A Last Glacial Maximum pollen record from Bodmin Moor showing a possible cryptic northern refugium in southwest England

JOURNAL OF QUATERNARY SCIENCE (2010) 25(3) 296–308Copyright � 2009 John Wiley & Sons, Ltd.Published online 18 August 2009 in Wiley InterScience
(www.interscience.wiley.com) DOI: 10.1002/jqs.1309
A Last Glacial Maximum pollen record fromBodmin Moor showing a possible cryptic northernrefugium in southwest EnglandANN KELLY,* DAN J. CHARMAN and REWI M. NEWNHAMSchool of Geography, University of Plymouth, Plymouth, UK

Kelly, A., Charman, D. J. and Newnham, R. M. 2010. A Last Glacial Maximum pollen record from Bodmin Moor showing a possible cryptic northern refugium insouthwest England. J. Quaternary Sci., Vol. 25 pp. 296–308. ISSN 0267-8179.

Received 15 February 2009; Revised 11 May 2009; Accepted 15 May 2009

ABSTRACT: A late Devensian palynological record is presented from Dozmary Pool (Bodmin Moor,southwest England), beyond the southern limit of the Last Glacial Maximum (LGM) British Ice Sheet.The pollen assemblages indicate predominantly herbaceous tundra–steppe communities but alsoinclude elevated levels (typically 10–20%) of conifer tree pollen (Picea, Pinus, Abies) and lower butpersistent percentages of broadleaf tree pollen during the LGM. This record is seemingly at odds withthe orthodox view of an entirely treeless tundra–steppe environment for this region and elimination oftree species from the British Isles during glacial maxima. Long-distance pollen transport seems an
unlikely explanation for the tree pollen considering distance to the nearest known refugia, exceptpossibly for Pinus. Reworking of the tree pollen, often invoked in these circumstances, remains apossible alternative, especially given the abundance of these trees in the region during earlyDevensian interstadials. However, this explanation has been challenged by studies reporting plantmacrofossil and faunal evidence for survival of temperate biota during glacial maxima and fromclimate modelling work that suggests some trees could have survived the glacial extremes in areas wellbeyond the recorded glacial refugia. Assuming reworking was not a major factor, the DozmaryPool pollen record is consistent with the ‘cryptic northern refugia hypothesis’ that invokes survival oftrees in small, scattered populations under locally favourable conditions during glacial maxima.Copyright # 2009 John Wiley & Sons, Ltd.
Additional supporting information may be found in the online version of this article.

KEYWORDS: late Devensian; palynology; glacial tree refugia; conifer trees; Dozmary Pool; Mystery Interval.

Introduction

The palaeoecological record of environmental change in theBritish Isles for the last glacial (Devensian) period is surprisinglyincomplete. To the south of the glaciated zone (Fig. 1) only ageneralised picture of the changing landscape and bioticcommunities is available, owing to a lack of well-dated,continuous sequences. The fragmentary records (e.g. Simpsonand West, 1958; Coope et al., 1961; Morgan, 1973; Hall, 1990;Maddy et al., 1998) are derived mostly from fluvial deposits andare biased towards warmer intervals more likely to result indeposition and preservation of finer sediment under low-energyconditions. In detail, large gaps remain for certain regions andintervals and there is limited direct evidence for the Last GlacialMaximum (LGM; ca. 24–18 cal. ka BP; Mix et al., 2001) inparticular.

* Correspondence to: A. Kelly, School of Geography, University of Plymouth,Drake Circus, Plymouth PL4 8AA, UK.E-mail: [email protected]

Nevertheless, from these fragmentary records, a prevailingview (e.g. Godwin, 1975) has emerged of unglaciated regionsof Britain characterised by treeless tundra–steppe communitiesduring long glacial periods, interspersed with milder intervalsor interstadials, when woody vegetation temporarily expandedin less inhospitable areas. These interruptions to the treelesslandscape were more prominent in the early stages of theDevensian glaciation, and by the late Devensian glacialmaximum it seems that trees had been eliminated completelyfrom the British landscape. This view is consistent withtraditional concepts of vegetation response during glacial–interglacial cycles, where tree species from the northerntemperate zone of today retreated to southern glacial refugia,from which they expanded during periods of climateamelioration. This conventional view has been questionedrecently using various lines of evidence, including plantmacrofossils (e.g. Willis and van Andel, 2004; Binney et al.,2009) mammalian faunas (e.g. Stewart and Lister, 2001) andpollen (e.g. Caseldine et al., 2007) which point independentlytoward the stronger presence of thermophilous trees andpersistence of woodland habitat during the last glacial period incentral and northwestern Europe including the British Isles. In

Figure 1 The southwest peninsula with Dozmary Pool and the UK LGM ice limit (inset). After Bowen (1999)

A LAST GLACIAL MAXIMUM POLLEN RECORD FROM BODMIN MOOR 297

this context, there is an obvious need to develop continuous,well-dated palaeoecological records from unglaciated sites inBritain that extend through the LGM.

The southwest of England is one region with potential formore continuous deposits of Devensian age to be preserved,owing to its being free of ice cover during this period (Fig. 1;Campbell et al., 1998; Bowen et al., 1999). Palynologicalinvestigations of potential pre-Holocene sediments in themainland southwest have so far only yielded sequencescovering the Lateglacial (Brown, 1977), with the oldest dateson continuously accumulated sediments extending back to anindefinite time before ca. 15 cal. ka BP (Hill et al., 2007).Additionally, some data are available on organic unitsinterbedded with soliflucted material in coastal sections onthe Isles of Scilly (Scourse, 1991, 1996). These records suggestBetula and Pinus may have been present during the Lateglacialinterstadial (Brown, 1977; Hill et al., 2007) and there arerecords of Pinus from mid/late Devensian sediments on the Islesof Scilly, albeit attributed to long-distance pollen transport(Scourse, 1991).

This paper presents details of a new pollen record fromlimnic sediments beneath the peat of Dozmary Pool mire onBodmin Moor, southwest England, dated to the LGM. Thepollen assemblages indicate tundra–steppe communities, butalso contain significant amounts of coniferous tree pollen andsome broadleaf tree pollen. We discuss the possible origins of

Copyright � 2009 John Wiley & Sons, Ltd.

this tree pollen in the context of the debate over glacial treerefugia in northern Europe.

Changing concepts of European glacialtree refugia

The concept of glacial refugia developed from observationsbased largely on pollen records that vegetation communitiestended to track their climatic optima during glacial–interglacial cycles. A simplistic expansion–contraction (EC)model emerged of the principal vegetation biomes adjustingtheir geographical ranges over time, predominantly in an N–Sdirection, in response to marked temperature changesbetween cold and warm stages. For example, the unglaciatedregions of northern Europe that today are characterised bytemperate broadleaf forests were occupied by steppe–tundraduring cold stages (West, 2000). The temperate forest treessurvived these long, unfavourable periods by contracting theirranges into ‘glacial refugia’, from which they subsequentlyexpanded during transitions to warm stages. In Europe, theassumption that these glacial refugia were all distantly locatedin southern or eastern regions is supported by long pollenrecords from Greece (Tzedakis, 1993; Tzedakis et al., 2002)

J. Quaternary Sci., Vol. 25(3) 296–308 (2010)DOI: 10.1002/jqs

298 JOURNAL OF QUATERNARY SCIENCE

and Italy (Follieri et al., 1988) in particular. Further supportingevidence stems from an early application of phylogeographicmethods to this problem by Hewitt (1996, 1999, 2000),who used analyses of genetic variation in temperate plantsand animals to demonstrate the existence of major refugia inthe Mediterranean peninsulas of Iberia, Italy and theBalkans.

Recent views suggest a more complex pattern of glacialrefugia, based in part on macrofossil evidence for tree survivalduring the LGM beyond these southern refugial regions, forexample in Hungary (Willis et al., 2000; Willis and van Andel,2004) and Slovakia (Ltynska-Zajac, 1995) and northern Eurasia(Binney et al., 2009). Fossil records of temperate mammals foundin stratigraphic context with cold-stage faunas, particularly fromcave sites, have also prompted suggestions that northernwoodland refugia must have existed. For example, the remainsof red deer (Cervus elaphus), a temperate to boreal woodlandspecies, have been found in several late Devensian cave depositsin the UK, including Kent’s Cavern, Devon (Fig. 1) (Lister, 1984).Charred fragments of wood identified as Quercus sp., Ulmus sp.and Rhamnus catharticus, dated to 14 295� 120 14C a BP, havealso been found at Kent’s Cavern (Campbell, 1977). Drawing onthis and other evidence, Stewart and Lister (2001) proposed theconcept of ‘cryptic northern refugia’: isolated and scattered areasof sheltered topography that provided suitable microclimates fortemperate plants and animals during cold stages and whichsupplemented the well-known southern and eastern refugia.Support for this concept has emerged from recent phylogeo-graphic studies, summarised by Provan and Bennett (2008), thatalso cast doubt on the idea that temperate areas were recolonisedsolely from southern refugia after the LGM. For example, adetailed analysis of fossil and genetic data confirmed theexistence of relatively northern refugia of temperate trees inEurope while, in contrast, the populations that survived the lastglacial period in the Mediterranean regions did not spread intocentral Europe (Magri et al., 2006; Magri, 2008).

These recent findings have generated renewed interest inpalaeobotanical investigations of the last cold stage that target

Figure 2 Topographic context and sampling location at Dozmary Pool


possible refugial areas in unglaciated parts of northern Europe,such as southwest England.

Site description and previous work

Dozmary Pool, Bodmin Moor (SX 192744; 508 320 5N00; 48 320

900 W) is a large, shallow, freshwater pool and associated mire ata present-day altitude of 265 m OD (Fig. 2) on the southernflanks of Bodmin Moor, one of several extensive granitebatholiths that largely define the upland areas of the Devon–Cornwall peninsula of southwest England. It is one of three sitesexamined by Brown (1977) to reconstruct Lateglacial andHolocene vegetation change. A date (9053� 120 14C a BP) wasobtained for the silty detritus mud at the base of Brown’s profile,and which is overlain by Holocene peats.

At present the lake is less than 2 m deep at its deepest part,with a surface area of 15.7 ha (Newbold and Hazlehurst, 2000).Input is from springs along the western edge of the lake and a mirehas developed on the southwest margin extending over 4.7 ha.

The lake and mire are contained within a clearly definedshoreline which is surrounded by gently sloping grasslandwhere an acid brown earth has developed overlain by a peatyhorizon. The mire supports an oligotrophic bog includingSphagnum mosses, Eriophorum spp., Molinia caerulea,Calluna vulgaris, Erica spp. and Juncus spp. Aquatic macro-phytes include Elatine hexandra, Isoetes echinospora, Littorellauniflora, Nitella flexilis var. flexilis, Fontinalis anitpyretica andDrepanocladus fluitans (Newbold and Hazlehurst, 2000). Bothfor biological and geological criteria (Campbell et al., 1998)Dozmary Pool is a Site of Special Scientific Interest (SSSI).

The current climate at Dozmary Pool is predominantlytemperate oceanic with 30-year average winter minimumtemperatures ranging between 1.5 and 1.78C and averagesummer maximum temperatures reaching values of 17.2–19.08C. Air frost occurs on fewer than 54 days per year andaverage annual rainfall is 1690 mm. Snow may lie on the groundfor up to 16 days in any year (UK Met Office, 2009). The data



provided by the UK Met Office are extrapolated from the nearestweather stations of St Mawgan and Princetown, Dartmoor.

Methods

Field and laboratory subsampling

Three cores were removed from Dozmary Pool mire at thesample point (Fig. 2) – DZ1, DZ2 and DZ10 – using a cobrapercussion auger with a 50 mm diameter lined corer head.Cores DZ1 and DZ2 were removed less than 2 m apart andDZ10, at a later date, 15 m away. The three cores reacheddepths between 5.90 and 6.20 m and coring halted once theunderlying granite surface was encountered. All cores showednear-identical stratigraphy. DZ1 provided material for a pilotstudy and DZ2 was subsampled for multiple laboratoryanalyses; DZ10 was used to provide material for a rangefinderluminescence date.

Palynology

91 subsamples were removed from DZ2. Samples of 2 cm3

were prepared following Moore et al. (1991) but were sieved at300mm rather than 180mm after the pilot study revealed thepresence of Abies pollen in some samples. Abies pollen issufficiently large (�175mm) to have been impeded by sievingat 180mm. It may have been previously overlooked, therefore,in British late Quaternary investigations.

The samples were examined using an Olympus CH-2 lightmicroscope at magnifications of �400 and oil immersion�1000. Identification was verified using pollen keys publishedin Moore et al. (1991) and Faegri and Iversen (1989) and theUniversity of Plymouth pollen reference collection. Taxonomyand nomenclature followed that of Bennett (1994). A count of300 total land pollen (TLP) was made where this was possible,or 1000 Lycopodium spores, whichever total was reached first.Counts of TLP grains were below 100 between 495.5 and508.5 cm and although most grains were well preserved someof those between these depths showed signs of degradation. Afurther nine samples had counts of between 150 and 300 TLP.Those samples containing low TLP counts were included in thefinal pollen diagram as it was considered that they wererelevant to the overall pollen profile but are interpreted with alower degree of confidence.

Luminescence dating

One optically stimulated luminescence (OSL) date wasobtained from a bulk sample of the basal sediments of coreDZ10 in order to inform the subsequent radiocarbon datingprogramme. Particular care was taken to prevent exposure todaylight by immediately covering the retrieved end of the corewith triple layers of black plastic and by subsampling in adarkened room. The sample was removed from the core at adepth of 570–575 cm and further subsampled under light-controlled conditions at the Luminescence Dating Laboratory,University of Oxford. Laboratory preparation involved theremoval of carbonates using HCl and then wet sieving to isolategrains between 90–125mm and 180–250mm. Further treatmentwith concentrated HF (48%) for 100 min dissolved feldspargrains and removed (etched) the outer surface of the quartz


grains. Heavy grains were removed by centrifuging in sodiumpolytungstate solution at 2.68 g cm�3. Final sieving removedthe heavily etched grains. The remaining quartz grains weremounted on 1 cm diameter aluminium discs using viscoussilicon oil and analysed following methodology described byRhodes et al. (2003). Concurrently, a separate sample was sentfor commercial analysis to determine the dose rate (mGy a�1),and a measure of the water content of the sediments made. Theresults of the analysis are presented in Table 2.

Accelerator mass spectrometric 14C dating ofsediments removed from DZ2

Because of a lack of plant macrofossils, bulk sediments weredated from 1 cm thick slices removed from DZ2. Care wastaken to remove samples from points identified as significant inthe pollen diagram. Discs of sediment were removed using aclean scalpel and the outer 1–2 mm of material removed. At theNERC Radiocarbon Laboratory at East Kilbride, samples weredigested in 2M HCl (808C, 8 h), washed free of mineral acid withdeionised water and then dried and homogenised. The sampleswere heated with CuO in sealed quartz tubes and the resultantCO2 gas was converted to graphite by Fe/Zn reduction, prior to14C analysis by accelerator mass spectrometry (AMS). 14C AMSresults were calibrated using the radiocarbon calibrationprogramme Calib 5.02 (Stuiver and Reimer, 1993). Thecalibrated results of the analyses are shown in Table 2.

Charcoal particle counts

Charcoal particles were counted on the prepared pollen slidesand results calculated using a formula described by Clark(1982) to give cm2 charcoal cm�3 sample, for each level. Onlythose particles between 10 and 300mm size were counted, asparticles below 10mm were difficult to identify positively ascharcoal. The results are presented on the pollen diagram(Fig. 4). Alternate samples throughout core DZ2 were alsoexamined to identify macro particles of charcoal but none wereobserved.

Percentage organic carbon analysis

Subsamples from DZ2 were removed from the profile at thesame depths as the pollen samples and were analysed tomeasure the organic carbon content throughout the profileusing a Shiadzu total carbon analyser. Two sieved (<1 mm)samples from each level were combusted, one at 9008C for totalcarbon measurement and one at 2008C for minerogenic carbonmeasurement (Fig. 4).

Results

Description of core lithostratigraphy andorganic carbon

At the base of the core, the granite substrate is overlain by�20 cm of coarse sand and gravel and between 600 and330 cm depth by minerogenic grey-white clays with occasional



coarse sand or fine gravel. The sediments contain very lowlevels of organic carbon below 330 cm depth, typically 0.1–0.3%. Between 329 and 330 cm level, an inclusion of fine blackparticles within the clay matrix coincides with the start of anupward trend in proportion of organic carbon and a change togrey clay. Isolated roots (�10 cm) are visible in the profile,extending from 295 to 310 cm. The truncated upper surface ofthese roots at �295 cm coincides with a visible horizontalsurface at this horizon that may indicate a possible hiatus. Thechange from predominantly minerogenic sediments to struc-tureless organic material occurs at 284 cm (Table 1).

Examination of the sediments through thin sections from coreDZ2 showed mineralogy consistent with the natural granite ofthe area together with 2.5–10.4% clay size fraction, themajority of which consists of kaolinite. These findings suggestthat the origins of the sediments were likely to be from a kaolindeposit mobilised within the immediate area by subsurficialsprings/streams which then deposited the sediments in the lake.The lake basin depression may have originated throughdifferential erosion of this kaolin deposit relative to thesurrounding granite.

Chronology

OSL and 14C ages are presented in Table 2 and illustrated asan age/depth profile in Fig. 3. The 14C ages range from25 416� 672 to 17 569� 523 cal. a BP but show two agereversals (Fig. 3). Two of the 14C AMS ages (SUERC-9590 andSUERC-9595) were measured from samples containing verysmall amounts of carbon (0.07% and 0.08%) and are thus moresusceptible to contamination, in particular by younger C.Omission of these two dates from the chronology removes theage reversals evident in the age–depth profile (Fig. 3). The OSLdate provides an independent check of this age model (Fig. 3).An alternative explanation that calls upon an ageing effect inother dated samples (SUERC-9589, SUERC-9591, and SUERC-9592) due to inwash of older material seems less likely. In eithercase, it is clear that the sediments were deposited during thelatter stages of the Late Devensian, broadly corresponding tothe LGM (24–18 ka BP).

The age–depth profile (Fig. 3) includes a further datapoint ofca. 11 cal. ka BP at �305 cm, based on comparison betweenthe pollen assemblages presented here and Brown’s (1977)14C-dated early Holocene pollen diagram from the site. We

Table 1 Stratigraphic description of core DZ2

Depth Description

262–274 cm Dark/black disintegrated humus substance274–278 cm Homogeneous highly humified mud, yellowish colour278–284 cm Homogenous, highly humified mud, dark-brown in colour284–289 cm Evidence of clay particles, yellowish colour, very plastic289–295 cm Greater mineral component, evidence of clay particles, v

yellowish-brown in colour295–310 cm Clay-like particles, grey in colour but inclusions of long (1

rootlets310–329 cm Clay-like particles, grey in colour but no inclusions329–330 cm Clay-like particles, grey in colour, many inclusions of fine bla330–600 cm Undifferentiated clay-like particles, grey in colour, plastic, no

laminations but several inclusions of larger granite pieces600–610 cm Solid particles, grey/white in colour, coarse sand within fi

matrix610–620 cm Solid particles, grey/white in colour, fine gravel size with

within fine-grained matrix


include this additional age estimate to indicate a majorchange in sediment accumulation rates between the parts of thesequence below 324 cm (�31 cm ka�1) and above 324 cm(�3 cm ka�1). This marked difference in sediment accumu-lation between the LGM and Lateglacial parts of the record isconsistent with sedimentological and palynological evidencefor a hiatus during the latter period, as discussed further below.

DZ2 pollen assemblage zones (PAZ)

Cluster analysis (CONISS, based on total sum of squares)provided a basis for zonation of the diagram (Fig. 4) into fivePAZ. The pollen record above 284 cm depth is not presentedhere since it corresponds closely to the Holocene sequence forthis site which has been comprehensively described by Brown(1977). The pollen diagram for core DZ2 (Fig. 4) presents onlythe principal taxa observed and the full dataset is given insupporting information available online.

DZ2-PAZ 1: 610–516 cm

Through this zone, pollen assemblages are dominated byPoaceae (up to 60%) and Cyperaceae (up to 40%), these twotogether always contributing almost 90% TLP. There is,however, a very gradual increase in both Pinus and Picea,up to a combined maximum of 10% TLP and occasionaloccurrences of small numbers of broadleaf taxa such as Betula,Corylus and Salix. At 605 cm, a small number of Juniperus wereobserved. At modest levels also, Artemisia, Chenopodiaceaeand Saxifragaceae together contribute variably up to 10% of theoverall TLP count. The aquatics Isoetes and Myriophyllumfeature strongly throughout the zone, with a peak of Isoetes at525 cm. Charcoal levels increase gradually from very lowvalues, and some zero counts, towards a maximum of32 cm2 cm�3 by the end of the zone.

DZ2-PAZ 2: 516–492.5 cm

The TLP count for the assemblages in this zone is lower than100 TLP grains per sample, with most grains poorly preserved.

Troels-Smith (1955) nomenclature

Sh.4; Humo.4;Nig.4; Strf.0; Elas.0; Sicc.1; Lim.4Sh.2; As.2; Humo.3; Nig.2; Strf.0; Elas.0; Sicc.1; Lim.4Sh.4; Humo.4; Nig.3; Strf.0; Elas.0; Sicc.1; Lim.4Ag.3; As.1; Nig.1; Nig.1; Strf.0; Elas.0; Sicc.2; Lim.4

ery plastic, Ag.3; As.1; Nig.2; Strf.0; Elas.0; Sicc.2. Lim.2

0 cm) black Ag.3; As.1; Dl.1;Nig.2; Strf.0; Elas.0; Sicc.2; Lim.3

Ag.3; As.1; Nig.2; Strf.0; Elas.0; Sicc.2; Lim.3ck particles Ag.3; As.1; Dg.1; Nig.2; Strf.0; Elas.0; Sicc.2; Lim.3evidence of As.3; Gs.1; Nig.1; Strf.0; Elas.0; Sicc.2; Lim.3

ner-grained Gs.2; Ag.2; Nig.1; Strf.0; Elas.0; Sicc.2; Lim.1

coarse sand Gg.2; Gs.1; Ag.1; Nig.1; Strf.0; Elas.0; Sicc.2; Lim.2


Figure 3 Plot of 14C and OSL ages against core depth. Dashed line depicts the suggested age/depth relationship (see text). This figure is available incolour online at www.interscience.wiley.com/journal/jqs


All taxa (including aquatics) appear to decline during thiszone except for Picea, which is likely to have inflated valuesdue to the low pollen count combined with thecomparatively robust, easily recognised features of this taxon.Poaceae and Cyperaceae remain significant at valuesapproaching 20% TLP.

DZ2-PAZ 3: 492.5–320 cm

The zone is defined by a constant level of 10–15% TLP for Piceaand about 10% TLP for Pinus, together with isolated grains ofAbies (450–425 cm), which also occurs in small numbers inzones 1 and 2 between 525 and 505 cm. The curve for thecombined coniferous species reaches peaks of 25–30% TLP,between 41 and 91 grains per sample, the most substantialvolume seen in the diagram for these species. Small numbers of

Table 2 The results of the 14C AMS and OSL dating. 14C AMS results were cand Reimer, 1993). Calibrated ages shown as midpoint of age range � 2s

14C AM

Samplecode

Sampledepth (cm)

Labcode

DZ2-1 324.5 SUERC-9588DZ2-2 414.5 SUERC-9589DZ2-3 498.5 SUERC-9590DZ2-4 513.5 SUERC-9591DZ2-5 569.5 SUERC-9592DZ2-6 609.5 SUERC-9595

OSL

Samplecode

Sampledepth

Labcode

D

DZ10 570-575 X1159 77


Alnus, Betula and Corylus are also present throughout thiszone. There is an increase in Artemisia, Chenopodiaceae andSaxifragaceae, which is observed throughout zone 3. Myr-iophyllum levels are much lower in this zone than in theprevious one, while Isoetes levels fluctuate markedly through-out.

DZ2-PAZ 4: 320–305 cm

The boundary between PAZ 3 and PAZ 4 at 320 cm depthmarks a considerable change. Although there is an increase inPoaceae (to �50%) and Cyperaceae (to �20%), there is amarked decline in most of the other taxa, particularly theconiferous tree curves. Of these, Pinus declines to much lowerlevels than previously, while Picea virtually disappears at thelower zone boundary. Abies is no longer present. There are low

alibrated using radiocarbon calibration programme Calib 5.02 (Stuiver

S

Carboncontent (%)

14C age(a�1 SD)

Calibrated age(cal. a� 2s)

0.307 14568�129 17569�5230.197 18325�216 21769�6020.074 17177�185 20310�4370.102 19122�240 22746�6380.188 21144�314 25416�6720.083 19121�240 22745�638

e (Gy) Dose rate(mGy a�1)

Age (years before)AD 2000

� 9.2 3.23� 0.26 23900� 3500


Figu

re4

Core

DZ

2poll

endia

gram

show

ing

sele

cted

taxa

,ch

arco

alin

flux

and

per

centa

geof

org

anic

mat

ter.

Dat

esar

eca

l.a

BP

Copyright � 2009 John Wiley & Sons, Ltd. J. Quaternary Sci., Vol. 25(3) 296–308 (2010)DOI: 10.1002/jqs



values for herbaceous taxa, notably Artemisia and Saxifraga-ceae, while Chenopodiaceae and Empetraceae maintain theirprevious levels. There are low values recorded for Betula,Ulmus and Corylus, and tree pollen overall is much reducedthroughout the zone. The zone is further characterised by alarge increase in the aquatic Isoetes, which reaches 80% of theTLP, aquatics and spores total. Total pollen concentration andcharcoal influx increase considerably in this zone, presumablya consequence of the marked reduction in sedimentationaccumulation rate noted earlier.

DZ2-PAZ 5: 305–284 cm

Poaceae and particularly Cyperaceae values are reduced inzone 5 but steep increases in Corylus and Betula, and theemergence of Quercus and Ulmus, characterise the zoneand clearly differentiate it from zone 4. Corresponding tothe increase in Corylus there is a decrease in Salix andEmpetraceae as well as a substantial reduction in Chenopo-diaceae. The aquatics Myriophyllum and Isoetes are alsoreduced at this time. Charcoal levels are highly variable. Boththe pollen concentration curve and the organic carbon contentof the sediments remain elevated throughout this zone.

Discussion

The pollen diagram for Dozmary Pool (Fig. 4) shows significantchange between the upper and lower portion of the profile. PAZ5 is dominated by the pollen of broadleaved tree taxa andPoaceae, similar to the deepest pollen spectra recovered byBrown (1977), and can be attributed to the early Holocene. Incontrast, PAZ 1 to PAZ 4 are dominated by pollen of Poaceaeand Cyperaceae, which together consistently make up 70–80%TLP and include numerous other herbaceous taxa such asArtemisia, Chenopodiaceae and Saxifragaceae. Aquatic taxaare present throughout, in particular Isoetes, whereas Myr-iophyllum pollen declines markedly after PAZ 2. Theseassemblages, within highly minerogenic sediments, indicatean open tundra–steppe environment and persistent lacustrineconditions at Dozmary Pool throughout PAZ 1–3.

Zones PAZ 2 and PAZ 3 are marked by a substantial increasein coniferous tree pollen to unexpected levels. The compara-tively high Picea percentages during PAZ 2 must be viewedwith caution owing to the low pollen concentrations and lowpollen counts (<100 total land pollen grains per sample) duringthis zone. In PAZ 3, however, where total land pollen countsexceed 300 grains throughout, both Picea and Pinus are presentcontinuously and usually at levels >10%, along with moresporadic occurrences of Abies, Alnus, Betula and Corylus. Atthe same time, Cyperaceae values decrease and Poaceae valuesincrease. The radiocarbon and OSL age estimates spanningapproximately 26–17 cal. ka BP constrain the sediments in PAZ1 to PAZ 3 to the LGM. At the top of PAZ3 (ca. 17.5 ka), bothPicea and Pinus pollen levels decline abruptly and Piceadisappears from the record.

The high levels of coniferous tree pollen at Dozmary Pool areinteresting in light of the current debate over glacial tree refugiain northern Europe. Below we consider the three most likelysources for this tree pollen.

Long-distance transport

Conifer trees are wind pollinated and hence typically producelarge volumes of pollen dispersed well above ground level,


while most species have saccate pollen which is comparativelybuoyant and hence more prone to long-distance transport bywind and by water. This phenomenon is especially well knownin the case of Pinus and, as a consequence, high percentages ofPinus pollen within cold stage assemblages otherwise domi-nated by herbaceous plants are not uncommon and typicallyinterpreted as long-distance transport, as, for example, byScourse (1991), rather than reflecting local presence of trees.

In contrast to Pinus, Picea is not considered a strong pollenproducer and is not known for long-distance transport inmodern pollen rain studies. Previous work has used 5%(Huntley and Birks, 1983; Magri, 1999) 4% (Ravazzi, 2002) and2% (Latalowa and van der Knaap, 2007) as minimum Piceapollen percentage levels for indicating local presence of treesand 2% in the case of Abies (Huntley and Birks, 1983). In theirdiscussion of Picea pollen representation, Latalowa and van derKnapp (2007) consider that �2% pollen indicates with nearcertainty that Picea was locally present but might fail to detectsmall populations or those present at low density. Suchthresholds should be viewed with caution if applied to coldstages, when redeposition of pollen through erosion processeswas more likely (Terhurne-Berson et al., 2004) and when treepollen could travel further across a more open landscape oftenunder stronger wind regimes. Nevertheless, the persistence ofhigh Picea pollen levels (often >10%) during PAZ 3, well abovethese thresholds, would seem to discount long-distancetransport as a factor, especially as the nearest reported refugialsources for Picea are as far away as northern Italy (Ravazzi,2002). Similar arguments may apply to Abies, despite lowerpollen levels for this taxon at Dozmary Pool, because the largeAbies pollen is reported to have a limited dispersal range(Godwin, 1975).

It is much harder to reject long-distance transport as aprimary source for the LGM Pinus pollen at Dozmary Pool,owing to its greater capacity for long-distance dispersal andproximity of possible refugia. Palaeoecological evidence (Birks,1989) and recent phylogeographic studies (Sinclair et al., 1998)point to the existence of multiple glacial refugia for Pinussylvestris that include probably Ireland or western France. Thenearest analogous records to the Dozmary Pool late Devensianpollen assemblages are from Carn Morval and Watermill Coveon the Isles of Scilly, where high levels of Pinus pollen (up to40–50%) have been attributed to long-distance travel (Scourse,1991). Moreover, Scourse interprets increasing Pinus values asevidence of climatic deterioration as a result of lower localpollen productivity in the flora generally and selectivepreservation of sacs and intact grains for Pinus. Thisinterpretation may similarly apply to the Pinus pollen atDozmary Pool, where low pollen counts and concentrationsmay have contributed to artificially high percentages of morerobust and easily recognised grains. Alternatively, the co-occurrence of elevated Picea and sporadic Abies pollen levelswith elevated Pinus pollen levels at Dozmary Pool during theLGM raises the possibility that long-distance transport of Pinusto that site and to the Isles of Scilly may not have been the solefactor.

Reworking

Similar arguments apply in consideration of reworking as apossible explanation for elevated LGM coniferous pollen levelsat Dozmary Pool. The contiguous and consistently elevatedpercentages of Pinus and Picea pollen seemingly argue againstthis explanation. Reworking would be expected to result in amore variable pattern, with pulses of sediment flux represented



by prominent peaks dominated by the more hardy coniferouspollen, separated by in situ deposition characterised by farfewer or no coniferous pollen. An exceptional situation can beenvisaged for this extreme environment however, wherebypervasive reworking of older catchment sediment persisted asthe dominant means of pollen delivery to the basin, during aninterval of sparse local vegetation cover, low pollen pro-ductivity and hence low airborne pollen input. The highlyminerogenic sediments deposited at a comparatively fast ratesupport this scenario, while the likely occurrence in thecatchment of early Devensian (e.g. Chelford Interstadial;Simpson and West, 1958) soils or sediments rich in Picea,Pinus and Abies pollen provides a possible source forreworking.

Another feature of the Dozmary Pool pollen diagram thatmay be consistent with this reworking scenario is the highlyelevated percentage of Picea pollen during PAZ 2 (estimatedage 22.5–ca. 24 cal. ka BP, i.e. early LGM). During this zonePicea pollen percentages reach 30% TLP despite a greatlyimpoverished pollen flora, indicated by very low pollenconcentrations, low pollen counts and comparatively poorpreservation. This pattern is consistent with a major flux of oldersediment or interval of increased inwash into the basin at a timewhen there was comparatively little aerial pollen input fromin situ vegetation, resulting in pronounced distortion of themore resilient and identifiable Picea grains. Similar processesmay have operated but to a lesser degree during PAZ 3, wherepollen counts improve and Picea percentages decline to 10–20%. Reworking may also help to explain the reversals in the14C age model (Fig. 3) through the delivery of older carbon tothe depositional site.

Survival of trees

How likely is it that coniferous tree populations survived duringthe LGM in southwest England or the adjacent continentalshelf? In addition to phylogeographic insights (e.g. Provan andBennett, 2008), there are two lines of evidence that might beused to address this question: other palaeocological data andresults of climate modelling.

As discussed earlier, the strongest supporting palaeoecolo-gical evidence for the existence of northerly forest refugiacomes from mammalian remains, including those from Kent’sCavern, also in southwest England (Stewart and Lister, 2001).Ecological niche modelling of potential ranges for caribou(Rangifer tarandus) and red deer (Cervus elaphus) during theLGM over northwestern Europe also suggests these speciescould have survived in southwest England (Banks et al., 2008).Plant macrofossil evidence provides further support for north-erly refugia of trees in central and eastern Europe at up to 508 N(Willis and van Andel, 2004) and in northern Eurasia (Binneyet al., 2009), further north than the well-established refugia inthe southern peninsulas of Europe (Bennett et al., 1991).Suggestions of tree refugia on the continental shelf nearScandinavia based on Lateglacial radiocarbon dates on plantmacrofossils of Betula, Pinus and Picea (Kullman, 2002)contradict all the other palaeoecological and glaciologicalevidence (Birks et al., 2005), although the case for presence isstrengthened by further recent data (Kullman, 2008; Stewartand Cooper, 2008). Closer to the study area, the only other treemacrofossil data are provided by a Lateglacial interstadial dateon Quercus, from Belgium (Otte, 1994, cited in Stewart andLister, 2001). Support for this hypothesis from other palaeoe-cological evidence in the region is therefore present butuncertain, and pollen records, unless supported by macrofossilevidence, have been equivocal (Birks and Birks, 2000). As


discussed above, much depends upon the validity of theprevailing assumption that elevated tree pollen levels in cold-stage assemblages must derive from reworking or long-distancetransport – an assumption often justified in terms of an implicitprecautionary approach, but which carries an inherentlycircular argument.

Turning to climate modelling results, climate simulationsduring the coldest phases of Marine Isotope Stage (MIS) 3provide the closest analogue available at appropriate spatialresolution (Barron and Pollard, 2002). These simulations showa strengthened N–S gradient, with winter temperatures up to208C cooler than today in the north but with mid-latitudeEurope (45–508 N) only 7–108C cooler and even less insouthwesterly regions such as the Iberian peninsula. Summertemperatures may have only been 4–78C cooler than today inthis region. Although precipitation was lower than present, itwas higher in mid-latitudes than in southern Europe. The treespecies with elevated pollen levels at Dozmary Pool occurtoday in regions with more severe climates. For example, Piceaabies can tolerate frosts up to�388C as long as annual rainfall isat least 400 mm (Ravazzi, 2002) – limits that lie well within themodelled glacial climate for southwest England. Similarly,Cheddadi et al. (2006) show that the potential LGM distributionof Pinus sylvestris, determined by applying modern climaticlimits of this species to modelled climate reconstructions,extends as far north as Cornwall. Investigations of palaeoeco-logical evidence for the timing and provenance of postglacialtree migration in the British Isles (Birks, 1989) and genetics(Ennos et al., 1997; Sinclair et al., 1998) have both concludedthat a glacial refugium (or refugia) for P. sylvestris in north-central Europe is likely. The steep climate gradients modelledfor cold stages have been used to explain the local presence oftrees in central and eastern Europe during the LGM (Willis andvan Andel, 2004). If the climate model results are correct,conditions in the far southwestern peninsula of Britain andoffshore areas may not have been very much different fromthose in regions such as southern Poland, where charcoalremains of Pinus, Abies and Larix have been dated to the LGMat 26 170� 220 and 27 590� 700 cal. a BP (Willis and vanAndel, 2004, based on radiocarbon ages in Damblon et al.,1996, and Musil, 2003). The sites are at approximately the samelatitude as each other (�508 N) so this does not seemunreasonable. It is feasible therefore that tree populationssurvived in microclimatically suitable areas of southwestEngland and the adjacent continental shelf. Further supportis provided by recent climate modelling work showingpotential LGM ranges for Alnus incana, Betula pendula, Betulapubescens, Picea abies, Pinus sylvestris and Salix caprea as farnorth as the southern limits of the British Ice Sheet (Svenninget al., 2008).

What does this tell us about likely patterns and characteristicsof northern cryptic refugia? The likely species represented bythe three coniferous taxa with elevated pollen levels atDozmary Pool (Picea abies, Pinus sylvestris and Abies alba)share certain common ecological affinities that provide someclues. All three occur in montane–alpine regions of central andsouthern Europe and throughout northern Europe, with Piceaabies and Pinus sylvestris extending to the northernmost limitsof forest in Europe. Interestingly, Bhagwat and Willis (2008)demonstrate through hierarchical cluster analysis that pollen ofthe conifer tree taxa found at Dozmary Pool are those whosecharacteristics are fully consistent with the life history andhabitat preferences that make them suitable to have survived inmore northerly refugia. In some respects these regions may beviewed as interglacial refugia. Severe temperature depressioncan occur in these regions but the key to survival of these trees ismore likely to be moisture availability. Picea, for example, is



most sensitive to extreme continental climate and in particularto winter desiccation (Ravazzi, 2002) while Abies alba isdrought intolerant (Terhune-Berson et al., 2004). Pinussylvestris can tolerate temperatures as low as �188C but doesnot survive today in areas that experience less than 400 mmannual rainfall (Cheddadi et al., 2006). Moisture dependencymay explain the postulated occurrence of these three conifersin the oceanic setting of southwest England during the LGM. Ifso, northern cryptic refugia are more likely to be found insituations where local precipitation or moisture availability isenhanced, such as in sheltered valleys of the low hills ofBodmin Moor.

One further question might be asked at this point: if smallremnant populations of Picea survived near Dozmary Poolduring the LGM, why did the species disappear so completelyat the end of the LGM, represented by PAZ 3, and not persistthrough the Lateglacial into the Holocene warm phase? Thispart of the Dozmary Pool pollen record is subject to greateruncertainty as just 20 cm sediment accumulation separates thehorizon where Picea abruptly declines, dated at ca. 17 ka andthe earliest Holocene sediments, by which time Picea hasdisappeared from the record. As argued above, sedimentaccumulation slowed dramatically or at times may have ceasedduring this interval. Almost certainly the lake level fell at times,exposing the lake bed; loss of sediment throughwinnowing accompanying desiccation of the lake may alsohave occurred.

The marked reduction in coniferous pollen, includingelimination of Picea andAbies, at Dozmary Pool thus coincideswith a major change in the local depositional environment(Fig. 5). The period of no, or markedly reduced, sedimentaccumulation, extending from ca. 17 ka to the early Holocene,includes the so-called deglacial ‘mystery interval’ (17.5–

Figure 5 Dozmary tree and shrub pollen using age model of Fig. 3 compacoincide with shift to lighter isotope values in NGRIP. Sedimentary hiatus atcoincides with marked Lateglacial fluctuations in NGRIP record (and elsewDenton et al. (2006)


14.5 ka) of Denton et al. (2006) and the subsequent Lateglacialoscillation. During the ‘mystery interval’, ice-rafted debris wasdischarged extensively into the North Atlantic (the Heinrich 1event; Hemming, 2004), North Atlantic meridional overturningcirculation was shut down (McManus et al., 2004), sea surfacetemperatures in the northern Atlantic Ocean and Mediterra-nean Sea were significantly depressed (Bard et al., 2000; Cachoet al., 2001) and hypercold winters occurred in northwesternEurope (Renssen and Isarin, 2001) and in Greenland (Dentonet al., 2005). Paradoxically, during this interval of extreme coldin western Europe, extensive deglaciation occurred in theEuropean Alps (Denton et al., 1999) and global atmosphericCO2 levels, measured from Antarctic ice cores, rose markedly(Monnin et al., 2001). Denton et al. (2006) explain thiscontradiction by suggesting that the cold oceanic and terrestrialconditions reflect extensive winter sea ice cover in the northernAtlantic, perhaps as far south as 488 N, while the retreat oftemperate mountain glaciers reflect warmer summers linked torising atmospheric CO2.

The effect of these severe conditions on the climate andvegetation of southwest England must have been marked.Hypercold winters would have posed a major threat to anywoody vegetation that, having survived the LGM, was alreadyclose to its climatic limits. Extreme cold would have beenexacerbated by dry and windy conditions that are likely to haveaccompanied expansion of sea ice and southward displace-ment of the Inter-Tropical Convergence Zone (Lea et al., 2003).For survival of woody vegetation, these highly detrimentalwinter conditions would have outweighed the favourableconditions that accompanied warmer summers. The decline oftree pollen and disappearance of Picea pollen at Dozmary Poolduring the ‘mystery interval’ therefore are consistent withwider-scale regional changes in the North Atlantic and western

red with NGRIP (Greenland) d18O. Elevated LGM levels of tree pollenDozmary Pool occurs soon after decline in coniferous tree pollen andhere) that commence with the ‘mystery interval’ (ca. 17.5–14.5 ka) of



Europe regions, and in particular with hypercold winters andreduced moisture availability. Similarly, the evidence for lakedesiccation concurs with a scenario of warmer summers withhigher rates of evaporation and hypercold winters when thelake basin is likely to have frozen.

A similar pattern was observed in marine sediment cores inthe Gulf of Lions interpreted as indicating LGM survival ofAbies, Picea and deciduous Quercus in the drainage basins ofthe Pyreneo-Languedocian rivers of southeastern France(Beaudoin et al., 2007). These trees were subsequently absentfrom the Rhone drainage basin during the deglaciation and alsodisappeared from the Pyreneo-Languedocian drainage basinsfrom ca. 17–15 cal. ka BP.

The usefulness and accuracy of the term ‘glacial refugia’ hasrecently been questioned by Bennett and Provan (2008). Theyargue that the term ‘bottleneck’ is more appropriate fordescribing the temporarily reduced glacial populations fol-lowed by population size increases that characterise theindividualistic response of taxa to climate change. If the recordof Picea at Dozmary Pool reflects local trees, than it providesfurther evidence that individual taxa may vary markedly in theirresponse to climate change. It also reveals an additionalproblem with the term ‘glacial refugia’ by suggesting survival ofPicea during much of the last glaciation in a region beyond itspresent-day limits followed by contraction, not expansion,during postglacial times. In this respect, the term ‘delayedcontraction’ may be a better descriptor for the behaviour ofPicea at Dozmary Pool.

Also contained within the Dozmary Pool LGM sediments arepollen grains of broadleaf trees and shrubs, Betula, Alnus andCorylus and, although in low numbers, they are recordedthroughout PAZ 3 at a time when the coniferous tree pollencurves reach their maximum levels. As with the conifer trees,these taxa are currently presumed to have been eliminated fromthe British Isles during glacial maxima, but the Dozmary Poolrecord raises the possibility that these broadleaf trees andshrubs also occurred near the site in cryptic refugia.

Conclusion

Our understanding of environmental and biotic changes in theunglaciated mid–high-latitude regions of the Northern Hemi-sphere during the most extreme cold stages of the Quaternary isconstrained by a paucity of sedimentary and palaeocologicalrecords. In regions bordering the great northern ice sheets alimited array of scattered sites and fragmentary records isperhaps an inevitable legacy of a combination of harsh,changing climate and environmental instability. Often nodirect archive survives of the most extreme intervals andreconstructions rely upon untested assumptions and extrapol-ation across broad intervals of space and time. It is notsurprising, therefore, that alternative views about glacial treerefugia in northern Europe have emerged.

The palaeoecological record from Bodmin Moor in south-west England near the maximum extent of Late Devensian iceprovides a near-continuous pollen record through the LGM.Interpretation of this pollen record in terms of vegetationchange is hampered by comparatively low pollen concen-trations and the paucity of comparable, coeval records from theregion. The inferred vegetation record must also be qualified bythe untested assumption that alternative pollen-taphonomicsources were not a significant factor in delivering pollen to thecore site. Although long-distance transport of pollen seemsunlikely, reworking of at least some pollen from older strata


remains a strong possibility. Nevertheless, recent climatemodelling showing the potential for tree survival in this regionduring the LGM compels us to consider this more straightfor-ward explanation for the Dozmary Pool pollen record.

Assuming the pollen assemblages reasonably reflect thecontemporary vegetation around the site, two features standout. First, the occurrence of high percentages of coniferous treepollen and lower but persistent percentages of broadleaf treepollen between 25 416� 672 and 17 569� 523 cal. a BP,broadly coinciding with the LGM. This evidence is contrary tothe orthodox view of treeless steppe–tundra grasslands, butconsistent with the ‘northern cryptic refugia hypothesis’ thatinvokes survival of temperate biota during glacial maxima andwith climate modelling work that suggests some trees couldhave survived the glacial extremes in areas well beyond therecorded glacial refugia where moisture availability was not alimiting factor.

Second, the delayed contraction of tree pollen (at ca. 17 ka)and inferred extensive period of lake desiccation imply moresevere environmental conditions were experienced in south-west England during the deglaciation than during the LGM. Thisapparent contradiction nevertheless is consistent with widerregional evidence for climate deterioration and strong seasonaldifferentiation across western Europe and the North Atlanticregion during the so-called ‘mystery interval’, ca. 17.5–14.5 ka.An obvious corollary is that bioclimatic conditions were moresevere in southwest England (and perhaps further afield) duringthe deglaciation ‘mystery interval’ than during the LGM.

The critical assumption underpinning these observations canbe tested by further palaeocological investigation at other sitesin the region, including offshore. Substantial replication of theDozmary Pool LGM record would go some way to confirmingthe cryptic refugia hypothesis, which has profound ramifica-tions for biogeography and prehistory. Similarly, there is a clearneed to find and investigate other sedimentary sequencesrepresenting the last deglaciation in southwest England thatmight provide further insight into the marked environmentalchanges during the so-called ‘mystery interval’.

Acknowledgements We would like to thank the NERC RadiocarbonFacility for funding six AMS 14C dates, allocation number 1137.1005.We also thank the Quaternary Research Association and the School ofGeography, University of Plymouth, for funding the rangefinder OSLdate and the Cartography Unit of the School of Geography, Universityof Plymouth, for the preparation of Figs 1 and 2. Thanks are due also toJohn Stewart and an anonymous referee of this paper for their informedand helpful comments. We are grateful to Ralph Fyfe for comments onthe manuscript and to James Scourse, Donatella Magri and ChronisTzedakis for helpful discussion.

References

Banks WE, d’Errico F, Peterson AT, Kageyama M, Colombeau G. 2008.Reconstructing ecological niches and geographic distributions ofcaribou (Rangifer tarandus) and red deer (Cervus elaphus) duringthe Last Glacial Maximum. Quaternary Science Reviews 27: 2568–2575.

Bard E, Rostek F, Turon J-L, Gendreau S. 2000. Hydrological impact ofHeinrich events in the Subtropical Northeast Atlantic. Science 289:1321–1324.

Barron E, Pollard D. 2002. High-resolution climate simulations ofOxygen Isotope Stage 3 in Europe. Quaternary Research 58: 296–309.



Beaudoin C, Jouet G, Suc JP, Berne S, Escarguel G. 2007. Vegetationdynamics in southern France during the last 30ky in the light ofmarine palynology. Quaternary Science Reviews 26: 1037–1054.

Bennett KD. 1994. Annotated catalogue of pollen and pteridophytespore types of the British Isles. Department of Plant Sciences,University of Cambridge, UK.

Bennett KD, Provan J. 2008. What do we mean by refugia? QuaternaryScience Reviews 27: 2449–2455.

Bennett KD, Tzedakis PC, Willis KJ. 1991. Quaternary refugia of northEuropean trees. Journal of Biogeography 18: 103–115.

Bhagwat SA, Willis KJ. 2008. Species persistence in northerly glacialrefugia of Europe: a matter of chance or biogeographical traits?Journal of Biogeography 35: 464–482.

Binney HA, Willis KJ, Edwards ME, Bhagwat SA, Anderson PM,Andreev AA, Blaauw M, Damblon F, Haesaerts P, Kienast F, Kre-menetski KV, Krivonogov SK, Lozhkin AV, MacDonald GM,Novenko EY, Oksanen P, Sapelko TV, Valiranta M, Vazhenina L.2009. The distribution of Late-Quaternary woody taxa in northernEurasia: evidence from a new macrofossil database. QuaternaryScience Reviews. doi: 10.1016/j.quascirev.2009.04.016

Birks HJB. 1989. Holocene isochrones maps and patterns of tree-spreading in the British Isles. Journal of Biogeography 16: 503–540.

Birks HJB, Birks HH. 2000. Future use of pollen analysis must includeplant macrofossils. Journal of Biogeography 27: 31–35.

Birks HH, Larsen E, Birks HJB. 2005. Did tree-Betula, Pinus and Piceasurvive the last glaciation along the west coast of Norway? A reviewof the evidence, in light of Kullman (2002). Journal of Biogeography32: 1461–1471.

Bowen DQ, Lewis G, Maddy D, Gibbard PL, Preece RC, Campbell S,Hunt CO, Scourse JD, Keen DH, Croot DG, Thomas GSP, SutherlandDG, McCabe AM, Cameron TDJ, Holmes R. 1999. A Revised Cor-relation of Quaternary Deposits in the British Isles. GeologicalSociety Special Report 23.

Brown AP. 1977. Late-Devensian and Flandrian vegetational history ofBodmin Moor, Cornwall. Philosophical Transactions of the RoyalSociety of London B 276: 251–320.

Cacho I, Grimalt JO, Sierro FJ, Shackleton NJ, Canals M. 2001.Evidence for enhanced Mediterranean thermohaline circulationduring rapid climatic coolings. Earth and Planetary Science Letters183: 417–429.

Campbell JB. 1977. The Upper Palaeolithic of Britain. Clarendon Press:Oxford.

Campbell S, Hunt CO, Scourse JD, Keen DH. 1998. Quaternary ofSouth-West England, Geological Conservation Review Series. Chap-man & Hall: London.

Caseldine CJ, McGarry SF, Baker A, Hawkesworth C, Smart P. 2007.Late Quaternary speleothem pollen in the British Isles. Journal ofQuaternary Science 23: 193–200.

Cheddadi R, Vendramin GG, Litt T, Francois L, Kageyama M, Lorentz S,Laurent J-M, Beaulieu J-L, Sadori L, Lost A, Lunt D. 2006. Imprints ofglacial refugia in the modern genetic diversity of Pinus sylvestris.Global Ecology and Biogeography 15: 271–282.

Clark RL. 1982. Point count estimation of charcoal in pollen prep-arations and thin sections of sediments. Pollen et Spores 24: 523–535.

Coope GR, Shotton FW, Strachan I. 1961. A late Pleistocene flora andfauna from Upton Warren, Worcestershire. Philosophical Trans-actions of the Royal Society of London B 244: 379–421.

Damblon F, Haesaerts P, Van der Plicht J. 1996. New datings andconsiderations on the chronology of Upper Palaeolithic sites in theGreat Eurasiatic Plain. Prehistoire Europeenne 9: 177–231.

Denton GH, Heusser CJ, Lowell TV, Moreno PI, Andersen BG, HeusserLE, Schluchter C, Marchant DR. 1999. Interhemispheric linkage ofpaleoclimate during the last glaciation. Geografiska Annaler 81a:107–153.

Denton GH, Alley RB, Comer GC, Broecker WS. 2005. The role ofseasonality in abrupt climate change. Quaternary Science Reviews24: 1159–1182.

Denton GH, Broecker WS, Alley RB. 2006. The mystery interval 17.5 to14.5 kyrs ago. Pages News 14: 14–16.

Ennos RA, Sinclair WT, Perks MP. 1997. Genetic insights into theevolution of Scots pine, Pinus sylvestris L., in Scotland. BotanicalJournal of Scotland 49: 257–265.


Faegri K, Iversen J. 1989. Textbook of Pollen Analysis, (4th edn with KKrzywinski). Wiley: Chichester.

Follieri M, Magri D, Sadori L. 1988. 250,000-year pollen record fromValle di Castiglione (Roma). Pollen et Spores 30: 329–356.

Godwin H. 1975. Quaternary History of the British Flora, (2nd edn).Cambridge University Press: Cambridge, UK.

Hall K. 1990. Quaternary glaciations in the southern ocean: sector 0-long.–180- long. Quaternary Science Reviews 9: 217–228.

Hemming SR. 2004. Heinrich events: massive late Pleistocene detrituslayers of the North Atlantic and their global climate imprint. Reviewsof Geophysics 42: RG1005.

Hewitt GM. 1996. Some genetic consequences of ice ages, and theirrole in divergence and speciation. Biological Journal of the LinnaeanSociety 58: 247–276.

Hewitt GM. 1999. Post-glacial re-colonization of European biota.Biological Journal of the Linnaean Society 68: 87–112.

Hewitt GM. 2000. The genetic legacy of the Quaternary ice ages.Nature 405: 907–913.

Hill DJ, Haywood AM, Hindmarsh RCA, Valdes PJ. 2007. Characteris-ing ice sheets during the mid Pliocene: evidence from data andmodels. InDeep-Time Perspectives on Climate Change:Marrying theSignal from Computer Models and Biological Proxies, Williams M,Haywood AM, Gregory J, Schmidt D (eds). MicropalaeontologicalSociety Special Publication. Geological Society: London; 517–538.

Huntley B, Birks HJB. 1983. An Atlas of Past and Present Pollen Mapsfor Europe 0-13,000 Years Ago. Cambridge University Press: NewYork.

Kullman L. 2002. Boreal tree taxa in the central Scandes during theLate-Glacial: implications for Late-Quaternary forest history. Journalof Biogeography 29: 1117–1124.

Kullman L. 2008. Early postglacial appearance of tree species innorthern Scandinavia: review and perspective. Quaternary ScienceReviews 27: 2467–2492.

Latalowa M, van der Knaap WO. 2007. Late Quaternary expansion ofNorway spruce Picea abies (L.) Karst. in Europe according to pollendata. Quaternary Science Reviews 25: 2780–2805.

Lea DW, Pak DK, Peterson LC, Hughen KA. 2003. Synchroneity oftropical and high-latitude Atlantic temperatures over the Last Glacialtermination. Science 301: 1361–1364.

Lister AM. 1984. Evolutionary and ecological origins of British deer.Proceedings of the Royal Society of Edinburgh 82: 205–229.

Ltynska-Zajac M. 1995. Anthracological analysis. In Complex of UpperPalaeolithic Sites near Moravany, Western Slovakia, Hromada J,Kozlowski J (eds). Jagiellonian University Press: Krakow, Poland; 74–79.

Maddy D, Lewis SG, Scaife RG, Bowen DQ, Coope GR, Green CP,Hardaker Y, Keen DH, Rees-Jones J, Parfitt S, Scott K. 1998. TheUpper Pleistocene deposits at Cassington, near Oxford, England.Journal of Quaternary Science 13: 205–231.

Magri D. 1999. Late Quaternary vegetation history at Lagacione nearlago di Bolsena (central Italy). Reviews of Palaeobotany and Paly-nology 106: 171–208.

Magri D. 2008. Patterns of post-glacial spread and the extent of glacialrefugia of European beech (Fagus sylvatica). Journal of Biogeography35: 450–463.

Magri D, Vendramin GG, Comps B, Dupanloup I, Geburek T, GomoryD, Latalowa M, Litt T, Paule L, Roure JM, Tantau I, van der KnaapWO, Petit RJ, de Beaulieu J-L. 2006. A new scenario for the Qua-ternary history of European beech populations: palaeobotanicalevidence and genetic consequences.New Phytologist 171: 199–221.

McManus JF, Francois R, Gherardi J-M, Keigwin LD, Brown-Leger S.2004. Collapse and rapid resumption of Atlantic meridional circula-tion linked to deglacial climate changes. Nature 428: 834–837.

Mix AC, Bard E, Schneider R. 2001. Environmental processes of the iceage: land, oceans, glaciers (EPILOG). Quaternary Science Reviews20: 627–657.

Monnin E, Indermuhle A, Dallenbach A, Fluckiger J, Stauffer B,Stocker TS, Reynaud D, Barnola J-M. 2001. Atmospheric CO2

concentrations over the Last Glacial Termination. Science 291:112–114.

Moore PD, Webb JA, Collinson ME. 1991. Pollen Analysis. Blackwell:London.



Morgan A. 1973. The Pleistocene geology of the area north and west ofWolverhampton, Staffordshire, England. Philosophical Transactionsof the Royal Society of London B 265: 233–297.

Musil R. 2003. The Middle and Upper Palaeolithic game suite in centraland south-eastern Europe. In Neanderthals and Modern Humans inthe European Landscape during the Last Glaciation, van Andel TH,Davies SW (eds). McDonald Institute for Archaeological Research:Cambridge, UK; 167–190.

Newbold C, Hazlehurst D. 2000. Dozmary Pool, Bodmin Moor, Corn-wall: A Survey of the Aquatic Macrophytes. English Nature: Peter-borough and Truro, UK.

Otte M. (ed.). 1994. Le Magdalenien du Trou de Chaleux ERAUL 60,Liege.

Provan J, Bennett KD. 2008. Phylogeographic insights into crypticrefugia. Trends in Ecology and Evolution 23: 564–571.

Ravazzi C. 2002. Late Quaternary history of spruce in southern Europe.Review of Paleobotany and Palynology 120: 131–177.

Renssen H, Isarin RFB. 2001. The two major warming phases of the lastdeglaciation at �14.7 and �11.5 ka cal BP in Europe: climatereconstructions and AGCM experiments. Global and PlanetaryChange 30: 1–2.

Rhodes EJ, Bronk Ramsey C, Outram Z, Batt C, Willis L, Dockrill S,Bond J. 2003. Bayesian methods applied to the interpretation ofmultiple OSL dates: high precision sediment ages from Old ScatnessBroch excavations, Shetland Isles. Quaternary Science Reviews 22:1231–1244.

Scourse JD. 1991. Late Pleistocene Sratigraphy and Palaeobotany of theIsles of Scilly. Philosophical Transactions of the Royal Society ofLondon B 334: 405–448.

Scourse JD. 1996. A field guide for the Isles of Scilly excursions. In FieldGuide: Late Quaternary Coastal Change in West Cornwall, HealyMG (ed.). Quaternary Research Association: London.

Simpson IM, West RG. 1958. On the stratigraphy and palaeobotany of aLate-Pleistocene organic deposit at Chelford, Cheshire. New Phytol-ogist 57: 239–250.


Sinclair WT, Morman JD, Ennos RA. 1998. Multiple origins for Scotspine (Pinus sylvestris L.) in Scotland: evidence from mitochondrialDNA variation. Heredity 80: 233–240.

Stewart JR, Cooper A. 2008. Introduction: Ice Age refugia andQuaternary extinctions: an issue of Quaternary evolutionarypalaeoecology. Quaternary Science Reviews 17: 2443–2448.

Stewart JR, Lister AM. 2001. Cryptic northern refugia and the origins ofthe modern biota. Trends in Ecology and Evolution 16: 608–613.

Stuiver M, Reimer PJ. 1993. Radiocarbon calibration program CALIBRev5.0.2. Radiocarbon 35: 215–230.

Svenning J-C, Normand S, Kageyama M. 2008. Glacial refugia oftemperate trees in Europe: insights from species distribution model-ling. Journal of Ecology 96: 1117–1127.

Terhurne-Berson R, Litt T, Cheddadi R. 2004. The spread of Abiesthroughout Europe since the last glacial period: combined macro-fossil and pollen data. Vegetation History and Archaeobotany 13:257–268.

Troels-Smith J. 1955. Characterisation of unconsolidated sediments.Danmarks Geologiske Undersoegelse 4: 1–73.

Tzedakis PC. 1993. Long-term tree populations in northwest Greecethrough multiple Quaternary climatic cycles. Nature 364: 437–440.

Tzedakis PC, Lawson IT, Frogley MR, Hewitt GM, Preece RC. 2002.Buffered tree populations in a Quaternary refugium: evolutionaryimplications. Science 297: 2044–2047.

UK Met Office. 2009. www.metoffice.gov.uk [January 2009].West RG. 2000. Plant Life of the Quaternary. Cambridge University

Press: Cambridge, UK.Willis KJ, van Andel TH. 2004. Trees or no trees? The environments of

central and eastern Europe during the Last Glaciation. QuaternaryScience Reviews 23: 2369–2387.

Willis KJ, Rudner E, Sumegi P. 2000. The full-glacial forests ofCentral and Southeastern Europe. Quaternary Research 53: 203–213.


Documents

A Last Glacial Maximum pollen record from Bodmin Moor showing a possible cryptic northern refugium in southwest England