Ireland was largely ice free between the last interglacial (oxygen isotope sub-stage 5e) and the Late Devensian, when the LGMoccurred. As such the BIIS was effectively out of phase with Laurentide and Scandinavian ice sheets, as well as inferences of ice volume from oxygen isotope stratigraphy. The BIIS during the Late Devensian maximum was also perceived as having been relatively stationary. New evidence shows that the LGMwas an important event during the evolution of an earlier BIIS when the extent of ice was greater. Repeated iceberg rafting events over the past 50 ka are shown by marine cores, while the derivative inference of numerous corresponding glacial advances is supported by several clusters of 36Cl ages on glaciated surfaces and glacial boulders, that are indicative of deglacial events between 40 and 12 ka. These appear to be associated with Heinrich events, the earliest being inferred as Heinrich 4 at about 40 ka. During this advance, the BIIS and Scandinavian Ice Sheet (SIS) were in contact and all of Ireland was glaciated. The ice sheet appears to have fluctuated several times between 40 and 25 ka, although evidence for this is poorly preserved. But the 36Cl and 14C evidence is clear that the BIIS reached its LGMmaximum size about 22 ka, soon after Heinrich Event 2, when the BIIS and SIS were not in contact. One cluster of 36Cl and 14C ages, at 21.471.3 ka, records an initial pulse of deglaciation that was followed by extensive deglaciation about 17.470.4 ka just before Heinrich Event 1, when the ice sheet readvanced. Contrary to previous views, the BIIS probably existed throughout much of Devensian time as a mobile and sensitive ice sheet, during which the LGMadvance was but one important event. In places, glacial deposits of the earlier Devensian glaciation have previously been incorrectly identified as products of the later LGMglaciation. r 2001 Elsevier Science Ltd. All rights reserved. 1. Introduction the precise timing of the maximum extent of ice during the Last Glacial Maximum (LGM). The British and Irish Ice Sheet (BIIS) consisted of The definition of the LGMused here is that lowland and upland centres of ice dispersal that were the recommended by the EPILOG project (Environmental main driving forces of ice sheet configuration and flow Processes of the Ice age: Land, Oceans, Glaciers) at on to the continental shelf. Thickened sediment wedges, ‘Chronozone Level 1’, between 19 and 23 cal ka (Mix ice-contact morainic ridges and terminal outwash et al., 2001). Following the recommendation of that spreads show that deglaciation was characterized by project, radiocarbon ages are calibrated to calendar ages shrinking ice sheet margins that contracted back to most using the CALIB-4 calibration scheme. of the original centres of ice dispersion. But until The standard view of the Devensian (Wisconsinan/ recently little sedimentologic, paleoclimatic and geo- Weichselian) was that the maximum extent of LGMice chronologic evidence has been available to understand largely coincided with the limit of ‘fresh’ relatively the environments of deposition, the climate forcing or unmodified glacial and fluvioglacial landforms. This stemmed from Wright’s (1914, 2nd edition 1937) distinction between ‘older’ and ‘newer drifts’, followed by Charlesworth (1928) in Ireland with the delineation *Corresponding author. Tel.: +44-29-2087-4830; fax: +44-29-2087- 4326. of the ‘South Irish End Moraine’ that marked the LGM E-mail address: bowendq@cardiff.ac.uk (D.Q. Bowen). limit. It was then correlated with other ‘newer drift’ 0277-3791/02/$ - see front matter r 2001 Elsevier Science Ltd. All rights reserved. PII: S 0277-3791(01)00102-0 90 D.Q. Bowen et al. / Quaternary Science Reviews 21 (2002) 89–101 landforms in England and Wales (Charlesworth, 1929). 2. Cosmogenic rock exposure ages Mitchell et al. (1973) defined the Late Devensian, when the LGMoccurred, as between 26,000 and 10,000 In many cases, pre-existing geochronologic limita- radiocarbon years ago. But they believed that all of tions have been overcome through the application of previous Devensian time, from the ‘last interglacial’ to cosmogenic nuclide surface-exposure dating. Cosmo- 26 14C ka BP, was ice free. This implied that the BIIS genic nuclides (3He, 10Be, 21Ne, 26Al, 36Cl) are produced was out of phase with both Laurentide and Scandina- in rocks at the surface of the earth by the action of vian ice sheets, as well as with ice volume changes cosmic radiation (Cerling and Craig, 1994). Most of the inferred from oxygen isotope stratigraphy (Shackleton cosmic ray flux is attenuated within the top one to two and Opdyke, 1973). meters of the surface of the solid earth and thus The CLIMAP reconstruction of the LGM (CLIMAP, processes such as glaciation that excavate rock 1981) adopted views then current on the location from greater depths and deposit it on the surface can of the ice margin in England (Boulton et al., 1977), be dated by measuring the accumulation of cosmogenic Wales (Bowen, 1973), and southern Ireland (Synge, nuclides. Gosse and Phillips (2001) have recently 1970). But in western Ireland it depicted ice that reviewed the principles and methods of cosmogenic extended offshore. It also showed the BIIS and nuclide dating. Scandinavian Ice Sheet (SIS) to be in contact in the For this study, glacial erratic boulders and glacially North Sea (Andersen, 1981). The IGCP-24 project smoothed bedrock were sampled around the former ice (Glaciations in the Northern Hemisphere) showed an margins. Criteria described in Gosse and Phillips (2001) LGMice margin corresponding to those shown in Fig. 1 were used in selecting sample sites. The possibility of for Ireland and Scotland. But in England and Wales it coverage by peat was a primary consideration but it is corresponded with the ‘earlier Devensian ice maximum’ believed that the sample sites presented here had, at shown on Fig. 1. This project also concluded that the most, a thin peat blanket. Close attention was given to BIIS and SIS were not in contact during the LGM weathering and surface degradation, attempting to (Bowen et al., 1986). sample whenever possible surfaces exhibiting glacial The timing of the ‘LGM’ was adduced from only five smoothing or striations. The samples were processed for ‘maximum’ radiocarbon ages from four localities. These 36Cl analysis by dissolution in hydrofluoric acid. An were 18,2407250 14C yr BP (21.97 cal ka) and isotopically enriched, stable 35Cl spike was added to the 18,5007499 14C yr BP (21.7 cal ka) at Dimlington, east- samples during processing so that the 36Cl and stable Cl ern England (Penny et al., 1969); 30,500 14CyrBP at concentrations could be determined by isotope dilution Four Ashes, West Midlands of England, (Shotton, mass spectrometry during analysis by accelerator mass 1967b); 18,000 14C yr BP (21 cal ka) in northeast Wales spectrometry at Purdue Rare Isotope Measurement (Rowlands, 1971); and 30,500 14C yr BP in northwest Laboratory (PRIME Lab) (Elmore et al., 1979). Sur- Ireland (Colhoun et al., 1972). Ubiquitous radiocarbon face-exposure ages were calculated using the data ages for deglaciation throughout the British Isles were reduction program CHLOE (Phillips and Plummer, available from peat samples that infilled kettle holes, 1996) according to the methods and production although these mainly provided Blling/Allerd ages. constants given in Phillips et al. (2001) and Gosse and The dearth of reliable ages for establishing the age of the Phillips (2001), and are given in Tables 1 and 2. Surface- LGMcontrasts greatly with North America where the exposure ages are affected by rock-surface erosion rates, advection of moist air from the Gulf of Mexico and thus ages are listed for possible erosion rates promoted tree growth close to the southern Wisconsinan ranging from 0 to 3 mm/kyr. Sample-specific erosion ice margin. Except for ‘interstadial’ tree growth during rates were estimated based on field observations. The oxygen isotope stage 5 (Jones and Keen, 1993) there is ages corresponding to the best-estimate erosion rates are no evidence in Great Britain of tree growth during the in bold type in Table 1. Devensian other than during the Blling–Allerd. Valuable information on extent of ice has been Europe may have been isolated from warm air masses obtained in Scotland based on cosmogenic dating of by west-to-east topographic barriers, and advection of trim lines around nunataks, from which ice profiles can warm moist air masses occurred only from the west be extrapolated (Stone et al., 1998). In Ireland and during times of relatively brief interstadial warmth and Gower, however, sampling was carried out close to the strong thermohaline circulation. Thus, without appro- limits of glacial stratigraphic units that define the former priate organic samples for radiocarbon ages, most ice sheet margins. The distribution of ages resulting sectors of the Devensian ice sheet lacked adequate from this sampling approach have yielded coherent geochronologic control.