High Resolution Palynological, Multiple Profile and Radiocarbon Dating Studies of Early Human Impacts and Environmental Change in the Inner Hebrides, Scotland
High resolution palynological, multiple profile and radiocarbon dating studies of early human impacts and environmental change in the Inner Hebrides, Scotland
Q
Heather Sugden Department of Archaeology and Prehistory University of Sheffield
Thesis submitted for the degree of Ph.D. December 1999 ý ý Yt .ýt. ý ti Ais,ý i'.. l Y.ti. ýl )
Nach rud söileacha'drasda n'a b'hnn hirnia b'hbhsuibanaite dhuibhbi' phalbh.
"Would it not be a beautiful thing now if you were just coming rather than going" A traditional Hebridean farewell VOLUME 1 012 ABSTRACT
The Inner Hebrides comprise a diverse range of environmentsand vegetation, and archaeologicalevidence suggests that people were interacting with thesefrom early Holocene times. There are relatively few detailed palynological investigations from the islands for the early Holocene and not all the published data include quantification of microscopic charcoalwhich may assistin the interpretation of human impacts. Radiocarbon datesare also lacking from a number of the published profiles so that inter-site comparisonsand comparisonswith the archaeologicalrecord are difficult.
Some pollen profiles from the Inner Hebrides contain possible indications of human impact in the first half of the Holocene (Lowe and Walker, 1986a;Andrews et al., 1987; Heronsand Edwards, 1990; Edwards and Berridge, 1994). Theseprofiles lack detail however, and it was clear that a multiple profile approachwould provide a clearer picture of vegetation changeand allow more confident interpretations of the pollen data. The coring of several sites would assistin defining the spatial differences in early Holocene vegetation within the islands and differences in the scaleof human impacts which may reflect different types of interference.
Multiple profiles were obtained from Loch a'Bhogaidh (Islay), A'Chrannag bog and Livingstone's Cave bog (Ulva) and Kinloch (Rum), all of which are close to areasof known Mesolithic occupation. A single core was obtained from Loch an t'Suidhe at Bunessan,south west Mull, from where there is currently no archaeologicalrecord for the Mesolithic. All cores were analysedfor pollen and microscopic charcoal and AMS dateswere obtained for all profiles.
The results provide evidence for changesin vegetation due to climatic impacts and inferred Mesolithic activity. The possible effects of human influence vary from temporary woodland reductions to the creation of heathlandand cereal cultivation. The use of multiple profiles is validated in that it provides an indication of spatial variation in inferred land use patternsover short timescales.The results are compared with previously published studies and the factors influencing the early Holocene spreadof arboreal taxa, and the elm decline, are re-evaluated. ACKNOWLEDGEMENTS
I am indebted to a large number of people and organisationswho have assistedin
many ways with the production of this thesis. First and foremost I must thank my
supervisor, Prof. Kevin Edwards, for his encouragement,sound advice and resilience
in the face of early drafts. Thanks are also due to NERC for my personal funding and
for providing the 43 AMS datesused in this project under the supervision of Dr. Charlotte Bryant.
Archie Wood, Scottish Nature, Rum ReserveTrust and the Hon. J. Howard kindly
granted accessto the coring sites, and assistedwith transportationof equipment and
cores to and from site. Vic Hughes,Ross Dean, Vicki Vale, Rob Craigie and John
Tweddle all assistedwith coring, despite the wet and the midges. I would like to thank
the islanders for all their hospitality, particularly Marie Kirk, Donald and Scruff from
Eigg. Additional thanks must go to Rob Craigie for assistancein the laboratory, to
John Tweddle for his commentson my data and ideas, and to Dr. Rex Harland for
dinoflaggelate analysis of samplesfrom Ulva. I am also grateful to Prof. Mike Walker
and Alan Savile for answeringmy e-mails and providing informal advice.
I am extremely grateful to all those who provided a baby-sitting service, and to my son Nathaniel, for learning to sleep through the night from an early age and tolerating his mother's long absences.A big thankyou also to Vic Hughes,Neil Wigfield, Rachel and Dave Roebuck, EiblQs, Neil Oliver, Althea Davies and Ben Hensman,all of whom helped me maintain my sanity through a long and difficult year.
Most of all I must thank my parentsfor emotional and financial support and my husband Seanfor all his support, and for maintaining his senseof humour. LIST OF CONTENTS i
LIST OF FIGURES vii
INTRODUCTION 7 The Project 2 Structure of the thesis 4 Conventions and Definitions 5
CHAPTER 1 Geographical,archaeological and palaeoenvironmental background to the Inner Hebrides, Western Scotland 6
1.1 Geographyof the Inner Hebrides 6 1.1.1 Climate 6 1.1.2 Geology, Geomorphologyand Topography 7 1.1.3 Soils 10 1.1.4 Vegetation 11 1.2 Archaeological evidencefor early Holocene human occupation in the Inner Hebrides 14 1.2.1ThePalaeolithic 14 1.2.2The Mesolithic 15 1.2.1.1Archaeological sites 16 1.2.1.2 Excavationsand evidencefor resourceutilisation 17 1.2.1.3Dates and other evidence 19 1.2.3 The Neolithic 21 1.3 The early Holocene pollen record for the Inner Hebrides 22 1.3.1 Islay and Jura 23 1.3.2 Oronsay and Colonsay 25 1.3.3Mull 26 1.3.4Rum 28 1.3.5 Skye 31 1.4 Reasonsfor further pollen analysis in the Inner Hebrides 32
CHAPTER 2 Methods and Site Selection 36
2.1 Methods 36 2.1.1Multiple coring 36 2.1.1.1 Pollen distribution and multiple coring in lake sediments 37 2.1.1.2 Multiple coring of peat 41 2.1.2 Microcsopic charcoal: problems of distribution and taphonomy 43 2.1.2.1 Microscopic charcoal in lake sediments 46 2.1.2.2Microscopic charcoal in peatbogs 49 2.1.3 Interpretation of microscopic charcoal in pollen preparations 49 2.1.4 Quantification of microscopic charcoal 50 2.1.5 AMS radiocarbondating of pollen profiles 51 2.1.5.1 Choice of levels for dating 51 2.1.5.2 Processingof samples 51 2.1.5.3 Calibration of dates 52 2.1.5.4 Extrapolation of datesand construction of age-depthcurve 52 2.1.6 The detection of cereal-typepollen grains 53 2.1.6.1 Cereal scanning 53 2.1.6.2 Criteriafor the identificationof cereal-typepollen 54 2.1.7 Damagedpollen 55 2.1.7.1 Presentationof damagedpollen data 58 2.1.8 Dinoflagellate analysis 58 2.1.9 Sample treatmentand laboratory methods 59 2.1.10 Loss on Ignition (LOI) 60 2.1.11 Construction of pollen diagrams 60 2.2 Choice of sites 61 2.3 Coring 63
CHAPTER 3 Loch a Bhogaidh, Islay. Discussion and interpretation of pollen from lake sediments 64 3.1 Site location and description 64 3.2 Background to the current project 65 3.3 The possible extent of the earlier loch 65 3.4 The pollen diagrams 66 3.5 AMS dateson the Loch a'Bhogaidhprofiles 68 3.5.1 LABI 68 68 3.5.2 LABIII - LABVIII 3.5.3 The early hazel peak, LABIb(i) 69 3.5.4 LABIIIc 70 3.5.5 Extrapolation of age-depthcurves and pollen influx calculation 71 3.6 Isoetes lacustris sporesas an indicator of erosion 72 3.7 Damagedpollen analysis 74 3.8 General representationsof the pollen profiles 74 3.8.1 Pollen concentrationand influx 75 3.8.2 Microscopiccharcoal concentrations and influx 79 3.9 Interpretation of the profiles 81 3.9.1 LABIa(i), LABIa(ii), LABIIb (Pre 10,000BP) 81 3.9.2 LABIb(i), LABIb(ii), LABIIc, LABIIIa, LABIVa, LABVa,
LABVIa, LABVIIa. LABVIIIa 81 ,0 3.9.3 LABIc(i), LABIIb(i), LABIVb(i), LABVb(i), LABVIb, LABVIIb(i), LABVIIIb 85 3.9.4 LABIc(i), LABVIIb(i) andb(ii), LABIIIb(ii), LABVb(i) and(ii), LABVb(i), b(ii) and b(iii), LABVIb(ii) 86 3.9.5 Discussion of the different featuresin the profiles 88 3.9.5.1 Feature la, LABIIIb(i) 88 3.9.5.2 Feature 1 89 3.9.5.3 Features2,3,4,5 and 6 91 3.9.5.4 Feature7 LABIVc, LABVIc. and d, LABVIIc and d, LABIc(i), LABVIIIc and d, LABIId(ii), LABIIIc and d, LABVc and d 96 3.10 Infilling of the loch with peat 102 3.10 Conclusions 103
11 CHAPTER 4 Pollen analysis of sedimentsfrom Loch an t'Suidhe, Ross of Mull 105
4.1 Site description 105 4.2 Backgroundto presentstudy 106 4.3 Pollen diagrams and zones 107 4.4 Discussion of zones 107 4.4.1 2LSa c. 9800 - 7900 BP (8910 - 8500 cal. BC) 107 4.4.2 2LSb c. 7900 - 6850 BP (8500 - 6810 cal. BC) 109 4.4.3.2LSc Alnus - Corylus - Betula - Quercusc. 6850 - 5200 BP (6880 - 4100 cal. BC) 115 4.4.4 2LSd Betula - Alnus - Corylus - Poaceaec. 5200 - 4550 BP (4010 - 3260 cal. BC) 117 4.5 Summary of Holocene vegetation history at Loch an t'Suidhe to c. 5000 BP (3860 cal. BC) 118
CHAPTER 5 ULVA. Discussion of multiple profiles from two peat filled basins 120
5.1 Site descriptions and backgroundto the project 120 5.2 Minerogenic sediments 122 5.3 AMS dates 123 5.3.1 A'Chrannag bog profiles 123 5.3.2 Livingstone's Cave bog profiles 125 5.4 Presentationand discussionof A'Chrannagpollen data 128 5.4.1 ACIa Betula - Cyperaceae- Filipendula; AClb(i) Sphagnum; Corylus - Calluna - 6250 BP AClb(ii) Corylus - Poaceae- Cyperaceaec. 8700 - 129 (7790 - 5160 cal. BC) 5.4.2 AClb(iii); AClc; AC2a; AC2b c. 6250 - 5300 BP 132 (5160 - 4150 cal. BC) 5.4.3 ACId Betula-Calluna;AC! e(i) Betula; ACIe(ii) Corylus-Calluna;AC2c(i) Alnus - Corylus- Poaceae 136 c. 5300- 4200BP (4150- 2790cal. BC) Potentilla 5.4.4 AC2c(ii) and AC2c(iii) Betula - Calluna - Poaceae- 140 c. 4200 - 2500 BP (2790 - 600 BC) 5.5 Presentationand discussionof the Livingstone's Cave profiles 140 Cyperaceae; 5.5.1 LC 1a Cyperaceae- Cichorium intybus-type; LC2a 141 c. 11,000- 10,000BP (11050 - 9530 cal. BC) 5.5.2 LC1b and LC2b (no pollen) c. 10,000- 5500 BP 142 9530 - 4380 cal. BC) 5.5.3 LClc(i) Corylus - Cyperaceaeand LC1c(i) Corylus - Cyperaceae, 142 c. 5515 - 5350 BP (4350 - 4160 cal. BC)
iii 5.5.4 LClc(ii) Ulmus - Quercus - Abius - Corylus and LC2c(ii) Ulmus - Quercus- Alnus - Corylus; c. 5350 - 4900 BP 144 (4160 - 3730 cal. BC) 5.5.3 LClc(iii) Alnus - Corylus - Cyperaceae; LC2d Alnus-Corylus-Poaceae; LC2e(i) Betula - Alnus - Corylus - Poaceae; c. 4900- 4000BP (3730- 2530cal. BC) 147 5.5.6 LCld Alnus - Corylus; LCle Betula - Poaceae- Potentilla; LC2e(ii) Betula - Cyperaceae- Filipendula c. 4000 - 3200 BP (2530 - 1480 cal. BC) 151 5.5.7 LC2fAlnus - Corylus - Poaceae;3200 - 2600 BP (1480 - 670 cal. BC) 153 5.6 Contrastsbetween bogs 154 5.6.1 General observations 154 5.6.2 The expansionof arboreal taxa 156 5.6.3 The Ulmus profiles 157 5.6.4' Human impact 160 5.6.5 Summary 161
CHAPTER 6 RUM. Analysis of pollen from Kinloch 162
6.1 Site description 162 6.2 Background to the presentstudy 162 6.3 Reconstruction of vegetation history at Kinloch using profile K (Hirons andEdwards, 1990) 164 6.4 Comparisonof KL 1, KL3 andKL4 with K 165 6.5 Differences between the profiles 167 6.5.1 The potential for an hiatus in KL1 167 6.5.2 The pollen assemblageprior to reduction of woodland at c. 3950BP (2440cal. BC) 168 6.5.2.1KL3a(i) andKIId 6.5.2.2 KLIa(i), KL4a(i), KL3a(ii) and KIIe 169 6.5.3 Reduction of Alnus-Corylus woodland c. 3900 BP (2440 cal. BC) KL1b, KIIIa, KL3b, KL4a(ii) 171 6.5.4 The post woodland decline flora from c. 3900 BP (2440 cal. BC) KL1 c, KIM, KL3c, KL4c 174 6.5.4.1 Alnus increase,KL3c(ii), KL1c(ii) 175 6.5.4.2 Cereal-typepollen and anthropogenicindicators in the later profiles 178 6.6 Summaryof the Kinloch data 179
CHAPTER 7 Synthesisof pollen data from the Inner Hebrides, Outer Hebrides and west coast of Scotland 181 7.1 Introduction 181 7.2 Early Holocene vegetation history in western Scotland 181 7.2.1 The early Holocene c. 10,000BP (9530 cal. BC) 181
IV 7.2.2 The Corylus rise (from 9800 BP [9300 cal. BC) 184 7.2.2.1 The influence of soils 186 7.2.2 2 Human influence and the Corylus expansion 188 7.2.2.3 Summary of evidencesurrounding the Corylus expansion 188 7.3 Holocene vegatation patternsc. 9000 - 5000 BP (8225 - 3870 cal. BC) 189 7.3.1 The Alnus expansion 193 7.4 The distributionof humanactivity assuggested in pollen diagrams 195 7.4.1 The Mesolithic 197 7.4.2 The Neolithic 200 7.5 The Ulnius declines 202 7.5.1 Climatic influence and the Ulmus declines 203 7.5.2 Soils and site location 204 7.5.3 Human impact and the Ulmus declines 205 7.5.4 Pathogenicattack 206 7.5.5 Summary of evidencefor the Ulmus declines 206 7.6 Summary of pollen records for the Inner Hebrides, western Scotland and the Outer Hebrides 207
CHAPTER 8 Review of methods and the implications of results for future research 209 8.1 Review of methods 209 8.1.1 Multiple profiles 209 8.1.2 AMS dates 212 8.1.3 Microscopiccharcoal quantification 213 215 8.1.4 Damagedpollen analysis 8.1.5 Cereal scanningand identification 216 8.2 Conclusions 216
Bibliography 220
Figures 244
Tables 500 501 3.1 Pollen zonesin the Loch a'Bhogaidh profiles 3.2 Radiocarbondates for LABI (Edwardsand Berridge, 1994) 504 505 3.3 Radiocarbondates for LABIII - LABVIII 3.4 Summary of the characteristicsof Feature 1 507 in the Loch a'Bhogaidh profiles 3.5 Summary of the characteristicsof Features2-6 within 508 the Loch a'Bhogaidh profiles 3.6 Summary of the characteristicsof Feature7, Loch 509 7700 - 7300 BP (6600 - 6190 cal. BC) at a'Bhogaidh BP 3.7 Summary of the characteristicsof Feature7,7300 - 7100 Loch 510 (6100 - 5910 cal. BC) at a'Bhogaidh
4.1 Radiocarbon datesfor 2LS 511 4.2 Comparison of zones from LS (Lowe and Walker, 1986b)
V and 2LS 512
5.1 Radiocarbondates from the Ulva profiles 513 5.2 Comparison of zonesfrom profiles AC1 and AC2 514 5.4 Comparison of zones from profiles LC1 and LC2 515
6.1 AMS datesfor the Kinloch profiles 516 6.2 Pollen assemblagezones from Kinloch 517
7.1 Dates for the Corylus rise in the Inner Hebrides and western Scotland mentioned in the text 519 7.2 Dates for the Ulnius decline in the Inner Hebrides and western Scotland mentioned in the text 520
Plates 521 3.1 Loch a'Bhogaidh, Islay 522 4.1 Loch an t'Suidhe showing core position 523 524 5.1 View of A'Crannag bog, Ulva looking towards AC2 5.2 View of Livingstone's Cave bog, Ulva 525 6.1 Kinloch pollen site, Rum 526
APPENDICES 527 APPENDIX A Calibrations of radiocarbondates 528 APPENDIX B Summaryof sampledlevels scanned for cereal-typepollen 551
vi List of Figures Page
1.1 Map of Scotland and the location of the Inner Hebrides 244 1.2 Topographic map of Skye 245 1.3 Topographic map of Rum 246 1.4 Topographic map of Mull 247 1.5 Topographic map of Islay and Jura 248 1.6 Map of Mesolithic and earlier sites referred to in the text 249 1.7 Map of Neolithic sitesreferred to in the text 250
2.1 Map of pollen sites chosenfor this study 251
3.1a Location map of Loch a'Bhogaidh,Islay 252 3.1b. Site map of Loch a'Bhogaidh, showing core locations and transect line 253 3.2 Diagram of the east-westtransect of Loch a'Bhogaidh 254 3.3 Diagramof the north-southtransect of Loch a'Bhogaidh 255 3.4 Pollen and sporepercentage diagram for LABI 256 3.5 Pollen and sporepercentage diagram for LABII 260 263 3.6 Pollen and sporepercentage diagram for LABIII 3.7 Pollen and sporepercentage diagram for LABIV 267 3.8 Pollen and sporepercentage diagram for LABV 271 3.9 Pollen and sporepercentage diagram for LABVI 275 3.10 Pollen and sporepercentage diagram for LABVII 279 283 3.11 Pollen and sporepercentage diagram for LABVIII 3.12 Summarydiagram of additionaldata for LABI 287 3.13 Summary diagram of additional data for LABII 289 3.14 Summary diagram of additional data for LABIII 290 3.15 Summary diagram of additional data for LABIV 292 3.16 Summary diagram of additional data for LABV 293 3.17 Summary diagram of additional data for LABVI 294 3.18 Summary diagram of additional data for LABVII 295 3.19 Summary diagram of additional data for LABVIII 296 3.20 Concentration diagram of selectedtaxa forLABI 297
vii 3.21 Concentrationdiagram of selectedtaxa forLABII 299 3.22 Concentrationdiagram of selectedtaxa for LABIII 301 3.23 Concentrationdiagram of selectedtaxa for LABIV 303 3.24 Concentrationdiagram of selectedtaxa for LABV 305 3.25 Concentrationdiagram of selectedtaxa for LABVI 307 3.26 Concentrationdiagram of selectedtaxa for LABVII 309 3.27 Concentrationdiagram of selectedtaxa for LABVIII 311 3.28 Influx diagram of selectedtaxa for LABI 313 3.29 Influx diagram of selectedtaxa for LABIII (original dates) 315 3.30 Influx diagram of selectedtaxa for LABIII (amendeddates) 317 3.31 Influx diagram of selectedtaxa for LABIV 319 3.32 Influx diagram of selectedtaxa for LABV 321 3.33 Influx diagram of selectedtaxa for LABVI 323 3.34 Influx diagram of selectedtaxa for LABVII 325 3.35 Influx diagram of selectedtaxa for LABVIII 327 3.36 Diagram of Corylus percentage curves 329 3.37 Diagram of Corylus influx curves 330 3.38 Diagram of charcoalconcentration curves for Loch a'Bhogaidh 331 3.39 Diagram of Ch:P curves from Loch a'Bhogiadh 332 3.40 Diagram of charcoal influx curves from Loch a'Bhogaidh 333 3.41 Composite diagram of TLP concentrationcurves 334 3.42 Composite diagram of TLP influx curves 335 3.43 Percentagesof damagedpollen types for selectedtaxa in LABIII 336 3.44 Percentagesof damagedpollen types for selectedtaxa in LABIV 339 3.45 Percentagesof damagedpollen types for selectedtaxa in LABV 342
3.46 Percentagesof damagedpollen types for selectedtaxa in LABVI 344
3.47 Percentagesof damagedpollen types for selectedtaxa in LABVII 346
viii 3.48 Percentagesof damagedpollen types for selectedtaxa in LABVIII 348
3.49a DCA plot of post-Corylus rise zonesfrom Loch a'Bhogaidh 350 3.49b DCA plot of zonesc and d from the Loch a'Bhogaidh profiles 351 3.50a Spreadof datesfor the Loch a'Bhogaidhprofiles 352 3.50b Dates from the Loch a'Bhogaidh profiles grouped by core 353 3.51 Age-depth curve for LABI 354 3.52 Age-depth curve for LABIII 355 3.53 Age-depth curve for LABIV 356 3.54 Age-depth curve for LABV 357 3.55 Age-depth curve for LABVI 358 3.56 Age-depth curve for LABVII 359 3.57 Age-depth curve for LABVIII 360 3.58 Summary Corylus pollen percentagediagram 361 c. 7700 - 7000 BP (6660 - 5910 cal. BC) 3.59 Map showing summary pollen diagram from the Loch
a'Bhogaidh profiles, c. 7700 - 7000 BP (6660- 5910cal. BC) 362 3.60 Diagram showing relationships betweenprofile zonesin the Loch a'Bhogaidh profiles 363 3.61 Diagram showing estimated times for peat inception at cores sites from Loch a'Bhogaidh 364
4.1 Locationmap of Loch an t'Suidhe,Bunessan, Mull 365 4.2 Surface sample pollen percentagesfrom Loch an t'Suidhe 366 4.3 Pollen and sporepercentage diagram of LS (Lowe and Walker, 1986b) 367 4.4 Percentagediagram of selectedtaxa from LS (Lowe and Walker, 198Gb),zones LS5, LSG and LS7 368 4.5 Pollen and sporepercentage diagram from Loch an t'Suidhe369
ix 4.6 Summary of LOI, TLP concentrationand influx, charcoal, sediment accumulation and damagedpollen from 2LS 372 4.7 Pollen and sporeconcentration diagram for selectedtaxa from 2LS 374
4.8 Pollen and spore influx diagram from selectedtaxa from 2LS 376 4.9 Percentagesof damagedpollen type for selectedtaxa from 2LS 378 4.10 Age-depth curve for 2LS 381 4.11 DCA plot of Subzonemeans for 2LS 382 4.12 DCA plot of mean taxon scoresfor the 2LS profile 383 4.13 DCA plot of mean taxon scoresfor zone 2LSb 384
5.1 Map showing Ulva bogs, transectsand core locations 385 5.2 Transectprofiles from A'Chrannag bog, Ulva 386 5.3 Transectprofiles from Livingstone's Cave bog, Ulva 387 5.4 Age-depth curve for AC1 388 5.5 Age-depthcurve for AC2 389 5.6a Spreadof datesfrom the Ulva profiles 390 5.6b Spreadof datesfrom the Ulva profiles, grouped by core 391 5.7 Age-depthcurve for LC1 392 5.8 Age-depthcurve for LC2 393 5.9 DCA plot of zone mean scoresfor AC1 and AC2 394 5.10 Pollen and spore percentagediagram for AC1 395 5.11 Pollen and spore percentagediagram for AC2 398 5.12 Summary of LOI, charcoal,TLP concentrationsand influx, sedimentaccumulation and damaged pollen for AC1 403 5.13 Summary of LOI, charcoal,TLP concentrationsand influx, sedimentaccumulation and damaged pollen for AC2 405 5.14 Pollen and spore concentrationdiagram of selectedtaxa for AC I 407
5.15 Pollen and spore concentrationdiagram of selectedtaxa for AC2 409 5.16 Pollen and spore influx diagram of selectedtaxa for AC1 411 5.17 Pollen and spore influx diagram of selectedtaxa for AC2 413
X 5.18 Percentagesof damagedpollen types for selectedtaxa from AC1 415
5.19 Percentagesof damagedpollen types for selectedtaxa from AC2 419
5.20 Pollen and sporepercentage diagram for LC 1 420 5.21 Pollen and sporepercentage diagram for LC2 425 5.22 Summary of LOI, charcoal,TLP concentrationsand influx, sediment accumulation and damagedpollen for LC1 429 5.23 Summary of LOI, charcoal,TLP concentrationsand influx, sediment accumulation and damagedpollen for LC2 431 5.24 Pollen and spore concentrationdiagram of selectedtaxa for LCl 433
5.25 Pollen and spore concentrationdiagram of selectedtaxa for LC2 435
5.26 Pollen and spore influx diagram of selectedtaxa for LC1 437 5.27 Pollen and spore influx diagram of selectedtaxa for LC2 439 5.28 Percentagesof damagedpollen types for selectedtaxa from LCI 441
5.29 Percentagesof damagedpollen types for selectedtaxa from LC2 445
5.30 Map of inferred vegetation patternsat A'Chrannag and
Livingstone's Cave bogs, Ulva, c. 8500 - 6000 BP 449 (7770 - 4910 cal. BC) 5.31 Map of inferred vegetation patternsat A'Chrannag and Livingstone's Cave bogs, c. 6000 - 5500 BP (4910 - 4380 cal. BC) 450 5.32 Map of inferred vegetation patterns at AChrannag and Cave bogs, BC) 451 Livingstone's Ulva, c. 5500 - 4000 BP (4380 -2530 cal. 5.33 Map of inferred vegetation patterns at AChrannag and Livingstone'sCavebogs, Ulva, c. 4000 - 2500 BP 452 (2530 - 600 cal. BC) 5.34 Diagram of Ulnius and Cereal-typepercentage curves and Ch:P for the Ulva profiles 453 5.35 Composite diagram of Uhnus influx for the Ulva profiles 454
6.1 Location map of Kinloch, Rum, showing core locations and transects 456
xi 6.2 Diagram of the north-south transect,Kinloch 457 6.3 Diagramof the east-westtransect, Kinloch 458 6.4 Percentagesof taxa in a surfacesample from KL1 459 6.5 Percentagesof taxa in a surfacesample from KII 460 6.6 Percentagesof taxain a surfacesample from KU 461 6.7 Percentagesof taxain a surfacesample from KU 462 6.8 Pollen and sporepercentage diagram for K 463 6.9 Pollen and sporepercentage diagram for KU 466 6.10 Pollen and sporepercentage diagram for KU 470 6.11 Pollen and sporepercentage diagram for KU 474 6.12 Summary diagram of LOI, TLP concentration,influx and charcoal for K 478 6.13 Summary diagram of LOI, TLP concentration, charcoal nd damagedpollen for KU 479 6.14 Summary diagram of LOI, TLP concentration,charcoal nd damagedpollen for KL3 480 6.15 Summary diagram of LOI, TLP concentration,charcoal nd damagedpollen for KL4 481 6.16 Pollen and sporeconcentration diagram for KL1 482 6.17 Pollen andspore concentration diagram for KL3 484 6.18 Pollen and spore concentrationdiagram for KL4 486 6.19 Percentagesof damagedpollen for selectedtaxa from KL1 488 6.20 Percentagesof damagedpollen for selectedtaxa from KL3 490 6.21 Percentagesof damagedpollen for selectedtaxa from KL4 492 6.23 Diagram linking similar zonesbetween profiles 494
7.1 Map of sitesin westernScotland, Inner Hebridesand Outer Hebrides referred to in the text 495 7.2 Chartof datesfor the Corylusrise at varioussites in the Inner Hebrides and western Scotland 497 7.3 Diagram correlating inferred human impacts at sites from Arran, the Inner Hebrides and Outer Isles 498 7.4 Chart of dates for the Ulmus declines at sites in the Inner Hebrides and Arran 499
xii INTRODUCTION
The Inner Hebrides are located west of the Scottish mainland and consist of numerous islands of diverse geology and topography which were shapedby Tertiary volcanic activity and Quaternaryglaciation. Contemporaryvegetation rangesfrom blanket bog communities to mature deciduouswoodland, and arctic-alpine plants survive in the mountainousregions on Mull, Rum and Skye. The early Holocene witnessed a rich woodland cover.
Archaeological sites and finds from the islands indicate that humanswere inhabiting and utilising the abundanceof natural resourcesthroughout the Holocene. The
Mesolithic period is particularly well representedon most of the islands, though less so on Mull and Skye, and sites suggestingextensive and prolonged occupation are found at Bolsay, Islay (Mithen and Lake, 1996) and Kinloch, Rum (Wickham-Jones,
1990). Radiocarbon dateson material from Kinloch provide evidence of the earliest known occupation of the islands at 8590±45 BP (7640 cal. BC) (ibid. ), but it is probable that people were presentin the Inner Hebrides much earlier than this
(Edwards and Mithen, 1995; Edwards, 1996) as they were in the Scottish mainland (SAN, 1998).
The abundanceof Mesolithic archaeologyon the islands and the availability of a large number of sites suitable for coring encouragethe view that Mesolithic impacts on vegetation are likely to be detectedby pollen analysis.Palaeoecological investigations on some of the islands have already been undertaken.The work of Birks (1973), Birks and Williams (1983) and Walker and Lowe (1990; 1991a; 1991b) on Skye, and Lowe and Walker (1986), Walker and Lowe (1985; 1987) on Mull is the most extensive researchin the area and focusesespecially on the Lateglacial and early Holocene times. Later investigations by Hirons and Edwards (1990) on Rum, Andrews et al., (1987) on Colonsay, Bennett (1989) and Edwards and Berridge (1994) on Islay were concernedwith Holocene vegetation reconstructionwithin a context of human occupation. Most of thesestudies suggestreductions in local woodland which may have arisen from the activities of Mesolithic people.
Given the extreme geographical variations of the Inner Hebrides, much greater spatial variation of palynological sites is required in order to fully appreciate the changing patterns of vegetation, including human impacts, throughout the Holocene. It was anticipated that higher resolution pollen sampling would assist in the provision of basic data and allow secure inter-site comparisons, especially if supported by multiple profiles and a sufficient number of radiocarbon dates.
The project
The project aims to detect the impacts of people on early Holocene vegetation in the
Inner Hebrides, and to provide more detailed information on changing vegetation patterns in the islands from c. 10,000BP (9500 cal. BC) covering the Mesolithic at all sites, and the Neolithic and later (from c. 5000 BP [5650 cal BC] onwards) on Rum and Ulva. In order to maximise the potential for detection of small scalevegetation changesmultiple profiles from single sites have been used where appropriate.
The main questionsaddressed in this thesisare as follows:
1. Do multiple coring methods expand greatly on the detail provided by single cores?
What are the differences (if any) betweenprofiles from a site and what can be inferred from these in terms of local vegetation composition, sedimentationand pollen distribution and accumulation?
2 2. Do the pollen records contain evidenceof human impact and does this vary betweenprofiles within a site andbetween sites?
3. How do the early Holocenepollen recordsfrom the Inner Hebridescompare with those from the Outer Hebrides and western Scotland?
In order to illuminate the above, several sites were chosenfor coring. Loch a'Bhogaidh on Islay and Kinloch on Rum are both located close to areasof known
Mesolithic occupation and cores from both suggesttemporary woodland reduction during the Mesolithic sites (Edwards and Hirons, 1990; Edwards and Berridge, 1994).
A site on Mull was chosenas there is very little evidenceof Mesolithic occupation on the island, yet a core from Loch an t'Suidhe (Lowe and Walker, 1986a)displays potential evidence of human activity. Greater sampling resolution and microscopic charcoal quantification was required at this site to assist interpretation. A'Chrannag and Livingstone's Cave bogs on Ulva were chosendue to their proximity to a known site of Mesolithic occupation and the lack of pollen data from the island.
The sites also cover a range of coring environmentsand incorporate lake sediments and peats.This will allow an assessmentof the value of multiple coring methods within different sediments.Implicit in this technique are the requirementsof close resolution sampling and AMS datesto allow comparisons,not only of phasesof potential human impact, but also of woodland developmentand vegetation patterns within the islands over thousandsof years.This detail may assistin the refinement of isopollen maps for the areaand in assessingthe degreeof impact that Holocene climatic changesmay have had.
3 Structure of the thesis
The thesis is divided into 8 chapters.Chapteri provides the geographicalbackground of the Inner Hebrides and summarizesthe archaeologicalevidence and published pollen data from the islands which are to be used as the context for this thesis. Chapter
2 reviews the methods employed in this study and includes commentary on pollen and microscopic charcoal taphonomy in the context of multiple profiles. This is followed by a description of the sites selectedfor study. The data from each site are presented and discussedin Chapters3-6. A synthesisof the vegetation reconstructionsis included in Chapter 7 in the context of data from other Inner Hebridean sites, as well as sites from the Outer Hebrides and the west coast of Scotland. Chapter 8 comments on the successof the methods employed and makes suggestionsfor future research.
Conventions and Definitions
Dates
All dates are presentedas uncalibratedradiocarbon years before present (BP) to allow comparison with published pollen data. As archaeologicalpublications often refer to calendar years, calibrated datesBC and AD (Stuiver and Reimer, 1993) are included in parentheseswithin the text for actual and extrapolateddates. Calibration details
(Stuiver and Reimer, 1993) of dateson the new profiles in this study are included in
Appendix A.
The term Lateglacial refers to the transition period between the last glacial and the current interglacial occurring c. 14,000- 10,000 BP (14,900 - 9500 cal. BC). The present interglacial (c. 10,000 BP [9500 cal. BC] to the present) is referred to as the
Holocene or postglacial.
4 Nomenclature
Vascular plant nomenclaturefollows Stace(1991) and pollen and pteridophyte spore nomenclaturefollows Bennett, 1994.The term 'arboreal' includes both trees and shrubsand the terms'heath' and'heathland' refer to dwarf shrubsof the Ericaceaeand Empetraceaefamilies.
5 CHAPTER 1- GEOGRAPHICAL, ARCHAEOLOGICAL AND
PALAEOENVIRONMENTAL BACKGROUND TO THE INNER
HEBRIDES, WESTERN SCOTLAND
1.1 Geography of the Inner Hebrides 1.1.1 Climate
The Inner Hebrides (Fig. 1.1) are situated off the western coast of Scotland where the thermal regime has a strong maritime influence from the North
Atlantic Drift. Mean annual temperaturesare comparableto those in lowland
England, but the seasonalrange is smaller (Green and Harding, 1983). Mean January between 20C 7.50C minimum and maximum temperaturesrange and , and the range of minimum and maximum temperaturesfor July is 10.5°C -170C
(Boyd and Boyd, 1990). Islay, the most southerly of the Inner Hebridean islands and the most exposedto maritime influences lies at the warmer end of the scale.
The islands are exposedto east-trackingcyclones and fronts, resulting in frequent and high levels of rainfall which are increasedby orographic uplift over the more mountainous islands of Mull (Jenny, 1978), Rum and Skye (Green and
Harding, 1983). Rainfall occurs an averageof 200 days per year throughout the islands and at least 1mm of rain per day is recorded.Wind speedsare high with an annual range of 15-20 knots and gales are common (ibid. ).
Although the Inner Hebrides mainly experiencecyclonic rainfall (and occasional snow when cyclones are followed by Polar Maritime and Arctic air), the extension of easterly lying anticyclonescan bring drier weather (Green and
Harding, 1983). A mean maximum for the Inner Hebrides of 20 days per year of
6 snowfall is recorded(Boyd and Boyd, 1990) and snowfall is more frequent and
greateron Rum, Mull andSkye than on otherislands due to the altitudesof their summits(Green and Harding, 1983;Jermy, 1978).
1.1.2 Geology, Geomorphology and Topography
The solid geology of the Inner Hebridean islands mainly consistsof Lewisiar
Gneiss and Tonridonian sandstones(Craig, 1991). Lava floe s from Tertiary volcanic activityoccurredon Skye (Birks, 1973), Mull (Woolley and Jermy,
1978), Eigg and Rum (Wickham-Jones,1990a); a result of theseis Bloodstone
Hill (Emeleus, 1987; 1991; Wickham-Jones,1990a, 1990b) a potential source of material for stone tool manufacturein prehistory (Wickham-Jones,1990a and b). Tertiary volcanic activity in the Ben More areaof Mull also resulted in the high basalt cliffs and columns which are dominant in the coastal landscapesof
Mull and the Isle of Ulva (Woolley and Jenny, 1978; Howard and Jones, 1990; Emeleus, 1991). -
The landscapeof the Inner Hebridean islands has been largely shapedby the climates of the late Devensianglaciation. Following the Devensiancold stage maximum c. 18,000BP (19,450 cal. BC), climatic amelioration reduced the area of ice cover over the islands.
A period of climatic deterioration between c. 11,400and 10,000BP (11,430 and
9500 cal. BC) known as the Loch Lomond ' Stadial (Gray and
Lowe, 1973; Ballantyne and Dawson, 1997) may have been preceededby other readvancesin western Scotland (Robinson and Ballantyne, 1979; Dawson,
1982; Peacock, 1983) including the Wester Ross Readvance(Robinson and
Ballantyne, 1979). However, the Loch Lomond Readvanceis the most recent
7 and clearly defined episodeof glacial activity and there is evidenceon Skye,
Mull and Rum that valley glaciers formed during this time (Peacock, 1983; Price, 1983).
The cold climates of the Loch Lomond Readvancewere responsible for the periglacial features,talus, landslips and screeslopes evident in the Inner
Hebridean landscapes.Many of thesephenomena occur in the mountainous areassuch as the Cuillins of Skye, easternMull and Rum, where altitudes have exacerbatedthe climatic effects of the Lateglacial period. Glacial drift and glacimarine sedimentsalso remain exposed(Dawson, 1983), the latter being a sourceof pebble-washedflint from which prehistoric stone tools have been made (Mercer, 1970,1974,1978-80; Wickham-Jones,1990; Edwards and
Berridge, 1994; Edwards and Mithen, 1995).
Climatic changeat the end of the Lateglacial and in the early Holocene have also shapedthe coastlinesof the Inner Hebridean islands and western Scotland.
Despite isostatic rebound of the northern landmass,sea levels were temporarily higher than they are today (by as much as 40 m around 13,000 BP [13,870 cal.
BC]) (Peacock, 1983; Price, 1983). Subsequentreductions in sea level resulted in the raised beacheswhich are frequent occurrenceswith examplesin eastern
Skye, Jura and south western Rum (Peacock, 1983; Dawson, 1983) and where
Lateglacial storm beachesare also found (Peacock, 1983). One of the most important of these featuresis the Main Lateglacial shoreline which is currently exposedas cliffs in the Oban areaof western Scotland and continues sloping under present sea level into the Irish Seabasin (Ballantyne and Dawson, 1997).
Many of the caveswithin the cliff face were occupied by people in the early Holocene (refer to Section 1.3.2). As climatic amelioration continued from the Lateglacial into the Holocene,
glacio-eustatic adjustmentsresulted in a rise in sea levels, which are thought to
have reacheda maximum betweenc. 7200 and 6200 BP (6050 and 5130 cal.
BC) (Ballantyne and Dawson, 1997). Isostatic readjustmentof the landmasses
subsequentlyproduced relatively lower sea levels, and Holocene raised beaches
resulting from relative sealevel changesare frequent in the coastlinesof the
Inner Hebrides (Dawson and Dawson, 1994).
Inner Hebridean landscapesare varied due the volcanic origins of much of the
geology and the impact of glaciation. Skye (Fig. 1.2) Mull (Fig. 1.3) and Rum
(Fig. 1.4) are the most mountainousof the islands with altitudes exceeding900 m O.D. on Skye and Mull, and 800 m on Rum. Topography is more marked on
Rum and its coastline is dominatedby steepcliffs. In the south of the island mountain slopesrun directly into the sea(cf. McCann and Richards, 1969).
Rum is not easily accessible,the easiestroutes to the interior being Kinloch
Glen which runs east-westand Kilmory Glen which runs north-south. A relatively flat and low-lying areais located to the north west of Rum at Farm
Fields where the mouth of the Kinloch River meetsLoch Scresort.
Coastal sceneryon Skye and Mull varies between cliffs and sandy beaches.The smaller island of Ulva also featuresbasalt cliffs and raised beachesalong its coastline, and hills reach a maximum of 313 m O.D. at the summit of Beinn Creagh.
Islay andJura (Fig. 1.5)are the most southerlyof the Inner Hebrideanislands.
To the north of Islay and on the western peninsula of the Rhinns, the landscape
9 consistsof low rolling hills, the highest point on the Rhinns being Beinn Tart
a'Mhill at 232m O.D. SoutheastIslay is more mountainous,with summits
reaching 491 m O.D. in the caseof Beinn Bheigeir. Jura is also mountainous,
summits typically ranging between300 and 500 m O.D. and the Papsof Jura
reach 785 m. The coastlinesof both islands display numerousraised beaches
and cliffs and small coves and pebble beachesare also frequent.
1.1.3 Soils
The soils of the Inner Hebrides are fairly uniform due to the similarity of parent
materials and the extreme wetnessof the climate. Parentmaterials have been
identified as varieties of drift including glacial and morainic tills, raised
beaches,aeolian sand and montane frost-shattereddetritus. Nine soil
classifications have been cited for the islands (Hudson and Henderson, 1983);
1. Brown rendzinasand brown calcareoussoils.
2. Brown forest soils.
3. Non-calcareousgleys.
4. Peaty gleys.
5. Humus-iron podzols.
6. Peaty podzols.
7. Organic soils.
8. Montane soils. 9. Regosols.
Rendzinasand brown calcareoussoils are confined to the limestonesof Skye andIslay. Brown forestsoils arederived from Tertiary basaltsand occur on
10 Skye,Mull (Hudsonand Henderson, 1983) and the smallerislands including
Rum (Jenny et al., 1978) and Colonsay (Birks, 1987).
Peaty gley soils are derived from many drift types and are'the most extensive soils of the Inner Hebrides' (Hudson and Henderson,1983: p. 115). Non- calcareousgleys, humus-iron podzols and peaty podzols are also common on all of the islands with humus-iron podzols in particular developing over the sands and gravels of raised beaches.Also common, due to the wetnessof climate, are organic soils or peatswhich are dominatedby blanket bog and bog heathermoor
(Hudson and Henderson, 1983).
Montane soils are categorizedby altitude as sub-alpine (over 400 m) and alpine
(over 700 m) and alpine soils usually show evidenceof active or fossil periglacial activity (Hudson and Henderson, 1983) examplesof which are found in the Cuillins of Skye, and on Rum.
Regosols are soils which are immature and developing on unconsolidated material. An example is machair which developson blown sand (Hudson and
Henderson, 1983) and may be seenon Tiree (Peacock, 1983), Mull (Jenny et al., 1978), Oronsay and Colonsay (Birks, 1987). Regosolsof unstable colluvium have been noted on Tertiary rocks to the west of Rum (Emeleus, 1987).
1.1.4 Vegetation
The vegetation of the Inner Hebridean islands is closely linked to soils, thus variations on acid-tolerant wetland communities are prevalent. Major classifications are heathermoor, bent-fescuebog and bent-rush-fescuegrassland although greater detail specific to each island may be found in Birks (1973,
11 Skye),Jenny and Crabbe (1978, Mull), Ball (1987,Rum), Howardand Jones
(1990,Ulva), Ogilvie (1995,Islay), andBoyd andBoyd (1990)and Currie and Murray (1983).The altitudesexperienced on Rum, Mull and Skyeresult in
pocketsof sub-alpincflora which includeAlchendlla alpina (Alpinc Lady's
mantle),Cardandnopsis petrdea (northernrock cress)(Ball, 1987;Bangcrtcr et al., 1978),Afinuarlia sedoides(cyplicl) (Curric andMurray, 1983)and only Rum hasspecimens of 77ilaspialpestre (alpine pcnny-crcss) (Curric and
Murray, 1983).The islandsalso have arctic- subarctic elements including Koenigia islandica(Iceland purslanc) on Rum andMull (Ball, 1987;Bangerter et al, 1978)and Arenaria norvegicassp. norvegica (Arctic sandwort)on Rum andEigg (Curric andMurray, 1983).The exposureexperienced on the west coastof Colonsaymeans that this islandalso retains similar arctic-alpincrcfUgia (Birks, 1987).
According to paleoenvironmentaldata, the Inner Hebrides were more widely wooded in the past (Birks, 1983; Ball, 1987; Birks, 1989; Boyd and Boyd,
1990; Tipping, 1994; Edwards and Whittington, 1997), but most of the natural woodland of the islands is now extinct. Ball (1983) statesthat the current distribution of natural woodland is closely linked to shelteredareas, altitude and soils.
Relicts of natural woodland survive on Rum in steepgullies and this consists of
Betula pubescens(downy birch) scrub including Ilex angustifoliun: (holly),
Crataegus inonogyna (hawthorn) and Salix spp. (willow) (Ball, 1987; Rum
ReserveOffice, 1996). Since 1957, when Rum becamea National Nature
Reserve,the north side of Kinloch has been extensively replanted with species including Alnus gluti rosa (alder), Corylus avellana (hazel), Salix spp. (willow),
12 Fraxinus excelsior (ash), Ulmus glabra (wych, elm) and Quercus spp. (oak).
Other planting was undertakenin the Kinloch and Papadil areasby the Bullough family in the last century(Ball, 1987).
Both Colonsayand Skyeare well woodedin parts,mainly with Quercus petraea,and Corylusavellana, although Sorbus aucuparia (rowan) and flex angustifoliumare also present (Birks, 1973;1987). The northernnatural limit of Quercusin the Inner Hebridesoccurs in the ccntreof Skye(Birks, 1973).Some pocketsof naturalBetula-Corylus scrub survive on Islay althoughthere is now almostno nativewoodland existing on neighbouringJura (Ball, 1983)and there are small areasof naturalwoodland on Mull which consistof Corylusscrub and Betula-Quercuswoodland (Jermy, 1978; Ball, 1983).Replanting of mixed woodlandin the last centuryhas taken place in the SornValley at Bridgcnd,
Islay (Edwards,pers. comm. ) andcommercial conifer plantation occurs on the southernRhinns of Islay andon Jura.Similar plantations dominate the Mishnish areain the north of Mull (Jermyet al., 1978).
The coastlinesof the Inner Hebridean islands obviously support maritime plant communities although somemontane speciesmay be found on the seacliffs of
Islay, Skye, Mull and Eigg. Theseinclude Saxifraga oppostifolia (purple saxifrage) and Silene acaulis (moss campion) (Currie and Murray, 1983). Salt flats are frequently colonized by salt-tolerant moorland speciessuch as Molinia caerulea (purple moor grass)and Eriophorum angustifolium (common cottongrass)in addition to halophytic speciessuch as Armeria maritima (thrift) and Plantago maritima (seaplantain) (Currie and Murray, 1983). Machair grasslandsare also found in the Inner Hebridean islands with exampleson
13 Oronsay, western Colonsay (Birks, 1987) and the north coast and Ross of Mull
(Jenny et al., 1978).
1.2 Archaeological evidence for early Holocene human occupation in the Inner Hebrides
1.2.1 The Palaeolithic (Fig. 1.6)
For most of the EuropeanUpper Palaeolithic period, Scotland was covered in glacial ice. Any archaeologicalevidence for the occupation of Scotland during interglacial episodeswill have been destroyedby subsequentglaciations
(Ballantyne and Dawson, 1997) and there is currently no acceptedevidence for a Palaeolithic human presencein Scotland. Suggestionshave been made by
Mercer (1970) that possible tangedpoints found on Jura are similar to
Ahrensburgian-typepoints which can date to the late Upper Palaeolithic, and he speculatesthat the ameliorating lateglacial climate and environment of the Inner
Hebrides would have been conducive to human occupation. Edwards and
Berridge (1995) make similar speculationsin the light of palynological and artefactual evidence from the islands.
Other evidence for a Palaeolithic presencein Scotland has come from an assemblageof reindeer antlers from ReindeerCave, Inchnadamph,Assynt, northwest Scotland. The fragmentation of the antlers was originally interpreted by Movius (1940) as artefactual,the result of human activity. Subsequent radiocarbon dating of a bulked sample of antler fragmentsprovided an age of c. 10,080BP (11,030 cal. BC) for the assemblageand the presenceof arctic vole and other lateglacial-type faunal remains from the samecontext confirmed that
14 the antlers could have been of Upper Palaeolithic origins (Morrison and Bonsall,
1980). However, reinterpretationof the assemblage(Lawson and Bonsall, 1986;
Murray et al., 1993) suggestedthat fragmentation occurred as a result of post-
depositional processesand that the antlers were naturally shed.
The remaining evidencefor a Palaeolithic presencein Scotland is
circumstantial. Following the find of a scraperin a North Seacore which may potentially date to c. 18,000BP (19,450 cal. BC), Long et al. (1986) suggested that evidenceof Palaeolithic occupationsmay now be below sealevel, and that coring of submarinesediments may eventually confirm this (cf. Coles, 1998).
1.2.2 The Mesolithic (Fig. 1.6)
The impetus for researchinto the Scottish Mesolithic has evolved from the work of Lacaille (1954). 'The Stone Age in Scotland'provided a comprehensive environmental backgroundand archaeologicalsummary for the Scottish
Mesolithic and is currently the only synthesisof the Mesolithic in Scotland
(Morrison, 1996). Many of the sites mentioned by Lacaille (1954) had been investigated by antiquariansof the late 19th and early 20th centuries.These included the middens on Oronsay and cavesalong raised beachdeposits on the west coast of Scotland near Oban. Movius (1940) was one of the early excavators,also with an interest in northern Irish Mesolithic sites, and was the first to use the ten-nsLarnian (for the Irish Mesolithic industries) and Obanian
(for the Scottish cave and midden sites), both of which were used by Lacaille
(1954). 'Obaniaif is now recognisedas the term.for a Mesolithic industry of western Scotland and the Inner Hebrides, although debatecontinues as to what exactly comprises an'Obanian! site or assemblage(Finlayson, 1995; Bonsall, 1996; Finlayson, 1996).
15 Since Lacaille (1954) there have been major advancesin the discovery and
researchof Mesolithic sites in Scotland,particularly in the last 10-15 years.
Developmentsin the interpretation of site functions and improved dating of
Mesolithic depositshave addedto the debateon various controversies,such as the date for the earliest human settlementin Scotland,the potentially varying functions of different sites and the provenanceof artefactual finds and their significance (Finlayson, 1996).
1.2.1.1 Archaeological sites (Fig. 1.6)
Much of the evidencefor the Scottish Mesolithic has come from the west coast of Scotlandand the Inner Hebrideanislands in particular.The archaeologycan be split into two main groups;middens which arefrequently associated with caves,and lithic scatters(Fig. 1.2). Most cavemidden sites are locatedon the coastof the westernmainland near Oban (Lacaille, 1954;Bonsall, 1996), althoughLivingstone's Cave on Ulva (Bonsallet al., 1991,1995)is an exception.Open midden sites are located on Oronsay(Lacaille, 1954; Mellars,
1987),and a middenwhich is currentlyunexcavated was recently discovered nearDervaig on Mull (Robins,1998), an islandwhere no substantialMesolithic archaeologicalsites were previously known. A few isolatedlithic finds from the Aros areaof Mull hadbeen ascribed to the Mesolithicby Alan Saville (pers. comm.).
Lithic scattersin the Inner Hebrides have been excavatedat Staosnaigon
Colonsay and at severalplaces on Islay, including Bolsay Farm and Gleann
Mor, as part of the SouthernHebrides Mesolithic Project (Mithen, 1995; Mithen and Lake, 1996). Other excavatedlithic sites include Newton on Islay
16 (McCullagh, 1990), Kinloch on Rum (Wickham-Jones,1990) and Lealt Bay,
Lussa Wood and Glenbatrick Waterhole on Jura (Mercer 1970; 1974; 1978-80).
Severalother isolated lithic finds have beenmade on Rum (Wickham-Jones,
1990a; 1990b) and Ulva (Bonsall et al., 1991,1995).
Most midden sites from both the west coast of Scotland and the Inner Hebridean
islands are ascribedto the `Obanian' tradition. The artefactsassociated with
'Obanian'sites are bevel-ended stone tools andbone tools, whilst lithic scatters aretypified by narrowblade technology (Finlayson, 1995; Finlayson, 1996). Interpretationof thesedifferent artefactual assemblages originally led to the conclusionthat the Obanianindustry was later in datethan the narrowblade
industry,and evenearly Neolithic ratherthan Mesolithic (Woodman, 1990), but radiocarbondates suggest that the two industriesare contemporary and that their
differencesmay eitherbe a reflectionof function(Bonsall and Smith, 1990;
Bonsall, 1996)or preservationof material(Finlayson, 1996). However, there remainproblems with the allocationof sitesto certainindustries as some middensites such as that at Livingstone'sCave, Ulva, havenot presenteda
'typically Obanian'assemblage of finds (Finlayson,1995; Finlayson, 1996). Nevertheless,radiocarbon dates and artefactual assemblages secure all Inner Hebrideansites mentioned above and shown in Figure 1.2within a Mesolithic timescale.
1.2.1.2 Excavations and evidence for resource utilisation
The abundanceof Mesolithic sites in the Inner Hebrides is perhapsnot surprising in view of the variety of resourceswhich the islands would have been able to provide. The recovery of finds from excavationsof the Inner Hebridean sites differs according to the type of site investigated,preservation, and the area
17 of excavation,but overall the finds reflectthe diversityof habitatsexploited by Mesolithic people.
Palacocnvironmcntaldata indicate that woodlandcover was far greaterin the pastin the Inner Hebrideanislands, consisting mainly of a mixed woodlandof birch-hazel-oak(Andrews et al., 1987;Whittington and Edwards, 1997) which would haveprovided firewood and a varietyof foodstuffs.Hazelnuts are abundanton Mesolithic sitesand Livingstone's Cave, Ulva andthe lithic scatter siteson Colonsay,Jura, Islay andRum haveproduced wood charcoaland charredhazel nuts and shellsin additionto lithic artefacts(Mercer, 1974; 1978- 80; Bonsall, 1989;McCullagh, 1990; Wickham-Jones, 1990b; Mithen andLake,
1996).Other vegetation types may alsohave been utilised despitetheir absence from the archaeologicalrecord. These include acorns, elderberries, Vaccinium
andEmpetrum berries and the rhizomesof aquaticplants (Price, 1990; Edwards, 1998),all of which were availablein the Inner Hebrides.
Prehistoric woodlands would also have supporteda wide variety of birds and
land mammals including red deer, which are presenton most of the Inner
Hebridean islands today. Middens on Oronsay and at Livingstone's Cave, Ulva have yielded land mammal remains which provide evidenceof hunting, and
include red deer antler and pig bone (Grigson and Mellars, 1987; Bonsall, 1989,
Russell et al., 1995). Other finds from the midden sites include the bones of marine mammals, fish bonesand otoliths, and remnantsof shellfish (Mellars,
1987; Bonsall et al., 1989,1995). Thus Mesolithic people were exploiting marine, coastal and woodland resourcesat various times in their occupationsof the islands.
18 The stonetools recoveredfrom the InnerHebridean excavations comprise flints,
bloodstoneand quartz. Beach-washed flint occursnaturally today on the east coastsof Islay andColonsay, and to the westof Jura.It is arguedthat this
resourcewould havebeen utilised by Mesolithicpeople (Mercer,1970; Wickham-Jones,1990a, 1990b; Edwards and Mithen, 1995),although assuming similar beachconditions then as for the presentday, it is debatablewhether sufficientnodules would havebeen available to supportthe local industries
(Mithen, 1995)and it may be that supplieswere greater in the past.Bloodstone is likely to havebeen more abundant than flint in the islands,and although no direct links havebeen made between bloodstone tools recoveredduring excavationsand potential sources on the islands,Wickham-Jones (1990a, 1990b)suggests that BloodstoneHill, Rum,is the most likely sourceof stone
for the finds discoveredduring the excavationsat Kinloch.
1.2.1.3 Dates and other evidence
One of the earliest known datesfor a Mesolithic presencein Scotland is that
from Kinloch, Rum, at 8590±45 BP (7640 cal. BC) (Wickham-Jones, 1990b)
although a site near Fife in the eastof Scotlandhas provided a similar date
(Wickham-Jonesand Dalland, 1998) and a recently excavatedsite in the
Lowther Hills, Clydesdalehas provided the oldest date associatedwith lithics in
Scotland so far at 9075±80 BP (c. 8060 cal. BC) (Garavell, 1998; SAN, 1998).
Lithic finds from Mesolithic sites in the British Isles may be ascribedto two main typologies; `broad blade' assemblagesdating to c. 9000 BP (8225 cal. BC) and earlier, and `narrow blade' assemblageswhich are post c. 9000 BP (8225 cal. BC). (Morrison and Bonsall, 1990). Isolated finds of tanged points associatedwith broad blade technologieshave been found in Scotland,
19 including finds from Jura(Mercer, 1978-80), Ballevullin on Tiree (Morrison andBonsall, 1990)and Newton on Islay (Edwardsand Mithen, 1995).The
identificationof the find from LussaBay on Juraas a tangedpoint (Mercer,
1978-80)has been disputed as the point is severelyabraded and broken (Morrisonand Bonsall, 1990; Finlayson and Edwards, 1997). There are also problemswith the point from Tiree,which wasnot foundin a securecontext (Morrisonand Bonsall, 1990)and the tanged-typepoint from Newton,Islay which was foundthrough fieldwalking (Edwards and Mithen, 1995).The lithic assemblagefrom the excavatedsite at Newtoncomprises mainly narrowblade lithics which post-datec. 9000BP (8225cal. BC) (Edwardsand Mithen, 1995) andthus cannotconfirm a humanpresence on Islay prior to this (cf. Finlayson andEdwards, 1997).
Other tangedpoints in Scotlandhave been found on the Scottish mainland and
Orkney (Morrison and Bonsall, 1990). However, these are also isolated finds, and the only `broad blade' assemblagerepresenting the early Mesolithic in
Scotland comes from Morton, Fife, although radiocarbondates are possibly later than that for Kinloch at 8050+355 BP (8910 cal. BC) (Coles, 1971).
Thus, there is an abundanceof evidencefor early human settlementfrom the
Inner Hebrides and west coast of Scotland. Whilst Kinloch on Rum provides the earliest date for human occupation in the islands, it is possible that in the absenceof archaeologicalsites and artefactual evidence,the palaeoenvironmentalrecord may shed additional light on the suggestionthat humans were presentin the Inner Hebrides prior to c. 9000 BP (8225 cal. BC)
(Mercer, 1979,1978-80; Edwards and Mithen, 1995) in addition to contributing to a reconstruction of the environmental background to Mesolithic occupation.
20 1.2.3 The Neolithic (Fig. 1.7)
Neolithic cultures are generally associatedwith settled agriculture rather than the hunter-gathererpractices of their predecessors,with a more sedentary existencefacilitating the social and practical meansof monument building. It may therefore be anticipated that evidenceof Neolithic occupationswould be greaterthan for the Mesolithic, as monumentsand longer occupation periods of domestic sites are expectedto be more durable in the archaeologicalrecord.
However, known Neolithic sites in the Inner Hebrides are relatively few.
The most common Neolithic remains in the Inner Hebrides are the chambered cairns. Many of theseare found on Skye and Islay, although a few examplesare located on Colonsay, Jura and Mull (Henshall, 1972; Ashmore, 1996). Stone circles are rare in western Scotland (Ritchie and Harman, 1985), but an exception is Lochbuie stone circle on Mull. Standing stonesare also uncommon; a line of standing stonesdating to c. 4000 BP (2500 cal. BC) is found at Dervaig on Mull and there are standing stonesof supposedsimilar date at Ballinaby, to the north of Loch Gorm on the Rhinns of Islay.
There is someNeolithic settlementevidence from afore-mentionedMesolithic sites although this is not extensive.Finds were mostly recovered from small pits and hollows similar to those of the earlier Mesolithic contexts, but blade assemblagesand pottery shardssuggest Neolithic origins and thesehave been confirmed by radiocarbon datesin many instances(summarized in Armit,
1996). The archaeologyreflects agricultural and pastoral subsistence.Ovicaprid bones and pottery were recoveredfrom the upper levels of the midden at
21 Livingstone'sCave on Ulva (Bonsall,1989) and pottery was also recovered from the Oronsayshell middens (Mellars, 1987).Cereal-type pollen was retrievedfrom pottery shardsat Kinloch, Rum (Moffatt, 1990)and charred cerealgrains including those of cultivatedbarley (Hordeum vulgare) and rye
(Secalecereale)ýwere recovered from a pit on Ulva (Bonsallet al., 1991,1995). Charcoalat the baseof the pit wasradiocarbon dated to 499L+60BP (5690cal.
BC) althoughthe cerealgrains will presumablybe youngerthan this (Bonsallet al., 1995).
Although it is not possible to identify whether the original Mesolithic inhabitants gradually adoptedagricultural practicesin the islands or whether they were replacedby immigrants (cf. Armit and Finlayson, 1992) this cultural changeis significant both in archaeologicalterms and in terms of potential differences in environmental exploitation evident.
1.3 The early Holocene pollen record for the Inner Hebrides (cf. Fig. 7-1)
There have been many pollen studiesof sites in the west coast of Scotland and near coastal areas(e. g. Nichols, 1967; Rymer, 1977; Birks, 1980; Robinson,
1983; Affleck et al., 1988; Edwards and McIntosh, 1988; Robinson and
Dickson, 1988; Haggart and Sutherland, 1992; Rhodeset al., 1992; Tipping,
1992; Walker et al., 1992), particularly in connection with archaeology.
Palynological investigations of the Inner Hebridean islands have focused on the larger islands of Skye and Mull and there is little published pollen data for most of the smaller islands (Eigg, Muck, Tirec, Coll, Canna,Ulva). Much of the pollen data from Skye and Mull were used in the reconstruction of Lateglacial environments, and little detailed attention has beenpaid to Holocene vegetation changes.Emphasis has often been placed on regional vegetation successions
22 rather than local ecologiesand potential human impacts, and radiocarbon dates are few, although more recent pollen profiles from Rum and Islay are the exceptions (Hirons and Edwards, 1990; Birks, 1993; Edwards and Berridge, 1994).
1.3.1 Islay and Jura
Little palynological work had taken place on Islay until fieldwalking and excavationslinked to the southernHebrides Mesolithic Project produced discoveries of many areasof Mesolithic and early Neolithic occupation (Mithen and Lake, 1996).
Most of the palynologicalinvestigations have taken place on the Rhinnsof Islay wherethe most extensivearchaeological sites were discovered and a summary of pollen datafrom thesesites is providedby Edwards(in press).Analyses of two coresfrom Loch Conailbheand Coultoon Bog in the southernRhinns, were hamperedby an hiatusin the sedimentsat Loch Conailbheand high quantities of deterioratedpollen in the Cultoonmaterial (Bennett, 1989). Nevertheless, the pollen profiles suggestthat Corylusavellana was dominantin local vegetation in the earlyHolocene. This comparesfavourably with pollen datafrom Loch Gorm (Bunting et al., in press)in the northernRhinns which alsoshow a high abundanceof earlypostglacial Corylus avellana-type pollen, but evengreater amountsof ericaceouspollen.
The above pollen profiles are undatedbut cores from Coulererach(Carp, 1984;
Edwards, in press), Gruinart Flats (Edwards, 1998) and Loch a'Bhogiadh
(Edwards and Berridge, 1994) on the Rhinns of Islay, and a core from Sorn
Valley, Newton (McCullagh, 1991) have been radiocarbon-dated.The pollen
23 from theseshows an earlypostglacial open environment of grasses,herbs and birch which is supercededby the expansionof hazel.The datefor the hazel expansionat GruinartFlats is given as917CL+50 BP (8380cal. BC) (Edwards,in press)which compareswell with datafrom the southernHebrides and dates suggestedby Birks (1989)and Boyd andBoyd (1990).The publisheddate for the Corylusrise at Loch a!Bhogaidh of 10,910±450BP (10,740cal. BC) is too old andunlikely to be correct(Edwards and Berridge, 1994; Sugden and Edwards,in press).Although the Corylusexpansion on Colonsayis relatively early (9850+110BP [9420cal. BC]) (Andrewset al., 1987),elsewhere to the north of Islay andColonsay the Corylusexpansion is recordedas later (sections 1.3.3,1.3.4,1.3.5).
All the pollen data from Islay show that hazel was a major component of vegetation throughout the early Holocene. There are however some temporary reductions recordedin the hazel curves. At Loch a'Bhogaidh a reduction datesto c. 7500 BP (6330 cal. BC) (Edwards and Berridge, 1994) and a similar reduction in the Som Valley profile occurs c. 7000 BP (5870 cal. BC) (McCullagh, 1991).
On Islay, Alnus pollen increasesto its rationallimit asother arboreal pollen, including Corylusavellana-type, declines (McCullagh, 1991; Edwards and
Berridge,1994; Bunting et A, in press),although there appear to be localised differencesbetween some of the cores.Calluna vulgaris is betterrepresented in the Loch Gonn profile (Buntinget al., in press)than elsewhere but decreasesas
Abius expands.The pollen datafrom BealachFroige, albeit undated,show low amountsof Corylusand Alnus pollen alongsidehigh levelsof Poaceaeand
Cyperaceae(Edwards, in press).While this assemblagemay representlater
24 Holoceneenvironments, at Loch a!Bhogaidh on the Rhinnsof Islay, arboreal pollen percentagesat this time arelow andthose of Callunavulgaris high. Carp
(1994)also noted that pollen from sedimentsdated to 401G±75BP (2500cal.
BC) at Loch Gorm containedhigh amountsof Corylusand Betula pollen when by this time thesetaxa, would be expectedto be muchreduced. Although it is suggestedthat the pollen derivesfrom re-workedolder sediments,it is not inconceivablethat pocketsof woodlandremained undisturbed on Islay until much later in the Holocene.
It may be assumedthat vegetation on Jura was similar to that on Islay for the early Holocene period, but there is little pollen evidenceavailable to substantiatethis. Pollen from Bird Loch, Jura provides a generalpicture of
Holocene vegetation (Mercer, 1970; Edwards, in press). Pollen counts however were low (ISO grains) and the profile was not dated (Mercer, 1970). Pollen in the lower levels is mainly Betula and Corylus, which reducesas Alnus and
Ericaceaeexpand and suggeststhat the profile pre-datesc. 6500 BP (5460 cal.
BQ. All arboreal pollen is greatly reducedin the later profile as Poaceaeand
Cyperaceaeincrease. Other pollen analyseson Jura were undertakenon a pit fill as part of the archaeologicalinvestigations at Lealt Bay, but the pollen is likely to post-datethe Mesolithic (Mercer, 1970).
1.3.2 Oronsay and Colonsay
Oronsay itself presentsfew opportunites for pollen analysis as much of the island is covered in sand.A small mire has been sampled for pollen by Andrews et al. (1987), the baseof which was radiocarbon dated to 8750±85 BP (7800 cal.
BC) where the majority of pollen recordedwas Betula and Corylus. This is likely to representbirch-hazel scrub, pockets of which remain on the islands
25 today(Section 1.1.4). From c. 8000BP to c. 3000BP (6190-1200cal. BC) thereis an hiatusin the peatand no pollen contemporaneouswith the middens on Oronsaysurvives. Andrews et al. (1987)also investigated a sectionof marinesilts 50 cm deepfrom an inter-tidalbasin on the Strandat Oronsay.
Corylus,Quercus, Alnus andBetula pollen waspresent, but the shortdepth of the profile andthe absenceof radiocarbondating make it impossibleto correlate with othersites, although Andrews et al. (1987)suggest a dateof c. 6500BP (5460cal. BC) in view of the amountof Alnuspollen present.
The only published pollen data from ColonsayarefromLoch Cholla.to the south eastof the island (Andrews et al., 1987).Tht;; se display the usual suite of taxa,for the early Holocene of Corylus, Betula, Poaceaeand Cyperaceae.The hazel expansionoccurs shortly after 9850+110 BP (9420 cal. BC) and is comparable with evidencefrom Skye (Birks, 1973; Walker, 1984) for this event. Although not identified as such by Andrews et al. (1987), a reduction in the hazel profile shortly after 7870+80 BP (6800 cal. BQ may reflect Mesolithic human activity
(Edwards, in press) and further pollen analysesfrom Loch Cholla and the surrounding peat are currently underway (Boresjza,pers. comm.). The original pollen investigation suffcrs;from the low resolution sampling common in other
Inner Hebridean pollen profiles though radiocarbondates are available.
1.3.3 Mull
Mull is the secondlargest island of the Inner Hebrides and has been the focus of palynological investigations by Lowe and Walker (1986a, 1986b) and Walker and Lowe (1985,1987). Pollen analysis of severalcores (Fhuaran, Coire
Clachach and Tomess) from Glen More (central Mull), Mishnish and Beinn
Reudle (northern Mull), Gribun (southwestMull) and Loch an t'Suidhe (Ross of
26 Mull, southwestMull) were undertakenby the above authors.Most of the investigations concentratedon the lateglacial environmentsand the Lateglacial - Holocene transition for which radiocarbondates were obtained, though low resolution Holocene pollen records for all sites were also published (Lowe and
Walker, 1986a;Walker and Lowe, 1985; 1987).
The pollen profiles for the Lateglacialand early Holocene are similar for all sites.Betula is the main arborealpollen type presentand with Juniperus, Poaceaeand Cyperaceae, reflects an environmentof openscrub and grassland. Betulapollen increasesas Empetmm nigmm and Juniperuscommunis are reduced.However, there are local variationsin the pollen curves.For example, the Beinn Reudlepollen assemblagecontains far IýssBetuld andJuniperus pollen thanthe othersites (Lowe andWalker, 1986a).Percentages of
Lateglacialand earlyHolocene herbaceous taxa alsovary betweensites. For example,Filipendula pollen is recordedin greateramounts in the early Holoceneat Coire Clachach(Walker and Lowe, 1985)than in any of the other profiles.
While all cores show fairly uniform pollen sequencesrepresenting the standard
Holocene successionsof Corylus, Quercus, Ulmus and Alnus, local variations in the pollen taxa are again evident. Calluna vulgaris and other ericaccouspollen is better representedat Mishnish and Beinn Reudle (Lowe and Walker, 1986a), and whilst at most sites Calluna pollen expandsfollowing the major expansion of Alnus, at Beinn Reudle the Calluna expansionpreceeds that of Ablus (Lowe and Walker, 1986a).Only one radiocarbon date is available for the above changes,and this datesthe Corylus expansionat Torness to 876L+140 BP (7900 cal. BQ (Walker and Lowe, 1985). This is late comparedwith most of the
27 evidencefor Islay andis possiblya local phenomenondue to localised environmental conditions. However, anomalously late datesfor the Corylus expansionalso occur on Skye(Birks, 1973;Vasari, 1977;Walker, 1984).
All sites on Mull show a reduction in arborealpollen and expansionsin Calluna vulgaris and a variety of herbaceoustaxa in the later Holocene sediments.The degreeof expansionof ericaceousand herbaceouspollen differs between sites but there is an overall indication of a reduction in tree cover and an expansionof moor and grassland(Walker and Lowe, 1985; Lowe and Walker, 1986a, 1986b;
Walker and Lowe, 1987).
1.3.4 Rum
Thereare few pollen datafrom the islandof Rum.A corefrom Long
Loch, Rum providesa recordof earlyHolocene environments characterisedby an expansionsof Betula,Juniperus and subsequently Coryluspollen (Ford, 1976cited in Parish,1990), but no otherpollen evidencefrom the islandis asearly asthis. The restof the palynological investigationson Rum havebeen connected to the excavationof the site at FarmFields, Kinloch (Hironsand Edwards, 1990). This sitehas a basal radiocarbondate of 7800+75BP (6640cal. BQ at which time arboreal pollen is poorly representedand the assemblageis dominatedby
Poaceae,Cyperaceae and Filipendula. Corylus and Salix arepresent in low amountsand an expansionof Coryluswhich probablypost-dates the rationalhazel rise occursslightly later.Betula and Pinus pollen are also recorded.
28 The expansionof Alnuspollen in the FarmFields monolith is datedto 6430+90BP (5400cal. BQ andalder is hardly recordedprior to this despitethe amountof pollenpresent in the earlierdated contexts from the excavationarea to the east(see below). Subsequentfluctuations in the Alnus curveand other data, especially marked charcoal peaks, are interpretedby Hirons andEdwards (1990) as being possible reflections of humanimpacts. Alnus pollen occursin a pollenprofile from the
Kinloch Glento the westof FarmFields (Parish, 1990) but its percentage valuesare very low anderratic (as are those of otherarboreal taxa) and do not displaythe perturbationsseen in the FarmFields profile.
After c. 3950 BP (2540 cal. BC), both Alnus and Corylus pollen decline
(Hirons and Edwards, 1990).A similar decline is seenin the Kinloch
Glen pollen profiles (Parish, 1990). At Farm Fields, the decline in woodland is accompaniedby an expansionin Poaceaeand herbaceous pollen, notably Potentilla-type, and the first cereal-typepollen occurs.
Pollen and macrofossil analysis of contexts from the Farm Fields excavation were also undertakenand although most contained little or no identifiable pollen, other samplescomplementthe environmental reconstruction of the Farm Fields area(Moffatt, 1990). Two of the contexts analysedwere radiocarbon-dated.The earlier context datesto
7925±65 BP (6830 cal. BC) and this precedesthe basal date of the Hirons and Edwards (1990) monolith. Poaceaeand ericaceouspollen have the highestrepresentation in the sample,although some Betula and Quercus pollen is present.The secondsample had a later dateof 7850150BP
29 BC) (Moffatt, 1990). This again shows similarities to the pollen from other contexts at the site, the main componentsof the assemblagebeing Alnus,
Poaceae,Cyperaceae, Ericales and a range of ruderal pollen types. It is likely that all the pollen from the excavation site derived from very local sources.
Cereal pollen was also detectedon pottery from Neolithic contexts from the excavation(Moffatt, 1990).This suggeststhat humanagency may havebeen responsible for the final reduction in woodland speciesat the site.
30 The pollen data from the Farm Fields area,Kinloch, Rum provides a very different picture of past vegetation cover comparedto pollen records from other
Inner Hebridean islands. Corylus avellana-type pollen is recordedin much lower amounts,as is arborealpolien in general,and ericaceousand herbaceous pollen are recordedin greaterquantities earlier than elsewhere.The relative paucity of woodland representationin the pollen at the site may be a reflection of the unsuitability of the environment of Rum for the growth of trees, or may be linked to early and continued human occupation.
1.3.5 Skye
Skyeis the largestand most northerly of the islandchain andpollen investigationshere have been undertaken by Vasariand Vasari (1968), Birks (1970,1973,1993),Vasari (1977), Williams (1977),Birks andWilliams (1983) and Walker andLowe (1990,1991a,1991b). Many of the studiesfocus on the
Lateglacialto Holocenetransition and indicate that Poaceae,Cyperaceae,
Runiex,Thalictrum, Empetrum and Betula nanawere the major taxapresent in the Lateglacial.In the successionto early Holocene,, Juniperuscommunis andBetula are succeededby Corylusavellana and Salix. Despitethe similarities betweenprofiles, thereare local differencesin the pollen assemblages.Walker
31 and Lowe (1990) note that Pinus pollen is better representedin a core from
Elgol than elsewhere,and here Juniperus communisand Salix pollen percentagesare low.
Of the main Holocene vegetation successions,the focal point of interest has been that of the expansionof Corylus avellana-type pollen where this has reachedits rational limits. Radiocarbondates for this event are available on a number of cores and in general this occurs around 9500 BP (8900 cal. BQ
(Birks, 1973; Vasari, 1977; Walker and Lowe, 1990). However, it is noted by
Vasari (1977) that severaldates for the hazel rise on Skye are anomalous.The hazel rise at Loch Fada datesto 7500+120 BP (6340 cal. BQ (Birks, 1973) and at Loch Choira Ghobhain to 8650+150 BP (7840 cal. BQ. It was suggestedby
Vasari (1977) that thesedates could be wrong but subsequentdates on cores from Coire Laggan and Glen Varragil of 7780+180 BP (6660 cal. BQ and
8175+85 BP (7250 cal. BC) respectively (Walker, 1984), suggestthat there is local variation in the expansionof hazel on Skye as witbLateglacial taxa mentioned above, and this may be linked to local environmental conditions.
Human activity inferred from the pollen diagrams from Skye relatesto the last half of the Holocene only (Vasari and Vasari, 1968; Birks and Williams, 1983; Birks, 1993).
1.4 Reasons for further pollen analysis in the Inner Hebrides
The above demonstratesthat some of the Inner Hebridean islands are rich in archaeologicalevidence for early Holocene human occupation and it is unlikely that all islands have both Mesolithic and Neolithic finds, even if they remain to be discovered.There is also an abundanceof sites suitable for
32 palaeoenvironrnental,particularly archacopalynological, investigations and there is consequentlya high probabilitythat earlyprehistoric human impacts will be detectablein the pollen recordsfrom the area.This meansthat thereis the potentialfor the discoveryof a formerhuman presence where there is currently little or no archaeologicalevidence available. Indeed, there have been frequent inferencesfrom pollen dataof the OuterHebrides, Shetland and Orkney of early, especiallyMesolithic, occupation (Bennett et A, 1990;1992; Bunting,
1994;Edwards, 1996; Fossitt, 1996). From theseislands only Orkneyprovides evidenceof Mesolithic occupationin the form of stonetools which are largely unprovenanced(Saville, 1996).The validity of an anthropogenicinterpretation in suchinstances may be testedmore confidently by comparisonwith pollen datafrom sitesin proximity to areasof known humanoccupation.
Pollenanalysis of siteswithin the Inner Hebridesis not only importantin terms of archaeologicalapplication. The delimitationof vegetationalsuccessions is alsoimportant to the understandingof pastand changing environments in the islandsand in northernEurope as a whole (Huntleyand Birks, 1983).It was mentionedin Section1.3 that manypollen diagramsfrom the Inner Hebridesare poorly dated,and estimates for vegetationaltransitions have been made by comparisonwith diagramsfrom elsewherein Scotlandand the British Isles.
This may meanthat pollen isochronesfor the InnerHebrides are inaccurate.
Indeed,isochrones suggested by Birks (1989)rely on datafrom only 8 sites,and
5 of theseare from Skye.Modifications to. previously suggested isochrones for Corylusand Ulmusbased on informationfrom a greaternumber of morerecent andradiocarbon-dated pollen diagramswere made by Whittingtonand Edwards
(1997)but therestill remainsa needfor more detail to producea more accurate picture of palaeovegetationalsuccessions.
33 The arrival andexpansion of Corylusavellana in the British Islesin the early Holocenehas long beenthe subjectof debate.Original estimatesfor the colonizationof the Inner Hebrideanislands by this taxonsuggest that it was establishedby 9500BP (8900cal. BQ (Birks, 1989).Whilst it hasusually been assumedthat climatewas the controllingfactor in the spreadof this taxon, Huntley (1993)suggests that otherfactors including soil stability andhuman assistanceshould be considered.The topographicvariability of the Inner Hebrideanislands has been mentioned in Section1.2. Geomorphology and topographywill obviouslyhave influenced soils in the past,hence there may be considerabledifferences between islands (and indeed within islands)for the timing of their colonizationby hazeland other species, as suggestedabove. Thereis clearly a neednot only for morepollen diagramsfrom the Inner
Hebrides,but for moreradiocarbon-dating of profiles.
Peaksin microscopiccharcoal have frequently been interpreted as the productof humandomestic activity (Edwards,1990a) or the burning of vegetation(Smith,
1970;Simmons et al., 1981). However,there may be non-anthropogenic explanationsfor high levelsof microscopiccharcoal (e. g. reworkingof sediment;increase in airborneparticulates due to changein climateand weather patterns[Edwards, 1990a; Tipping, 1996],and lightning, strikes[Patterson et al., 1987;Edwards, 1990a)), which may alsohave implications for the interpretationof pollen diagrams.Most of the pollen datafrom the Inner Hebrideslack associatedmicroscopic charcoal quantification, yet this would appearto be a necessarycomponent of palaeoenvironmentalinvestigation in orderto contributeto the evidencefor the influencesof climatic, ecologicaland human events.
34 Palynologicalinvestigation has been undertaken on manyof the Inner
Hebrideanislands, but thereare limitations to the interpretationof thesedata.
The problemsof low samplingresolution, lack of quantificationof microscopic charcoaland insufficient radiocarbon dates have already been mentioned, as well asa concentrationon Lateglacialrather than Holocene environments. Henceadditional pollen profiles combinedwith microscopiccharcoal data, closeresolution sampling and radiocarbon dates would enhancethe palaeoenvironmentalrecord for the islands.In addition,the multiple coring of singlesites has a greaterpotential than a singlecore to producean accurate reconstructionof local vegetationthrough time (Edwards,1983). Intra-site variationsin pollen profiles may aid interpretationof what is regionaland local, or of naturalor humanorigins. Multiple profiles will alsoincrease the probability of detectingsmall scalelocal impactsand may suggesta spatial biassingof theseevents. An elaborationof this is given in Section2.1.
35 CHAPTER 2- METHODS AND SITE SELECTION
2.1 Methods
In discussing the requirementsof further pollen work in the Inner Hebrides in the previous chapter, it was mentioned that certain methods should be employed, the main methodsbeing;
1. Multiple coring of single sites.
2. Close resolution pollen sampling.
3. Microscopic charcoal quantification.
4. AMS radiocarbon dating of sedimentprofiles to provide a chronology and to allow reliable intra- and inter-site comparisons.
These are used here and other methods employed are:
S. Loss on ignition analysis.
6. Low power (x100) scanningof complete 1cm3 samplesto maximise cereal pollen detection.
7. Classification and quantification of damagedpollen.
8. Dinoflagellate analysis of some sediments.
2.1.1 Multiple coring
Although comparisonsbetween pollen profiles havelong beenmade over large areasincluding the Inner Hebridesand islands such as Mull and Skye(cf. Birks, 1973;Vasari, 1977;Walker and Lowe, 1985,1987,1990,1991a,1991b; Lowe andWalker, 1986),multiple profile studiesof singlebasins are rare. Three pollen profiles areavailable from Loch aýBhogaidh,Islay (Agnewet al., 1988;
36 Edwardsand Berridge, 1994) and two from MachrieMoor on Arran (Edwards andMcIntosh, 1988;Robinson and Dickson, 1988).Multiple profile studies may be considerednecessary for projectswhere small scalelocalised changes in pollen profiles needto be detectedas, in general,studies relying on a singlecore will not providea goodrepresentation of very local vegetationcommunities and their dynamicsthrough time. Centrallyplaced cores may reflect wider regional vegetationchanges due to the taphonomicfactors influencing pollen dispersal anda differentrange of taxamay be recordedfrom differentparts of a site due to local ecologicaldifferences (cf Turner, 1975;Edwards, 1982,1983). In addition,early Holocene human impacts are likely to havebeen short-lived and spatiallyconstrained, hence their impacton the pollen recordmay not be wide- ranging.When the taphonomicvariables affecting pollen input to a lake or bog arealso taken into account,then a singlecore may not be representativeof eitherregional or local vegetation.
Multiple coresfrom a peatbasin will normally incorporatetemporal and spatial differencesin the pollen profiles (Turner,1975; Edwards, 1983; Turner et al.,
1993)but it is often assumedthat sedimentfocusing in lakesresults in an overall smoothingof the fossil pollen record,and that lake-derivedpollen profiles haveuniform characteristics(Jackson, 1994). Evidence for the behaviourof pollen in lakesand peat bogs is givenbelow, in additionto evidencesupporting multiple profiling methods.
2.1.1.1 Pollen distribution and multiple coring In lake sediments
Studies on the dispersal and influx of fresh pollen into lake basins have shown that wind direction, wave action, pollen grain morphology and density, lake morphometry, seasonalityand thermal stratification may influence initial pollen
37 depositionat the water-sedimentinterface (Pennington, 1947,1973,1979; Davis et al., 1969;Brush and Brush, 1972;Davis, 1973;Davis andBrubaker, 1973;
Bonny, 1978)but sedimentfocusing is usuallyassumed to negatesubstantial differencesresulting from the effectsof thesevariables (Davis et al., 1969; Davis andBrubaker, 1973; Jackson, 1994). Nevertheless, Davis et al., (1969), Edwards(1983) and Bonny et al., (1984)have queried the reliability of a single coreas an adequatesample of a pollen recordin a lake.While Davis (1973) showedthat pollen is not mixed independentlyof sedimentonce deposited and thereforemay not be subjectto the samevariables as freshpollen, it appearsthat the extentof sedimentfocusing is dependentupon the morphometryof the lake basinand this may occurto differentdegrees in different lakes.Indeed, morphometrymay influencewhether or not sedimentmixing andfocusing occursat all (Davis, 1973;Chen, 1987). Sediment focusing is greaterwhen seasonalthermal stratification of lakewaters occurs which createswaves at depthand annualwater turnover. In non-thermallystratified lakes, there is little or no turnoverand sediment focusing is reducedor conceivablyabsent. Large lakesmade up of smallerindependept basins are unlikely to becomethermally stratified(e. g. Lake Windermere,(Pennington, 1947]).
Wheresediment focusing is low or absent,then the factorswhich affect the initial distributionof pollen over lake surfacesediments are of greater significance.The pollen contentof samplesof sedimentunits from within a lake were statisticallytested for uniformity by Davis et A (1969)and significant variationsin pollen contentwere detected in strataonly a few metresapart.
Davis andBrubaker (1973) have shown that smallpollen grainsof low density aremore susceptibleto wind-drivencurrents than heavier pollen. Davis and Brubaker(1973) also showedthat aerially derivedpollen is likely to be greater
38 in surfacesediments at the centreof a lake asit is delivereddirectly to the watcr body andmay sink quickly, whereaslocally derivedpollcn andstream-bome inputswill be greaterin the sedimentat a lake edge.Pennington (1947) recorded greateramounts of Pinuspollen in the northernbasin of lake Windermerethan in sedimentsat its perimeter.As Pinusgrains arc particularlybuoyant in air, this suggeststhat the lake centrewas receiving greater amounts of airbornepollen. Pinusmay not havebeen redistributed quickly by wave actionif water movementwas low andrapid sinkagemay haveoccurred due to the the air sacs of the pollen grainsfilling rapidly with water.Chen (1987) also noted that pollen distributionin a non-stratifiedlake will be affectedby the pollen source.
Pennington(1973,1979) found that largerlakes with no inflow streamsreceive a greateramount of regionalpollen whereaslakes with inflow streamsmay receivemore local pollen althoughif inflow streamsare many anddrain a large catchnientthen pollen from a largearea would be transportedinto lake waters and containedwithin the fossil pollen recordof its sediments.Variations in pollen influx rate arounda lake edgemay over-representstream-side plants in the pollen record(Bonny, 1978)and although proportions of fossil pollen betweencores may be similar dueto sedimentfocusing, concentrations and influx ratesmay differ (Bonny et al., 1984).It may be arguedthat a knowledge of the differentpollen concentrations,pollen influx ratesand sediment accumulationrates within a lake basinis necessaryfor an understandingof the dynamicsof a systemthrough time, includinginterpretation of the pollen curves and definition of a relatedcatchment area (Bonny et al., 1984).Indeed,
Pennington(1972) suggests that climatic andenvironmental changes through time may alter the throughputof waterin a lakebasin and thus sediment focusingand pollen influx.
39 Considerationmust alsobe givento the pollen catchmentareas represented by the airborneand strcam-bomepollen components.As airbornepollen is more likely to derivefrom a wider area,then local variationsin lake-sidevegetation will be diluted in the pollen rain from a wider areain centraliscdcores. Stream- bornepollen representinga smallercatchment area is more likely to reflect lake- sidechanges, although multiple lake edgecores will perhapsbe neededto detect these(Bonny, 1978),particularly if the instigatorof ecologicalchange, such as humanoccupation or the grazingof animals,moves through time (cf. Edwards, 1982).
Much of the aboveevidence is derivedfrom studieson freshpollen within lake basins,and there are few studieson the variationsof fossil pollen spectrawithin a lake.Edwards (1982,1983) and Edwards and Thompson (1984) suggested that singlecores from lakesmay missdetail andthis is borneout by the resultsof Whittingtonet al., (1991 a) from Black Loch, Fi fe. An eventregistered in zone
BLIVd on a core from the loch edgeis not apparentin any othercores from the site andthe suggestionis madethat this zonehas registered the effectsof localizedMesolithic impacts.A similar effectwas recorded by Bos andJanssen (1996)when pollen investigationsof two cores,one from deepwater sediments in a lake andthe otherfrom lake edgesediments, were undertaken. Despite providing the greatestdepth of resolution,the deepwater corelacked the detail recordedat the lake edge.The shortercore contained reductions in Pinuspollen
(locally present)which appearedto be cyclical andcoincided with Palaeolithic occupationsindicated by the local presenceof Palaeolithicartcfacts.
40 Therecan be little certaintyabout which variablesmay haveaffected pollen and sedimentdistribution in the past(Edwards and Whittington, 1993)as lake morphometry,the thermalregime of lakewaters and wind directionscan change throughtime, andhiatuses in sedimentationwithin lake basinscan also occur, thus interruptingthe pollen record(e. g. Bennett,1989; Whittington et al., 1991; Edwardsand Berridge, 1994). Such factors mean that the generalrules usually followed for corelocation may be inadequate(Jacobson and Bradshaw, 198 1), particularlywhen the detectionof small scaleimpacts in the pollen recordis the investigator'saim.
2.1.1.2 Multiple coring of peat
It has long been recognisedthat fossil pollen from peat cores may vary spatially depending on the core position within the peat body (Turner, 1975). This is the result of pollen dispersalmechanisms. Large bogs with open centrcs may receive greater amountsof airborne pollen which is likely to derive from trees and shrubsin the wider catchmentarea, whereas bog edgesmay trap more local pollen if screeningby local woodland occurs (Tauber, 1965,1967; Turner,
1975; Janssen,1977; Jacobsonand Bradshaw, 1981; Edwards, 1982,1983).
Therefore, in peat which was previously well wooded, it is probable that most fossil pollen will have originated close to the coring site (Tauber, 1965,1967; Edwards, 1991).
Caseldine(198 1) hasshown from surfacepollen studiesthat woodland surroundinga bog actsas an effectivefilter to mosttypes of pollen derivedfrom pasturelandfurther afield. His work alsosupports the (distancedecay) theory that higherratios of woodlandpollen occur at a bog edgeand reduces in proportionto local pollen towardsthe centreof the bog. Tinsley and Smith
41 (1974) showed a similar effect on Quercuspollen which decreaseswith distance from the woodland edge.
Caseldine(198 1) alsodemonstrated that largerPoaceae pollen grainsare concentratedat the bog edgeand that grain diametersdecrease with distanceto the centreof the bog. This meansthat althoughsome pollen typesmay permeate the woodlandboundary, there is preferentialdistribution of the grainsby size.
Edwards(1982) suggested that this is anothercomponent is the problematic detectionof cerealpollen, which is largerand heavier than most othertypes of Poaceaeand is not widely dispersed.
Multiple profile studiesof fossil pollen in peatdo not alwaysshow differences.
For example,profiles from Bloak Mossand Kennox Moss, Ayrshire were intendedto demonstratethe variability in pollen contentof coresfrom a single bog but, the pollen recordfrom all threecores was remarkablysimilar (Turner, 1975).The coreswere positioned approximately 500 m.apart and in central positionsand aretherefore more likely to havereceived a largeregional pollen componentcompared to coressituated at the bog edge(Edwards, 1982 and cf
Janssen,1977; Cascldine, 198 1). Othermultiple profile studiesin peatmay providerecords of'local differences,but the coresare often from separate basinsand kilometres apart (e. g. RannochMoor, (Walkerand Lowe, 1981]; coastalplain of north Ayrshire [Boyd, 1988]).
As smallerbogs may receivea lower componentof aerialpollen they may also providerecords of very local vegetationchanges. Studies at CrossMere, Ayrshirewere undertaken Turner (1984) where five coresno morethan 500 ra apartwere analysed.Whilst certainsimilarities were displayed in the pollen
42 records,statistically significant differences were also noted in someof the profiles.These were interpreted by Turneras the reflectionsof different local environmentalconditions. Smith and Cloutman'(1988) also noted broad scale similaritiesbut many local differencesin a studyof sixteencores from a small bog of 250 m in area,Waun-Fignen-Felen in SouthWales. Simmons and Innes (1988)also advocated fine resolutionmultiple profile analysisof peatbogs, particularlywhere the aim is to detectsmall scale,local changesin vegetation.
There are therefore many reasonsfor the differential dispersal of Pollen over a bog. Screeningout of certain pollen types by local vegetation, pollen size and local ecological differences will all influence pollen distribution as may basin size and the relation of pollen sourceto the basin. Multiple coring of a single peat-filled basin will thereforeprovide the most complete record of changesof local vegetation through time, provided cores are not confined to the centre of the basin, and include sites at the bog edge.This will increasethe probability of the detection of early prehistoric human impacts, as these are likely to have produced localised effects.
2.1.2 Microscopic charcoal: problemsfor distribution and taphonomy
The occurrenceof microscopic charcoal in pollen preparationsis usually interpreted as an indication of fire in the history of a pollen catchment(which may, ormay not, correspondwith the charcoal catchment area [cf. Edwards and
Whittington, in press]). Fires may be causedby various natural meansincluding lightning strikes (Pattersonet al., 1987) and by accidental or intentional burning by humans (cf. Edwards, 1990). Differences in the types of material combusted in (e.g. hardwood, softwood, grass) and in fire intensity may produce charcoal varying sizes and amounts(Patterson et al; 1987; Moore, 1996).
43 Microscopiccharcoal may be representativeof local fires, but thereare
argumentsthat it may derivefrom further arield.Patterson ct al., (1987)
summarizedthe variablesinfluencing charcoal production and distribution and
suggestthat in generalterms, small particles of charcoalare spreadover a wider areathan largeparticles. Clark (1988)also showed that smallerparticles of the sizeusually seen in pollen preparations(5-20 pm) are far more likely to become
airborne.This hasled, to argumentsthat microscopiccharcoal in pollen preparationsderives from a wide areaand that peaksmay be a reflectionof increasesin fire distributionand periodicity due to naturalfactors such as climatic change(Tipping, 1996).
It hasalso been suggested that humanagency (accidental or intentional)may
havebeen responsible for fire throughoutprehistory in orderto promotefresh
growthof vegetationor to drive game(Smith, 1970;Jacobi et al., 1976;
Simmonset al., 1981). Edwards(1990) received evidence where some increases
of Alnusglutinosa pollen were coincidentalwith increasesin microscopic
charcoal,but in many casesthe Alnus pollen andmicroscopic charcoal profiles did not appearto be interdependent.Similarly, Edwards(ibia) found that increasesin Corylusavellana-type pollen werenot accompaniedby increasesin microscopiccharcoal although an anthropogenicinterpretation has often been
ascribedto this phenomenon(Smith, 1970,1984).Even when microscopic charcoalis recordedin pollen deposits,there can be no direct evidenceto
confirm that this wasproduced by local humanactivity. Tipping (1996)
acknowledgesthat wherearchaeological evidence is present,particularly when closeresolution and spatial sampling methods are employed in palaeoecological
analysis,then links betweenfire episodesand human activity may be drawn.
44 This doesnot necessarilymean that charcoalabundance is a direct result of the direct burningof vegetation.An abundanceof domesticfires may also increase microscopiccharcoal, both in the atmosphereand in soils, andthis may reach pollen sitesif local, or from furtherafield whenwoodland surrounding a site, which may haveacted as a screento microscopiccharcoal, has been removed (Edwards,1990).
Whilst airbornecharcoal may be locally producedor may be derivedfrom sourcesfurther afield, microscopiccharcoal may reacha lake or peatbog by othermeans, such as inwash by streams,erosion due to increasedrunoff caused by a reductionin surfacevegetation (Patterson et al., 1987)and landslips (Whitlock andMilspaugh, 1996). In someinstances the fire may occuron a peat bog itself (Jacobiet al., 1987;Caseldine and Hatton, 1993;Turner et al., 1993), or at a lake edge,when burnt vegetationmay fall directly onto sediments
(Whitlock andMilspaugh, 1996). Clark (1988)proposed that the small sizesof charcoalusually seenin pollen slidesare unlikely to derivefrom stream-and surface-bornesources, and that the largerparticles transported by thesemethods arebetter detected by thin sectionanalyses of sediments.He alsocited evidence which showsthat little charcoalis removedfrom its sourceby erosionand runoff. An opposingview is providedby Pattersonet al., (1987)and studies of by sedimentsfrom variouslake catchmentsin the YellowstoneNational Park
Whitlock andMilspaugh (1996) showed that erosionand runoff aremajor mechanismsfor the introductionof charcoal,but particularlymacroscopic charcoal,into lake sediments.Evidence from Patterson(1978) indicated that charcoalin lake sedimentsis highernear inlet streamsand quantities of microscopiccharcoal are also known to be higherin minerogenicrunoff.
45 Increasesin charcoal in laminated lake sedimentshave been correlated with fire eventsrecorded as fire scarson treesin North America(Swain, 1973;Clark,
1988)and northeastern Europe (Tolonen, 1983; Lchtonen and Huttunen, 1997).
Whilst fires in a lake catchmentare normally representedby increasesin macroscopiccharcoal (e. g. Clark, 1988;Lehtonen and Huttunen, 1997), increasesin microscopiccharcoal (such as that countedin pollen preparations) do not alwaysoccur (Swain, 1973; Clark, 1988).Swain (1977) suggeststhat the samplingresolutions used in pollen analysismay preventpeaks in microscopic charcoalbeing detected,but taphonomymay alsobe a cause.Clark(1988) remarksthat microscopiccharcoal is far more likely to derivefrom the fallout of airborneparticles from a wide areawhereas macroscopic charcoal will travel lessfar andis more likely to erodeinto a lake.Nevertheless, increases in microscopiccharcoal which correlatewith known fire eventsmay still be detectedin pollen preparations(Cwynar, 1977).
It cannotbe assumedthat microscopiccharcoal is subjectto the same distributionprocesses as pollen, asit hasa differenttaphonomy and behaves differently in air andwater (Clark, 1988;Edwards and Whittington, in press).
The taphonomyof microscopiccharcoal may alsopromote its differential distributionin lake sedimentsand peat bogs, and charcoalrecords from both environmentsare discussed below.
2.1.2.1 Microscopic charcoal in lake sediments
Once charcoal is incorporated into lake sediments,then any sediment focusing which occurs should act on the charcoal content of the sediment in a similar manner to the pollen (refer to section 2.1.1.1). The potential for the differential distribution of pollen in lake sedimentssuggests that microscopic charcoal may
46 not be evenly distributed within a lake and may appearin variable quantities in different profiles from the basin (Bradbury, 1996).
The mostrecent and comprehensive study of fire eventsand charcoaldistribution in lake sedimentsis that of Whitlock andMilspaugh (1996).Whilst it includedan analysisof spatialdistribution it alsoinvolved analysison a short-termtemporal scale. The studyquantified macroscopic charcoalin a numberof lakesin the YellowstoneNational Park, North America, associatedwith local fires in 1988,and incorporates lakes of differentdepths with both burnedand unburned catchments where sampling for charcoalwas undertakenalong transects following wind directionsat the time of burning.In additionto the informationregarding the methodsby which charcoalmay enter lake sediments,their studyis invaluablein identifying someof the factorswhich may influencecharcoal distribution within the lake itself. Whilst the study for relatesto macroscopiccharcoal, different size ranges were quantified and all the sizeranges charcoal distribution patterns appeared to be the same,although a greateramount of small sizedfragments were present in all samples.Despite the problemsregarding the taphonomyof microscopiccharcoal mentioned above(Section 2.1.2) it is possiblethat microscopiccharcoal originating from fire eventswithin a lake catchmentmay registera similar signalto macroscopic charcoalentering the lake.
Conclusions drawn by Whitlock and Millspaugh (1996) are summarizedas follows: -
1. Charcoal was incorporatedinto sedimentsof lakes with unburned catchments suggestingan airborne fallout.
47 2. Deep lakes show higher charcoal concentrationsin the shallow littoral zone
(even when the catchmentis unburned and charcoal input is aerially derived).
3. Deep lakes with unburnedcatchments show no increasein charcoal in deep water sedimentsthrough time in accordancewith fire events.
4. Shallow lakes show no clear spatial patterns in charcoal distribution, but show increasesthrough time linked to fire events.
S. Peaksin charcoal (i. e. increasesabove the background noise) are representativeof actual fire events.
The variationsin charcoaldistributions in lakeswithin YellowstoneNational Park suggestthat multiple coringof lake sedimentsis imperativeforachieving a representativerecord of local fire events,particularly when lakesare infilled and the previousmorphology of the basinin not completelyknown. Indeed,the resultsof Whitlock andMilspaugh (1996) indicate that the bestcharcoal record, particularlyfor local fire events,may be detectedin lake edgesediments. Edwardsand Whittington (in press)also suggest that it may be possibleto infer the distanceof a fire or seriesof fire eventsfrom the coring site usingmultiple lake profile pollen andcharcoal data, although inferences from current palaeoevironmentaldata are inconclusive. As singlecores from lake sitesare usuallycentrally placed (and may thereforereceive a greateramount of airborne particulates[ Section2.1.1.1 ]), the variationsin charcoaldata in the above studiesindicate a needfor multiple profiles in orderto elucidateinterpretation of the fossil charcoalrecord.
48 2.1.2.2 Microscopic charcoal in peat bogs
Unfortunately, most of the studieson the taphonomy and distribution of microscopic charcoal have been made on lake sediments.There is little information about the behaviour of charcoal in peat (Pattersonet al., 1987). It may be assumedthat peat bogs receive a generalbackground of aerial fallout and that local fire eventswill register by increasingthe input of aerially dispersedcharcoal to bogs. In topographic mires charcoal input may also be increasedby runoff and stream deposition. Spatially and temporally diverse charcoal representationsin multiple profiles from North Gill, Yorkshire, suggest that very local fires increasemicroscopic charcoal at the site of burning (Turner et al., 1993). In some cases,where intenseburning occurs directly on a peat bog, then the burnt layers should be visible in the peat stratigraphy, and hiatuses should be apparentin the palaco'ecological record (Moore, 1975).
2.1.3 Interpretation of microscopiccharcoal in pollen preparations
A certaindegree of cautionover anthropogenicinterpretation of charcoalevents mustbe maintainedwhen archaeologicalevidence is not presentand considerationmust be given to changesin the fallout of long distancecharcoal andthe reworkingof old charcoal(Edwards, 1990a; Tipping, 1996).It is noted by Whitlock andMilspaugh (1996) that thereis currentlyno reliablemethod for the determinationof distanceof a fire eventfrom the charcoal-receivingsite and little is known aboutEuropean fire historiesand connections to climate.Most datarelating to the fire in ecologicalsystems derives from North American,
Australianand Scandinavianexamples, where climate and ecologydiffer from thosefor northwest13ritain. However, the abovesections suggest that the quantificationof microscopiccharcoal in pollen preparationsmay aid the
49 interpretation of pollen profiles. When reinforced with archaeologicaldata and consideredas part of an environmentalassemblage, then charcoal records may assistin the understandingof vegetationchanges.
2.1.4 Quantification of microscopiccharcoal
Microscopic charcoal was assessedon all levels sampledand counted for pollen from the cores in this study. It was initially intended to use the point count estimation method (Clark, 1982) but concentrationswere not sufficient to provide a statistically reliable sample size in the first few levels tested and this method was abandoned.Charcoal particles were quantified by two dimensional measurement(length and breadth) and concentrationsexpressed as total surface areaper sample volume (CM2cm-3). Charcoal to pollen ratios (Ch:P) were also calculated in order to identify false charcoal peaks arising from variations in sediment accumulation rates (Swain, 1973). Charcoal influx is defined as charcoal accumulation rates in units Of CM2CM-3 yil.
Edwardsand Whittington (in press)presented microscopic charcoal data from multiple lake profiles usingthe threemethods described above. For single profiles the curvesfor charcoalconcentration, Ch: P andcharcoal influx appearedvery similar and whereone curve increased the othersincreased.
Althoughmicroscopical charcoal has been quantified using all threemethods iii this study,Ch: P is usuallyreferred to for easeof interpretation.Comments regardingcharcoal concentrations and influx refer to uniqueevents in their profiles which may enhanceinterpretation.
50 2.1.5 AMS radiocarbon dating of pollen profiles
The necessityto comparepollen profiles from this study with each other and existing profiles within the Inner Hebridean islands and elsewhere,ýand the potential for matching pollen profile changeswith periods of prehistoric occupation near the pollen sites, demandsthat the cores should be dated.
Radiocarbon datesare also necessaryto confirm the contemporaneityof palynological.features recorded in multiple cores from a single site. Close resolution sampling, including I cm.contiguous samplesin places, meansthat
AMS datesare preferable as they are capableof providing a higher resolution than is normally the casewith radiometric determinations.
2.1.5.1Choice of levelsfor dating
The choiceof levelsfor datingwas based on featuresin the pollen diagrams wherereductions and recoveries of certaintaxa occurred, and where initial expansionsin taxa appeared(e. g. Corylusrise, Alnus rise). Dating of the latter would contributeto refinementof pollen isochronemaps and the timing of the expansionof thesetaxa within the earlyHolocene.
2.1.5.2 Processing of samples
Processingof 49 14Csamples to graphite targetswas carried out by the NERC
Radiocarbon Laboratory at East Kilbride, Scotland, under the supervision of Dr.
Charlotte Bryant. This was followed by AMS dating of 48 samplesat the
University of Arizona USF-AMS facility and of I sample by the Lawrence
Livermore Radiocarbon Laboratory. Tables of results including laboratory codes are provided for each site (Vol. 2).
51 2.1.5.3 Calibration of dates
The calibration of AMS dateswas achievedusing Calib3 (Stuiver and Reimcr,
1993; Appendix A). Calibrated dates,generally on spot dates,are provided in brackets within the text, and the main interpretationsrely upon uncalibrated datesin radiocarbon years BP. This is to facilitate comparisonswith the wider palaeoecological literature, and datesfor much of the archaeologicalevidence in the Inner Hebridean islands are also presentedin this form. The calibration of datesresults in a range of error over which the actual date may occur which makes comparisonsextremely difficult.
2.1.5.4 Extrapolation of dates and the construction of age/depth curves
The extrapolation of datesto allow the calculation of pollen influx and sediment accumulation rates for eachprofile was achievedusing TILIA (Grimm, 1991).
Linear extrapolation appearedto be most appropriateas pollen influx for published Inner Hebrideanpollen profiles has been calculated using this method
(e.g. - Edwards and Berridge, 1994) which makes fewer assumptionsof the data comparedto other methods such as cubic spline interpolation or construction of a polynomial curve (Bennett, 1994). As linear interpolation does not take into account standarddeviation errors or reversalsof dates(unlike a polynomial curve) some dateswhich appearederroneous have been excluded from calculations and are discussedin the following chapterswhere appropriate.
Pollen influx is defined as the amount of pollen present in lcm3 of sediment for 1). each radiocarbon year (= pollen accumulation rate of grains Cm73yf
Age/depth curves basedon uncalibrated,radiocarbon dateswere constructedfor eachprofile where an appropriatenumber of dateswas available.
52 2.1.6The detectionof cereal-typepollen grains 2.1.6.1 Cereal scanning
Cereal-typepollen grainsin the original corefrom Kinloch, Rum (Hirons and Edwards,1990) indicated agriculture as early asc. 3950BP (2540cal. BQ. As cerealpollen is producedin very low amounts(OConnell, 1987;Edwards and McIntosh, 1988) and doesnot travel far from its source(Vuorela, 1973;
O'Connell,1986; Hall, 1989),it was decidedto maximisethe detectionrate of early cereal-typepollen on coresfrom Rum, andsubsequently on coresfrom Ulva (in which cereal-typepollen alsooccurred during routinecounting) by employinga low powerscanning method.
Originally developedand discussed by Edwardsand McIntosh (1988), the methodinvolves the low power(x 100)microscope scanning of pollen preparations.As cereal-typepollen grainsare large (> 37Am)they aremore easilyidentified at low powerthan most other pollen grains.This methodallows a greaterproportion of eachsampled level to be searchedfor cereal-typepollen, wherethousands of grainsmay be presentand where the amountof time involved in pollen countingor scanningat a higherpower (e.g. x 400) would be prohibitive.
Estimationsof total pollen contentof scannedsamples were obtained by countingland pollen over transectson eachslide viewed at x 400.The area coveredin countingcould thenbe comparedto the whole areascanned and an estimationof total pollen scannedwas calculated.Total amountsof land pollen presentin I cm3samples are dictatedby pollen concentrations,and pollen densitieson slidesmay alsovary dueto dilution with siliconeoil. It was usually
53 possibleto scanbetween an estimated5000 and 10,000land pollen, but in some samples,extremely low concentrationsmeant that asfew as 1000land pollen were scanned.The estimatedamounts of pollen scannedin eachsample are providedin AppendixB.
I cm thick, contiguous sampleswere scannedfor cereal-typepollen on parts of monoliths KLI, KU and KL4 from Rum and additional samplesfrom below the earliest level where cereal-typepollen was previously detectedon K (Edwards and Hirons, 1990) were also processed.I cm.sampling resolution was also employed on parts of coresAC I, AC2, LC1 and LC2 from Ulva. Where additional cereal-typepollen was detectedthe samplewas also counted for land pollen to the standard500 land pollen sum (seebelow, Section 2.1.7). This enabledthe range of pollen types to be assessedfor other indications of
agricultural activity.
2.1.6.2 Criteria for identification of cereal-typepollen
The identificationof cereal-typepollen wasbased initially on the size classificationgiven in Moore et al., (1991). Poaceaegrains exceeding 37 gm in diameterwith an annulus>8 gra werethus considered for classificationas cereal-typepollen, althoughthe surfacepattern (scabrate) and the strengthof the annulusboundary were alsotaken into consideration(Andersen, 1979). Grains exceeding40 Itm in diameterwith an annulus> 10 [tm and a verrucatesurface patternwere considered for inclusionin the Avena-Triticunjtype category (Andersen,1979; Moore et al., 1991).
Unfortunatelythere were too few grainsrecorded to test statisticallyfor the reliablity of identification, and large Poaceacpollen types such as the Glyceria
54 groupmay fall into the cereal-typecategory using the abovecriteria (Andersen, 1979;O'Connell, 1987). Hence the accompanyingpollen spectra,radiocarbon datesand proximity of archaeologicalevidence to the site areall the more importantto interpretation.
2.1.7 Damaged pollen
The physical and chemical processeswhich pollen grains may undergo prior to and during their sub-fossilisation may alter their physical appearance.These processesdiffer according to the circumstancesand environment in which fossilisation takes place, and recognition of the resultant featuresmay assist in the interpretation of a fossil pollen assemblage.
Thereis little known aboutthe effectsof taphonomicand depositional processes on the'physicalappearance of pollen grainsalthough degraded pollen is often usedas an interpretationaltool (e.g. Cushing,1964,1966,1967; Birks, 1970; Tolonen,1980; Lowe, 1982;Tipping, 1984,1987;Walker andLowe, 1990)and assumptionsare frequentlymade about the processeswhich havecaused damageto pollen. Someexperimentation has taken place in attemptsto identify which processescause which typesof deterioration.Pollen grainsstored in leaf mold may developvarious features of corrosiondue to bacterialattack
(Havinga,1964; Rowley et al., 1990).Rowley et al. (1990)suggest that differencesin the sporopollencontent of grainscan lead to differential preservation,as somegrains are more resistant to attackthan others. Similar conclusionswere drawn by Sangsterand Dale (1961,1964)in experiments wheredifferent grains were buried in differentnatural environments and the ratesand types of decayof the grainsrecorded.
55 Experimentsby Holloway (1989)and Campbell and Campbell (1994) have shownthat wet-dry cyclescan have a considerableeffect on pollen preservation, andmay result in corrosionand thinning of the exine.It is often thoughtthat the buffetting of pollen grainsin coarsematerials such as sandmay alsoresult in corrosionor at leastbreakage (Fall, 1987),but Campbell(199 1) was unableto demonstratethis underlaboratory conditions and concluded that wet/dry cycles arethe most importantcontributors to pollen grain decay.
The different types of pollen exine deterioration have been identified (Havinga, 1964; Cushing, 1964,1966; Tolonen, 1980). Most classifications follow that originally defined by Cushing (1964,1966) and that used in this study, and defined below, is basedon Cushing's deterioratedpollen types.
fused 1. Degradedor thinned- exinesappear altered as structural elements are together.The whole of the exineis affected.
by 'pits' the 2. Corrodedor pitted - corrosionis characterized random over surfaceof the grain which vary in sizeand coverage. Areas of the grain not pitted retainthe normalcharacteristics of the pollen type.
3. Crumpled- grainsare badly foldedor wrinkled.
Occasionally 4. Broken- the exineis rupturedand the grain split. only a proportion of the whole pollen grain will be present.
Some deterioratedpollen grains may be so badly affected that ascribing them to a pollen type is not possible. Although these grains are assignedto the
56 'indeterminate'category, this is subdividedinto damagetypes as defined above. The type of damage,if any, for eachgrain identifiedwas alsorecorded accordingto the aboveclassifications.
Classificationof deterioratedpollen grainsis a subjectiveprocess (cf. Cushing, 1966;Lowe, 1982;Tipping, 1984).Some pollen types,including Juniperus communis,Poaceae and Cyperaceae crumple easily, and the breakageof saccate grainssuch as Pinus, where air sacsmay becomedetached from the body of the grain,is common.The chemicalprocedures used in preparationof pollen samplesmay alsoaffect the grainsto somedegree, and acetolysis may induce featuresakin to corrosion.These factors should be consideredwhen interpreting damagedpollen data,and some damaged pollen types,such as crumpled Cyperaceaegrains may be excluded.
Cushing(1966) has suggested a hierarchicalclassification system, whereby certainclasses of deteriorationtake precedence over others.Whilst this avoids duplicationof numbersof deterioratedpollen grains,this meansthat sometypes of deteriorationwill be under-represented(Lowe, 1982).
Despitethe problemsof subjectivityassociated with deterioratedpollen classification,the recordingof damagedpollen hasbeen included in this study asit may be useful in the identificationof re-workedsediments, changes in sedimentationand changes in pollen taphonomy.The damagedpollen classificationsare given abovein hiererarchicalorder, with thinned'as the most importantand 'crumpled' the leastimportant. Few degradedor thinnedgrains wererecorded in the pollen countsand corrosion or pitting appearedto be the main componentof the deterioratedgrain assemblages.Crumpled and broken
57 grainswere also poorly represented,particularly as Juniperus connnunis and Cyperaccaewere excludedfrom the crumpledcategory. Poaceae was recorded ascrumpled when this wasextreme and not a merefolding of the grain.Pinus grainswere included in the brokencategory; grain bodies or grain bodieswith oneor more sacswere treated as a singlegrain. Single air sacswere recorded separatelyand 2 sacswere recorded as representing a singlePinus grain.
2.1.7.1 Presentation of damaged pollen data
The damagedpollen data are presentedas percentagesof raw counts of individual taxa, and total damagedpollen percentagesof TLP (Lowe, 1982).
Although damagetypes to which certain grains are particularly susceptibleare excluded, the susceptibility of eachtype of grain to a particular type of damage may be consideredwhen Total Damagedpollen percentages(TD%) are interpreted. Increasesin damageto a single taxon might also indicate a different taphonomy for this pollen type, but large percentagesof damagedpollen can occur due to the presenceof very low amountsof a single taxon (e.g. a single pitted Alnus grain at one level will be representedas 100 % pitted). Damaged pollen is not representedas concentrations,as this would have the effect of increasing damagedpollen concentrationswhen taxon concentrationsincrease.
2.1.8 Dinoflagellate analysis
Dinoflagellatecyst analysiswas undertakenby Dr. Rex Harlandon sediments underlyingAC1, AC2, LCI andLC2 from Ulva. Minerogenicsediments sandwichedbetween organic material on coresLC I andLC2 were also assessed for dinoflagellatccontent. The analysiswas undertakenin an effort to determine whetherthe sedimentsoriginated from the erosionof terrestrialsurfaces in the
Lateglacial,or weredeposited in marineconditions in the earlyHolocene.
58 A total of 23 sampleswere processed using palynological procedures outlined by Wood et al., (1996)although oxidising agentswerc avoidedin orderto preventthe lossof delicatecysts (Harland, 1998). In orderto allow the calculationof the numberof cystsper grammeof sediment,the dry weight of eachsample was notedand aliquot samples of the organicresidues were taken after acid digestion,mounted in Elvacite,and counted using ax 10 objective (Harland1989,1998).
2.1.9 Sample treatment and laboratory methods
Volumetric samplesof 1CM3were processedfor analysis of pollen content.
Sampling intervals vary from 4 cm down to I cm contiguous samplesand the refinement of sampling resolution was undertakenafter initial pollen counts were achieved.Refinements targetted observed changes in the pollen spectra which were of ecological interest and where the possibility of human impact was consideredlikely.
All sampleswere suspended in HC1,followed by 10 % KOH andacetolysis treatment(Faegri and Iversen, 1989; Moore et al., 1991).Samples from lake sedimentsand peat from the baseof prorilescontaining particles of silica were boiled in 40 % HF for 20 minutes.Tablets of Lycopodiumclavatum spores were addedto eachpreparation to allow the calculationof pollen concentrations (Stockmarr,1971).
Slides were preparedfrom processedsamples and viewed under a light microscopeat a magnificationof x400. Individual grainsrequiring greater
59 attentionfor identificationwere viewed at x1000with the slide underoil immersion.
Pollen from Loch aBhogaidhand Farm Fields, Rum, was countedto a minimum of 500 total land pollen (TLP) (i. e. excluding aquatics and spores) whereconcentrations allowed. Low concentrationsin somesections of the Loch a!Bhogaidh cores meant that countsof somesamples reached only c. 300 land pollen (althoughspores were numerous). Pollen counts from Ulva werebased on a land pollen countof at least300, but Alnuswas excludeddue to its extreme abundance.However, most of the levelscounted exceed 500 land pollen. Identificationsof fossil pollen andspores were based on the type slide collection of the Departmentof Archaeologyand Prehistory, University of Sheffield,aided by the keys, inter alia, of Birks (1973),Faegri and Iversen (1989) and Moore et al., (1991).Pollen andspore type nomenclaturefollows Bennett(1994) and Bennettet al., 1994).
2.1.10 Loss on ignition (LOI)
All levelssampled for pollen andsome from non-polleniferousmaterial in the Ulva coreswere measured for percentageorganic content by losson ignition (LOI). Samplesof I cM3were dried for at least12 h at I OSOCin cruciblesof known weight.These were then weighed and ignited at 5000Cfor 3 h. Ash weightswere recorded and organic content calculated and expressedas a percentageof dry weight (Bengtselland Enell, 1986).
2.1.11 Construction of pollen diagrams
All pollen diagramswere constructed using TILIA 2 spreadsheetsand
TILIA. GRAPH 2 (Grimm, 1991) and zoned with the aid of CONISS.
60 Percentagesare based on TLP. Detrendedcorrespondance analysis (Dccorana [Hill, 19791)of taxonvariables from individual cores,and of zonemean scores, was alsoundertaken using the ordinationroutine of TILIA.
2.2 Choice of sites
Most of the sites chosenfor coring as part of this study are close to known areas of Mesolithic and Neolithic occupation (Armit, 1996). These are Loch a!Bhogaidh on Islay, Farm Fields at Kinloch, Rum and A! Chrannagand
Livingstone's Cave bogs on Ulva (Figure 2.1).
Pollenanalyses of two coresfrom Loch aBhogaidhhad alreadyproduced profiles which suggesthuman impact on vegetationin the Mesolithic (Edwards andBerridge, 1994; Edwards and Mithen, 1995),and additionalcores were obtainedfor closerresolution sampling in orderto provide greaterspatial and temporaldetail on the original interpretations,and in an attemptto detect smallerscale changes which could not be perceivedin the original cores.
The proximity of A! Chrannagand Livingstone's Cave bogs on Ulva to the midden deposits at Livingstone's Cave (Bonsall ef al. 1991,1995) suggested that the detection of disturbancein the pollen record relating to the occupation of the cave was likely. Both Mesolithic and Neolithic people may have altered vegetation near the site as the archaeologicalremains extend into the latter Ulva period according to finds and radiocarbon dates(ibid). No pollen data from are currently published, and palynological investigations from thesesites was anticipated to be useful in completing the Holocene pollen record for the Inner
Hebrides. The small areaof the island (approx. 28 km2) could produce
61 vegetationtypes and successions different from thoseof the largerislands during the Holocene.
The islandof Mull wasincluded in this studyas it was anticipatedthat pollen profiles may providesuggestions of Mesolithicactivity in the absenceof archaeologicalevidence (the middenat Dervaig[Robins, 1998] was not known andwas unpublishedwhen fieldwork wasundertaken). Sites originally corcdby
Lowe andWalker (1986aand b) andWalker and Lowe (1985,1987)were consideredfor additionalcoring. A peat-filleddepression at Gribun,western Mull andLoch an t'Suidheon the Rossof Mull wereassessed as offering the greatestpotential for providing a completeearly Holocene pollen record.These sitesalso followed a north-southtransect from Ulva, both being on the western sideof the island.Unfortunately, pollen from Gribunproved too sparseto provide a usefulrecord, and the core from Loch an t'Suidhenear Bunessan (Lowe andWalker, 1986aand b) is the only one from Mull to be includedin the project.Loch an tSuidhe is small,thus multiple coring was considered inappropriate.However, close resolution sampling was intendedto provide greaterdetail for the earlyHolocene part of the core.
The pollen from a singlecore from FarmFields, Kinloch, Rum was also suggestiveof Mesolithic impacts,and occurrences of cereal-typepollen in combinationwith radiocarbondates show that agriculturalpractices may have occurredhere early in the Neolithic from c. 3950BP (2540cal. BC) (Hirons and Edwards,1990). Further profiles obtainedfrom the Farm Fields siteswere intendedto providegreater spatial and temporal detail for theseevents, but the recoveredpeat monoliths proved unable to providea pollen recordcovering the
62 Mesolithic.Thus, the focusof investigationon Rum wasrestricted to the detectionof the earliestccrcal cultivation in the pollen records.
2.3 Coring
Coring was carried out at Loch a!Bhogaidh, Islay, A! Chrannagand Livingstone's
Cave bogs, Ulva, and at Loch an t'Suidhe, Mull using a wide bore (8 cm. diameter) Russian-typecorer (Jowsey, 1966) where possible. A small bore (5 cm diameter) corer of similar type was used for very deep sedimentsat Loch a!Bhogaidh and on Ulva where extraction with the larger device was otherwise impossible. Due to the shallow depth of the peat on Rum, pits were dug and all samplesobtained in monolith tins from a cleanedpeat face.
A generalsurvey of surfacevegetation was undertaken for eachsite and samples obtainedat the surfaceof eachcore and other strategic places. Site descriptions appearin chapters3-6.
63 CHAPTER 3- LOCH AlBIIOGAIDII, ISLAY. Discussionand
Interpretation of multiple profiles from lake sediments
3.1 Site location and description
Loch a!Bhogaidh (Figs. 3.1a and b; Plates 3.1 and 3.2), NGR NR 224 577, is an
area of fen peat surroundedby blanket peat of dimensions 500 rn north-south
and 350 m east-west.It lies approximately 500 m to the cast of Beinn Tart
a!Mhill on the Rhinns peninsula.The archaeologicalsites of Bolsay and Gleann
Mor are located 500 m south and 1050 m northeastof the mire respectively. The bog is surroundedto the north and eastby low gentle slopes dominated by
Pleridium aquilinum. Elsewherethe mire becomesdrier and extendsinto
Calluna vulgaris heath. Betula-Corylus woodland is located to the west of the
site in a shelteredvalley at the foot of Beinn Tart a!Mhill.
The surfaceof Loch a!Bhogaidh contains areas of openwater surroundedby
Phragmitesaustralis. Thickets of Salix spp.and Rubus spp. agg. are common
althoughthe majority of the bog surfacesupports hummocks of Molinia sp. and
Eriophorumangustifolium interspersed with Sphagnumspp. pools. Other commonbog plantsinclude Myrica gale, Succisapratensis, Paniassiapalustris andHypericum elodes.
The currentpeat surface overlies an expanseof gyttja or organiclake sediment
andan east-westtransect by Agnewet al. (1988)shows the depthsof gyttja and
overlyingpeat and basal deposits of pink clay which areprobably of Lateglacial origin (Fig. 3.2).
64 3.2 Background to the current project
CoresLABI (NR 22435782), LABII (NR 22605745 ) (Edwardsand Berridge, 1994)and LABX (NR 22545805) (Agnew et al., 1988)have already been used in the reconstructionof vegetationat Loch a!Bhogaidh. These have been includedin a generalvegetation history for Islay in Section1.3.1. As LABX post-datesthe Alnus rise (Agnewet al., 1988)it is not comparablewith the other early Holoceneprofiles. LABII appearsto exhibit a substantialhiatus in
sedimentationand is not radiocarbon-dated(Edwards and Berridge, 1994), but is neverthelessuseful in the interpretationof multiple profile pollen studiesat the site.
Six furthercores were obtained for the purposesof this study,LABIII (NR 2236 5792), LABIV (NR 2245577 1), LABV (NR 22325777), LABVI (NR 2257
5777),LABVII (NR 22405776) and LABVIII (NR 2235 5782).Cores LABIII,
LABVII, LABI, LABIV andLABII arelocated at intervalsalong the north-
southaxis of the mire (Fig. 3.1), andalthough these are not on a linear transect
line the accumulatedlithologies provide an indicationof the north-south
stratigraphyof the site (Fig. 3.3). Obtainingcores from a north-southtransect was hamperedby surfacevegetation which includeddense thickets of Salix and Rubusspp. agg., deep pools of openwater dominatedby Phragmitesaustralis to
the southwest,and sub-surfaceaccumulations of gravel.
3.3 The possible extent of the earlier loch
The easternedge of the mire is boundedby a sharp break of slope (indicated by the 70 m contour line in Figure 3.1b). The west of the mire rises more gently, and the drier ground is coloniscd by Calluna vulgaris and other ericaceoustaxa.
The extent of the loch in the early Holocene was therefore limited by slopes to
65 the east,and the southeasterlyprofile LADVI is probablylocated close to the loch'seastern margin.
The east-westprofile (Fig 3.2; Agnew et aL, 1988) indicates that in the shallower sediments,thin basal depositsof gyttja are overlain by clay, and are lithologically similar to the LABII profile. As LADII may indicate the southerly limit of the loch in the early Holocene (Edwards and Berridge, 1994) then similar profiles to the west of the site may delineate its former extent in that direction.
Although the stratigraphictransect of the north-southprofile is incomplete,the basinappears to havehad a smootherand more uniform topography.If the depthof gyttja is a reflectionof waterdepth, it appearsthat a deeperbasin of water occurredto the north andan elongatedarea of shallowwater extendedto the southas indicated by the lithologiesof LABIII, LABVII, LABIV andLABII (Fig. 3.3).
3.4 The pollen diagrams
Percentagepollen diagrams for LABI - LABVIII comprise Figures 3.4 - 3.11.
Summary diagrams of LOI, charcoal concentrations,influx and Ch:P, TLP concentrationand influx and total damagedpollen percentagesfor LABI -
LABVIII are shown as figures 3.12 - 3.19. Concentration diagrams for selected taxa from the profiles are presentedin Figures 3.20 - 3.27 and pollen influx diagrams in 3.28 - 3.35. Figure 3.36 shows the Corylus percentagecurves for profiles LABI and LABIII - VIII plotted against age, and Figure 3.37 the influx curves for the sameprofiles (LABII was excluded in view of the absenceof
66 dateson the profile anddifficulties of comparisonwith othercores). Figure 3.38,3.39 and3.40 respectively display the charcoalconcentration, influx and
Ch:P datafor all profiles exceptLABII, andUP concentrationsand influx rates
for profiles LABI andLABIII - LABVIII arepresented as Figures 3.41 and 3.42.It shouldbe notedthat whereAMS datesarc usedin extrapolation,some dateson LABVIII are acquiredfrom comparisonwith otherprofiles. Damaged
pollen percentagesfor selectedtaxa from eachprofile LABIII - LABVIII are shownin figures3.43 - 3.48.
Scanning low from the levels displayed in and pollen counts samplespreceding , the diagrams (from earlier gyttja and clays) showedthat Betuld and
Myriophyllum alterniflorum dominated the pollen assemblagesand the earliest
Corylus pollen detectedis displayed on all the profiles.
Samplingintervals are refined to I cm,contiguous samples for the Corylusrise and expansion,and where reductions in the Coryluscurve became evident.
Wider samplingintervals occur elsewhere. Pollen counting was hamperedin someof the profiles by very low pollen concentrationsor the absenceof quantitiesof organicmaterial after processing. Such levels were sampledmore thanonce to ensurethat chemicalprocessing was not responsible.The absence of pollen in somesampled levels, and very low pollen countsin others,means that in someplaces in the diagramsthere are uneven sampling intervals and wider thanintended sampling resolution.
Comparisonsbetween diagrams have been drawn on the basisof the percentage pollen profiles aidedby referenceto the concentrationdata (Table 3.1). AMS dateswere also used as a guide for comparisons,but where pollen percentage
67 datashowed strong similarities but datesbetween profiles conflicted,reliance wasplaced on the percentagedata (refer to Section3.4). A DCA plot of post- hazelrise meanscores (Fig. 3.49a)and one of zonesc andd (Fig. 3.49b)were alsoused to assistin the comparisonof zonesand subzoncs between profiles.
3.5 AMS dates on the Loch a'Bhogaidh profiles 3.5.1 LABI
Six radiocarbondates were obtained for coreLABI (Table3.2) by Edwardsand Berridge(1994) but only threeof theserelate to the first half of the Holocene. Thesedates were obtained from conventionalradiocarbon analyses of 10cm thick portionsof core,which makesit difficult to ascribea precisedate to the centreof eachsample. Therefore these dates may thereforenot be asreliable as the AMS datesdetermined for smallervolumes of sedimentrestricted to I cm depth.Nevertheless, other than the basaldetermination, they seemto provide an adequatetimescale for vegetationchanges as reflected by the pollen from the
Loch a! Bhogaidhcatchment. The basaldate on LABI may be erroneousdue to laboratoryerror or an inadequatecarbon content for the sample.
3.5.2 LAB111 - LABV111 Twenty three AMS dateswere obtained for cores LABIII to LABVIII (Table
3.3; Fig. 3.50). Thesewere intended to allow comparisonbetween cores from
Loch a!Bhogaidh and elsewherein the Inner Hebrides, but specific featureson the percentagepollen diagramswhich may representclimatic and natural ecological changesor human impact were also targeted.
68 3.5.3 The early hazel peak, LABIb(i), 10910+450BP (109740cal. BQ
Comparison of the earliest date on LABI to other pollen profiles from the Inner
Hebrides and western Scotland suggeststhat on the basis of initial increasesin
Corylus avellana-type pollen the date of 10,910+450BP (10,470 cal. BC)
(Edwards and Berridge, 1994; Sugdenand Edwards, in press) is too old even if the standarddeviation is not ignored. Edwards and Berridge (1994) note that the range of pollen of thermophilous taxa which occurs in LABlb(i) includes Corylus avellana-type, Juniperus communis,Eniparum and Betula.
Thesereflect temperateconditions yet datesof c. 10,910 BP (10,740 cal. BC) for other sites in the Inner Hebrides suggestLateglacial Loch Lomond Intcrstadial conditions (e.g. Birks, 1973; Birks and Williams, 1977; Lowe and Walker,
1986; Walker and Lowe, 1985,1987). The preceedingclay lithology incorporating'the pollen of Poaceae,Cyperaceae and a variety of herbs on
LABIa(i) and a(ii) are more representiveoftateglacial tundra conditions
(Edwards and Berridge, 1994) and an older date. Indeed, the mixing of earlier sedimentswith later gyttja may explain the anomalousdate at the base of
LABIb(i) if low carbon levels or laboratory error are not the cause.
Comparisonof LABI with the otherpollen profiles from Loch a!Bhogaidh also confinnsthe aboveanomaly. Expansions of Corylusavellana-type pollen appearmuch later than 10910± 450 BP (10,740 cal. BC) in the additional profiles (i. e. c. 9500BP; 9615+ 65 (8990cal. BC) on LABVIIa; 9415± 65
8800cal. BC) on LABIIIa; 9565+ 75 (8930cal. BC) LABIVa). Pollen data from Arran (hazelrise c. 9500BP (8900cal. BQ, Robinsonand Dickson, 1988; Edwards,1990), the mainland(hazel rise 9230+ 120BP (8500cal. BC); Peglar
1977in Birks, 1980;Boyd andDickson, 1988)and other Inner Hebridean
69 islands(eg. Skye,Birks andWilliams, 1983;Walker and Lowe, 1991a, and Mull, Lowe andWalker, 1985)also suggest that the Corylusexpansion at LABI shouldbe later.
3.5.4 LABIIIc
There are three datesin LABIIIc (8215 ± 80 BP (7290 cal. BQ; 7720 ± 70 BP
(6600 cal. BQ and 7710 + 100 BP (6570 cal. BQ) which do not correlate with dateson the other profiles (including interpolated dates) for what are considered to be related zones.LABIIIc displays a distinctive reduction in Corylus with accompanyingchanges in other taxa and increasesin TD%. This reduction and associatedcharacteristics also appearin LABIVc and LABVIc, LABVIIc and
LADVIIIc, and in LABIc betweenthe depthsof 650 cm.and 620 cm. The DCA plots of zone mean scores(Fig. 3.40a and b) indicate that LABIIIC compares well with LABIVc, LABVIc, and LABVIIIc on the DCAI axis on both charts.
Comparisonsof all profiles suggestthat this Corylus reduction begins c. 7700
BP (6600 cal. BQ, making the earliest date at the baseof LABIIIc too old. This date overlaps at two standarddeviations with the later date for this zone (Fig.
3.50b), and may representa period of rapid sedimentationdue to erosion.
LABIR is located close to a stream at presentand may also have been close to a stream inlet at the loch edge in the early Holocene. LABIIIc may therefore incorporate re-worked material, indicated by a changein lithology, which has made the datesolder. There is no other evidenceof erosion as LOI does not decreasebut older organic material may have been introduced by a slow- flowing stream, such as that which currently crossesthe mire. Anomalies in the datesin LABIII may also have arisen from the accidental incorporation of older carbon to the sample in the field or laboratory.
70 3.5.5 Extrapolation of age-depth curves and pollen influx calculation
Linear extrapolation betweenthe AMS dateson the profiles was used to provide influx age-depthcurves (Figs. 3.51 - 3.57) and to calculatepollen rates and sediment deposition times. Where only one or two dateswere available on a profile (e.g. LABV and LABVIII) additional dateswere estimatedby comparing dated featurescommon to all the profiles. This was to allow the visual comparison of pollen and microscopic charcoal data for the complete suite of profiles, and it is appreciatedthat further radiocarbon datesare required on these profiles to firm the conclusionspresented later.
Given the likelihood that certaindates on LABI andLABIII aretoo old, two setsof age-depthcurves were calculated for theseprofiles, the first using the actualdates acquired, and the secondusing dates inferred by comparisonwith the otherprofiles. The DCA plot of zonemean scores Fig. 3.40aindicates that LABlb(ii) compareswell with zonea for the otherprofiles, where the Corylus is rise occursdating to c. 9700BP (9040cal. BQ. The earlierLABIb(i) seperatedfrom theseand this may reflect the differentpollen assemblageand earlierdate. Thus, a secondage-depth curve for LABI is suggestedusing 9700 BP (9040cal. BQ at the secondCorylus rise on the profile andomitting the earlierdate of 10,910+450BP (10,740cal. BQ.
As the DCA plots (Fig. 3.49aand b) indicatethat dateson LABIlIc shouldbe later,a secondage-depth curve has also been estimated for this profile for ease of comparisonwith otherprofiles. The secondprofile with assumeddates used 7700BP (6600cal. BC) asthe basaldate for the Corylusreduction and 7 100BP
(5930cal. BQ asthe datefor the Corylusrecovery, although given variation in
71 dateson the otherprofiles the true datescould fall within this range,if the original datesare indeed incorrect.
3.6 Isoetes lacustris spores as an indicator of erosion
Microsporcs of the submergedaquatic Isoetes lacustris are prominent in most of
the pollen diagrams from the site. Vuorela (1980) has suggestedthat increasesin
the microspores of Isoetes lacustris in pollen diagrams from lake sedimentsmay be representativeof short erosive episodeslinked to agriculture. Gacia,and
Ballesteros (1994) showed that although light levels and temperaturesaffect the
growth and productivity of Isoeteslacustris spores,the primary limiting factor is nutrient availability. Vuorela (1980) also suggestedthat increasesin mineral
input to a lake will increasenutrient availability and favour the expansion of
communities of Isoetes lacustris, although eventual eutrophication due to
lengthy periods of erosion will significantly depletepopulations. A short-lived pre-agricultural disturbancein a lake catchmcnt should therefore produce temporary increasesin Isoetes lacustris in the pollen record. Where there are also reductions in LOI, the indication is that somedisturbance to vegetation within a pollen catchmcnt has occurred (Vuorcla, 1980).
Szmeja (1994)showed that Isoctes lacustris communities favour lakes or pools with slow sedimentationrates and an input of fine detritus,and theseconditions areunlikely to occurfor any lengthof time whenminerogenic input is increased.Although the abundance of damagedIsoetes lacustris microsporcs wasnot quantified,owing to theprohibitatively great abundance of countsfor thistaxon, it wasobserved that many, perhaps 70 % or more,of theexines were split or broken.Whilst breakage may be an intrinsiccharacteristic of this typeof
72 sporedue to its flimsy perine,the rc-workingof sporesin usuallake sediment focusingor during erosionalpcriods could increasebreakage.
Whilst water levelsat Loch a!Bhogaidh are unknown, and will havechanged throughtime, Isoeteslacustris may havebeen absent from the centralbasin due to lower light levelsand temperatures if the loch waterswere particularly deep (Gaciaand Ballesteros, 1994). The Isoeteslacustris microspores in LABI,
LABVII andLABVIII could thereforebe reworked.In LABI andLABVII, Isoetesmicrospores are present at everylevel from c. 9000BP (8225BQ and reachvalues of between150 - 200 % TLP. In LABVIII Isocteslacustris microsporesexpand later c. 8000BP (6860BQ (LABVIllb) anddisplay a uniforin anduninterupted curve reaching values of c. 750 % TLP. Profiles from the outlying coresLABII, LABIII, LABIV, LABV andLABVI show comparitivelylow levelsof Isoeteslacustris microspores c. 10 % TLP, but there aresporadic increases which at timesreach > 1500% TLP in LABIV and LABV.
All the major increasesin Isoeteslacustris in the outlying corescoincide with reductionsin LOL and frequentlywith reductionsin other taxa,and increases in TD%. AlthoughIsoetes lacustris may havebeen more abundantin the shallow watersof the loch edge,the increasesin its sporesmay signify disturbancesto vegetationin the loch catchment,particularly when these occur in conjunction with othersignificant data. Increases in Isoeteslacustris may arisedue to natural disturbancesor humanimpact causing erosion into the loch.
73 3.7 Damagedpollen analysis
Damagedpollen was assessedfor profiles LABIII - LABVIII as detailedin Section2. I. S. It was notedin pollen countingthat most of the damagedpollen
was pitted.Differences in percentagesof damagedpollen for individual taxa within profiles arenot extreme,and between profiles TD% is consistently
between20 - 30 % exceptin LABIII wherefor mostof the proflle damaged pollen percentagesare lower, at around10 % TLP. This suggeststhat a constant
proportionof pollen is damagedby taphonomicprocesses and pollen in LABIII may be betterpreserved due to a lower influx of re-workcdpollen from sedimentfocusing.
Increasesin TD% occurin LABIIIc which could reflect an increasein re-
workedpollen, eitherfrom the loch sedimentsor from the surrounding Slight increasesin TD% in LABIVc, LABVIc, LABVd catchments.. occur and andmay be linked to erosiondue to woodlandreduction in thesezones.
3.8 General representationsof the pollen profiles
Pollenprofile LABI (Edwardsand Berridge, 1994) is the mostcentral of all the coresfrom the site, andis thus likely to providea broaderand more regional representationof vegetationthan the otherprofiles. The outlying coreswere
intendedto providedetails of very local disturbancesand changes in vegetation.
A comparisonof zonesbetween profiles is providedin Table 3.1. Despite evidentdifferences between profiles, the generalvegetation history of Loch
a!Bhogaidh remains similar in all the cores.Betula, Poaceae and Cyperaceae comprisethe main earlyHolocene assemblage, then Corylusavellana-type rises
rapidly to becomethe dominanttaxon for the remainderof the profiles. Other arborealtaxa present from c. 9000BP (8225cal. BC) include Ulniusand
74 Quercus,taxa frequently associated with Corylusand components of contemporaryCorylus woodlands in theInner Hebrides (Birks, 1983; Lowc and Walker,1986a and b). A tall-hcrbundcrstorcy is suggestedby thepresence of Filipmdula,Apiaceae and Rumax spp., and the spores of Polypodiumvulgare- typeand Ptcropsida (monolete) indet. (ferns) reflect the damp shady environrnentof thewoodland floor. Aquatic taxa are few with theexception of Isocteslacustris.
3.8.1 Pollen concentration and influx
In chapter 2 it was explained that spatial differences in pollen recruitment and the effects of sediment focusing within a lake basin could result in differences betweenpollen profiles, and that variability in sediment accumulation within different parts of a lake basin should be apparentin the pollen concentrations and pollen influx curves of multiple profiles.
Calculationsof total pollen concentrationsand influx rates(Figs. 3.41 and 3.42) incorporateonly land pollen, asthe largeamounts of the sporesof Isoetes lacustris,which areeither consistent or sporadicallyapparent in all the profiles, havea markedinfluence on the overall concentrationand accumulationdata curve.
It is notedthat the comparisonof absolutevalues for LABI andLABII with thosefrom the otherprofiles may not be satisfactory,as lower TLP countsand the differing numberof exoticspores added to the samplepreparations will compoundstatistical errors inherent in the calculationof suchvalues (Section 3.7.1.2).
75 Low pollen concentrationsin the earlyHolocene, (LABIb, LABIlb, LABIlla,
LABIVa, LABVa, LABVIa, LABVIla andLABVIlla [Figs. 3.20 - 3.27; Fig. 3.41]) areprobably linked to fast sedimentaccumlation rates. The rise of
Coryluscoincides with an expansionof TLP concentrationswhich averagec. 100,000grains cm-3 in LABIllb(i) and(ii), LABIVb(i), LABVb(i) and LABVIb, LABVIIb(i) and(ii) andin LABVIRb. The equivalentportion of LABIc(i) displayslower concentrationsof c. 50,000 grainscm-3. As the most
centralof the profiles, LABI may havereceived lower amountsof pollen (cf Davis andBrubaker, 1973), particularly in the earlyHolocene, when trees were few andthe flora was dominatedby Poaceaeand Cyperaccae whose pollen hasa limited rangeof distributioncompared to treepollen. However, similar low
concentrationsof Corylusin LABIld(i) suggestthat anomalousCorylus
concentrationsin LABI andLABII arerecorded compared with the
subsequentlysampled profiles, possibly due to statistical inaccuraciesderived
from the estimationof pollen concentrations.Nevertheless, the shapesof the
TLP concentrationcurves between all the profiles (Fig. 3.41)are similar
althoughaverage concentrations range from 50,000to > 300,000grains cm-3 for
all the profiles.
TLP concentrationsare greaterin LABVII and LABVI (Fig. 3.41).LABVII
may be expectedto recruit greateramounts of pollen than elsewhereas it was placedin profundalsediments which derivepollen from the loch perimeterand
receivelarge amounts of airbornepollen. High amountsof TLP concentrations
in LABVI of betweenc. 200,000- >300,000grains CM-3 contrast with average 3 concentrationsof c. 100,000grains CM, in the otherperipheral profiles. These
may be explainedif wind andwave direction transported pollen to the
76 southwcsternarea of the loch, but low sedimentaccumulation rates at LABVI ( "; 0.0125 cm yr Fig. 3.17)may alsoaccount for high UP conccntrations.
Attentionhas been paid to Pinuspollen concentrationsas this pollen is
particularlybuoyant and more susceptiblethan other grains to the actionof wind andwaves. Thus, variable concentrations of Pinuspollen betweenprofiles may assistin explainingdifferent concentrations of otherpollen typesand microscopiccharcoal which may alsobe particulailyprone to focusingby wind andwave action.Pinus concentrationsin the deepercentral profiles LABI, LABVII andLABVIII averagec. 5000grains cm73 (Figs. 3.20,3.26,3.27).In LABII, LABIII andLABV Pinusconcentrations average c. 2500 grainsCm73
(Figs. 3.21,3.22,3.24)and LABIV andLABVI Pinus concentrationscompare with the centralprofiles at c. 5000grains Cnf3 (Fig. 3.23 andFig. 3.25).
Pennington(1949) suggests that the air sacson Pinus grainsmay fill with water andrapidly sink, which could explaingreater amounts of Pinus pollen in profundallake sediments.The greaterconcentrations in the peripheralprofiles to the south-westof the loch basinmay be the resultof wind direction,but more probablythe slowersediment accumulation rates at LADIV andLABVI (c. 0.0125cm, yfl; Figs. 3.15 and3.17) compared with the otherperipheral profiles (c. 0.33cm. yfl; Figs. 3.13,3.14,3.16)are responsible. There is thereforeno evidencefrom the Loch a'Bhogaidhprofiles to suggestthat wind andwave directionsignificantly affected pollen distributionover otherprocess such as sedimentfocusing.
Whenplotted against depth, the pollen influx datafrom Loch a'Bhogaidh(Figs.
3.21 - 3.27) closelymirror the concentrationdata. Variations in sediment deposition rates coincide with expectedchanges in pollen influx within each
77 profile, as calculation of both is affected by the specift location of dateson the
profiles. For all the profiles, Corylus experiencesthe highest influx rates,
particularly in the earlier Holocene. When plotted against age (14C yr. BP) it is
noticeable that the perimeter proflles LADIII, LABIV, LABV and LA13VI have
lower TLP influx (averagec. 2500 grains cm-2 yr-1) than LABVII, LABVIII
and LABI (averagec. 5000 -> 20,000 grains cm-2 yr-1) (Fig. 3.42). Influx values for individual taxa, especiallyBetuld, Ulmus and Quercus, are also in greater the deeperprofiles LABI, LADVII and LABVIII (Figs. 3.28 - 3.35). As sediment deposition is faster in the deepercores at c. 0.05 cm yr-1 (compared
to c. 0.016 cm.yr-1 in peripheral cores) (Figs. 3.12 - 3.19), then it could be expectedthat pollen influx should be lower, there being less time for pollen to become incorporated into the sediment.However, if the sediment of the
profundal profiles was largely derived from material focused from the loch
perimeter then some of the material (including pollen) from the outlying regions
of the loch may have been removed and focused into the areaswhere greater
depths of gyttja now exist. Very high TLP influx for LABVII (> 20,000 grains 2 CM, yi) suggeststhat this part of the loch received the bulk of sediment
transportedin from the perimeter.
Most of the pollen influx profiles from Loch a!Bhogaidh show no marked changesthrough time. However,between c. 7900BP (6810 cal. BQ and 7700
BP (6600cal. BC) pollen influx ratesat LADVII increasemarkedly, influenced mainly by the Corylusinflux curve.After c. 7700BP (6600cal. BQ, pollen
influx at LABVII remainshigher, (> 20,000grains cm-2 yr-I)than in the other profiles. Corylusaccumulation rates thus increaseover this time from C.10,000 to 25,000grains cm-2 yr-I andsedimentation from 13 to 5 yr cm'I- This
78 suggeststhat from this time the main site of sedimentfocusing in the loch was in the region of LABVII.
3.8.2 Microscopic charcoal concentrations and Influx
Microscopic charcoal curves (including concentrations,Ch: P and charcoal influx) are presentedagainst depth for eachprofile in Figures 3.12 - 3.19 and in composite diagrams in Figures 3.38 - 3.40. For individual profiles, the three different representationsof microscopic charcoal show similar patterns,with reductions and increasesoccurring simultaneouslywithin the curves. The mode of charcoal representationdoes not appeartherefore to be significant (cf
Edwards and Whittington, in press) and where referenceis made to Ch:P in the following interpretation, changesin absolutecharcoal values may also be assumed.
The microscopic charcoal influx curves show a similar pattern to that of the TLP influx curves (Fig. 3.42), in that the central profiles LABVII, LABVIII and the new charcoal samplesfor LABI incorporate greateramounts of microscopic charcoal than the peripheral profiles. (The original LABI charcoal concentrationsand influx are not companýbledue to reasonsmentioned earlier in
3.6.1 and different methods of microscopic charcoal quantification. Edwards and Berridge (1994) used point count estimation). The greater amounts of microscopic charcoal in centrally positioned profiles could be explained by their greaterpotential for the receipt of airborne particulates and mechanism of sediment focusing mentioned above.It is also possible that the peripheral profiles could have received greater amountsof stream-bornepollen and microscopic charcoal, this method of transport and deposition effectively diluting pollen and charcoal influx.
79 The influx diagram(Fig. 3.40) that high composite . charcoal shows relatively charcoalinflux ratesoccurred in the early Holoceneat the baseof the profiles, betweenc. 9700BP (9300cal. BQ until c. 9000BP (8180cal. BC) although influx is rclatively low in LABV andLABIV despitecorrespondingly high concentrationsand Ch: P at this time in thesetwo profiles (Figs. 3.38 and 3.39). Later in all the profiles,between c. 7700BP (6660cal. BC) and 7100BP (5930 cal. BQ substantialincreases in charcoalinflux areevident in LADVII and LABVIII. LABV, LABIV and LABVI alsoshow slight increasesin charcoal influx, althoughoverall charcoalinflux ratesarc much lower in theseprofiles (Fig. 3.40).
High charcoalinflux in the earlyLoch a!Bhogaidh profiles may derivefrom the receiptof high levelsof airbornecharcoal from a wide areain a relatively open environment.A greaterinput of erodedmaterial may alsohave assisted in increasingcharcoal influx. Later in the Holocenethe expandingareas of Corylus dominatedwoodland may haveintercepted airborne charcoal, and stabilised soils therebyreducing erosion. Charcoal influx may be persistentlyhigher in the profundalprofiles asthese could havehad greater exposure to the fallout of airbornecharcoal. Streams feeding into the loch edgemay havediluted charcoal influx in the peripheralprofiles, and sedimentfocusing to the loch centremay alsohave deprived the edgeprofiles of charcoal.Temporary increases in charcoalinflux in the profiles,particularly in the peripheralprofiles, often correlatewith reductionsin woodlandat the loch edgewhich may have increasedthe amountsof airborneor erodedcharcoal reaching the site. Charcoal increasesmay alsoderive from fires at the loch edge,either natural or anthropogenicin origin.
80 3.9 Interpretation of the profiles
3.9.1 LABla(i) and LABIa(ii) Cyperaceae - Cichorium; LABIM
Cyperaceae - Asteraceae - Caryophyllaceae (Pre c. 10,000 BP [9400 cal. BCI)
These zones are earlier than those recorded in the other profiles, and the pollen record extendsinto the clay underlying the gyttja. Poaceae,Cyperaceae, Empetrum and tall-herbs dominate the pollen assemblagesreflecting an open environment. Small amountsof Betuld pollen suggestthat birch scrub or copses of birch were locally present,although the pollen may derive from Beluld nana.
Pinus pollen is also presentand the later introduction of Corylus in the following zone suggeststhat LABIa(i) and a(ii) representa tundra-like environment immediately preceding climatic amelioration of the early
Holocene (Edwards and Berridge, 1994).
3.9.2 LABIb(i) and LABIb(ii) Betula - Poaceae- Cyperaceae;LABIIc Betuld Betuld - Poaceae,LABIIIa Betula - Poaceae- Cyperaceae;LABIVa - Cyperaceae; Poaceae- Cyperaceae- Filipendula; LABVa Betula - Poaceae- Betula LABVIa Betula - Poaceae- Cyperaceae- Filipendula; LABVIIa, - (c. 10,000 Poaceae- Cyperaceae;LABVIIIa Betuld - Poaccae- Cyperaceae;
- 9300BP [9400- 8540cal. BCJ) LABlb(i) andLABIb(ii) displaysa pollen assemblagesimilar to the earlier (sub)zonesof LABla, LABIb(i) andLABIlb, andinclude Betuld, Enipetrum,
Poaceaeand Cyperaceae.Herbs include Lactuceae, Filipenduld, Plantagospp. andRumax acetosa representing a tall-herbflora characteristicof other early
81 Holoceneenvironments in the Inner Hebrides(e. g. Mull - Walker and Lowe, 198501987; Lowc andWalker, 1986a;Skye - Birks, 1973).A similar flora to thosein LABIb andLABIlb is recordedin the otherprofiles, although percentagevalues may differ, andthe minor taxavary in frequency.
The initial expansionof Corylus is recorded in all the profiles. In LABVIIa
Corylus increasesabove 2% TLP at 9615 + 65 BP (8990 cal.BC). In LABIVa
Corylus is presentin the early part of the zone at values <2% but this is possibly the result of the reworking of penccontemporaneoussediments or pollen.
Higher percentages,concentrations and influx occur in the latter part of
LABIVa, and the first Corylus record >2% datesto 9565 + 75 BP (8930 cal.
BC). Extrapolation of datesfor LABVIIIa, LABIIIa, LABVa and LABVIa places the initial expansionof Corylus at c. 9500 BP (9400 cal. BC). The hazel rise at LABIb(i) has an estimateddate of c. 10,000BP (9400 cal. BQ and the secondrise in LABIb(ii) is cstimatcd at c. 9500 BP (9400 cal. BQ as mentioned in Section 3.4.3-3.
LABVIla displayshigher Coryluspercentages than the otherprofiles and is the profile mostsimilar to LABIb(i) andb(ii). The rise of Corylusavellana-type pollen in LABlb(i) andb(ii) showsdistinct fluctuationswhich aredefined by thesesubzones. LABlb(i) doesnot appearto be distinguishedin concentration termsfor individual taxa,and does not featurein the TLP concentrationcurve. It is thus interpretedas an artefactof the percentagedata (Edwards and Berridge,
1994;Edwards and Mithen, 1995).LABIb(ii) appearsas a distinct subzonein the concentrationdata and this suggeststhat a real declinein Corylusavellana- type pollen occurred.The patternin the early Coryluscurve is replicatedover a
82 shorterdepth in percentageterms in LABVIIa althoughless clearly in the concentrations.
For LABVIIa, the peakin Corylusprior to its secondreduction dates to c. 9400 BP (8670cal. BC). At similar timesin LABIVa andLABVa, the Coryllis percentageprofiles displaysingle level reductionsalthough these do not appear in the LAM concentrations,and LABIIIa andLABVIa showa rapid rise in eachCorylus curve with no distinct fluctuations.LABVIIIa displaysCorylus percentagesand concentrations prior to c. 9300BP (8540cal. BQ which arc too low to reliably detectany fluctuations.It is probablethat the poorcrresolution of LABIII, LABIV, LABV andLABVI meansthat any fluctuationsin the early
Coryluspopulations have not beendetected.
High valuesof Ch:P arealso a featureof LABIlla - LABVIIIa, LABIb and LABIIc, but theseare probably a functionof low pollen influx ratesand charcoalinflux is probablya morereliable indication of amountsof charcoal presentin thesezones. Charcoal influx is greatestin LABIb (- 1000cm -2 cm -3 1) yf andis high in LABVIIIa (- 250 Cm*2Cnf3 yfl) but it is lower in the other profiles (Fig. 3.40).The muchhigher influx valuesin LABIb could result from its centrallocation and could thereforeindicate that most of the charcoal receivedwas of aerialderivation.
LABIb displaysgreater pollen influx ratesthan comparable zones in the other prorilesincluding LABVIIa andLABVIIIa. Greaterpollen influx may meanthat
LA131breceived a greaterproportion of aerially derivedpollen thanthe other prorilesin the early Holocene,perhaps from a wider area,and this argumentis reinforcedby the high levelsof charcoalin LABIb (Fig. 3.40; Section3.6.2).
83 Differential pollen influx may alsobe linked to the distributionof pollen over the lakessediment surface by watermovement, strcam. inputs and sediment disturbanceby wave actionor biota.
Edwards and Berridge (1994) and Edwards and Mithen (1995) have suggested that the Corylus reduction and increasein Ch:P in LABlb(ii) may rcflcct human activity. Although there is no archaeologicalevidence for human occupation this early on Islay, this doesnot mean that humanswere not presenton the island.
However, the Ch:P curve displays a gradual decline as Corylus increases.This suggeststhat either burning decreasedto allow the expansionof woodland or indicates that the charcoal was derived from aerial fallout and screenedout by expanding Corylus woodland, or was reworked (Edwards, 1990; Edwards and Berridge, 1994).
Inferredclimatic oscillationsas recorded in stableisotope and pollen data appearto haveinfluenced the earlyHolocene development of Corylusin Fife, easternScotland (Whittington et al., 1996;Edwards and Whittington, 1997b; andcf. Walker et al., 1994).The Coryluspollen profiles in LABlb(ii) and LABVIla may alsoderive from the influenceof climatic cooling (Sugdenand Edwards,in press)but the impacton the pollen curvesmay not be asmarked at Loch aBhogiadhas oceanic influences on the climateof Islay could have amelioratedany reductionin temperatures.The differencesbetween the Loch a!Bhogaidh profiles canalso be explainedby differencesin resolutionand the centralcores LABI andLABVII may providean earlierand more regional recordof Corylusas the slightly earlierdate for the hazelexpansion on LABVIIa appearsto indicate.
84 3.9.3 LABIc(i) Betuld - Corylus - Isoetes; LABIllb(i) Corylus - Cyperaceae; LABIVb(i) Corylus; LABVb(i) Corylus; LABVIb, Corylus;
LABVIlb(i) Corylus; LABVIIlb Corylus (c. 9400 - 8800 BP 18540- 7910 cal. BCI)
The transition to LABIc(i), LABVIlb(i), LABVII]b, LABIllb(i), LABIVb(i),
LABVb(i) and LABVIb is marked by a rapid expansionin the Corylus curves from c. 10 % up to c. 60 % UP from c. 9400 BP (8540 cal. BC). This is accompaniedby reductions in Betula, Poaceaeand Cypcraccaepollcn and herbs.
The earlier expansionof Corylus in LABVIII and LABI is discussedabove
(Section 3.9.2). On LABIII, LABIV and LABVI and LABVIII the Corylus expansionappears rapid in percentageterms, and more gradual in LABV and
LABVIII and this may be partly explained by different sedimentationrates for the profiles. A more rapid expansionof Corylus in the littoral, profiles may also be explained by an increasedlocal abundanceof this taxon. For the early
Holocene c. 9400 BP (8540 cal. BQ Corylus pollen influx in marginal profiles
LABIII and LABVI is as high as in the central profiles LABVII and LABI.
Humanactivity hasfrequently been linked to the expansionof hazelin the early Holocene(Smith, 1970;Simmons and Tooley, 1981;Huntley, 1993),and particularlythe rapid increasessustained after its introductionto the pollen profiles.It hasbeen suggested that burningof Coryluswoodlands may have assistedin propagatingyoung treesand high pollen productionmay arisefrom increasedflowering. However, high levelsof microscopiccharcoal in pollen diagramsdo not alwayscoincide with the main expansionin Corylus(Edwards, 1990),and at Loch a!Bhogaidh Ch: P is low at the Corylusrise andactually declinesin most of the profiles (LABIlIb(i), LABIVb(i), LABVb(i),
LABVlb(i)) comparedto previouslevels. For the reasonsgiven above(3.9.2), it
85 is unlikcly thatmicroscopic charcoal at the Corylusrisc at Loch a'Bhogaidh was the resultof humanactivity andthere is thereforeno evidenceto link humanimpact with the main expansionof Corylusat the site.
Huntley (1993) has suggestedvarious other reasonsfor the rapid expansion of
Corylus in the early Holocene, including climatic influences, soil type and the presenceof suitable ecological niches. As the Rhinns of Islay was not rcglaciated during the Loch Lomond Rcadvance(Ballantyne and Dawson, 1997) soils here may have been well developedcompared to upland areasof Mull and
Skye. With the exception of Beinn Tart aMhill there are no extremesof topography in the Loch a!Bhogaidh basin which is shelteredby low lying hills and the areamay not have experiencedextremes of erosion and associatedsoil instability. Soils may have beennutrient rich due to local limestone geology
(Ballantyne and Dawson, 1997) and hazel can prefer calcareoussoils (Moar,
1969). These factors may all have contributed to allow the rapid expansion of
Corylus at Loch a!Bhogaidh from c. 9400 BP (8540 cal. BQ and the lack of competition from other arboreal taxa (the pollen profiles indicate that only
Baula was presentin any significant amount prior to c. 9400 BP (8540 cal.
BQ would also have allowed Corylus to become the predominant arboreal taxon at the site.
3.9.4 LABIc(l) Betula - Corylus - Isoetes; LABVIlb(i) Corylus; LABb(II)
Corylus - Isoetes; LABIllb(ii) Corylus; LABIVb(i) Corylus; LABIVb(ii)
Corylus - Cyperaceae; LABVb(i) Corylus; LABVb(ii) Corylus -Isoetes;
LABVb(iii) Corylus; LABVlb(ii) Corylus (c. 8800 - 7700 BP [7910 - 6600 cal. BCj)
86 Following its initial expansionbetween c. 9700BP (9300cal. BC) andc. 9400 BP (8540cal. BC), Corylusbecomes the dominanttaxon in all prorilcs
(subzoncsdenoted above) reaching percentage values of up to 80 % TLP. Its concentrationsaverage between 50,000 to 100,000grains cm-3 and these do not differ greatlybetween profiles. The Corylusand TLP influx is greaterin 1, LABIc(i) andLABVIIb, over 10,000grains Cm&2 Yf comparedwith c. 2500- 2 I 5000grains CM, Yf in comparablezones in the peripheralprofiles.
Nevertheless,in all the subzones,Corylus influx is muchhigher thanpollen influx for any othertaxa andthis confirmsthe predominanceof hazelin the vegetationat Loch a!Bhogiadh from c. 8800BP (7910cal. BC).
There are a number of fluctuations occurring in the Corylus percentageprofiles
LABVIlb(i), LABVIlb(ii), LABIlIb(ii), LABIVb(ii), LABVb(i) andb(ii) and
LABVIb which arercflcctcd in the concentrationand influx data.None of these featuresare clearly recordedin eitherLABIc or canbe detectedin the LABII profile, probablydue to the lower samplingintervals at LABI andlow sampling intervalscombined with a shortdepth of resolutionat LABIL
AMS datingof someof the featureson the profiles and the extrapolation of dateson othersallows comparisonof featuresbetween cores, although there are no dateson LABIL Edwardsand Berridge (1994) suggested that a reductionin
Coryluspollen in LABIId(i) acCurredc.7300 BP (6190cal. BC) corresponding with a similar eventdated by extrapolationof datesin LABIc(i). Comparisonof thesefeatures on both profileswith a later reductionin LABIld(ii) indicatesthat the initial comparisonmay not be correct.It is suggestedhere that the Corylus reductionin LABIId(i) is earlierand corresponds with one of the Corylus reductionsevident between c. 8800BP (7910cal. BC) and c. 7700BP (6600
87 cal. BQ in the otherprofiles. The DCA plot of meansubzone scores (Fig. ývith 3.49a)suggests that LABlId(i) correspondsclosely zoneb in the new profiles.Edwards and Berridge (1994) also suggested that a transitionto peat occursin the LABIId(i) profile afterthe clay horizonat about300 cm, although
LOI percentagesat this point arelower thanwould be expectedfor peatat - 40 %. LOI valuesincrease later in the profile (260cm) to 80 % TLP, which is more representativeof peatand evidence in the otherprofiles indicatesthat the infilling of peattook placein the southwestof the basinfrom c. 7000BP (6010 cal. BQ. This would alsoindicate that the featurein LABIld(i) is earlierthan suggestedby Edwardsand Berridge (1994).
3.9.5 Discussion of the different features in the profiles
The featuresidentified in the pollen profiles and discussedbelow are indicated on the Corylus curves Figures 3.4 - 3.11,3.20 - 3.37 and on the right hand side of Figures 3.1 - 3.19,3.38 - 40,3.43 - 3.48. The depths at which the features occur are provided in Tables 3.4 - 3.6 and noted in the following discussion. in Tables 3.4 - 3.6 show the datesallocated to eachof the featuresidentified the profiles and summarizethe characteristicsof each feature. The overlap in standarddeviations for some dates(Fig. 3.50) meansthat some of the features could refer to the sameevent on different profiles. However, there are sufficient differences betweenmost of the datesto suggestthat these are more likely to representdiscrete events.
3.9.5.1 Feature]a, - LOIIM(i)
A reductionin Coryluspercentages and increases in Poaceaeand Cyperaccae aredenoted as Feature Ia (FI a; denotedin Table3.4 underFeature 1) in
LABIIIb(i) c. 9200BP (8440cal. BQ. A reductionin concentrations I
88 correspondswith this feature,and there are slight increasesin TD%. The characteristicsin FeatureIa aresimilar to thoseof the main FeatureI andboth arc likely to haveresulted from the sameinfluences, hcncc the interpretationof
FIa is given below,with that for Feature1.
3.9.5.2 FeatureI (Table3.4)
AMS datesfor similar eventsboth allocatedas Feature I (F I) in LABVIlb(i) and LABVIb are9030 ± 65 BP (8190cal. BC) and9000 + 75 BP (8180cal. BC) respectively.Linear extrapolationof dateson othercores indicates that F1 occursin the percentageprofiles of LABlIlb(ii), LABVb(i) and LABVlb(i). In LABVIIlb c. 9000BP (8230cal. BQ, a reductionin Corylusconcentrations only may alsorepresent F I, anda minor reductionin concentrationsat this time is alsorecorded in LABIVb(i).
Although temporaryreductions occur in the Corylusconcentrations in many of the profiles,LABIllb(i) andLABVIb concentrationsdisplay the startof a gradualdecline. Simultaneous reductions in total pollen influx of LABVIIlb,
LABIVb(i), LABVb(i) and LABVIb alsomark the beginningof a progressive decline. I
LOI valuesshow a greaterdegree of variationin the outlying profiles compared to LABIc(i) andLABVIlb. Early in LABVII]b, LABVIb, andLABIllb(i), the LOI curvesincrease whilst in the otherrelevant profiles LOI is reduced,but not necessarilyin conjunctionwith the Coryluspercentage reductions. There are no lithological changesnoted in the respectivezones in any of the coresand no significantincreases in Ch:P or TD% are evident.LABVIllb, LABIllb(ii) and LABVIb showpercentage increases in sometaxa including Ulmusand Quercus,
89 but thereis no comparativcincrease in the concentrations,suggesting that these percentageincreases are not a true representationof vegetationdynamics. There arc small increasesin Isoeteslacustris in LABlIlb(i) andLABVb(i) which correlatewith fluctuatingLOI values,and possibly result from increased nutrient availability dueto localisederosion or the inflow of streams.
The basesof zonesLABlIlb(i), LABIVb(i), LABVb(i), LADVlb(i), LABVIIb(i) and LABVII]b (in which FI occurs)date to c. 9300BP (8540cal. BQ where Corylusbecomes dominant in the pollenprofiles. High percentagesand concentrationsof Corylusavellana-type pollen in thesezones indicate that this taxonwas locally abundantand flowering profuselyat this time. Given the indicationfrom the pollen profilesthat Coryluswoodland was well established, the fluctuationLOI valuesassociated with FeatureI in LABIVb(i), LABVb(i) andLABVlb(i) may havearisen due to the input of erodedmaterials by streams.
LABVI is locatedclose to the slopesat the easternloch marginfrom where streamsare likely to havedrained the higher ground.There are no increasesin minor taxawhich may representstream side flora, andno increasesin damaged pollen which might indicatethe erosionof streamside sediments in FeatureI- However,increases in Ulmuspollen in mostof the peripheralprofiles may result from streamscarrying its pollen from the wider catchmentarea, as Mitts pollen is robustand preserves particularly well. Small increasesin Pinuspollen in
LABVIb(i) may alsohave been washed down by streamsfrom the better drainedslopes above the loch.
Although some reductions in Corylus percentagesappear in the profiles between c. 9300BP (8540cal. BC) andc. 8900BP (8010cal. BC), the fluctuatingpollcn concentration,pollen influx andLOI valuesfor FI and FIa do not clearly
90 replicatethe percentagedata, and there are no consistentincreases in other taxa which could indicatevegetation succession. This would reinforcethe indication that woodlandpatch dynamics and the influenceof strcam-bornepollen were responsiblefor FeatureI in the profiles.There is no suggestionof human impactat Loch a!Bhogaidh which canbe determinedfrom the profiles prior to c. 8900BP (8010cal. BQ. If peoplewere present in the loch catchmcntthen any impactthey may havehad on vegetationappears negligible and not reflectedin the pollen records.Indeed, small scalehuman impacts on vegetationcould producesimilar effectsin the percentagepollen profiles to thoseof early Holocenevegetation dynamics and pollen taphonomicprocesses as already described,and would thusnot be readily differentiatedin the pollen profiles.
3.9.5.3 Features2,3,4,5 and 6 (Table3.5) LABVIIb(ii), L4BVIIIb, L4BII]b(ii), LOI*(U), LARVb(i),L4BVIb, MBIM(i)
Thereare a numberof clearly definedfeatures of different datesrecorded in profiles LABIVb(ii), LABVb(i) andLABVIb. Theseare characterised mainly by reductionsin Coryluspollen percentagesand are denoted as Feature 2 (LABIVb(ii)), Feature3(LABVb(i)), Feature4 (LABVIb) andFeature 5 (LABIVb(ii)). Noneof theseare represented in LABIc, due to the lower samplingresolution and probably due to what appearto be very local impacts representedby eachfeature. A small Corylusreduction late in LABITIb(i) c.
8500BP (7540cal. BQ couldbe the sameevent as Feature 2 on LABIVb(ii), andreductions in Corylusc. 8500BP (7540cal. BQ and 8000BP (6910cal.
BQ in LABVIIb(ii) could be representationsof the eventscausing Feature 2, LABIVb(ii) andFeature 5, LABIVb(ii) respectively.However, on LABIIlb(i) andLABVIlb(ii) the reductionsare slight and arenot clearly presentedin
91 concentrationcurves which displaymany fluctuations.A reductionin Corylus alsooccurs in LABIld(i), but this rcmainsundatcd and is refcffcd to asFcature 6.
The dates for featuresin LABIVb(ii), LADVb(i) and LADVIb have potential overlap when standarddeviations are considered(Figure 3.26). However, the summary data (Table 3.4) presentno clear matchesbetween featuresand despite the overlap in standarddeviations, thesefeatures are as likely to represent different events as the sameepisode. The featuresare therefore treated separately.If the date of 8215+80 BP (7290 cal. BQ for the transition of
LABIlIb(ii) to the transition to LABIIIc is correct, then the reduction of Corylus pollen within LABIIIc may correspondwith Feature4 (LABVb(ii)). However, the pollen assemblagein LABIlIc suggeststhat this is in fact a later event correspondingwith zone c in the other profiles, and that Features2-6 do not presentthemselves in the LABIIlb(ii) pollen curves. Therefore the Coryhis reduction in LABIIIc is discussedbelow as a later event (Section 3.9.5.4).
Feature2 in LABIVb(ii), 8545+ 70 BP (7600cal. BC), showsa reductionin Coqlus percentagesand concentrations, and pollen influx alsoreduces at the startof an ovcrall dcclinc.LOI falls from 20 %- 10 % andthcre arc incrcascsin Cyperaceaeand Isoetes lacustris. Estimates of sedimentdepostion and pollen influx ratessuggest that the completereduction and recovery of Coryluslasted c. 120radiocarbon years.
Feature3 in LABVb(i), 8270+ 60 BP (7300cal. BQ may be characteristicof a greaterdegree of disturbancein the loch catchmentthan Feature 2 or may have been closer to the loch. There are again reductions in the Corylus percentageand
92 concentrationcurves, a decreasein pollen influx ratesand a decreasein LOI from 60 %- 40 %. The lithology of dark brown gyttja is interruptedby yellow silty gyttja between400 cm and403 cm andmarked increases in TD% occur.
Concentrationsof Isoeteslacustris increase substantially from 50,000grains cm73to 2,000,000grains cm-3 and other taxa includingPoaceae, Cyperaceae andFilipenduld increase.According to sedimentationrates the durationof this episodemay be calculatedas c. 100radiocarbon years.
Feature4 in LABVIb datesto 8155 + 70 BP (7160 cal. BQ. This feature also shows the characteristic Corylus reductions of the previous features,plus increasesin Cyperaceaeand a reduction in LOI from 30 - 18 %. Sediment deposition appearsto increaseslightly with this feature,probably due to erosion of the slopesto the eastof LABVI associatedwith decreasesin woodland there.
Sediment deposition rates and pollen influx suggestthat this phaseof reduction may have lasted c. 120 radiocarbonyears.
Feature5 in LABIVb(ii) occursfrom 8000+ 70 BP (6880cal. BQ. Corylus percentagesand concentrations and TLP influx arereduced as Cyperaceae and
Filipendula increase.A rise in LOI from 30 %- 60 % may be explainedby slowersediment accumulation rates, although it is alsopossible that additional organicmaterial was alsointroduced to the loch by erosion.The lengthof this
Corylusreduction is c. 800 radiocarbonyears.
Feature6 in LABIld(i) is defined by a reduction in Corylus percentagesand concentrationsand remains undated.However, datesfor the Corylus expansion early in LABIld(i) and the later Corylus reduction in LA13Ild(ii) may be estimatedby comparisonwith the other profiles as c. 9200 IBP(8440 BC) and c.
93 7700BP (6600BC) respectively.Assuming a uniform rateof sedimentationfor
LABlId(i), Fcaure6 hasan estimateddate of c. 8450BP (7470cal. BC) and may thereforebe the sameevent as Feature 2 in LABIVb(ii) or Feature3 in
LABVb(i). In additionto a reductionin Corylusin LABIld(i), a reductionin LOI valuesand change from gyttja to clay stratigraphyis recorded.Isoctes lacustrisincreases sharply to c. 800 % TLP andincreases in percentagesof Pinus, Quercusand Cyperaceae are alsocharacteristics of this feature.
Comparedto the earlierprofile zones,the combinedLOI andpollen percentage, concentrationand influx datasuggest less erosion and disturbance to vegetation in the catchment,and more stablesediment accumulation rates, despite
occasionaldisturbances (Features 2-6). The small reductionsin Corylus percentagesand occasional increases in Poaceae,Cyperaceac and herbs may be dueto the dynamicsof lochsidevegetation where temporary clearances may
occurdue to windthrow of trees,aging or disease.Fossitt (1996) estimates that storm eventsmay produceclearings in woodlandwhich may take 100- 200 yearsto recover.
However,the Corylusreduction in Feature5 (LABIVb(ii)) is much longerthan the otherdisturbances at c. 800 yearscompared with c. 100years, and is probablyrcflected in the centralprofile LABVIIb(ii) asa result of its duration.
Feature3 in LABVb(i) andFeature 4 in LABVIb arc of slightly earlierdate but the overlapsbetween dates at the 2 standarddeviation level suggesta continued period of interruptionin the Coryluswoodland. The changein stratigraphy coincidingwith Feature3 in LABVb(i) indicatesproximity to the loch edgeor a periodof intenseimpact on vegetation,and may appearin LABIld(i) as feature 6.
94 Although climatic changescould be responsiblefor the recordedreduction in
Corylus,there is no major eventcurrently recorded which would coincidewith this at c. 8000BP (6910 cal. BQ. Neitheris thereany direct evidencefor humanoccupation on Islay betweenc. 8500BP (7540cal. BQ andc. 8000BP (6910cal. BC), AMS datesfrom BolsayFarm and Gleann Mor confirm a Mesolithic presencein the areaof Loch a'Bhogaidhfrom c. 7500BP (6330cal.
BC) (Mithen, pers.comm. ). Stonetool assemblagesfrom Islay and Juracould dateto the earlierMesolithic (Mercer,1970,1978-80; Mithen andLake, 1996) andthe Mesolithic occupationof Rum is confirmedfrom 859CL+45BP (7640 cal. BQ (Wickham-Jones,1990). Pollen evidence from Arran (Robinsonand
Dickson, 1988)also indicates possible human impacts on Coryluswoodlands c. 8500BP (7540cal. BC). This circumstantialevidence raises the possibility that humanactivity at the loch edgecould havebeen responsible for at leastsome of the disturbancesdenoted by Features2-6, andespecially features 3-5 which areof long duration,may overlapin time andinclude the greatestnumber of disturbanceindicators. Corylus woodland may havebeen cleared for useas fuel or otherdomestic purposes, and hazelnuts gathered for consumption.The absenceof an increasein Ch:P in any of the profilesbetween c. 8500BP (7540 cal. BC) andc. 8000BP (6910cal. BC) indicatesthat burningwas not usedas a modeof clearance,if thesefeatures are anthropogenic in origin, ashas been assumedin the past(Smith, 1977;Simmons and Tooley, 1981).Although some charcoalmay derivefrom domesticfires (Edwards,1990) the low levelsof charcoalassociated with the abovefeature may not havebeen of local derivation andprobably represent a backgroundof aerial fallout or stream-borne componentfrom the wider catchmentarea. Although Features3-6 may be indicationsof the clearanceof Coryluswoods by Mesolithic peopleand could
95 be a precurserto a substantialphase of anthropogenicwoodland reduction at Loch a!Bhogaidh, (Section 3.6.2.5), in the absenceof any firni evidenceof humanimpact, an interpretationof naturalcauses must have at leastequal validity andthe windthrowof treesand grazing and trampling by animalscould alsohave had similar effects.
3.9.5.4 Feature 7 LABIVc,LAB VIc and d, LABWc and d, LABIc(i), LAB VIIIc and d, LABIld(H),LABIIIc and d, LAB Vcand d (Tables3.6 and 3.7)
It wasnoted by Edwardsand Berridge (1994) that reductionsin Coryluspollen andLOI andsmall increasesin Betuld,Pinus andherbs occurred in LABIc(i) anda Corylusreduction in LABIld(ii) (Figs. 3.4 and3.5). Increasesin Isoetes lacustris,which may indicateerosion, appear in LABIc(i) andLABIld(ii), and Ch:P increases,corresponding with the Corylusreduction, are evident in LABlId(ii) (Fig. 3.13).
Both BolsayFarm andGleann Mor excavationsprovided evidence of human When occupationin proximity to Loch a!Bhogaidh during the Mesolithic. and Lake (1996)suggest that the lithological evidencefrom Bolsayrepresents a long lengthyperiod of humanoccupation or at leastfrequent occupations over a periodof time. AnAMS dateon carbonizedhazelnut shell from a contextat
Bolsayof 7400+55BP (6300cal. BQ was obtained(Mithen, pers.comm. ).
This confirmsthe presenceof peopleat Loch a!Bhogaidh within the temporal rangeof the Corylusreductions discussed below. Indeed,Edwards and Berridge (1994)postulate that the fluctuatingCorylus values in LABIc(i) andLABIId(ii) (denotedas Feature 7) could haveresulted from humanactivity at the site.
96 The reductionin Coryluspollen andincreases in hcrbsand charcoal may also havearisen due to a changeto a cooler,drier conditionsfrom c. 7700BP (6600 cal. BC) until c. 7100BP (5930cal. BC) (Alley et al., 1997;Barber el al.,
1999;Willemse and T6mqvist, 1999).The increasesin microscopiccharcoal in LADIId(ii) may reflect increasesin naturalfires in a dry climate.
The additionalprofiles may assistin interpretingthe causeof the disturbances originally detectedin LABIc(i) andLABIId(ii) anddetail of Feature7 is providedin Figures3.58 and3.59 andTables 3.6 and3.7. In Feature7, increasesin Belula, Poaceaeand herbs are evident in all the profiles, and increasesin Pinus alsooccur in all profiles exceptLABIc(i). Thereare however, spatialand temporal variations within Feature7 in the profiles, particularlyin LOI, Isocteslacustris and Ch: P, anddifferences in stratigraphy(summarised in
Tables3.6 and3.7; Fig. 3.58).Although someof the temporaldifferences may relateto imprecisionsin radiocarbondates and the inability to precisely correlatesampled levels between the profiles,there nevertheless appears to be a shift in the locationof Corylusreduction and receipt of Ch:P at Loch a Bhogaidhfrom c. 7300BP (6190cal. BQ.
The initial Corylusreduction occurs c. 7700BP (6600cal. BQ until c. 7300BP
(6190cal. BQ in profundalprofiles LABIc(i), LABVIIc, LABVIc andprobably in LABVIIIc. (Althoughdates on this profile areestimated, the DCA plot Figure
3.49breinforces this correspondance).Whilst LABIIIc suggestsearlier dates for the Corylusreduction, it is probablethat theseare too old, andFeature 7 in LABIIIc may alsodate from 7700BP (6600cal. BC). In the otherperipheral profiles LABIVc also indicatesa reductionin Corylusfrom c. 7700BP (6600
97 cal. BQ but the reductionin LABVc(i) andc(ii) is very slight in percentage termsand does not occurin absoluteterms (cf. Figs. 3.8,3.21,3.32).
LOI is reduced in LABIc(i), LABVIIc, LABVIIIc, LABIVc and LABVIc whilst
LOI values in LABIIIc and LADVc(i) and c(ii) are relatively stable (Figs. 3.12
- 3.19). A reduction in woodland cover would result in greater erosion on slopes to the east of the site comparedto the flatter northerly and westerly flanks of the loch (cf. Fig. 3.1b) and this could explain the differences in LOI between the profiles. Indeed, LABIVc has the lowest LOI values of c. 5% and its stratigraphy changesfrom gyttja to grey silty clay. LOI percentagesappear to reduce slightly later in LABVIIc than in the peripheral profiles, and the timing of LOI reductions in LABIc(i) and LABVIIIc may be similar to that in
LABVIIc. The comparatively later reduction in LOI in the central profiles could be linked to the time taken to redistribute sediment originally deposited at the loch edges,and the subduedeffects of loch edge erosion in the central profiles.
Increasesin TD% in most of the zones(Tables 3.6 and 3.7; Figs. 3.12 - 3.19) may also be representativeof erosion; reworked, older pollen may have been subjectedto a greater degreeof oxidation and fungal attack than younger pollen, and thus show greater signs of pitting. LABVIc shows the greatestincreases in
TD% between c. 7700 BP (6600 cal. BQ and 7300 BP (6190 cal. BQ.
Ch:P and charcoalinflux increasesare only presentin LABVIIc (Fig. 3.18),
LABIIIc (fig. 3.14)and LABVIc (Fig. 3.17) c. 7700- 7300BP (6600- 6190 BQ. LABVIc may incorporatere-worked charcoal due to its receivinglarge amountsof erodedmaterial, and this may havebeen redistributed to LABVIIc. LABIlIc may alsohave received redistributed charcoal through wave actionand sedimentmovement but it as likely that the profile receiveda componentof
98 stream-bornepollen andparticulates derived from the wider catchmcntarea. This might explainwhy LABIIIc showsa greaterreduction in LOI anda greater increasein charcoalcompared to LABVc(i) andc(ii) despiteboth coresbeing situatedaway from the easternslopes and obvious sources of erosion.
From c. 7300BP (6190cal. BQ Coryluspercentage and absolute reductions intensify in LABIVc and occurin LABVd (referto Figs. 3.7,3.803.36,3.58). It is suggestedthat the Corylusreduction in LABIld(ii) probablyoriginated at a similar time. In LABIVc (Fig. 3.15) andLABIId(ii) (Fig. 3.13),LOI percentagesreduce and the stratigraphyof LABIVc changesfrom gytjja to grey silty clay. The LABVd stratigraphychanges to peatand LOI increasesto c. 80 % (Fig. 3.16).Microscopic charcoal increases are particularly evident in
LABIVd, LABVd andLABIld(i) (Fig.s 3.15,3.16,3.13).
In the otherprofiles (LABIlId(i), LABVId, LABIc(i), LABVIId(i) and LABVI11d)(Figs. 3.36,3.5 8) Corylusand LOI recover,although in LABVIId(i) andLABIIId(i) Corylusand LOI fluctuate,as do the microscopiccharcoal curves.These fluctuations can be explainedby the sporadicredistribution of sedimentacross the loch by waveaction, and by streamsincorporating pollen from the wider catchmentwhere Corylus woodland has recovered.
The additionaldata from multiple profiles enablesa firmer interpretationof Feature7 thanpreviously provided by LABI andLABIL it is probablethat colderclimatic conditionsfrom c. 7700BP (6600cal. BC) (Alley et al., 1997; Barberet al., 1999;Willemse and T6mqvist, 1999)resulted in a reductionin woodlandtaxa including Corylus,by increasingsusceptibility to disease, reducingflowering andtherefore propagation, or by direct lossesassociated
99 with cold andexposure. Trees on exposedslopes where erosion is more likely to occurwould havebeen more susceptible the effectsof climate.Most of the profiles showincreases in Betuld,Poaceae, Cyperaccae and tall-herbs at the
Corylusdecline, representing wider areasof birch scrubin the vicinity of Loch a!Bhogaidh. Slight increasesin otherarboreal taxa may derivefrom re-worked materialdue to increasederosion, or wind- andstrearn-borne pollen transported from a wider areain a moreopen landscape. Similar mechanismscould be responsiblefor the increasesin microscopiccharcoal seen in someof the profiles.An increasedfrequency of naturalfires in a dry openenvironment could alsohave contributed to charcoalincreases.
Nevertheless,the archaeologicalevidence confirms a humanpresence at Loch a!Bhogaidh which coincideswith a phaseof climatic deterioration.In cold, dry conditionsthe removalof woodlandby peoplewould havecompounded the effectsof a harshclimate. It may alsobe consideredthat earlierdisturbances in woodlandat Loch a'Bhogaidhrecorded c. 8000BP (Feature5) may have resultedfrom humanactivity but,in a warmer,wetter climate, the effectsof this activity would not havehad sucha greatimpact. If humanactivity was at least partially responsiblefor Feature7 from c. 7700 - 7300BP (6600- 6100cal. BC), it is unlikely that fire was usedto clearwoodland close to Loch aBhogaidh in view of the low valuesof Ch:P and low charcoalinflux in the peripheral profiles.If the increasesin charcoalare not a reflectionof naturalfires or the re- working of older deposits,then these may derivefrom the fallout of domestic fires concentratedto the north andeast of the loch (cf Edwards,1990; Edwards andBerridge, 1994) or may derivefrom the burningof vegetationat a distance from the site.
100 Evidencefrom the GISPice coresuggests that a gradualclimatic warming beganfrom c. 7300BP (6190cal. BC) (Alley et al., 1997)which could explain the recoveryin Coryluswoodland from this time in the widcr Loch a!Bhogaidh catchmentas indicated by LABIlld(i), LABVIld(i) and LABVId. The continued reductionin Corylusto the southand west of the loch as indicatedby LABIVc, LABVd andpossibly LABlId(ii) may not thereforebe explainedby the direct effectsof climate.The changeto peatin LABVd c. 7100BP (5930cal. BQ may haveresulted from climatic changesand the flatter areassurrounding the loch may havebecome too waterloggedto supportCorylus. However, the earlier phaseof Corylusreduction also appears to havebeen concentrated to the north- west of the site, andthe Coryluscurve is relativelyunchanged in LABVc betweenc. 7700-7300BP (6600- 6190cal. BQ whenwoodland reduction would haveencouraged peat formation. The increasesin microscopiccharcoal
(both Ch:P andinflux; Figs. 3.15and 3.16) associatedwith the latter Corylus reductionin LABIVc andLABVd from c. 7300BP (6190cal. BC) suggestthat peoplemay havecleared woodland at the loch edgesand this may have encouragedthe growthof peat.
The dataprovided by multiple profiles at Loch a!Bhogaidh therefore suggest that, althougha periodof cooling in the Atlantic may havereduced woodland in the loch catchment,and produced episodes of erosionfrom c. 7700BP (6600 cal. BC) peoplemay alsohave cleared areas of vegetationat Loch a'Bhogaidh andmay havebeen responsible for enhancingthe effectsof climatic changes from c. 7700BP (6600cal. BQ.
101 3.10 Infilling of the loch with peat If the completestratigraphics of eachprofile areconsidered (Fig. 3.61),then the developmentof peatoccurs in somethe peripheralprofiles (excludingLADVI) shortly after the inferredepisode of woodlandclearance discussed above, c. 7000BP (5870cal. BQ. Peatappears in the stratigraphyLABVd at 360 cm,at c. 7100BP (5930cal. BQ andin LABIle at 255 cm c. 7000BP (5870cal. BC), with associatedincreases in LOI, Cyperaceaeand in LABIIe increasesin Ranunculaccaeand EquisetuniYeat occurs in the LABIII profile c. 6500bP (5460cal. BC) althoughthis might be later,given the possiblyerroneous dates on this profile, or earlier.The latter is perhapsmore likely given the subdued topographyin the westof the Loch aBhogaidhcatchment, as this areawould be moreprone to waterlogging.A changein hydrologicalconditions could have limited the subsequentregeneration of Coryluswoodland to the north of the loch andthe Corylusrecovery in LABIIId probablyincorporates' Corylus pollen from furtherafield.
Peatdevelopment occurs c. 4000BP (2530cal. BQ in LABI andLABVII and c. 5000BP (3860cal. BQ in LABVIII. Given the potentialinaccuracies in have extrapolatedradiocarbon dates, peat inception for theseprofiles appearsto occurredaround the sametime. The relatively late developmentof peatin LABIV (ca. 3000BP [ 1200cal. BQ andLABVI (ca. 1000BP [1060cal. AD]) wasprobably influenced by the positionof thesesites close to slopesat the east of the loch. Theseslopes appear to haveremained wooded later thanelsewhere, given the relativelyhigher percentages of Corylusin LABVId andLABlVd until c. 6500BP (5460cal. BQ.
102 It is therefore possible that the activities of Mesolithic people at Loch a!Bhogaidh may have influenced the expansionof peat in the area.Mesolithic impacts elsewherehave been linked to peat development(Caseldinc and Hatton,
1993; Simmons and Tooley, 1981) and at the Bolsay site near Loch aBhogaidh, podzolisation followed by peat developmentis evident in post-occupation horizons. Climatic changesmentioned earlier may also have assistedin peat developmentby reducing Corylus woodland at Loch a!Bhogaidh. It is unlikely, however, that sufficient waterlogging for peat formation could haveoccurred c. BC) dry 7700 - 7300 BP (6600 - 5910 cal. which was particularly cold and (Alley et al., 1997). From c. 7300 BP (5910 cal. BC) climatic conditions becamegradually warmer and wetter, and the reduction in Corylus to the southwestof the loch from this time may have encouragepeat growth, although woodland reduction may also have occurred due to human activity compounding the time-lag effects of earlier climatic events.
3.11 Conclusions interpretations The pollen profiles from Loch a!Bhogaidh reinforce the original of LABI and LABII (Edwards and Berridge, 1994) with regard to the succession of Holocene taxa. However, different featuresoccur in separateprofiles, and thesevariations enhanceinterpretation.
The multiple representationsof the Corylusrise at Loch a'Bhogaidhsuggest that centrallyplaced profiles may receivegreater amounts of airbornepollen from a
wide area,but other central profiles do not show changes.
The variations in influx between theprofiles through time also indicate that sediment focusing and pollen distribution may not be constant over the catchment.
103 Additional detail of vegetationhistory was created by the useof multiple profiles, andbetween c. 8500BP (7540cal. BQ andc. 8000BP (6910cal. BQ,
a numberof fluctuationsoccur in the variousprofiles at different times,
althoughthere is no evidenceto directly confirm the causeof these.Openings in woodlandmay result from naturalenvironmental changes but it is alsopossible that at leastsome of theserepresent Mesolithic activity. In contrast,what appearedto be a small scalereduction in Corylusin LABIc(i) is providedin much greaterdetail in the multiple profiles asa climatic eventwith possible humanimpact superimposed. This detail andthe arrayof AMS dateson the profiles enhancesan interpretationof climatic changeand human interference alteringcatchment vegetation, and raises the possibility that temporalchanges in
the concentrationand location of humanactivity may be detectedat other sites.
The Loch a'Bhogaidhprofiles tend to confirm the hypothesisthat deep,centrally placedprofiles may not detectsmall scaleimpacts at a loch edge,and where considerablevegetation changes are evident, their representationsin central
coresmay be mutedand lack detail.The resultsfrom Loch a!Bhogaidh present a
strongcase for the employmentof multiple profiles in pollen studiesprovided theseare combined with rigorous,AMS datingand close resolution sampling.
104 CHAPTER 4- POLLEN ANALYSIS OF SEDIMENTS FROM LOCH AN T'SUIDIIE, ROSSOF MULL
4.1 Site description
Loch an t'Suidhe(Figure 4.1; Plate4.1), NGRNM 371215, is locatedon the Rossof Mull in the southwest of the islandand directly southof the Isle of Ulva. Approximately50m in diameter,it lies lessthan 500 rn from the northerncoastline of the Rossof Mull in an areaof flat bog surroundedby low hills (maximum85 m O.D. ). The hills supporta sparsevegetation of Poaceacspp., Cyperaceae spp. and Calluna vulgaris. The surroundingbog is dominatedby Myrica gale andother vegetation includes Sphagnum spp., Eriophorumangustifolium, E. vaginatum,Erica tetralix, Juncusspp.,
Potentilla erectaand P. palustris.The openwater of the loch supports Nymphaeaalba andthe fringing hydrosereis dominatedby Sphagnumspp. The extentof openwater in the pastat Loch an t'Suidheis unknown,and the
surroundingmire probablyrepresents an areaof infilling.
Currentlythere are no streamsflowing into the loch, but it may havebeen stream-fedin the past.If this wasthe case,then stream input may not have beenlarge due to the subduedtopography surrounding the loch andthe flat valley floor in which it is situated(See Plate 4.1).
A pollen surface sample(Fig. 4.2) from the southernedge of Loch an t'Suidhe show the predominanceof Myrica gale in the surrounding mire.
The Bunessanarea on the Ross of Mull is virtually treeless,and the lack of arboreal pollen in the surface samplesindicates that most of the Corylus avellana-type pollen in the assemblageis probably Myrica, despite the low pollen productivity of this taxon (Birks and Huntley, 1983).
105 4.2 Background to the presentstudy
Lowe andWalker (1986a,1986b) carried out pollen analysison a core from the southernedge Loch an t'Suidhe. Their pollen samplingintervals were between8 and 10 cm with TLP countsof about3 00 TLP, andthere was no
quantificationof microscopiccharcoal. Their profile, LS, is undated,and a
proposedchronology was constructedby comparisonwith otherpollen profiles from Mull. Only the percentagedata were published. For the present study,a furthercore (2LS) was obtainedfrom the southernedge of Loch an t'Suidhe.Pollen samples were extracted at intervalsof 2-4 cin andTLP countsof at least500 were achieved. Microscopic charcoal quantification anddamaged pollen analyseswere also undertaken and 6 AMS dates(Table 4.1) allow the estimationof sedimentdeposition times and pollen influx
rates.
The lithologiesof LS and2LS aresimilar, consistingof dark brown to black
gyttja of Holocenedate, which overlaysa seriesof grey andgreen clays, of
Lateglacial origin.
As this researchaims to detecthuman impacts in the early Holocenepollen record,samples for pollen wereextracted from the Holocenegyttja sediment until a declinein Ulniuspollen andincrease in Poaceaepollen weredetected.
Thesewere evident in the Lowe andWalker (1986b)profile andwith an estimateddate of c. 500013P (3860 cal. BQ, the Ulmusdecline may be used
as a markerto ensurethe captureof the vegetationhistory of the Mesolithic period.
106 4.3 Pollen diagrams and zones
The relevantportion of the publishedLS pollen diagramis providedin Figures4.3a and b. The percentage,concentration, pollen influx, damaged pollen andsummary data profiles for 2LS areprovided in Figs. 4.4 - 4.8. An age-dcpthcurve for the Loch an t'Suidheprofile showing2SD errorbars is presentedin Figure4.9 anddetails of radiocarbondates are presented in Table4.1. The zonesallocated to 2LS correspondclosely to thosefrom Lowe andWalker's (1986b) and comparisons of appropriatezones and allocateddates are summarised in Table4.2.
4.4 Discussion of zones
4.4.1 2LSa c. 9800 - 7900 13P(8910 - 6810 cal. BQ
Subzonc2LSa(i) Betyly - Cyperaccae- Filipendula c. 9800 - 9200 BP (8910
- 8500 cal. BQ The baseof the profile in 2LSa(i) dates from 9870+75 BP (8910 cal. BQ and the subzoncincorporates high percentagevalues of Betula, Poaceae,
Cyperaceaeand Filipendula in addition to the aquatic taxa Myriophy1him alterniflorum and Littorella uWflora. Corylus avellatia-type pollen is present in low amounts and although Ch:P values are high, c. 100 x 10-8,low charcoal influx compareswith low TLP influx values. These are probably a 1 productof rapid sedimentationrates which average125 cm yf in 2LSa.
LOI risesfrom 20 % andcontinues to rise throughthe transitionto 2LSa(ii).
A DCA plot of mean subzonescores (Fig. 4.10) suggeststhat 2LSa(i) is very different to the other profile zones and if more than two sarnpledlevels had been available then 2LSa(i) would have been classified as a zone in its own right. The arboreal pollen is limited to Betula; low influx for this taxon implies that Betula was scarcein the early Holocene and damp grassland with tall-herbs appearsto have been the predominant vegetation. Steadily
107 increasingLOI valuesreflect soil developmentand the presenceof Myriophyllumalternij7orum indicates the ameliorationof climate from earlierLatcglacial conditions. The microscopic charcoal in 2LSa(i) may derivefrom reworkedcharcoal deposited through erosion, or from aerial fallout in a relativelyopen environment. The reductionsof Ch:P at the transitionto 2LSa(ii) probablyresult from the screeningout of charcoal particulatesby expandingwoodland. A reductionin charcoalinflux is also discernible,despite very low values.There is thereforeno suggestionat Loch an t'Suidheof anyburning at the Corylusrise which may be linked to its expansionand possible human interference (cf Smith, 1977;Edwards, 1990).
The datefor the Corylusrise is estimatedby linear extrapolationas c. 9250 BP (8500cal. BC). This is slightly later thanthe datefor the Corylusrise at
Loch a!Bhogaidh c. 9400BP (8650cal. BC) (Chapter3) andon Arran, c. 9400BP, (8650cal. BC)-(Robinsonand Dickson, 1988),but is earlierthan recordedelsewhere on Mull (Walkerand Lowe, 1985,1987;Lowe and
Walker, 1986a,1986b), and Skye(Birks, 1977;Walker andLowe, 1990;
1991).Local environmentalfactors are probably responsible for the varying timesof the Corylusexpansions in the Inner Hebridesand these are discussedin Chapter7.
LSa(ii) Betyly - Coalus. c. 9200 - 7900 BP (8500 - 6810 cal. BC) Within 2LSa(ii) Betula and Corylus are the dominant taxa and SaUxis also present,but decreasesin the latter part of the subzone.Ulmus and Quercus expand c. 8700 BP and 8200 BP (7790 and 7280 cal. BC) respectively.
These datescompare well with those from other Inner Hebridean sites (Birks and Huntley, 1983; Birks, 1989; refer to Chapter 7). Poaceaeand Cypcraceae persist in low amounts,and occasionalherbs <2% TLP are recorded,
108 includingPotentilla, Ruinex acetosa and Succisa pratensis. Equisetuni and ferns(Pteropsida (monolete) indet. ) arealso present. The low arborealpollen influx values(including Betuld and Corylus)in 2LSa(ii) suggestthat woodlandnear the site remainedscarce until c. 8000BP (6860cal. BQ althoughrapid sedimentaccumulation rates will alsohave influenced pollen influx values.Poaceae gradually reduce in percentageand absolutevalues within 2LSa(ii), probablyas a responseto the expansionof woodland,but the dampgrassland communities recorded earlier in 2LSa(i) persist throughoutthis zoneas is evidentfrom the assemblageof herbs,Equisetum andferns.
From c. 8000 BP (6860 cal. BC) there is a reduction in LOI from 50 % to 25
%, which indicates the introduction of minerogenic material to the loch and values remain low (- 25 O/o)within the subsequentsubzone, 2LSb(i). There are no reductions in the pollen profiles which could be associatedwith this although a contemporaryincrease in pollen influx values at c. 8000 BP
(6860 cal. BQ suggestsa decelerationof sedimentationwhich could have changedorganic accumulationsand influenced LOL
4.4.2 2LSb c. 7900 - 6850 BP (6810 - 5870 cal. BC) 2LSb(i) c. 7900 7000 BP (6810 5870 cal. BCI CyDcraceae - - -Poaceae- - Calhina.
At the base of 2LSb(i), Betuld and Corylus pollen are predominant, but
Pinus sylvestris, Ulmus and Quercus are well represented,and the
assemblagesuggests that a mixed deciduouswoodland inhabited the Loch an t'Suidhe catchmentat 7875+60 BP (6720 cal. BC) and pine may have been locally present. Shortly after this time reductions in Betula and Corylus percentages,concentrations and pollen influx values occur, and these correspondwith increasesin Poaceae,Cyperaceae, Calluna vulgaris and
109 Sphagnuni.The pollen of variousherbs is alsorecorded within this subzonc, andinclude Potentilla, Rumex acetosa and Chenopodiaccae. A DCA plot of taxon scoresfor 2LSb (Fig. 4.12)is notdissimilartoone for taxa from the whole profile (Fig. 4.13)although axes one andtwo arc reversed.This might indicatethe generaluniformity of the 2LS profile (dominatedby arboreal pollen).Nevertheless, Calluna vulgaris andFilipenduld aremore closely associatedwith Poaceaeand Cyperaceae in Figure4.12 which reflectsthe decreasein arborealpollen andincrease in herbsin 2LSb.
The pollen data therefore indicate that local Betula-Corylus woodland at
Loch an t'Suidhe declined, and was replacedby a herb rich grassland.The increasesin Sphagnummay representlocal waterlogging resulting from a decline in woodland cover. This is dated to between 7700 and 7100 BP (cal.
8440 and 7890 BP) and coincides with a period of climatic cooling detected in the GISP2 ice core record from central Greenland(Alley et al., 1997), 14C marine sediment and records from the Hudson Bay area, easternCanada
(Barber et al., 1999) and lake sedimentsfrom Greenland(Willemse and
T6mqvist, 1999). The reduction in arborealpollen is also similar to Feature
7 identified in the Loch a!Bhogaidh profiles (Chapter 3). 2LSb(i) is therefore likely to representcolder, drier conditions which occurred in the early Holocene.
However,it is Betula and Coryluswhich experiencethe greatestreductions in both percentageand absolute terms. Other arboreal taxa sustaintheir values,and even if theserepresent the backgroundnoise of long distance pollen transport,a reductionin otherarboreal taxa, particularly Quercus and Ulmus,could be expectedin a periodof sustainedcooling, particularly one so severeas to affectBetula, a pioneeringtaxon capable of surviving relativelypoor climatic conditions.
110 LOI values at the transition to 2LSb(i) appearstable c. 28 %, and although
LOI continues to increasethrough the mid to latter part of the zone, the
curve is punctuatedby severalextreme reductions in LOI, suggestingthe periodic inwash of mineral matter into the loch, although no minerogenic
layers or stratigraphic changeswere apparent.Erosion would be expectedif woodland cover in the local areawas reduced.Ch: P increasessubstantially 3 1 in 2LSb(i) and charcoal influx increasesto from c. 50 CM"2CM, Yj, to c. 300 2 3 1 1. CM, CM, yf and reachesa maximum of 1340 CM-2CM-3 ýf These' increasesin the charcoal recordsmay have arisen from a reduction of the woodland screensurrounding the site due to the effects of an increasingly cold climate. In 2LSb(i) however, charcoal influx is much greater than in
2LSa, where the environment appearsto have been less wooded and where a
greaterproportion of charcoalwas probably of aerial derivation. The
reduction of woodland in 2LSb(i) could have createda greater amount of mineral inwash to the loch, and thus introduced reworked charcoal, but again, LOI was lower in the earlier zone where Ch:P and charcoal influx were lower. If the charcoal increaserecorded in this subzonehad resulted
from washed-incharcoal then it would appearto be contemporarywith the pollen, andis likely to representlocal burning.Indeed, the appearancesof
Callunavulgaris, Potentilla (possiblyerecta) and Pteridium aquilinum reinforcethe casefor local burning,as these may all be favouredby fire
(Page1976, Hobbs and Gimingham, 1988; Edwards et aL, 1995).
Although a drier, colder climate could have increasedthe potential for natural fires between c. 7700 and 7100 BP (cal. 8440 and 7890 BP)
(Edwards et al, in press) the woodland reduction at Loch an t'Suidhe between c. 7800 and c. 6900 BP (6900 and 5780 cal. BC) in 2LSb(i) may reflect Mesolithic cultural activity. The potential anthropogenicindicators
111 Aster-type,Cirsium-type, Arlendsia, Chenopodiaccae and Run= acelosaarc
all recordedin this subzonealthough these taxa would alsooccur naturallyin woodlandopcnings and in the disturbedground resulting from cold, dry
climatic regimes.There is currentlyno evidencefor a Mesolithic presencein
the southof Mull, but Livingstone'sCave on the nearbyisland of Ulva was occupiedduring this period(Bonsall et al., 1991,1995),a middenof probableMesolithic datewas recently discovered at Dervaigin northern Mull (Robins,1998) and there is a wealth of evidencefor Mesolithic
occupationin the otherInner Hebrideanislands (Chapter 1, Section1.2). It seemsinconceivable that manyother parts of Mull, and especiallycoastal areas,would not havebeen visited by humancommunities at this time. Other pollen profiles from westernScotland display reductions in woodlandtaxa of
similar Mesolithic dates.Human activity may featurein the profiles from Loch a!Bhogaidh, Islay, superimposedon the effectsof climatic changes
which occurredbetween c. 7700- 7000BP (6600- 6010cal. BQ (Chapter 3). A similar phaseof inferredhuman impact also occurs at Loch Cholla, in Colonsaybetween c. 7800- 6050cal. BP (Andrewset al., 1987;Edwards, press)although this may alsohave been initiated by climatic changes.Two
shortphases of woodlandreduction and increases in Calluna vulgaris and charcoalare recorded in the pollen recordfrom Loch an VS11,South Uist
from c. 8025BP to c. 7900BP (6810- 6770cal. BC) and from c. 7850 BP
to c. 7790BP (6640- 6770cal. BQ (Edwards,1996; Edwards, 1998). Woodlandreductions of possibleanthropogenic origin on Rum occur later,
c. 6000- 5750BP (4560- 4610cal. BC) and 5250- 4950 BP (4100- 3800 cal. BC). The pollen profiles from Loch an VS11in particularcorrespond closelyto thosefrom Loch an t'Suidhe in that reductionsin Betuld and Corylus.and increasesin Calluna vulgaris, Poaceac,Polentilla andCh: P
occur(Edwards, 1996,1998). The shorterreductions in woodlandand increasesin herbsare perhaps more likely to be the result of humanactivity,
112 but, whencombined with the abundantarchaeological evidence for Mesolithic occupationof the InnerHebrides, that humanactivity could be incorporatedwith the effectsof climatechange in the Loch an t'Suidhe profile. A multiple profile strategysuch as that employedat Loch a!Bhogaidh may haveassisted in determiningthe extcntof any spatial distributionin vegetationpattems within 2LSb(i) andcould thereforehave assistedin identifying whetheror not the subzoneis purely a rcflection of climate,or whetherhuman impacts on vegetationalso occurred.
If the increasesin microscopiccharcoal at Loch an t'Suidhe andLoch an t'Sll areanthropogenic in origin, they may representthe useof fire to create or maintainheath and grasslandas grazingfor gameand fowl (Simmonsand
Tooley, 1981;cf. Edwards,1996,1998). However, woodland may havebeen clearedby othermethods, and fallout from the domesticfires of hunter- gathererscould haveincreased amounts of fossil charcoal(cf Edwards, 1990;1996; 1998) although absolute charcoal values and Ch:P at Loch an t'Suidhe appearvery high, andmight be thoughtunlikely to derivesolely from campfires. A drier, colderclimate may haveincreased the potentialfor naturalfires, but alsothe Mesolithic useof fire, whetheroriginating as intentionalor accidental.
2LSb(ii) Betyla - Coalus c. 7000- 6850BP (5870- 5770cal. DO In the transitionto 2LSb(ii), Befuldand Corylusdisplay a progressive recoveryas Poaceae and Callunavulgaris reduce.However, within the subzone,Betula percentages,concentrations and influx decline,possibly as the result of an expansionof Corylusand Salix. Fewerherbs are recorded comparedto 2LSb(i) andPoaceae, Cyperaceae and Callutia vulgaris are substantiallyreduced from formervalues and absolute charcoal values and
113 Ch:P remainlow. This subzoneassemblage indicates a recoveryin woodland following the cessationof the proposedprecccding human activity.
A fall in LOI from 40 % to 5% in 2LSb(ii) reflectserosion and the depositionof mineralmatter into the loch. Thereare small increasesin TD% within this subzoncwhich reinforcethe suggestionof erosion,although the only notableincreases in damagefor a singletaxon occur for Calluna vulgaris.These occur as its overallpercentages fall, andthe valuesfor damageare possibly a functionof the low numbersof Callunapollen recordedrather than any changein taphonomyfor heatherpollen. There are no otherindications of erosionfrom the pollenrecord and indeed, most of the arborealpollen curvesincrease. It is possiblethat the arborealpollen is of very local origin whereasactual erosion may havetaken place in the wider catchment.The fall in LOI may appeardue to increasedminerogenic input from streamsflowing into the loch. If erosiontook placeclose to the loch, thenthis wasprobably small scale,and the compazýtiveabundance of arborealpollen over non-arborealpollen may haveprevented the pollen indicatorsof taxa from openand disturbed ground being detectedin the pollen record.
Alnusglutinosa reaches its rationallimit at Loch an VSuidhein 2LSb(ii) c.
6800BP (5700cal. BC), a dateearlier than predicted by isopollenmaps
(Birks andHuntley, 1983)and pollen isochrones(Birks, 1989).Although thereis no evidenceof humanactivity in the pollen spectraof 2LSb(ii), the possiblehuman impacts recorded earlier in 2LSb(i) could havebrought aboutchanges in the Loch an t'Suidhecatchment, and createdan ecological niche for the expansionof Alnus in the area.Indeed, Alnus is consistently presentin the earliersubzone 2LSb(i) in low amounts(< 2% TLP) andthis may signify thatAlnus was a minor componentof local woodland,or was at
114 least presentin the catchmentalthough at some distance from the pollen core
site (cf. Bennett, 1983). The earlier reductions in Betula-Corylus woodland
in 2LSb(i) may have createdwetter downslope habitats increasesin Salix in
2LSb(ii) could also reflect this.
4.4.3 2LSc Ahtus - Corylus - Betuld - Quercus c. 6850 - 5200 BP (5770 - 4100 cal. BQ
In the transition to 2LSc at 6845+55 BP (5720 cal. BQ, the Betula, Pinus,
Ulmus and Quercus profiles maintain the percentageand absolutevalues of
the previous subzone,but Ulmus and Quercus expand later in the zone.
Although a temporary decline appearsin the Corylus percentagecurve as
Alnus glutinosa values rise, Corylus influx values are not greatly reduced
and the pollen assemblageindicates the local presenceof a mixed woodland.
Only low values of Poaceae,Cyperaceae and herbs are recorded, and
arboreal pollen influx curves are comparatively high, confirming the predominanceof woodland in the area.
A reductionin LOI occursat the baseof 2LSc,and a similar reduction occurslater in the zoneat 655 cm, indicatingtwo short episodesof minerogenicinwash to the loch. Thereare no simultaneousdecreases in the pollen percentageprofiles or increasesin damagedpollen which could
signify erosion,but reductionsin the pollen influx curvesc. 6500BP (5460
cal. BQ andc. 6100BP (4820cal. BC) coincidewith the reductionsin LOL Althoughmicroscopic charcoal is not recordedin notablequantities at every level, comparativelyhigh charcoalinflux, concentrationsand Ch: P are inten,nittently recordedcompared to 2LSb andin particularat the first reductionin LOI at the baseof the zone.Increases in charcoalmay derive
115 from episodesof inwashinto the loch andperiodic increases in sediment accumulationmay alsohave occurred, thus reducingpollen influx,
From c. 6000 BP (4560 cal. BQ, LOI values remain around 70 % for the remainder of the zone and charcoal concentrations,influx and Ch:P fall, suggestingless erosion. The LOI values are high for gyttja and could representthe infilling of the loch with peat. Alternatively the high LOI values could suggesta build-up of rich, organic matter and low decomposition rates resulting from the very local presenceof damp woodland. The location of core 2LS may not have been far from the loch edge at the south of the basin from c. 6000 BP (4560 cal. BQ.
Pinuspollen percentagesand absolute values decrease from 7% UP to <2 % UP in the lower part of 2LSc.This occursshortly after the decreasein
LOI, but Pinuspercentages are probably too low to reflect a local presence (Bennett,1984; Willis el al., 1998)and the Pinuspollen influx curve is also
low comparedto thoseof otherarboreal pollen types.This indicatesthat
Pinuswas not a major componentof local woodlandand most of its pollen
in the profiles probablyderives from long distancetransport. The reduction
of Pinus pollen occursc. 6650BP (5600cal. BC) in 2LSc, andthis is early comparedwith reductionsat othersites in Scotlandwhere climatic deteriorationis probablyresponsible (Bennett, 1984; Gear and Huntley,
1991; Willis et al., 1998).Pinus influx valuesthroughout the zoneare not
muchlower thanpreviously recorded, and the apparentreduction of Pinus may be an artefactof the percentagediagram in responseto increasesin Alnus, Ulniusand Quercus.Increased woodland cover could alsohave
screenedout Pinus pollen from the site, thusmitigating its representationin 2LSc.Woodland cover would alsohave screened out charcoalparticulates
andmay thereforebe partly responsiblefor the reductionsin charcoalinflux
116 in the laterpart of 2LSc. In this zone,the pollcn of Poaccacand Cypcraccac is recordedmostly in valuesof <2% TLP, andherbs such as Filipendula andPotentilla featuresporadically and are also only presentin very low amounts,reinforcing the suggestionthat woodlandbecame denser at Loch an t'Suidhe from c. 6800BP (5770cal. BC).
4.4.4 2LSdBetula-Ahius-Coryltis-Poaceaec. 5200-4550BP(4010-
3260 cal. BQ
Within 2LSd, the arboreal pollen percentagesand absolute values recorded in the previous zone are maintained for Betuld, Pinus, Quercus and Corylus.
Alnus glutinosa expandsslightly however, and there is a clear decline in
Ulmus at c. 5 100 BP to <2% TLP. UInjus recovers in the later part of the zone, but does not achievethe values recordedearlier in the profile. The
Poaccaccurve exhibits an expansiontowards the end of 2LSd, and herbs are recordedmore frequently within this zone, and include Plantago lanceolata and Ruinex acetosa.LOI values are high - 75 %, although percentages reduce towards the end of the zone.
The datefor the Ulmusdecline at Loch an t'Suidhe correspondswith the
dateof c. 5100BP (3860cal. BC) attributedto the main British elm decline. A variety of causeshave been hypothesised for this event(refer to Chapter 7), includingdisease, climatic deteriorationand increased human activity. It
is entirelypossible that climatic changeswere responsible for the declinein Uhnusat Loch an t'Suidhe.The site is in an exposedposition on the Rossof
Mull peninsulaand subjected to salt-ladenwesterly winds andmin from the Atlantic Ocean.Increases in rainfall could haveincreased the Icachingand
acidificationof local soils andthe expansionof. 41nuswithin 2LSd may be a responseto increasedwetness. Ulmus may haveinhabited the slopes surroundingLoch an tSuidhe ratherthan the wettervalley floor, which will
117 havehad a greatersusceptibility to erosionand leaching due to increased
rainfall. Diseasemay alsohave caused the deathsof Ulmusspecimens and resultedin woodlandopenings which could alsohave contributed to increasederosion into the loch.
Although humanactivity may alsohave contributed to the Uhnusdecline in someareas of the British Isles,either directly or by exacerbatingthe cffccts of naturalenvironmental changes, there is no strongevidence in 2LSd which
could supportthe theoryof a humanpresence at Loch an t'Suidhc c. 5 100 BP (3860cal. BQ. Small increasesin Poaceae,Plantago lanceolata and Rumexacelosa and a reductionin LOI probablyresult from the exposureof soils andassociated erosion. This could haveresulted from the mechanisms
describedabove, but could alsohave been caused by humanactivity. Although Ch:P remainslow, thereare small increasesin charcoal
concentrationsand influx althoughthese are not substantialand charcoal is only recordedintermittently in this zone.
Whilst the loch doesnot currentlyreceive stream-bomed inputs, periods of higherrainfall may havecaused streamflow into Loch an t'Suidhe.The weak
signalof cultural activity, includingthe intermittentcharcoal record, associatedwith the Ulmusdecline in 2LSd may thereforebe explainedas havingarisen from the far edgesof the catchmentif stream-bornepollen is a
componentof this zoneassemblage.
4.5 Summary of Holocene vegetation history at Loch an t'Suldhe to c. 5000 BP (3860 cal. BQ
The pollen data indicate that Betuld scrub and grasslandcomprised the early
Holocene flora at Loch an t'Suidhe from c. 9800 BP (8910 cal. BC). Corylus
expandedshortly aRer c. 9200 BP (8500 cal. BQ and by c. 8000 BP (6860
118 cal.BC) Ulmusand Quercus were introduccd to thewoodland flora. A decreasein woodlandfrom c. 7750BP to 6900BP (6620- 5780cal. BQ andexpansions in Callunavulgaris, Poaccae and herbs may have been causedby theactions of Mesolithicpeople. A coincidentalincrease in Ch:P suggeststhat burning could have occurred locally, and was probably anthropogcnicin origin.
A regenerationof woodlandis indicatedfrom c. 6900BP (5780cal. BQ and the expansionof Alnus shortly after this may be linked to increasedwetness within the Loch an t'Suidhecatchment. The reductionof Pinus following the Alnus expansionmay alsobe linked to increasedwetness of the valley floor, andAlnus outcompetingPinus at the site, althoughfalls in Pinuspollen
could equallybe explainedby the screeningout of its pollen by expanding woodlandor as an artefactof the percentagedata. The earlierreduction of
Betula- Coryluswoodland between c. 7750to c. 6900BP (6620- 5780cal. BQ could havebeen influential in hydrologicalchanges at Loch an tSuidhe andassisted in the spreadof Alnus there.
From c. 6800BP (5750cal. BQ a mixed woodlandof Alnus, Corylus,
Quercus,Ulmus and Betuld persistedat Loch an t'Suidheand after this time thereis no strongevidence in the profiles for local fires or humanimpacts.
At 5235± 55 BP (5100cal. BC), Ulniuspollen valuesdecline as NAP valuesincrease and LOI valuesreduce. Whilst humanactivity may have causedthis a combinationof climatic and local cdaphicinfluences and diseasemay alsohave been responsible for this event.
119 CHAPTER 5- ULVA. Discussionof multiple profiles from two peat filled basins
5.1 Site descriptions and background to the project
Two peat-filledbasins on Ulva to the north andwest of Livingstone'sCave wereselected for sampling.
A!Chrannag bog (Figure5.1; Plate5.1), NGRNM 431391, is a topographic mire at 74m O.D. located250 rn eastof the summitof A!Chrannag on Ulva and500 rn north of the archaeologicalsite of Livingstone'sCave. The dimensionsof the bog are approximately300 m east-westand 200 m north-
southand it is borderedby higher,rocky groundwhich is exposedto southwesterlywinds. The bog surfaceappears to havebeen cutover in the past.Small pocketsof Betulapubescens and Salix spp.occur around its perimeterwhich is drier, andPleridiuni aquilinumis presenton the rockier slopes.The surfaceflora includesEriophorum angustifolium and E. vaginatum,Potentilla erecta,Myrica gale, CallunaVulgaris, Dactylorhiza maculatassp. ericitorum andD. purpurella amongstother species.In wetter areasJuncus spp., Potamogeton spp., Hydrocotyle vulgaris andMyosotis scorpioidesare common.
Both east-westand north-south depth transects were measured to provide an indicationof sub-peattopography (Figs. 5.1,5.2) and to assistin locating coring positions.Two coreswere taken in the A!Chrannag basin, one from the deepestpart of the bog in the north-west(AC 1; NGR NM 432 391), and the other from a pocketwhere peat deepens in the north-southprofile at the southernbog edgeclosest to the cave(AC2; NGR NM 432 389) (Fig. 5.1).
120 The peatin both profiles overlaysa greenclay horizonwhich gradcsdown into sandierdeposits.
Livingstone'sCave mire (Fig. 5.1; Plate5.2), NGR NM 4313 86, was formedin a topographicdepression to the westof Livingstone'sCave. It is situatedto the southof A'Chrannagbog at c. 30 m,O. D. and 300 m to the westof Livingstone'sCave. The bog is also500 m.north of the basaltcliffs of Ulva's southcoast which areexposed to southwesterlywinds from the sea.The basinitself is relatively shelteredas it is boundedby low rocky outcropsto the west andshallow slopes to the east.
The slopesaround the bog supportCorylus avellana, Betula pubescens and
Salix spp.woodland with an understoreyof ferns,mosses, and Poaccae. Floweringherbs include Allium ursinum,Oxalis acelosella, Primuld vulgaris and Campanuldrotundifolla. The north of the bog is drier than the south,and vegetation is dominatedby Calluna vulgaris andMyrica gale. As wetnessincreases to the souththe flora becomesmore diverse.Species presentinclude Eriophorum angustifolium, Filipendula u1maria,Juncus effusus,Potentilla crecta,P. palustris andIris pseudacoris.Polantogeton spp.and Sphagnum spp. are alsopresent.
The dimensionsof the mire are approximately250 m.north-south and 150m east-west.Depth transects were measured across the bog, oneacross the north-southdimensions, and three at intervalsacross this from castto west
(Figs. 5.1; 5.3). Two coreswere extracted, LCI to the middle of the bog (NGR NM 4313 86), andLC2 to the easternedge of the bog (NGR NM 432
385) (Fig. 5.1).The peatin both coresoverlays sandy green clay under which is a furtherdepth of organicsediments. These also overlay sandy
121 greenclays which becomesyellow brown andsandier with depth.Profile stratigraphicsare included in the pollen diagrams,(Figs. 5.20 and5.21).
5.2 Minerogenic sediments
The mineralsediments underlying all the cores,and those intersecting the Livingstone'sCave peat profiles (refer to lithologieson Figs. S.10,5.11, 5.20,5.21) apearto be similar to those from an un-namcdmirc 200 m west of Livingstone'sCave at 20 m O.D. anddescribed by Bonsall et aL, (1991: 9-10) as
" marine sandsgrading upwards into shallow lagoonal facies brackish marine silts with (possibly intertidal) peat deposits,these giving way to brackish and eventually freshwaterpeats"
It is mentionedin ChapterI that early Holocenesea levels were higher than thoseat present,and the Main Postglacialshoreline for Ulva is estimatedas c. 10 ra O.D. (Dawson,1984; Ballantyne and Dawson, 1997). Prior to peat formationthe basinwould be lower thanthe present20 m O.D. andcould haveexperienced marine and lagoonalconditions. There are deposits in Livingstone'sCave ýog (to the westof the caveat 30 m O.D. ) which could be of marineorigin asthey appearsimilar to thosedescribed by Bonsall el al., (1995).AMS datessuggest that the accumulationof minerogenic materialsunderlying the mire peatoccurred in Lateglacialtimes (see below) when sealevels were lower (Ballantyneand Dawson, (1997). An adjustment for marineerror on datingof 450±50radiocarbon years (Harkness, 1979; Lowe andWalker, 1984)does not move the datesinto the early Holocene.
The minerogenicsediments underlying the organicdeposits in the Livingstone'sCave profiles were assessedfor their dinoflagellatecontent
(Harland,1998; Section 2.1.7) in an effort to determinepast levels of
122 salinity. The underlyingmineral sediments from the AChrannagbasin were alsoinvestigated, as these were visually similar to thosefrom Livingstone's Cavebasin. The searchfor dinoflagellatecysts was unproductive(Harland,
1998)and although sand is not the ideal mcdiurnfor dinoflagellate preservation,there is no conclusiveevidence to indicatethat marine conditionswere responsible for the sandyaccumulations in Livingstone's Cavebog. The currentaltitude of the mire is c. 30 rn OD with a maximumof 3 rn of Holocenepeats, and, even allowing for erosionof early Holocene sedimentsand an hiatus(Section 5.3), it is unlikely that early Holocenesea levelswould havebeen sufficiently high to penetratethe area.The similarity betweenthe Livingstone'sCave sediments and those underlying the peatat A'Chrannagbasin suggests similar origins,and A! Chrannag basin at 74 rn
OD must surelybe abovethe rangeof the earlyHolocene sea levels.
Nevertheless,it is possiblethat highersea level in the early Holocenecould haveincreased the salinity of the water tablein the Livingstone'sCave basin andthis may havehad implicationsfor sedimentationand vegetation successionat the site (Sections5.3.2,5.5.2).
Whencombined with AMS datesand pollen evidence(discussed below,
Sections5.3 and 5.5) the aboveevidence suggests that the lower part of the
Livingstone'sCave profiles, underlyingthe mostrecent peat, dates to the Lateglacialand is probablynot directly of marineorigin.
5.3 AMS dates
5.3.1 AlChrannag bog profiles
A total of six AMS radiocarbon dateswas obtained for profile AC I and two datesfor profile AC2 were secured.These are provided in Table 5.1. Age - depth curves for ACI and AC2 are shown as Figures 5.4 and 5.5
123 rcsPcctivcly.A diagramof the sprcadof datcsfrom Ulva,at 2 standard deviationsis presentedin Figure5.6.
Dates on AC I confirm that this provides the oldest policniferous sediments of the two profiles. The lowest date on ACla of 8625±75 BP (7740 cal. BQ is slightly younger than the succeedingdate of 8705±75 BP (7870 cal. BQ,
although when the standarddeviations on eachdate arc considered,both are
statistically indistinguishable at the Ia level (Figs. 5.4 and 5.6). If older
minerogenic material was mixed into the peat through erosion then this
could account for the reversal in dates,but the contemporaneousLOI
percentagesare high and there is no indication that erosion occurred. The
dated horizons all come ftom the samecore and thus undulating sediment
horizons would not explain the reversal in dates.As the next date of
8540+95 BP (7770 cal. BC) on ACI (at 282 cm) is also very similar to the dates is two preceding-, dates it is possible that the overlap of the earlier a
result of rapid sedimentation(Fig. 5.4) or compaction of peat (cf Aaby,
1986) and the range of statistical error on the earlier datesmeans that, at 2
standarddeviations, the first could be older than the second(Fig. 5.6). In
order to calculate pollen influx and sediment depostion rates, the date at the
baseof the profile AC I, 8625±75 BP (7740 cal. BQ at 304 cm has been
used and the later date of 8705+75 BP (7870cal. BC) is discardedas these
two dates are virtually the same.
Linear extrapolationsuggests that the baseof AC2 datesto c. 6500BP (5460
cal. BQ andpeat inception here is far later thanat AC I. As AC2 is sited at the southernedge of the bog thenit is possiblethat the infilling of the basin wasrelated to topography,progressing from a depressionin the bog centre is andoutwards to the edges.Although the depthof sedimentsat AC2
similar to that at AC I, linear extrapolationof datessuggests that the top of
124 the peatat AC2 is approximately2500 years old wlicrcasat ACI, the top of the profile is estimatedto dateto c. 4000BP (2530cal. BQ (Figs. 5.4 and 5.5). Although sedimentaccumulation rates may havevaried abovethe uppermostdates of eachprofile, the bog surfaceappears to havebeen cutovcrin the past.
The remainderof dateson AC I andAC2 displayno dctcctablcanomalies. Indeed,the expansionof Ulnius,Quercus and, 41nus on AC I andAC2 is virtually synchronousat 5685±60BP and5695+55 BP respectively(both 4530cal. BC). This suggeststhat correspondencesbetween the profilcs are reliable,despite the apparentlate introductionof thesetaxa to pollen catchmcntarea compared to elsewherein the Inner Hebrides(see below and
cf. Lowe:and Walker, 1986;Birks, 1989;Edwards and Bcrridge, 1994; Chapter7).
5.3.2 Livingstone's Cavebog proriles
AMS dateswere alsoobtained for the two profiles from the Livingstone's
Cavemire; five on LC I andtwo on LC2. Theseare shownin Table 5.1 and
asage-depth curves in Figures5.7 and5.8. The spreadof datesat 2 standard deviationsfor the Livingstone'sCave cores is includedin Figure 5.6.
Profile LC I hasthe only dateson what appearto be Lateglacialorganic
deposits,possibly well humified peats(LCI a, LC2a) which sit beneathan horizonof greensandy clays (LClb; LC2b). The uppermostorganic deposits in LC Ia havean estimateddate of 11,120+100BP (11,260cal. BQ and
theseprobably mark the endof the LateglacialInterstadial (although the date could be too old if the sedimentsampled included a degreeof re-worked
older material[cf. Edwards,1990]). The baseof the LatcglacialInterstadial organichorizon was datedto 10,760±85BP (10,770cal. BQ which is
125 evidentlytoo youngand the dateis unreliable.The claysoverlying the organicdeposits in LCIb andLC2b probablyaccumulated in the Loch LomondStadial. An hiatusin sedimentationoccurs bctwccn the top of the
Loch LomondStadial clays and the overlyingpeat within the basin.
The base of the peaty depositsoverlying the clay was dated on each of the profiles LCI and LC2. Despite the similarity of the pollen assemblagesin
LC I c(i) and LC2c(i), the datesdiffer by over 2000 years (7765±60 BP (6580 cal. BC) and 5515 + 55 BP (4350 cal. BC) respectively). Whilst the contamination of the basal peatsby the infiltration of min water and humic acids could have produced dateson both profiles which are erroneously young (cf. Walker, 1984; Edwards et al., 1991) subzonesLClc(i) and
LC2c(i) incorporate similar taxa, suggestinga similar date for peat initiation
at both core sites.
Either of the datesrecorded at the baseof the Holocenepeats could be
correct,therefore two age-depthcurves for LCI andLC2 havebeen
constructedusing the datesrecorded at the baseof eachcore (Figs. 5.7 and 5.8). When the dateof 7765+60BP (6580BC) is usedfor linear
extrapolation,the curvesfor LCI andLC2 aresimilar whenextrapolated backto the LateglacialInterstadial dates, although this would assumea
continuousand steady rate of sedimentationinto the Holocenewhich is
obviouslynot the case.Ascribing the dateof 7765+60BP (6580cal. BC) to the baseof the Holocenesediments in the Livingstone'sCave basin also
meansthat Ulmus,Quercus and Alnus appearearlier than in the A!Chrannag basin.As Ulmusand Quercus are introducedto most pollen profiles in the Inner Hebridesand west coast of Scotlandby c. 8000BP (6910cal. BC)
(Birks, 1989;Whittington and Edwards, 1997) this is not especially
problematic,but the expansionof, 41nusin the profiles LC1c(ii) and LC2c(ii)
126 shortly after 7765+60BP (6580cal. BC) is earlierthan most of theAlnus pollen recordsfor the islandsalthough the expansionof this taxonis notoriouslyasynchronous (Birks, 1989;Chambers and Elliott, 1989;Bennett andBirks, 1990).
An alternativescenario is that the basaldate of 5515±55BP (4350cal. BC) is the correctdate for the inceptionof peatin the Livingstone'sCave basin.
At 2 standarddeviations, this dateoverlaps those on AC I andAC2 (5685±60BP and5695+55 BP (both 4530cal. BC); Fig 5.6) when Ulmus, Quercusand Alnus expand.This suggeststhat from this time the slopesand higherground surrounding Livingstone's Cave basin could havebeen woodedwith hazel,birch, oak andelm, which could havereduced minerogenicsediment input andpromoted the accumulationof humic materialand peat formation in the basin.The slopesare not too steepto supportwoodland cover, as they are coveredin a deciduouswoodland flora today.Also, if the later dateof 5515+55BP (4350cal. BC) on LC2c(i) is correct,then reductions in Ulmuspollen in LC I c(ii) coincideswith thosein the A'Chrannagbasin, occurring between c. 4800 - 4600 BP (3570- 3310 cal. BC). Although local ecologicalfactors may havecaused the declineof Ulmusat different times in the two basins(refer to chapter7), other reductionsin Pinus and Corylusin the latterparts of LC I c(ii) andLC2c(ii), appearat comparabletimes accordingto linear extrapolationif the dateof
5515±55BP (4350cal. BC) is used,and suggest that this is probablycorrect. If the dateof 7765±60BP (6580cal. BC) is used,the the Pinus and Corylus reductionsappear at very differenttimes in eachprofile.
Thereis no readily apparentsolution as to which, if either,of the datesfor the Livingstone's Cave basal peatsis correct. Although it is possible that neither is erroneous,the similarities of the Livingstone's Cave pollen profiles
127 suggestthat their basaldates should be the same.Confirmation of similaritiesbetween subzones LClc(i) andLM(i), and LCIc(ii) and LC2c(ii), is providedin the DCA plot of subzonemean scores (Fig. 5.9). If peatinitiation beganon a physicallydisturbed clay surfacea mixing of older andyounger materials could have produced the different dates. Sedimentationmay alsohave been very slow, andthe datesmay be the result of averagingfor a very thin layer.Contamination, either in the field or laboratory,might alsobe consideredas a factorcausing the different dates.
As the taxa in the Livingstone'sCave pollen profiles havevery different percentagesand concentrations to thosefrom A!Chrannag, it is not easyto employintcr-site comparisons. However, if a basaldate of 5515+55BP (4350cal. BC) is appliedto both of the Livingstone'sCave profiles, the expansionof arborealtaxa. in the Livingstone'sCave basin matches that in the A!Chrannag profiles, the Vrn declines'correspond between basins, and the comparativereductions in Pinusand Corylusin LC1c(ii) andLC2c(ii) coverthe sametimescale. The preferreddate for the baseof LC I c(i) and
LC2c(i) is thus 5515+55BP (4350cal. BC) andthe main presentationand interpretationof LCI andLC2 ispredicateduponthis assumption.
5.4 Presentationand discussionof A'Chrannag pollen data Percentage,concentration and pollen influx diagramsfor both the
A!Chrannag profiles arepresented as Figures 5.10,5.11 and5.14 - 5.17. Alnus is excludedfrom the pollen sumin the Ulva percentageprofiles dueto the dominanceof its pollen in many of the sampledlevels, relating to the high pollen productivityof this taxon.Summary diagrams of LOT,charcoal data,TLP concentrationand influx andtotal damagedpollen are shownas Figures5.12 and5.13 and damaged pollen percentagediagrams as Figures
5.18- 5.19.Local assemblagezones and comparisons of profiles AM and AC2 are detailedin Table5.2.
128 5.4.1 ACIa Betula- Cyperaceae- Filipendula; ACIb(i) Colylus - Calluna - Sphagnunt;AClb(fi) Corylits- Poaceae- Cyperaceac;c. 8700-
6250BP (7790- 5160cal. BQ The major pollen typesrecorded in AC Ia areBetula, Cyperaccae and
Filipendula. Othertaxa present in low amountsinclude Ranunculus acris- type, 71alictrum,Plantago majorIniedia and Rumex acetosa, in additionto the aquaticsLittorella uniflora, Potamogetonnatans-type and Menyanthes trifoliata. The assemblagesuggests the presenceof birch scrubpunctuated by openareas dominated by sedgesand tall hcrbs.The presenceof aquatic taxa indicateslocaliscd wetness and possibly ponding on the mire surface.
In the latter part of AC I a, Corylusand Callunavulgaris expandand Betula percentagesand concentrations slightly decline.Values of Ch:P show substantialincreases from low amountsrecorded earlier in the zoneand suggestthat fire may haveengineered these taxon changes. Sphagnunj sproresalso increase in the upperpart of the zone.
The Corylusexpansion at A!Chrannag (i. e.a consistentpresence >2% TLP) shortly beforec. 8500BP (7470cal. BC), is later thanat manyof the other sitesin the Inner Hebrides(Birks, 1973;Birks, 1989;Edwards and Berridge,
1994;Loch an t'Suidhe,this volume),but is earlierthan at Kinloch, Rum
(possibly7800+75 BP (6700cal. BC); Hirons andEdwards, 1990) and Loch
Fada,Skye (7500+120 BP (6330cal. BC); Birks and Williams, 1983).
Nevertheless,the datesfor the A'ChrannagCorylus expansion correspond with thosefor othersites on Skye(such as Loch Choira Ghobhain [8650±150BP (7840cal. BQ, Birks andWilliams, 1983])and from the west of Mull (e.g. Torness,Glen More [8760±140BP (7900cal. BC), Lowe andWalker, 1986a;1986b]). The significanceof the spreadof datesfor the
129 Corylusrise is discussedin Chapter7, but the dateof c. 8500BP (7470cal. BC) for the Corylusexpansion on Ulva is not untoward.
It is interestingthat the rise in Coryluscoincides with an apparentexpansion
in Callunaheath and increases in charcoalconcentrations, influx and Ch:P. Smith (1970)suggested that the postglacialCorylus expansion in the British Islesmay havebeen assisted by fires, possiblystarted by Mesolithic people. This claim hasbeen largely refuted, as microscopic charcoal values usually declineor are low or absentat the Corylusrise in mostpollen profiles (Edwards,1990; Huntley, 1993).However, increases in charcoalinflux and
Callunavulgaris at the Corylusrise in AC Ia providea goodindication of local fires. Burning may assistthe propagationof Calluna seedlingsand thus
the expansionof Callunaheath (Hobbs and Gimingham, 1987), and burning at A!Chrannag could be anthropogenicin origin due to peoplesetting fires to createbrowse for game(Simmons and Tooley, 1981). As Corylusavellana is fire tolerantcompared to otherarboreal species, fires may alsohave been
utilised to promoteits growth aboveother arboreal taxa. Corylusis easily
coPpicedand provides both food andwood suitablefor the constructionof tools andother domestic uses. However, the north of A'Chrannagbog is
relatively exposed,and natural fires may alsohave arisen from lightning strikesduring storms,producing the sameeffects.
An alternativeexplanation for the behaviourof Corylusat AC Ia is
migrationallag. Ulva is a small islandand although close to Mull, it is separatedfrom it by a channelof deepand turbulent water andhigh cliffs. Boyd andDickson (1986) suggest that this as one of the reasonsfor the late
expansionof Coryluson Arran. Nevertheless,tidal currentshave been
suggestedas a methodby which Coryluspropagules may spread(Salmi,
130 1963cited in Huntley andBirks, 1983).Further discussion of the expansion of Corylusin the Inner Hebridesis providedin Chapter7.
Following the Corylusexpansion in ACIb(i) thereare reductions in the
Corylusand Callunavulgaris percentage and absolutepollen valuesbetween
277 cm.and 264 cm (c. 8400- 7900BP [ 7450- 6810cal. BQ which coincidewith reductionsin absolutecharcoal values and Ch:P, andincreases in Betula,Osmunda regalis, Pteropsida (monolete) indet. and Sphagnum.
The laterpart of AC I b(i) showsa recoveryin valuesof Coryhisand Calluna vulgaris,but in AClb(ii) Corylusvalues reduce again although Calluna valuesremain stable. Betuld percentagesand influx valuesalso decrease in AC1b(ii) and arecontemporaneous with increasesin Salix, Sphagnumand
Pteridium,and Pterposida (mono. ) indet. andCyperaceae in the latter part of the subzonewhere a temporaryreduction in LOI from - 90 % to - 65 % also occurs.Charcoal influx andCh: P valuesremain low in AClb(ii).
The intial decreasein Corylusand Callunain AC I b(i) may be linked to the cessationof fires in the area.The appearanceof Osmundaregalis and Sphagnumin ACIb(i) areindicative of an increasewetness which could havereduced the flammability of local vegetation.If humanactivity was responsiblefor the earlierinferred localized burning, then a cessationin fires possiblycaused by the out-migrationof peoplefrom the areawould presumablyresult in lower charcoalrepresentation, and reductions in Corylusand Calluna.There is no reasonwhy increasedwetness itself should havedirectly reducedCorylus populations as hazel is tolerantof damp conditions.Its reductionalong with thosein the charcoalcurves may reinforcethe suggestionthat early HoloceneCorylus populations at A!Chrannag may havebeen influenced by fire.
131 ACIb(ii) datesfrom c. 7900 BP (6810 cal. BC) which coincides with the start of a period of climatic cooling in the Atlantic mentioned in earlier chapters(Alley et al., 1997; Barber et al., 1999; Willcmse and T6mqvist,
1999). This could be responsiblefor the further reduction in Corylus within this zone, although in contrast to Loch an t'Suidhe, microscopic charcoal values are low. The increasesin ferns, willow and Sphagnuin in AClb(ii) mentioned earlier suggesta waterlogged environment which may have adversely affected local birch-hazel woodland. High levels of damaged pollen in thesesubzones may result from pollen trapped in water on the mire surface for some time, or from re-worked pollen washed into the bog by numerousstreams which would increaselocalised waterlogging.
Thereisno recoveryof Betula and Corylusin ACIb(ii) betweenc. 7300-
7100BP (6190- 5930cal. BQ which occursat Loch a!Bhogaidh and Loch an t'Suidhe,and may be associatedwith a recoveryin temperaturesin the
North Atlantic. This suggeststhat if the vegetationreflected in AClb(ii) is the result of climatic impacts,then the changeswere sufficientto prevent any considerablerecovery of arborealspecies. However, Loch a!Bhogaidh andLoch an VSuidhewere both lakes,and may havereceived pollen from a muchwider areathan the peatfilled basinof AChrannagwhereasAC I b(ii) may reflect local changes.
5.4.2 AClb(iii)(Poaceae-Cyperaceae-Sphagitum), AClc; AC2a,
AM; c. 6250 - 5300 BP (5160 - 4150 cal. BQ
In AClb(iii) the reduction in Corylus and Betuld pollcn values continues, but
Sahr, Calluna and Ptcropsida,(mono. ) indet. also reduce. There is a small but temporary increasein Pteridium aquilinum within this subzonebut the main increasesoccur in Poaceae,Cyperaceae, Filipendula, Potentilla-type
132 and Succisapratensis. There are also increasesin microscopic charcoal values within ACIb(iii) particularly Ch:P, although charcoal influx is not high.
In correspondingAC2a Betula values are also low and Corylus declines
from high percentagesat the baseof the zone as Cyperaccae,Pteropsida
(mono.) indet. and Pteridium increase.Although Ch:P is high in AC2a.
charcoal influx is low and comparablewith that in Mb(iii). A variety of
herbs are also recorded in AC2a and include Aster-type, Filipendula, Succisa pratensis and Apiaceac undiff. Calluna vulgaris is also present in this zone.
Very intenseburning may havereduced birch andhazel woodland at A!Chrannag, and this could alsosignificantly affect heathlandecology by
enforcinga transitionto Pteridiumand then Poaceae (Hobbs and Gimingham,1987; Legg et al., 1992).Greater increases in microscopic
charcoal,particularly charcoal influx, might be expectedif suchintense local burninghad occurred, but unknowntaphonomic factors mean that local fires
cannotbe discountedin reducingthe local extentof Coryluswoodland and heathlandat A'Chrannagc. 6250BP (5160cal. BC). Low influx valuesfor
Betula and Corylusuntil c. 5700BP (4350cal. BC) in AC I andAC2 indicatethat this woodlandwas not particularlydense, and the arrayof herbs
recordedin AC I b(iii) andAC2a could occurin naturalclearings and on the
rocky slopesto the southof A!Chrannag basin. There is evidencefrom Livingstone'sCave that peoplewere present on Ulva c. 5700BP (4650cal. BC). Whilst low charcoalinflux suggeststhat they did not employ fire to
clearwoodland it is possiblethat tree felling could haveassisted in expandinggrassland, possibly to encouragegrazing. Certainly grazing pressurecan maintain and widen areasherb-rich grassland (Bennett el al., 1992).
133 Pinus pollen is presentin both profilesbut in higherpercentages and concentrationsin AM. Howeverthis is likely to be a statisticalartcfact as Pinus influx in both profilesis similar. If local,Pinus may haveinhabited the thinnersoils of rocky outcropssurrounding the bog but thereare no major differencesin damaged(broken) Pinus pollen percentagesin eitherAM andAClb(iii) which might indicatelocal or regionalderivation.
Ulmus,Quercus and Alnus glutinosa expand c. 5750BP (4590cal. BQ within the latter part of AC I b(iii) andat the transitionto AM, with increasesin percentages,concentrations and influx. Shortly after this date,at the transitionto AC I c, Betula displaysa markedexpansion in percentages, in concentrationsand influx, anddespite percentage reductions, an increase Corylusinflux occurs.Calluna, Poaceae, Cyperaceae and herbs are
considerablyreduced in both profiles in additionto Ch:P andcharcoal influx.
It appearsthat from c. 5750BP (4530cal. BC) episodesof local burning The may haveceased thus allowing the expansionof woodlandover the site. increasein woodlandwould alsohave provided an effectivescreen to
charcoalparticulates from further afield. Whilst someheath and grassland habitatsmay havedeclined at the expenseof woodlandthe absenceof any
notablereduction in influx for Calluna,Poaceae and Cyperaceaesuggests that the percentageand concentration decreases in thesetaxa is a statistical
artefactdue to largeamounts of arborealpollen in the assemblagesof AC Ic
andAM. Wood macrofossilsin the peatof both profiles from c. 5750BP (4530cal. BC) (mainlyAlnus glutinosa) indicate that at leastAblus
woodlandwas growing at the bog surface.
134 Belula, Corylus,Ulnius and Quercusinflux ratesare greater in AC I c, comparedwith AC2b, but Alnusglutinosa displays much higher influx rates in AC2b. This suggeststhat Alnus populationswere greaterto the southof the bog, anda mixedwoodland dominated by Beluld was probablypresent in the north of the basin.There is alsoa disparityin percentagesof damaged pollen for arborealtaxa between AC Ic andAC2b. Total damagedpollen percentages,particularly total pitted percentages,are greaterin AC2b, and this may be an indicationthat the pollen is derivedfrom further aricld and
not in the vicinity of AC2b. However,Alnus influx ratessuggest that this taxonwas local to AC2b, yet pitted percentagesfor Alnus reachc. 30 % in this zone,compared with c. 20 % in AC I c. The differencesin damaged pollen percentagesbetween the profiles may thereforerelate to sediment
accumulationrates. Slower sediment accumulation at AC2 may meanthat
pollen was exposedto the atmosphereand surface wct/dry cyclesfor longer
periodsof time. The high amountsof Alnus comparedto otherarboreal taxa in AC2b may meanthat the southof the bog wasparticularly wet, and Wilkinson andHuntley (1987)found that pollen may deterioratefaster in
bog pools ratherthan in raiseddry areas,although they acknowledgethat
this conflicts with otherevidence (e. g. Dimbleby, 1985;Moore and Webb, 1991).Such pools may containlarge amounts of Sphagnumwhich could trap
pollen grainsand expose them to oxidisationand fungal attackat the pool
surface.
At c. 5750BP (4530cal. BC), the expansionof Ulmlis, Quercusand Alnus in the A'Chrannagprofiles is considerablylatcr thanrecorded for other
profiles in the Inner Hebrides.The Ulmusexpansion is usuallyanticipated at c. 8500BP (7470cal. BC) and Quercusshortly after, c. 8000BP (6910cal. BC) (Huntley andBirks, 1983;Birks, 1989;Edwards, 1996). The timing of the Alnus expansionis notoriouslysporadic, but pollen profiles suggestthat
135 this taxonwas very much establishedwithin the Inncr Hebridesby c. 6000
BP (4910cal. BQ (Huntleyand Birks, 1983;Birks, 1989).It is possiblethat the earlierburning of vegetationat AThrannagprevented the expansionof thesetaxa and althoughan earlierintermittent presence of Ulmusis recorded at c. 5% UP in AClb from c. 8000BP (6910cal. BQ, which may rcflcct a local presence,it is alsolikely to result from the long distancetransport of Ulmuspollen, asare the low amounts(< 2% TLP) of Quercusand. 41nus. The late expansionof thesetaxa may be relatedto a fall in sealevel around c. 6000BP (4910cal. BC) (Ballantyneand Dawson, 1997). This may have meantthat the exposedA! Chrannag basin received less salt sprayfrom c. 6000BP (4910cal. BQ which may haveprevented earlier extensive colonizationby the major arborealtaxa. It is notedthat arborealpollen influx at Loch an t'Suidheon Mull, directly to the southof Ulva, is also low until c. 6000BP (4910cal. BC) andthis might alsobe explainedby earlierhigh sea levelsand increased salinity restrictingwoodland species.
Furtherdiscussion of the behaviourof thesetaxa in a wider regionalcontext is presentedin Chapter7.
5.4.3 ACIdBetula-Calltiiia; ACle(i)Betula; ACle(ii)Coryltis-
Calluna; AC2c(i) Alnus - Corylus- Poaceaec. 5300- 4200 BP (4150- 2790BQ
Although greaterdetail of this time period is offeredby AC I dueto better resolution,both profiles showthat a mixed woodlandof Betuld, Ulmus,
Quercusand Coryluswas present at AChrannagfrom c. 5300BP (4150cal. BQ. Pinus may havebeen a minor componentof local woodland,or its pollen may derivefrom furtherafield on Ulva or nearbyMull (refcr to 5.4.2). As with the previouszone, Betula influx is greaterin AC I thanAC2, but
136 Ahius influx is greaterin AC2 which, assuminga reliablechronology, suggestsdifferences in the local distributionof thesetaxa.
At c. 5300BP (4150cal. BQ in ACId, Callunapercentages and absolute valuesincrease. These coincide with rising charcoalinflux, concentrations and Ch:P, andtemporary reductions in percentagesand concentrations of Ulmus,Quercus and Abius, althoughonly Alnus pollen influx valuesarc reduced.There are no largeincreases in Callunaor charcoalrepresentations from c. 5300BP (4150cal. BC) in the comparativezone AC2c(i), but percentages,concentrations and pollen influx valuesof Poaccaeincrease and a greatervariety of herbs,including Potentilla, Ranunculus acris-type, Filipendula,Rubiaccae and Apiaceae undiff., is recorded.Plantago lanceolataappears at nearlyevery level in and fromAC2c(i) albeit in low amountsof <2% TLP; this may be an indicationof grazingby domestic animalsand could signify a humanpresence. Ch: P could be lower in AC2c(i) comparedwith AC Id andAC I c, asthe latter zonesprovide better resolution,but is is moreprobable that the microscopiccharcoal recorded representslocalised burning to the north andeast of A!Chrannag mire.
From c. 5000BP (3860cal. BQ in ACle(i), Calluna vulgaris and Corylus arereduced and charcoal influx, concentrationsand Ch:P valuesfall, but all increaseagain in AC I e(ii) from c. 4800BP (3570cal. BQ. The relationshipsof the charcoaland Calluna vulgaris curvessuggest that fires wereresponsible for increasesin heatherpollen. It is possiblethat humans could haveintentionally burned heathlands to promotethe growth of new shootssuitable for the grazingof livestockand to promoteareas of browse for gamesuch as grouse (Hobbs and Gimingham, 1987). The occurrenceof a
Hordeum-typepollen grain in AC I e(ii) at c. 4400 BP (3120cal. BC) providespotential evidence of cerealcultivation near AThrannag, and the
137 baseof a pit containingthe charredremains of cultivatedbarley and oats was datedto 4990+60BP (3730cal. BC) in Livingstone'sCave. The confirmed presenceof Neolithic peoplereinforces the possibility that fires wereused in
the managementof heathlandson Ulva.
Nevertheless,it is possible that natural fires arising from lightning stri es
could also have set heathlandalight and Pennington et al. (1972) provide
evidencefor increasedrainfall in the British Isles from c. 5200 BP (4010 cal.
BQ. Seasonaltemperatures may also have been lower. If Atlantic storms
becamemore frequent from c. 5200 BP (4010 cal. BQ, then this could
account for an increasein fires arising from lightning strikes. There are
peaks in Sphagnumin ACle(i) and ACle(ii) which coincide with lower
values of Calluna and Ch:P and might indicate increasedwetness, perhaps
due to higher rainfall. If fires were anthropogenicin origin, then increased
wetnessmay have reducedthe flammability of Calluna heathland and
preventedburning betweenc. 5000 BP (3860 BC) and c. 4800 BP (3570 BC).
Thereare other indications of a lessfavourable climate in the A!Chrannag
profiles. The ACI pollen influx profiles displayreductions in all arboreal pollen types(including Corylus)from c. 5200BP (4010BQ, andtherefore climatic effectscould haveaffected woodland from this time. The numberof
treescould havebeen reduced due to increasedwaterlogging and lower temperatures,flowering may alsohave been reduced by this andwith
woodlandspecies becoming more susceptible to disease.If all woodland typeswere affected,then there would be little changein the pollen
percentagedata. Increases in Plantagolanceolata and herbs in AC2c(i) couldbe dueto reductionsin woodlandand more soil exposure.
138 A decreasein Ulmuspollen percentagesand concentrations is recordedat c. 5100 BP ( 4010 cal. BC; 178cm) in AC Id andat c. 4800 BP (3570cal. BC;
132cm) in AC I c(i). A declinein Ulmuspollen in AC2c(i) from 21Ocm (c.
4800BP [3570cal. BQ could correspondto eitherof the declinesnoted in ACle(i) dueto the potential Inaccuraciesof the dates,but a simultaneous
reductionin Pinuspollen percentagesand concentrations is notedin AC2c(i) which suggeststhat it probablycorresponds with the secondarydecline of
AC I e(i). An earlierdecline may not be recordedin AC2 dueto low pollen influx andtaphonomic influences which affectthe distributionof pollen acrossthe bog.
SecondaryUlmus declines c. 4800BP (3570cal. BC) havebeen recorded in
Scotland(Robinson and Dickson, 1988;Edwards and McIntosh, 1988;
Edwardsand Whittington,1997), and Northern Ireland (Edwards and Hirons, 1986).Increases in rainfall associatedwith climatic changeas
indicatedby recordsfor Pinuswould, in somelocations, increase
waterlogging,podzolisation and acidification of soils. Falls in seasonal temperaturescould alsohave reduced Ulmus populations, particularly those
of Ulmusglabra, which hasgreater potential than other speciesof this genus to colonizepoorer soils andgreater extremes of slope(Birks andHuntley,
1983)and is most likely to inhabitmarginal environments. A peakin. 41nus
pollen after the Ulmusdecline in AC I e(i) may be a responseto wetter conditionsat the bog surfaceand a reductionin LOI indicatesthe inwashof
mincrogenicsediments. However, increases in Corylus,Calluna and Ch-P increasein AC I e(ii) andPlantago lanceolata is recordedshortly aflcr the Ulmusdecline in AC I e(i). It is possiblethat this representshuman activity
which may be linked to the elm decline.However, all the evidencefor this occursafter the Ulmusdecline at AThrannag despitedates from the
139 archaeologicalsite confinning that peoplewere present at c. 4800BP (3570 BC). Additional evidenceof Ulmusdeclines on Ulva is presentedfor Livingstone'sCave profiles, Section 5.5.4 and furtherdiscussion of Ulmus declinesin the InnerHebrides is madein Chapter7.
5.4.4 AC2c(fi) and AC2c(iii) Betuld - Calluna - Poaccae- Potentilla
4200 - 2500 BP (2790 - 600 cal. BC) From c. 4200 BP (2790 cal.BC) in AC2c(ii), Betula increasesin both percentageand absolutevalues, but there is relatively little changeto the
Quercus,Alnus, Corylus and Poaceaecurves. In AC2c(ii), Potentilla-type expandsto be consistently presentin values >2% TLP. Charcoal is not recorded at every level and charcoal influx and Ch:P are low, and there is no indication that substantialburning occurred locally which could indicate human activity. However, the sustainedpresence of Plantago lanceolata in
AC2c(ii) may representthe effects of grazing of domestic animals, and suggestsa continuous human presencenear the site. The zone dates to the late Neolithic and Early Bronze Age, when pastoral activity was well established(Simmons and Tooley, 1981; also cf. Edwards and Ralston,
1996). The transition to AC2c(iii) is marked by an anomalousspike in Alnus pollen percentagesand absolutevalues, but the overall pollen assemblage changeslittle from AC2c(ii). Most notable is the presenceof an Avena-
Triticum-type pollen grain in a sample dating to c. 2800 BP (1000 cal.BC) which might indicate that mixed farming was taking place.
5.5 Presentation and discussion of the Livingstones Cave Bog profiles
Percentage,concentration and pollen influx diagrams for the two profiles from Livingstone's Cave bog, LC I and LC2, are presentedin Figures 5.20,
5.21 and 5.24 and 5.27. Summary diagrams of LOI, microscopic charcoal,
TLP concentrations,pollen influx and TD% for each proflle are presentedas
140 figures5.23 - 5.24 anddamaged pollen percentagesas figures5.28 and5.29. Table 5.3 compares(sub)zones between the Livingstone'sCave profiles.
5.5.1 LCI a Cyperaceae- Cichorium intybus-type;Ma Cyperaceac; c. 11,000- 10,000BP (11,050- 9530cal. BQ The pollen assemblagesLCla andLC2a aredominated by Cypcraceaeand Cichoriumintybus-type (probably Taraxacum officinale), and Poaceac and
Parnassiapalustrispollen is alsopresent. The pollen is representativeof an open,damp environment which is likely to haveincorporated other taxa not recordedin the pollen profiles.Much of the pollen in this zonemay be reworked,as suggestedby a potentialreversal in dates(Fig. 5.6, Table 5.3.) althoughthese may be similar at 2 SD. If re-workingof sedimenthad taken placethen higher percentages of damagedpollen might be expected,and TD% doesnot exceed5% for most of the zone.Although high Ch:P might indicatehigh levelsof reworkedcharcoal (cf Edwardsand Berridge, 1994) charcoalcould alsoderive from the fallout of largeamounts of aerial particulatesin a virtually treelessenvironment, and low pollen concentrationsmay augmentCh: P values.Charcoal influx in LC Ia and
LC2a is low, but higherthan in the later Holoceneprofiles. In the absenceof re-workingthen the low levelsof Betula,Pinus and Coryluspollen (< 2% TLP) could indicatethe local paucityof woodlandtaxa, or long-distance transportof pollen during the LateglacialInterstadial.
LOI reaches- 70 % in LC I a, but doesnot exceed- 20 % in LC2a andlow sedimentaccumulation rates are indicated (although these are unreliable as they arebased on extrapolationfrom dateson later Holocenesediments which doesnot allow for an hiatusin sedimentation).
141 5.5.2 LC1b and LC2b (No pollen) c. 1%000- 5500 BP (9530- 4380 cal. BQ
ZonesLClb andLC2b arecharactcrised by a decreasein LOI to <5% and a changein stratigraphyto greensandy clay. The inorganicsediments yielded insufficientpollen to allow a reliablevegetational reconstruction, thus there is no pollen representedin thesezones on the diagrams.Of the little pollen present,most was severelycorroded and crumpled, making identification difficult. However,it wasnoted that Cyperaccacpollen dominatedthe lower levelsof LClb andLC2b. This could possiblyresult from intermixingwith underlyingorganic sediments or the reworkingof older pollenifcrous horizonsfrom elsewhere.The samplesfrom the upperlevels of the zones containedthe pollen of Cyperaceae,Poaceae, Belula, possibly Runiax sp., andwhat may be Caryophyllaceaeundiff. or Thalictnim.Sphagnuin was also presentin apparentlylarge quantities in many samples,suggesting damp conditions.Although not quantified,microscopic charcoal was also notedin manyof the samples,and high levelsare recorded in many Lateglacial pollen profiles (cf Edwards,1978; Whittington et al., 1991 a; Edwardsand Berridge,1994). Iron sulphatespherules suggesting wet conditions,possibly ponding(Wiltshire et al., 1991),were also present in most samples.
LC Ia andLC2b areprobably of Loch LomondStadial date from c. 11,000to
10,000BP (11,050- 9530cal. BQ.
5.5.3 LCIc(i) Corylus- Cyperaceae and LC2c(i) Cotylus-Cyperaceae; c. 5515 - 5350 BP (4350 - 4160 cal. BQ The baseof the peat overlying the Lateglacial mineral sedimentshas radiocarbon datesof 5515±55 BP (4350 cal. BC) and 7765±65 BP (6580 cal.
BQ. Although the later date is preferred, both dates suggesta discontinuity
142 of sedimentationwithin the basinstratigraphy and it is possiblethat the
erosionof earlyHolocene sediments could havetaken place.
Althoughtidal activity dueto marineincursion into the basincould alsohave
removedearly Holocenesediments, this hasalready been dismissed (Section 5.2). It is mostprobable that sedimentationin the basinwas absentfor the earlierHolocene. Although the indicationof pondingand waterlogging by spherulesin the later mineralsediments underlying the Holocenepeat
suggestsextremely wet conditions,subsequent drying of the relatively porous,sandy sediments could haveprevented the expansionof vegetation into the basin,or the preservationof organicdetritus. Although thereis no
clearevidence that an hiatusin sedimentationoccurred in the Livingstone's
Cavebog profiles,it is the mostplausible explanation of the basin
stratigraphy.
SubzonesLClc(i) andLC2c(i) representthe earliestpolleniferous Holocene
sedimentsin the Livingstone'sCave basin. In thesesubzones, both LOI
curvesrise steadily,signifying the accumulationof pcat over earliersandy
clays.Corylus, Salix, Cyperaccaeand Poaceae are recordedin both
subzones,with a variety of herbsincluding Filipendida, Aplacead undiff., Brassicaceae,Potentilla-type and Ranunculus acris-type. The presenceof
Calluna vulgaris, Vaccinium-typeand Emptrum. nigrurn in LClc(i) and
LC2c(i) suggestsa heathytype flora at the bog surface.Abius glutinosa rises
from low valuesat the baseof LClc(i) but doesnot expanduntil the
succeedingsubzone in LC2, andit may be that LC2c(i) is slightly earlier thanLClc(i). Ch:P andcharcoal influx aregreater in LClc(i) than LC2c(i)
but falls tocomparativclylowlevels towards the end of the subzone.The differencesmay be explainedif LC I c(i) datedslightly earlierthan LC2c(i),
143 or if a greaterdegree of mixing with the underlyingLoch LomondStadial depositshad occurredin LC I c(i).
The pollen spectraindicate that c. 5500BP (4380cal. BQ the mire surface vegetationconsisted of areasof sedge-dominatcdgrassland with a tall-herb flora andpockets of Salix scruband heath.Alnus andCorylus may not yet havebeen locally present;low percentagesqnd influx in the early profiles suggestthat thesetaxa were relatively sparse.
Pinussylvestris pollen is alsorecorded in LC I c(i) andLC2c(i). Although its percentagerepresentation reaches between 15 - 20 % TLP, low influx values suggestthat it was probably not presentat Livingstone's Cave bog. Indeed, the sparselywooded environment would have allowed the influx of Pinus pollen transportedfrom further afield. 5.5.3 LC1c(ii) Omits - Quercits -Ahms - Cotylus and LC2c(ii) Uhnus Quercus Alnus Corylus; - - - c. 5350 - 4900 BP (4160 - 3730 cal. BC) At the transition to LC I c(ii) and LC2c(ii), LOI values increaseto - 90 % and this value is maintained for the duration of the subzone.Corylits and
Alnus are the predominant taxa in percentageterms, although their pollen influx values remain low, averaging 20,000 grains cm-1 yr-1, compared to those in A'Chrannag basin averaging 50,000 grains cm-1 yr-I for the same period. This suggeststhat woodland in Livingstone's Cave bog was sparse.
Nevertheless,Alnus macrofossils in the peat stratigraphy show that this taxon was presentat the mire surface. Although a slightly earlier presenceof
Ulmus was recorded in subzoneLCIc(i), the main expansion of Uhnus and
Quercus occurs in LClc(ii) and LC2c(ii). At an estimated date of 5200 BP
(4010 cal. BC) this is slightly later than recorded in the A'Chrannag profiles c. 5600 BP (4500 cal. BC) but the dates for both basins overlap at 2 SD.
Betula persists in LCIc(ii) and LC2c(ii) in similar percentagesto those recorded in the previous subzone.
144 Despitethe rangeof percentagevalues recorded for Belula, Ultnus,Quercus and Corylus,pollen influx valuesare similar for thesetaxa (within each profile andbetween the LC profiles).This may indicatethat their pollen derivesfrom a wide area,possibly further afield thanUlva, or that nonewas predominantover the otherswithin the local woodlandflora. However, Alnus influx valuesare higher than for the otherarboreal taxa, and this taxon was undoubtablyof greaterlocal abundanceat andprobably around the mire surfaceas is indicatedby Alnusmacrofossils in the peat.
The range of herbs recorded in the earlier subzonesis reduced in LC I c(ii) and LC2c(ii) and both Salix and Calluna decline to values of <2% TLP. although increasesin ferns are recorded.The expansionof Alnus onto the
Livingstone's Cave mire may representincreasingly wet conditions which may not have suited Calluna heath. Salix may have persisted locally as an understoreyshrub, but its pollen was not widely distributed, therefore it is not present in the pollen record and a damp understoreyof sedgesand ferns would have been present.Betula may also have intermixed with Alnus and
Corylus at the mire surfacebut Ulmus and Quercus were probably more abundanton the better drained soils at the basin edges.
Microscopic charcoal is virtually absentin LC I c(ii) and recordedonly in very low values in LC2c(ii). Most of the charcoal is probably background charcoal rain and higher representationsin LC2c(ii) may derive from the fallout of airborne particles reaching the easternslopes at the mire edge.
In the latterpart of LC1c(ii) (from 290 cm) and from early in LC2c(ii) (from in 280 cm), the. 41nus,Pinus, and Corylus curves display reductions percentage,concentration and influx values and a reduction in Ulmus occurs
145 in LCIc(ii). Although an Ulmus decline appearsin LC2c(ii) this is the result of being unable to achievereliable pollen counts between 241 and 243 cm on the is The in this core , so part of pollen record missing. actual reductions arboreal pollen in LC I c(ii) and LC2c(ii) coincide with increasesin
Cyperaceaeand ferris, and in LC2c(ii) Poaceaevalues also increase.Ch: P remains low in both LC I c(ii) and LC2c(ii), although there are two independentspikes in Ch:P within LC2c(ii). The LOI percentagesof both profiles appearconstant. The reductions in arboreal pollen in both profiles occur c. 5 100 BP (4010 cal. BC) if the estimatedchronology is correct which is also the date for the classic British Ulnius decline. which has been attributed to climatic deterioration (cf. Whittington et al., 1991b; Walker and
Lowe, 1985 and Chapter 4, Loch an t'Suidhe; Hirons and Edwards, 1986). If the distribution of woodland speciesassumed above is correct, it appearsthat the local vegetation of the mire surfacewas affected which implies that changesin mire conditions probably causedthe reductions in the taxon (i. profiles. Whether theseresulted from the effects of a changing climate e. increasedrainfall) or other local factors (such as the changein drainage patterns or bog growth) cannot be determined.
Thereare alsoreductions in Alnus and Corylusin the A'Chrannagproriles AC I andAC2 at a similar time to thoseabove (estimated c. 5200BP [4010 cal. BC]), andin ACId, theseappear to be linked to increasesin burning and the propagationof Callunaheath. If this was the result of humanactivity, thenthe arborealpollen reductionsin LClc(ii) andLC2c(ii) could alsoresult from this. However,large amounts of Alnus macrofossilsin the peatin both coresindicate low levelsof humificationwhich could be linked to waterlogging.Within LClc(ii) andLC2c(ii) pollen influx valuesand in concentrationsare very low and,at somesampled levels, this resulted in preparationswith insufficientpollen to producereliable counts. LC I c(ii)
146 particularcontains high amountsof pitted pollen which hasbeen linked to wet-dry cycles(Wilkinson andHuntley, 1987;Campbell and Campbell, 1994).TD% for the completeLC I profile is higherand more sporadicthan for LC2, suggestingthat the centralmire areaexperienced greater extremes of wet-dry cyclesthan the easternmire edge.Thus the reductionin arboreal pollen in LC I c(ii) andLC I c(ii) may be the resultof a period of increased wetnessin the mire and/orincreased wct-dry cyclesat the mire surface.The spikesin Ch:P in LC2c(ii) could derivefrom the inwashof charcoal fragmentsof possiblesecondary origin. Increasedrainfall due to climate changemay havecontributed to the possiblehydrological changes in Livingstone'sCave bog, but thesecould alsobe a result of the naturalcycles of mire growth affectedby its morphologyand environmental setting.
5.5.5 LCU(M)Ahjus- Corylus- Cyperaceae;LC2d(i) and (H)Alnus- Abius Corylus Poaceae Corylus- Poaceae;LC2e(i) Betuld - - - c. 4900- 4000BP (3730- 2530cal. BQ This period is coveredby LClc(iii) andin greaterresolution by LC2d and LC2c(i). Samplingin LClc(iii) was interruptedby severalAhms macrofossilsbetween 225cm and 238 cin in the core,and there is no pollen recordfor this sectionof the profile.
LC I c(iii) andLC2d(i) incorporatea similar pollen assemblageto that of the previous(sub)zones, where, 41nus, Corylus, Poaceae, Cyperaceae and fems arethe main componentsand Alnus and Corylusvalues recover from the reductionsnoted earlier. Although a contractionin, 41nusis suggestedin LC2d(i), this arisesfrom an anomalousspike in Alnus percentagesand absolutevalues at the baseof this subzone,and when comparedto the previouszone, LC2c(ii), no real reductionis indicated.Calluna is virtually
147 absentfrom MOO andLC2d(i), andthis may signify that the mire was particularlywaterlogged (hence the predominanceof Ahms in the percentage profiles).A greatervariety of herbsis recordedcompared to the earlier subzones,and includes Apiaceae undiff., Potentilla andFifipcndula. These indicatethat woodlandat the mire surfacewas not particularlydense, and that clearingsof darnpgrassland and tall-herbs were present. Ch: P values increasewithin LClc(iii) andLC2d(ii) (evenwith increasesin TLP concentrations)although charcoal influx remainslow. Suchincreases in charcoalcould alsoresult from sparsewoodland cover, and may signify increasedopenings in local woodland.Low charcoalinflux suggeststhat extensiveburning was unlikely, andincreases in Ch:P coukucrivefrom increasedaerial fallout if screeningwoodland was reduced.A human presenceon Ulva from c. 4990BP (3730cal. BQ is confirmedby dateson artefactsfrom Livingstone'sCave and at this time peoplemay have intentionallyremoved areas of woodlandfor fuel, or to provide areasfor cultivation.Grazing by domesticanimals may alsohave opened up woodlandin the Livingstone'sCave basin. Nevertheless, the pollen may also representnatural openings in woodlandon the bog surface,and woodland doesnot appearto havebeen particularly dense according to pollen influx data.
The main featureof LCIc(iii) is a seconddecline in Uhnuspollen at 270 cm c. 4800BP (3570cal. BC) which correspondswith the timing of the UInjus declinesin the A!Chrannag profiles. This is accompaniedby a marked reductionin LOI over a singlelevel from 95 % to 15 % andincreases in Pteridiumaquilinum, Ch: P andcharcoal influx. As LOI valuesrecover c. 4650BP (3450cal. BC), Ulmuspercentages resume those previously recorded.Ulmus declines occur at 240 cm,c. 4600 BP (3440cal. BQ in LC2d(i) and at 220 cm c. 4200BP (2790cal. BC) in LC2d(ii) following a
148 largeincrease in Ulniuspercentages and absolute values at 235 cm c. 4400
BP (3120cal. BC). Giventhe unreliabledating of the Livingstone'sCave profiles,it is possiblethat eitherof the elm declinescorrelate with that on LCIc(iii) at 270 cm c. 4800BP (3570cal. BC).
The causesof the Ulmus declines in the Livingstone's Cave profiles from c.
4800 BP (3570 cal.BC) may have an underlying climatic causeas suggested for the A'Chrannag profiles, but diseasemay also have been a factor in temporarily reducing Uhnus populations on Ulva and variability in the distribution of pathogensmay explain the potential spatial and temporal distributions of the Ulmus declines in the profiles.
Considerationmust alsobe given to the potentialfor humaninfluence on Ulmuspopulations from c. 4800BP (3570cal. BC). The increasesin Ch:P,
Pleridium aquilinumand in herbs,particulary Potentilla (possiblyA crecta) in LC I c(iii) andLC2d(ii) could suggestthat burning occurredlocally, althoughthis appearsunlikely given the low charcoalinflux in both profiles. Nevertheless,Livingstone's Cave was inhabitedin the Neolithic from c.
4990BP (3730cal. BC) (Bonsallet al., 1995)and people may have expandedopen areas of woodlandfor cultivationor grazing.A possible Hordeum-typepollen grain in combinationwith increasesin Plantago lanceolatain LC2d andanAvena-Triticunt-type grain in LC I c(iii) reinforce the evidencefor a humanpresence, and from c. 4400 BP (3120cal. BC) thereis evidencein A'Chrannagbog (AC I e(ii)) that fire may havebeen employedin the managementof Calluna heath.It is thereforepossible that peoplecould alsohave influenced local Ulmuspopulations. The gatheringof leavesand the strippingof bark andbranches alone could havereduced Ulmusbut this would haveexacerbated effects arising from any climatic influencesand rendered trees prone to disease.if Uinjuswas an important
149 resourceto Neolithic people, then its decline could also have been prompted by some form of management,perhaps coppicing (cf Hirons and Edwards, 1986), which could increaseflowering and could therefore be the causeof the temporary increasein Ulmus c. 4400 BP (3120 cal. BQ at 235 cm in
LC2d(ii).
Thereis no clearevidence, however, for humaninvolvement in the Ulnuis declinesat Livingstone'sCave bog, andclimatic andedaphic influences would be expectedto influencethe behaviourof Ulmusin the Ulva profiles. The complexityof interpretationsurrounding the Ulmusdeclines is extremelyevident in the aboveand furtherdiscussion of the Ulmusdeclines on Ulva is providedin Chapter7, in the contextof the wider Inner Hebridean region.
Following the Ulmusdeclines and the interruptionof LC I c(iii) by Alnus macrofossils,after c. 4100BP (2660cal. BQ Alnus, Befuld, Ulmus, Quercusand Corylusand fernspersist in similar percentagesand absolute valuesto thoserecorded earlier. However, in the later part of LClc(iii) Filipendula andPotentilla-type are recorded in low values,and Silene vulgaris-typeis alsopresent reaching maximimum values of c. 5% TLP. A grain of Avena-Triticum-typein the latter part of LClc(iii) c. 4100 BP (2660 cal. BQ andcould reflect cultivationin the catchmentarea. Other
anthropogenicindicators are rare, however, in the pollen profiles andgrasses andherbs remain low in percentageand absolute terms. This suggeststhat cultivationmay havetaken place some distance from the pollen site, or the Avena-Triticumtype grain representsa wild speciesof grass,possible wild
oat.
150 At the transitionfrom LC2d(ii) to LC2c(i) c. 4100BP (2660cal. BC), a singleHordeuni-type grain coincides with increasesin Plailtago lanceolata, Pleridium aqtdhiumand Ch: P. Plantagolanceolata and Pleridium pcrsist within the LC2e(i) profile wherean expansionin Belula percentagesis recordedand a furtherHordeum-type grain occurs later in the zone.The indicationfrom LC2e(i) is of increasinghuman activity in the vicinity of Livingstone'sCave basin. Although the cereal-typepollen grainscould derivefrom wild grasses,carbonized macrofossils of Hordeunisp. and Avenasp. havebeen recovered from depositswithin Livingstone'sCave which may be contemporarywith the pollen profiles in thesesubzoncs, suggestingthat cultivationon Ulva occurredin the Neolithic. The increases in herbpollen, particularly Plantago lanceolata in LC2e(i), indicateopen areaswithin the woodland,and vegetation succession within thesecould be indicatedby increasesin Betula,a frequentsecondary pioneer taxon. The greateramount of evidencefor humanimpacts appears in LC2c(i). This is probablydue to the locationof core LC2 at the bog edge,close to the known site of humanoccupation. LC2 is alsocloser to betterdrained areas above the mire which may havebeen cultivated and is also locatedat the foot of slopesto the eastwhich may havebeen grazed by domesticanimals.
5.5.6 LCldAlitus-Corylits; LCleBetula-Poaceae-Poteittilla;
LC2e(ii) Belitla - Cyperaceae- Filipenduld
4000- 3200BP (2530- 1480cal. BQ The periodof 4000- 3600BP (2530- 1950cal. BC) is representedby LCId,
LC 1eand LC2c(ii). The earlierpart of LC2f coversthe laterperiod 3600- 3200BP (2530- 1480cal. BQ. Betweenc. 4000 and3600 BP (2530- 1950 cal. BQ, LCId displaysmarked increases in Alnus andCorylus pollen percentagesand influx values,whilst Pinus almostdisappears from the profile. This is representativeof the regionalPinus declineevident in other
151 1 pollen profiles from westernScotland (cf Bennett,1984; Hirons and Edwards,1990; Willis et al., 1998)linked to an increasinglywet climate.
The increasesin Alnus could thereforebe a responseto increasedwetness at the mire surfaceand the higher Corylusvalues may actually reflect an expansionin Myrica gale. Although Edwards(198 1) suggeststhat Myrica is not a significantcomponent of Corylusavellana-type pollen in Scotland until c. 2000BP (100 cal. BC),, the datadrelimitedto a few sitesin eastern Scotland.The warmeroceanic climate of the Inner Hebridesmay have favouredan earlierexpansion of Myrica, which thrives in wet conditions, andthe behaviourof manytaxa on Ulva doesnot conform to established precedents(e. g. the expansionof Quercusand Ulmus).Myrica twigs and leaveswere alsorecovered from peatfrom MachrieMoor, Arran in significantamounts dating from c. 3500BP (1820cal. BQ (Robinsonand Dickson, 1988).
Otherarboreal taxa of Betula, Ulmusand Quercusare still presentin LC1d, andPoaceae and Cyperaceae are represented in high values.Calluna vulgaris andLonicerapericlymenuni also increasein LC I d, although percentageand absolute values are relatively slight, andthere is a peakin Ch:P althoughcharcoal influx is very low. If this representsa fire event,then it may be linked to the increasein Callunawithin this zone.The presenceof Lonicera suggeststhat an openingmay haveoccurred in the local woodland canopy,but this may be a naturaloccurrence and is not necessarilythe result of humanactivity. Nevertheless,human actions could be the sourceof burning and increasein Calluna representedin LC I d.
From c. 3800BP (2220cal. BQ in LCle thereare markedreductions in percentagesand absolute values of Alnus, Corylus, Ulinus-andQuercus which coincide with increasesin Betula and Poaceae.Lonicera is high at the
152 baseof this zone,and Callunais alsorecorded in valuessimilar to tile previouszone. There is a greaterabundance and range of herbswhich includeFilipendula, Potentilla-type, Plantago lanccolata, Runlex spp.,
Ranunculusacris-type andApiaccacundiff. Noticeable peaks in Ch:P also occurand a small increasein charcoalinflux. In LC2e(ii), Alnus and Coryllis valuesdecrease, and Ulmusis presentin very low amounts.A similar range of herbsto thosein LC 1eis recorded,and in the latter part of LC2c(ii) charcoalinflux andCh: P increase.
Pollcn assemblagesin LCle and LC2c(ii) intimate that from c. 3800 BP
(2220 cal. BQ an intensification of human activity resulted in the reduction of much of the woodland in the pollen catchmentarea, creating areasof
Betula scrub and herb rich grassland.Wood may have been collected for domestic purposes,and for grazing animals, and grazing itself would have assistedin maintaining open areas.
(1480 670 5.5.7 LC2f Alnus - Corylus - Poaceae; 3200 - 2600 BP - cal. BQ
(1480 670 BC), Corylus From c. 3200 - 2600 BP - cal. zone LCUAlnus and recover from previous values, and Poaceaeincreases, but fewer herb pollen taxa are evident. Nevertheless,Ch: P remains high and charcoal influx
increasesto a profile maximum. Two incidencesof Hordcum-type pollen are
recorded.Despite a recovery in local woodland, human activity may have
persistednear the site, and the mire edge location of LC2 may have enabled
this to be detected.However, this pollen assemblagemay also indicate an
increasein wetnessin Livingstone's Cave bog and the Corylus recorded in
this zone may representMyrica gale at the mire surface as discussedabove.
The Hordcuin-type pollen grains recorded in LC2f may therefore be large
grasspollen such as Glyccriafluitans which prefers a wet environment, and
153 Poaccaepollen is abundantin this zone.Ahms is alsotolerant of wct conditions,and Ch:P valuesmay havebeen maintained by inwashdown the slopesto the castof the coresite. However,Ch: P valuesarc comparableto thosein the earlierzone which suggeststhat humanoccupation did not entirely disappearfrom Ulva after c. 3200BP (1480cal. BC). Indeed,a shift in the locationof humanactivity could havecleared vegetation elsewhere andconsequently increased runoff into the Livingstone'sCave basin at LC2. An amalgamationof naturaland human factors probably influenced the local vegetationbetween c. 3200BP (1480cal. BC) and2600 BP (670 cal. BC).
5.6 Contrasts between bogs
Although the pollen profiles from AChrannag and Livingstone's Cave bogs extend to covcr different periods of time, all the profiles cover the period c. Within 6000 - 4200 BP (4910 - 2790 cal. BQ. this period, the profiles indicate variations in local vegetation within and betwcen basins.
5.6.1 General observations
A seriesof mapsindicating the spatialvariations in inferredvegetation patternson Ulva,between c. 8500and 4000 BP (7470and 2530 cal. BQ are presentedas Figures 5.30 - 5.33.
In A'Chrannagbasin from c. 8500- 6000BP (7470- 4910cal. BC) (Fig. 5.30), Corylusis the dominantarboreal taxon. Calhina vvlgaris seemsto havebeen prevalent to the northwestof the bog, and charcoalrecords in ACI suggestfircs, eithernatural or anthropogenic,influenced episodic Calluna growth,and may alsohave favoured Corylus and preventedthe expansionof otherarboreal taxa. Birch scrubwas alsoprevalent to the northwestof A!Chrannag basin, and may havebeen interspersed with Callunaheath.
154 Shortly after c. 6000BP (4910cal. BQ Quercus,Uhnus and Ants cxpand but percentageand pollen influx curvesindicate that Atus populationswere greaterto the southof the bog, perhapsbecause the environmenthere was wetteror lessexposed. There is relatively little Alnus recordedin AC I, whereBetula dominatesthe pollen profiles,perhaps due to tile degreeof exposureto south-westerlywinds experiencedby the north of AChrannag bog, Betuldbeing able to withstandthis betterthan other arborealtaxa.
From c. SSOOBP (4380 cal. BQ in the Livingstone's Cave basin, Salix and hcrb-rich grasslandwere replacedby Alnus-Corylus woodland. Alnus may have been the dominant taxon in the basin according to pollen percentages, but Alnus influx is very low comparedwith AChrannag, and high values of
Poaceae,Cypcraceae and herbs suggestthat the basin was wetter and woodland more open than A'Chrannag.
TLP andcharcoal influx in the A'Chrannagbasin are much higher than at the Livingstone'sCave sites, and damaged pollen percentages are greater in the lower bog profiles.This may be an indicationof the differencesin bog growth andhydrological characteristics. Peat formation in the A'Chrannag basinprogressed earlier than in the Livingstone'sCave basin and the former mire probablydeveloped raised bog characteristics.In contrast,the younger Livingstone'sCave basin peats may havebeen rheotrophic, or at the transitionto mesotrophicstatus (Moore et al., 1991),and more waterlogged thanthose in the A!Chrannag basin, deriving more nutrientsfrom runoff and
streams.The differentbog statuseswould haveinfluenced vegetation patterns,but differentialpeat accumulation and hydrological characteristics could haveimplications for the deteriorationof pollen grains.Wilkinson and Huntley (1987)find that pollen preservationmay be considerablyworse in
155 wetterareas of bog whereTLP concentrationsarc also low, andthat better preservationand higher TLP concentrationsoften occur in drier, hummocky areasdominated by Callunaheath. This would seemto be borneout by the evidencefrom the Ulva basins.
Charcoal influx is much lower in Livingstone Cave profiles than in the
A'Chrannag profiles. This is a possible reflection of the distribution of fires, and the high amountsof Calluna in the AC I profile may result from burning. A! Chrannagbog may also have been drier and its vegetation more easily combustible than that in the Livingstone's Cave basin.
5.6.2 The expansionof arboreal taxa
The major expansionof arborealtaxa (Quercus,Ulnius andAlnus) in the
A'Chrannagprofiles takes place c. 5600BP (4500cal. BC). A similar expansionoccurs in the Livingstone'sCave profiles and,if the datesassumed for the baseof the Holocenesediments in LCI andLC2 arecorrect, this also takesplace c. 5600BP (4500cal. BC). This expansionis late comparedwith thoseat otherInner Hebrideansites (e. g. Loch a!Bhogaidh, Islay; Loch an t'Suidhe,Mull; Gribun,Mull [Walker andLowe, 1987])and pollen isochronerecords for westernScotland (Huntley andBirks, 1983).Although small islandecologies may differ from thosein mainlandareas and the larger islands,it is unlikely that this could explainthe late expansionof arboreal taxaon Ulva, andgiven the earlierexpansion of Quercusand Ulmusc. 7800 BP (6640cal. BQ at remotesites in neighbouringMull (Tornessand Coire Claccachin Glen More, Walker andLowe, 1985)it is unlikely that the high
cliffs and fast flowing channelsof water which separateUlva from Mull could havesignificantly delayed the transportof propagules.
156 As the Holocenebasal dates for the Livingstone'sCave bog prorilcs may be incorrect,Quercus, Ulmus and Alnus could haveexpanded earlier than 5600
BP (4500cal. BC) at the site (possiblyc. 7700BP [6600 cal. BC]) although severewetness in the mire floor could havedelayed and restrictedtheir expansionin comparisonto otherInner Hebrideansites.
If an earlierwoodland expansion occured in the Livingstone'sCave catchment,then it is not easyto explaina muchlater woodlandexpansion at the A!Chrannag catchment. It is possiblethat the dateof 5685±60BP (4530 cal. BC) in the A'Chrannagprofiles is alsoincorrect, and that woodland expandedearlier than this in the A'Chrannagcatchment. However, it is possiblethat the A!Chrannag basin was stronglyfavourable to birch-hazel woodlandand Callunaheath, and that frequentfires, whetherof naturalor anthropogenicorigin, assistedin maintainingthis form of vegetationat the expenseof otherwoodland taxa.
The problematicdates on the Ulva profiles makeinterpretation of woodland expansionin the areasof the A!Chrannag and Livingstone's Cave basins difficult, andthere is no singlereason which readily explainsthe possible delayin the arrival of Quercus,Ulmus and Alnus in the two basins.An amalgamationof many factorsis probable,and underlies the complexnature of the ecologyof Holocenearboreal colonisation.
5.6.3 The Mims proriles
The Ulmus percentageprofiles from both the Ulva basins are presentedwith
cereal-pollen and charcoal profiles in Figure 5.34a. Ulm= influx curves are presentedin Figure 5.34b. Ulmus influx is slightly higher in the A! Chrannag
than in the Livingstone's Cave profiles, and the more exposedA! Chrannag
site may have received greater amounts of airborne pollen transported for
157 long distances.It is also possible that the drier environment of the
AThrannag basin comparedto the Livingstone's Cave basin was more favourableto Ulmus.
In the A!Chrannag profiles an Ulmusdecline occurs in AC I c. 5 100BP
(4010cal. BC) anda later dccliný occursin ACI andAC2 c. 480013P (3ý70 BC) wherethe taxonis reducedto virtual absenceappears in eachprofile. Multiple Ulmusdeclines appear in the Livingstone'sCave profiles as early as
5100BP (4010cal. BC) andas late as4200 BP (2790cal. BC) althoughthe taxonpersists until c. 3800BP (2220cal. BC) in LCL Various interpretationswere suggested for thesedifferences, including the cffccts of climateand disease combined with ecologicaland cdaphic factors. The shelterednature of the Livingstone'sCave basin may haveprovided better conditionsfor Ulmuscompared to the exposedA! Chrannag basin so that edaphicconditions could have moderated the effectsof climatic deterioration or pathogenicattack. Indeed, Ulmus populations around the sheltered Livingstone'sCave basinmay havebeen less susceptible to diseaseif their environmentwas more favourable.Grace (1987) states that southfacing and shelteredslopes may be 30Cwan-ner in summerthan north facing slopes, which may be equiv.-itent to a latitudinal differenceof a hundredkilomctrcs of more.Hence the effectsof conditionson Ulniusin the different Livingstone'sCave and AThrannag basins could havebeen signiflcant.
Humaninfluence on Ulmuspopulations was also suggestedbut the indicationsof humanactivity in all the Ulva profiles occurshortly after the Ulmusdeclines. If Ulmus in the A!Chrannag catchment had sufferedthe
impactof disease,then subsequent clearance of this taxonby Neolithic
peoplemay havecontributed to its continuedabsence in the later profiles. The suiteof datafrom the Ulva profiles indicatesthat Neolithic peoplemay
158 have burnedvegetation to createor maintainareas of licathland,and cleared woodlandfor grazinganimals and for cultivation from as carly asc. 5200 BP (4010cal. BQ.
Scanning for early cereal-typepollen in contiguous I cm samplesmeans that the earliest cereal-typepollen in the Ulva profiles is probably detected.
Ccreal-type pollen occurs shortly afler the Ulmus decline at ACI, and, although Ch:P is also high, earlier peaks in Ch:P occur where Ulnuis was constantly present, and the charcoal probably relates to the burning of heathlands.At AC2 there is no cereal-typepollen connectedto the Uhnus decline when Ch:P falls. In the initial Ulmus decline c. 4800 BP (3570 cal.
BQ in LCI there is a small increasein Ch:P but this is not substantial and may result from the fallout of fires to the north of AThrannag.
Whilst the pollen andcharcoal evidence suggests that peoplecould have modified vegetationcommunities on Ulva from the Mesolithic, andthat such activity may haveintensified into the Neolithic, thereis no evidenceto directly link humanactions with the demiseof Ulmusin the A'Chrannagand Livingstone'sCave basins up to c. 4800BP (3570BC). A later Ulnuts declinewhich coincideswith declinesin otherarboreal taxa recorded in LC I c. 3800BP (2220cal. BC) may be linked to humanactions as other arboreal taxa,decrease at the sametime whilst herbsand Ch: P increase.Nevertheless,
the impactsof climatechange and disease in the contextof site locationand
its characteristicswere probably the most influential factorsin the Ultnus declineon Ulva.
5.6.4 Human impact (Referto Figs. 5.30- 5.33) The Ulva pollen profiles show that very different vegetation communities in existed around the A'Chrannag and Livingstone's Cave basins the early
159 Holoceneand the effectsof humanactivity arc also spatiallydifferent.
Between8500 and 7000 BP (7470- 6010cal. BC) thereis evidencethat to the northwestof the A!Chrannag basin, burning of Calluna hcathland occurred.In the southof the basin,from c. 6200- 5700BP (5090- 4350cal. BC), charcoalin AC2 may be the fallout from fires in the northwestor could representlocal buming.The pollen indicatesthat openingsoccurred in local Corylus-Betulawoodland, and possibly resulting from the fc1lingof trees throughburning. If thesedifferent pollen assemblagesare the result of humanimpacts, they may representdifferent land usage,or they may be the resultof similar activitieson different areasof vegetation.
Later in the profiles from c. 5300BP (4150cal. BC), ACI shows'further evidenceof fire which may be linked to the expansionof Callinia heathland. This againcoincides with increasesin herbsin AC2. The first possible evidenceof cerealcultivation occurs c. 4600BP (3440cal. BC) in AC I and coincideswith a reductionin charcoaland Calluna,which may indicatea temporarychange in Neolithic land use,from heathlandmanagement to crop dates production.Cereal-type pollen is detectedin the otherprofiles at later (Section5.6.3) from c. 4200BP (2790cal. BC) andcoincides with reductionsin arborealpollen, expansionsin herbs(including Plantago lanceolata)and increased charcoal influx. Charcoalrecords suggest that burningwas much greaterin the vicinity of ACI andMesolithic and later Neolithic communitiesmay haveset fires to refreshareas of heathland.To the southof A!Chrannag (AC2) andto the west of the Livingstone'sCave basin(LC2), temporaryreductions in treesmay indicatethat people occasionallycleared small areasfor cultivation,or to usewood for domestic purposes.The grazingof domesticanimals may alsohave caused openings in the woodland.
160 5.6.5 Summary
The Ulva pollen profiles showthat a rangeof vegetationalcommunities may 2 haveoccurred over a relatively small areaof c. 2km in southeastUlva. If the dateson the Ulva profiles arecorrect, the timings of the expansionsof the major arborealtaxa do not correspondwith existingpollen isochroncand isopollcnpatterns (Huntley andBirks, 1983;Dirks, 1989)and indicate that small islandecologies may be very differentto thoseon largerislands and the Scottishmainland. The Ulva profiles alsoinclude indications of human impactsfrom asearly asthe Mesolithicperiod, but evidenceof human disturbanceincreases into the Neolithic, and frequentoccurrences of ccrcal- type pollen suggestthat arablecultivation was undertaken.
161 CHAPTER 6- RUM: ANALYSIS OF POLLEN FROM KINLOCII.
6.1 Site description
The FarmFields pollen site (Figure6.1; Plate6.1), NGRNM 399 399, islocated 500 m westof the FarmFields area of excavation,close to the mouth of tile Kinloch River at Loch Scresort.At 10 m O.D. the site is comprisedof soligcnous peatdeposits fon-ncd over a raisedbeach platform of sandstonesloping gently north to south.Transects undertaken by Hirons andEdwards (1990) and for this researchshowed that the peatvaried from 0.5 m to 1.8m deep.To the north and eastof the site areareas of plantedwoodland made up of Alnusglutinosa, Corylus avellana,Quercus spp., Pinus sylvestris and Sainbucus nigra. SaUxspp. saplings arepresent on the peatitself but theseare small andgrowth is checkedby grazing animals.Trampled areas around the field boundarydisplay vegetation characteristicof disturbedground such as Plantago lanccolata, P. coronopusand Run= acelosella.The mire surfaceincludes a rangeof Poaceaeand Cypcraccac, plus Equisetunispp., Rubiaceae, Filipendula uhnaria,Potentilla erccta, Ranunculusacris anda variety of orchids.Sphagnian spp. are common in tile wetterareas.
6.2 Background to the presentstudy
Initial surveysof the areaby Hirons andEdwards (1990) located the deepestparts of the peat(Fig. 6.2) and subsequentpollen analysesfrom two monoliths,K and KII (Fig 6.1), showedthat K was suitablefor intensivestudy; the pollen in KII was severelydegraded and much of it wasunidentifiable. Subsequent depth probing in the locationof K showedthat peatdcpths varied between 0.40 m and 1.10m on an cast-westtransect, whereas on the north-southtranscct the maximum peatdepth found was 0.60m, andthe peatwas occasionallyinterrupted by clusters
162 of stones.The east-westline was therefore determinedthe most suitable for sampling (Fig. 6.3). Sampling positions to the west of existing fields and to the south of former cultivation ridges in the surrounding slopes and to the cast and west of site K were thought likely to provide a'thrcc-dimcnsional'asscssmcnt of vegetation change.It was intcnded to obtain severalmonoliths at rcgular intervals along the transect for comparisonwith K. Monolith extraction was limited however, as subsurfacewater rapidly flooded some of the pits, and the positions of theseare shown in Figure 6.3. The position of the original profile K (Hirons and
Edwards,1990) was cithcr not located or the surfacehad sunk as a result of drainageor the settling of the large inspection pit excavatedin Hirons and
Edwards (1985), as depth measurementsover much of the site did not record peat deeperthan 1.10 m
(1990), KLI KU Furtherto the work of Hirons andEdwards threemonoliths, , andKL4, (seeSection 2.3.1) were extractedfrom the FarmFields sitc (Figs. 6.1, 6.2). The locationsof KLI (NGR 401999), KU (NGR 399 999)sKU (NGR
400 999),Hirons and Edwards's(1990) K (NGR 401999) andthe excavatedarea 6.4 6.7) of humanoccupation are shownin Figure 6.1. Surfaces=plcs (Figs. - wereobtained from eachnew profile location.
Percentagepollen diagramsof selectedtaxa for K, KLI, KU andKU including peatlithologies are shown in Figures6.8 - 6.11 andconcentration diagrams are presentedin figures6.13 - 6.15.Summary diagrams of LOI, microscopiccharcoal, TLP concentrationsand TD % pollen for KLI, KU andKL4 arc providcdin figures6.12 and6.16 - 6.18 anddamaged pollen percentagesfor KLI, KU and K KL4 are shownas Figures 6.19 - 6.21.Radiocarbon dates for the original profile and for the new profiles are shownin Table6.1.
163 6.3 Reconstructionof vegetationhistory at Kinloch using profile K (Illrons and Edwards, 1990) The pollen datafrom monolith K at Kinloch (Hirons and Edwards,1990) provided a generalvegetation history for the site andstrong evidence of local anthropogenic impacts.The pollen spectraindicate that treecover was not extensivein the early
Holocenebetween c. 7800and c. 6500BP (6640- 5460cal. BQ; Coryluswas probablythe main arborealtaxon. Hirons andEdwards (1990) suggestedthat the establishmentof hazelscrub at the site from c. 7800BP (6640cal. BQ is reflectedin zoneKIa, in which the pollen of Cyperaceaeand spores including Equisetunt,Osmunda, Sphagnum, and ferns (Pteropsida (monoicte) indet. ) rcflcct a dampground flora. Tall-hcrb communitiesare also represented by the pollen of Filipendula,Rumax acelosa and Apiaceae.
Alnusglutinosa was establishedaround 6500 BP (5460cal. BC) whencea reductionin the pollen and sporesindicative of dampenvironments suggests progressivedrying. During the main phaseof Alnits-dominatcdwoodland there are two major reductionsin arborealpollen derinedby subzonesKllb and Mid. Simultaneousincreases in microscopiccharcoal may representphases of burning instigatedby humanactivity, and given the known proximity of Mesolithic pcopic to the pollen site,this explanationhas at leastcqual validity to one of natural causes(cf. Hirons andEdwards, 1990).
The predominanceof Alnusglutinosa in the pollen assemblageis reducedc. 3950 BP (2440cal. BQ andis accompaniedby increasesin Poaccae,Cypcraccae and herbs,notably Potentilla (zoneKIIIa). A singleccreal-typc pollen grain was recordedat the baseof zoneKIIIa, andPlantago lanceolata andA majorlmedia
164 are present. Charcoal concentrationsalso increasein this zone, either as a result of increasedlocal fire frequenciesor the reduction of the screening effect of woodland to the fallout of airborne charcoal (Hirons and Edwards, 1990), but there is strong evidencefor Neolithic impacts. However, climatic dcterioration c.
4000 BP (2530 cal. BQ could also have instigated reductions in arboreal taxa, particularly those existing in marginal environmentssuch as are found on Rum.
Nevertheless,the strength of the cultural indicators in the pollen profilc K indicates that human impacts could at least have exacerbatedthe ecological cffccts of climate change.
Throughoutzone KIII the dominanttaxa arePoaceae and Cypcraccac,and the presenceof cereal-typepollen associatedwith the pollen of ruderaltaxa from c. 1500BP to the presentimplies a periodof continouscultivation. Reductionsin
LOI during KIIId arelikely to reflect increasedsoil erosionassociated with agriculturalactivity andrelict cultivationridges are visible in the present landscape(Hirons andEdwards, 1990; Wickham-Jones, 1990a).
6.4 Comparison of KL1, KL3 and KU with K
It was mentionedin Section2.2 that the monolith profiles KLI, KL3 and KU werenot sufficiently deepto providea recordof Mesolithic agevegetation for comparisonwith K althoughthe deepestpeat on andaround the transectwas sampled.However, there are characteristics in all the pollen profiles which may be attributableto Neolithic activity. In the laterpollen profilcs, spatialdiffcrcnccs arc evidentwhich may highlight the needfor multiple profile studics,and may also demonstratethe potentialuse of damagedpollen in their intcprctation.
165 The zonationof pollen diagramsKLI, KU andKU identifieschanges reflecting generalvegetation successions at Kinloch which correspondswith major zonesin
Hirons andEdwards' (1990) original coreK. Subzoncsidcnti fy what appearto be localisedvegetational differences. The local pollen assemblagezones of cach profile are shownin Table6.2 andFigure 6.23 links comparableassemblage zones betweenthe differentprofiles.
High levelsofAlnus glutinosapollen in Ula(i), Ma andKIAa indicatethat the profiles post-datec. 6500BP (5460cal. BC) when comparedwith K, anddates of c. 3900BP (2430cal. BQ on the carliestcercal-type pollen grainsin KLI and KL4, which coincidewith the reductionin Alnus, suggestthat the inceptionof polleniferousdeposits at KLI andKU is not much earlierthan this, andperhaps dateto c. 4500BP (2370cal. BC). KL3a(i) displaysmuch higher values ofAhIUS pollen thanMa andKL4a(i), in additionto higher valuesof Filipenduld, fcrns and Ch:P. This pollen assemblageis similar to that of KIld which, by extrapolation,dates from c. 5200BP to c.4800 BP (4010- 3570cal. BQ, and suggeststhat KL3a(i) is older thanKLla andKL4a(i).
Following the reductionin woodlandevident in all the profiles,Poaccae and Cyperaceaedominate the pollen assemblagesand a variety of herbsarc representedincluding Filipendula, Succisa pratensis, Brassicaccac and Plantago lanceolata.Cereal-type pollen was detectedin all the profiles at variouslevels. It is probablybetter represented in K (Hironsand Edwards,1990) as the complete profile was subjectedto low power scanningfor cercal-typegrains as part of this project.On the otherprofiles, this techniquewas limited to detectingthe earliest possibleccreal-type pollen aroundthe woodlanddecline, and not usedin the later sectionsof the profiles.
166 6.5 Differencesbetween profiles
Althoughthe pollenprofiles from all monolithsat Kinloch arc similar in their representationsof local vegetation,from c. 5000BP (3860cal. BQ in K andKU and from c. 4000BP (2530cal. BQ in KLI andKL4 asdiscussed above, Scction 6.4, thereare small scaledifferences in the pollen recordedbetween monoliths (anddenoted by the subzonesof eachprofile) which requireexplanation and may assistin the reconstructionof local environmentalchanges at Kinloch.
6.5.1 The potential for an hiatus In KLI in In KLl, the transition KLla - KLlb(i) displays large and marked reductions the percentagesof Alnus followed by Coryhis pollen, a suddendecrease in Lonicera pollen and marked increasesin Poaccaeand Cyperaceae.This compareswith the other profiles K, KL4 and KL3, where Alnus and Corylus display apparently prolonged and more gradual reductions. However, all the concentration prorilcs indicate steepreductions in Alnus in what are consideredto be contemporary KL3a(ii) horizons between the profiles, (KLIa - Ulb, KL4a(i) - KL4a(ii), - less in KL3b and KIle - KIIIa), although the reduction is perhapsslightly markcd in lithology be KL3a(ii) - KL3b. Evidence of a change or LOI values could expectedin KLla(i) if an hiatus had occurred. Although a decreasein LOI values and temporary changein lithology to grey sandy silt occurs in KLlb(i), this is afler the major decline in arboreal pollen and there are no lithological or LOI changesin any of the profiles at the actual decline.
Thereare radiocarbon dates for two of the four profiles (KLI andKL4) aroundthe declinein arborealpollen, asthis coincideswith the carliestdetected cereals at Kinloch. A datemay alsobe extrapolatedfor this featureon K. Thus the declinein
167 woodlandbegins before 3840 ± 45 BP (2330cal. BC) in KL4a(ii), but woodland persistsin the profile until shortly afler this. The woodlanddecline at KLI a occurs at at 3950± 60 BP (2430cal. BC). Both thesedates comparc favourably with c. 3950BP (2440cal. BC) at the startof the arborealpollen declineat K. Although thereis statisticaloverlap at 2 SD of eachdate, there may remaina hint that woodlanddecline at KLI beganearlier than in the otherprofiles.
The amalgamationof concentration,lithological andradiocarbon data thus suggeststhat no major hiatusin sedimentationoccurred in KLI. It is probablethat the differencesbetween the percentageprofiles reflect the variability in pollen depositionbetween the profiles andwoodland appears to havedeclined earlier in the vicinity of KLI. thanin the otherprofiles (seebelow, 3.3.4).Differential sedimentaccumulation and pollen influx ratesbetween the profiles could also havecontributed to the patternsin the percentageand concentration prortIcs, and theseremain undetected as there are insufficient radiocarbon dates for pollen influx calculationsin KLI, KU andKL4.
6.5.2 The pollen assemblageprior to reduction of woodland at c. 3950 BP
(2440 cal. BQ Zones KL1a, KIM, Me, KL4a(i) and KL3a(i) and (11)
6.5.2.1 Ma(i) and KIld
Prior to c. 3900BP (2440cal. BC) on all profiles, the pollen recordindicates that mainly Alnus and Coryluswoodland was present at Kinloch. KL3a(i) is earlier thanthe basalzones KLIa andKL4a(i) andprobably correlates with the carlicr part of KIId (discussedabove; section 3.3). Sub-zoncKL3a(i) representsa woodlandflora dominatedby Alnus and Corylus,where Beluld andPinus sylvestriswere probably minor components.As Pinjis averagesc. 20 % TLP for
168 this subzone,then it is likely that the pollen is of local origin andnot acrially transportedfrom furtherafield. It could alsobe expectedthat the long distance transportof Pinuspollen grainswould producean evendistribution across the site, andin the later sub-zoncKL3a(ii), Pinus appearsin greaterpercentages and concentrationsthan in the otherprofiles, again suggesting that Pinus was locally presentto the westof Kinloch.
Poaceac,Cyperaceae, Filipendula andferns could be presentin the pollcn subzonc assemblagesdue to the proximity of the site to the Kinloch River and may representa dampenvironment. They may alsobe indicativeof woodlandclearings andit is possiblethat theseareas were extended by humanactivity as Ch:P levels arehigh in KL3a(i), andin KIM. However,KL3a(i) andKIld may not be contemporaryand KL3a(i) could be of later dateand both sub-zonescould reflect extremelylocal impactson woodlandby prehistoricpeople at different timcs.
6.5.2.2 KL1a, KL4a(i), KL3a(ii) and Me
In KLI a, KL4a(i), KL3a(ii) andthe lattcr part of KLIlc, the profilcs indicatethe continuedlocal presenceof Alnusglutinosa and Corylus,and of Pinus in KL3a(ii). However,percentage representations of arborealtaxa in the profiles vary. Slightly higher arborealpollen percentagesare recordedin KLI a, but concentrationsof thesetaxa arehigher in KL4a(i) andKL3a(i). In contrast, concentrationsof Poaceaeappear much greater in the later part of KLI a. In Kllc, pollen concentrationsoverall areat least 10 times greaterthan those in the subsequentlysampled profiles.
Although it is acknowledgedthat concentrationsmay not be a reliablecomparator astheir calculationdoes not take into accountsedimentation rates, the differences
169 in concentrationsbetween the profilcs could suggestthat, evenprior to the reductionin Alnus andCorylus woodland represented in the later zones,woodland wasperhaps sparser in the vicinity of KLI comparcdto west of this prorlIc.
Indeed,a moredivcrse woodland is representedin KL3a(ii), wlicrc Bcada, Qucrcusand Pinus s Y Ncstris pollcn hascompamtivcly greater values in both percentageand concentration terms.
Although varying sedimentaccumulation rates and pollen influx may accountfor
someof the disparitiesin pollen concentrationswithin the profilcs, there neverthelessappear to havebeen spatial differences within woodlandcomposition at Kinloch accordingto the pollen data.There is no indicationthat localiscd
burningcould havereduced woodland at KLI a, as Ch:P is low in all profiles. However,KLIa includeshigh values(10 % TLP) of LoniceraperlClyinenton
pollen which is hardlypresent in the otherprofiles (valuesdo not exceed2% TLP). As Lonicerais insect-pollinated(Grime et at., 1988)and produces low
amountsof large,heavy grains which will rarely travel far from source.The taxon
alsorequires substantial amounts of light to flower (Grime cl al., 1988),thus the
abundanceof its pollen in KLI a probablyreflects an openingup of woodlandin this vicinity. The pollen of Rexangustifolium and Iledera helix, also undcrstorcy taxawhich requirelight to flower, alsooccur in KLI a. Although the pollen of all
theseunderstorey taxa may havederived from the erosionof woodlandsoils to the
north andupslope of the narrowmire, openingsin woodlandto the north of KLI would alsobe requiredfor the flowering of thesetaxa andfor any substantial
erosionto occur.
The spatial representationsof woodland in the Kinjoch profiles prior to c. 3900
BP (2370 cal. BC) indicate that trees were sparse o the cast of Kinloch, close to
170 theknown site of humanoccupation. The earlier reduction in woodlandcover and comparativelylow arborealpollen sum recorded in KLIa maytherefore be linked to woodlandclearance, possibly to openup areasfor cultivationor to obtainwood for domesticuse.
6.5.3 Reduction of Abitts-Corylus woodland c. 3900 BP (2440 cal. BQ KL1b,
K111a,KL3b, KL4a(ii)
At the transition to KLIb(i), KL4a(ii), KL3b and KIM a reduction in percentages and concentrationsof arboreal pollen is evident. This appearsmore markcd in
KLI, and at 3840 + 45 BP (2330 cal. BQ in KL4b(i), and at c. 3900 BP (2440 cal.
BQ in KlIa, much higher values of Alnus and Corylus are recorded compared to
KLl, where woodland is substantially rcduccd at 3950 ± 55 BP (2430 cal. BQ.
Although there is no confirmatory date for the woodland decline in KL3, there neverthelessremains an indication that woodland decline to the west of the
Kinloch site was comparativelylate.
Alnusglutinosa macrofossils are present in the peatof monolithsKL3b and KL4b(i) but not in KIIIa andKLIb(i). In KLIb(i) thereis a significantreduction in LOI from 90 % to 70 % andgrey silty peatwas recordedin the lithology. A similar horizonof silt particlesoccurs in the lithology of K betwccn67 cm and73 cin andalthough small reductionsin LOI appearin KL3b, thereis nothing as significant.The reductionin LOI andthe presenceof silt particlesin K andKLI aresymptomatic of the erosionof minerogenicmaterial from the slopesto the north,of the mire. If a greaterproportion of woodlandwas reducedto the castand north of the site (asthe pollen and lithology may be suggesting),then more eroded materialwould be expectedto be introducedto depositsto the castof the mire.
171 Datarelating to damagedpollen in the profiles may also be significantto interpretation.Unfortunately thercarcno damaged pollen dataavailable for K.
Thereare no significantdifferences in TD% betweenthe profiles, evenwhen
Pinus is excluded(as this pollen type breakseasily and may distort figurcs). Nevertheless,the percentageof damaged(mainly pitted) arborealpollen typesis
greaterin KLIb, whereAlnus pollen is 60 - 80 % pitted and Coryluspollen 20 -
50 % pitted comparedwith c. 20 - 30 % pitted for eachtaxon in KL4a(ii) and KL3b. Total damagedpollen percentagesmay be affectedby local factors
affectingpollen preservation,and a greateramount of Poaccaeand Cypcraccac pollen is damagedin KL4a(ii) andKL3b. However,greater damage to arboreal pollen in KL1 may meanthat this is derivedin part from upslopelocations. Increasesin the ratio of damagedto undamagedarboreal pollen also occur in
KLlb comparedwith KL4a(ii) andKLA
It is possiblethat the reductionin woodlandfrom c. 3950BP (2440cal. BQ
occurreddue to increasedwetness at the site andHirons andEdwards (1990)
mentionthat increasedstorminess in the Atlantic could havecaused this. High
amountsof Cyperaceaein all profiles following the main declinein woodlandare indicativeof a dampenvironment but thereare few otherindications of increased
wetness.In KL I b(i), Sphagnumspores increase and Succisa pratensis is recorded
at around5% TLP for most of the subzone,but not in any of the otherprofiles. This localisedreflection of wetnessmay be linked to the crosionof slopesto the
north of KLI and associatedrunoff andincreased waterlogging in the castof the
mire.
Humanactivity may alsobe consideredas a factor influencingwoodland dcclinc
at Kinloch c. 3950 BP (2440 cal. BQ. Hirons and Edwards (1990) suggcstcdthat
172 humanactions will at leasthave augmented the effectsof climatic change.The reductionin woodlandcoincides with the appearanceof early ccrcal-typcpollen at K, KL4a(ii) andKLIb. It is possiblethat the ccreal-typepollen is in fact large-
sizedPoaceae pollen, possibly that of Glyceriafluitans,but thereis other circumstantialevidence supporting the casefor cerealcultivation on Rum. As arborealpollen is reducedand Poaceae and Cyperaccae become the predominant
pollen types,Plantago lanceolata pollen increasesin KLlb(i) andKL3b and the
grazingof domesticanimals could havebeen responsible for this (cf. Newell, 1988;Bennett et aL, 1990).The pollen assemblageincludes Brassicaccac, R11"lax
acetosella,Succisa pratensis and Ranuncuhisacris-type, all taxawhich correspondwith Hicks' (1988) suggestedbarley (Hordeuni) field flora. Cereal-
type pollen of Neolithic datewas found on pottery shardsrecovered during the
excavation(Moffatt, 1990)also suggesting that ccrealcultivation occurredat Kinloch.
Increasesin Ch:P in KIIIa andKLIb may alsoreinforce the casefor human
activity. As thereare no contemporaneousincreases in Ch:P in KL4a(ii) and the
earlierpart of KL3b, microscopiccharcoal appears to be concentratedtowards the eastof the pollen site transectand closest to the areaof excavationand known humanoccupation. This patternis alsoreflected in the pollen assemblages.It was
mentionedabove that KL3b andKL4a(ii) incorporateda greaternumber of
woodlandtaxa than KIIIe andKLIb. In contrast,percentages of Calluna vulgaris
pollen aregreater in KIIIa, andKLIb andthis specieswas alsodiscovered as
pollen andmacrofossil evidence in contextsfrom the archaeologicalsite of Neolithic date(Wickham-Jones, 1990). Calluna vulgaris heathsare easily
combustible,either by humanaction or naturalevents, but burning of heathlands
173 in prehistorymay haveoccurred as a form of management(Edwards cl al., 1995, Stevensonand Birks, 1995;Whittington and Edwards, 1997a).
High levelsof Lonicerapericlymenuntpresent in KLI a, whilst possiblyrclating to naturalopenings in woodland,arc perhapsmore likely to be a rcflcction of an openingof the woodlandat this point by humans,and a decreasein LOI in KLlb(i) reflectserosion likely to result from the exposureof the surfaceof slopes to the north.
Whilst climatic deteriorationmay havecontributed to the reductionin woodland throughoutthe region,the accumulationof evidenceat Kinloch suggeststhat
activitieslinked to humanoccupation can best explain vegetation changes in the area.The pollen profiles andlithologies suggest that the focusof impactduring the final stagesof reductionin woodland,c. 3950BP (2440cal. BQ, was to the
eastand northeast of the sitewhich is closestto the known areaof Neolithic
occupation.
6.5.4 The post woodland decline flora from c. 3900 BP (2440cal. BQ KLIc, KIIIb, KL3c, KL4c
Following the woodlanddecline, the profiles becomedominated by Poaccaeand
Cyperaceaepollen, which indicatesa wet grasslandenvironment. Some herbs arc
present,particularly the cultural indicatorsBrassicaceae, Plantago lanceolata and Rumexacetosa which, in all profiles, frequentlycoincide with occurrencesof
cereal-typepollen. This implies that the Kinloch areawas at leastcpisodically cultivatedfrom the Neolithic to the present.Ch: P valuesare high comparedto the
earlierzones when treecover was greater.Microscopic charcoal may odginatc from increasedlocal burning dueto humanactivity (Edwards,1990) or may be the
174 result of increasedmicroscopic charcoal fallout dueto the removalof the fdnging woodland.
Low levels of the pollen of Alnus and Corylus in all profiles indicate a sparsebut continuous presenceof woodland within the region. An expansion in cricaccous pollen in KIIlb is interpreted by Hirons and Edwards (1990) as representing the colonization of drier sandy soil around the mire by heathland vegetation following the removal of hazel scrub. Little ericaccouspollen is recorded in any of the new
Kinloch monoliths, but small increasesin Calluna vulgaris and Vacciniuni-type pollen in KLIc, KL3c and KL4c may reinforce this interpretation.
Increasesin Pinuspollen appearat the top of the profiles in KLlc(iii), KL4c(iii), andKL3c(iii), andprobably represents recent conifer plantationaround the Kinloch area.
Although the later profiles do not fall within the rangeof early humanimpacts, thereare variations in the laterpollen spectrawhich areworthy of interpretation.
6.5.4.1 Ahjus increase,,KL3c(ii),, KL1c(ii)
An increasein percentagevalues of Alnusghtfinosa pollen occurswithin KL3c(ii) andKLIc(ii) but this is not seenin profiles K or KU It is assumedthat these zonesare contemporaneous and concentrations of Ants arehigher in KL3c(ii). The LOI profiles areconsistently high at around98 % so thereis no indicationof erosionand the associatedinflux of mincrogcnicmaterial to the mire surfaceat this time. However,the depositsof KLlc(ii) containAhms ghtfinosa wood fragmentswhich do not appearin the othermonoliths. This suggeststhat a
175 regenerationof Alnus woodland occurred in the locality of KLI despite less Alnus pollen being recordedhere comparedto KL3c(ii).
It is possible that the mechanismsof distribution of Alnus pollen mean that little is depositedwithin or near the woodland itself. Vuorela (1973) notes that Alnus pollen is easily transportedabove the woodland canopy and quickly distributed by wind action away from wooded areas.It is not impossible that a similar cffcct may occur for all arboreal pollen as the surface samplesfrom Kinloch (figs. 6.4 - 6.7) indicate that those towards the centre of the mire surface and away from areasof woodland to the north and east of the site show the greatestamounts of arboreal pollen. However, Alnus pollen fallout is substantial and Vuorela (1973) found that in surface samplesfrom cultivated fields close to woodland, Abuts pollen was to be found in similar values regardlessof distance from source. If Alnus woodland was located either to the eastor west of the bog as the subzonesKL3c(ii) and KLlc(ii) suggest,then the airborne redistribution of pollen from these areasshould result in a recovery of Alnus percentagesand concentrationsin profiles KII and KL4.
Thereis no indicationin the damagedpollen recordsto suggestthat pollen was betterpreserved at any of the profile positionsin the bog. Total damagedpollen percentagesare similar in all profiles wheredamaged pollen is recorded,and similar percentagedamage to Alnus grainsoccurs in both KLlc(ii) and KL3c(ii). If damagedpollen is not a significantfactor in explainingthe recordeddifferences betweenprofiles, than it is possiblethat fast sedimentaccumulation rates and low pollen influx andsampling resolutions have resulted in the omissionof the A11111S rise from KL4c andKIIIa or b profiles.
176 A further considerationin explaining the differences is that the Ahms profiles reflect a patchy local distribution of Alnus. It is noted that Coryhis avellatia-type pollen increasesin conjunction with Alnus pollen in KL3c(ii) whereaselsewhere it maintains steadylow levels consistentwith a continued regional presence.
Therefore, it is also likely that regenerationof Alnus - Corylus woodland to the west of the site, either in the more fertile Kinloch Glen, or on the slopes to the northwest, resulted in an increasedinput of thesepollen types to the west of the mire. Only Alnus appearsto have been presentin KL I c(ii) to the cast.
In both subzonesKL3c(ii) andKLlc(ii) thereare also increasesin Ch:P, and cereal-typepollen andruderal species are detected. Calluna V111garisand Vaccinium-typepollen increasein profiles at KLlc(ii), KL4c(i) andKL3c(ii), but
particularlyin KIIId asmentioned above (Section 6.5.4). The increasein heathlandpollen typesmay be directly relatedto burning of eithernatural or
humanorigins andthis may alsohave caused the temporaryexpansions in A11111S
and Coryluspollen in the outlying profiles asit hasbeen suggested that
expansionsin Alnus and Corylusmay occurafter fire (MeVcan, 1956a;1956b). However,Alnus is alsorecorded as a sourceof wood for the productionof
charcoal(Brown andBarber, 1985) and it may havebeen been coppiccd alongside Corylus.This may alsoaccount for high levelsof Ch:P andincreases in woodland
within K111b,KL3c(ii) andKLlc(ii).
The overall evidencesuggests that Alnus and Coryluswoodland was rc-
establishedat the mire edges.Pollen preservation factors are unlikcly to be responsiblefor the differentpollen profiles and the paucityof Ahjus in the pollen recordat KLlc(ii), despitethe presenceof its macrofossilsin the lithology of this subzone,may be explainedby taphonomicfactors and the aerialdistribution of
177 pollen awayfrom its source.Different ratesof sedimentaccumulation and pollen influx may explainthe absenceof a recoveryin Alnus and Coryluspollen In K and
KL4, but very local representationsof thesetaxa and restrictedpollen distribution could also explainthe variationsbetween the differentKinloch profiles.
6.5.4.2 Cereal-type pollen and anthropogenic indicators In the later profiles
There are variations in the timing of occurrencesof cercal-type pollen and
associatedcultural indicators between the cores. Although the inherent problems
of the detection of cereal-typepollen (i. e. restricted dispersal; low pollen
production) mean that it may not always be detected,even when low-powcr
scanning of samplesis employed, the distribution of ccreal-type pollen between
profiles and the occurrencesof rudcral speciesassociated with cultivation, may
provide an indication of the areaunder cultivation.
It was mentionedearlier that cereal-typepollen is recordedalongside the temporaryrecovery in Alnus pollen. A singlegrain of ccrcal-typcpollen is
recordedin KLlc(ii) andtwo grainsin the latter part of KL3c(ii), and these
coincidewith increasesin Brassicaceae,Plantago lanceolata and Ranuncidus
acris-typepollen. Brassicaceaeand Plantago lanceolata arc particularly evidentin KL3c(ii). Cereal-typepollen is recordedat nearly everysampled level in KIllb,
althoughPlantago lanceolata is more frequentlyrecorded than othcr flora
associatedwith cultivation.The family Brassicaccaeconsists of mainly insect-
pollinatedspecies (e. g. CardaminePratensis,Sinapsis anensis) whosepollcn is unlikely to be distributedfar from original sourceswhereas Plantago lanceolata is wind-pollinatedand its pollen may travel somedistance. The consistentpresence
of Brassicaceaeand Plantago lanceolata in KL3c(ii) suggeststhat cereal cultivationwas very local. The absenceof ccrcal-typepollen and lower instances
178 of ruderalspecies in KL4c may be a functionof detectionmethods, but may also rcflect the distanceof this profile from areasof cultivation. KII, KLI andKL3 arc all nearerthe bog edgeand thus were probably closer to areasof cultivatedland.
6.6 Summary of the Kinloch data
The palaeoenvironmentalevidence from Kinloch shows that woodland at the site was drastically reduced from c. 3950 BP (2440 cal. BC) and replaced by a damp open flora dominated by Poaceaeand Cypcraceae.Localiscd erosion and flushing of the mire surface may also have occurred. Cereal-type pollen in K, KLI and
KL4 and the correspondingpresence of anthropogenicindicators in some of the profiles suggeststhat woodland reduction was at least partly linked to human
impacts, which originally appearsto have been concentratedto the cast of the site,
close to a known areaof Mesolithic and Neolithic human occupation.
Cerealcultivation continued throughout the post-Neolithicperiod until the present
day althoughareas under cultivation may havevaried through time. A temporary
recoveryin Alnus and Coryluswoodland occurs in KL3c(ii) following the earlier
woodlanddecline and a similary recoveryin Alnus occursin KL I c(ii). Increases in anthropogenicindicators, Ch: P andEricaccae coincide with tile woodland
recoveryin both theseprofiles, andheathiand may haveexpanded in responseto
management.
Differencesin the pollen betweenmonoliths may be interpretedas the result of a
complexinterplay of taphonomy,local ecologicalvariations, pollen preservation andmire hydrology.Despite this complexity,a multiple profile reconstructionat Kinloch hasproduced a moredetailed picture of Neolithic humanactivity and
subsequentenvironmental changes on andaround the site thanwas availablefrom
179 the original singlemonolith. However, the needfor a suite of radiocarbondates in multiple profile studiesis highlighted,as these would allow the estimationof sedimentaccumulation and pollen influx rates,and result in more confident comparisonsand interpretations of the profiles.
180 CHAPTER 7- SYNTHESIS OF POLLEN DATA FROM THE INNER
HEBRIDES, OUTER HEBRIDES AND WEST COAST OF
SCOTLAND
7.1 Introduction
Thc, preccding,,- chapters presented pollen diagrams from various sites in the
Inner Hebrides spanningthe period c. 10,000BP to c. 2500 BP (9530 - 600 cal. BQ. The following summarizesdata covering the Mesolithic and
Neolithic periods and comparesthem with other pollen evidence from the
Inner and Outer Hebrides and the west coast Scottish mainland.
7.2 Early Holocene Vegetation in western Scotland
7.2.1 The early Holocene c. 10,000 BP (9530 cal. BQ
A summary of published data concerning the composition of early Holocene vegetation in the Inner Hebrides was provided in Chapter 1, Section 1.3. This and the data acquired from the new pollen profiles fits the general precedents indicated by isopollen maps for the Inner Hebrides (Huntley and Birks, 1983;
Birks, 1989). Early in the Holocene, c. 10,000BP (9530 cal. BQ the infcrrcd vegetation at Loch a!Bhogaidh and Loch an t'Suidhe was juniper and willow scrub, with birch forming localised copses,and open areasof herb-rich grassland.This vegetation is similar to that indicated for Skye (Birks, 1973;
Williams, 1977; Lowe and Walker, 1991) and north and south central Mull, although Empetrum heath appearsto have been presentin the Glen More area
(Lowe and Walker, 1986aand b; Walker and Lowe, 1985,1987) and probably had a localised presencein many other parts of the islands at this time. At Loch Cholla on Colonsay, in the south-wcst of the Inner Hebrides, early Holocene vegetation appearsto have been composedof birch woodland and sedge-richgrassland.
181 The woodlandin the Inner Hebridesc. 10,000BP (9530 cal. BQ was initially dominatedby Betuld,but Corylusis the next arborealtaxon to be found in mostprofiles andin manycases it undergoesa rapid expansionto becomethe
predominanttaxon in percentageterms (e. g. 50 - 80 %). This is rcinforccdat Loch a!Bhogaidh by muchhigher influx ratesfor Corylusthan othcr arboreal
taxabetween c. 9000and 8000BP (8225- 6860cal. BC), althoughpollcn influx ratesat Loch an t'Suidheand A'Chrannag are similar for Corylusand
Betula,suggesting that in westernMull andUlva thesetaxa were present in
similar quantities.The timing of the Corylusexpansion may vary, and possiblereasons for the varioustimings of the expansionof Corylusarc discussedbelow.
The early Holocenevegetation assemblages and successionsof the Inner Hebridescompare closely with thosefor westernScotland although the
numberof well-datedsites with Holocenepollen recordson the western
mainlandis restricted.A pollen profile from Drimnagall,on the Scottish
mainlandeast of the Isle of Jura(Rymer, 1977), and pollen prorilcs from RacksMoss andAros Mosson the Kintyre peninsulato the southof Drimnagall(Nichols, 1967)suggest that birch woodlandwas establishedin
the early Holocenealthough environments remained largely openareas of herb-richgrassland and juniper scrub.Willow was probablypresent locally
andPinus appearsto havebeen significant at Aros Moss,although it should be notedthat neitherthis or the RacksMoss profile is radiocarbon-datcdand the Pinus pollen may be of Lateglacialdate. In tile Inner Hebrideanpollen diagramsfor the early Holocene,Pinus is recordedonly in very low amounts
comparedto its earlypresence in Lateglacial-Holocenctransitional prorilcs
on Mull andSkye (Birks, 1983;Lowe andWalker, 1992a,1992b; Walker and Lowe, 1985,1986a,1986b, 1987).
182 Otherpollen profiles from the SolwayFirth (Moar, 1969), Dubli Lochan, Loch Lomond(Stewart et al., 1984),Pulpit Hill, Oban(Tipping, 1992)and Loch Maree,Loch Assyntand Sionascaig (Birks, 1980)amongst others (rcfcr
to Figure 7.1) confirm the generalpicture presented above for early Holocene
vegetationassemblages, although local variationsin amountsof willow scrub, Empetnimheath and tall-herbs occur. Following the expansionof Corylus, the Scottishmainland percentage profiles suggestthat Coryluspopulations
werenot asdense with increasinglatitude, or at leastthat flowering of this taxonwas suppresseddue to lessfavourablc climatic conditions.The
percentageCorylus curves at Loch Maree,Loch Assynt and An Druim only reachesbetween 20-30 % UP (Birks, 1980). In othernorthcrly pollen profiles, suchas Loch Clair, Loch Assyntand An Druim Battla pollen
reacheshigh percentages(c. 50 % TLP) (Birks, 1980),which alsoreflects the cold andwet climateof northwestScotland.
The early Holoceneflora of the OuterHebrides is representedin a numberof
pollen diagramsfrom variouslocations in the islands(Figure 7.1) which indicatelocal variationsin the extentof birch woodland,Empantin heath,
juniper scruband grasslandsimilar to thoseelsewhere in the Inncr Hebrides
andthe westernmainland. Although Corylusis presentin most of the Outer Hebrideanpollen recordsfrom c. 9000BP (8225cal. BQ, neitherhazel nor
birch appearto attainlevels achieved in the more shelteredInner Hcbridcsor
in the southand west coastmainland, as percentages and concentrations of
thesetaxa arecomparatively low (e.g. Birks and Madsden,1979; Bennett ct
aL, 1990;Fossitt, 1996).
183 7.2.2 The Corylus rise (from c. 9800BP 19300cal. BCJ)
The earlypostglacial behaviour of Corylusavellana and its rcprcscntationin pollen profiles hasbeen the subjectof muchdebate (Smith, 1970;Boyd and Dickson, 1986;Huntley, 1993;Edwards and Berridge, 1994;Edwards, in press;Sugden and Edwards, in press).Isopollen maps (Huntley and Birks, 1983)and pollen isochrones(Birks, 1989;Whittington andEdwards, 1997) indicatethat Coryluswas establishedin westernScotland and the Inner and OuterHebrides by c. 9500BP (8900cal. BQ. In pollen diagramsfrom the
OuterHebrides, Corylus never reaches the high percentagevalues seen in
profiles from the Inner Hebridesand western Scotland (as much as 80 % TLP at Loch a!Bhogaidh, this volume)where rapid increasesto thesehigh values may occur.The reasonfor theserapid expansionsis not entirelyclear, andit
is assumedthat climate,migrational lag of othertaxa, local ecology,soils and hydrologymay all haveinfluenced Corylus to different degrees(Huntley and Birks, 1983;Boyd andBoyd, 1986;Huntley, 1993).Human assistance has
alsobeen cited aspossibly assisting the rapid expansionof this taxon (Smith, 1970;Huntley andBirks, 1983;also cf. Edwards,1990; Huntley, 1993).
An assessmentof the differenttimings of the Corylusrise in westernScotland
andits islands,and the differentconditions associated with this may divulge further informationson the possiblecauses of the early Holocenebehaviour
of Corylus.The rangeof datesfor the hazelrise at selectedsites from the Inner Hebrides and somefrom the westcoast of Scotlandis prcscntcdin Figure7.2. The low Corylusvalues in many OuterHebridean pollen prorlics
meanthat the Corylusrise is not aspronounced and dates for its rise in these islandshave been excluded. The hazelrise or expansionis derincdsensu
Smith andPilcher (1973) as the point wherethe taxonbegins a climb to hazcl sustainedhigh percentagesin the pollen profiles. The datesfor the expansionsdenoted in Fig. 7.2 arc spreadover c. 2500radiocarbon years. The
184 earliest expansionsof hazel occur at Loch an t'Suidhe and Loch Cholla in the
Inner Hebrides c. 9800 BP (8900 cal. BC) and most of the expansionsoccur before c. 9000 BP (8225 cal. BC). These include a range of sitcs from the southern Inner Hebridesto the north of Skye. Dates for the spreadof Corylus at A'Chrannag, Ulva and Glen More, Mull (Walker and Lowe, 1985) arc later, later dates for c. 9000 - 8500 BP (8225 - 7470 cal. BC), and are recorded Glen Varrigill and Loch Fada on Skye (Birks and Williams, 1983;Walkcr and,
Lowe, 1991a)' as well as Loch Clair on the west coast mainland (B irks,
1980).
The apparentrapidity of the Corylus expansionat some sites may be the result of sub-optimal sample quality. The Corylits rise at An Druirn in northwest Scotland datesto c. 10,000BP (9530 cal. BC) and also corresponds with a rapid increasein Betula (Birks, 1980). The dates on the profiles indicate extremely rapid sedimentationin the early Holocene and as sampling resolution is low it is difficult to determine the chronology of events from this profile.
Polleninflux may be preferablein assessingthe early HoloceneCorylus expansion.At Loch a!Bhogaidh there are differences in the percentage representationsof the Corylusrise in the variousprofiles from the site, but Corylusinflux is high in all the profiles.In contrast,Coryhts influx is low at
Loch an t'Suidhedespite this taxonrising rapidly to high percentagevalues c. 9800BP (9300cal. BC). The influx curvessuggest that the Corylus woodlandwas denserand established more quickly at Loch a'Bhogaidhthan at Loch an t'Suidhe,although the percentagecurves would suggestotherwise.
Thereare no indicationsthat latitudinal differencesaffected the spreadof early Corylus populations although flowering or the extent of Corylus
185 populationsmay havebeen limited with northerlylatitude and with a greater degreeof exposureto westerlywinds andrain. In northwestScotland tile date of the Corylusrise is comparablewith that of the Inner Hebridesislands but percentagesremain lower (c. 20-30% TLP) thanin Inner Hebridesand
southwestcoast profiles (c. 50 %).
Altitude is not likely to haveaffected the Corylusexpansion and its rate of
spreadat the sitesin question(cf, Birks, 1973).Site positionmay have influencedthe Corylusexpansion in differentlocalities. Many of the sites
wherethis occursprior to c. 9000BP (8225cal. BQ are coastal(e. g. Loch an t'Suidhe,Mull) or in areasof subduedtopography (e. g. Loch a'Bhogiadh,
Rhinnsof Islay; Loch Cholla,Colonsay). In contrast,Coirc Claccachand Tornessin the GlenMore regionof Mull andLoch Fadaon Skyeare
surroundedby steep,high slopesand are situated in the interior of the islands. Glen Varrigill on Skyeis alsosituated inland. The physicalbarrier of
topographycombined with the inlandpositions of thesesites could have
delayedthe expansionof Corylusto theseareas, and low percentagevalues of
this taxonin the earliersections of the pollen profiles could result from pollen
derivedfrom the wider region.Expanses of waterwith strongcurrents
separatingthe islandsfrom the mainlandhave also been suggested as delayingthe expansionof Coryluson Arran (Boyd andBoyd, 1986)and this
waspossibly the caseon Ulva, whereCorylus at A'Chrannagdoes not expand
until c. 8700BP (7790cal. BQ.
7.2.2.1 The influence of soils
Soils may alsohave imposed limitiations on early HoloceneCorylus
populations.The spreadof Corylusmay havebeen affected by the distributionof acidic andthin soils, asthis taxonprefers base-rich soils. Ijowever, Corylusoccurs relatively early in the outer Hebridesand northwest
186 Scotlandwhere high levelsof precipitationmay havehindered soil developmentand caused early acidificationand leaching.In additionto climate,other influences on soils may havecontributed to the uneven expansionof Corylusin the early Holocene.Glenn Varrigill and Loch Fada on Skye,and the GlenMore regionof Mull, whereCorylus expandslate, are in areasknown to havebeen glaciated in the Loch LomondRcadvance (Lowe andWalker, 1986a;Walker andLowe, 1990;Ballantyne and Dawson, 1997). Huntley (1993)suggests that soil developmentcould havebeen initiated in the Allerod interstadial,but in areassubjected to subsequentglacial advance andperiglacial activity, pcdogenicdevelopment would havebeen impeded until the early Holocene.The extremesof topographynear these pollen sites would alsohave hindered soil stability following deglaciationand the developmentof soils ableto supportCorylus woodlands could havebeen delayed.
It is unlikely that the effectsof glaciationduring the Loch LomondStadial
andthe extremesof slopecould havedelayed the Corylusexpansion at A!Chrannag and Livingstone's Cave basins on Ulva. The southerncoastline of
the islandcontains many shelteredvalleys with no extremesof topography
andit is locatedclose to the westerncoastline of Mull whereat Loch an t'Suidhe,to the south-westof Ulva, the Corylusexpansion occurs c. 9800 BP
(9300cal. BC). Therefore,notwithstanding difficulties arising from possible
hiatuses,or the unreliability of the chronology,other factors arc probably
responsiblefor the later expansionof Corylusin southeastUlva. The absence of Corylusin the early sectionof the AC I profitledoes not excludethe island, possibility that the spreadof Corylusoccurred earlier elsewhere on the andgiven the earlierCorylus rise at many siteselsewhere in the Inner and OuterHebrides, hazel was probably present on Ulva much earlierthan is
currentlyrecorded.
187 7.2.2.2 Human influence and the Coqlus expansion.
The contributionof humaninfluence on the Corylusrise hasbeen largely discounted(cf. Edwards,1990; Huntley, 1993)and at nearly all the sites
includedin this studythere is no evidenceto suggestthat peoplemay have influencedthe earlyHolocene Corylus expansion. However, at A'Chrannag on Ulva thereare increases in microscopiccharcoal associated with tile Corylusrise, and Callunavulgaris increasessimultaneously, suggesting
probablelocal burning.Although the expansionof Calluna heathmay be coincidental,and fires (whetherof naturalor humanorigins) may havebeen confinedto areasof heathland,there remains a possibility that burning on Ulva may haveassisted the spreadof Corylusavellana, and that this may
havebeen a resultof the useof fire by Mesolithic people
7.2.2.3 Summary of the evidencesurrounding the Corylus rise
The abovesuggests that the spreadand expansion of Corylusin the Inner Hebrides,Outer Hebrides and western Scotland is not as straightforwardas
pollen isochronemaps might indicate.Low samplingintervals and the interpretationof poorly datedpercentage data may haveprovided erroneous informationin the past.Variations in topography,soils and anteccdcntsite
conditionscould all haveinfluenced the spreadand behaviour of the taxon. Climateand latitude may havealso affectedthe post-colonizationbehaviour
of Corylusbut thereis little evidencein the currentlyavailable pollen datato
supportthis. From the availablepollen datait is not possibleto draw any firm conclusionsabout the factorsresponsible for the spreadand expansionof Corylusin the Inner Hebridesand Scotland.Further data is obviously
required,and should not be restrictedto pollen analysisbut should incorporatewider palacoenvirorunentalinvestigation (Scction 8.3).
188 7.3 Holocenevegetation patterns C.9000 - 5000 BP
(8225- 3870cal. BQ Betweenc. 9000and 8000 BP (8225- 3870cal. BC) Mints and Quercus expandin mostprofiles from the Inner Hebrides,Outer Isles andwestern Scotland,indicating a diversificationof woodlandwhich fits the general schemefor the spreadof thesetaxa in isopollenmaps (Huntley andBirks, 1983),and isochronc maps (Birks, 1989;Whittington andEdwards, 1997). The variationsin the timings of the expansionsof thesetaxa at specific sites may be explainedby differencesin local topography,soils, competitionand the accessibilityof a site to taxonpropagules.
Comparativelylate expansions of Ulniusand Quercus at CoireClacchach in theGlen More region of Mull, datingto c. 7800BP (6640cal. BC) and7000 BP (6010cal. BC) respectively, provide an example of thepotential of environmentalfactors as influences on theabundance of thesetaxa. Although a limiting factorto thespread of thesetaxa may have been the delayed transportof propagulesinto thismountainous region, these taxa expanded earlierat Torness,also in theGlen More region, c. 8200BP (7280cal. BC) (Walkerand Lowe, 1985). However, Tomcss in moresheltered and the authorssuggest that site exposure could have been a factorlimiting the successof Quercusand Ulmus at CoireClachach. Similar environmental constraintsmay have operated at BcinnReudle in north-wcstMull. The undatedpollen profile from this sitecontains very little Ulmusand Quercus pollen(similar in percentagesto those recorded in theOuter Hebrides) and BeinnReudle is exposedto salt-ladenwesterly winds and rain. This has resultedin theleaching, acidification and thinning of soils(Lowe and Walker, 1986).At Lochan tSuidhe on theexposed Ross of Mull peninsula,Ulnuis andQuercus influx is verylow until c. 7800BP (6640cal. BC) suggestinga delayin thecolonisation of thesite by thesetaxa. Kinloch, Rum and Loch
189 Cholla,Colonsay also provide examples of exposedsites where Ultnusand Quercusare poorly represented in thepollen profiles and may nevcr have beenlocally present.
A substantialincrease in Ulmusand Quercus occurs in theLoch an t'Suidhe profilesc. 5700BP (4530cal. BC) andthis correspondswith thelate expansionof thesetaxa at A'Chrannag,Ulva. Reasons given in Chapter5 for thedelayed expansion of Ulmusand Quercus at thesite were the persistence of a fire climaxcommunity of Corylusand Calluna, possibly resulting from humaninterference, or thedifficulty of propagulesreaching the site due to the islandbeing surrounded by cliffs andfast flowing water. However, increased arborealpollen influx at Lochan t'Suidhe, although possibly coincidental, mayalso be linkedto thelate arboreal expansions of thesetaxa on Ulva. Althoughall arborealpollen influx ratesincrease slightly in the2LS profile c. 5700BP (4530cal. BQ, it is Ulmusand Quercus which show the greatest increases.As sealevels were higher in theearly Holocene reaching a maximumbetween c. 8000and 6000 BP (6860and 4910 cal. BQ, falls in localsea level may have resulted in lesssalt spray at thesites, to which Quercusand Ulmus could be particularly sensitive. This may then have allowedthese taxa to increaseas components of woodlandin westernMull andon Ulva.Earlier high sea levels may have infiltratcd the water table at coastalsites, and subsequent falls would have produced lower ground water salinitywhich may also have encouraged the spread of Qucrcusand Ulnuis at coastalsites.
It was alsomentioned in Chapter5 that the datesfor the woodlandexpansion on Ulva may be erroneousand an earlierexpansion, possibly c. 7700BP (6600cal. BQ is possible.This would correlatewith expansionsof Qucrcus and Ulmusin remoteor exposedregions of Mull (above).A period of
190 climatic cooling (Alley et al., 1997;Barber et al., 1999;Willcmsc and T6mqvist, 1999)has been identified as occurring from c. 7700BP (6600cal.
BQ which may haveadversely affected woodland at Loch a'Bl1ogaidhand
Loch an t'Suidhc(this volume),where human interference in vegetationmay
alsohave occurred. Given that in someareas mixed deciduouswoodland may haveincreased in someareas despite a colder,drier climate,this reinforces the suggestionthat at LochaBhogaidhand Loch an t'Suidhc,human impact
may havebeen more important than climatic conditionsin producingchanges in local vegetation.
The later expansionsof Ulmusand Quercussuggested by pollen diagrams from the north-westmainland of Scotland(e. g. Loch Assynt,An Druim, c. 7000BP [6010cal. BC] both Birks, 1983),are probably connected to a
slowerrate of spreaddue to climatic constraints(Birks, 1989).The limitations imposedby climateon both thesetaxa meanthat they only appear
in low amountsin mostpollen diagramsfrom the OuterHebrides from c. 8000BP (6860cal. BC) (e.g. Loch Lang, S. Uist, Bennettct aL, 1990;Loch Phuinnd,S. Uist, Fossitt,1996; Borve, Benbecula, Edwards and Whittington,
1997;Callanish, Lewis, Bohnckc,1986. The low frequenciessuggest that elm
andoak wereminor componentsof woodlandin the OuterHebrides rather thanthe result of long distancetransport of pollen grainsas suggested by
Birks andMadsden (1979), but they wereprobably restricted to shcltcrcd
valleysand better soils, allowing birch-hazelwoods to dominateover the
greaterparts of the islands.
Mixed woodlandpersisted in mostplaces in the wcst coastof Scotlandand the Inner Hebridesthroughout the later Holocene.The activitiesof Neolithic
peoplebegan to increaseopenings in woodlandfrom c. 5000 BP (38 10 cal. BQ (e.g. Loch a'Bhogaidh,Islay, Edwardsand Bcrridge, 1994; Kinloch,
191 Rum, Hirons andEdwards, 1990; Machrie Moor, Arran, Robinsonand Dickson, 1988).Increased oceanicity of climatemay also haveassisted tile
spreadof heathlandand bog over manyareas, but at somesites woodland
persisteduntil very recently(c. f. Walker andLowe, 1985).
In contrastto the above,the birch-woodlandcover of the OuterHebrides declinedcomparatively early. From c. 6200BP (5130cal. BQ Calluna heath
andgrassland increase at the expenseof treesin manypollen diagrams(e. g. Borve,Benbecula, Whittington and Edwards, 1997), and indicatorsof
disturbedground such as Plantago lanceolata and Potentilla-typc appear (Bennettet aL, 1990;Fossitt, 1996). In someinstances these changes in vegetationmay be attributedto Mesolithic activity (seebelow), but the
exposednature of manypartsof the OuterHebrides would provide fairly marginalconditions for woodlandand high levelsof rainfall, disturbanceby
stormsfrom the Atlantic, andcold winds could easilyreduce woodland and allow blanketmire andheath development.
In the morenortherly parts of the westernmainland, expansions in Pillus
occurin somepollen profiles from c. 7000BP (6010cal. B C) asBelula and Corylusdecline. Examples include Loch Maree,Sionascaig and Loch Clair,
wherePinus dominatesthe pollen profilesuntil c. 4400BP (3120cal. BC) (Birks, 1980). Relativelyearly reductionsin Ulnutsc. 5500BP (4380 cal.
BC) in someprofiles from the Scottishmainland indicate that climate and
exposuremay havestrongly influenced vegetation patterns here, and the later reductionsin Pinus areprobably also linked to increasedrainfall. However, the influenceson vegetationpatterns may be morecomplicated than currently
availablepollen profiles andradiocarbon dates suggest.
192 7.3.1 Abius expansion
The postglacialpattern of spreadfor Ants glutinosa is complicatedand it is largely acknowledgedthat site-specificfactors influenccd the expansionof alderrather than climatic amelioration(Birks, 1989;Chambers and Elliott, 1989;Bennett and Birks, 1990;Tallantire, 1992).Birks's (1989)pollen isochronesfollow a trendindicating that Ahms was present in substantial amountsin westernScotland and the Inner Hebridesby c. 6500BP (5460cal. BQ. Alnus wasprobably present in the OuterHebrides from this time. Ablus is recordedas macrofossil wood from Lewis andHarris (Fossitt, 1996),and it is presentin mostpollen profiles from the OuterHebrides in minor amounts (Birks andMadsden, 1979; Bolincke, 1988;Bennett et al., 1990;Fossitt, 1996;Whittington andEdwards, 1997). Nevertheless, it appearsto havebeen only a minor componentof woodlandand was possibly more widespread on SouthUist (at Loch Lang,Bennett et al.(1990) and Loch Phuinnd,Fossitt
(1996))from c. 6000BP (4910cal. BC) than elsewherein the OuterIsles.
Thesesites are situated on the eastof the islandand are relatively sheltered from prevailingwesterly winds. Many partsof the OuterHebrides do not havepollen recordsand Alnus may havehad a morewidespread distribution thanis currentlyrecogniscd.
At somesites in the Inner Hebrides,Alnus expandsearlier than pollcn isochronessuggest. The early expansionof Alnus occursc. 8000BP (6860
cal. BC) at Drimnagallin the north of the Kintyre Peninsula(Rymer, 1984), at Rhoin Farmalso on the Kintyre Peninsulaat 6910±90BP (5830cal. BQ (Edwards,1990) and at Moine Mhor, nearOban, c. 7000BP (6010cal. BQ (Haggartand Sutherland, 1992). At Loch an t'Suidhe,western Mull, Ablus
expandsc. 6800BP (5720cal. BC). Alder is recordedas pollen and macrofossilsin archaeologicalcontexts from tile FarmFields site at Kinloch on Rum from c. 7900BP (6770cal. BQ indicatingan early local presence
193 (Hironsand Edwards in Wickham-Jones,1990), although its expansionin thc Kinloch monolith profiles is not until 6430±90BP (Edwardsand Hirons, 1990).
The Inner Hebrideansitcs referred to aboveare all locatedin fairly coastal positionsand may haveexperienced a degreeof groundwatersalinity and waterloggingearly in the Holocene.Ahius is ableto tolcratcsaline conditions (Chambersand Elliott, 1989)and high levelsof salinity could havercstrictcd the growthof otherarboreal taxa at the samesites with Abutsemerging as the successfulcompetitor.
Associationsbetween the expansionof Abutsand human activity havebeen
inferredfor somesites including Kinloch andRhoin Farrn(Edwards, 1990)
whereincreases in microscopiccharcoal occur at the main Ahmsrise. McVean(I 956a,1956b) associated the spreadof Alnus ghtfinosawith
clearanceof othervegetation by burning,which may also haveincreased runoff andnutrient inputs to low-lying areas(Moore, 1986).Although thcrc is
no increasein charcoalat the Ahmsrise at Loch an t'Suidhe,the expansionof Alnus occursshortly after a phaseof woodlandreduction linked to possible
anthropogenicactivity, andhigh levelsof charcoaland increases in Calluna may indicatethat areasof the catchmentwere burned. This may haveled to increasedwaterlogging and peat formation in the valley floor (Chapter4) thus
assistingthe spreadof Ahms.Increases in Alnus in the centralLABI and LABVII profiles at Locha!Bhogaidh also occur after a prolongedperiod of
inferredanthropogenically-induced woodland reduction and coincide with indicationsof peatformation at the loch edges.
In contrast to the above, there are strong indications of burning in the
A! Chrannagprofiles from Ulva, but the Alnus expansionis dated to c. 5700
194 BP (4530cal. BQ, althoughthe datemay not be correct.Following its expansion,pollen influx valuesindicate that alderhad a stronglocal distribution.It appearsto havepopulated the potentiallywetter andmore
shelteredareas of A!Chrannag bog, andin the drier partsof the bog to the north,birch becamepredominant. The Livingstone'sCave profiles indicate that alderwas present in substantialamounts in the basinitself, which was probablysubjected to extremesof waterloggingaccording to the preceding. stratigraphyand pollen assemblages.This suggeststhat Ahms may also be
stronglyinfluenced by local hydrologicalconditions and by competitionfrom existingvegetation. Although anthropogcnicinfluences can occasionally result in conditionsfavourable to the expansionof Alnus, the possiblelinks at most sitesremain enigmatic, and where these are suggested,an interpretation
of naturalvegetation succession due to changinghydrology and ccological dynamicsmay be equallyas valid.
7.4 The distribution of human activity as suggestedin pollen diagrams DetailedHolocene vegetation histories from the west andnorthwest coastal
areasof Scotlandare few andthose published lack resolution,microscopic
charcoalcounts and radio-carbon dates. Fluctuations in the Corylusand Betula curvesof a diagramfrom Aros Moss,Kintyre could result from Mesolithic activity, but moreconvincing evidence of humanimpact occurs in
the later partsof the profile andsuggests Neolithic clearanceand cereal
cultivation (Nichols, 1967).Unfortunately there arc no datesattached to the
profile to confirm thesehypotheses. Early cereal-typepollen is recordedfrom
pre-elmdecline deposits from Rhoin Farm,Aros Moss,Kintyre (5640+80BP [4480cal. BC], Edwardsand McIntosh, 1988)and from GallanachBeg,
southwestOban (undated, Rhodes et al., 1992)which could indicatepioneer cultivation (Edwardsand McIntosh, 1988;MacDonald and Edwards, 1991). Fluctuationsin the earlyHolocene Calluna profile from An Druim,
195 Sutherland(Birks, 1980)may reflectMesolithic iilflucnccs oil licatliland,but thereis no charcoalrecord from the site, andin any case,natural cycles linked to exposureand climate of high latitudesmay haveinfluenced expansionsin Callunaheath. Other than the above,there are no published pollen recordsfrom the westcoast of Scotlandwhich indicatehuman activity, althoughthis is probablydue to the lack of high resolutionand radiocarbon datedinvestigation rather than an absenceof people.Pollen profiles from near coastalsites such as Loch Sionascaigand Loch Clair containincreasing amountsof Callunavulgaris, Plantago lanceolata, Polvailla andother herbs from asearly asc. 9000BP (8190cal. BC) (Birks, 1980)probably as a result of their exposedlocations. The removalof woodlandby peoplecould also result in increasesin thesetaxa, but humanimpact will remainundetected as
a resultof this equifinality.
In contrast,there are many indications of humanactivity in the Inner Hebrideanpollen records,although some episodes of woodlandreduction
which may oncehave been interpreted as human impacts may havebeen largely influencedby climatic changes.Mesolithic activity may nevertheless
haveexacerbated woodland reduction and erosion during this period of
climatic cooling,particularly given the antecedentevidence for possible Mesolithic impactsand the archaeologicalevidence for Mesolithic occupation
at somesites. Mesolithic andlater Neolithic impactscould appearin all the
pollen diagramsfrom this studyand are inferred as ranging from small scale clearancesin woodlandto reductionsof woodlandover a wide areawhere burningto createCalluna heath and grassland may haveoccurred. Cereal
cultivationmay havetaken place on Ulva from c. 4600BP (3540cal. BQ in andon Rum from c. 3950BP (2440cal. BC). Cerealpollen appears the LABI profile from Loch a'Bhogaidhshortly after c. 3500BP (1880cal. BC). Otherpollen recordsfrom Arran (Robinsonand Dickson,1988; Edwards and
196 McIntosh, 1988),Colonsay (Andrews et al., 1987)and Newton,Islay (McCullagh,1991) display evidence of humanactivity andthere arc indicationsof similar eventsin pollen recordsfrom the Outer11cbridcs (e. g. Edwards,1997a; Callanish,Bohnckc, 1988). Loch an t'Sil, ,
Figure 7.3 summarizesthe indications of human impact for the Holocene to c.
3500 BP (1880 cal. BQ at sites from the Inner and Outcr Hebrides. For conveniencethe transition from the Mesolithic to the Neolithic is placcd at c.
5000 BP (3810 cal. BC) but it is acknowledgedthat the cultural change was probably diffuse and may have been initiated earlier or later than this in
different areas.
7.4.1 The Mesolithic
It is evident that most of the evidence for Mesolithic activity appearsin the
Inner Hebrides, notably at Loch a!Bhogaidh, Loch an t'Suidhe, Loch Cholla,
A'Chrannag and Kinloch. The earliest possible impacts are recorded at Loch
a!Bhogaidh from c. 8500 BP (7540 cal. BQ and take the form of small
reductions in birch-hazel woodland, although the lack of microscopic
charcoal in the profiles suggeststhat at least some of thesecould be of natural
origin. A much longer phaseof woodland reduction occurs at Loch is a'Bhogaidh from c. 7700 - 7000 BP (6600 - 5780 cal. BC). Although this
probably a reflection of a colder, dry period, there is an indication that
Mesolithic activity may also have influencedvegetationdianges at tile site,
particularly from c. 7300 BP (5910 cal. BC). A phaseof woodland reduction
of similar antiquity and probably of similar origins, occurs at Loch an
t'Suidhe, although in addition to increasesof Poaceaeand Cyperaceacat the
expenseof birch-hazel woodland, Calluna vulgaris also increases,possibly
arising from fires set by Mesolithic people as indicated by increasedCli: P and
charcoal influx. At Loch Cholla, a reduction in birch-hazel woodland is of
197 longerduration (c. 7500- 6000BP [6320- 4910cal. BQ thanat Loch a!Bhogaidh and Loch an t'Suidhe,but this may be due to the largespacing of radiocarbondates on the core leavingchanges in sedimentationand possible hiatusesundetected. The stratigraphyof the Loch Cholla profile for this period is silty peat,rather than gyttja, andimpact here may havebeen more localisedthan at Loch a!Bhogaidh and Loch an t'Suidhe(both lakes),and may thereforehave registered in the pollen recordfor a longerperiod.
The increasesin Callunavulgaris in the Loch an t'Suidheand Loch Cholla profiles may be the resultof differentvegetation succession or the proximity of activity to the coresites. Calluna vulgaris pollen doesnot travel far, and human from c. 7700- c. 7300BP (6190- 5780cal. BC) at Loch a!Bhogaidh, activity may havebeen some distance from the loch edge,thus Calluna increasesmay remainundetected. However, the comparativelack of
microscopiccharcoal in the Loch a!Bhogaidh profiles suggeststhat tile direct burningof vegetationdid not occurextensively there. Charcoal increases in
the laterprofiles from c. 7300- 7000BP (6190- 5780cal. BC) may result from local campfires, andproximity of theseto the locli edge.In contrast,
burning appearsto havebeen sustained at Loch an t'Suidheand was possibly
widespreadin orderto produceincreases in Calluna.By implication,burning
of vegetationmay alsohave taken place at Loch Cholla and similar practices may havetaken place at A'Chrannagon Ulva, ashigh levelsof charcoalfrom
c. 8500BP (7540cal. BC) correspondwith increasesin Calluna and grasses. Burning may alsohave maintained an abundanceof Coryhispollen in the
Ulva profiles by effectivelycoppicing hazel (cf. Edwards,1990) and reducing the spreadof otherarboreal taxa. High levelsof charcoaland Calluna at tile baseof the MachrieMoor profile c. 8700BP (7790cal. BC) may alsohave
resultedfrom similar Mesolithic activity (Robinsonand Dickson, 1988).
198 In the OuterHebrides there is comparativelylittle evidenceof potential
Mesolithic impacts,particularly on the scaleseen in the Inner Ilcbridcan profiles.Two relativelyshort pcriods of woodlandreduction occur at Loch an t'Sll, SouthUist, c. 8000BP (6910cal. BC) and 7800 BP (6640cal. BQ, lastingc. ISOyears and 70 radiocarbonyears respectively. Both of these indicatethat localiscdburning produced increases in grasses,herbs and Calluna(Edwards et al., 1995;Edwards, 1996). Increases in hcrb-rich grasslandand Callunaheath occur at Reincval,South Uist, andthese appear to be linked to increasesin microscopiccharcoal. This could be evidenceof humanactivity, but lightning strikeswould producethe sameeffects, and the openenvironment and exposure to westerlyrainstorms could havereduced treecover andincreased bog andheathland growth. A declineof woodland andthe developmentof heathand machair grassland occurs at Borvc on Benbeculac. 6200BP (5130cal. BQ, andincreases in microscopiccharcoal andPteridium at this transitionled Whittington andEdwards (1997) to suggestthat Mesolithicpeople may havebeen responsible, at leastin part, for this transition
Only Callanish,Lewis providesany firm indicationof substantialMesolithic activity in the OuterHebrides. Here fluctuationsin the birch-hazelwoodland arematched by increasesin charcoal,grasses, herbs and Calluna.Dates indicatethat this may havestarted c. 8300BP (7330cal. BQ andalthough
intermittent,this disruptionof woodlandat Callanishcontinues through the
Mesolithic to the earlyNeolithic at c. 4500BP (3140cal. BC).
It is possiblethat little evidencefor Mesolithic impactsis recordedin the OuterHebridean profiles dueto the lesseramounts of woodlandfor many
sitesin the islandscompared to thoseof the Inner Hebrides,and the potential for naturaldisturbances in an exposedand fragile ecosystem.Indeed, the
199 pollen indicatesthat in manyareas of the OuterHebrides, birch-liazcl woods declinedfrom c. 6000BP (4910cal. BC) followed by the developmentof blanketbog and Callunaheath over many areas(Birks and Madsdcn,1979;
Bennettet al., 1990;Edwards et al., 1995;Fossitt, 1996;Whittington and Edwards,1997).
If all the disturbancesin the pollen diagramsreferred to aboveand in Figure
7.4 arethe resultof humanimpact, then the evidencepoints to the widespread useof fire during the Mesolithic to initiate vegetationchanges. It is arguable asto whetheror not this usewas intentional,particularly in caseswhere woodlandreduction and recovery occurs in a short spaceof time. It is not unlikely that the regenerationof treesfollowing a single firc may take 70 -
100years, such as is indicatedat Loch an t'Sll, particularlyin a marginal
environmentwhere pressure from grazinganimals may takeits toll. Charcoal may alsoderive from the secondaryburning of felled treesand may or may not be natural.However, it may be arguedthat sustainedburning over a periodof c. 800 yearsas is indicatedat Loch an t'Suidhe,if of anthropogcnic
origin, is unlikely to result accidentally.Microscopic charcoal records on Ulva alsoindicate that fires occurredover lengthyperiods of time. Thus there
is everyindication that Mesolithicpeople in the Inner andOutcr Hebrides could haveintentionally used fire as a meansof modifying their
environments,possibly to creategrazing for game,increase fowl, andperhaps to increaseareas of berriesand rhizomous plants for consumption.
7.4.2 The Neolithic
At many of the sites which bear evidenceof Mesolithic impacts, there are
indications that vegetation disturbancescontinued through the Neolithic. At
A! Chrannagon Ulva it appearsthat similar cycles of burning linked to the
200 propagation of Calluna heath continued until at least c. 4200 BP (2790 cal.
BQ, but there is additional evidencein this and other profilcs on Ulva that woodland clearanceand cereal cultivation occurred frorn as early as c. 4400
BP (3120 cal. BQ. Intermittent reductions in trees combined with incrcascs in grasses,and herbs such as Poictitilla and Plantago lanceolata continuc to the top of the sampledprofiles AC2, LC I and LC2.
The extensiveopening of woodlandat Kinloch, Rum occursfrom c. 3950 BP (2440cal. BC) andappears to havebeen at leastpartially causedby human
activity, particularlyas cercal pollen andincreases in Plantagolanceolata andBrassicaceae amongst other herbs, are detected in the four proriles from the site. LABI at Loch a!Bhogiadh shows ccreal-type pollen andthe opening
of woodlandfrom c. 3590BP (2440cal. BQ.
In the OuterHebrides, indicators of openenvironments such as grasses and herbs,including Plantago lanceolata, increase from c. 4000 BP (2530cal.
BQ (Birks andMadsdcn, 1989; Newell, 1988;Bennett el d, 1990;Fossitt,
1996)which could suggestwidespread grazing of domesticanimals and increasedland usepressures. However, increased occanicity in the Atlantic
from c. 4000 BP (2530cal. BC) (Hirons andEdwards, 1990) may also have
playeda part in changingvegetation.. Nevertheless, the presenceof Ncolithic
peoplein the OuterHebrides is confirmedby numerousarchaeological rinds (cf. Newell, 1988)and stone monuments, particularly the Callanishstone
circle (Hcnshall,1973). A corecloscby provided indications of a considerable
periodof Mesolithic impactsand cvidcnccfor humanactivity in the profilc resumcsc. 4000BP (2530cal. BC). This includesincreases in microscopic charcoal,herbs including Plantago lanceolata, Trifolitins and Brassicaccac,
andccreal-type pollen (Bohnckc,1986). This, andevidence from elsewhere,
201 seemsto confirm that Neolithic fanning took placein the OuterIsles (Barclay, 1997;Mulder, 1999).
7.5 The Ulmus decline
There are few pollen profiles from the Inner Hebrides and west coast of
Scotland which document the Ulnius decline in detail. Reliably dated pollen profiles featuring the Ulmus decline are from Moorlands, Machric Moor,
Arran (Robinsonand Dickson, 1988;Edwards and McIntosh, 1988)and Rhoin Farm,Aros Moss,Kintyre Peninsula(Edwards and McIntosh, 1988).
Securelydated Ulmus profiles for the Inner Hebridesare limited to the sites discussedin earlierchapters and listed in Table7.2, whereAMS datesapply directly to Ulmusdecline levels, or interpolateddates can be applied.Sites
from Skyewith interpolateddates for Ulmus declinesare Loch Meodal,Loch
Cleat andLoch Ashik (Birks andWilliams, 1983).There are no Ulnuis
declinesin profiles from Kinloch, Rum (Hirons andEdwards, 1990) or Loch Cholla,Colonsay (Andrews et aL, 1987)where Mims pollen is recordedin
very low quantitiesand this is alsothe casefor the OuterHebrides.
Table7.2 lists the Ulmusdeclines and Figure 7.4 showsthe spreadof dates for these.Error barsare at a rangeof 2 and I StandardDeviations in
respectivegraphs where actual dates are available.
Although a rangeof datesmay occurdue to the statisticalerrors inherent in the datingof pollen profiles andthe lack of accuracyin defining the actual
point of declinein the pollen diagram(cf. Hirons andEdwards, 1986), there is neverthelessmore than one phase of Ulniusdecline with multiple declines declines occurringin someprofiles. Multiple or single later Mats arc not limited to the Inner Hebrides(Ulva, Arran, Islay) andoccur in pollen diagramsfrom northeastIreland (Hirons and Edwards,1986), Wales (Smith
202 andCloutman, 1988) and Fife, easternScotland (Whittington et A, 1991) amongstothers. All of theseare multiple profile studieswhich bolstersthe reality of the phenomenon.
Thereare various hypothesized causes for the declineof Ulmuspollen in Britain andIreland. These incorporate climate change, erosion and soil deterioration,pathogenic attack and human impact (e. g. Pilcheret al., 1972; Garbett,1981; Groenman-van Waateringe, 1983; Girling andGreig, 1985;
Hirons andEdwards, 1986; Peglar and Birks, 1993;Digerfeldt, 1997).It is largely acknowledgedthat a combinationof factorscould be reponsiblefor thesedeclines and that different factorsmay operateat different sites.
7.5.1 Climatic influences and the Ulmus decline
Climate changecharacterised by increasedrainfall in the British Isles at c.
5000 BP (3810 cal. BC) is frequently cited as a reason for the classic elm decline. However, in Scotland and Ireland a Pinus decline initiated in some areasas early as c. 4800 BP (3570 cal. BC) and peaking c. 4200 - 3800 BP has been increased (2740 - 2250 cal. BQ attributed to rainfall and substrate waterlogging (Bennett, 1984; Dubois and Ferguson; 1985; O'Connell et aL,
1986; Whittington and Edwards, 1997; Willis et aL, 1997). There are simultaneousreductions in Pinus and Ulmus c. 4800 BP (3570 cal. BC) recorded in published pollen diagrams from northeastIreland (Hirons and
Edwards, 1986), Black Loch, Fife (Whittington et al., 1991) and Machrie
Moor, Arran (Robinson and Dickson, 1988) and in the A! Chrannag and
Livingstone's Cave profiles in this study. As both taxa could have responded to the sameenvironmental stresses,it is extremely plausible that climate changehad an adverseeffect on Ulmus populations. However, the recovery of
Ulmus in someprofiles and the variable timings of the Ulmus declines
203 indicatethat the effectsof climatewere not immcdiatcor direct.Factors such aslocal differencesin soils, exposure,ecological competition, disease and humanimpact may havemoderated or exacerbatedclimatic cffccts.In some instancesUlmus persists until until much later (e.g. 3590±100 BP [ 1940cal.
BC] at Loch a'Bhogaidh;3600 BP [ 1940cal. BC] at LC I, Livingstone's Cave).Pinus has also been known to surviveinto the historic period in lowland andsheltered locations of northernIreland (Bradshaw and Browne, 1987).This suggeststhat in someinstances, site-spcciric factors may outweighclimatic influenceson sometaxa (cf Smith and Cloutman,1988; Whittingtonet al., 1991;Edwards and Berridge, 1994).
7.5.2 Soils and site location
The effects of soils and site location were mentioned earlier in connection with the initial expansionof Ulmus. It is not surprising therefore that the earliest pennanentreductions of Ulmus occur at exposedlocations such as
A'Chrannagand Loch an t'Suidhe(refer to Fig. 7.4; Table7.2). Soils here may havebecome eroded and leached relatively early dueto exposureto
Atlantic stormsand peat formation occurred early in the Holocenefrom c. 8500BP (7470cal. BC) at A'Chrannagon Ulva, andc. 6800BP (5700 cal.
BC) at Loch an t'Suidhe.A recoveryin Uhnusdemonstrated by multiple declinesat somesites including those from Livingstone'sCave basin, Ulva, suggeststhat the initial declinesof this taxoncould be the result of episodes of diseaseor reducedflowering broughtabout by environmentalstresses. Most of the siteswhere Ulmus recovers from early declinesor wherea single declineoccurs later thanc. 5000- 4700BP (3810- 3460cal. BC) are shelteredor positionedinland where topography is not extreme.These factors may meanthat soils therewere lessprone to waterlogging,Icaching and erosion,thus allowing Ulmusto recoverfrom the initial impactsof climatic deterioration.
204 7.5.3 Human impact and the Untus declines
Indicationsof humanactivity includingcercal-type pollen and increasesin herbsand microscopic charcoal coincide with or directly prcccedthe U1111113 declineat somesites in Scotlandand northern Ireland (e.g. Pitcheret aL, 1972;Hirons andEdwards, 1986; Edwards and McIntosh, 1988;Robinson andDickson, 1988;Whittington et al., 1991).In the pollen dataresulting from this study,the earlier Ulmusdeclines at A'Chrannagand Livingstone's Cave,Ulva andLoch an t'Suidhe,Mull do not coincidewith substantial evidenceof humanactivity. Increasesin microcopic;charcoal in AC I at A'Chrannagc. 4600BP (3440cal. BC) probablyrelate to the burning of heathlandand occur after the main Ulmusdecline.
As Neolithic land useintensified, then elm populationsalready under pressurefrom the effectsof climateand soil deteriorationwould be placed undergreater pressure. The later andsecondary Ulmus declines in the Inner Hebrideanpollen recordmay thereforebe more closelylinked to human impactthan earlieroccurrences, and from c. 4000 BP (2530cal. BC) thereis a broadrange of evidencefor cerealcultivation andgrazing. Cereal-type pollen occursin somepollen profiles at or beforethe Uhnusdeclines (Livingstone'sCave, Ulva; Loch a'Bhogaidh,Islay; Rhoin Farm,Aros Moss,
Arran) andcarbonized cereal remains were foundin a Neolithic context in Livingstone'sCave, Ulva (Bonsallet al., 1991).Herbs and Calluna expand in manypollen profiles, andincreases in Plantagolanceolata may result from the pressureof grazingof domesticanimals. Microscopic charcoal rises with theselater Ulmusdeclines, but may result from an increasedfallout from domesticfires linked to morewidespread occupation and increasing populations.In the later Ulmusdeclines from most sitcsreductions in other arborealtaxa areevident, representing the openingof woodlandas cultivation
205 andgrazing increases. Turner et A, (1993)find asynchronousand rclativcly lateelm declines in profilesin a smallarea from the North York Moorswhich areattributed to the responseof elm to humanimpacts and discase.
7.5.4 Pathogenic attack
Diseasehas frequently been cited as a possible mechanism for the rcduction of Ulmus in pollen spectrafrom the British IsIcs. Finds of the elm bark bcetle
(Scolytusscolytus) in elm decline deposits at HampsteadHeath, London Icd
Girling and Greig (1985) to proposethat Dutch elm disease(carried by the beetle) could be responsible for the classical clm decline of c. 5000 BP (4010 cal. BQ. However, Scollylus scolylus only attacks trccs already under strcss, and if Dutch elm diseaseor other pathogensinfluenced the UIMUSdecline, then thesewere probably secondaryto the effects of climate, soil deterioration and possibly human activity.
7.5.5 Summary of evidencefor the Uhnits declines
Although the Inner Hebrideanpollen profiles with datedUhnits declines arc limited in number,the problemsof the identificationof possiblecauscs and the complexitiesof interpretationare evident. Multiple profiles from Ulva indicatethat exposureand local ecologicalfactors may havehad a greater influenceon Ulmusthan previously suggested. The widely differing datcsfor the Ulmusdeclines in A!Chrannag and Livingstone's Cave profiles on Ulva demonstratehow multiple profiles canbe usedto detectnumerous Ulmus declines.It may thereforebe usefulto extendall the Loch a!Bhogaidh profiles to coverthe Ulniusdeclines, and obtain dates for these.Currently, only LABI from Loch a!Bhogaidh shows an Unius decline.
206 Althoughthe rangeof datesattributed to the clm declinesat diffcrcnt sitesin the Inner Hebridesmay partly be dueto inaccuraciesin radiomctricdating andstandard deviation, there is neverthelessa suggestionthat local environmentalinfluences, which may includeexposure, soil type andhuman impact,were all significantin affectingUlmus in additionto climate.As
Whittingtonet aL, (1991:85) concluded:
"a myriad of local factorscould be evokedto explainthe behaviourof c1m pollen curves."
The variationin the timings of the Ultnusdeclines and the detectionof multiple declinesat many sitesmeans that referenceto'the elm decline'asa datinghorizon and an indicatorof changingprehistoric cultures may be misleading(Whittington et al., 1991).The arrayof datesfor the Mills declinesin the InnerHebrides and the differentpatterns in the Mills profiles demandmore detailed records for this taxon.Implicit in this is the necessity for moremultiple profile studiesto assessthe intra-siterepresentations of
Ulmuswhich may providea clearerindication of the causeor causesof the elm decline.
7.6 Summary of pollen records for the Inner Hebrides, western Scotland,and the Outer Hebrides
The pollen datadiscussed above indicate that a rangeof vegetationtypes was presentat varioustimes in different locationsin the west of Scotlandand its islands.In the Inner Hebrides,human impacts on vegetationmay have occurredfrom as early asc. 8500BP (7470cal. BQ andmuch of the dated evidencefor humanactivity coincideswith datesfor the occupationof archaeologicalsites. There is also evidencefor Mesolithic activity in the OuterHebrides, although this may be lessconvincing, but little from siteson the mainland,probably due to the lack of detailedpollen diagrams.Neolithic
207 impactsappear to be morewidespread throughout the region,although tllcsc aremore easilydetected in pollen profiles thansmall scale,transient Mesolithic disturbances.
The differentchronologies for the expansionof variousarboreal taxa in the regionchallenges current perceptions regarding their early Holoccne behaviourand the datahighlight the complexecological factors which may haveinfluenced changing patterns of vegetationin the rcgion.
208 CHAPTER 8- REVIEW OF METHODS AND IMPLICATIONS OF RESULTS FOR FUTURE RESEARCH
8.1 Review of methods
Chapter 2 detailed the methodologiesto be employed in undertaking this researchand highlighted the importance of multiple profiles in palynological investigations. Implicit in this is the employment of close resolution sampling and radiocarbon AMS dating.
The quantification of microscopic charcoal, assessmentof damagedpollen types and scanningfor cereal-typepollen were also advocatedas useful tools for palaeoecologicalinvestigation. The inclusion of the above methods allows an assessmentof the usefulnessof eachin the context of this and future studies.
8.1.1 Multiple profiles
Multiple profile studies in the Inner Hebrides were previously limited to Loch from a!Bhogiadh, Islay and Machrie Moor, Arran, but two of the three cores Loch has been a'Bhogaidh were undated and no synthesisof the Machrie Moor profiles Rum four published. Additional profiles from Loch a!Bhogaidh, Islay, Kinloch, and in islands. profiles from Ulva increasethe range of multiple profile studies the
Similaritiesand differences in intra-siteprofiles for the sametaxon (in percentageand influx terms)can assistin explainingthe behaviourof a taxon,and may be useful in
detectinganomalies within individual profiles.This is demonstratedby the Loch
a!Bhogaidh profiles, where an earlier Corylusrise is suggestedby the centralprofilc, LABI, comparedto surroundingprofiles which often lay closerto the loch margins. Althoughthe basalradiocarbon dates from LABI are subjectto obviouserror, the
differentcharacteristics of the early LABI Corylusprofile comparedto the othersfrom
the site indicatethat LABI receiveda greateramount of Coryluspollen from a wider
209 region.The LABI profile may thereforereflect the dynamicsof Cor),his prior to its actualexpansion at Loch a!Bhogaidh.
Although A'Chrannagbog andLivingstone's Cave bog, Ulva are differentbasins, these displaymarked differences in taxonassemblages, and the differential behaviourof the respectiveUlmus curves between the two basinsat what could be similar times (e.g. c. 5 100BP [ 4010cal. BC], c. 4800BP [3570cal. BC], c. 3800BP [2220 cal. BC]) demonstratesthe potentialfor multiple profiles asan aid to determiningtaxon behaviour.
It was suggestedin Chapter 2 that multiple profile studies may assist in determining whether featuresin the pollen profiles were of natural or human origin. Although in many instancesinterpretation remains speculative,such as for the small reductions in
Corylus in the Loch a!Bhogaidh profiles prior to c. 7800 BP (6650 cal. BQ, in other casesinterpretation of featuresmay be enhanced.This is certainly the case for the main Loch from phaseof Corylus reduction at a!Bhogaidh c. 7700 - 7000 BP (6600 - 5870 cal. BC), which appearedless clearly in the central profile LABI due to its central position, greater depth and lower sampling resolution. Although Edwards and Berridge
(1994) mentioned the possibility that this could be a reflection of Mesolithic impacts, this is now inferred as reflecting a period of climatic cooling (Alley ct al., 1997; Barber et al., 1999; Willcmse and T6mqvist, 1999), although multiple profiles indicate that somehuman impact may have exacerbatedthe effects of climate in the lattcr part of the climatic event.
The final episodeof woodlandreduction at Kinloch is confirmedas being largely the resultof humanactivity, asmultiple profiles indicatea progressionfrom castto wcst, startingclose to the known site of humanoccupation.
210 There are two aspectsto the potential for rcprcscntationsof spatial variation in vegetation offered by multiple profilcs. Differences may occur between the overall representationsof individual taxa, such that percentages,concentrations and influx rates for a single taxon are markedly different between profiles. The other aspcct is that different representationsof single featuresof disturbancein the profiles may be detected.
In the two profiles from A'Chrannagbasin, Ulva, therearc distinctly different percentage,concentration and influx representationsin somearboreal taxa which suggeststhat very differentvegetational communities occurred in different partsof the bog andsurrounding area. For most of the sampledACI profile, Bendais dominant andexceeds 80 % TLP. For the comparablepart of the AC2 profile, Benda avcragcs30 % TLP, but Alnus reachespercentages of 100% TLP in AC2 comparedwith c. 30 % TLP in ACL Concentrationsand influx ratesfor thesetaxa display similar patternsto the percentageprofiles.
In contrastto the above,the Loch a!Bhogaidh percentage and absolute profiles have uniform representationfor the major arborealtaxa, especially and predominantly Corylus.The predominanceof Corylus,particularly in the early partsof the profiles, may be partially responsiblefor this by reducingthe pollen countsof other taxa to backgroundlevels which may be fairly consistentin percentageterms over long periodsof time. Sedimentaccumulation may alsohave contributed to the homogenisationof the fossil pollen assemblagewithin the lakebasin. Nevertheless, theredo appearto be both small andlarge scale disturbances in the Corylusand other
taxoncurves from Loch a!Bhogaidh and these are differentially representedbctwccn the profiles.The Corylusreduction which occursbetween c. 7700BP andc. 7000 BP Loch Bhogaidhis (6600- 5840cal. BQ at a! of particularnote.
211 The profiles covering this event indicate that the source of the Corylus reduction changedfrom the northwest to the southeastof the basin at c. 7300 BP (6190 cal. BC) and in this context spatial variation in the Corylus and other profiles is apparent.At
Kinloch, Rum, a reduction in woodland c. 3900 BP (2370 cal. BC) was initially greatestto the cast of the mire, close to the known site of human occupation, and woodland may have persisted for severalhundred years later than this to the wcst of the site. However, the lack of radiocarbon datesin the profiles restricts confidence in this interpretation, and highlights the need for the application of AMS dates in multiple profile studies.
On Ulva, the profiles from two different basins indicate that approximately I krn to the north of the known site of human occupation, burning to promote the continued renewal of heathland could have occurred betweenc. 4400 and c. 4000 BP (3120 and
2530 cal. BC). Nearer the site of occupation, the pollen indicates that small scale woodland clearancesoccurred within the sametime period, probably to increase grazing for livestock and allow cereal cultivation which is also indicated in the pollen profiles and by the presenceof cereal macrofossils at the occupation site.
Theseexamples illustrate the rangeof variationin vegetationwhich may be detected throughmultiple coring, both in lake sedimentsand peat. There is no reasonto suggest that multiple profiles from lake sedimentsare any lessuseful than those from peat(cf. Jackson,1994). Overall, multiple profiles appearto be very effectivein reconstructing vegetationhistories, their variations,and differential behaviour of a singletaxon in severalpollen curvesmay assistin identifying its ecologicalrequirements in tile early Holocene.
8.1.2 AMS dates
Radiocarbondates are obviously neededin the context of multiple profiles to ensure crosscorrelation, but the absenceof well-datedprofiles for manypollen corcsfrom
212 westernScotland and the Inner Hebridesmakes comparisons difficult (Chapter7). AMS datesare preferable to enablethe correlationof profiles which haveclose samplingresolutions, as thinner, smaller samples are taken for AMS dating.
The absenceof a suiteof AMS datesfor eachprofile from Kinloch, Rum meansthat the interpretationof a progressiveeast-west woodland reduction from c. 3900BP
(2370cal. BQ is basedon assumption,and more dates on all the profiles would strengthenthe argumentsfor this.
In some of the profiles with many AMS dates,reversals in dates (at AC I, Ulva) and what appearto be erroneousdates (at LABIII, Loch aBhogaidh), are problematic, but well-dated multiple profiles have assistedin detecting and explaining these. For example, dateson LABIIIc at Loch a!Bhogaidh appearto be too early when compared to other dated profiles from the site, but would have appearedreasonable had this been the only core. Similarly, had only a single profile from Livingstonc!s Cave basin been dated, then one of the Holocene basal dateson either of the cores would probably have been consideredacceptable and used in interpretation.
8.1.3 Microscopic charcoal quantification
Microscopic charcoal quantification is an important element of palacoccological investigation, and charcoal concentrations,Ch: P and, where possible, charcoal influx were calculated for each of the profiles. The different methods of quantification provide results which would lead to similar interpretations. Although at Livingstone's
Cave bog on Ulva, charcoal influx is extremely low in comparison to concentrations and Ch:P, it neverthelessshows the samedeclines and increasesas these.
Although the continued presenceof microscopic charcoal in most of the profiles suggestsa continued fallout of particulates - probably derived from f ires occurring over a wide, regional area- peaks in charcoal within the profiles probably reflect local fircs.
213 This is certainlyindicated by profiles from Ulva, whereincreases in charcoallinked to fires in heathlandappear in AM in the A'Chrannagbasin, but not in the corresponding partsof AC2 from the samebasin, or in correspondingparts of profiles from
Livingstone'sCave basin. If this charcoalwas derived from an increasein f ircs over a wide region,then increases in charcoalin all profiles could be expected.
Microscopiccharcoal analyses in the Loch an t'Suidheprofile also enhance interpretationof eventsrecorded in the pollen curves.In particular,large incrcascs in b) charcoalare recorded in the new profile; whereasin the original Lowe and Walker (1986aand in profile, charcoalis not recordedalthough a reductionin arborealpollen andexpansion grasses is detecteddespite the low sampling Although the episodemay -resolution. reflectvegetation responses to a colder,dryer climate(Alley et al., 1997)and the increasedcharcoal could be the resultof increasednatural fircs in a dry climate,the
charcoalincreases may alsoreflect a humanpresence. Spatial and temporal changes in
the microscopiccharcoal profiles at Loch a!Bhogaidh coincide with changesin the
pollen curvesfor a similar periodto that at Loch an t'Suidhe,but suggestthat human
actioncould havecontributed to a reductionin arborealvegetation at this site. Archaeologicalevidence suggests a strongMesolithic presencein the Inner Hebrides,
thusthe increasesin microscopiccharcoal between c. 7900BP andc. 6850BP (6810
and5870 cal. BQ at Loch an t'Suidhecould alsohave arisen from humanactions as well asarising from naturalevents.
Although the taphonomy of microscopic charcoal is not fully understood(Section
2.1.2), when combined with pollen data, microscopic charcoal aids interpretation to
some extent. Although a persistentprescricc of charcoal is usually recorded, peaks in
the microscopic charcoal curves appearto representlocaliscd fire events in many cases,
and often reinforce or cnhanceinterpretationsofthe pollen profiles. Also, the more
frequently microscopic charcoal is quantified in palaeoccologicalanalysis, the more
214 informationmay beacquiredaboutits taphonomyand the role of fircs in past ecosystems.
8.1.4 Damagedpollen analysis
The resultsof the identificationand quantification of damagedpollcn typcswcrc includedin the profile interpretationsin previouschapters. Tile type of damagemost frequentlyrecorded was that of pitted grains.Thinned pollen was scarcelydctcctcd, andbroken and crumpled grains occurred mainly in pollen typesof high susceptibility to suchdamage. For example,Pinus pollen is mainly brokenand contributes to a high percentageof the total brokenpollen of eachsample. Crumplcd pollen, (given the exclusionof Cyperaceaefrom this category)is representedmainly by Poaccaewhich crumpIcseasily.
Althoughtotal damagedpollen percentages and the ratio of damagedto undamagcd pollen increasein mostcases when reductions in arborealpollen occurand erosion may be indicated,given the lack of variationin damagetype, thereis no evidenceto suggestthat onetype of damagehas a specificcause. Pitting is normally ascribedto aerobicoxidation and fungal actionwhen a grain is exposedat the sedimentsurface prior to burial. It is possiblethat long distancetransport may exposepollen grainsto damageof this nature.There appears to be a generallevel of'backgroundnoise' for damaged,particularly pitted, pollen recordedin eachprofile, andnormal taphonomic processeswill result in a proportionof pollen beingdamaged. Nevertheless, damaged pollen quantificationmay be usefulto ascertainwhether, for certainperiods of timc, the pollen of onetaxon has arrived at a site by a different taphonomicroute than others.
Although damagedpollen profiles may be usedas an interpretativetool, their main functionappears to be limited to reinforcingthe main palynological,LOI and microscopiccharcoal data. In someinstances, where reductions in arborealpollen arc evident,there are no increasesin damagedpollen percentages(e. g. Loch an t'Suidhc),
215 andthe behaviourof damagedpollen doesnot alwaysfollow expectedpatterns. The processeswhich leadto the differenttypes of damagein pollen grainsneed to be bcttcr understoodbefore firmer concusionscan be drawn.
8.1.5 Cereal scanning and identification
The method of low-powcr scanningof contiguous samplesfor cercal-typc pollen resulted in the detection of additional cereal-typegrains which would not have bccn discovered during routine counting. The main problem encounteredwas low TLP concentrationsin some samples,which meant that only low pollen counts were scannedin some samplesand made it difficult to establish a fixed amount of grains to scan.However, the main problem lies not with the detection of cereal-typepollen, but its identification. Most of the cereal-typepollen detectedfell within the Hordcunl-type category and could be the pollen of large grassesrather than cultivated plants.
Although the archaeologicalconfirmation of a contemporaneoushuman presencefor cereal-typepollen detectedin this study suggeststhat this is Hordculn, the evidence is circumstantial and caution in interpretation is still needed.In pollen studies where no archaeological evidence is available to reinforce interpretations, then thesemust remain even more circumspect.
8.2 Conclusionsand implications of results for future research
The aboveindicates that multiple coring of singlesites is ableto provide a much clearerreconstruction of local vegetationthan a singleprofile, althoughclose resolutionsampling and a suiteof AMS datesare necessary to formulateany valid conclusions.Interpretations may alsobe enhancedby incorporatingadditional techniquessuch as charcoalquantification, damaged pollen analysis,and low-powcr scanningfor cereal-typepollen in a multiple profile study.
The previouschapter summarised the palaeoecologicaldata produced from the Inner
Hebrides,Outcr Hebridesand the west coastof Scotland.The multiple profile studies
216 undertakenin this programmeof researchprovide a greaterinsight into the early Holocenebehaviour of the major arborealtaxa in the Inner Hebrides.Nevertheless, the dataare limited to a few islandsand many more sites from Mull andSkye, as well as the smallerislands of Canna,Eigg andMuck, arerequired to fully appreciatethe ecologicaldynamics of earlyHolocene woodlands in the archipelago.
Multiple profiles from more sitesmay assistin identifying the impactsof climatic changes,and well datedmultiple profiles over a rangeof latitudesfrom western Scotlandand its islandsmay be comparedto thosefrom the easternScotland to assess the degreeof impactwhich climatemay havein different environmentalsettings. It may alsobe usefulto applypalaeomagnetic analysis to multiple pollen corcs from lake lake sediments,as this may assistin determiningthe processesacting within the and which may haveaffected pollen distribution.
Scotland The explanationfor the rapidly attainedhigh levelsof Corylusin western and the Inner Hebridesin the earlyHolocene remains largely unanswered.Whilst multiple profiles from areasof Skyeand Mull in additionto the smallerislands may clarify investigations matters,this shouldprobably be combinedwith landscapescale of soil developmentand chemistry, and studies of local meteorologicalconditions influenced by topography,in additionto climatic data.Also of note in this study arethe various timings for the Ulmusdecline at the different sites.The complexrange of factors
surroundingthe Ulmusdecline is very muchestablished, but multiple profile studies
from a wider areaof westernScotland could provide greaterdetail aboutecological influenceson Ulmusand its reactionsto thesein different environments.Inter- disciplinarystudies such as those suggested for investigationthe Corylusrise should
alsobe applied.
Although there are many pollen profiles from Skye and Mull, these are single corcs,
often with low resolution sampling, and lacking records of microscopic charcoal. Most
217 will be centrally located in the site basin which meansthat , their pollcn may be derived from a wide region and is not rcprcscntativc of local vcgctation, cspccially in lake profiles. Single corcs, particularly those which are centrally positioncd, may mcan that evidenceof human impact is less likely to be detectedand multiple prorilcs would increasedetection rates in addition to possibly rcprcscnting such evcnts in grcatcr detail.
The currentlack of evidencefor early Holocenehuman impacts in prorilcs from the wcst coastof Scotlandmay be correctedby futuremultiple profile studies.Mesolithic andNeolithic archacologyconfinns the presenceof people,particularly in the Oban areaand Kintyre pcninsula,and further multiple profile investigationsin theseareas would providematerial for comparisonwith that from the Inner Hebrides.Similar investigationswould alsobe appropriatein the OuterHebrides whcrc humanimpact may be moredifficult to discern.
The pollen evidenceobtained as part of this studymay alsobe usedto identify areas for archaeologicalinvestigation. The Mesolithic activity suggestedin the Loch an
VSuidheprofile occursin an areawhere there are currently no Mesolithic finds, and it may be appropriateto direct future fieldwalkingactivities in the Buncssanarea. Thcre is certainlya needto find moreMesolithic archaeologyon Mull, which this island lackscompared to the otherInner Hebrideanislands. Future discoveries of archaeologicalmaterial could also influencethe choiceof sitcs for pollen analysisin the future.
Although the pollen datafrom this studyhas provided a greatdeal of dctail for the palacoecologyof specificsites in the Inner Hebrides,comparison of thesesites with othersin the wider region is difficult due to differencesin the quality of datafrom the variousstudies. The datafrom multiple profile studiesarc ableto provide firmer interpretationsthan those from singleprofiles, but a greaterknowledge of the
218 taphonomicprocesses affecting pollen andcharcoal, and landscape studies of soils within a catchmentmay improvethe interpretationof multiple profiles in the futurc.
219 BIBLIOGRAPHY
Aaby, B., 1986.Palaeoecological studies of mires.In Berglund,B. E. (ed), Handbookof HolocenePalaeoecology and Palaeohydrology. 145- 164.Chichester: John Wiley and Sons.
Afficck, T.A., Edwards,K. andClarke, A., 1988.Archaeological and palynological studiesat the Mesolithic pitchstoneand flint site of Auchareoch,IsIc of Arran. Proceedingsofthe SocietyofAntiquaries ofScotland 118,37-59.
Agnew, A. D.Q., Edwards,K. J., Hirons,K. R., Hulme,P. D. andProctor, M. C.F., 1988. The Mire of Loch a!Bhogaidh, Islay. Transactionsof theBotanical Society ofEdinburgh 45,187-201.
Alley, R.B., Mayewski, P.A., Sowers, T., Stuiver, M., Taylor, K. C. and Clark, P.U., 1997. Holocene Climatic Instability: A prominent, widespreadevent 8200 yr. ago. Geology 25,483 - 486.
Andersen,S. T., 1979.Identification of wild grassand cereal pollen. Datimarb GeologiskeUndersogelselrbog 1978,69-72.
Andrews,M. V., Beck, R.B., Gilbertson,D. D. andSwitsur, V. R., 1987.The Pastand PresentVegetation on Oronsayand Colonsay, Section B: Palaeoenvironmental Investigations.In Mellars,P., Excavationson Oronsay.Edinburgh: Edinburgh University Press,57-71.
Artnit, 1., 1996.The Transitionto Agriculture 3: Scotland.In Pollard,T. andMorrison, A., (eds.) TheEarly Prehistoryof Scotland.Edinburgh: Edinburgh University Press.280- 291.
Armit, I. andFinlayson, W. F., 1992.Hunter-gatherers transformed: the transitionto agriculturein northernand western Europe. Antiquity 66,664-676.
Ashmore,P. J., 1996.Neolithic and BronzeAge Scotland.London: B. T. BatsfordLtd.
Ball, M. E., 1983.Native Woodlandsof the Inner Hebrides.Proceedings of the Royal SocietyofEdinburgh 8313;319-340.
Ball, M. E., 1987.Botany, Woodland and Forestry. In Clutton-Brock,T. H. andBall, M. E. (eds.), Rhum: YheNatural History ofan Island. Edinburgh;Edinburgh University Press, 43-62.
220 Ballantyne,CK andDawson, A. G., 1997.Geomorphology and Landscape Change. In Edwards,K. J. andRalston, I. B. M. (eds),Scotland: Environment and Archaeology,8000 BC -AD 1000.Chichester, John Wiley and SonsLtd., 23-44.
Ballantyne,C. K. andWain-Hobson, 1980. The Loch LomondAdvance on the Islandof Rhum.Scottish Journal of Geology16,1 -10.
Bangerter,E. B. andCannon, J. F. M., 1978.Selected distribution patterns shown by the vascularplants 2.2. In Jermy,A. C. andCrabbe, J. A. (eds.), TheIsland ofMull. - a survey of itsflora and environment.London: British Museum(Natural History).
Barber,D. C., Dyke, A., Hillaire-Marcel,C., Jennings,A. E., Andrews,J. T., Kerwin, M. W., Bilodeau,G., McNeely,R., Southons,J., Morehead,M. D. and Gagnon,J. -M., 1999.Forcing of the cold eventof 8,200years ago by catastrophicdrainage of Laurentide lakes.Nature 400,22/7/99,344-348.
Barclay, G.J., 1996. The Neolithic. In Edwards, K. J. and Ralston, I. B. M. (eds), Scollan& Environment and Archaeology, 8000 BC - AD 1000. Chichester,John Wiley and Sons Ltd., 127-150.
Bartlein,P. J. andPrentice, I. C., 1989.Orbital variations,climate and palacoecology. Trendsin Ecologyand Evolution4,195-199.
Batcy, C.E. (cd.), 1993.Discovery and Excavationin Scotland:An Annual Surveyof ScottishArchaeological Discoveries, Excavation and Fieldwork. Council for Scottish Archaeology,Edinburgh.
Bengtsen,L. andEnell, M., 1986.Chemical Analysis. Berglund, B. E., Handbookof HolocenePalaeoecology and Palaeohydrology.Chichester: John Wiley and Sons.
Bennett,K. D., 1983.Postglacial population expansion of foresttrees in Norfolk, UK. Nature 303,164-167.
Bennett,K. D., 1984.The post-glacialhistory of Pinussylvestris in the British Isles. QuaternaryScience Reviews 3, p. 133-155.
Bennett,K. D., 1988.Post-glacial vegetation history: ecologicalconsiderations. In Huntley,B. andWebb, T. III (eds.), VegetationHistory. Dordrecht:Kluwer Academic Publishers 699-724.
Bennett, K. D., 1989. Postglacial vegetation history of the Rhinns of Islay. In Mithen, S.J. (ed.), Yhe Southern Hebrides Mesolithic Project, 2nd interim report, Unpublished.
Bennett,KD., 1994.Annotated catalogue of pollen andpteridophyte spore types of the British Isles.httplwww. kv. geo. uu. se1pc-intro. htm1
221 Bennett,K. D., 1996.Late-Quatemary Vegetation Dynamics of the Caimgorms. BotanicalJournal ofScotland,48,51- 63.
Bennett,K. D. andBirks, H.J. B., 1990.Postglacial history of alder(Alnus glutinosa (L. ) Gaertn.) in the British Isles.Journal of QuaternaryScience 5,123-133.
Bennett,K. D., Boreham,S., Sharp,M. J. andSwitsur, V. R., 1992.Holocene history of environment,vegetation and human settlement on CattaNess, Lunnasting, Shetland. Journal ofEcology 80,241-273.
Bennett,K. A., Fossitt,J. A., Sharp,M. J. and Switsur,V. R., 1990.Holocene vegetational and environmentalhistory at Loch Lang, SouthUist, WesternIsles, Scotland. New Phytologist 114,281-298.
Bennett, K. D. and Lamb, H. F., 1984. The history of pine and oak in S.W. Scotland. British Geological Society Bulletin 5,16-17.
Bennett,K. D., Whittington,G. andEdwards, K. J., 1994.Recent Plant Nomenclature Changesand Pollen Morphology in the British Isles.Quaternary Newsletter June 1994, 1-6.
Birks, H.H., 1984.Late-Quaternary pollen andplant macrofossilstratigraphy at Lochan an Druim, north-westScotland. Haworth, E. Y. andLund, J.W. G. (eds), Lake Sediments and EnvironmentalHistory. LeicesterUniversity Press.
Birks, H.J. B., 1970.Inwashed pollen spectraat Loch Fada,Isle of Skye.New Phytologist 69,807 - 820.
Birks, H.J. B., 1973.Past and PresentVegetation ofthe Isle ofSkye.A Palaeoccological Study.Cambridge: CUP.
Birks, H.J. B., 1980.Quaternary Vegetation History of WestScotland. Fifth International PalynologicalConference, Guidebook for ExcursionC8. Cambridge.
Birks, H.J. B., 1987.The Pastand Present Vegetation of Oronsayand Colonsay,Section A. Mellars, P., Excavationson Oronsay-Edinburgh: Edinburgh University Press.
Birks, H.J. B., 1989.Holocene isochrone maps and patterns of tree-spreadingin the British Isles.Journal ofBiogeography16,503-540.
Birks, H. J.B., 1990. Changesin vegetation climated during the Holocene of Europe. Boer, M. M. and DeGroot, R. S. (eds.), Landscape-Ecological impact ofcliniatic change. Amsterdam: IOS Press, 133-158.
222 Birks, H.J. B., 1993.Loch Ashik, Loch Cleatand Loch Meodal.In Gordon,J. E. and Sutherland,D. G. (eds.), QuaternaryofScotland. London: Chapman and Hall, 399-407.
Birks, H.J. B. andMadsden, B. J., 1979.Flandrian vegetational history of Little Loch Roag,Isle of Lewis, Scotland.Journal ofEcology 67,825-842.
Birks, H.J. B. andWilliams, W., 1983.Late-Quaternary vegetational history of the Inner Hebrides.Proceedings ofthe RoyalSociety ofEdinburgli 8313,269-292.
B6hncke,S. J. P., 1988.Vegetation and Habitation History of the CallanishArea, Isle of Lewis, Scotland.In Birks, H.H., Birks, H.J. B., Karland,P. E. andMoe, D.(eds), 771c Cultural Landscape:Past Present and Future. Cambridge:Cambridge University Press, 445-461.
Bonny, A. P., 1978. The effect of pollen recruitment processeson pollen distribution over the sediment surface of a small lake in Cumbria. Journal ofEcology 66,385-416.
Bonsall, C., 1996. The 'Obanian Problem': CoastalAdaptation in the Mesolithic of Western Scotland. In Pollard, T. and Morrison, A., (eds.) The Early Prchistory of Scotland. Edinburgh: Edinburgh University Press, 183-197.
Bonsall,C. and Smith, C., 1990.Bone andantler technology in the British Late Upper Palaeolithicand Mesolithic: the impactof acceleratordating. In Vermeersch,P. M. and Van Peer,P., (eds.), Contributionsto the Mesolithic in Europe.Leuvenileavcn University Press,359-368.
Bonsall,C., Sutherland,D. G., Lawson,T. J. andRussell, N. J., 1991.Excavations in Ulva Cave,western Scotland 1989: a preliminaryreport. Mcsolithic Misccllany 13,7-13.
Bonsall,C., Sutherland,D. G., Lawson,T. J., Russell,N. J. andBarnetson, L., 1989. Ulva Cave:Excavation Report No. 2. University of Edinburgh,Department of Archaeology. Unpublished.
Bonsall,C., Sutherland,D. G., Russell,N. J., Coles,G., Paul,C. R. C., Huntley,J. and Lawson,T. J., 1995.Excavations in Ulva Cave,western Scotland 1990-9 1: a preliminary report. MesolithicMiscellany 15,8-21.
Bos,J. A. A. andJanssen, C. R., 1996. Local Impactof PalaeolithicMan on the EnvironmentDuring the End of the Last Glacial in the Netherlands.Journal of ArchaeologicalScience 23,731-739.
Boyd, J.M. andBoyd, I.L., 1990.Yhe Hebrides. A Natural History. London:William Collins Sonsand Co.
223 Boyd, W. E., 1988. Early Flandrian vegetational developmenton the coastal plain of north Ayrshire, Scotland: evidence from multiple pollen profiles. Journal ofBiogcography 15, 325-337.
Boyd, W.E. andDickson, J. H., 1986.Patterns in the GeographicalDistribution of the Early FlandrianCorylus rise in southwestScotland. New Phytologist 102,615-623.
Bradbury, J.P., Tarapchak, S.J., Waddington, J.C. B. and Wright, R.F., 1975. The impact of forest fire on a wilderness lake in northeasternMinnesota. Verhandlungen Internationales VereinigungfdrLimnologie 19,875 - 883.
Bradbury,J. P., 1996.Charcoal deposition and redeposition in Elk Lake,Minnesota, USA. In TheHolocene 6,339-344.
Bradshaw, R.H. and Browne, P., 1987. Changing patterns in the post-glacial distribution of Pinus sylvestris in Ireland. Journal ofBiogeography 14,237-248.
Brush, G.S. and Brush, L. M., Jr., 1972. Transport of pollen in a scdiment-ladcnchannel: a laboratory study. American Journal ofScience 272,3 59-3 8 1.
Buckland,P. C. andEdwards, K. J., 1984.The longevityof pastoralepisodes of clearance activity in pollen diagrarns:the role of post-occupationgrazing. Journal ofBiogeography 11,243-249.
Bunting,MI, 1994.Vegetation history of Orkney,Scotland; pollen recordsfrom two small basinsin west Mainland.New Phytologist, 128,771-792.
Bunting, M. J. andDavies, A. L., in press.Pollen from Loch Gon-n,Islay. In Mithen, S.J. (ed.), TheSouthern Hebrides Mesolithic Project. Cambridge:MacDonald Institute Monographs.
Campbell,I. D., 1991.Experimental mechanical destruction of pollen grains.Palynology 15,29-33.
Campbell,I. D. andCampbell, C., 1994.Pollen preservation: experimental wet-dry cycles in salineand desalinated sediments. Palynology 18,5 - 10.
Carp,E., 1994.The pollen sequencefrom the Uppersection at Coulererach.In Mithen, S-J.(ed. ), 71e SouthernHebrides Mesolithic Project 1994(W) Interifn Report.Dept. of Archaeology,University of Reading,71-72.
Carruthers,W., 1994. The charredplant remains from Feature 24, Staosnaig:A preliminary assessment.In Mithen, S.J. (ed.), Vie Southen, Hebrides Mesolithic Project 1994 (7th) Interim Report. Dept. of Archaeology, University of Reading, 98-10 1.
224 Caseldine,C., 1991.Surface pollen studiesacross Bankhead Moss, Fife, Scotland. Journal of Biogeography8,7-25.
Cascldine,C. andHatton, J., 1993.The developmentof high moorlandon Dartmorr:fire andthe influenceof Mesolithicactivity on vegetationchange. In Chambers,F. M., Climate Changeand HumanImpact on the Landscape.London: Chapman and Hall, 119- 132.
Chambers,F. M. andElliott, L., 1989.Spread and expansion of Ahius Mill. in the British Isles:timing, agenciesand possible vectors. Journal ofBiogeography16,541-550.
Chen,Y., 1987.Pollen and sedimentdistribution in a small craterlake in Northeast Queensland,Australia. Pollen et Spores28,89 - 110.
Clark, J.S., 1988. Particle Motion and the Theory of Charcoal Analysis: Source Area, Transport, Deposition and Sampling. Quaternar Research30,67-80. ,y Clark, J.S., Merkt, J. and MUller, H., 1989. Post-glacial fire, vegetation and human history on the northern alpine forelands, south-westcrnGermany. Journal ofEcology 77, 897-925.
Clark, R.L., 1982.Point countestimation of charcoalin pollen preparationsand thin sectionsof sediments.Pollen et Spores,Vol. 24, No. 3-4,523-535.
Clutton-Brock,T. H. andBall, M. E., 1987.Introduction. In Clutton-Brock,T. H. andBall, M. E.(es. ), Rhum: YheNatural History ofan Island. Edinburgh;Edinburgh University Press, 1-10.
Coles,B. J., 1998.Doggerland: a speculativesurvey. Proceedings of the Prehistoric Society64,45 -8 1.
Coles, J.M., 1971. The early settlementof Scotland: excavationsat Morton, Fife. Proceedings ofthe Prehistoric Society 37,284-366.
Currie, A. andMurray, C., 1983.Flora andVegetation of the Inner Hebrides. Proceedingsof the RoyalSociety ofEdinburgh 8313;293-318.
Cushing,E. J., 1964.Redeposited pollen in Late Wisconsinpollen spectrafrom east- centralMinnesota. Amercian Journal ofSdence 262,1075-1088.
Cushing,E. J., 1967.Evidence for differentialpollen preservationin late Quaternary sedimentsin Minnesota.Review ofPalaeobotany and Palynology,4,87 - 101.
Dalland,M., 1997.Radiocarbon dates from Fife Ness.In Turner,R. (ed.), Discoveryand Excavationin Scotland.Edinburgh: Council for ScottishArchaeology, 113.
225 Davis, M. B., 1973.Redeposition of pollen grainsin lake sediment.Linmology and Oceanography18,44-52.
Davis, M. B. andBrubaker, L. B., 1973.Differential sedimentationof pollen grainsin lakes. Lininologyand Oceanography18,635-646.
Davis, M. B., Moeller, R.E. andFord, J., 1984.Sediment focusing and pollen influx. Haworth,E. Y. andLund, J.W. G. (eds.), Lake Sedimentsand EnvironnientalHistory. LeicesterUniversity Press.
Davis, R.B., Brewster,L. A. and Sutherland,J., 1969.Variation in pollen spectrawithin lakes. Pollen et Spores11,557-571.
Dawson,1984. Quaternary sea-level changes in westernScotland. Quatemary Science Reviews3,345-368.
Dawson, A. and Dawson, S., 1994. Late Pleistoceneand Early Holocene Sea level change on Islay. Mithen, S.J. (ed.), The SouthernHebrides Mesolithic Project 1994 (7111) Interim Report. Dept. of Archaeology, University of Reading, 61-70.
Day, P., 1996.Devensian Late-glacial and early Flandrianenvironmental history of the Vale of Pickering,Yorkshire, England. Journal of QuaternaryScience 11,9-24.
Douglas Price, T., 1990. The Reconstruction of Mesolithic Diets. In Bonsall C. (cd.), 71c Mcsolithic in Europc. Edinburgh: John Donald Publishers,48-59.
Dubois,A. D. andFerguson, D. K., 1985.The climatic history of pine in the Cairngorms basedon radiocarbondates and stable isotope analysis, with an accountof the events leadingup to its colonization.Review qfPalaeobotany and Palynology46,55-80.
Edwards,K. J., 1981.The seperationof Coryltis andMyrica pollen in modemfossil samples. Pollen et Spores23, No. 2,205-218.
Edwards,K. J., 1982. Man, spaceand the woodlandedge: speculations on the detection andinterpretation of humanimpact in pollen profiles.In Bell, M. andLimbrey, S. (eds.), ArchaeologicalAspects of WoodlandEcology. BAR Int. Series146,5-22.
Edwards,K. J., 1983.Quaternary palynology: multiple profile studiesand pollen variability. Progressin PhysicalGeography Vol. 7,587-609.
Edwards, K. J., 1988. The Hunter-Gatherer/Agricultural Transition and the pollen record in the British Isles. In Birks, H. H., Birks, H. J.B., Karland, P.E. and Moe, D. (cds.), The Cultural Landscape: Past Present and Future. Cambridge: C.U. P.,255-266.
226 Edwards,K. J., 1989.Meso-Neolithic Vegetational Impacts in Scotlandand Beyond: PalynologicalConsiderations. In Bonsall,C. (ed.), TheMesolithic in Eltrope.Edinburgh: JohnDonald Publishers Ltd.
Edwards,K. J., 1990.Fire andthe ScottishMesolithic: Evidencefrom Microscopic Charcoal. Versmeersch,P. M. andVan Peer,P, (eds.), Contributionsto theMesolithic in Europe,Leuven: Leuven University Press, 71-79.
Edwards,K. J., 1991.Spatial scale and palynology: a commentaryon Bradshaw.In Harris, D.R. andThomas, K. D. (eds.), Modelling ecologicalchange. London: Institute of Archaeology,University CollegeLondon, 53-59.
Edwards,K. J., 1996.A Mesolithic of the Westernand Northern Isles of Scotland? Evidencefrom Pollen and Charcoal.In Pollard,T. andMorrison, A., (eds.) 77ieEarly PrehistoryofScotland. Edinburgh: Edinburgh University Press, 23-38.
Edwards, K. J., in press. Vegetation history of the southernInner Hebrides during the Mesolithic period. In Mithcn, S.J. (cd.), The Southeni Hebrides Mesolithic Project, Cambridge; MacDonald Institute Monographs.
Edwards,K. J., Ansell, M. andCarter, B. A., 1983.New Mesolithic Sitesin south-west Scotlandand their importanceas indicators of inland penetration.Transactions of the Dumfrieshireand GallowayNatural History and AntiquarianSociety 8,9-15.
Edwards,K. J. andBerridge, J. M. A., 1994.The Late-Quatemaryvegetational history of Loch a'Bhogaidh,Rinns of Islay S.S. S. I., Scotland.New Phytologist 128,749-769. deposits: Edwards,K. J. andHirons, K. R., 1984.Cereal pollen grainsin pre-elmdecline implicationsfor the earliestagriculture in Britain andIreland. Journal ofArchaeological Science11,71-80.
Edwards,K. J. andHirons, K. R., 1990.Environmental analyses of samplesfrom Trench AM anda blanketpeat remnant, Farm Fields excavations,Kinloch, Isle of Rhum. In Wickham-Jones,C. R. (ed.), Rhum:Mesolithic and later sitesat Kinloch, Excavations 1984-1986,Society of Antiquariesof ScotlandMonograph Series 7.
Edwards,K. J., Hirons, K. R. andNewell, P.J., 1991,The palacoccologicaland prehistoric contextof minerogeniclayers in blanketpeat: a studyfrom Loch Dee,southwest Scotland.The Holocene 1,29 - 39.
Edwards, K. J. and McIntosh,, C.J., 1988. Improving the detection rate of Cereal-typc pollen grains from Ulmus decline and earlier deposits from Scotland. Pollen ct Spores 30, 179-188.
227 Edwards,K. J. andMithen, S., 1995.The colonizationof the HebrideanIslands of WesternScotland: evidence from the palynologicaland archaeological records. World ArchaeologyVol. 26,348-365.
Edwards,K. J. andRalston, 1., 1984.Post-glacial hunter-gatherers and vcgctational history in Scotland.Proceedings of the SocietyofAntiquaries of Scotland,114,15-34.
Edwards,K. J. andThompson, 1984. Magnetic, palynological and radiocarbon correlation anddating comparisons in long coresfrom a northernIrish lake. Cateiia 11,83-89.
Edwards,K. J. andWhittington, G., 1993.Aspects of the Environmentaland DepositionalHistory of a Rock BasinLake in EasternScotland, UK. In McManus,J. and Duck, R.W. (eds.), Geomorphologyand Sedimentologyof lakesand reservoirs. Chichester:John Wiley andSons.
Edwards, K. J. and Whittington, G., 1994. Lateglacial pollen sites in the Western Isles of Scotland. Scottish Geographical Magazine 110,33-39.
Edwards,K. J. andWhittington, G. 1997a.Vegetation Change. In Edwards,K. J. and Ralston,I. B.M., Scotland.,Environment and Archaeology, 8000 BC - AD 1000. Chichester:John Wiley andSons, 63 - 82.
Edwards, KJ. and Whittington, G., 1997b.Vegetation History and Archacobotany. lake: Edwards,K. J. andWhittington, G., in press.Multiple charcoalprofiles in a Scottish taphonomy,fire ecology,human impact andinference. Palaeogeography, PaIdeoclimatology,PaIdeoccology.
Edwards,K. J., Whittington, G. andHirons, K. R., 1995.The relationshipbetween fire and long-termwet heathdevelopment in SouthUist, OuterHebrides, Scotland. In Thompson, D.B. A., Hester,A. J. andUsher, M. B. (eds.), Heathsand Moorland: Cultural Landscapes. Edinburgh:HMSO, 240-248.
Edwards,K. J., Whittington, G. andTipping, R., in press.The incidenceof microscopic charcoalin Lateglacialdeposits. Palaeogeography, Palaeoclinialology, Palaeoecology.
Emeleus,C. H., 1987.The RhumVolcano in Clutton-Brock,T. H. andBall, M. E., Rhum: TheNatural History ofan Island. Edinburgh:Edinburgh University Press.
Emclcus, C.H., 1991. Tertiary Igneous Activity in Craig, G.Y. (cd.), Geology ofScolland 3rd edn., London: The Geological Society. by Faegri, K. and Iversen, J., 1989. TextbookofPollen Analysis, 4th edn. (revised K. Faegri, P.E. Kaland and K. Krzywinski). Chichester: John Wiley and Sons.
228 Fall, P.L., 1987.Pollen taphonomyin a canyonstream. Qualcritary Rcscarch 28,393- 406.
Finlayson,B., 1995.Complexity in the Mesolithic of the WesternScottish Seaboard. In Fischer,A. (ed.), Man and Scain the Mcsolithic. Oxford: Oxbow Monograph53.
Finlayson,B., 1996.The Transitionto Agriculture2: The Basisfor Change.In Pollard,T. andMorrison, A., TheEarly PrehistoryofScotland. Edinburgh: Edinburgh University Press,213-224.
Finlayson,B. and Edwards,K. J., 1997.The Mesolithic.In Edwards,K. J. andRalston, 1. B. M., Scotland:Environment and Archaeology 8000 BC - AD 1000.Chichester: John Wiley and Sons,109-126.
Flenley, J.R. and Pearson,M. C., 1967. Pollen analysis of a peat from the island of Canna (Inner Hebrides). New Phytologist 66,299-3 06.
Fossitt,J. A., 1996.Late Quaternaryvegetation history of the WcstemIsIcs of Scotland. NewPhytologist 132,171-196.
Gacia,E. andBallesteros, E., 1994.Production of Isocleslacustris in a Pyrmeanlake: seasonalityand ecological factors involved in the growing period.Aquatic Botany,48, 77-89.
Garavell,D, 1998.Stone Age cherrytree find setto rewrite history books.Scolland on Sunday,25/10/98.
Garbett,D., 1981.The Elm Decline:the depletionof a resource.New Xytologist 88, 573-585.
Gear,AI andHuntley, B., 1991.Rapid changes in the rangelimits of Scotspine 4000 yearsago. Science 251,544-547.
Girling, M. A. andGreig, J., 1985.A first fossil recordfor Scolytusscolytus (F.) (elm bark beetle):its occurrencein elm declinedeposits from Londonand the implicationsfor Neolithic elm disease.Journal ofArchaeologicalScience 12,347-35 1.
Grace,J., 1987.Climatic toleranceand the distributionof plants.In Rorison,I. H., Grime, J.P., Hunt, R., Hendry,G. A. F. andLewis, D.H. (eds.), Frontiers of Colliparitivc Plant Ecology.(Reprint of New Phytologist106 [Supplement],1987). London: Academic Press.
Gray,J. M. andLowe, J.J., (eds.), 1977.Studies in the ScottishLateglacial Environment. Oxford: Pergamon.
229 Grccn, D. G., 1982. Firc and stability in the postglacial forcsts of southwcst Nova Scotia. Journal of Biogeography, 9,29-40.
Green,D. G., 1983.The EcologicalInterpretation of Fine ResolutionPollcn Rccords.Now Phytologist94,459-477.
Green,F. H. W. and Harding,R. J., 1983.Climate of the Inner Hebrides.Procce(lings of the Royal SocietyofEdinburgh 83B, 121-140.
Grigson,C. andMellars, P., 1987.The MammalianRemains from the Middcns. Mellars, P., Excavationson Oronsay.Edinburgh: Edinburgh University Press.
Grime,J. P., Hodgson,J. G. andHunt, R., 1988.Coniparitive Plant Ecology.London, Unwin Hyman.
Grocrunan-vanWaateringc, W., 1983. The early agricultural utilization of the Irish Landscape:the last word on the elm decline? In Rccvcs-Smytlle, T. and Hamond. F., LandscapeArchaeology in Ireland, B. A. R. British Series 116,217-232.
Haggart,B. A. andSutherland, D. G., 1992.Moine Mhor. In Walker,M. J.C., Gray, J.M. andLoe, J.J., (eds.), Thesouth-west Scottish Highlands: Field Guide.Cambridge: QuaternaryResearch Association.
Hakenson,L. and Jansson,M., 1983.Lake Sedimentology.Berlin: Springer-Verlag.
Hall, V. A., 1988.The Role of HarvestingTechniques in the Dispersalof Pollen Grainsof Cerealia.Pollen et Spores30,265-270.
Hall, V. A., 1989.A study of the modempollen rain from a reconstructed19th century farm.Irish Naturalists'Journal.
Harland,R., 1989.A dinoflagellatecyst recordfror the last 0.7 Ma from the Rockall Plateau,northeast Atlantic Ocean.Journal of the GeologicalSociety, London, 146,945- 951.
Harland,R., 1998.Dinoflagellate Cyst Analysis ofSuspected Marine SedimentsfromTwo Siteson the Island of Ulva, North WestScotland. Report RH/98/02/01. Unpublished.
Havinga,A. J., 1964.Investigation into the differentialcorrosion susceptibility of pollen and spores.Pollen et Spores6,621-635.
Henshall, A. S., 1972. The ChamberedTombs ofScolland, vol. 2. Edinburgh: Edinburgh University Press.
230 Hicks, S., 1988.The representationof differentfanning practicesin pollen diagramsfrom northernFinland. In Birks, H.H., Birks, H.J. B., Karland,P. E. andMoc, D.(cds. ), 77ic Cultural Landscape:Past Present and Future. Cambridge:Cambridge University Press, 189-207.
Hill, M. O., 1979.DECORANA -a FORTRANprograinforDelrended Correspondance Analysisand ReciprocalAveraging. Ithaca: Comell Unimsity.
Hirons, K. R. and Edwards,K. J., 1990.Pollen and related studies at Kinloch, IsIc of Rhum,Scotland, with particularreference to possibleearly humanimpacts on vegetation. New Phytologist116,715 - 727.
Hobbs,R. J. andGimingham, C. H., 1987.Vegetation, fire andherbivore interactions in heathland.Advances in EcologicalResearch 16,87-173.
Holloway, R. G., 1989. Experimental mechanicalpollen degradationand its application to Quaternary age deposits. TexasJournal ofScience 41,131-145.
Howard,J. andJones, A., 1990.The Isle of Ulva: A VisitorsGui(le. Ulva: Howard and Jones.
Hudson,G. andHenderson, D. J., 1983.Soils of the Inner Hebrides.Proceedings of the RoyalSociety ofEdinburgh 83B; 107-120.
Hulme, P.D. and Shirriffs, J., 1994. The Late-glacial and Holocene vegetation of the Lang Lochs Mire area, Gulberwick, Shetland: a pollen and macrofossit investigation. New Phytologist 128,793-806.
Huntley,B., 1993.Rapid early-Holocenemigration and high abundanceof hazel(Corylus avellanaL. ) alternativehypotheses. In Chambers,F. M., ClimateChange and 1114man Impact on the Landscape.London, Chapman and Hall, 205-216.
Huntley,B. andDirks, H.J. B., 1983.An Atlas ofPast and PresentPollen Afapsfor Europe: 0-13000yearsago. Cambridge:Cambridge University Press.
Huntley,B., Bartlein, P.J. andPrentice, I. C., 1989.Climatic controlof the distribution andabundance of beech(Fagus L. ) in Europeand North America.Journal of Biogeography16,551-560.
Jackson,S. T., 1994.Pollen and Sporesin Quaternarylake sedimentsas sensors of vegetationcomposition: theoretical models and empiricalevidence. In Traverse,A. (ed.), Sedimentationof OrganicParticles. Cambridge: Cambridge University Press,253-285.
Jacobi,R. M., Tallis, J.H. andMellars, P.A., 1976.The southcmPennine Mesolithic and the ecologicalrecord. Journal ofArchaeologicalScience 3,307-320.
231 Jacobson,G. L. andBradshaw, R. H. W., 1981. The selectionof sitcs for palacovegetationalstudies. Quaternary Research 16,80-96.
Jannssen,C. R., 1973.Local andregional pollen deposition.In Birks, HIB. andWest, R.G., QuaternaryPlant Ecology.Oxford: Blackwell ScientificPublications, 31-42.
Jannsscn,C. R., 1984.Modern pollen assemblagesand vegetation in the Myrtle Lake peatland,Minnesota. Ecological Monographs, 54,213-252.
Jardine,W. G., 1975.Chronology of Holocenemarine transgression and regression in south-wcstemScotland. Boreas 4,173-195.
Jermy,A. C., 1978.Climate, 6.1- 6.11.Jermy, A. C. andCrabbe, J. A. (eds.), 77icIsland of Mulh a surveyof itsflora and environment.London: British Museum(Natural History).
Jenny, A. C. and Crabbe, J.A. (eds.), 1978.71e IsImid qfMull: a survey of itsflora and environnient. London: British Museum (Natural History).
Jenny,A. C., James,P. W. andEddy, A., 1978.Terrestrial Ecosystems, 10.1-10.77. Jermy,A. C. and Crabbe,J. A. (eds.), TheIsland ofMulk a surveyof itsflora and environment.London: British Museum(Natural History).
Jones,R. L., 1978.Late-Quatemary vegetational bostory of the North York Moors. VI. The ClevelandMoors. Journal ofBiogeography,5,81-92.
Jowsey,P. C., 1966.An improvedpeat sampler. New Phytologist 65,3 5-55.
Kaminski,J., 1994.Charred plant remainsfrom BolsayFarm. In Mithcn, S.J. (cd.), Yhe SouthernHebrides Mesolithic Project 1994(7th) Interim Report.Dept. of Archaeology, University of Reading:91-94.
Kohler, E. andLange, E., 1979.A contributionto distinguishingcereal from wild grass pollen grainsby LM and SEM. Grana, 18,133 - 140.
Kvarnme,K. L. andJochim, M. A., 1985.The EnvironmentalBasis of Mesolithic Settlement. Bonsall,C. (ed.), TheMesolithic in Europe.Edinburgh; John Donald PublishersLtd., 1-12.
Lacaille, A. D., 1954. The StoneAge in Scotland. London: Oxford Unimsity Prcss.
Lang, G., 1992. Some aspectsof European late- and post-glacial flora history. Acta Botanici Fennici, 144,1-17.
232 Larsen,C. P. S., 1996.Fire andclimate dynamics in the borealforest of northernAlberta, Canada,from AD 1850to 1989.Ae Holocene6,449 - 456.
Lawson,T. J. andBonsall, C., 1986.The Palaeolithicof Scotland:a reconsiderationof evidencefrom ReindeerCavc, Assynt. In Colcutt,S. N. (cd.), ThePalaeolithic ofBritain and its NearestNeighbours: Recent Trends. Shcffield: Dcpt. of Archaeologyand PrehistoryUniversity of Sheffield,85-89.
Legg, C.J., Maltby, E. and Procter,M. C.F., 1992.The ecologyof severemoorland fire on the North York Moors: seeddistribution and seedling establishment of Calluna Vulgaris. Journal ofEcology 80,737-752.
Lchtonen, H. and Huttunen, P., 1997. History of forest fires in easternFinland from the fifteenth century AD - the possible effects of slash-and-burncultivation. 211cHolocene 7, 223-228.
Long, D., Wickham-Jones,C. R. andRuckley, N. A., 1986.A flint artefactfrom the northernNorth Sea.In Roe.,D. A. (ed.), Studiesin the UpperPalaeolithic ofBritain and NorthwestEurope. BAR InternationalSeries 296,55-62.
Lowe, JI, 1982.Three Flandrian pollen profiles from the Teith Valley, Perthshire, Scotland.II. Analysis of deterioratedpollen. New Nytologist 90,371-385.
Lowe, J.J., 1993.Isolating the climatic factorsin early- andmid-Holocene palaeobotanicalrecords from Scotland.In Chambers,F. M. (ed.), ClimateChange and HumanImpact on the Landscape.London, Chapman and Hall, 67-82.
Lowe, J.J. and Walker,M. J.C., 1986a.Flandrian Environmental History of the Isle of Mull, Scotland.II, Pollen Analytical Data from sitesin westernand northern Mull. New Phytologist103,417-43 6.
Lowe, J.J. andWalker, M. J.C., 1986b.Lateglacial and early Flandrianenvironmental hisstoryof the Isle of Mull, Inner Hebrides,Scotland. Transactions of the Royal Society ofEdinburgh: Earth Sciences77,1-20.
Maycwski,P. A., Buckland,P. C., Edwards,K. J., Meeker,L. D. andO'Brien, S., 1996. ClimateChange Events as Seenin the GreenlandIce Core(GISP2): Implications for the Mesolithic of Scotland.In Pollard,T. andMorrison, A., (eds.), TheEarly Prehistoryof Scotland.Edinburgh: Edinburgh University Press,74 - 86.
McCann,S. B. andRichards, A., 1969.The coastalfeatures of the islandof Rhum in the Inner Hebrides.Scottish Journal of Geology5,15-25.
233 McCullagh,R., 1990.Excavation at Newton,Islay. GlasgowArchaeological Journal, 15, (1988-1989),23-50.
McVcan, D.N., 1956a.Ecology of Alnusglutinosa (L. ) Gacrtn.V Noteson someBritish alderpopulations. Journal ofEcology 44,321 - 330.
McVean,D. N., 1956b.Ecology of Alnusglutinosa (L. ) Gaertn.VI Post-glacialhistory. Journal ofEcology 44,331-333.
Maenza-Gmelch,T. E., 1997.Holocene vegetaion climate, and fire history of the Hudson Highlands,southeastern New York, USA. 27icHolocene 7,25-3 7.
Mellars, P.A., 1987.Excavations on Oronsay:Prehistoric Human Ecology on a Small Island. Edinburgh:Edinburgh University Press.
Mercer, J., 1970. Flint tools from the presenttidal zone, Lussa Bay, Isle of Jura, Argyll. Proceedings ofthe Society ofAntiquaries ofScotland 102,1-30.
Mercer,J., 1978.Lussa Wood 1: the Late-Glacialand EarlY Post-Glacial Occupation of Jura. ProceedingsoftheSocoetyof, 4ntiquariesofScollaiicl, 110,1-32.
Mercer,J., 1974.Glcnbatrick Watcrhole, a microlithic site on the Isle of Jura. Proceedingsofthe PrehistoricSociety, 105,9-32.
Mithen, S., 1995.Mesolithic settlementand raw materialavailablifly in the southem Hebrides. In Fischer,A. (ed.), Man and Seain theMesolithic. Oxford: Oxbow Monograph53.
Mithen, S., 1996.Ae SouthernHebrides Mesolithic Project, 1996Progress Report. Unpublished.
Mithen, S.J. andLake, M., 1996.The SouthernHebrides Mesolithic Project: ReconstructingMesolithic Settlementin WesternScotland. In Pollard,T. andMorrison, A., (eds.), TheEarly PrehistoryofScotland. Edinburgh: Edinburgh University Press, 123-151.
Moffatt, B., 1990.Report on pollen and ancillaryanalyses in supportof the excavations at Kinloch, Rhum, 1984,1985,1986.In Wickham-Jones,C. R. (cd.), Rhuni:Afesolithic and later sitesat Kinloch, Excavations1984-1986, Society of Antiquariesof Scotland MonographSeries 7.
Moore, J., 1996. Damp Squib: How to Fire a Major Deciduous Forest in an Inclement Climate. In Pollard, T. and Morrison, A. (eds.), The Early Prehistory ofScolland. Edinburgh: Edinburgh University Press,62-73.
234 Moore, P.D., 1977. Stratigraphy and pollen analysis of Claish Moss, North-wcst Scotland: significance for the origin of surface-poolsand forest history. Jounial of Ecology 65,375-397.
Moore, P.D., 1986.Hydrological changes in mires.In Berglund,BE (ed), Handbookof HolocenePalaeoecology and Palaeohydrology.John Wiley and Sons:Chichester.
Moore, P.D., Webb,J. A., andCollinson, M. E., 1991.Pollen Analysis.Oxford, Blackwell Scientific Publications.
Morrison, A., 1996.The northwardmarch of Palaeolithicman in Britain. In Pollard,T. andMorrison, A. (eds.), YheEarly PrehistoryofScotlaild. Edinburgh:Edinburgh University Press.
Morrison, A. and Bonsall, C., 1989. The Early Post-Glacial Settlement of Scotland: a Review. Bonsall, C., YheMesolithic in Europe. Edinburgh: John Donald Publishers Ltd.
Movius, H. L., 1940. YheIrish StoneAge: It's Chronology, Developmentand Relations. Cambridgc: Carnbridgc Unimsity Prcss.
Mulder, Y., 1999.Aspects of Vegetationand Settlementhistoty in the OuterHebrides, Scotland.Unpublished Phd thesis, University of Sheffield.
Murray, N.A., Bonsall,C., Sutherland,D. G., Lawson,T. J. andKitchener, A. C., 1993. Furtherradiocarbon determinations on reindeerremains of Middle andLate Devensian agefrom the Creagnan Uamh caves, Assynt, NW Scotland.Quateniary Newsletter 70,1- 10.
Myers, A. M., 1988.Scotland inside and outside of the British mainlandMesolithic. ScottishArchaeological Review No. 5,23-39.
Newell, P.J., 1988.A buriedwall in peatlandby Shesader,Isle of Lewis. Societyof AntiquariesofScotland 118(1988), 79-93.
Nichols, H., 1967.VI. Vegetationalchange, Shoreline displacement and the Human Factorin the Late QuaternaryHistory of South-WcstScotland. Transactions of the Royal SocietyofEdinburgh 67,145-185.
O'Connell,M., 1986.Reconstruction of local landscapedcvelopmcnt in the Post-Atlantic basedon palaeoecologicalinvestigations at Carrownagloghprehistoric field system, CountyMayo, Ireland.Review ofPalaeobotany and Palynology49,117-126.
O'Connell,M., 1987.Early Cereal-Typepollen recordsfrom Connemara,Western Ireland andtheir possiblesignificance. Pollen et Spores29,207-224.
235 O'Connell,M., Molly, K. and Bowler, M., 1988.Post-glacial Landscape Evolution in Connemara,Western Ireland with particularreference to woodlandhistory. In Birks, H.H., Birks, H.J. B., Karland,P. E. andMoe, D.(cds. ), TheCultural Landscape:Past Presentand Future. Cambridge:Cambridge Univcrsity. Prcss, 267-287.
Ogilivie, M., 1995.Ae Wild Flowersof1slay. Islay NaturalHistory Trust.
Page,C. N., 1997.The Ferns ofBritain and Ireland. Cambridge:Cambridge University Press.
Parish,R., 1990.A PalynologicalAnalysis of a pcatcore from the Kinloch Glcn, Isle of Rhum.In Wickham-Joncs,C. R. (ed.), Rhuni:Mesolithic and later sitesat Kinloch, Excavations1984-1986, Society of Antiquariesof ScotlandMonograph Series 7.
Patterson,W. A. 111,1978.The effects ofpast and current land disturbance oil Squaw Lake, Minnesota and its watershed.Unpublished PhD thesis, University of Minnesota, Minneapolis.
Patterson111, W. A. andBackman, A. S., 1988.Fire and diseasehistory of forests.In Huntley,B. and Webb,T. III (eds.), VegetationHistory. Dortrecht,Kluwer Academic Publishers, 603-632.
PattersonIII, W.A., Edwards,K. J. andMaguire, D. J., 1987.Microscopic Charcoal as a FossilIndicator of Fire. In QuaternaryScience Reviews 6,3-23.
Peacock,J. D., 1983.Quaternary Geology of the Inner Hebrides.In Procce(lingsofthc RoyalSociety ofEdinburgh 8313,83-90.
Pennington,W., 1947.Studies of the Post-glacialhistory of British VegetationVII. Lake Sediments:Pollen diagramsfrom the bottomdeposits of the North basinof Lake Windermere.In PhilosophicalTransactions of theRoyal Society ofLondon, 233 B596, 137-175.
Pennington,W., 1973.Absolute pollen frequeciesin the sedimentsof lakesof different morphometry.In Birks, H.J. B. andWest, R. G., QuatemaryPlaid Ecology.Oxford: Blackwell ScientificPublications, 79-104.
Pennington,W., 1979.The origin of pollen in lakc sediments:an enclosedlakc compared with one receivinginflow streams.New Nytologist 83,189-213.
Pennington, W., Haworth, E.Y., Bonny, A. and Lishman, J.P., 1972. Lakc Sedimentsin Northern Scotland. In Philosophical Transactiom ofthe Royal Society ofLondon Series B, 264 B. 861,191-295.
236 Pilcher,J. R., Smith,A. G., Pearson,G. W. andCrowder, A., 1971. Land clearancein the Irish Neolithic: new evidenceand interpretation. Science 172,7/5/71,560-562.
Prentice, I. C-t 1988. Recordsof Vegetation in time and space:the principles of pollcn analysis. In Huntley, B. and Webb, T. RL (cds.), VegetationIfistory. Dortrccht: Kluwer Academic Press, 17-42.
Price,R. J., 1983.Scotland's Environnient during the last 30,000years.Edinburgh: ScottishAcademic Press.
Rhodes,AN, Rumsby,B. T. andMacklin, M. G., 1992.Gallanach Beg, Oban.In Walker, M. J.C., Gray, J.M. and Loe, J.J., (eds.), Thesouth-west Scottish Highlands: Field Guide. Cambridge:Quaternary Research Association, 121-125.
Rhum ReserveOffice, 1996. WoodlandRestoration on Rhum, Factshect,Isle of Rhum National Nature Reserve.
Ritchie, J.C., Cwynar,L. C. and Spear,R. W., 1983.Evidence from north west Canadafor an early HoloceneMilankovitch thermalmaximum. Nature 305,126-128.
Ritchie, G. andHarman, M., 1996.Argyll andthe WesternIsles. Edinburgh: HMSO.
Robins,P, 1998,A new flint assemblagefron CnocDubh, Dcrvaig,Mull NM 40925311. Historic Argyll, Lorn.Archaeological and Historical Society,3,15.
Robinson,D., 1983.Possible Mesolithic activity in the west of Arran: evidencefrom peat dcposits.Glasgow Archaeological Journal, 10,1-6.
Robinson,D., 1984.The estimationof the charcoalcontent of sediments:a comparisonof methodson peatsections from the Islandof Arran. In Circaea2.
Robinson,D. E. andDickson, J. H., 1988.Vegetational history andland use:a radio- carbondated pollen diagramfrom MachrieMoor, Arran, Scotland.New Phytologist 109, 223-251.
Rorslett,B. andJohansen, S. W., 1995.Dynamic response of the submergedmacrophyte, Isocteslacustris, to alternatinglight levelsunder field conditions.Aquatic Botany 5 1, 223-242.
Rowley, K. R., Rowley, A. and Skvarla, JI, 1990, Corroded exines from Havinga's Leaf Mold Experiment. Palynology 14,73 - 79.
Rummcry, T. A., 1983. The use of magnetic mcasurcmcntsin interpreting the fire histories of lakcs. In Hydrobiologia 103,53 - 58.
237 Russell,N. J., Bonsall,C. andSutherland, D. G., 1995.The exploitationof marine molluscsin the Mesolithic of westernScotland: evidence from Ulva Cave,Inner Hcbrides.In Fischcr,A. (ed.), Man and Seain theMesolithic. Oxford: OxbowMonograph 53.
Rymer,L., 1977.A Late-glacialand early post-glacial pollen diagramfrom Drimnagall, North Knapdale,Argyllshire. New Phytologist, 79,211-22 1.
SAN, 1998.The earliestsite in Scotland?Scottish Archaeological News 28,1.
Sangster,A. G. and Dale, H. M., 1961. A preliminary study of differential pollen grain preservation. Canadian Journal ofBotany 39,3 5- 43.
Sangster,A. G. andDale, H.M., 1964,Pollen grainspreservation of underrepresented speciesin fossil spectra.Canadian Journal ofBotany 42,437 - 449.
Saville, A., 1996. Lacaille, Microliths and the Mesolithic of Orkney. In Pollard, T. and Morrison, A., The Early Prehistory of Scotland. Edinburgh: Edinburgh University Press, 213-224.
Saville,A., 1998.Studying the Mesolithic periodof Scotland:A bibliographicgazeteer. In Ashton,N., Healy, F. andPettitt, P. (eds.), StoneAge Archaeology: Essays in honour ofJohn Wymer.Oxford Monograph102/Lithic Studies Society Occasional Paper 6.
Scaife,R., 1986.The Elm Declinein the pollen recordof southeast England and its relationshipto early agriculture.In Jones,M. (ed.), Archaeologyand Flora of the British Isles. Oxford: Oxford University Committeefor Archaeology.
Simmons,I. G. andInnes, JR, 1981.Tree remains in a North York Moors peatprofiles. Nature294,5/11/81,76-78.
Simmons,I. G. andInnes, JR, 1987,Mid-Holocene adaptations and laterMesolithic forestdisturbance in northernEngland. Jounial ofArchaeologicalSciencc 14,385-403.
Simmons,I. G. andInnes, J. B., 1988.Late Quaternaryvegetational history of the North York Moors VIII. Correlationof FlandrianII litho- andpollcn stratigraphyat North Gill, GlaisdalcMoor. Journal ofBiogeography15,249-272.
Simmons,I. G. and Innes,J. B., 1988.Late Quaternaryvegetational history of the North York Moors IX Numericalanalysis and pollen concentrationanalysis of Flandrian11 pcat profiles from North Gill, GlaisdaleMoor. Journal ofBiogeography15,273-297.
Simmons,I. G. andTooley, M. J., (eds.), 1981.The Environinvit itsBritish Prehistory. London:Duckworth.
238 Smith, A. G., 1970.The influenceof Mesolithic andNeolithic man on British Vegctation. In Walker, D. andWest, R. G., (cds.), Studiesin the VegetatiotialHistory ofthe British Isles. London: CambridgeUniversity Press,81-96.
Smith, A. G., 1984.Newfcrry andthe Borcal-Atlantictransition. Arew Phylologist 98,35- 55.
Smith, A. G. and Cloutman,EX, 1988.Reconstruction of Holcenevegetation history in thrcc-dimcnsionsat Waun-Figncn-Felcn,an uplandsite in SouthWales. Philosophical Transactionsofthe Royal SocietyofLondon SeriesB, 322,159-2 10.
Stace,C., 1991. New Flora ofthe British Isles. Cambridge:Cambridge University Press.
Stevenson,A. C. and Birks, H. J.B., 1995. Heaths and Moorland: Long-term ecological changes,and interactions with climate and people. In Thompson, D. B. A., Hester, A. J. and Usher, M. B. (cds.), Heaths and Moorland: Cultural Landscapes.Edinburgh, HMSO, 224-239.
Stewart,D. A., Walker,A. andDickson, J. H., 1984.Pollen diagramsfrom Dubh Lochan, nearLoch Lomond.New Phytologist98,531-549.
Stockmarr,J., 1971.Tablets with sporesused in absolutepollen analysis.Pollen ef Spores13,615-62 1.
Stuiverand Reimer,1993. Extended 14C database and revised CALIB3.0 14Cage calibrationprogram. Radiocarbon35,215-230.
Sugdcn,H. andEdwards, K. J., in press.The earlyHolocene vegetational history of Loch a'Bhogaidh,southern Rhinns, Islay, with specialreference to hazel(Corylus avellana L. ). In Mithen, S. (ed.), Hunter-gathererlandscape archaeology: the SouthernHebrides MesolithicProject, 1988- 1998.Cambridge: McDonald Institute for Archaeological Research.
Swain,A. A., 1973.Lake SedimentHistory of Fire. QuaternaryResearch 3,390-396.
Szmeja,J., 1994.Effect of disturbancesond interspecificcompetition in isoetid populations.Aquatic Botany48,225-238.
Tallantire,P. A., 1992.The aldcr (Nnus glutinosa(L-) Gaertn.) problemin the British Isles:a third approachto its palaeohistory.New Phytologist 222,717-731.
Tinsley, H. M. and Smith, R.T., 1974. Surface pollen studies acrossa woodland/heath transition and their application to the interpretation of pollen diagrams.New Phytologist 73,547-565.
239 history deglacial Tipping, R., 1984.Late Devensian- early Flandrian vegetational and chronologyof WesternArgyll. Unpublished:Phd thesis. City of LondonPolytechnic.
Tipping, R., 1987.The origins of corrodedpolicn grainsat five early postglacialpollcn sitesin westernScotland. Review qfPalaeobotany and Palynology53,151-161.
Tipping, R., 1992.Pulpit Hill, Oban.In Walker,M. J. C., Gray,J. M. andLowe, J.J. (cds.), 77teSouth-west Scottish Highlands: Field Guide.Cambridge: Quaternary Research Association.
Tipping, R., 1994. The Fonn and Fate of ScotlandsWoodlands. Procecdings ofthe Society ofAntiquaries ofScotland 124,1-54.
Tipping, R., 1996. Microscopic Charcoal Records,Infcrrcd Human Activity and Climate Changein the Mesolithic of Northernmost Scotland. In Pollard, T. and Morrison, A., (eds.), The Early Prehistory ofScotland. Edinburgh: Edinburgh University Press,39-6 1.
Tolonen, M., 1980. Degradation Analysis of pollen in sedimentsof Lake Lamminjumi, S. Finland. Annales Botanici Fennici 17,11-14.
Turner,J., 1974.Pollen diagrams from CrossFell andtheir implicationsfor former tree lines.In Haworth,E. Y. andLund, M. G. (eds.), Lakc Sedinientsand Environniental History. LeicesterUniversity Press.
Turner,J., 1975.The evidencefor land useby prehistoricfarming communities: the use H. (cds.), of three-dimensionalpollen diagrams.In Evans,J. G., Limbrey, S. andCleere, No. 11. Theeffect ofman on the landscape:the Highland Zone.CBA ResearchReport
Turner,J., 1982.The evidencefor land useby preshistoricfanning communities:the use of three-dimensionalpollen diagrams.In Hands,A. R. and Walker,D. R., Archaeological Aspectsof WoodlandEcology. BAR ReportS 146,8 6-95.
Turner,J., Innes,J. B., and Simmons,I. G., 1993.Spatial diversity in the mid-Flandrian vegetationhistory of North Gill, North Yorkshire.In Nciv Aytologist, 123,599-647.
Vasari,Y., 1977.Radiocarbon dating of the Lateglacialand early Flandrian vegetational successionin the ScottishHighlands and the Isle of Skye.In Gray,J. M. andLowe, J.J. (eds.), Studiesin the ScottishLateglacial Environment. Oxford: PergamonPress, 143 - 162.
Vasari, Y. and Vasari, A., 1968. Late- and post-glacial macrophytic vegctation in te lochs of northcm Scotland.Acta Botanica Fennica 80,1-120.
Vuorela,1., 1973.Relative pollen rain aroundcultivated fields. InAcla BotanicaFennica 102,1-27.
240 Vuorcla, 1., 1980. Microspores of Isocies as indicators of human scttlcrncnt in pollen analysis.Memoranda Societasprofauna etflora Fennica 56,13-19.
Walker,MIC., 1984.A pollen diagramfrom St. Kilda, OuterHebrides, Scotland. New Phytologist,97,99-113.
Walker,MIC. andLowe, U., 1977.Postglacial environmental history of Rannoch Moor, Scotland,I. Threepollen diagramsfrom the Kingliousearea. Journal of Biogeography4,333-35 1.
Walker,MIC. andLowe, JI, 1979.Postglacial environmental history of Rannoch Moor, Scotland,II. Pollen diagramsand radiocarbon dates from the RannochStation and Corrourareas. Journal ofBiogeography6,349-362.
Walker, M. J.C. and Lowe, J.J., 1981. Postglacial environmental history of Rannoch Moor, Scotland III. Early- and Mid-Flandrian pollen stratigraphic data from sites on western Rannoch Moor and near Fort William. Journal of Biogeography 8,47S-49 1.
Walker, M. J.C. and Lowe, J.J., 1985. Flandrian Environmental History of the Isle of Mull, Scotland. I. Pollen-stratigraphic evidenceand radiocarbon dates from Glen More, south-centralMull. In New Phytologist 99,587-610.
Walker,M. J.C. andLowe, J.J., 1987.Flandrian Environmental History of the Isle of Mull, Scotland.III. A high resolutionprofile from Gribun,western Mull. New Phytologist 106,333-347.
Walker,W. C. andLowe, J.J., 1990.Reconstructing the environmentalhistory of the last glacial-interglacialtransition: evidence from the Isle of Skye,Inner Hebrides,Scotland. QuaternaryScience Reviews 9,15-49.
Walker,M. J.C. andLowe, J., 1991a. VegetationalHistory of the Isle of Skye:I. The Late DevensianLateglacial period. In Ballantyne,CK, Benn,D. I., Lowe, J. andWalker, M. J.C., 21e Quaternaryofthe Isle ofSkyeField Guide.Quaternary Research Association:Cambridge.
Walker,M. J.C. andLowe, J., 1991b.Vegetational History of the Isle of Skye:II. The Flandrian.In Ballantyne,CK, Benn,D. I., Lowe, J. and Walker,M. J.C., TheQuatmary ofthe Isle ofSkyeField Guide.Quaternary Research Association: Cambridge.
Walker, M. J.C., Coope, G.R. and Lowe, J.J., 1994. The Devensian(Wcichsclian) latcglacial palaeoenviromncntalrecord from Gransmoor, East Yorkshire, England. Quaternary ScienceReviews 12,659-680.
241 Whitlock, C. andMillspaugh, S. H., 1996.Testing the assumptionsof firc-history studies: and examinationof modern,charcoal accumulation in YellowstoneNational Park, USA. 71c Holocene6,7-15.
Whittington, G. andEdwards, K. J., 1995.A Scottishbroad: historical, stratigraphic and numericalstudies associated with polleniferousdeposits at KilconquharLoch. In Butlin, R.A. andRoberts, N. (eds.), EcologicalRelations in Historical Tinies,The Instituteof British GeographersSpecial Publications Series, 68-87.
Whittington, G., Edwards,K. J. andCundill, P.R., 1991. Late- andpost-glacial vegetationalchange at Black Loch, Fife, easternScotland -a multiple core approach.New Phytologist,118,147-166.
Whittington, G., Fallick, A. andEdwards, K. J., 1996.Stable oxygen isotope and pollen recordsfrom eastern.Scotland and a considerationof Late-glacialand early Holocene climatechange for Europe.Journal of QuaternaryScience 11,327-340.
Whittington, G., andEdwards, K. J., 1997.Climate Cliangc. In Edwards,K. J. and Ralston, L, Scotland.- Environment and History 8000BC-1000 AD. Chichester:John Wiley and Sons,11-22.
Wickham-Jones,C. R., 1990a.Recent work on the islandof Rhum, Scotland.In Bonsall, C., (ed.), Ae Mesolithic in Europe.Edinburgh: John Donald Publishers Ltd.
Wickham-Jones,C. R., 1990b.Rhum: Mesolithic and later sitesat Kinloch, Excavations 1984-1986,Society of Antiquaricsof ScotlandMonograph Series 7.
Wickham-Jones,C. R., 1994.Scotlands First Settlers.London; Batsford Ltd. (Historic Scotland).
Wickham-Jones,C. R. andDalland, M., 1998.A small Mesolithic site at Fifeness,Fife, Scotland.http: //l*iitarch. ac. ukljournallissucSlwickliai? i-itidex. htnil
Wilkinson, D.M. andHuntley, B., 1988.Differential pollcn preservationin a topogcnous mire by High Force,Teesdale, northern England. In Pollen ct Spores66,455-460.
Willemse,N. W. andT6mqvist, T.E., 1999.Holocene century-scale temperature variability from West Greenlandlake records.Geology 27,580-584.
Williams, W., 1977. Ae Flandrian vegetational history ofthe Isle ofSkye and the Aforar Peninsula. Unpublished PhD thesis, University of Cambridge.
Willis, K. J., Bennett,K. D., andBirks, H.J. B., 1998,Late Quaternarydynamics of pines in Europe.In Richardson,D. M. (ed.), Ecologyand BiogeograpliyofPinus, 107-121.
242 Wiltshire, P.E. J., Edwards,K. J., andBond, S., 1994.Microbially-defived MetaIVIC sulphidespherules, pollen andthe waterloggingof archaeologicalsites. Association of AmericanStratigraphic Palynologists Contributions Series 29,207-22 1.
Wood, G.D., Gabriel,A. M. andLawson, J. C., 1996.Palynolgical techniques - processing andmicroscopy. in Jansonius,J. andMcGregor, D. C. (eds.), Palynology:principles and applications.American Association of StratigraphicPalynologists Foundation, Publishers Press,Salt Lake City, 1,29-50.
Woodman, P.C., 1990. A review of the Scottish Mesolithic: a plea for normality. Proceedings ofthe Society ofAntiquaries ofScotland 119 (1989), 1-32.
Woolley, A. R. and Jermy, A. C., 1978. Geology. 4.1-4.25. In Jermy, A. C. and Crabbe, J.A. (eds.), The Island ofMull: a survey of itsflora and environment.London: British Museum (Natural History).
Wormell, P., 1970.Establishing woodland on the Isle of Rhum,Scottish Forestry 22, 207-220.
Zagwijn, W.H., 1994.Reconstruction of climatechange during the Holocenein westem andcentral Europe based on pollen recordsof indicator species.Vegetation History and Archaobotany3,65-68.
243