The Biostratigraphy, Palaeoecology and Geochemistry of a 
Long Lacustrine Sequence from NW Greece 
Dissertation submitted to the University of Cambridge 
for the degree of Doctor of Philosophy 
by 
Michael Reginald Frogley 
Fitzwilliam College 
September, 1997 
DECLARA TION 
I hereby declare that this thesis is not substantially the same as any submitted for a 
degree or diploma or any other qualification at any other University. No part of this 
work has already been or is being concurrently submitted for any degree, diploma 
or other qualification. I further state that the text of this dissertation does not exceed 
80,000 words. 
Except where otherwise stated, this dissertation is the result of my own original 
work and is not the outcome of work done in collaboration. 
Michae1 R Frogley 
September, 1997 
ABSTRACT 
Examination of an impOltant new 319m core of lake sediment recovered from Ioannina 
in NW Greece has attempted to relate changes in the lake to variations in the regional 
climate of south-central Europe over the last 600,000 years. The site is known to have 
been extremely sensitive to past climatic change for three reasons: (i) temperate 
vegetation persisted throughout glacial stages (albeit at low frequencies), so the 
vegetational response to climatic change would therefore have been almost immediate; 
(ii) the extreme thickness of the sediments suggests that accumulation rates were high 
(at times, > 1m per thousand years), which has enabled high-resolution palaeoclimatic 
reconstructions; and (iii) precipitation of authigenic carbonate has preserved a 
remarkably sensitive proxy record of productivity variations for most of the lake's 
history. Well-defined shifts from glacial - interglacial mode have been correlated with 
vegetational changes identified in a core previously analysed from the same basin (using 
magnetic susceptibility profiles), enabling tentative correlations to be suggested with 
other European terrestrial sequences and with the marine oxygen isotope record, back to 
marine isotope stage 16. Twelve AMS radiocarbon determinations from the upper part 
of the core, together with the identification of a series of reversed palaeomagnetic events 
within the Brunhes chron, support the proposed age model for the sequence. 
The sediments at Ioannina, unlike most of the other long terrestrial European 
sequences, are calcareous and contain mollusc and ostracod assemblages. Part of this 
project has involved a comprehensive review of Quaternary and modem aquatic faunas 
from the lake, as well as the description, illustration and critical assessment of several 
poorly-known endemic taxa. Faunal assemblage data have been used to provide 
valuable information concerning the variable response of lake-level to climatic change 
over time. Convincing new mollusc an evidence indicates low lake-levels at the Last 
Glacial Maximum, agreeing with regional pollen data, but conflicting with 
geomorphological evidence derived from Kastritsa, a well-documented nearby 
Palaeolithic cave site. It is suggested that this discrepancy may be a result of 
subsequent tectonic uplift of the rockshelter. In addition, stable isotopic analyses of 
both the ostracods and the bulk carbonate within the sediments have contributed 
towards deriving a comprehensive palaeoenvironmental history for the site. 
Although the study analysed physical, biological and geochemical aspects of the entire 
core, two distinct parts of the record were selected for more detailed investigation. 
High-resolution analysis over the last interglacial (the Eemian) has revealed evidence for 
a clear, two-step deglaciation at the beginning of the period, known from elsewhere as 
the Zeifen-Kattegat Oscillation. Climatic instability has also been detected within the 
full interglacial. Comparisons are drawn with a range of other Eemian records from 
across Europe, as well as the Greenland ice cores. High-resolution analysis of the 
period from the end of the last glacial to the present day has also revealed evidence for 
climatic instability. A cool and arid oscillation is demonstrated by several climatic 
proxies that may constitute the first recognition of the Younger Dryas stadial from 
Greece. A shorter, but more subdued cooling event has also been detected during the 
first half of the Holocene, which may correspond with a widespread climatic oscillation 
from high-resolution terrestrial, marine and ice core records that has been dated to 
between 7,500 and 8,000 years BP. 
ACKNOWLEDGEMENTS 
This project would not have been possible without the support, encouragement and 
enthusiasm of my supervisor, Richard Preece, to whom I extend my most sincere 
thanks. Over the past four years he has successfully walked a fine supervisorial 
line, being neither too overbearing nor too inaccessible. His constant stream of 
ideas, considered criticism and good humour (even down to his continued interest 
in the molluscan content of the core when a raging field-fire was advancing upon us 
with alarming rapidity), have provided inspiration even in the darkest moments. 
On embarking upon this project, I was warned that trying to please two supervisors 
was probably going to prove the most difficult aspect of all. Thankfully, the 
tolerance and patience of my second supervisor, Jonathan Holmes, has meant that 
this has never been a problem. His constant enthusiasm for the project has been 
matched only by his enthusiasm for e-mail, meaning that despite him being based in 
Kingston, I have been able to call on his advice with immediate ease. 
Additional thanks must also go to Chronis Tzedakis, not only for demonstrating the 
original potential of the loannina sequence and thereby making this project possible, 
but also for taking on the effective role of an unofficial third supervisor. His 
infectious wit and good humour have been ably complemented by his constant 
encouragement, broad knowledge and liking of al fresco lunches. 
I would also like to take this opportunity to acknowledge the Director of IGME, for 
making the core material available for study. In particular, the advice of Y. 
Broussoulis and the assistance given by the loannina staff was invaluable. 
The multi-disciplinaly nature of this project has meant that expert advice and 
assistance has been generously provided by a wide range of people and Institutions 
(although any errors or omissions herein remain entirely my own responsibility). 
The following deserve special mention (in no particular order): Tim Heaton (NERC 
Isotope Geosciences Laboratory, Keyworth) for undertaking all the stable isotopic 
measurements (lightning strikes notwithstanding); Liping Zhou (GIQR, University 
of Cambridge) for help with the magnetic susceptibility measurements; Barbal'a 
Maher and team (UEA, Norwich), for providing palaeomagnetic facilities; Huw 
Griffiths (University of Hull), for patiently helping with the identification of Balkan 
ostracods and for allowing me to mine his collection of obscure Serbian papers; Tim 
Atkinson and Peter Rowe (UEA, Norwich), for undeltaking the U-series 
determinations; Glenn Goodfriend (University of Washington, USA) for 
undertaking the amino acid analyses (and for providing a stream of cheering 
anecdotes whilst being stranded overnight in a coach in a Turkish blizzard); Piene 
and Nicole Lozouet for discussion, hospitality and fine wine whilst examining 
mollusc an specimens in Paris; Tom Meijer (RGD, Netherlands) for additional 
molluscan help; Tjeerd van Andel (GIQR, University of Cambridge), for advice, 
encouragement and for providing a pre-print of his paper on Greek soils; Charles 
Turner (Open University), for introducing me to the sights and sites of Epirus; Phil 
Gibbard (GIQR, University of Cambridge) and Jamie Woodward (University of 
Leeds), for discussion and the retrieval of bedrock samples from Ioannina; James 
Scourse (University of Wales, Bangor) for support, advice and timely criticism; 
Mike Hall (GIQR, University of Cambridge), for being persuaded to run several 
pilot isotopic samples in the Godwin Lab; Geoff Bailey (University of Newcastle 
Upon Tyne), for access to the Kastritsa samples; Virgin Airways, for providing the 
means to get the core material back to the UK without bankrupting NERC; Steve 
Boreham (GIQR, University of Cambridge), for help with the particle-size 
analyses; Uli von Grafenstein (Technische UniversiUit, Mtinchen), for enthusiasm, 
encouragement and access to unpublished ostracod isotopic data; David Pyle 
(GIQR, University of Cambridge), for providing unpublished information on the 
Morphi tephra; Simon Mansell, for providing floorspace whilst working in the 
magnetics lab at UEA; and Bill Lee and John Rodford (both Department of 
Zoology, University of Cambridge), for patient work on the SEM and in scanning 
and preparing the photographic plates, respectively. Thanks must also go to Ray, 
Yvonne, Monica and Anne (Museum of Zoology), for providing technical support. 
The constant encouragement and support of good friends have ensured that my time 
in Cambridge has been extremely happy. Thanks must go to tolerant lab colleagues 
Rich Meyrick (for graciously allowing me to dig holes for him in unlikely European 
locations) and Dave Home (for a great Wallace impression). Special thanks must 
also go to Rich for being particularly supportive during the bloody, final days of 
writing-up. Chris Fogwill, Glenn Tattersall and Chris Glaister proved most 
excellent drinking companions, as did Dave Aldridge, Eleanor Weston and a host of 
other Zoology reprobates. The loyal Kingston lads also deserve particular· mention, 
for providing a regular· safe haven for when Cambridge became too much. 
A NERC Studentship is gratefully acknowledged. Additional financial support was 
provided by the Department of Zoology and Fitzwilliam College. A Resear·ch 
Fellowship at St John's College is also gratefully acknowledged, for providing me 
with the time, facilities and a suitable environment in which to complete this report. 
CONTENTS 
1. INTRODUCTION 
1.1 Background 1 
1.2 Ioannina 3 
1.3 Project Aims 4 
1.4 Rationale of this Study 5 
1.4.1 The Last Interglacial 5 
1.4.2 The Last Glacial - Holocene Transition 5 
1.5 Organisation of this Report 6 
1.6 Conventions 7 
2. PHYSICAL SETTING 
2.1 Study Site 9 
2.2 Regional Geology 9 
2.3 Basin Evolution 13 
2.4 Climate 15 
2.5 Basin Hydrology 17 
2.6 The Modern Lake Basin 18 
2.7 Summary 19 
3. METHODS 
3.1 Introduction 21 
3.2 Physical Analyses 21 
3.2.1 Core Recovery and Logging 21 
3.2.2 Magnetic Susceptibility 22 
3.2.3 Palaeomagnetic Analysis 23 
3.2.4 Particle-Size Analysis 27 
3.2.5 SEM Analysis 28 
3.2.6 Modern Faunal Analysis 28 
3.2.7 Fossil Faunal Analysis 28 
3.3 Chemical Analyses 30 
3.3.1 Sediment Geochemistry 30 
3.3 .2 Modern Water Chemistry 35 
3.3.3 Ostracod Shell Chemistry 36 
4. RESULTS 
4.1 Physical Analyses 40 
4.1.1 Lithological Description 40 
4.l.2 Magnetic Susceptibility 42 
4.l.3 Particle-Size Analysis 46 
4.1.4 SEM Analysis 49 
4.2 Chemical Analyses 50 
4.2.1 Sediment Geochemistry 50 
4.2.2 Modern Water Chemistry 56 
4.2.3 Ostracod Shell Chemistry 58 
4.3 Summary 63 
5. 
6. 
7. 
CHRONOLOGY 
5.1 Introduction 65 
5.2 Correlation with Core 249 65 
5.2.1 Basis of the Age Model for 249 66 
5.2.2 Correlation of 284 with 249 70 
5 .3 Radiocarbon Age Determinations 74 
SA Magnetostratigraphy 80 
504 .1 Gothenburg excursion 86 
504 .2 Mono Lake excursion 86 
504 .3 Laschamps excursion 87 
50404 Norwegian/Greenland Sea and Fram Strait excursions 87 
5.4.5 Blake event 88 
504 .6 Jamaica event 89 
504 .7 Pringle Falls excursion 90 
504 .8 Levantine event 90 
504.9 Biwa III excursion 90 
504.10 Emperor excursion 92 
504.11 Big Lost excursion 92 
5.5 Amino Acid Epimerization Data 93 
5.6 Uranium-Series Dating 95 
5.7 Tephra 97 
5.8 Summary and Proposed Age Model 99 
FAUNAL BIOSTRATIGRAPHY AND MODERN ASSEMBLAGES 
6.1 Introduction 
6 .2 Aquatic Molluscan Fauna 
6 .2.1 Systematic Review of the Aquatic Molluscan Fauna 
6 .2.2 Modern Aquatic Molluscan Fauna 
6.2.3 Fossil Aquatic Molluscan Fauna 
6.204 Aquatic Molluscan Fauna from Core 256 
6 .3 Ostracod Fauna 
6.3.1 Taxonomy 
6.3 .2 Systematic Review of the Ostracod Fauna 
6.3.3 Modern Ostracod Fauna 
6.304 Taphonomy 
6 .3.5 Fossil Ostracod Fauna: Last Glacial - Holocene 
6 .3.6 Fossil Ostracod Fauna: Eemian 
6 .3 .7 Other Intervals of 284 
604 Summary 
DISCUSSION 
7.1 
7 .2 
7 .3 
704 
Introduction 
Sediment Analyses 
7.2 .1 Carbonate Content 
7 .2.2 Organic Matter Content 
7 .2 .3 Magnetic Susceptibility 
Faunal Analyses 
7.3 .1 Lake-Levels 
7.3 .2 The Kastritsa Problem 
Chemical Analyses 
704.1 Modem Waters 
704.2 Sediment Isotope Geochemistry 
704 .3 Ostracod Shell Chemistry 
102 
103 
108 
122 
124 
126 
128 
128 
129 
142 
143 
145 
149 
153 
154 
155 
155 
156 
158 
159 
161 
161 
162 
164 
164 
165 
169 
7.4.4 Open or Closed Lake System? 
7 .5 Synthesis 
7.5.1 Eemian 
7.5.2 MIS-5d - Last Glacial Maximum 
7.5.3 Last Glacial - Holocene Transition 
7.5.4 Response Times 
7.6 Summary 
8. COMPARISON OF 284 WITH OTHER LONG RECORDS 
9. 
8.1 
8.2 
8.3 
8.4 
8.5 
8.6 
Introduction 
TelTestrial Sequences 
Marine Record 
Ice Cores 
Rapid Climate Fluctuations 
8.5.1 The Eemian 
8.5.2 The Montaigu Event 
8.5.3 The Last Glacial 
8.5.4 The Last Glacial- Holocene Transition 
Summary 
CONCLUSIONS 
REFERENCES 
PLA TES I - VII 
APPENDICES 
Appendix A: 
Appendix B: 
Appendix C: 
Modern Water Chemistry Data 
Sampling Site Details 
Eemian Pollen Diagram from Core 284 
172 
175 
176 
180 
180 
186 
187 
190 
190 
194 
195 
197 
197 
200 
201 
202 
203 
205 
209 
1. INTRODUCTION 
1.1 BACKGROUND 
Interest in past climatic change has acquired an important profile in recent years. We 
have gradually become aware through issues such as global warming that human impact 
upon the environment can have far ranging consequences. By gaining clearer insights 
into the mechanisms that drove ancient climatic systems we can hopefully achieve a 
greater understanding of how the current climate operates and might then apply such 
understanding in the prediction of future climatic change. 
The study of the Quaternary is a multidisciplinary endeavour largely concerned with 
piecing together the evidence for ancient climatic change that has become preserved in 
oceanic, continental and ice core sequences. Traditionally, the 'yardstick' for the 
Quaternary has been the long continuous record of the deep oceans, where huge 
thicknesses of sediment have accumulated, largely undisturbed by the periodic 
glaciations that have rendered much of the continental sedimentary record highly 
disjointed and fragmentary. Over the last thirty years or so, an effective mm·me 
chronology has been established (the mm·ine oxygen isotope record) that details 
palaeoclimatic variability throughout the Quaternary (and much em·lier periods). 
Recently however, attention has refocused on continental sequences, especially those 
that have accumulated in lake basins beyond the reach of the ice sheets, such as those in 
southern Europe. Under favourable geological conditions, a limited number of thick 
sedimentm), sequences have accumulated at far faster rates than those operating in the 
oceans, so preserving the palaeoclimatic record in much greater detail (fig. 1.1). In 
addition, these lacustrine sequences cover much longer periods of time than most of the 
fragmentm), continental records from northern Europe. One of the fundamental 
challenges now facing Quaternary researchers is the detailed analyses of these long, 
continuous sequences. 
The new core that forms the focus of this study is from Ioannina, a lake basin in 
northwestern Greece. It represents one of these rare, continuous continental southern 
European records and is unusual in terms of its climatic sensitivity as well as its 
physical and strati graphical length. 
1 
1 La Grande Pile 
2 Les Echets 
3 Lac du Bouchet / Prac1aux 
4 Padul 
5 Valle di Castiglione 
6 Monticchio 
7 Ioannina 
8 Tenaghi Philippon 
-' 
Fig. 1.1 Location map of principal long European terrestrial sequences. 
S2 
;;5 
o 
tl 
c:: 
Cl 
:::l 
o 
=<: 
• 
N 
INTRODUCTION • 3 
1.2 IOANNINA 
Previous work on the Quaternary of the loannina region has been rather sparse. The 
first detailed research was conducted in the 1960s by the archaeologist Eric Higgs and 
his co-workers, who largely concentrated their efforts on the Palaeolithic site of 
Kastritsa in the south of the basin (e.g. Higgs and Vita-Finzi, 1966; Higgs et al ., 
1967). It was Bottema (1974) who first demonstrated the potential of the loannina 
sequence. He analysed pollen from two short cores located in the northern prut of the 
basin and showed that they extended back to the last glacial. 
More recently, the area has been extensively cored by the Greek Institute of Geology 
and Mineral Exploration (IGME), who have been investigating the lignite potential of 
this and adjacent basins. They originally drilled almost 300 individual boreholes of 
varying length, although many have since been discarded. From the surviving archived 
material, Tzedakis (1991, 1993, 1994) selected and analysed the pollen record from a 
long core taken from the south of the basin (no. 249, 185m in length, unbottomed) . It 
covered a series of glacial - interglacial cycles during the last 430ka which enabled 
cOlTelation with other European continental sequences and the mru'ine oxygen isotope 
record. In addition, he demonstrated that the continued presence of temperate ru'boreal 
taxa at loannina even during glacial stages, albeit at low frequencies, identified the site 
as a long-term refugial area (e.g. Bennett et aI., 1991; Tzedakis, 1991 , 1993). This is 
cleru'ly extremely important, since the continued presence of temperate taxa not only 
ensures a rapid response to climatic wruming at the beginning of an interglacial 
(minimising migration lags which ru'e seen elsewhere), but also confirms the extreme 
sensitivity of the region to climatic variability. This work further highlighted the 
enormous potential of the loannina record and paved the way for this new project. 
The original intention had been to analyse other climatic proxy indicators at high­
resolution from core 249 and compare them with the response of the vegetation record. 
Unfortunately, it transpired that the original core material had been inadvertantly 
discarded since the sub-sampling for the palynological investigation of 249 (Tzedakis, 
1991). Attention was therefore focused on two alternative long IGMB cores taken in 
close proximity to 249. These were made available for study and returned to 
Cambridge in 1993. 
The longest, core 284, was located less than half a kilometre to the southwest of 249 
and at 319m unbottomed length, was over lOOm deeper and therefore likely to have a 
longer stratigraphical range. Core 284 appears to have formed continuously in a part of 
INTRODUCTION • 4 
the basin that was, until recently, always submerged. Core 256, located approximately 
6km to the southwest of 284, reached bedrock at 130m. It is thought to represent a 
more marginal environment, a conclusion based on its geographical location, the nature 
of its lithologies and the presence of a series of hiatuses within the record. Preliminary 
palaeontological and palaeomagnetic work has shown that it covers a longer time-span 
than 284 (the bottom 60m is palaeomagnetically reversed). However, because of the 
gaps in the record, there is uncertainty over its exact chronology. It is therefore 
considered unsuitable for additional detailed analysis and will not be considered further. 
As a result, the work presented in this report is entirely based on the detailed analysis of 
core 284. 
1.3 PROJECT AIMS 
The principal aims of this study can be summarised as follows: 
• To establish a robust chronology for the entire sequence that can be verified by 
independent means. 
• To reconstruct the palaeoclimatic history of the Ioannina region by means of high­
resolution, multi-proxy analyses that complement and can be correlated with the 
exisiting pollen stratigraphy from the adjacent 249 sequence. 
• To investigate the evidence for high frequency climatic variability through selected 
parts of the record. 
• To use the high-resolution data as a means of assessing the variable response times 
('leads and lags') between pollen and other proxies to climatic change. 
• To undertake a comprehensive study of the modern and fossil mollusc an and 
ostracod faunas in order to assess the possibility of using faunal changes as 
biostratigraphical markers and to provide detailed palaeoecological information on 
lake history. 
• To assess the importance of the sequence in relation to other long European 
terrestrial records, the marine record and the ice cores and to suggest correlations 
where appropriate. 
• To assess the potential for future work on the sequence. 
INTRODUCTION. 5 
1.4 RATIONALE OF THIS STUDY 
The extreme physical length of a sequence such as Ioannina 284 presents celtain 
logistical problems with regard to formulating an efficient programme of study. The 
approach taken in this case was to apply a series of analyses to the entire sequence, in 
order to characterise the overall stmcture of the sequence and to help derive a 
chronological framework. Two selected parts of particular interest were then chosen 
for high-resolution investigation. These were the last interglacial (the Eemian) and the 
last glacial - Holocene transition. 
1.4.1 The Last Interglacial 
Known as the Eemian in the northwest European continental record, and marine oxygen 
isotope stage (MIS) 5e in the oceanic record, the last interglacial began about 130,000 
years ago and lasted approximately 10 - 13,000 years (e.g. Imbrie et al. , 1984). Until 
recently, it has been considered a relatively stable, warm period, with temperatures only 
slightly higher than at present. However, recent high-resolution isotopic evidence from 
the Greenland Ice-core Project (GRIP) has suggested that this period was in fact 
punctuated by a series of high frequency climatic oscillations (GRIP members, 1993) . 
This suggestion has prompted much scientific debate (e.g. Grootes et al., 1993; Bender 
et al., 1994; 10hnsen et al., 1995), particularly as another Greenland ice core (GISP2), 
located less than 30krn to the west of the GRIP site, has failed to corroborate evidence 
for these oscillations. Researchers have so far been unable to confirm unambiguously 
the existence of these climatic fluctuations in either marine or continental sequences. 
One of the fundamental aims of this project is therefore to examine the last interglacial at 
Ioannina for evidence of rapid climatic oscillations. 
1.4.2 The Last Glacial - Holocene Transition 
Following the height of the last glacial (the 'Last Glacial Maximum', about 20ka BP), 
the Earth's climate steadily warmed until about 11ka BP, when there. was a short-lived 
but intense cooling event, known as the Younger Dryas (YD). This lasted for about 
1,000 years, during which temperatures returned to almost full glacial values. At about 
lOka BP, global temperatures subsequently climbed steadily and rapidly, and the 
Earth's climate entered the current interglacial (the Holocene). Although conditions 
have remained relatively stable during this time, several rapid climatic oscillations have 
INTRODUCTION • 6 
been recorded, in particular a brief cooling event during the first half of the Holocene 
(about 7.5-8ka BP) that may be global in exent (e.g. Alley et al., 1997). 
The YD was originally thought to be solely a North Atlantic phenomenon, but 
subsequent records from elsewhere (e.g. nOlthern Europe, Greenland, the Pacific, 
Asia, North and South America) have established that it was global in extent (Frakes et 
al., 1992). According to Willis's (1995) review of Balkan vegetational history, no 
clear and unequivocal YD signal has yet been detected from Greek pollen records, 
which she explains might be due to elevational factors. Although the pollen record 
from core 249 did not reveal the presence of the YD (probably because the uppermost 
Palt of the record was missing), Tzedakis (1993) noted that the Ioannina site appeared 
to be palticularly sensitive to climatic change (the reasons for which have been outlined 
above). A brief intra-Holocene cooling event at around 7.5ka BP has previously been 
recorded in Greek pollen sequences and linked to the abmpt disappeal'ance of Pistacia 
from the vegetational record (e.g. Rossignol-Strick, 1995). This project will therefore 
attempt to establish the existence of both the YD and evidence for intra-Holocene 
climatic instability in the Ioannina record. Additionally, an assessment of the timing 
and nature of any anthropogenic (i.e. human) disturbance in the region will be 
attempted. 
1.5 ORGANISATION OF THIS REPORT 
This report is arranged into chapters as follows: 
• Chapter 2 Introduces the study site and presents details of its physical setting 
with regal'd to regional geology, basin evolution, climate and hydrological factors. 
The chapter concludes by considering aspects of the modern lake basin. 
• Chapter 3 Presents detailed accounts of the physical, chemical and faunal 
analytical methods employed during this project. 
• Chapter 4 Presents the results from the chemical and physical analyses of the 
sediments. 
• Chapter 5 Outlines the different methods by which a chronological framework 
has been derived for the sequence and concludes by presenting a detailed age model 
that is used throughout the rest of the study. 
• Chapter 6 Provides details of the modern and fossil molluscan and ostracod 
faunal analyses. Previous faunal work is reviewed and updated. The importance 
INTRODUCTION • 7 
and potential of the fauna for biostratigraphical and palaeoenvironmental purposes is 
assessed. 
• Chapter 7 Discusses and interprets the results with particular reference to the way 
in which lake systems operate. A synthesis of all results is then presented in order 
to derive a comprehensive palaeoenvironmental history of the lake basin. 
• Chapter 8 The Ioannina sequence is assessed in relation to other long terrestrial, 
marine and ice core records. Potential correlations are suggested where appropriate. 
• Chapter 9 Draws together the main conclusions of the project and suggests 
possible avenues for future research. 
1.6 CONVENTIONS 
Since there is no detailed and widely accepted stratigraphical scheme for southern 
Europe, reference is occasionally made to the local scheme proposed by Tzedakis 
(1991) from the Ioannina 249 sequence, although this is highlighted wherever 
appropriate. Although not applicable in the strictest stratigraphical sense, the last 
interglacial is referred to throughout as the Eemian. This convention is for the sake of 
convenience, reflecting the broad, informal use that the name currently enjoys 
throughout the literature. The terms 'glacial' (or 'cold stage') and 'interglacial' (or 
'warm stage') are often used informally to denote periods of relatively cold or warm 
(temperate) climatic conditions, respectively. 
Radiocarbon dates or ages are quoted as uncalibrated years before 1950 AD (BP). 
Otherwise, dates or ages are given as years ago, where the abbreviations Ma and ka are 
used to represent millions and thousand of years ago, respectively. Other abbreviations 
in the text are explained in full at their first usage: examples include Younger Dryas 
(YD); Last Glacial Maximum (LGM); marine isotope stage (MIS); Zeifen-Kattegat 
Oscillation (ZKO); and dissolved inorganic carbon (DIe). Greek symbols used to 
denote certain quantities or parameters (for example, the standard delta notation for 
representing isotopic ratios) are explained at their first usage. 
Geographical place names present celtain problems, in that there usually exist several 
possible translations from Greek into English (for example, Ioannina is often referred to 
as Janina, Jannina or Yannina). Throughout this report therefore, one spelling is used 
consistently throughout. Where appropriate, Yugoslavia is taken to consist of Serbia 
and Montenegro. 
INTRODUCTION • 8 
The extended character set used in the production of this report does not permit the use 
of certain accented letters that are common in the Serbian and Croatian languages (for 
example, an's' or a 'c' beneath an inverted caret 1\ symbol). In such cases, the un­
accented English equivalent is given throughout (including the references). 
2. PHYSICAL SETTING 
2.1 STUDY SITE 
Lake Pamvotis (470m above sea-level) is situated in the Ioannina basin within the 
region of Epirus in north-western Greece. The site is situated on the western flank of 
the Pindus Mountain Range, approximately 60km from the Ionian coast (fig. 2.1). The 
basin extends from Protopappa in the north-west to Ampelia in the south. It is 
approximately 30km in length and approximately 15km at its maximum width. 
During the late 1980s, the Institute of Geology and Mineral Exploration (IGME) 
investigated the Ioannina basin for commercially exploitable lignite deposits. Other 
basins in the Pindus had previously yielded economically viable deposits and the 
Ioannina basin was also considered worthy of detailed investigation. Almost three 
hundred cores were eventually drilled across the basin. Unpublished reports by IGME 
record the presence of several thin, discontinuous lignite beds that have been broken up 
by the sub-basin tectonics (see below). They are currently not considered worthy of 
of 
commercial exploitation becauseftheir depth and severely limited thickness. 
The core that forms the focus of this study is number 284 (39°45'N 20°51 'E, 472.69m 
above sea-level). It was selected on the basis of several criteria that included continuity, 
condition, overall length and proximity to core 249 (section 1.2). At 319m total length, 
core 284 is one of the longest cores recovered from the Ioannina basin, but despite this 
it still does not reach bedrock; drilling was ceased when it became obvious that no 
lignites were likely to be encountered. Subsequent IGME stratigraphical reports 
(unpublished) show that the core had been drilled in one of the deepest sub-basins, 
making it an ideal candidate for this particular study. Located less than a kilometre from 
core 249, it also appeared to be free from obvious lithological breaks and had been 
preserved in a reasonable condition. It is described in considerably more detail in 
section 4.1.1. 
2.2 REGIONAL GEOLOGY 
The Ioannina basin is a synclinal structure that was formed during tectonic movement in 
the Ionian isopic zone, prut of the broad external zone of the Hellenides (Aubouin, 
2. PHYSICAL SETTING 
2.1 STUDY SITE 
Lake Pamvotis (470m above sea-level) is situated in the Ioannina basin within the 
region of Epirus in north-western Greece. The site is situated on the western flank of 
the Pindus Mountain Range, approximately 60km from the Ionian coast (fig. 2.1). The 
basin extends from Protopappa in the north-west to Arnpelia in the south. It is 
approximately 30km in length and approximately 15km at its maximum width. 
During the late 1980s, the Institute of Geology and Mineral Exploration (IGME) 
investigated the Ioannina basin for commercially exploitable lignite deposits. Other 
basins in the Pindus had previously yielded economically viable deposits and the 
Ioannina basin was also considered worthy of detailed investigation. Almost three 
hundred cores were eventually drilled across the basin. Unpublished reports by IGME 
record the presence of several thin, discontinuous lignite beds that have been broken up 
by the sub-basin tectonics (see below). They are cUlTently not considered worthy of 
of' 
commercial exploitation becauseftheir depth and severely limited thickness. 
The core that forms the focus of this study is number 284 (39°45'N 20°51 'E, 472.69m 
above sea-level). It was selected on the basis of several criteria that included continuity, 
condition, overall length and proximity to core 249 (section 1.2). At 319m total length, 
core 284 is one of the longest cores recovered from the Ioannina basin, but despite this 
it still does not reach bedrock; drilling was ceased when it became obvious that no 
lignites were likely to be encountered. Subsequent IGME stratigraphical reports 
(unpublished) show that the core had been drilled in one of the deepest sub-basins, 
making it an ideal candidate for this particular study. Located less than a kilometre from 
core 249, it also appeared to be free from obvious litho logical breaks and had been 
preserved in a reasonable condition. It is described in considerably more detail in 
section 4.1.1. 
2.2 REGIONAL GEOLOGY 
The Ioannina basin is a sync1inal structure that was fOlmed during tectonic movement in 
the Ionian isopic zone, part of the broad external zone of the Hellenides (Aubouin, 
Kato Lapsista 
Extent of lacustrine 
deposits in basin 
o Skm 
PHYSICAL SETTING • 10 
N 
I 
Fig. 2.1 Location maps illustrating critical places and core sites referred to in the text. 
N 
t 
o 
ke 
PHYSICAL SE1TING • 11 
Skm 
+ 284 Core Location 
Alluvial Fans 
Quaternary Deposits 
Flysch 
Palaeogene Limestone 
Upper Senonian Limestone 
Vigla Limestone 
Posidonia Beds 
Pantokrator Limestone 
Fig. 2.2 Summary geological map of Ioannina basin (after IGSR-IGP, 1966 and Perrier et al. , 
1967). Inset shows isopic zones of Hellenides (after Aubouin, 1959). 
PHYSICAL SEITING • 12 
1959). These zones are generally thrust-bound and composed of similar facies (fig. 
2.2). Mesozoic carbonates were defOlmed during a period of thmsting known 
informally as the Hellenide contractional event (related to the Alpine Orogeny further 
west) during the mid-Eocene - Oligocene. A westward-migrating thrust sheet overrode 
the Gavrovo zone to the east on the Pindus Thmst and folded the Ionian zone into a 
series of north north-east trending anticlines and synclines (Clews, 1989). The result 
was the fOlmation of the Pindus Mountains and an associated foreland thmst basin to 
the west. During the Oligocene - Lower Miocene, erosion of the newly formed Pindus 
resulted in a rapid influx of clastic sediments into the foreland basin, intermpting the 
carbonate sedimentation that had dominated since the early Mesozoic. These ranged 
from coarse, quartz-rich clastics in the east to mads in the west. The sediments have 
been interpreted largely as submarine fan deposits, with palaeocurrent data suggesting 
input from the east (Piper et al., 1978). Upper fan and basin slope deposits dominate in 
the east (Gavrovo zone), whilst further west, more distal turbidite sequences of shales, 
silts and current-rippled sands dominate (Ionian and Pre-Apulian zones). It seems 
likely that the flysch deposits currently outcropping in the east of the Ioannina basin 
(mostly silts, sands and medium- to fine-grained conglomerates) represent a mid-fan 
environment. 
Inner arc extension trending N-S behind the Hellenic Arc to the south (associated with 
subduction) in the Late Pliocene allowed a series of WNW and ENE non-marine basins 
to open (Clews, 1989). Transfer zones (consisting of NNW and NE tt·ending faults) 
that linked the basins enabled them to elongate. During the Early Pleistocene, 
compressional tectonics became dominant in Epims due to the continued collision of the 
Greek (Anatolian-Aegean) block with the Apulian continental block to the north and 
west (e .g. Clews, 1989; King et al., 1993; Jackson, 1994; King et al., 1994; Le Pichon 
et al., 1997). This regime is still active today, as demonstrated by the rejuvenated 
profiles of rivers such as the Thiamis and Louros (King and Bailey, 1985), as well as 
evidence of uplift from the Ioannina basin itself (discussed below). 
The Pindos experienced localised glaciation during the Quaternary, although clear 
geomorphological evidence only exists for the last glacial period (e.g. Denton and 
Hughes, 1981; Lewin et al., 1991). Even so, little research has been carried out as to 
the extent and number of localised glacial events and more fieldwork and detailed 
mapping are necessary (Tzedakis, 1991). 
The main lithological units seen outcropping in the Ioannina basin are detailed below. 
Their distribution is illustrated in fig. 2.2 (after IGSR-IGP, 1966; Perrier et al., 1967). 
PHYSICAL SETTING • 13 
Pantokrator Limestone (Lower Jurassic). White, maSSIve onchoidal limestone, 
containing algal remains and brachiopods from the Lower - Middle Lias, mostly 
outcropping in the north-western part of the basin. 
Posidonia Schists (Upper Jurassic). Thinly bedded silica-rich argillaceous deposits 
containing abundant radiolaria and bivalves. They outcrop to the north-western part of 
the basin. 
Vigla Limestone (Upper Jurassic - Lower Cretaceous). Well-bedded pinky-cream 
limestone with flint nodules, seen across the basin. 
Upper Senonian Limestone. White, well-bedded limestone containing rudist (reef­
building bivalve) fragments and flint nodules. Dominates the periphery of the basin. 
Palaeogene Limestone. Yellow-white, massive limestones outcropping to the south­
east of the basin, displaying minor karst development. 
Tertiary Flysch (Upper Eocene-Lower Miocene). Marlstones, siltstones, sandstones 
and intercalated fine-grained conglomerates outcropping to the eastern part of the basin. 
Quatemary deposits. Alluvial and lacustrine sediments (mostly clays and silts) present 
across the main floor of the basin. These are discussed in detail below. 
2.3 BASIN EVOLUTION 
Karst solution- of the Ioannina syncline carbonates and continual movement along the 
boundary faults fOlmed the original basin or polje. IGME suggest that in Late Pliocene / 
Early Pleistocene times, the basin floor was sealed by braided river deposits which 
allowed the formation of the lake. Cores show fluvial sediments that give way rapidly 
to lacustrine deposits; this lithological succession has been designated the Anatoli 
FOlmation (Broussoulis et al., in prep.). Consisting of silts, clays, calcareous gyttja 
and two main lignite bands, these lacustrine deposits can be found in cores stretching 
from Kato Lapsista in the north to Ampelia in the south. IGME have attributed a Lower 
Pleistocene age to these deposits by comparing pollen data from core 123 (taken from 
the northern prut of the basin) with similar palynological studies from elsewhere in 
Greece (e.g. Benda et al., 1987). Current lack of any independent chronological 
control for the Ioannina pollen data however, means that any dating must remain 
questionable. 
A depositional hiatus is overlain by beds of the Kastritsa Formation, a conformable 
sequence of sands, silts, clays, calcru'eous gyttja and another two lignite bands 
(Broussoulis et al., in prep.). These deposits are found in cores taken from two 
separate areas of the basin: around Kato Lapsista in the north; and around the present-
PHYSICAL SE1TING • 14 
day lake, stretching from south of Katsikas, Kastritsa and Loghades, to Perama in the 
north. The geometry of the sand deposits has been interpreted as indicative of alluvial 
fans bringing eroded flysch material in from the east. These fans are present south of 
the Anatoli - Loghades line. As mentioned above, apart from some tentative pollen 
comparisons, IGME have only stratigraphic evidence to justify their attribution of the 
Kastritsa Formation to the Middle and Upper Pleistocene. 
By studying the lignite stratigraphy of the whole basin, IGME have suggested that 
several sub-basins developed, each of which was susceptible to variable and 
independent tectonic subsidence. The sub-basin in which core 284 was drilled is 
considered to have been the deepest (Y. Broussoulis, pers comm.). In addition, the 
dominantly silt lithologies of 284 (see section 4.1.1) suggest that it was drilled through 
the Kastritsa Formation, beyond the furthest extent of the alluvial fans. 
It is interesting to note at this point that a scheme to classify Mediterranean valley 
sediment sequences was first proposed by Vita-Finzi (1969). He suggested that phases 
of alluviation throughout the Mediterranean region were climatically controlled and 
resulted in two main sedimentary units: the 'Older Fill', characterised in Epims by the 
red beds of the Kokkinopilos Formation; and the 'Younger Fill', characterised in Epirus 
by the drabber Valley Floor Alluvium. This overly simple classification has since been 
widely questioned, as regional alluvial stratigraphies are often much more complex than 
this two-fold system might suggest (e.g. Butzer, 1980; Wagstaff, 1981; Pope and van 
Andel, 1984; Gomez, 1987; Harvey and Wells, 1987; Lewin et al., 1991; van Andel, in 
press). These doubts are ably supported at Ioannina by the complex alluvial history 
detailed above. Accordingly, no further use of any of these terms will be made during 
this study, however an interesting review of the development of the OlderIYounger Fill 
story is given by Grove (1997). 
The soils developed on the catchment geology will be an imp0l1ant component of any 
alluvial deposit. According to the somewhat confusing study made by MacLeod and 
Vita-Finzi (1982) on the soils and alluvial deposits of Epirus, they broadly fall into two 
main categories: terra rossa red beds and so-called brown Mediterranean soils. 
However, a detailed study on the palaeosols and red beds of Greece (van Andel, in 
press) points out that terra rossa sensu stricto is now a relatively rare in situ deposit in 
the region. True terra rossa is developed on karstic limestone terrain as a direct result of 
dissolution, its strikingly characteristic red colouration being derived chiefly from the 
iron-rich insoluble residue. Most terra rossa seen in the region is therefore a secondary, 
redeposited sediment. It typically has a bimodal composition, with up to 30% 
consisting of silt-sized quartz material and around 70% consisting mainly of clay-sized 
PHYSICAL SETTING. 15 
material. The mineralogy of this fine fraction is dominated by kaolinite, illite and to a 
lesser extent, smectite (van Andel, in press). The clay fraction is derived from 
dissolution of the limestone (the colour resulting from abundant iron oxides in the 
residue, mostly haematite and goethite), whereas the silt fraction is thought to be 
derived from aeolian material, brought by scirocco winds from the deserts of north 
Africa (MacLeod, 1980). 
Brown Mediterranean soils, on the other hand, develop on both flysch areas and the 
chert and shale bands associated with some of the limestones. Both of these lithologies 
are present in the basin, the flysch outcropping extensively to the east and the Posidonia 
beds outcropping in a limited area to the north-west. MacLeod and Vita-Finzi (1982) 
note that these soils typically have a higher fine-sand component than the red beds 
(-50% as opposed to -10%). Their clay-sized component, whilst generally lower than 
the red beds (-25 % as opposed to -70%), is characterised by relatively high 
proportions of the clay minerals illite and montmorillonite. The iron hydroxide mineral 
goethite (often derived originally from haematite) is thought to be primarily responsible 
for imparting the drab brown colour to the soil. For a comprehensive discussion of the 
palaeosols and red beds of the region, see van Andel (in press). 
Fans consisting of a broad range of alluvial material can mainly be seen along the 
northern margin of the present day basin (at the base of the Mitsikeli ridge), as well as 
just north of -Arnpelia (fig. 2.3) . Other geomorphological features of note are the 
terraces that can be seen occasionally around the margins of the basin, for example 
those mapped to the far east, south of Loghades (IGRS-IFP, 1966). Higgs and Vita­
Finzi (1966) also record several terraces that they interpret as raised beaches located to 
the west of the airport (north-west of the lake), the lowest of which is 12m above the 
current lake-level. IGRS-IFP (1966) however, map these deposits as the remains of 
ancient siliceous alluvial fans, rather than raised beaches. Similar fan deposits are also 
recorded in the far south of the basin, east of Ampelia. The possibility of tectonic 
movements within the basin are discussed in section 7.3.2. 
2.4 CLIMATE 
The Mediterranean experiences a complex transitional climate, since it is effectively 
sandwiched between temperate maritime, temperate continental and arid subtropical 
desert regions. The climatic system is largely controlled by the westerlies in winter and 
the subtropical anticyclone in summer (Barry and Chorley, 1992). This results in the 
PHYSICAL SETTING • 16 
characteristic hot, dry summers and the mild, wet and windy winters associated with a 
typical 'Mediten'anean climate'. It should be noted that the actual definition of a 
'Mediterranean climate' is not easy. Different climate classification schemes invariably 
invoke different criteria, reflecting the complexity and limited understanding of the 
current climatic system. For example, much of the Meditenanean falls into the Cs 
category of the classic Koppen scheme, which requires rainfall in the wettest winter 
month to be at least three times that of the driest summer month (Wigley and Farmer, 
1982). Aschmann's (1973) definition is much narrower however, specifying a range 
of annual precipitation (between 275-900mm), a mean winter temperature below 15°C 
and the hours per year at which the temperature is below freezing to be no more than 
3% of the total. A detailed discussion is beyond the scope of this report, but both 
Wigley and Farmer (1982) and Barry and Chorley (1992) provide comprehensive 
reviews. A brief summary is given below. 
Following a long and generally unpredictable spring (March - May), collapse of the 
Eurasian high-pressure cell is accompanied by northern and eastwards expansion of the 
Azores anticyclone from the west. This results in a northwards movement of 
depressions into continental Europe, markedly reducing the chances of precipitation. 
By mid-June, the expanding Azores anticyclone becomes the dominant factor 
controlling the weather (Barry and Chorley, 1992). A low-pressure trough that 
develops across northern Africa ensures that most of the winds are from the north (e.g. 
the etesians of the Aegean), derived from an extension of the north-east trades. 
Regional winds are also important, such as the hot, dry and dusty airstreams of tropical 
continental origin blowing northwards (e.g. the sciroccos). Precipitation is generally 
low and many meteorological stations record very little or no rain for at least one of the 
summer months. 
Towards the end of October, the onset of winter is usually heralded by the swift 
collapse of the Azores anticyclone (Barry and Chorley, 1992). Air masses from the 
north-west also invade frequently, derived from a southerly-displaced branch of the 
Polar Front Jet Stream. These push the Atlantic and Genoa-type depressions (the latter 
forming in the lee of the Pyrenees and Alps) to the east. Incoming winds from north 
Africa and Asia mean that the movement of these depressions is often complex, 
particularly in the eastern Meditenanean (Wigley and Farmer, 1982). Convective 
instability is ensured by a combination of relatively warm sea-surface temperatures and 
cool air temperatures (up to 2°C difference during January), meaning that intense 
precipitation often results (Barry and Chorley, 1992). 
PHYSICAL SETTING. 17 
Orographic effects (the topographical 'blocking' of weather fronts) are extremely 
important local factors throughout the Mediterranean region that are capable of 
producing small-scale but generally sharp contrasts in precipitation (Wigley and 
Farmer, 1982). For example, the Pindos Mountains in Epirus are oriented parallel and 
in close proximity to the Ionian coast and perpendicular to the easterly-tracking winter 
depressions. This has a marked orographic effect, causing high precipitation over the 
region and an effective rain-shadow further to the east. A marked moisture gradient 
therefore exists west - east over northern mainland Greece. 
As a west-facing site at -470m above sea-level in the western reaches of the Pindos, 
Ioannina falls well within the zone of high precipitation referred to above (it is one of 
the wettest mainland meteorological stations in Greece). Annual precipitation is 
1204mm, with December the wettest month (176mm) and August the driest (29mm). 
The high Pindos ensure that Ioannina receives a degree of shelter from cold north­
westerly winds (derived from the Polar Front Jet Stream) during winter months, 
meaning that the average January temperature is around 5°C. Lake Pamvotis therefore 
freezes only a few times per century (Tzedakis, 1991). During the summer, mean 
temperatures reach -25°C, although during the period 1951-1980 the maximum 
recorded temperature was 41°C, in 1973 (all data from Tzedakis [1991], originally 
derived from records of the National Statistical Service of Greece). In experiencing 
warm, dry summers and cold, wet winters therefore, the climate of Ioannina can be 
broadly described as 'Mediterranean' with a degree of continentality, to account for its 
elevation and distance from the coast (Tzedakis, 1991). 
2.5 BASIN HYDROLOGY 
No detailed studies on the hydrology of the lake basin have been published, but a few 
overviews do exist, the main points from which are summarised here. The lake can be 
broadly classified as an effectively closed hydrological system, since it has no 
outflowing streams or major basin drainage. This is a crucial factor, since closed lake 
basins have been shown to display lake-level changes that are largely a response to 
climatic changes in precipitation and evaporation. Anagnostidis and Economou-Arnilli 
(1980) summarise some of the main hydrological features of the modern basin. They 
highlight the fact that the lake forms the base-level of thekarst aquifer underlying the 
Mitsikeli ridge (altitude: 1,81Om), which is drained by the springs that feed the lake on 
the nOlthern and eastern shores. Two of the main ones are located on the shore just 
north of the island of Nissi and near the settlement of Goritsa to the north-west. There 
PHYSICAL SEITING • 18 
is no permanent outflow, although it is thought that blocked sink-holes (or katavothrae) 
present in the floor of the lake occasionally become unblocked for a short time (e.g. 
Fels, 1957; Higgs et al., 1967). It may therefore be difficult to distinguish climatically­
driven lake-level variation from that caused by, for example, earthquake activity 
temporarily unblocking the sink-holes. 
Runoff from the mountains to the east is minimal, since surface waters disappear into 
karstic sink-holes and reappear through the springs (Conispolatis et al., 1986). 
However, run-off from the lowland plain to the north-west is more substantial 
(Anagnostidis and Economou-Amilli, 1980). The modern water chemistry of the lake 
is discussed in detail in section 4.2.2. In summary, it can be considered as a typical 
freshwater temperate hard-water lake, slightly alkaline (pH = -7.6) and dominated by 
Ca2+ and HC03· ions. The thermoc1ine is not very pronounced (Overbeck, 1980). 
Some of the main hydrological parameters of the catchment are summarised in table 
2.1. 
Total mean annual precipitation 
Evapotranspiration 
Infiltration 
Runoff 
Annual yield of springs 
1236mm per yr 
545mm per yr 
495mm per yr 
196mm per yr 
45 x 106m3 per yr 
Table 2.1 Water budget for total sUlface of catchment area of Lake Pamvotis . Data taken from 
Marinos (1974) and Anagnostidis and Economou-Amilli (1980) . 
2.6 THE MODERN LAKE BASIN 
The city of Ioannina, with a population of approximately 100,000 inhabitants, is 
situated on the western shore of Lake Pamvotis. The agricultural area of the basin 
(approximately 133,000 acres) is mostly planted with corn, tobacco and vegetables 
(Albanis et al., 1986). The lake itself is eutrophic, with abundant phytoplankton in the 
surface waters thriving from the pollutants introduced from various sources (human and 
industrial waste, agricultural pesticides). Falling fish stocks and the constant presence 
of unpleasant dark green algal blooms mean that pollution of the lake is becoming a 
serious ecological concern (e.g. Albanis et al., 1986; Kalogeropoulos et al. , 1994). 
PHYSICAL SETTING. 19 
The modern lake cunently has a maximum depth of around lOm and an areal extent of 
22.8km2• It is approximately 8km in length (along its longest axis) and has a maximum 
width of approximately 5km. Throughout historical times, the lake was more extensive 
than it is today. At the end of the nineteenth century, part of the lake and the marsh 
around Lampsista to the north were attificially drained for agricultural purposes. 
Completed in 1944 (Fels, 1957), this was achieved via a 17km long canal linking to a 
4km long tunnel constlUcted through the Tomarokhoria Plateau in the west, emptying 
into the Kalamas River. The 10MB boreholes were therefore all drilled on dry land. 
There is little surrounding natural vegetation remaining in the basin due to cultivation, 
logging and over-grazing of the land by domestic animals. These activities have also 
largely contributed to the destabilisation and erosion of the soils. Around the lake itself, 
the vegetation consists lat°gely of the free-standing reeds Phragmites communis, along 
with Potamogeton peifoliatus, Typha augustifolia, Scirpus holoschoenus, S. lacustris, 
Cyperus longus, and Sparganium erectum amongst others (Higgs, et al., 1967; 
Bottema, 1974). Sparse open-canopy blUshwoods and low thickets (dominated by the 
evergreen oak, Quercus coccifera) at°e the main vegetation types on the lower 
uncultivated slopes and alluvial fans below about 1,OOOm (Higgs et al., 1967). Other 
taxa, such as Juniperus oxycedrus, Pistacia terebinthus, Ostrya carpinifolia and several 
other kinds of thorny shlUb also occur. Above this level and up to the tree-line 
(approximately 1,700m), remnants of a mixed forest can be determined, composed 
chiefly of Fagus sylvatica, Pinus nigra, Abies cephalonica and in the upper reaches, 
Juniperus communis (Higgs et al., 1967). It has been suggested (Higgs et al., 1967; 
Bottema, 1974) that in the absence of human activity, the natural vegetation of the basin 
would be a deciduous forest dominated by Quercus coccifera up to about 1000m. 
Above this level, a mixed beech-pine-fir forest would dominate, with the possibility of 
some sub-alpine taxa (such as Daphne oleoides) at the highest elevations. 
2.7 SUMMARY 
This chapter described the physical setting of the study site, by first considering the 
regional geology of EpilUs and then by outlining the detailed evolution of the lake basin 
itself. Tectonic movements during the mid-Eocene to Oligocene deformed the extensive 
Mesozoic cat°bonates of the region and folded them into a series of anticlines and 
synclines. The newly formed Pindos Mountains provided substantial quantities of 
clastic material to the region, which were deposited as a series of turbidite fans into the 
foreland thrust basin that had developed to the west of the new mountains. Extensional 
PHYSICAL SETTING • 20 
deformation in the Late Pliocene encouraged a series of elongated non-marine basins to 
form, followed by a return to a compressional tectonic regime during the Early 
Pleistocene. The region was probably subject to localised glaciation throughout all 
glacial phases of the Quaternary. 
The Ioannina basin itself developed on a syncline. Solution of the limestone basement 
created a depression known as a polJe which, coupled with tectonic subsidence along 
the boundary faults, led to the development of the lake basin. Extensive coring by 
IGME revealed that the floor of the basin was sealed, probably in Plio-Pleistocene 
times, by braided river deposits that were quickly succeeded by lacustrine sediments. 
Evidence from lignite bands suggests that several sub-basins developed, each of which 
was subject to a variable tectonic history. It seems likely that the basin through which 
core 284 was drilled was the deepest of these sub-basins . The geomorphology of the 
current basin was also discussed, along with evidence for previous lake levels. 
The climate of the region was then considered. Ioannina is subject to a 'Mediterranean' 
type climate, in that it experiences cold, wet and windy winters, along with watm and 
dry summers. The elevation of the site and its distance from the coast are also 
important, as they impatt a degree of continentality to the climatic regime. The 
orographic effect of the Pindos ensures high levels of precipitation from easterly­
tracking depressions during winter months, yet also shelter the region from the coldest 
north-westerly winds derived from the Polar Front Jet Stream. 
The little information that exists on the hydrology of the basin was reviewed. The lake 
forms the base-level of a karstic aquifer, much of which is present under the Mitsikeli 
Ridge. A typical temperate hat'd-water lake, it receives water both from Mitsikeli, via a 
series of springs on the northern and eastern shores, and runoff from the plain to the 
north-west. As there is no permanent outflow, the lake can be considered an effectively 
closed system, although a series of underground sink-holes (known as katavothrae) 
may become intermittently unblocked and allow the lake to drain. This could potentially 
cause problems in trying to distinguish climatically-driven lake-level changes from 
those instigated by temporarily unblocking the sink-holes (e.g. by earthquakes). 
Finally, the modern lake (Pamvotis) was described, along with a brief review of the 
present natural vegetation of the at·ea. 
3. METHODS 
3.1 INTRODUCTION 
The extreme length of borehole sequences such as 284 pose serious logistical 
problems when it comes to designing a logical and manageable programme of study. 
The approach taken with 284 was initially to apply a series of relatively quick and 
simple techniques to the entire core in order to establish an idea of overall structure. 
Loss-on-ignition (to assess organic matter and carbonate content) and magnetic 
susceptibility were principally used for these initial broad-scale analyses. These data 
were then used as a basis for a programme of further, more detailed investigations 
across selected intervals of the core (specifically, the Eemian and the last glacial -
Holocene transition). These included both physical and chemical techniques, such as 
particle-size analysis, isotope geochemistry and faunal analysis. All of these methods 
are described in detail below. 
3.2 PHYSICAL ANALYSES 
3.2.1 Core Recovery and Logging 
Core 284 was recovered by IGME in 1989 using a rotary EDECO truck-mounted 
drill. Retrieved in 3m lengths, it was first logged and then sectioned longitudinally 
into quarters. One of these quarters (average 50mm across) was then split into metre 
lengths and archived in open wooden boxes; the remainder was discarded. The core 
boxes were stored in an unheated shed where the core lengths were able to air-dry. 
Prior to its transfer to Cambridge, the core was re-logged and compared with the 
IGME record. Detailed inspection by eye (using a hand-lens) was complemented by 
the use of dilute HCI and a Munsell soil colour chart (Munsell Color, 1975) in 
appraising moist sediment. 
METHODS.22 
3.2.2 Magnetic Susceptibility 
Mass specific magnetic susceptibility (X) is a measure of the ratio of magnetisation 
induced in a sample to the intensity of the applied magnetising field (Yu and Oldfield, 
1989). Essentially, it is a measure of the magnetisation of a sediment sample from the 
temporary application of a weak (low-amplitude) magnetic field. The strength of the 
signal is dependent upon the concentration, grain-size and grain-shape of the 
magnetic minerals within the sample. In some cases, these minerals can form directly 
within the lake itself. Usually, however, they are derived from weathered-out detrital 
material from the catchment. One of the main applications of magnetic susceptibility 
is therefore as a proxy indicator for inwash (Thompson et al., 1975). During warm 
periods, vegetation flourishes, binding the ground surface, reducing the amount of 
detrital material washed into the lake from natural erosion. During colder periods, 
however, reduced vegetation cover results in higher rates of erosion, consequently 
leading to higher levels of inwash (Thompson et al., 1975). This is discussed in more 
detail in section 7.2.3. 
Another application of magnetic susceptibility is in providing a means of correlation 
between cores taken from the same basin (e.g. Thompson, 1973; Thompson et al., 
1975; Dearing et al., 1981; Flower et al., 1984). Even in catchments where the 
bedrock is lacking in primary magnetic minerals (such as karstic terrain), 
pedogenically enhanced soils can provide a source of secondary magnetic minerals 
(e.g. Longworth et al., 1979; Dearing, 1979; Flower et al., 1984). At Ioannina, core 
249 was originally studied for pollen by Tzedakis (1991), but no susceptibility 
measurements were made at that time. Sampling of both cores (284 and 249) was 
therefore undertaken as part of this study in an attempt to correlate the two records. 
Sampling was carried out on both cores at approximately 1m intervals. In the case of 
core 284 , the samples were intended to double for both palaeomagnetic and magnetic 
susceptibility determinations (a full discussion of the palaeomagnetic work is given in 
section 3.2.3, below). Cubes of sediment were carefully cut from the cores using a 
saw blade and trimmed to size with a stainless steel penknife before packing into 
2.3cm2 clear plastic sample cubes (8cm3 internal volume) and securing with clingfilm. 
The sediment cubes were oriented so that their way-up could be marked on the plastic 
cube faces for the purposes of palaeomagnetic measurement. A total of 318 samples 
was taken. In the case of core 249, only loose sediment sub-samples from the original 
core were available (courtesy of P.C. Tzedakis). Sediment was therefore crushed and 
packed into clear cylindrical plastic pots (IOcm3 internal volume). A total of 153 
METHODS.23 
samples was taken from this core. The mass of the sediment sample in each case was 
determined by accurately measuring the mass of the pot before and after packing. 
Mass-specific susceptibility is calculated using the following expression: 
reading - background average Xlf = ---=----=-----=--
mass (g) 
where Xlf = low frequency mass-specific magnetic susceptibility (m3kg- I); and 
backgroUl'ld average = (meter reading + previous reading) / 2. Susceptibility 
measurements were made on a Bartington Magnetic Susceptibility Meter MS2 
attached to a dual frequency Bartington Sensor MS2B (set to low frequency). 
Background levels were determined using a clean, empty pot (cubic or cylindrical, as 
appropriate) . 
3.2.3 Palaeomagnetic Analysis 
The magnetic field of the Earth has the charateristics of a dipole magnet and is 
thought to be generated from motion within the fluid outer core. Although it has been 
known since the middle of the 19th century that volcanic rocks acquire a 
magnetisation on cooling (e.g. Delesse, 1849; Melloni, 1853), it wasn't until the 
beginning of this century that it was recognised that some older lava flows preserved 
a magnetisation with a reversed polarity (e.g. David, 1904; Brunhes, 1906). Although 
the mechanism is poorly understood, it is now well established that throughout 
geological time the polarity of the Earth's magnetic field has periodically reversed, 
the last major reversal occurring~780ka ago (Shackleton et al., 1990; Baksi et aI., 
1992). Since these reversals occurred almost instantaneously and were global in 
nature, their recognition in both terrestrial and marine sequences has been important 
in establishing worldwide correlations and a lengthy chronological framework. 
It was hoped that by studying the palaeomagnetic signal from the Ioannina sequence, 
a comparison of reversals could be made with a 'master' global magnetostratigraphy, 
thereby deriving an independent chronology. The results from that investigation are 
presented and discussed at length in section 5.4. The methods used are detailed 
below. 
The suitability of the Ioannina material for palaeomagnetic measurement had initially 
to be considered. It is all too easy to analyse a sequence without giving full 
METHODS.24 
consideration to potential problems that might bias results (e.g. Kukla and Zijderveld, 
1977; Verosub, 1982; Lszjvlie, 1989). Ideally, sequences should be continuous and 
undisturbed. If sediments have been subject to subsequent tectonic activity, which 
has folded or tilted initially horizontally-lying deposits, then magnetic measurements 
uncorrected for these effects will be misleading. Bioturbation can affect the 
orientation of magnetic grains within the sediment, with the result that magnetic 
measurements will again be misleading. Sites with high sedimentation rates are 
therefore preferable, since any bioturbation effects will be minimised. Finally, coring 
techniques should minimise disturbance of the sediment and recover vertically­
aligned cores, so allowing accurate orientation of the samples. 
The Ioannina 284 core satisfies most of these criteria. The sequence appears to be 
continuous since there are no obvious lithological breaks or hiatuses (section 4.1.1). 
Whilst . the site is located within a tectonically-controlled basin, it is thought that the 
sediments have not suffered much post-depositional disturbance. There is no direct 
evidence to suggest that the sub-basin through which 284 was cored was subject to 
any tilting or folding as a result of basin subsidence. Discrete lithological bedding, 
which has a consistently horizontal orientation, occurs in places throughout the length 
of the core. The proposed pollen-derived age model (discussed in detail in section 
5.2) allows an estimation of accumulation rates. It appears that sedimentation was 
rapid, an average of O.5rnka- 1 (in some cases more), far higher than in most marine 
environments... It is likely, therefore, that bioturbation effects were minimal, 
especially since 284 is thought to have originated from the deepest sub-basin. This 
was a location that would have always experienced relatively little benthic activity in 
comparison to shallower littoral areas of the lake. Finally, given the nature of the 
coring equipment, it is a reasonable assumption that the cores were drilled vertically 
and were subject to minimal disturbance. 
A total of 318 samples were collected for analysis from the same horizons already 
analysed for loss-on-ignition and carbonate content were taken (i.e. -lm intervals). 
Samples were cut from the core using a steel saw blade and were 'way-up' oriented 
within plastic sample cubes (2.3cm3). Palaeomagnetic measurements were made at 
the University of East Anglia (under the supervision of Dr B.A. Maher) using a 
Cryogenics Consultants Ltd. cryogenic magnetometer, utilising GM400 software 
(version 5.4) and fitted with three SQUID devices. Samples were measured in two­
position mode throughout. A comprehensive summary of this technique is given by 
Collinson (1983). The drilling method originally employed meant that it was 
impossible to determine the azimuthal orientation of the core, automatically 
precluding any declination measurements. Consequently, the two most important 
METHODS.25 
parameters measured by the cryogemc magnetometer were the inclination and 
intensity of the magnetic field. 
Initially, the natural remanent magnetisation (NRM) of each sample was measured. 
This is a vector quantity, described by the inclination, declination and intensity of a 
magnetic field. The NRM of each sample is a vector sum of the primary remanence 
(acquired when the sediment was originally formed) and any secondary remanence 
(that may have been acquired at a later stage) . The primary remanence is the most 
important component of the NRM, since it is a record of the direction and intensity of 
the ambient magnetic field of the Earth at the time of sediment formation. 
Identification and removal of the overprinted secondary remanences is a major part of 
any palaeomagnetic analysis. 
There are several ways in which a sediment can acquire a secondary remanent 
magnetisation, the main ones being depositional (DRM), post-depositional (PDRM) , 
chemical (CRM) and viscous (VRM). DRM is acquired by magnetic grains as they 
settle in the water column, which become oriented in the direction of the ambient 
magnetic field. PDRM is acquired by magnetic grains after burial if they are free to 
rotate within the interstitial voids of the sediment, aligning themselves with the 
ambient magnetic field. CRM is again a post-depositional process and relates to those 
magnetic minerals that undergo chemical change during diagenesis, thereby acquiring 
a magnetisation. VRMs are generally thought of as time-dependent isothermal 
remanences (Jacobs, 1994). In other words , magnetic grains within a sediment can 
gradually acquire a (generally weak) remanent magnetisation in the direction of the 
ambient magnetic field without being heated. The rate and degree of magnetisation is 
dependent upon the temperature and grain size (Jacobs, 1994). Sediments can easily 
acquire VRMs in a variety of ways, such as during coring, sampling or long-term 
storage. 
A series of pilots from the entire core (15 samples) were selected and subjected to 
both alternating field (AF) and thermal stepwise demagnetisation procedures. A brief 
description of each of these techniques now follows; a more detailed account can be 
found in Tarling (1983). AF demagnetisation tries to isolate the primary 
magnetisation from the secondary by randomising the magnetisation directions of 
those magnetic particles within the sample that have low coercivities. Coercivity is a 
measurement of the ease with which the magnetisation direction of a particle can be 
induced to follow that of an applied magnetic field. It is often the case that secondary 
magnetisations of a specimen are carried by particles with low coercivities. If the 
magnetisation of these particles can be randomised so that they do not contribute to 
METHODS. 26 
the external magnetic field of the sample, then the remaining remanence is likely to 
represent the primary magnetisation. The process involves tumbling each sample 
along the axis of a coil through which an alternating electric current is passing (the 
tumbling ensures that all alignments of the sample are rotated into the peak magnetic 
field induced along the coil axis) . The coil and sample tumbler are themselves 
located within an area in which the ambient geomagnetic field has been zeroed. The 
field is steadily reduced from an initial strength (measured in milliteslas, mT). The 
magnetisation of those magnetic particles within the sample having a lower coercivity 
than the applied field will follow the direction of the field. Steady reduction of the 
field strength means that the magnetisations of these particles will become effectively 
'frozen' in the specimen in random orientations (once the field strength drops below 
the coercivity values of individual particles). Those magnetic particles with 
coercivities higher than the initial field strength will remain unaffected. The 
remanence of the sample is measured after each sample, after which the initial field 
strength is increased and the procedure repeated. Careful analysis of sample 
behaviour throughout this stepwise demagnetisation can often reveal the point at 
which only the primary magnetisation remains. 
Thermal demagnetisation operates in a similar way to AF demagnetisation, except 
that it relies on the relaxation times of the magnetic particles rather than the 
coercivity. Relaxation time, a temperature and grain-size related property, is a 
measurement of the time required for a magnetic particle to aquire a magnetisation in 
the direction of an applied field. Samples are heated to a predetermined temperature 
and maintained at that temperature for a certain period of time. On cooling (in a 
zeroed external magnetic field) the magnetisation of those magnetic particles with low 
relaxation times will effectively lie in random directions and will therefore contribute 
nothing to the external magnetic field of the sample. Those particles with higher 
relaxation times will be unaffected and so retain their original magnetisation 
directions. Repeated cycles of heating, cooling and measurement of remanence, with 
the target temperature being increased stepwise for each cycle, allows assessment of 
the point at which only the primary magnetisation remains. Different minerals 
(including magnetic ones) become unstable at specific temperatures and form new 
minerals that are stable under the altered conditions. For example, goethite alters to 
haematite at temperatures in excess of -200"C and haematite alters to magnetite at 
temperatures in excess of -550°C (Tarling, 1983). Chemical stability of the sample is 
therefore monitored throughout the demagnetisation procedure by measuring its low­
field magnetic susceptibility prior to beginning each new cycle. Once the 
susceptibility of a sample changes dramatically there is little point in continuing with 
METHODS.27 
the demagnetisation procedure, since its magnetic characteristics will now be 
determined by a new mineral assemblage. 
The pilot samples were all subjected to AF and thermal demagnetisation procedures. 
Stepwise AF demagentisation was carried out at 5mT, lOmT, 20mT and then at 20mT 
increments up to 100mT. Stepwise thermal demagnetisation was carried out at 50°C 
increments between lOO-400°C. Results indicated that the thermal method was more 
successful in removing any overprinted secondary remanences. A blanket 
demagnetisation was therefore applied to all of the core samples (300°C maintained 
for 15 mins) and their remanences remeasured. The results are presented and 
discussed at length in section 5.4. 
3.2.4 Particle-Size Analysis 
The particle-size characteristics of sediment sampled from selected parts of the core 
were examined by means of a Malvern Instruments Mastersizer-X laser analysis 
machine. By selectively removing the carbonate and organic fractions of the 
sediment by chemical means and re-running the particle-size analysis each time, the 
individual grain-size characteristics of each fraction could be established. 
Approximately 5g of sediment was sampled from the core at 1m intervals. This 
material was placed in a polythene test-tube with deionised water and left overnight in 
a warm (90°C) water bath to completely disaggregate. The resulting solution was 
then whirlimixed and a small sample immediately placed in the analyser and 
measured. The residue was centrifuged and excess water decanted. Hydrochloric 
acid (7%) was then added (to remove the carbonate fraction) and the tube returned to 
the water bath. Once all reaction had ceased (usually after about 30 minutes), the tube 
was again centrifuged, excess liquid decanted and deionised water added. Following 
whirlimixing, a small sample of the solution was placed in the analyser and measured. 
The residue was again centrifuged and excess water decanted. Hydrogen peroxide 
(lOO volume) was then added (to remove the organic component) and the tube 
returned to the water bath. Once all violent reaction had ceased, the tube was left for 
several hours in the water bath until all further reaction had stopped. After 
centrifuging and decanting once more, deionised water was added, the solution 
whirlimixed and a small sample immediately placed in the analyser and measured. 
METHODS.28 
3.2.5 SEM Analysis 
Interpretation of particle-size data can be greatly enhanced by the use of a scanning 
electron microscope (SEM), enabling the actual viewing of the individual constituent 
grains. In addition, the shape and surface textures of the grains can be used to assess 
the composition and proportion of sediment that is either allogenic or authigenic in 
origin. 
Selected sediment samples were disaggregated by adding deionised water and leaving 
them in a warm (90°C) water bath overnight. On the basis of the particle-size data, 
the resulting solution was wet sieved into two fractions (>125).lm and >57-125).lm) 
and dried. The fraction smaller than 57).lm was discarded. The dried material was 
placed onto an adhesive SEM stub and coated with carbon prior to analysis. Samples 
were viewed at the Multi-Imaging Centre, University of Cambridge, using a Philips 
XL30 FEG SEM fitted with an ISIS energy dispersive X-ray microanalyser. 
3.2.6 Modern Faunal Analysis 
During September 1994, eleven sites around the shore of the lake were selected on the 
basis of accessibility and variation in habitat (fig. 3.1) . Qualitative samples of 
molluscs were taken from all micro-habitats at the lake edge (e.g. water, shore, on 
aquatic and terrestrial vegetation) and preserved in plastic tubes and bags for later 
identification. The sites are described in appendix A. 
Qualitative ostracod samples were acquired by taking mud samples from the shallows 
«75mm water depth) and storing in ethanol in sealed plastic containers for later 
analysis. On return to the laboratory, the samples were dried, sieved and picked using 
the method outlined below for the fossil fauna. 
3.2.7 Fossil Faunal Analysis 
Samples for faunal analysis were taken at approximately 20cm intervals through those 
parts of the sequence singled out for detailed study. Slices of sediment 4cm thick 
were taken from the core, the full width being needed because of the minimal amount 
of material available (section 3.2.1). A few grams of the sample were retained for 
possible future analyses (e.g. pollen, diatoms, further magnetic susceptibility). As the 
fossil fauna was to undergo further chemical tests (stable isotope determinations and 
METHODS. 29 
N 
i 
Lake Pamvotis 
o 2km 
Fig 3.1 Modern sample sites (described in appendix A). Stippling denotes areas of dense Phragmites 
reeds. The locations of two springs are also shown. 
METHODS. 30 
so forth), the sediment was disaggregated using purely mechanical means. The 
largest part of the sample was immersed in deionised water for 24hrs before 
undergoing repeated gentle freeze-thaw cycles. Commonly, samples required two 
full freeze-thaw cycles to effect adequate disaggregation. Once broken down, 
samples were dried and sieved into four size fractions (>lmm; 0.5-1mm; 250J.lm-
0.5mm; and <250J.lm) for faunal analysis. Each size fraction was weighed prior to 
picking. 
Samples were picked under a Wild binocular microscope (up to lOOx magnification) 
using a triple-O sable-hair paint brush, moistened with deionised water when 
necessary. Specimens (molluscs, ostracods and so forth) were systematically 
separated from the sediment fractions, identified to species level where possible, 
counted and stored either in glass tubes (molluscs, macrofossils and so on), or on 
cavity slides (ostracods). In accordance with conventional practice, only those 
gastropod molluscs with intact apices were counted (non-apical fragments were 
counted only where a species was otherwise unrepresented). In the case of bivalve 
molluscs, complete valves or hinge fragments were counted. For ostracods, the first 
350 valves were counted (and the picked fraction weighed if this did not constitute the 
entire sample). 
3.3 CHEMICAL ANALYSES 
3.3.1 Sediment Geochemistry 
According to Dean (1981), the sediments of temperate hard-water lakes (such as 
Ioannina) can be broadly considered as consisting of four main components : 
carbonate minerals (usually low-Mg calcite); detrital clastic material (derived from 
the catchment); biogenic silica (mostly diatomaceous material); and organic matter. 
lones and Bowser (1978) recognise three distinct mineral sources from which these 
components are derived: allogenic material (minerals derived exclusively from 
outside the lake); endogenic material (minerals derived from processes occurring in 
the water column); and authigenic material (allogenic and / or endogenic material that 
has undergone a degree of alteration due to post-depositional diagenetic processes. 
Distinguishing between these three mineral sources is often extremely difficult. In 
particular, distinguishing between endogenic and authigenic fractions (even using 
advanced chemical techniques) can be almost impossible. For the purposes of this 
study therefore (and in the interests of clarity), the definition of Engstrom and Wright 
METHODS. 31 
(1984) is adopted throughout, in which the term authigenic is used to refer to all 
material formed within the lake, whether by strictly authigenic or endogenic means. 
Many techniques exist for analysing the chemistry of sediments. Most are bulk 
methods that do not distinguish between allogenic and authigenic components 
(Mackereth, 1966; Pennington, et al., 1972; AlIen, 1974; Engstrom and Wright, 1984; 
Robinson, 1994). With the development of techniques such as inductively-coupled 
plasma spectrometry, fractionation procedures have been increasingly employed that 
selectively digest the sediment in a stepwise fashion, permitting allogenic and 
authigenic components to be measured (e.g. Engstrom and Wright, 1984). However, 
these methods are time-consuming and were thought inappropriate for the length of 
core considered here. Since a combination of particle-size and SEM techniques can 
often be used to assess qualitatively the relative proportions of these two components 
within a sediment, two geochemical methods were selected that offered a realistic and 
efficient use of time: loss-on-ignition (to determine broad-scale variability) and bulk 
sediment isotope analysis (for detailed examination of selected parts of the record) . 
(a) Loss-an-Ignition 
Variations in the organic matter and carbonate content of lake sediments are often 
correlated with changes in organic productivity of the lake system over time. The 
relative validity of this assumption is discussed at length in section 7.2. Measurement 
of these components by loss-on-ignition is a relatively simple and standardised 
laboratory procedure. 
Sediment samples of approximately 19 crushed weight were initially taken every 1m 
down the core. Finer resolution was later achieved by sampling the top lOOm of the 
core at 10cm intervals and the remaining 219m at 20cm intervals. Determinations 
were carried out using a method slightly modified from Dean (1974) and Bengtsson 
and Enell (1986), as outlined below. 
After drying to constant mass in an air-circulation oven (lOYC for 12hrs), samples 
were burned in porcelain crucibles in a muffle furnace (Gallenkamp FSE-520) for 
7hrs at 550°C to determine the loss-on-ignition. All masses were measured to an 
accuracy of ±O.OOlg. The change in mass provides an approximation of the organic 
matter content of that sample (Dean, 1974). In calcareous sequences, various 
conversion factors from loss-on-ignition to organic carbon have been suggested. 
Generally organic carbon is taken as 40-60% of the loss-on-ignition (Bengtsson and 
METHODS.32 
Enell, 1986). In this study, no conversion factor has been applied to the data since an 
accurate value has yet to be determined. 
The ash residues were then burned for a further 7hrs at 950°C to determine the 
carbonate content. The change in mass (caused by evolution of CO2), multiplied by a 
conversion factor, reflects the carbonate content of that sample (Dean, 1974). 
Because of the variable volatility of carbonate compounds, the value obtained is only 
an approximation of the total carbonate content of the sample and does not allow the 
exact proportion of different carbonate compounds in the sample to be determined. 
Consequently, the conversion factor is generally taken to be 1.36 (Bengtsson and 
Enell, 1986), corresponding to the molecular weight of C03 (60) divided by the 
molecular weight of CO2 (44). In areas where the carbonate mineralogy of the 
sediment can be shown to be exclusively low-Mg calcite, a higher conversion factor 
(2.273) is sometimes advocated, corresponding to the molecular weight of CaC03 
(lOO) divided by the molecular weight of CO 2 (44). X-ray diffraction studies of the 
modern lake sediments at Ioannina (Conispolatis et al., 1986) have demonstrated that 
low-Mg calcite is the dominant carbonate mineral, whereas high-Mg calcite and 
dolomite (a magnesium carbonate) are present in only tiny amounts (average 3.1 % 
and 0.4% respectively). This is not surprising, given the lack of dolomites in the 
catchment and the overall chemistry of the lake waters (section 4 .2.2) which does not 
favour magnesium carbonate precipitation. Although a case could have been made 
for using the higher conversion factor in this study, the lower value has been used 
throughout, in order to make allowance for the possibility of minor amounts of 
magnesium carbonate being present. Values quoted for carbonate content are 
therefore likely to reflect minimum levels. 
Another potential drawback of this method is that clays within the sediment contain 
lattice-bound water (in the form of OH- ions). This is generally evolved at 
temperatures exceeding 550°C (Dean, 1974). In substantially clay-rich sediments, 
therefore, this loss of water could contribute substantially to the overall loss in mass 
of the sample. At Ioannina, however, the clay fraction of the sediment is not a major 
constituent (section 4.1.3). Evolution of lattice water is therefore assumed to have 
had a negligible effect on the overall loss in mass of the sample. 
The efficacy of the loss-on-ignition procedure was tested periodically throughout the 
study. A replicate set of samples from successive levels were taken after every 400 or 
so determinations to be measured at a later date. Results were then compared. All 
tests (for both organic matter and carbonate content) returned covariance r-values 
greater than 0.9. A typical example is shown in fig. 3.2. 
178.0 
178.5 
:g 
.<:: 179.0 
i5.. 
<1.l 
Cl 
179.5 
180.0 
o 
METHODS.33 
• Original 
................... Replicate 
Covariance: ,. = 0.93 
5 10 15 20 
Carbonate Content (%) 
Fig. 3.2 Replicability of loss-on-ignition technique to calculate carbonate content 
178.0 
Covariance: ,. = 0.99 
178.5 
LOI 
Calcimeter 
179.0 
179 .5 
180.0 
5 10 15 20 25 30 35 
Carbonate Content (%) 
Fig. 3.3 Comparison of loss-on-ignition and calcimeter techniques to calculate carbonate content. 
METHODS. 34 
To test the efficacy of the loss-on-ignition procedure as a robust indicator of relative 
change in carbonate content (as opposed to variation in absolute values), a replicate 
set of samples was measured using a calcimeter (at the Department of Geography, 
University of Cambridge). This relied on the evolution of CO2 from a known mass of 
sample at a known temperature and pressure after the addition of hydrochloric acid. 
The calcium carbonate values were calculated and compared with the values 
measured by the loss-on-ignition method. A remarkable correspondence between the 
two procedures was recorded (shown in fig. 3.3), with a covariance between the two 
curves of 0.99. 
(b) Sediment Isotope Geochemistry 
The study of stable oxygen and carbon isotopes from lakes where primary (i.e. 
authigenic) carbonates form has become a standard procedure in many 
palaeolimnological studies. Principles that are routinely applied in marine 
palaeoenvironmental research have increasingly been applied to lacustrine settings, 
with varying degrees of success (see review by Kelts and Talbot, 1989). Although 
some studies have used bioclastic material (such as mollusc and ostracod shells) as a 
primary carbonate source (see section 3.3.3 below), the technique used here 
concentrates on the authigenic carbonate fraction that has been produced either by 
direct precipitation from the water column, or as a result of inorganic biogenic 
precipitation. Coupled with a detailed knowledge of the regional and local factors 
affecting a specific lake system, stable oxygen and carbon isotope curves have been 
used to infer such criteria as variations in air and water temperatures (e.g. Eicher and 
Siegenthaler, 1984; Siegenthaler and Eicher, 1986), lake level history (e.g. lohnson et 
al., 1991), residence time of lake water (e.g. Stuiver, 1970), variations in evaporation, 
precipitation and moisture sources (e.g. Stuiver, 1970; Siegenthaler and Eicher, 1986) 
and changes in palaeoproductivity (e.g. McKenzie, 1985; Schelske and Hodell, 1991). 
An excellent review by Talbot (1990) discusses the relative merits of the technique 
and places particular emphasis on using the degree of covariance between 81S0 and 
813C to characterise different hydrological systems. Details of the methodology used 
in this study are given below. 
A minimum 20mg of calcite was required for analysis. For each level, approximately 
3g of sediment was sampled and added to a solution consisting of 40ml deionised 
water, plus 40ml of 40 volume hydrogen peroxide (to assist in sediment 
disaggregation) and stirred occasionally. After 24hrs, the resulting solution was again 
stirred, passed through a 150/lffi sieve (to remove any detrital mineral grains and / or 
METHODS.35 
fossil fragments), filtered via a Buchner funnel and rinsed with deionised water. The 
filter paper was then dried and the remaining solids transferred to a plastic tube. The 
stable isotopes were measured at the NERC Isotope Geosciences Laboratory, 
Keyworth. Carbonates were reacted off-line with phosphoric acid at 25°C, with the 
resulting CO2 samples measured from the manifold of a VG SIRA or VG OPTIMA 
mass spectrometer. Correction of results to the V-PDB scale was by comparison with 
laboratory marble standards calibrated against NBS-19. Analytical precision for both 
8180 and 813C was better than 0.1 %0. 
By convention, isotope ratios are not given as absolute values, since they would be 
very small and vary only slightly. Instead, they are given in delta notation as relative 
deviation (per mil, %0) from the isotopic ratio of a standard (Craig, 1957). Thus, the 
oxygen isotope ratio 180/160 is given by the expression: 
{PB 0 = R Slllllple - RSf(/lIdll rd xl 00 
R stalldard 
where R is the absolute ratio of 180 ;'60 . The standard used for both 8180 and 813C in 
the case of the Ioannina samples was V -PDB, Vienna - Peedee Belemnite (based on 
carbon dioxide collected from belemnites [Belemnitella americana] of the Cretaceous 
Peedee Formation of South Carolina). 
3.3.2 Modern Water Chemistry 
An appreciation of the modern water chemistry of the lake is essential if meaningful 
palaeoenvironmental reconstructions are to be attempted. Only by understanding how 
the current limnic system operates can trace-element and stable isotope data derived 
from fossil material be properly interpreted. To this end, a sampling programme was 
initiated that enabled several parameters of the modern water chemistry to be 
determined either directly in the field or later in the laboratory. 
Water samples were analysed from each of the 11 faunal sampling sites . 
Temperature, pH, conductivity, dissolved oxygen content and alkalinity were 
determined immediately . A 60ml unfiltered sample was taken for the purposes of 
later oxygen isotope analysis. A 130ml unfiltered sample was taken and 20ml BaCh­
NaOH solution added (to precipitate the total dissolved inorganic carbonate, TDIC) 
for the purposes of later carbon isotope analysis. Two 60ml filtered samples (0.45flm 
micro-pore filter) were taken. One was kept chilled for later use in the field 
METHODS. 36 
laboratory; the other was acidified using O.lml concentrated nitric acid and kept for 
later trace element analysis. At the end of each day, a Hach portable DRJ700 analyser 
was used to determine values of nitrate, sulphate, silica and phosphate from one the 
60ml samples from each site. Other variables noted at each site included the depth at 
which the water samples were taken, the extent and nature of terrestrial and aquatic 
vegetation, turbidity of the water, nature of the substrate, evidence for human 
disturbance, the general topography of the sample site, the weather and time of 
sampling. 
Water samples retained for stable isotope analysis were sent to the NERC Isotope 
Geosciences Laboratory in Keyworth. The 0180 and oD values were measured 
relative to V-SMOW (Vienna - Mean Standard Ocean Water); oI3C values of TDIC 
were measured relative to V-PDB (as above) . 
Water samples retained for trace-element determination were measured by means of a 
Jobin Yvon JY-70 Plus inductively-coupled plasma - atomic emission spectrometer 
(ICP-AES) at Kingston University. A full review of this technique can be found in 
Jarvis and Jarvis (1992) . A proprietary computer program (PCWATEQ, updated 
from Truesdell and Jones, 1974) was used to analyse the results. By using other 
sampling data (temperature, pH and so on), the accuracy of the field and laboratory 
measurements could be assessed by the calculation of an ionic charge balance error. 
The program also calculated the salinity (expressed as TDS, total dissolved solids), 
mineral speciation in the water and the corresponding saturation index. 
3.3.3 Ostracod Shell Chemistry 
Fully calcified ostracod valves are well established as an important source of 
authigenic carbonate and it has long been known that their chemical composition 
accurately reflects that of the water at the time of shell formation (see comprehensive 
review in Holmes, 1996). The trace-elements magnesium and strontium can be 
linked, with care, to changes in salinity and temperature of host waters (e.g. Chivas et 
al ., 1983; De Deckker et al., 1988; De Deckker & Forester, 1988; Holmes et al., 
1992; Palacios-Fest et al., 1994). Stable isotopes of oxygen have been shown to 
covary with salinity and temperature in hydrologic ally closed lake systems (e.g. Gasse 
et al., 1987; Lister, 1988; Schwarcz & Eyles, 1991), whereas stable carbon isotopes 
can be related to palaeoproductivity (e.g. McKenzie, 1985). Chivas et al. (1993) 
demonstrated that by combining trace-element and stable isotope data the salinity, 
METHODS. 37 
temperature and chemical history of such lake systems can be determined more 
successfully than by using anyone technique in isolation. 
Problems have been encountered with the trace-element analyses on the Ioannina 
fauna. One of the main criteria for selecting suitable ostracod valves for trace­
element measurement, is that they should be fully adult. Ostracods are tiny 
(commonly sub-millimetre) bivalved crustaceans. Like all Crustacea, they grow 
episodically by shedding their carapaces at intervals during their lifetime. There are 
usually eight moult stages (known as instars) between egg and adult, denoted A-I 
(largest) to A-8 (smallest). It has been shown that the uptake and concentration of 
trace-elements (particularly magnesium) in the shell varies during the lifetime of the 
animal (Chivas et al., 1983, 1986; Palacios-Fest et al., 1994), due in part to its 
variable growth rate. As a result, most workers advocate the use of fully adult valves 
in order to obtain comparable results. The ostracod fauna in core 284 contains a 
relatively low proportion of adults (the significance of which is discussed in section 
6.3.4), effectively precluding the use of trace-element geochemistry as a valid 
technique in this instance. 
No such restrictions apply when considering stable isotopes. Although relatively little 
work has been done on the isotopic variability between different moult stages, 
preliminary results (e.g. Heaton et al., 1995) have shown no significant differences. 
Nevertheless, for the purposes of this study, it was decided to sample the same moult 
stages (A-2 / A-3) throughout wherever possible, in order to minimise any 
complications from this source. Another important criterion for selection is that the 
same species should be measured throughout. Work has shown clearly that different 
species often either have different modes of life (e.g. Eyles and Schwarcz, 1991), 
occupy different microhabitats (e.g. Heaton et al., 1995), or display seasonal 
preferences when moulting (e.g. Xia et al., 1993; von Grafenstein et al., 1994), 
resulting in variable isotopic signals. 
The issue of whether to analyse a single ostracod valve at each level, or to analyse 
bulked samples (consisting of several valves per level) is another important 
consideration. Some authors (e.g. Eyles and Schwarcz, 1991; Dettman et al., 1995; 
Heaton et al., 1995) have analysed single valves. Since ostracods form their shells 
over a period of time that can often be measured in hours (Turpen and Angell, 1971), 
they capture a record of environmental conditions that is both highly temporally and 
spatially specific. Single valve analysis can therefore potentially reveal detail that is 
lost when valves are bulked (providing only an 'average' signal). However, Xia et al. 
(1993) and Heaton et al. (1995) warn that since isotopic variation invariably exists 
METHODS. 38 
between individual ostracods, caution must be exercised when using this method. 
They advocate a sampling strategy similar to that employed in the analysis of oceanic 
for a minifera, namely replicating measurements at each level (usually three 
individual valves are used). 
Where the selected ostracod species is too small to be measured using single valve 
analysis (or it is not desirable to do so), bulking of valves from each level is 
commonly undertaken. The 'average' isotopic signal derived in this manner must be 
interpreted in terms of the period of time that the thickness of each sediment slice 
represents. For example, Schwalb et al. (1994) examined 2cm slices of sediment 
from a core drilled in Lake Neuchatel, Switzerland, that represented between 25-30 
years of deposition. Dettman et al. (1995), on the other hand, analysed samples at 
20cm intervals from a series of cores from Lake Huron, that corresponded to 40-100 
years of deposition. 
Another criteria that needs to be considered is the preservation of the valves. Material 
that appears to have undergone any post-depositional dissolution or carries evidence 
of calcite overgrowth should be avoided. The weight of valves is also important. 
Modern mass spectrometers can now measure samples that are only a few tens of 
micro grams in weight. Nevertheless, even when single-valved analyses are being 
carried out, the process is made easier if the species being used is relatively large or 
has a thickly calcified shell. 
The preparation of ostracods for geochemical analysis is an intensely laborious 
process. It is essential that only authigenic carbonate is examined (i.e. the ostracod 
valve), so any adherent carbonate material must be removed, as must any residual 
organic matter. Most workers clean the valves manually under a binocular 
microscope, using high purity deionised water and a fine paint brush (after Holmes, 
1992) . This is often accompanied by additional procedures, either to help 
disaggregate the sediment initially, or to help remove organic matter. These include 
roasting in a vacuum (e.g. Lister et al., 1991), roasting in an oxygen plasma (e.g. 
Boomer, 1993), or applying various strengths of dilute hydrogen peroxide (e.g. Curtis 
and Hodell, 1993; Schwalb et al., 1994; von Grafenstein et al., 1994, 1996). 
In the absence of a detailed investigation into the relative effects of different cleaning 
methods on isotopic variability, purely physical methods were used throughout this 
study. Sediment slices (4cm thick) were taken from the core at approximately 20cm 
intervals. The sediment was disaggregated, sieved and picked according to the 
method set out in section 3.2.6. Between 8-10 valves were selected per level for 
METHODS. 39 
isotopic analysis of Candona permanenta (all either A-2 or A-3 instal's). All surface 
contamination was removed by means of a fine paint brush and high purity deionised 
water. The valves for each level were then placed into a small glass bucket, 
moistened with a small quantity of deionised water and crushed into tiny fragments. 
Stable isotopes were measured at the NE RC Isotope Geosciences Laboratory at 
Keyworth. Samples were reacted on-line with phosphoric acid at 90°C in a VG 
Isocarb coupled directly to a VG OPTIMA mass spectrometer. Correction of results 
to the V -PDB scale was by comparison with laboratory marble standards calibrated 
against NBS-19. Analytical precision for both 8180 and 8 13C was better than 0.1 %0. 
4. RESULTS 
This chapter presents the results of the physical and chemical analyses of the core 
material. There are two exceptions: results of the palaeomagnetic analyses are 
presented in chapter 5, since they relate directly to the chronology of the corej and the 
results of the faunal analyses (modern and fossil) are presented in chapter 6. 
Interpretations are discussed at length in chapter 7. 
4.1 PHYSICAL ANALYSES 
4.1.1 Lithological Description 
The lithologies of core 284 are detailed in table 4.1. It will be noted that from 8.44m 
to the bottom of the core, the lithology is a relatively homogeneous olive-grey silt. 
The subtle changes that occur within this main unit are detailed in table 4.2. 
Depth (m) 
0-1.36 
1.36-5.85 
5.85-7.50 
7.50-8.44 
8.44-319.0 
Description 
Brown calcareous silt with limestone clasts «7cm) . Very 
stony in places. 
Olive-grey clayey silt. Below -2.53m shell fragments 
seen (including Dreissena, Bithynia and Valvata). 
Dark greyish brown calcareous silt, very shelly in parts. 
Olive-grey calcareous clay, containing abundant tiny shell 
fragments. 
Olive-grey silt, calcareous in parts . 
Munsell Colour 
10YR 4/3 
5Y 412 
2.5Y 312 
5Y 412 
5Y 412 
Table 4.1 Description of the lithologies of core 284. Munsell colours determined on moist sediment. 
No distinct lithological breaks were detected in the core, despite careful examination. 
All changes in lithology (limited to the top -8.5m) were across apparently 
conformable and gradual boundaries. Apart from a missing section of core between 
164-164.50m (due to mechanical problems experienced whilst coring), the sequence 
appears to be continuous. 
Depth (m) 
8.50-14.50 
-18.50 
26.10-26.40 
29.0-29.50 
31.75-34.80 
42.0-45 .50 
66.0-66.35 
-72.0 
-90.0 
95.42-95.48 
125.20-144.0 
-160.0-178.0 
164.0-164.50 
172.6-172.9 
-192.0 
264.10 
-270.0-305.0 
308.0-309.50 
311.75-313 .30 
Description 
Slight change in colour from maIn sediment (no 
lithological change). Very shelly. 
Becomes slightly less shelly below this point. 
Darker, fossiliferous bed present with abundant Dreissena, 
along with lower numbers of Bithynia, Viviparus and 
Valvata. Bithynia opercula also common. Returns to 
occasionally shelly below this bed. 
Red-brown mottling within sediment. 
Very shelly, with beds of Dreissena. 
Very shelly, with beds of Dreiss ena. 
Viviparus also present. 
Sediment non-cohesive and soft whilst coring. 
Valvata and 
Only occasional shell fragments seen below this point. 
Abundant ostracods, but no molluscs. 
Organic material present, including wood fragments. 
Occasional molluscan shell fragments . Between 140.2-
140.5, fairly shelly bed, with Dreissena common. 
Subtle reduction in clay content , becoming slightly 
sandier. Occasional fine organic bands noted over this 
interval, along with occasional yellow-brown mottiing, 
especially between 165-169m. 
Missing section of core (poor recovery for this drive) . 
Slightly shelly bed. 
Clay component Increases slightly . Ostracods still 
abundant. 
RESULTS. 41 
Munsell Colour 
2.5Y 3/2 
5Y 212 
5YR 4/4 
2.5Y 6/4 
Thin (-3cm) band of olive-brown clay. 2.5Y 4/4 
Occasional plant debris noted. 
Colour change to very dark grey. No lithological change. 5Y 3/1 
Light olive-brown mottling. 2.5Y 5/4 
Table 4.2 Detailed litho logical description of core 284, from 8.50-319m (see text for further details) . 
Munsell colours determined on moist sediment. 
RESULTS. 42 
According to the limited data available from IGME regarding the degree of recovery 
of the original cores, the lengths within each metre compartment of storage box 
represented successive metre drives, even if the overall length of material did not 
exactly measure 1m. As described in section 3.2.1, the entire core had been air-dried, 
which had often resulted in fragmentation. The accuracy of any depth measurement 
made as part of this study is therefore subject to some degree of error. Since the 
preservation and quality of the core varies throughout, the degree of error also varies. 
However, it is estimated that at worst, depth values are accurate to ± lOcm, although 
for the majority of the sequence, a value of ±5cm is considered more realistic. 
4.1.2 Magnetic Susceptibility 
The mass specific magnetic susceptibility results for the whole core are presented in 
fig.4.1. Generally speaking, values average around 1.5x10-7m3kg- 1, which can be 
considered low. In comparison, Longworth et al. (1979) measured values ranging 
between 60-120x10-7m3kg- 1 from the bedrock limestones of the Lac d' Annecy region 
and Flower et al. (1984) measured values of 40-160x10-7m3kg- 1 from sediment cores 
recovered from the Dayat-er-Roumi catchment in Morocco. In the Kopais Basin, 
nearly 200km SE of Ioannina, AlIen (1986) attributed her low susceptibility values of 
between O.6-2.8x10-7m3kg-l to the high proportion of calcite in the sediment (up to 
93%) and accordingly applied a correction factor. In the case of Ioannina, however, 
carbonate content (presumed to be dominantly calcite) rarely exceeded 25% (see 
section 4.2.1 below), so the dilution effect of diamagnetic calcite is likely to be 
negligible and no correction factor has been applied. Low susceptibility values are 
therefore presumed to reflect either a low initial concentration of magnetic particles 
within the catchment soils, a low proportion of detrital material reaching the central 
lake site of core 284, or a combination of both. 
Between approximately 165-220m, susceptibility values are much higher (peaking at 
nearly 19x1O-7m3kg- 1 at 180.20m). Detailed examination of the sediment over this 
interval revealed the increased concentration of small, black aggregated iron-rich 
nodules (averaging -1-2mm diameter) that were rare elsewhere in the sequence (see 
SEM results, below). Further analysis of these nodules was undertaken to test 
whether they were responsible for the high susceptibility values at these levels . 
Several nodules were picked from the sediment, packed together in clingfilm and then 
placed in an empty susceptibility measuring pot (suspended in the centre of the pot, 
again using clingfilm). Their mass specific susceptibility was measured (using the 
method outlined in section 3.2.2) on a low frequency setting. The mass of the sample 
RESULTS. 43 
0.0 0.0 
10 .0 10 .0 
20 .0 20 .0 
30 .0 30 .0 
40.0 40 .0 
50.0 50.0 
60 .0 60 .0 
70 .0 70.0 
80.0 80.0 
90 .0 90 .0 
100.0 100.0 
110 .0 110 .0 
120 .0 120.0 
130 .0 130 .0 
140 .0 140.0 g 
150.0 150 .0 g 
-5 
-5 0.. 
., 160 .0 160.0 0.. Cl ., Cl 
170 .0 170.0 
180.0 180 .0 
190.0 190.0 
200.0 200.0 
210 .0 210.0 
220 .0 220 .0 
230.0 230 .0 
240.0 240 .0 
250 .0 250.0 
260.0 260 .0 
270.0 270.0 
280 .0 280 .0 
290 .0 290 .0 
300.0 300 .0 
310.0 3 10.0 
0 5 10 15 20 -I 0 2 3 
Magnetic susceptibility Magnetic susceptibility 
Fig. 4.1 Magnetic susceptibility profile of core 284 (sampled every -1 m) . Left-hand graph shows 
entire data set. Right-hand graph shows enlarged horizontal scale to illustrate lower amplitude 
variability. Magnetic susceptibility units are X 10-7 m3kg-l• 
RESULTS. 44 
was O.013g, the volume was -27mm3 and the meter reading (after subtracting 
background levels) was 23 .3. Because the volume of the sample was much less than a 
standard pot, a 20% correction factor was applied. Using the equation outlined in 
section 3.2.2, a mass specific magnetic susceptibility value of 2. 16xlO-5 xlO-7m3kg-' 
was obtained. This value is reasonably close to that obtained from the iron oxide 
minerals magnetite (5xlO-5 xI0-7m3kg-') or maghaemite (4x10-5x10-7m3kg-'). The 
origin of these potential carriers is discussed in section 7.2.3. 
To determine whether these high susceptibility values were associated with catchment 
erosion rather than any other process, the frequency dependent susceptibility of the 
original sediment samples was calculated, using the following relationship: 
where Xfd % = frequency dependent magnetic susceptibility; Xhf = high frequency mass 
specific magnetic susceptibility; and X'f = low frequency mass specific magnetic 
susceptibility. Commonly expressed as a percentage of total low frequency 
susceptibility (as here), frequency dependent susceptibility indicates the delayed 
response of viscous magnetic grains at the stable single domain/superparamagnetic 
grain-size boundary to the applied magnetising field (Yu and Oldfield, 1989). The 
original data are shown in table 4.3. The results (all values reasonably close to zero) 
suggest that the carrier in these levels (magnetite/maghaemite, as identified above) is 
fine-grained, of single domain and most likely to be of detrital origin (L.P. Zhou, 
pers. comm.) . It is therefore probable that these magnetic grains derive from the 
weathering of specific horizons in the catchment (most likely from within the flysch) 
and may represent times of increased pedogenic enhancement under 'Mediterranean' 
climatic conditions. This interpretation is discussed further in section 7.2.3 . 
Similar peaks can also be easily distinguished in the magnetic susceptibility curve for 
core 249 (fig. 4.2). Although the sampling resolution was not as fine as for 284 (due 
to the availability of remaining sub-samples), several major points of correlation 
between the two records are distinguishable. For example, the peak at 111 .50m in 
249 is directly correlated with the peak at 180.20m in 284. Other peaks are matched 
as shown. By using major susceptibility variations to correlate the two cores, the 
pollen data and chronological framework from 249 can be related to 284. This is 
discussed at length in section 5.2.2. 
RESULTS. 45 
10 
20 
30 
40 
50 
60 
70 
80 
90 
100 § 
..c:: 
0.. 
<1) 
110 Cl 
120 
130 
140 
150 
160 
170 
180 
190 
0 5 10 15 20 25 
Magnetic Susceptibility 
Fig. 4.2 Magnetic susceptibility profile of core 249 (sampled every -Im). Magnetic susceptibility 
units are X 10.7 m3kg- 1• 
RESULTS. 46 
Depth (m) Xhr Xlr Xrd (%) 
180.20 18.79 18.93 0.74 
181.35 10.40 10.48 0.76 
182.60 12.29 12.39 0.81 
183.30 12.01 12.08 0.58 
184.75 16.30 16.42 0.73 
185.70 9.85 9.95 1.01 
Table 4.3 Mass specific and frequency dependent magnetic susceptibility values for selected levels in 
284. Units of Xhfand Xlr are x 1O.7m3kg·' . 
A detailed chemical analysis of these grains was beyond the scope of this study, but 
spectral analysis (using an X-ray micro analyser attached to the SEM) indicated a 
dominantly iron oxide composition, with a coating of silica (see SEM analysis results, 
below). The aggregated grains were composed of individual angular octohedral 
particles less than half a micron in size (illustrated in plate I, fig. A). These results 
support the suggestion that the magnetic carrier is a fine-grained, detritally-derived 
ferrimagnetic iron oxide such as magnetite/maghaemite. It is not clear whether this is 
the same carrier as for the rest of the sequence, where much lower values are the 
norm. This may be a reflection of much lower concentration levels, the presence of a 
different carri~r, or may indicate an authigenic source, such as from magnetotactic 
bacteria (e.g. AlIen, 1986). Preliminary X-ray diffraction (XRD) analyses carried out 
by P. Rowe (UEA) in conjunction with the U-series dating work (section 5.6) reveal 
the presence of both goethite and magnetite between 83-100m. Further detailed rock 
magnetic analyses, coupled with additional XRD analyses and a comprehensive study 
of grain-size distributions, are necessary to determine the nature of the carrier(s) 
throughout the entire core. 
4.1.3 Particle-Size Analysis 
Particle-size analysis of the sediment was carried out between 11 0-7 Srn. The age 
model (developed in chapter 5) suggested that this interval spanned the end of the 
penultimate glaciation (MIS-6) and the whole of the Eemian (MIS-Se). A range of 
material from full glacial to full interglacial could therefore be analysed. 
80~  81 82 83 80 81 82 83 
84 84 
85 85 
86 86 
87 87 
88 88 g 89 89 g 
.<: 90 
c.. 91 .., 
Cl 92 
90 .<: 
91 c.. .., 
92 Cl 
93 93 
94 94 
95 95 
96 96 
97 97 
98 98 
99 99 
100 100 
101 101 
102 102 
I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I 
-2 -1 0 1 2 3 4 5 6 7 8 9 10 -2 -1 0 1 2 3 4 5 6 7 8 9 10 30 25 20 15 10 5 0 
Phi Units Phi Units Carbonate Content (%) 
Total Raw Sediment Isolated Carbonate Fraction 
Fig. 4.3 Particle-size data. Left-hand graph shows total sediment data, sampled at lm intervals. Centre graph shows isolated carbonate fraction, also sampled 
at lm intervals. Right-hand graph shows total carbonate content as measured every lOcm by the loss-on-ignition method, plottede in reverse. 
~ 
~ 
~ 
• 
.j::>. 
-..l 
RESULTS. 48 
The particle-size data are presented in fig . 4.3 . The plots represent typical particle­
size distributions for the raw sediment and the isolated carbonate fraction . The raw 
sediment plot suggests that the lithology for this interval should be classified as a fine 
silty sand (the silt - sand boundary is taken to be 4<1>, 63~m). This interpretation is 
slightly at odds with the visual examination (section 4.1.1), which classified the 
sediment over this interval as being a clayey silt (i.e. much finer). Subsequent SEM 
analysis (see below) confirmed that the sediment consisted largely of aggregated 
grains (composed of fine silt- and clay-sized particles), bound together by means of a 
siliceous cement. SEM analysis also confirmed that the procedure used to 
disaggregate the raw sediment was not sufficient to break all of the aggregated grains 
into their component particles. The siliceous coating could clearly act as a capable 
cementing agent. Only extended mechanical rinsing of the sediment fractions through 
a succession of finer sieves was found to be effective in separating the aggregated 
grains. As a result, the particle-size plots, as they stand, are biased towards a coarser 
overall grain size, although the relative distribution of size classes is likely to still be 
applicable. 
The overall grain-size distribution bears a good relation to the carbonate curve for the 
same interval (generally interpreted as a proxy for productivity), the implications of 
which are discussed in section 7.2.1. As might be expected, the isolated carbonate 
fraction plot also correlates well with the carbonate content curve (fig. 4.3). In 
general, however, the main feature of the plots is that between approximately 100-
85m, interpreted as being a 'warm' phase corresponding to the Eemian (section 5.8), 
the dominant particle-size coarsens noticeably. 
It is clear from this preliminary study that there is considerable scope for further 
detailed grain-size analyses. Future work might initially reassess the results of this 
study in light of the surprising resilience of the siliceous cement in holding the 
aggregated particles together. A more accurate idea of the difference in grain-size 
distribution between glacial and' interglacial periods could then be obtained. It should 
be possible to pick up some evidence of aeolian input, particularly during glacial 
phases. The soil profiles of the region display a marked bimodal distibution (see 
section 2.3), with a coarser, quartz-rich silt-sized fraction becoming more dominant 
during glacial periods (van Andel, in press). It is thought that this represents 
increased aeolian input from source areas such as North Africa and, possibly, from the 
loess deposits in central Europe. 
RESULTS. 49 
4.1.4 SEM Analysis 
SEM analysis of the sediment between llO-75m was used to investigate several of the 
specific questions posed by the particle-size results (in particular, the nature and grain 
size of the carbonate fraction) . In general, most of the sediment was composed of 
aggregated grains of fine silt-sized carbonate particles, clay minerals and a high 
proportion of fine siliceous material (including diatom fragments) . The grains were 
coated in silica, suggesting a post-depositional diagenetic effect (probably due to 
percolating groundwaters) that was responsible for aggregating and cementing the 
smaller sedimentary particles together. A minor contribution (in the order of a few 
percent) to the total sediment by detrital material in the 'colder' levels was evidenced 
by occasional feldspar, quartz and limestone grains that showed signs of abrasion due 
to transport. 
Plate I, fig . C, shows a typical aggregated grain from 89.02m. Some of the carbonate 
crystals are shaped on a range of scales like hexagonal plates, reminiscent of mica 
'books'. Some larger ones are illustrated in plate I, fig . D. They are typical of 
carbonates that have precipitated rapidly in waters with a very low concentration of 
magnesium (Folk, 1974). Other carbonate crystals within the aggregated grains 
display a more irregular polyhedral morphology, also typical of low-Mg, low salinity 
lacu~trine precipitates (e.g. Kelts and Hsti, 1978). 
Other components of the sediment included silica-coated pollen grains (a typical 
Pinus grain is shown in plate I, fig . B) and framboidal iron sulphide spherules (plate I, 
fig. E) . These latter are often formed in the uppermost few millimetres of lake 
sediments where relatively low-energy environments of deposition coincide with a 
source of metallic ions and decomposing organic matter (e.g. Vallentyne, 1963; 
Wiltshire et al., 1994). Under these quiet, locally anoxic conditions , microbial 
activity (both iron reducing and sulphur reducing) helps to form the authigenic (sensu 
stricto) spherules within the sediment. Several iron sulphide mineral species have 
been identified as forming spherules under these conditions (including marcasite, 
pyrrhotite and melnikovite), however, the most commonly found is pyrite (FeS2). A 
detailed chemical analysis of the spherules was beyond the scope of this study, 
although the X-ray microanalyser attached to the SEM clearly demonstrated that they 
were iron sulphides. They were therefore assumed to be pyrite. 
Several aggregated black mineral grains recovered from those parts of the sequence 
with high magnetic susceptibilities were also analysed under the SEM. They were 
composed of many tiny individual grains, less than half a micron in diameter (plate I, 
RESULTS. 50 
fig. A). The X-ray micro analyser indicated that they were iron oxides, coated in a 
si1ca cement. The distinct octagonal crystal form displayed by many of the grains is 
typical of minerals such as magnetite, an iron oxide identified as one of the potential 
magnetic carriers in this part of the core (section 4.1.2). 
4.2 CHEMICAL ANALYSES 
4.2.1 Sediment Geochemistry 
(a) Loss-an-Ignition 
Despite the lithological homogeneity of the core, the loss-on-ignition curves show a 
high degree of structure (figs. 4.4 and 4.5). Because of the high number of data points 
(n = 1,972), a nine-point running mean has been applied to smooth both curves and 
reveal some of the more dominant aspects of their structure. 
As explained in section 3.3.1, the loss-on-ignition curve for 550°C indicates 
variability in the organic matter content of the sediment. A verage values decrease 
from about 6% at the bottom of the core to about 4% at about lOOm. After this, 
values leap to an average of between 8-10% for the remaining part of the record. 
Superimposed on these broad-scale changes are higher-order oscillations. For 
example, the top lOOm exhibits substantial variation, as does the lowermost 80m. 
There are also distinct peaks in organic matter content centring around 190m and 
l50m. 
The loss-on-ignition curve for 950°C indicates variability of carbonate content in the 
sediment. Values vary between about 2% to 40%, although as with the organic matter 
content curve, higher order variability is clearly superimposed on more broad-scale 
first-order changes. Unlike the organic matter curve, the base-line average does not 
appear to gradually decrease upwards, but remains constant at around 5%. Smaller 
scale variability (in terms of wavelength) is obviously more complex than for the 
organic matter curve. A comprehensive interpretation is therefore given in chapter 7, 
where the connection between carbonate content and aquatic productivity is discussed 
at length. A composite plot of organic matter and carbonate contents is presented in 
fig. 4.6. 
RESULTS. 51 
0 0 
10 10 
20 20 
30 30 
40 40 
50 50 
60 60 
70 70 
80 80 
90 90 
100 100 
110 110 
120 120 
130 130 
140 140 
150 150 g g 
160 160 
.s 
.s 0. 0. <\) 170 170 <\) Cl Cl 
180 180 
190 190 
200 200 
210 210 
220 220 
230 230 
240 240 
250 250 
260 260 
270 270 
280 280 
290 290 
300 300 
310 310 
0 5 10 15 20 25 0 5 10 15 20 
Organic Matter Content (%) Organic Matter Content (%) 
Fig. 4.4 Organic matter content, sampled every 10cm for the top 100m, every 20cm for the remainder 
(n = 1,972). Left-hand graph shows raw data, right-hand graph shows smoothed data (9-pt. running 
mean). 
RESULTS. 52 
0 0 
10 10 
20 20 
30 30 
40 40 
50 50 
60 60 
70 70 
80 80 
90 90 
100 100 
110 110 
120 120 
130 130 
140 140 
150 150 
§ 160 160 § 
.s 170 170 .s 0. 0. (!) (!) 
Q Q 
180 180 
190 190 
200 200 
210 210 
220 220 
230 230 
240 240 
250 250 
260 260 
270 270 
280 280 
290 290 
300 300 
310 310 
0 10 20 30 40 50 0 10 20 30 40 
Carbonate Content (%) Carbonate Content (%) 
Fig. 4.5 Carbonate content, sampled every 10cm for the top lOOm, every 20cm for the remainder (n = 
1,972). Left-hand graph shows raw data, right-hand graph shows smoothed data (9-pt. running mean). 
RESULTS. 53 
0.0 
10.0 
20.0 
30.0 
40.0 
50.0 
60.0 
70.0 
80.0 
90.0 
100.0 
110.0 
120.0 
130.0 
140.0 
:§: 150.0 
160.0 
.<: 
0.. 
I!) 170.0 Cl 
180.0 
190.0 
200.0 
210.0 
220.0 
230.0 
240.0 
250.0 
260.0 
270.0 
280.0 
290.0 
300.0 
310.0 
o 10 20 30 40 50 60 70 80 90 100 
Percent of Total Sediment 
Fig. 4.6 Summary loss-on-ignition plot showing organic matter (stippled), carbonate content (shaded) 
and inorganic matter (white). 
-7 
82 
83 
84 
85 
86 
87 
88 
89 g 90 
..c 
0.. 
0) 91 
Cl 92 
93 
94 
95 
96 
97 
98 
99 
100 
101 
-3 -2 -I o 2 
8 13C (%0) 
-6 
3 4 
8180 (%0) 
-5 -4 -3 -2 
82 
83 
84 
85 
86 
87 
88 
89 
90 g 
..c 
91 0.. 
0) 
92 Cl 
93 
94 
95 
96 
97 
98 
99 
100 
101 
Fig. 4.7 Stable isotopes between lOO-83m derived from bulk carbonates. Isotope values quoted relative to V-PDB. Zonation details given in text. 
::>;:, 
~ 
~ 
2:l 
• 
Ul 
.j:>. 
RESULTS. 55 
(b) Sediment Isotope Geochemistry 
The stable isotope results derived from the analysis of bulk sediment carbonate 
between 83-100mare shown in fig. 4.7. Values of 8I3C and 8 180 are both given 
relative to V-PDB. 8I3C varies between about +3.5%0 and -2.3%0, whereas 8180 varies 
between about -2.2%0 and -6.2%0. Covariance between the two curves is reasonably 
strong (r = 0.58). The similarity in the behaviour of both curves allows the plot to be 
split into several discrete zones for the purposes of description. The interpretation 
and significance of these data is discussed at length in section 7.4.2. 
lOO-98.Sm 8I3C values oscillate initially between about +0.5 and +3%0, before falling 
to fluctuate around 0%0. Over the same interval, 8180 oscillates around -3.5%0. 
98.S-96.0m 8 I3C values decrease rapidly at the very beginning of the zone from 
around 0%0 to -2 .1%0 (at 98 .02m), before increasing gradually to around -0.5%0. 
Values increase slightly at the end of the zone. The 8 180 values display a similar 
pattern, decreasing to a minimum of -6%0 at 98.02m and then increasing to -3.8%0 at 
96.80m. However, the 8180 then undergoes a second oscillation before the end of the 
zone. 
96.0-90.90m Values of 8 13C remain at a fairly constant level throughout (--0.8%0), 
although values are slightly lower from about 92.5m. The values of 8180 gradually 
decrease at about 93 .lOm from an average of around -4.5%0 to an average of -4.8%0. 
90.90-86.10m 8 I3C falls sharply at the beginning of the zone, from values of around 
-0.8%0 to -1.2%0 and then increases steadily throughout (except for a single point at 
around 88m) to a value of about -0.75%0. Values of 8180 also decrease at the 
beginning of the zone, although more slowly than 813C. With the exception of the 
same point at about 88m, the value of 8180 remains roughly constant at around -5.2%0. 
86.10-83.0m Values of 8I3C fall sharply at the beginning of the zone and then 
oscillate around -0.5%0, followed by a very sharp increase to +3.5%0 at 83.5m. Values 
of 8180 decrease slightly at the beginning of the zone, then increase gradually until 
they repeatedly oscillate (to maximum of approximately -2.2%0) from 83 .80m. 
The isotopic signature of two bedrock samples was also analysed. One sample 
consisted of limestone pebbles collected from site 2, located on the north-eastern 
shore of the lake. The other sample consisted of limestone bedrock recovered from 
RESULTS. 56 
the base of core 256 (the marginal core drilled to the south-west of 284), at a depth of 
132m. The results were as follows: 
Site 2 material 
Core 256 bedrock 
OBC 
+1.97%0 
+1.37%0 
0180 
-1.04%0 
-2.63%0 
As before, 013C and 0180 are quoted relative to V-PDB. Faure (1986) suggests that 
the majority of marine carbonates from Cambrian to Tertiary age have 013C values 
close to zero (relative to PDB). He also suggests that 0 180 values for marine 
carbonates are commonly between +20-30%0 (relative to SMOW), depending on age. 
The carbonates in the Ioannina catchment are all marine Mesozoic limestones 
(described in section 2.2) and the differences between the two measured samples can 
be explained if they represent limestones of different age. The 256 sample is likely to 
be the Upper Senonian Limestone that underlies the majority of the basin. The site 2 
material could be derived from any of the limestone outcrops along the northern shore 
and could therefore be of Mid-Jurassic - Cretaceous in age. The ol3C values of both 
samples are broadly in accordance with values that might be expected from marine 
carbonates of these age, as are the 0180 values. The conversion between 0180Y_SMOW 
and 0180Y_PDB is given by the equation: 
Thus, the bedrock samples have 0180 values (relative to V-SMOW) of +29.78%0 and 
+28.12%0, completely in accordance with the expected values suggested by Faure 
(1986). 
4.2.2 Modern Water Chemistry 
The composition of the modern waters are typical of a temperate hard-water lake, 
being dominated by Ca2+ and HC03- ions. Table 4.4 shows the major chemical 
parameters of the lake (full details are given in appendix B). As might be expected in 
a karstic area, the lake is reasonably alkaline, having a pH that averages 7.64, a 
salinity that averages 0.35%0 (i.e. freshwater) and an Mg/Ca ratio that is low, at 0.34. 
The results are discussed at length in section 7.4.1. 
'C' 
~ 
Cl 
CoO 
-10 ........ -----------:-.'---...." 
.' ... 
-20 
-30 
-40 
-50 
I 
I 
I 
I 
-60 
-10 
I 
I 
I 
I 
I 
I 
I 
I /// :/ 
MWL/ 
I 
I 
I 
I 
,./ .. " ..... 
,. ... 
,. .... 6 
,','" 
I.:~:/ 
/' 
-7 .5 
I 
-5 
Gradient = 4.8 
1 
-2.5 
818 0 (%0) 
o 
RESULTS. 57 
Fig.4.8 Modern lake waters . 0 180 and oD both relative to V-SMOW. MWL = Meteoric Water Line. 
-3 
.. 
-4 
-5 11. 
-6 3. 
'C' 
-7 ~ 7. 
u 
<"l 
CoO -8 
-9 
-10 
-11 
-1 2 
6. 
I 
-8 -6 -4 -2 o 
8180 (%0) 
Fig. 4.9 Modern lake waters . 0 180 relative to V-SMOW, ol3e relative to V-PDB. 
"I 
RESULTS. 58 
Site 8180 813C 8D Mg Ca Na K pH Salinity 
1 -0.7 -3 .8 -13 10.37 38.76 19.77 3.49 7.53 0.27 
2 -0.5 -3.5 -12 10.38 44.02 20.01 3.43 7.48 0.27 
3 -0.6 -6.1 -12 10.39 52.78 20.72 3.07 7.77 0.41 
4 -0.5 -3 .2 -12 9.99 40.00 19.74 3.37 7.42 0.37 
5 -0.4 -3 .6 -12 10.06 39.81 19.67 3.51 7.53 0.30 
6 -7.0 -12.4 -43 5.39 70.37 57.93 0.60 7.64 0.54 
7 -1.1 -7.4 -14 14.21 91.70 25.26 1.64 7.83 0.56 
8 -0.6 -4.0 -12 10.11 40.20 19.08 3.46 7.53 0.27 
9 -0.6 -3 .8 -12 10.30 41.23 19.33 3.44 7.72 0.28 
10 -0.5 -3.5 -13 10.01 39.00 18.94 3.48 8.09 0.29 
11 -0.4 -5 .0 -12 10.14 39.57 19.44 3.28 7.54 0.33 
Table 4.4 Summary of water chemistry data for modern sites. 8180 and 8D values in per mil (%0) 
relative to V-SMOW. 8 13C values in per mil (%0) relative to V-PDB . All trace element values in mg/I. 
Salinity values in %0 (calculated as Total Dissolved Solids, TDS). See text for details. 
The stable isotope measurements indicate that for the most part, the waters are 
relatively homogeneous (fig. 4.8). They lie on an evaporation trajectory (gradient 
-4.8) from the meteoric water line (MWL) for the Mediterranean. This has a slightly 
different gradient from the global MWL, given by the equation oD = 8 x 0 180 + 20 
(Gat and Carrni, 1970). Values of oI3C are given relative to V-PDB. Values of 0180 
and oD are given relative to V -SMOW. The covariance between the 0180 and the 
oI3C values is reasonably strong (r = 0.91), illustrated in fig . 4.9. The significance 
that this has for interpreting the hydrological status of the lake is discussed in section 
7.4.1. 
Although this is the first isotopic study of the waters from Lake Pamvotis , the trace­
element results are comparable with those obtained by Marinos (1974) and Overbeck 
(1980). In particular, Overbeck (1980) notes the eutrophic status of the lake and the 
high levels of silica, which he attributes to the influence of siliceous material in the 
catchment. 
4.2.3 Ostracod Shell Chemistry 
The ostracod stable isotope results are presented in fig . 4.10. Only ostracods from the 
top 25m of the core were measured (corresponding approximately to the period from 
just before the Last Glacial Maximum into the Holocene: see chapter 5). Insufficient 
valves were preserved from the Eemian section of the core (approximately 100-84m) 
to allow isotopic determinations to be made. As with previous carbonate 
RESULTS. 59 
0 0 
2 2 
3 3 
4 4 
5 5 
6 6 
7 7 
8 8 
9 9 
g 10 10 g 
.s 11 11 ..c: 
Cl. i5.. C1.l 12 12 C1.l Cl Cl 
13 13 
14 14 
15 15 
16 16 
17 17 
18 18 
19 19 
20 20 
21 21 
22 22 
23 23 
24 24 
25 25 
-5 -4 -3 -2 -1 0 -3 -2 -1 0 2 3 4 
8 l3e (%0) 8180 (%0) 
Fig. 4.10 Stable isotope results derived from ostracod shell calcite (quoted relative to V-PDB). 
Raw data in thin solid line joined by points; smoothed data (3-pt. running mean) in heavy black line. 
RESULTS. 60 
818 0 (%0) 
-3 -2 -I 0 2 
3 3 
4 4 
5 5 
6 6 
7 7 
8 8 
9 9 
IQ IQ 
11 11 
12 12 
§ 13 13 § 
-5 ..c 
0. 14 14 P. 0) 0) Cl Cl 
15 15 
16 16 
17 17 
18 18 
19 19 
20 20 
21 21 
22 22 
23 23 
24 24 
-5 -4 -3 -2 -I 0 
Fig. 4.11 Comparison of smoothed stable isotope results (3-pt. running mean) derived from ostracod 
shell calcite (quoted relative to V -PDB). Covariances: below ISm, r = 0.31 ; above ISm, r = 0.75. 
RESULTS. 61 
measurements, 813C and 8180 values are given relative to V-PDB. Three-point 
running means have been applied to both curves for the purposes of clarity and to 
minimise noise. The oxygen curve varies over about 6%0 (from approximately +3%0 
to -3%0), although the smoothed curve reduces the overall variability to about 4%0. 
The 813C curve varies over about 5%0 (approximately 0 to -5%0), although again, the 
overall variability is reduced to about 3.5%0 by the smoothing. Overall covariance 
between the curves is generally weak (r = 0.37 for the raw data; r = 0.54 for the 
smoothed data). The two smoothed curves are compared in fig. 4.11. However, 
covariance varies between different parts of the record. Below 15m the curves covary 
with an r-value of 0.31. Above 15m, however, the r-value improves to 0.75. The 
possible significance of this is discussed in section 7.4.3. 
Duplicate sets of valves were sampled at selected intervals to test for replicability of 
results. Values were plotted as straight averages between the two determinations 
where appropriate. Lack of sufficient material meant that it was not possible to carry 
out enough measurements for the comparisons to be statistically significant. 
Therefore, whilst the results are shown in table 4.5 for the sake of interest, they have 
not been included in the final data set. It is clear that whilst the replicability of the 
8l3C is reasonable in two of the cases, the third value and at least two of the 8180 
values exhibit marked differences. It is likely that this is related to the variation 
inherent within the samples themselves, as each 4cm slice of sediment at this level is 
thought to represent at least 40 years (see chapter 5). These results highlight the 
necessity for further duplicate measurements, so that a statistically valid estimate of 
isotopic variability within samples can be calculated. 
Several duplicates were also used to compare different methods of removing 
contaminant organic matter from the sample. These entailed either using 40 volume 
hydrogen peroxide in the disaggregation stage, or vacuum roasting the sample (420°C, 
lO·lmbar, for 30 minutes) prior to isotopic measurement. The results, compared to the 
manual brush cleaning method used as the standard technique in this study, are shown 
in table 4.6. Once again, insufficient measurements were made for the results to be 
considered statistically significant and therefore the values were not included in the 
final data set. It appears that whilst there is some evidence to suggest that the 
peroxide pre-treatment has a relatively minor effect on 8l3C, its effect on 8180 is 
considerably more marked. The vacuum roasting treatment appears to significantly 
affect both 813C and 8180, suggesting that caution is needed in its application. As 
outlined in section 3.3.3, different workers employ a diverse range of preparation and 
pre-treatment techniques. There is clearly an urgent need to develop a standard pre-
RESULTS. 62 
treatment method (similar to that now used in foraminiferal studies) based on rigorous 
quantitative analyses of large data sets. 
Depth (m) No. Valves Average Average 
13.82 8 -3.77 -2.50 
13.82 6 -3.74 -3.76 -1.52 -2.01 
.... .... .......... ......... . ......... . ........ . .. . .. ........ .... .. ........ ....................... .. ............. ..... ... . . . ......................... " ••• ,', ........... . . . .. . . . .... . ....................................... 000 
16.80 9 -1.53 -1.21 
......... .. ... ~. ~:. §.9. .............................. ? ............................... :.~.: .1.9 ................... .... .... :.~.:.1. T ............... .. ......... :.9.:.?} .......................... :.9.:.7.~ ....... ... .. . 
23.0 
23.0 
9 
9 
-2.12 
-1.76 
-0.55 
-1.94 -0.35 -0.45 
Table 4.5 Comparison of duplicates. No. of valves = number of valves cleaned and used in a single 
sample. o l3C and 0180 values quoted as per mil (%0) relative to V-PDB. See text for details. 
Depth (m) 
16.80 
16.80 
No. Valves 
12 
12 
PN 
P 
V 
-1.52 
-1.80 
Variation 
3.4 
22.4 
-1.62 
-1.13 
VaI"iation 
125.0 
56.9 
.. ........ ... ~.~:.§9. .................. ... .. ........ ? ...... ... .. .. .. .............. M ....... ... ..... .. :}.:1T .................................................. :9.:7.~ .... .. .......... .......................... . 
21.20 
21.20 
12 
12 
P 
V 
-2.26 
-2.20 
26.2 
22.9 
-1.40 
-1.60 
141.4 
175.8 
........... }.~:}9. ................... .... ...... .. ? ........... .................. M ................. :J.:7.? .................................... ......... .. .. .. :9.::?§ .. ................ ....................... .. 
23.0 
23.0 
12 
12 
P 
V 
-2.09 
-2.16 
7.7 
11.3 
-0.63 
-1.10 
40 
144.4 
....... ....... ~.~:9. ................................ ? ............................. M ................. :}.:~1 ................................................... :9.:1? ...................................... ... .. 
23.79 
23 .79 
23.79 
12 
12 
9 
P 
V 
M 
-3.27 
-2.78 
-3.46 
5.5 
19.6 
-0.55 
-1.46 
-0.03 
1733.3 
4766.7 
Table 4.6 Comparison of ostracod preparation treatments. No. valves = number of valves cleaned and 
used in each single sample. P = hydrogen peroxide treatment. V = vacuum roasting treatment. M = 
manual brush cleaning. ol3C and 0180 values quoted as per mil (%0) relative to V-PDB . Variation = % 
variablity of values from the M-value (taken to be the 'true' value) . See text for details . 
RESULTS. 63 
4.3 SUMMARY 
This chapter presented the results of the physical and chemical analyses of the core 
material, in addition to modern water chemistry determinations. A detailed 
lithological description of the core was followed by a discussion of its magnetic 
susceptibility characteristics. Although susceptibility values were generally low, 
several discrete peaks were identified that were associated with small, black, 
aggregated grains in the sediment. Further analysis of these grains suggested that 
they were the source of the high susceptibility values and that they were composed of 
a ferrimagnetic iron oxide mineral, most probably magnetite/maghaemite. The 
possible origin of these carriers within the Ioannina catchment under normal 
Mediterranean climatic conditions was also discussed. 
The results of the particle size analysis (undertaken between lOO-83m) were then 
presented. Separate fractions were analysed, consisting of the total sediment, the 
isolated carbonate fraction, the isolated organic fraction and the remaining 
'minerogenic' fraction. Results suggested that the sediment should be classified as a 
fine silty sand. This was at odds with the visual lithological description of a clayey 
silt. The difference was explained by observing that the preparation procedure for the 
particle size analysis was insufficient to completely disaggregate the sediment. 
Overall results were therefore regarded as being artificially biased towards a coarser 
grain sizes and should therefore be treated with caution. 
SEM analyses revealed that · the problem with the disaggregation procedure was 
caused by a silica coating (probably of diagenetic origin) that cemented the individual 
grains of the sediment together into composite masses. The grains were largely 
aggregates of siliceous material (including many diatom fragments), clay minerals 
and carbonate particles. The carbonate grains were of particular interest, displaying 
either a polyhedral or a hexagonal structure, indicative of rapid precipitation in fresh 
waters that had a low concentration of magnesium. Other grains of interest included 
framboidal iron sulphide spherules (assumed to be pyrite) and the magnetic particles 
contributing to the intervals of high magnetic susceptibility. 
Work on aspects of the sediment geochemistry was then presented. Despite the 
lithological homogeneity of most of the core, the loss-on-ignition procedure revealed 
a high degree of structure in both the organic matter and carbonate content curves. 
The isotopic signals derived from bulk sediment carbonates between lOO-83m also 
revealed a distinct degree of structure, with a relatively strong covariance between 
RESULTS. 64 
813C and 8180 values. It was suggested that this might be indicative of a closed lake 
system. 
The results of the modern water chemistry analyses indicate that Lake Pamvotis can 
be described as a typical eutrophic freshwater temperate hard-water lake, with waters 
that have undergone evaporation. Reasonably strong covariance between 813C and 
8180 values suggested that the modern day lake might also be considered as a closed 
lake system. 
Finally, the results from the isotopic analyses of the ostracods from the top 25m of the 
core were presented. These showed interesting patterns of variability, particularly in 
their covariance, which markedly improved over the top I5m or so of the record. 
These results are discussed further in chapter 7, where, along with the age model 
derived in chapter 5 and the faunal data described in chapter 6, detailed interpretations 
are suggested and a synthesis of the all the results presented. 
5. CHRONOLOGY 
5.1 INTRODUCTION 
Establishing detailed chronological control for long terrestrial sequences is 
notoriously difficult. Even assuming that the record is continuous (as is the case with 
Ioannina 284), accumulation rates are likely to have varied between and within glacial 
and interglacial periods. No single technique exists that will provide a consistent 
chronology for a core of this age, yet clearly this is essential if any coherent 
palaeoenvironmental story is to be derived. The simple practice of matching proxy 
curves from different sequences is clearly unacceptable without some common basis 
of comparison and independent chronological control. 
The approach taken in this study is first to attempt a correlation with core 249. A 
comprehensive pollen-derived stratigraphy exists for the 249 sequence and this has 
been successfully linked to the marine oxygen isotope record (Tzedakis 1991, 1993, 
1994). It is intended that a provisional age-model should be established for 284, 
based on the chronology derived from 249. Other dating techniques will then applied 
to the sequence (e.g . radiocarbon, magnetostratigraphy, U-series) in order 
independently to verify and refine the proposed age-model. Each procedure is now 
discussed in turn. 
5.2 CORRELATION WITH CORE 249 
Pollen analyses have traditionally provided a useful method for correlating terrestrial 
sequences. Although a full palynological study has yet to be undertaken on 284, 
Tzedakis (1991) carried out a comprehensive study on core 249, located less than a 
kilometre away and within the same sub-basin (I. Broussoulis, pers. camm.). If a 
secure basis for correlation with 249 can be established, then the chronology 
developed for that sequence can be applied to 284. The veracity of the age model 
established by Tzedakis (1991) for 249 is first discussed, followed by a suggested 
basis by which the two sequences may be correlated. 
CHRONOLOGY • 66 
5.2.1 Basis of the Age Model for 249 
A total of 217 pollen analyses were undertaken for the whole 185.50m length of core 
249 by Tzedakis (1991), an average of one every -85cm. Not only was he able to 
demonstrate the presence of well-preserved pollen throughout the entire core, but he 
was able to show that the succession of different taxa was cyclic and structured. He 
was therefore able to propose a zonation of the sequence based on vegetational 
succession, which essentially represented alternations between forest and open 
vegetation communities. These higher order changes were interpreted as reflecting 
shifts in climatic conditions from interglacial to glacial mode. Lower order variability 
between individual taxa was interpreted as representing vegetational changes within 
each interglacial or glacial. Temperate arboreal taxa were present throughout even 
full glacial stages, albeit at low frequencies. This clearly demonstrated that the site 
was refugial, from which temperate vegetation could expand after climatic 
amelioration (Bennett, et al., 1991; Tzedakis, 1991, 1993). No independent 
chronology was established for 249. Instead, a timescale was erected first by means 
of correlation between the pollen record from 249 and those of the two radiocarbon 
dated cores drilled at loannina (Bottema, 1974) and then by comparison with three 
other long European pollen sequences. These are now considered in turn. 
Two short cores, designated loannina I and 11 (Bottema, 1974), were recovered from 
the northern part of the basin (fig. 2.1). loannina I was thought to represent the last 
glacial period and the Holocene, based both on the vegetational succession and three 
radiocarbon dates (obtained from bulk samples). However, Bottema (1974) suggested 
that there were problems with accepting the two lower dates, which defined an 
interstadial that he designated zone T (characterised by high frequencies of Pinus, 
bracketed by high frequencies of Quercus and Abies). The lowest date of 37,660±930 
yrs BP (GrN-6529) was thought to be unsatisfactory because of possible 
contamination by younger material. Bottema (1974) also suggested that the upper of 
these two dates, 40,000±1,000 yrs BP (GrN-4793), was too young, despite it being 
near the limit of the radiocarbon range. In any case, Tzedakis (1991) noted that there 
was no immediately obvious correlative of this interstadial in 249. The Lateglacial­
Holocene transition in loannina I was dated at 10,190±90 yrs BP (GrN-4875), 
marking a decrease in Artemisia and chenopods and a corresponding increase in 
Quercus. This transition was identified in 249 at a depth of 17m. Much of the 
Holocene record from 249 was absent as the top 10.20m of the sequence was missing 
(due to extensive previous sub-sampling by IGME). However, the forest phase 
subsequent to the Lateglacial-Holocene transition in 249 resembled the beginning of 
CHRONOLOGY • 67 
the Holocene record from Ioannina I, being characterised by early rises in Quercus, 
Ulmus, Corylus and Tilia, with a later rise in Ostrya frequencies (Tzedakis, 1991). 
Ioannina 11 was thought to extend back to the last interglacial, based on the 
characteristic vegetational structure of the basal section (very high initial Abies 
frequencies, followed by high frequencies of Quercus, Ostrya, Ulmus, Tilia and other 
taxa). Bottema (1974) proposed an Eemian age for this interval, explaining the 
radiocarbon date of 45,800 +2,5001-1,900 yrs BP (GrN-6181) obtained on bulk 
sediment as possibly being subject to root contamination. However, the pollen 
signature of this interval bore no resemblance to any forest periods in 249, including 
the Eemian. Above this section, a zone devoid of pollen was succeeded by a 
Holocene sequence that matched Ioannina I. The correlation of 249 with any part of 
Ioannina I or 11 (with the possible exception of the early part of the Holocene) was 
therefore extremely tenuous, a point freely conceded by Tzedakis (1991). 
A strategy for establishing long-distance correlations between 249 and other long 
European pollen sequences was then proposed. Four important sequences were 
considered: Tenaghi Philippon, a record from north-eastern Greece, that may extend 
into the Lower Pleistocene (Wijmstra, 1969; Wijmstra and Smit, 1976; van del' Wiel 
and Wijmstra, 1987a, 1987b); Valle di Castiglione, a 250ka record from central Italy 
(Follieri et al., 1988; Narcisi et al., 1992); and the central French records of Grande 
Pile (Woillard, 1-978; de Beaulieu and Reille, 1992) and Les Echets (de Beaulieu and 
Reille, 1984, 1989), both of which extend back at least 140ka (Woillard, 1978; 
Woillard and Mook, 1982; Guiot et al., 1989). Although broad-scale similarities were 
demonstrated between 249 and all four of these records, only the Tenaghi Philippon 
correlation will be discussed further here, since the age model for 249 was essentially 
derived from this sequence. This in turn had been correlated with the marine oxygen 
isotope record. 
Tenaghi Philippon is the longest terrestrial palynological record from Europe. Results 
have been published from two cores, designated TF2 and TF3 (Wijmstra, 1969; 
Wijmstra and Smit, 1976; van der Wiel and Wijmstra, 1987a, 1987b). TF2 was a 
120m long core drilled in the late 1960s through the Philippi Plain, part of the Drama 
Basin in northeastern Greece. A timescale for the top 30m was initially proposed, 
based on a series of ten radiocarbon dates derived from bulk sediment (Wijmstra, 
1969) and the characteristic structure of the pollen succession. The timescale for the 
lower part of this core was derived only after it had been correlated with TF3 (van der 
Wiel and Wijmstra, 1987), for which palaeomagnetic data existed. TF3 was a 280m 
long core drilled in 1976 approximately 150m northwest of TF2 and was correlated 
CHRONOLOGY • 68 
with it on the basis of vegetational and lithological similarities. Although no 
radiocarbon determinations were obtained from TF3, a series of palaeomagnetic 
measurements were made by N.D. Opdyke at Lamont Doherty. Full details have not 
been published, but they are referred to by van der Wiel and Wijmstra (1987), who 
suggested that the Brunhes/Matuyama boundary was located at a depth of 134m and 
another reversal, possibly the Jaramillo, was located occurred 195-210m. This 
enabled Wijmstra and Groenhart (1983) to propose an age model for TF3 that utilised 
the radiocarbon dates from TF2 (Wijmstra, 1969), the age of the Brunhes/Matuyama 
boundary (assumed to be 730ka) and an inferred age of 120ka for the Pangaion 
interstadial (assumed to be the last interglacial) as control points. In this way, a time­
depth relationship was established and a correlation with the marine oxygen isotope 
timescale of V28-239 (Shackleton and Opdyke, 1976) was suggested. This age model 
for the Tenaghi Philippon sequence was later revised slightly (van der Wiel and 
Wijmstra, 1987a, 1987b), once new biostratigraphical data for north-western Europe 
became available (e.g. Zagwijn, 1985). Although the choice of suitable control points 
was limited and not entirely satisfactory from a stratigraphical point of view, it was on 
the basis of the vegetational comparison with other well established and 
chronologically robust north-west European records that the final age model could be 
justified. 
The proximity and stratigraphical length of Tenaghi Philippon meant that it was the 
obvious 'yardstick' against which 249 could be compared. Nevertheless, Tzedakis 
(1991) pointed out several problems when comparing pollen spectra from different 
sites in ge~eral, and from these two sites in particular. Ioannina and Tenaghi 
Philippon, although only approximately 350km apart, are subject to markedly 
different climatic regimes. A changing moisture gradient from west to east (high 
precipitation in the west, much lower in the east) has a significant effect on the type of 
vegetation prevailing at anyone time. For example, forest periods in the Tenaghi 
Philippon record were dominated by Pinus and Quercus. At Ioannina, however, 
Abies, Fagus, UlmuslZelkova and Carpinus were also significant during these periods. 
In addition, arboreal taxa were often absent from Tenaghi Philippon during the full 
glacial, whereas at Ioannina they were continually present during these periods at low 
frequencies (the significance of which has already been outlined above). 
Nevertheless, despite these and other problems, a considered and balanced basis for 
comparing higher order vegetation changes between the two records was proposed, 
making possible the correlation of several major interglacials and forested 
interstadials (Tzedakis, 1991). 
CHRONOLOGY • 69 
The possibility was then considered of linking 249 with the marine oxygen isotope 
record via the correlations established for the Tenaghi Philippon sequence by 
Wijmstra and Groenhart (1983). The problems of comparing two different 
parameters were discussed. Strictly speaking, the marine oxygen isotope record is a 
measure of global ice volume (and to a lesser extent, local temperature) variability 
(e.g. Shackleton and Opdyke, 1973; Shackleton, 1987). However, it is often taken to 
be a broad proxy for climatic change, with various odd-numbered isotope stages 
corresponding to interglacial or interstadial periods and even-numbered isotope stages 
corresponding to glacial or stadial periods. As Tzedakis (1991) points out, the pollen 
record is a complex multivariate signal representing a range of factors (climatic, biotic 
and so forth) that are at best regional in nature. However, with care the summary 
APINAP curve from a detailed pollen record can also often be interpreted as a 
suitable proxy for climatic change. High frequencies of arboreal pollen are generally 
interpreted as indicative of warmer periods, lower frequencies being indicative of 
colder periods, although this is not always necessarily the case (e.g. Coope et al., 
1997). On this basis, comparison of higher order variability between the two climatic 
proxies (marine oxygen isotopes and APINAP) was considered reasonable, although 
certain assumptions had first to be taken into account. These included consideration 
of the response times of terrestrial vegetation and the marine isotope record to 
climatic change (e.g. Wright, 1984). It was assumed for the purposes of correlation 
that the response of the two systems was essentially synchronous at the onset of 
interglacial periods, although as Tzedakis (1991) points out, vegetation may have 
been responding to a completely different set of climatic variables than those 
responsible for variations seen in the isotopic record. A more detailed discussion 
concerning leads and lags between the response times of different systems is 
presented in section 7.5.4. 
Tzedakis (1991) then proposed using the age model suggested for the Tenaghi 
Philippon sequence by Wijmstra and Groenhart (1983) as a 'stepping stone' by which 
the marine oxygen isotope stratigraphy could be transferred to 249. To improve upon 
the original correlation, the V28-239 Pacific sequence used by Wijmstra and 
Groenhart (1983) was rejected in favour of a composite record (V19-30 and V19-28, 
both from the eastern Pacific) that possessed improved resolution and 
chronostratigraphical control. 
Several direct correlations between interglacials recorded in 249 and the marine 
record were suggested. Despite some uncertainties, particularly good correspondence 
was demonstrated for the locally-named interglacials Dodoni (with MIS-11), Zitsa 
(with MIS-7) and Metsovon (with MIS-5e). Because of the generally poor sampling 
CHRONOLOGY • 70 
resolution, the weakest part of the scheme was acknowledged as being between the 
end of MIS-5 and MIS-1, not surprisingly also the weakest part of the correlation 
between 249 and Tenaghi Philippon. Other possible correlations were also discussed, 
particularly for glacial periods. Because a varying degree of uncertainty existed in the 
correlation of parts of 249 with the marine record, no attempt was made to assign an 
age to every pollen zone. 
Tzedakis (1991) was therefore able to propose an age model for the 249 sequence 
incorporating: a radiocarbon date of 1O,190±90 yrs BP for the Lateglacial-Holocene 
transition from Ioannina I (Bottema, 1974); a comprehensive correlation with Tenaghi 
Philippon on the basis of both similarities in AP/NAP structure and vegetational 
succession within different pollen zones; and a reasonably confident correlation with 
the marine oxygen isotope record. This tentative chronology suggested that the top 
162.75m of the 249 sequence extended back 423ka. Although Tzedakis (1991) 
discusses at length certain problems that still exist with respect to this age model, 
given the degree of uncertainty inherent in the dating of any long terrestrial sequence, 
confidence in his chronology for 249 would seem reasonable. 
5.2.2 Correlation of 284 with 249 
Because no detailed pollen stratigraphy yet exists for 284, a suitable alternative 
method for comparing the two sequences must be employed. The procedure using 
magnetic susceptibility for correlating cores from within the same lake basin is now 
well established. Thompson (1973) successfully matched susceptibility profiles from 
a series of cores taken from Lake Windermere in NW Britain. It was later shown that 
under certain circumstances it was possible to match susceptibility profiles from cores 
taken from different basins (e.g. Thompson et al., 1975; Dearing, 1983; Dearing, 
1986). 
As mentioned In section 4.1.2, the magnetic susceptibility signal from 284 is 
generally weak, except for certain levels where markedly stronger peaks are seen. 
These are thought to represent the inwash from the erosion of discrete horizons within 
the flysch units (outcropping to the east of the lake) that contain a high proportion of 
magnetic minerals. The same pattern of distinct peaks is seen in the magnetic 
susceptibility profile from 249, enabling the identification of a series of time­
synchronous levels in both cores (fig. 5.1). 
CHRONOLOGY. 71 
0 10 
10 
20 
20 
284 249 
30 30 
40 
50 40 
60 
.... 50 
70 
.. ' 
80 60 
.. ' 
...... 
90 
100 
110 
70 
80 
? .............. ' 
~~=--...... /.//: .... / ... / .. 
120 
130 
90 ................. 
140 
§ 150 
-5 
0.. 
4) 160 Cl 
170 
180 
100 
110 
120 
..... : ..... ---- ._---------_ .. _-------
•••• ~ ••••••••••••••.•••• ::::::::::::::::::~: ~:::::~:: :: ::::::::S· .··· ······ ;i~'/~?::> .••• ••••• ••••• ••• ••••••••• •••••••••• ••••••• •••••. 
~ .... ,_ 
---- -----.-._----_ ... _-----------_ ... _------._---------- - 130 
190 .... ,-
200 140 
? 
210 _._ .: ... --_ ... "-- -
220 150 
------ ... _---_ ...... _-_ ... _-- -_ .... _-_ ... -.-. __ ... _--- - - _._- - .. -.. "" 
? 
230 
240 
160 --------------------------------J-----~-- ~;;;---~------------ --- ________________ 1 _________________ ---
250 170 
260 180 
270 
280 190 
o 2 4 6 8 
290 
300 
Magnetic Susceptibility 
310 
o 2 4 
Magnetic Susceptibility 
Fig. 5.1 Proposed correlations between magnetic susceptibility profiles of cores 284 (left) and 249 (right). 
Units for magnetic susceptibility are x107 m3kg- 1. 
CHRONOLOGY. 72 
Using the similarities of the magnetic susceptibility curves as a framework, a 
correlation can now be suggested based on the comparison of the APINAP curve (and 
its associated chronology) from 249 with the carbonate content curve from 284 (fig. 
5.2). As already mentioned, the APINAP curve in this case is considered to be a first­
order proxy for climatic variability (Tzedakis, 1991). For much of core 284, the 
carbonate content is largely indicative of lake productivity (section 7.2.1) and may 
therefore also be considered as a first-order proxy for climatic variability, enabling 
comparison between the two records . 
Such a correlation rests on the implicit assumption that both vegetational change and 
carbonate content respond synchronously to climatic change. In truth, the response 
times of the two systems are likely to be different and a lead/lag situation would exist. 
It must be remembered, however, that the sampling of the two cores is not continuous, 
but instead consists of a series of discrete 'snapshots' at different intervals of time. In 
addition, the time between 'snapshots' in 249 is not the same as between those of 284. 
Pollen from 249 was sampled at -85cm intervals, corresponding to a period of - 2ka 
(average accumulation rate"'" 0.39mka-'). The carbonate content in 284 was 
determined at 10 or 20cm intervals (section 4.2.1) and represents a much shorter 
period of time (approximately 200 or 400 years, using an average accumulation rate "'" 
0.53mka-'). With such disparity between sampling intervals and the inherent 
uncertainty that this generates, it is not possible to determine any lead/lag effect that 
may exist unless much finer sampling is undertaken for both cores. Under the 
circumstances therefore, it seems reasonable to assume that within the sampling 
constraints and over the timescales involved, variations in vegetational change and 
carbonate content are effectively synchronous. A more detailed discussion of leads 
and lags between the response times of different systems is presented in section 7.5.4. 
Analysis of the two records (fig. 5.2) reveals a remarkably good correlation for 
approximately the last 420ka, represented by the recovered part of 249. Beyond this, 
the age model for 284 can be tentatively extended by extrapolation and comparison of 
the carbonate content curve with the marine isotope SPECMAP record (Imbrie et al., 
1984). In conjunction with the timescale established by Prell et al. (1986), this 
suggests that 284 stretches back beyond 620ka, into MIS-16. By transferring the 
chronology of 249 in this way, a working age model for 284 can be established. This 
is able to serve as an initial framework against which other independently derived age 
models can be compared. 
0 
10 
20 
30 
4D 
50 
60 
70 
80 
90 
lOO 
110 
120 
130 
14D 
§ 150 
-5 
0-(1) 160 Cl 
170 
180 
190 
200 
210 
220 
230 
24D 
250 
260 
270 
280 
290 
300 
310 
Carbonate Content (%) 
o 10 
'.4 ....•...... ...•••. 
..... ::.: . E:;'.::~ 
':.',:';.:: ­,., 
~;I.~~~;~S·~::~:-
"'·~:::·;~~,\t~'.'.\'­
;.~~'" 
.. 1r 
20 
.. -. :~u_:::~':;':~:::; ';r~'" ..... ..... . 
r' 
30 
:~~:.:.:."-i:.~~;:~::::~ ..... __ =...:::::==---
....... ,;~; .. ~rN,;.-.-... 
" 0°/ 
{~:.!. 
\ 
i 
i 
.: 
••••••.• ~!' ..... 4;;;::;; .......... . 
;:~:::~~ ~~ ~ ~'-. 
,~:::~: ........ :::h 
.--...... ;;:;;~'::-.--.----.-.-.... -.--- ... -.-- ......... . 
~ 
4D 
::~~.:. ==::::;;;~~ 
·····:-::·:·:········-----..:f······-··· .. 
.:I. ..... 
.... 
2 
3 
4 
5 
.... 
.... 
6 
7 
8 
CHRONOLOGY. 73 
o 25 
-.f:::::-' 
..... 
APINAP % 
50 75 
..................... ~::;?: 
:,m;n~::.; . 
..... : 
. ........  :,. 
.... ~::::::s .. 
100 
..... _. ___ ....... ..... -· .... ~::::::!~!t ............ HH;;::; ••..•• ----.-
::::;::,.. 
<.~' " 
10 
20 
30 
4D 
50 
60 
70 
80 
90 
lOO 
110 
120 
130 
.......... 9 
...  ~,,~t~£,: ..........................................  14D ............................... 
~ 
.... ...... ...... :::::--:.If;;~::\~-' 
.. .r- _~=~;;;::.. ,:;;~::: :::::;::~ 
,.1 
.... ...... ; .• :::;1-{: ...•..... __ ..............•..... __ ....... -. 
,,' 
;. 
~~~;;'" 
150 
160 
170 
180 
-.... ::: ........... . 
"::~ 
............... ::;/~ II 
,. 
}~y~~~~~~~~:::. 
.~' 
<' 
190 
5 o 2 3 4 
Magnetic Susceptibility 
.. .. .. 
" l::::::':~:;::"" 
'::~:. 
-I o 2 3 
Magnetic Susceptibility 
Fig. 5.2 Suggested correlation between the APINAP pollen record from 249 (dashed line, right­
hand graph) and the carbonate content of 284 (dashed line, left-hand graph). Magnetic 
susceptibility (x 10-7 m3kg- 1) for both cores depicted by solid line. Suggested correlations with 
marine oxygen isotope stages are shown. 
§ 
.c 
0.. 
(1) 
Cl 
CHRONOLOGY • 74 
5.3 RADIOCARBON AGE DETERMINATIONS 
Radiocarbon dating is a well established technique that can be employed to date 
organic material that is younger than about 35ka (Evin, 1990). A comprehensive 
summary of the procedure can be found in Olsson (1986) and Pilcher (1991). Levels 
selected for dating from the Ioannina 284 record were identified on the basis of 
significant shifts in the high resolution carbonate record for the top 40m of the core, 
which appeared to match reconstructed temperature curves for approximately the last 
35ka BP (e.g. Atkinson et aI., 1987; Guiot et aI., 1989; de Beaulieu et aI., 1994; Lowe 
et al., 1994; Walker, 1995; Lowe and Walker, 1997). A preliminary interpretation of 
the carbonate curve identified the Last Glacial Maximum (LGM), the Late-glacial 
Interstadial, the Younger Dryas (YD) and the steep rise at the beginning of the 
Holocene. It was later possible to amend the preliminary interpretation once the 
results from the faunal analyses (chapter 6), isotopic analyses (section 7.4.3) and 
pollen analyses (P.e. Tzedakis, unpubl. data) became available. The revised 
interpretation is illustrated in fig. 5.3 . 
The sampling strategy was therefore designed to test this age model. As there were 
no terrestrial plant macrofossils present, the shells and opercula of aquatic molluscs 
were used for dating (specifically, the gastropods Viviparus janinensis and Bithynia 
graeca and the bivalve Dreissena (Carinodreissena) cf. stankovici, all of which are 
described in chapter 6). The small diameter of the core meant that insufficient 
material was recovered for conventional radiocarbon dating. A series of 12 dates was 
therefore obtained from a total of 11 levels using the accelerator mass spectrometry 
(AMS) technique (table 5.1 and fig. 5.3). 
There is a possibility that the dates, although in stratigraphical order, are nevertheless 
slightly too old. Such age anomalies may result from reservoir effects that can affect 
-
karstic lake systems, in particular, so-called 'hard-water error' (e.g. Deevey et ai, 
1954; Shotton, 1972; Karrow et al., 1984). A certain amount of 'old' carbon derived 
from the limestone bedrock can become assimilated into the shells of calcareous 
organisms (such as molluscs), giving rise to an age anomaly which produces a date 
much older than expected. For example, Harkness (1975) demonstrated that a 5% 
level of contamination by old carbon would produce a value approximately 400 yrs 
older than the true age; a 40% level of contamination would increase this error to 
4,100 yrs. Since different species of mollusc assimilate different amounts of old 
carbon, hard-water error can also be species dependent. Roberts (1983) considered 
the likely contamination of Dreissena shells from the Late Quaternary Lake Konya 
(Turkey) from old carbon to be significant, but suggested that even a 20% level of 
CHRONOLOGY • 75 
2 
3 ® 1,480±45/ 
4 1,915±45 
5 
6 
7 <U @ 2,465±45 c: <U 
8 u 0 
® 5,370±50 "0 9 ::t 
10 
11 ® 7,505±60 12 
13 
14 
15 @ 1O,080±70 
16 
17 @ 15,330±140 
18 
19 
§ 20 
21 
..c:: 
0.. 22 <U 
Cl 23 ® 20,760±230 
24 
25 :§ El U ;::l 
26 ..:'! El ® 30,340±460 0·-
27 t) ~ 
28 
j::E 
29 
30 
31 @ 38,600±1,OOO 
32 
33 
34 @ 40,100±1,300 
35 
36 
@ 44,200±2,000 37 
38 
39 
40 
0 10 20 30 40 50 
Carbonate Content (%) 
Fig. 5.3 Carbonate content curve for top 40m of core 284. Position of radiocarbon determinations 
denoted by @ symbol. Preliminary stratigraphical interpretattns also shown. 
CHRONOLOGY • 76 
Depth (m) 14C Age (yrs BP) Material 813C (%0) Lab No. 
3.30 1,480±45 Dreissena shell -3.5 AA 17662 
3.30 1,915±45 Viviparus shell -4.8 AA17661 
7.20 2,465±45 Bithynia opercula -lA AA23530 
9.20 5,370±50 Bithynia opercula -1.6 AA23529 
11.50 7,505±60 Bithynia opercula -2A AA23528 
15.10 10,080±70 Bithynia opercula -3.3 AA23527 
17.17 15,330±140 Viviparus shell -2.3 AA 17660 
23.18 20,760±230 Viviparus shell -4.7 AA 17659 
26.20 30,340±460 Dreissena shell -5.0 AA 17658 
30.15 38,600±1000 Dreissena shell -5 .0 AA 17657 
33.75 40,100±1300 Dreissena shell -5.1 AA 17656 
36.25 44,200±2000 Dreissena shell -5 .3 AA17655 
Table 5.1 AMS radiocarbon measurements from core 284. 813C values given relative to PDB . 
contamination to material with a true age of 20,000 yrs BP would still provide a 
'useful' date of 21,850 yrs BP (p.157). Therefore, although he examined his shell 
material using optical microscopy prior to measurement in order to detect diagenetic 
alteration of the calcite (i.e. contamination by 'young' carbon), he only applied a 
blanket 400 yr correction to his dates to account for reservoir effects. In view of the 
possibility of non-linear variation of hard-water error with age (e.g. Karrow and 
Anderson, 1975; KaITOW et aI., 1984), this would seem unwise. 
Hard-water error remains a serious problem when trying to date sequences from 
karstic areas in which terrestrial plant macrofossils are absent. There are, however, 
several potential ways of assessing the degree of error and applying corrections to 
radiocarbon dates thought to be affected in this way . One method involves the 
detailed modelling of carbonate sources in a water body and the assumption of 
chemical, isotopic and hydrological steady-states over time (e.g. Fontes, et al., 1983; 
Lamb et al., 1995). This was not considered to be a viable option in the case of 
Ioannina, principally because of the lack of sufficiently detailed hydrological 
information on the lake system. 
Despite the possibility that hard-water error may not vary in a linear fashion through 
time (see above), another approach sometimes employed is to use dates derived from 
effectively modern, immediately 'pre-bomb' Holocene specimens to quantify any 
possible error. In the case of Ioannina however, there wel:e no very recent pre-bomb 
shells in the core that would have enabled an estimate of the hard-water error to be 
made. The youngest occurrence of suitable material in the core was at 3.30m. Two 
CHRONOLOGY. 77 
shell samples from this level were analysed, but the age determinations were still too 
old for use in calibration. 
One promising method for circumventing the problems of hard-water error is to date 
pollen concentrates derived primarily from terrestrial sources. It has been 
successfully shown (Brown et aI., 1989; Regnell, 1992) that accurate AMS dates can 
be obtained from pollen grains. It is known that the Ioannina sediments are 
particularly rich in pollen (P. e. Tzedakis, pers comm.) and should provide sufficient 
material for AMS dating, even from stadial episodes when pollen productivity 
declined. The lack of substantial fluvial input into the basin suggests that most of the 
pollen preserved in the core was originally airborne and derived from the surrounding 
catchment area. This means that the chances of contamination by aquatic pollen was 
minimal. NERC support has been obtained to date pollen concentrates from each 
level in the sequence that has already yielded dates based on molluscan material. In 
this way, it is hoped to establish a chronology for the sequence that is independent 
from and which can be directly compared with, the shell-derived chronology. The 
processing of the pollen concentrates is currently proceeding. 
In the meantime, no attempt has been made to account for possible reservoir effects in 
the Ioannina sequence and so the values obtained (quoted as uncalibrated radiocarbon 
years BP) are therefore all likely to represent maximum ages. Nevetheless, the results 
are still meaningful and greatly assist in the interpretation of the sequence. For 
example, molluscan analysis has revealed a period of low lake-level at approximately 
26-27m (section 6.2.3). It is likely that this event corresponds to the time of the 
LGM, at approximately 20,000 yrs BP (this is discussed further in section 7.5). The 
interval 26.5-23.5m is bracketed by two maximum ages of 30,340±460 and 
20,760±230 yrs BP, confirming that the LGM does indeed fall within this section of 
core. 
The beginning of the Holocene (at 10,000 yrs BP) has been interpreted as occurring in 
the sequence at approximately 15m, on the basis of pollen and isotopic evidence 
(section 7.5.3) . The oxygen isotope signal suggests a sudden return to wetter 
conditions, whereas the pollen record indicates an increase in temperature and 
moisture levels by a marked rise in Quercus (P.e. Tzedakis, unpubl. data). The 
radiocarbon date of 1O,080±70 yrs BP at a depth of 15.1Om appears to confirm this 
interpretation. Isotopic evidence has also been used to interpret the possible position 
of the Younger Dryas (-11-10,000 yrs BP). This brief (probably global) stadial 
episode is recorded as a cool, arid period in the Balkans (e.g. Willis, 1994), but it has 
yet to be unequivocably recognised from Greece. The oxygen isotope signal suggests 
CHRONOLOGY • 78 
that the climate became distinctly arid between approximately 15-17m, but the 
sampling resolution was too coarse to enable this oscillation to be defined with any 
greater precision (section 7.4.3). The radiocarbon dates suggest that the Younger 
Dryas should fall between 15.1 and 17.2m, but until they have been corrected for 
hard-water error, it is not possible to calculate a meaningful age for the isotopic 
oscillation. Nevertheless, it would appear from these examples that the radiocarbon 
determinations (particularly the upper ones) have not been subject to excessive hard­
water error and even without correction they are still valuable in helping 
independently to verify the proposed age model. 
Assuming relatively minimal reservoir effects therefore, an age-depth plot based on 
these radiocarbon ages enables calculation of sediment accumulation rates over this 
period (fig. 5.4). Three main phases can be identified: (i) an initial period of high 
accumulation rates (-1. 13mka-'), lasting until approximately 38.6ka BP; (ii) a period 
of much lower accumulation rates (-O.39mka-'), lasting until around 20.8ka BP; and 
(iii) a final period of increased accumulation rates (-O.98mka-'), lasting until the end 
of the interval. The initial period of high accumulation rates coincides with the latter 
stage of MIS-3, which is known from several long European pollen records to have 
included a series of mild climatic phases (for details, see the review by van Andel and 
Tzedakis, 1996). The pollen record from Ioannina (Tzedakis, 1991, 1994) shows this 
to be a period of open Mediterranean woodland, indicative of reasonably moist 
conditions. The central period of low accumulation rate corresponds to a period of 
time that includes the very end of MIS-3 and the glacial MIS-2 (including the LGM). 
The pollen from Ioannina (Tzedakis, 1991, 1994) records steppe and desert-steppe 
conditions (dominated by Artemisia, grasses and chenopods), indicating a cold and 
arid climate. The final return to higher accumulation rates corresponds to a period of 
time that includes the Lateglacial Interstadial and the Holocene, known from pollen 
data to be a time of wetter and warmer climatic conditions. 
Whilst acknowledging the limited number of data points, it is clear that accumulation 
rates were higher during warmer periods, probably linked to the increased 
temperatures and precipitation during these times. In colder periods, arid conditions 
appear to be reflected by much lower accumulation rates. It therefore seems 
reasonable to assume that this situation might also apply to earlier glacial and 
interglacial periods and that similar accumulation rates could be estimated. 
0 
2 
4 
6 
8 
10 
12 
14 
16 
§ 18 
-5 20 
0. 
Cl) 
0 22 
24 
26 
28 
30 
32 
34 
36 
38 
40 
0 5 10 
CHRONOLOGY • 79 
Gradient = 0.98 
Gradient = 0.39 
Gradient = I .13 
15 20 25 30 
Age (kyr) 
35 40 
, 
, 
, 
45 50 
Fig. 5.4 Age-depth plot for 284 using uncalibrated AMS radiocarbon dates. See text for 
interpretation of variable gradients. 
CHRONOLOGY • 80 
S.4 MAGNETOSTRATIGRAPHY 
As outlined in section 3.2.3, palaeomagnetic analysis uses the fact that the Earth's 
magnetic field has undergone polarity reversals several times in the past. The last 
major reversal was at -780ka (Shackleton et al., 1990; Baksi et al., 1992), when the 
geomagnetic field flipped from being 'reversed' (the Matuyama chron) to 'normal' (the 
Brunhes chron). Core 284 does not appear to extend back to the Brunhes-Matuyama 
boundary (see below), but during the Brunhes there have been several magnetic 
events of relatively short duration, during which the geomagnetic field achieved 
almost full reversal (table 5.2). These brief events (generally lasting less than lOka) 
are thought to have been abortive attempts to reverse the geomagnetic field. They 
have been dated from lava flows, marine cores and other sequences using a variety of 
techniques (e.g. Ryan, 1972; Champion et al., 1988; Levi et al., 1990; Nowaczyk et 
al. , 1994; Glen and Coe, 1997), so enabling a reasonably consistent global 
magnetostratigraphy to be established. 
Name 
Gothenburg 
Mono Lake 
Laschamps 
Norwegian/Greenland Sea 
Fram Strait 
Blake 
Jamaica 
Pringle Falls 
Levantine 
Biwa III 
Emperor 
Big Lost 
Original Reference 
Mbrner et al ., 1971 
Denham and Cox, 1971 
Bonhommet and Babkine, 1966 
Bleil and Gard, 1989 
Thouveny et al. , 1990 
Smith and Foster, 1969 
WoIlin et al. , 1971 
Herrero-Bervera et al. , 1989 
WoIlin et al. , 1971 
WoIlin et aI., 1971 
Ryan, 1972 
Champion et al., 1988 
Age/ka 
(published) 
-12.4 
23-32 
30-43 
72-86 
98-102 
110-130 
170-200 
218 
280-310 
360-417 
460-490 
<610 
Age/ka 
(this study) 
35.3 
116-129.5 
189 
211-224 
291-301 
596 
Table 5.2 Comparison of published ages for palaeomagnetic polarity events with those derived from 
this study. Published ages are given as a range where appropriate (see text for references) . Ages for 
this study only given where formal correlations are suggested (see text for details). 
There still exists some disagreement as to the exact number, timing and duration of 
these events, largely depending on where and how they were measured (e.g. Kukla 
and Zijderveld, 1977; Verosub, 1982; Lpvlie, 1989). In addition, the terminology 
used to describe geomagnetic events has also become confused. 'Sub-chron', 
'excursion' and 'polarity event' are terms that are frequently used interchangeably. 
Throughout this study therefore, the nomenclature of Jacobs (1994) has been adopted. 
CHRONOLOGY. 81 
Short periods of full reversal are referred to as sub-chrons; the term 'excursion' is 
used to describe brief periods of shallow or reversed inclination (such as the Emperor 
or Mono Lake excursions); the term 'event' is used to describe a longer period of 
reversed inclination that could represent either a fully reversed sub-chron or an 
excursion (such as the Blake event). 
Matters are further complicated since many studies have used different values for the 
date of the Brunhes - Matuyama boundary, which has been used as a fixed point in 
time from which to interpolate the age of polarity episodes. This date, currently 
placed at -780ka (Shackleton et al., 1990; Baksi et al., 1992), has been subject to 
frequent adjustment backwards (from around 700ka), meaning that many of the older 
studies give ages for polarity episodes that are now considered as slightly too young. 
In addition, several excursions (e.g. Gothenburg, Biwa III and Emperor) are regarded 
as being of relatively short duration (discussed below). These could have been 
missed in sequences where sampling intervals were too coarse, or in the case of 
sedimentary sequences, where accumulation rates were too low. Finally, a consistent 
protocol for naming the shorter magnetic reversals has never been established, with 
the result that most can be referred to by several names. Since it is based on priority 
of usage, the nomenclature of Champion et al. (1988) has been adopted throughout 
this study. 
Despite these difficulties, broad worldwide correlations for several main reversals are 
now accepted. By comparing the geomagnetic polarity episodes in continuous 
sequences, such as Ioannina, with the 'master' magnetostratigraphy, an independent 
assessment of proposed age models may be made. 
Palaeomagnetic measurements were made using the procedure set out in section 3.2.3. 
Typical Zijderveld orthogonal plots are presented in fig. 5.5 which show the 
satisfactory behaviour of the samples under thermal demagnetisation. A brief 
analysis of the magnetic vectors suggests that at least one of the overprinted 
secondary magnetic signals has a substantial vertical component. This is interpreted 
as a VRM induced by the coring process. The full data set (consisting of thermally 
demagnetised inclination values) has been plotted against depth (fig. 5.6, left-hand 
graph). 
The relationship between inclination (l) and latitude (cp) is given by: 
tan (l) = 2 tan (cp) 
F1 
N,up 
W, E, 
S,down 
F7 
F13 
N, 
scale: 0.200 mAlIn 
0 
0 
X 
o to 300 DC) 
horizontal 
vertical 
NRM 
E,up 
scale: 0.400 mAIm 
010 300 DC) 
o horizontal 
o vertical 
X NRM 
F4 
W, 
F12 
CHRONOLOGY • 82 
N,up 
E, 
S,down 
E,up 
Icale: 0.500 mAl.n 
o to 300 QC) 
(I horizontal 
0 vertical 
X NRM 
Eale: 0.050 mAIm 
010 300 DC) 
o horizontal 
o vertical 
N.,r --'---r--.r--.--.---~~S, X NRM 
W,down 
W,down 
F20 
N,up 
E,up scale: 0.500 mAIm 
scale: 0.070 mAIm o to 300 DC) 
010 300 DC) 
0 horlz:iOntal 
0 horizontal 0 vertical 
0 vertical 
X NRM 
S, X NRM 
W, E, 
S,down 
W,down 
Fig. 5.5 Typical Zijderfeld demagnetisation plots across the polarity episode interpreted as the Blake 
event (see text). Thermal demagnetisation steps on plots represent NRM, 200·C, 250·C and 300°C. 
Sample numbers correspond to the following depths: Fl - 81.34m; F4 - 84.74m; F7 - 87 .26m; F1 2 -
92.28m; F13 - 93.23m; F20 - lOO.35m. 
CHRONOLOGY. 83 
0 0 
10 10 
20 20 
30 30 
40 40 
50 50 
60 60 
70 70 
80 80 
90 90 
100 100 
110 110 
120 120 
130 130 
140 140 g 
150 ISO 
g 
.s ~ 0.. 
Q) 160 160 Q) Cl Cl 
170 170 
180 180 
190 190 
200 200 
210 210 
220 220 
230 230 
240 240 
250 250 
260 260 
270 270 
280 280 
290 290 
300 300 
310 310 
-80 -60 -40 -20 0 20 40 60 80 -90 -60 -30 0 30 60 90 
Inclination Inclination 
Fig. 5.6 Inclination data for 284. Left-hand graph shows raw data, right-hand graph shows smoothed 
data (see text for details) . 
CHRONOLOGY • 83 
0 0 
10 10 
20 20 
30 30 
40 40 
50 50 
60 60 
70 70 
80 80 
90 90 
lOO lOO 
110 110 
120 120 
130 130 
140 140 
§ ISO ISO § 
.s .s 
0. 0. Q) 160 160 Q) Cl Cl 
170 170 
180 180 
190 190 
200 200 
210 210 
220 220 
230 230 
240 240 
250 250 
260 260 
270 270 
280 280 
290 290 
300 300 
310 310 
-80 -60 -40 -20 0 20 40 60 80 -90 -60 -30 0 30 60 90 
Inclination Inclination 
Fig. 5.6 Inclination data for 284. Left-hand graph shows raw data, right-hand graph shows smoothed 
data (see text for details). 
CHRONOLOGY • 84 
Ioannina is at a latitude of approximately 39 0 45', so the expected inclination value 
should therefore be around 59 0 • After removing the reversed data points from the 
calculation, the actual average inclination value obtained after blanket 
demagnetisation was approximately 56 0 , a value regarded as satisfactory. Further 
detailed demagnetisation of each individual sample may have increased the average 
inclination to be more in accordance with the predicted value. The raw demagnetised 
data set is rather noisy, with several apparently reversed data points. Some of these 
may represent genuine polarity episodes, although it is possible that some are the 
result of poor handling practice during recovery and storage of the core (which was 
not originally taken specifically for palaeomagnetic analysis). With this in mind, 
single reversed data points not supported by shallow inclination values either side 
have been rejected. Palaeomagnetic data derived from a single core box were also 
rejected since it had been dropped during transit, making the orientation of some of 
the core sections from that box somewhat suspect. The revised data set is plotted in 
fig. 5.6 (right-hand graph). This curve is used in all subsequent discussion. 
Although several of the data points have either shallow or negative inclinations, the 
fact that the majority are normally polarised would suggest that the entire sequence 
falls within the Brunhes normal polarity chron and is therefore younger than -780ka. 
This would be in accordance with the proposed age model based on the pollen­
derived chronology from 249. Six periods of shallow or reversed inclinations can be 
clearly recognised at depths of approximately 40m, 95m, 140m, 150m, 185m and 
297m. In addition, several single points also display shallow inclinations: a cluster 
between the top of the sequence and the fully reversed event at 40m; two or three 
points between this event and that at 95m; two points below the event at 140m; one 
point at approximately 212m; and one point at approximately 233m. 
The magnetic polarity reversals listed in table 5.2 are now considered briefly in turn. 
This is to assess any likely correlation between documented excursions or sub-chrons 
and the Ioannina record, rather than to attempt an exhaustive review (see e.g. 
Champion et al., 1988; Jacobs, 1994). The absence of declination measurements 
means that virtual geomagnetic pole (V GP) positions cannot be calculated. This 
would have enabled a better attempt at linking the periods of reversed inclinations 
with known excursions or sub-chrons. Nevertheless, it has still been possible to 
suggest likely correlations based on stratigraphical position and structure (summarised 
in fig. 5.7). 
CHRONOLOGY • 85 
Carbonate Content (%) Inclination (degrees) 
0 10 20 30 40 
-90-60-30 0 30 60 90 
0 0 
10 10 
........................... 12 
20 2 20 
30 30 
40 3 Laschamps 40 
50 50 
60 60 
70 70 
80 5 80 
90 Blake 90 
.............. .. ... . ............................. 128 
100 100 
110 110 
120 6 120 
130 130 
140 186 Jamaica 140 
150 7 150 § Pringle Falls § 
..c:: 160 245 160 ..c:: 
a 0. (1) 170 8 170 (1) Cl Cl 
180 180 Levantine 
190 190 
9 
200 200 
210 210 
220 220 
230 II 230 
240 240 
12 
250 478 250 
260 13 260 
270 
................. .. ............................ 524 270 
280 14 280 ............................................... 565 
290 290 
15 Big Lost 300 300 
310 ......•..............•...............•..... 620 16 310 
Fig. 5.7 Suggested magnetostratigraphy for 284. Centre column shows nonnal (black) and reversed 
(white) polarity episodes. Left-hand graph shows smoothed carbonate content (9-pt. running mean) 
with suggested marine oxygen isotope stages (ages in ka, after Prell et al., 1986). Position of MIS-9/8 
boundary after Tzedakis et al. (1997) (see section 8.2). Right-hand graph shows inclinations (in 
degrees) . 
CHRONOLOGY. 86 
5.4.1 Gothenburg excursion 
The Gothenburg excursion was first described by Morner et al. (1971) from the 
Botanical Gardens in Gothenburg, Sweden. Subsequently termed the Gothenburg 
Magnetic Flip (Morner and Lanser, 1974), it is thought to have been a particularly 
rapid excursion of less than 50 years duration. Radiocarbon dating and varve 
counting in a series of six Swedish cores put the age of the event at 12.4-12.35ka BP 
(Morner, 1977). However, its existence is still widely disputed, largely because of the 
difficulty in finding it elsewhere. Moreover, Verosub (1982) points out that such a 
short duration is in conflict with the current understanding of the fluid core dynamics 
required to produce a reversal. The Ioannina record has shallow inclinations at 
approximately ISm, one of which might possibly be correlated with the Gothenburg 
Flip, or else may simply represent a noisy signal. Such uncertainty means that no 
definite correlation can be proposed. 
5.4.2 Mono Lake excursion 
Described originally from Californian lake sediments (Den ham and Cox, 1971), the 
Mono Lake excursion was dated using radiocarbon techniques (on unspecified 
material derived from two ostracod-rich horizons) to 24.6-24ka BP. Ten further 14C 
dates (again, on unspecified material) later refined the temporal control for the 
sequence, allowing the age of the excursion to be revised to 29-32ka (Lajoie et al., 
1980). Although the existence of the Mono Lake excursion remains controversial 
(e.g. Verosub et al., 1980; Jacobs, 1994), polarity episodes deemed to be 
manifestations of the Mono Lake have been found in lake sediments from Nevada 
(Liddicoat et al., 1982) and Oregon (Negrini et al., 1984), as well as in marine 
sequences from the Arctic (Nowaczyk and Baumann, 1992; Nowaczyk et al., 1994). 
These studies have merely served to blur the dating of this event, with authors placing 
it anywhere between 32-23ka BP. It is also possible that an excursion recognised 
from prehistoric aboriginal fireplaces at Lake Mungo in south-east Australia (Barbetti 
and McElhinny, 1972, 1976) and dated at -30ka BP by radiocarbon techniques (on 
hearth charcoal) is the same excursion. It should be pointed out, however, that the 
Lake Mungo excursion has also been attributed to lightning strikes (Jacobs, 1994). 
Whilst no direct correlation is suggested with the Ioannina sequence, several points of 
low inclination are present above the fully reversed inclinations at approximately 
40m. It is possible that one of these points corresponds to part of the Mono Lake (the 
prime candidate being at approximately 27m), the coarse lm sampling resolution only 
CHRONOLOGY. 87 
having caught part of the excursion. Further detailed palaeomagnetic analysis across 
this interval may resolve this problem. 
5.4.3 Laschamps excursion 
Originally described from the Laschamps and Olby lava flows from the Chaine des 
Puys in the Massif Central of France (Bonhommet and Babkine, 1966), excursions 
deemed to be the Laschamps have subsequently been recorded from many locations 
worldwide, including the Gulf of California (Levi and Karlin, 1989) and the Arctic 
(e.g. Nowaczyk and Baumann, 1992; Nowaczyk et al., 1994). Most of these studies 
provide dates of around 30-40ka, although Levi et al. (1990), who used radiometric 
K-Ar techniques on Icelandic lavas, obtained a date of 42.9 ± 7.8ka. In the Ioannina 
sequence, a marked excursion occurs at approximately 40m. This corresponds to an 
interpolated age of 35.3ka based on the pollen-derived stratigraphy inferred from core 
249. It is therefore proposed that this excursion should be correlated with the 
Laschamps excursion. 
5.4.4 Norwegian/Greenland Sea and Fram Strait excursions 
These are two excursions that have only been studied relatively recently, as higher 
resolution investigations have been conducted. As a result, their existence as world­
wide polarity events is still somewhat controversial. The Norwegian/Greenland Sea 
excursion was originally described by Bleil and Gard (1989) and has subsequently 
been confirmed by other investigations both in the Arctic Ocean (e.g. Nowaczyk and 
Baumann, 1992) and elsewhere, e.g. south-western California (Glen and Coe, 1997). 
Possibly the best estimate of its age comes from a study by Nowaczyk et al. (1994), 
who used a combination of foraminiferal biostratigraphy, oxygen isotope stratigraphy, 
AMS radiocarbon dates and lOBe and 230Th stratigraphies on two sediment cores from 
the Arctic Ocean to derive an age of 72-86ka. 
The Fram Strait excursion was originally described by Thouveny et al. (1990) from 
the Lac du Bouchet sequence in France, but not formally named. As part of a study 
from the Fram Strait (due west of Svalbard), Nowaczyk and Baumann (1992) dated a 
magnetic excursion using oxygen isotope stratigraphy to 98-102ka. It has 
subsequently been recognised by Glen and Coe (1997) from south-eastern California, 
although they do not attempt to re-calculate its age. 
CHRONOLOGY. 88 
The brief duration of these excursions means that it is entirely possible that the 
sampling interval used on core 284 may have missed these excursions entirely. 
However, in the Ioannina record, two points of shallow inclination can be seen at 
74.45m and 82.9m. Using the pollen-derived stratigraphy derived from core 249, 
these points can be dated at 88.1ka and 91.5ka respectively. It is possible that the 
upper point correlates with the Norwegian/Greenland Sea excursion, the lower point 
with the Fram Strait excursion. However, it should be stressed that these shallow 
inclinations are not sustained over a series of samples and more detailed analysis is 
required before their validity as genuine excursions can be established. Therefore, no 
formal correlation is proposed at this stage. 
5.4.5 Blake event 
Possibly the best known (and most widely accepted) of all of the Brunhes polarity 
episodes, the Blake is nevertheless not without its problems. First described from a 
series of Caribbean and Indian Ocean cores by Smith and Foster (1969), it has 
subsequently been reported from a variety of sequences worldwide (see Nowaczyk et 
al., 1994), dated anywhere between between 110-130ka. In addition, there is 
evidence suggesting that the Blake may have a complex structure, being a composite 
of more than one reversed event. For example, Manabe (1977) records a marked 
polarity event in a marine terrace sequence from eastern Japan. Consisting of two 
reversed periods separated by a normal interval, the event is correlated with the Blake 
on the basis of biostratigraphical evidence. Two Mediterranean marine cores dated by 
oxygen isotope stratigraphy and tephrochronology (Tric et at., 1991) also record the 
Blake as consisting of two reverse periods separated by a normal interval. Creel' et at. 
(1980) claim that this can also be seen in a long (250m) sequence of marine clays 
from Gioia Tauro in southern Italy. Dated using forarniniferal biostratigraphy, one of 
several polarity episodes present in the core consists of two reversed periods 
separated by a normal interval. This they correlated with the Blake. Rampino (1981) 
points out that the Gioio Tauro sequence may actually be interrupted by a hiatus, in 
which case the entire polarity episode may not be present. 
In studying two marine cores from the Greater Antilles Outer Ridge, Denham (1976) 
noted that the Blake appeared to have a tripartite structure, consisting of three 
reversed periods separated by two normal intervals. Zhu et al. (1994) have studied 
the Blake event in a high resolution loess record from Xining in western China. 
Using previously established loess stratigraphy and thermoluminescence data to 
constrain the event chronologically, they also note that the Blake has a tripartite 
CHRONOLOGY. 89 
structure. More recently, Glen and Coe (1997) recalculated the age of the Blake event 
cited by Champion et al. (1988), using a revised value of 783ka (Baksi et al., 1992) 
for the age of the Brunhes - Matuyama boundary. They assigned their revised age of 
122ka to a complex zone of shallow inclinations in a long (323m) composite 
sedimentary sequence from Owen Lake in California. 
Interestingly, in the study mentioned above of two cores taken from the Greater 
Antilles Outer Ridge, Denham (1976) correlates his Blake event with most of the 'X' 
marine faunal zone of Ericson et al. (1961). This was taken to represent the whole of 
MIS-5, currently dated at 128-71ka (Prell et al., 1986). Because of high levels of 
carbonate dissolution, the microfossil stratigraphy of the cores was not clear 
(Denham, 1976). Later radiometric work using uranium and polonium techniques 
confirmed the original assignment of the polarity episode to the X faunal zone 
(Denham et al., 1977). It is possible therefore that these marine sequences recorded 
not only the Blake event, but also younger polarity episodes (such as the 
Norwegian/Greenland Sea and Fram Strait excursions). 
In the Ioannina record, a complex magnetic excursion occurs between approximately 
90-100m consisting of three reversed periods separated by two normal intervals. 
Using the marine isotope chronology (Prell et al., 1986) together with with the pollen­
derived stratigraphy inferred from core 249, this excursion can be dated at 129.5-
115.9ka. This represents a duration of approximately 13.6ka, in accordance with the 
-lOka suggested by most workers. It is therefore suggested that this polarity episode 
in 284 is correlated with the Blake event. 
5.4.6 Jamaica event 
First recognised (but not named) by Wollin et al. (1971) from a deep-sea NW Pacific 
core, this event was formally described by Ryan (1972) from a series of cores taken 
from the Caribbean and eastern Mediterranean. The Jamaica has subsequently been 
correlated with the Biwa I polarity episode from Lake Biwa in Japan (Kawai et aI., 
1972; Kawai, 1984), as well as other polarity episodes from elsewhere, e.g. Alaska 
(Westgate, 1988); the Norwegian-Greenland Sea (Bleil and Gard, 1989); California 
(Liddicoat, 1990); the Fram Strait (Nowaczyk and Baumann, 1992); and south-eastern 
California (Glen and Coe, 1997). Dates for this event vary from approximately 170-
200ka. The Ioannina record shows a marked polarity event at approximately 140m. 
Using the age model derived from 249 (as previously described), this event has an age 
of 189.3ka. It is proposed that this be correlated with the Jamaica event. 
~I 
CHRONOLOGY • 90 
5.4.7 Pringle Falls excursion 
This excursion was first described by Herrero-Bervera et al. (1989) from a 20m long 
core of lacustrine sediments from Pringle Falls near La Pine in central Oregon, USA. 
Although their sequence was poorly dated, the double structure of the excursion led 
them to tentatively suggest correlation with the Blake event. However, subsequent 
work on the chronology of the sequence using 40 ArP9 Ar dating techniques on a tephra 
layer (Herrero-Bervera et al., 1994) provided an age of 218±lOka, too old for it to be 
either the Blake or the Jamaica. Although the Pringle Falls excursion (as it has 
subsequently been called) has been correlated with other independently dated polarity 
episodes from Summer Lake, Oregon (Negrini et al., 1984; Negrini et al., 1994) and 
California (Liddicoat, 1990; Glen and Coe, 1997), there are few records of it from 
elsewhere. Its temporal proximity to the Jamaica may have meant that researchers 
analysing sequences at coarser resolutions may have been unable to distinguish 
between the two excursions. 
At Ioannina, a series of shallow inclinations can be seen at -150-155m. Using the 
pollen-derived chronology from 249, this interval can be dated to approximately 211-
224ka, suggesting that it should be correlated with the Pringle Falls excursion. 
5.4.8 Levantine event 
This event was again originally recognised by Wollin et al. (1971), although it was 
first formally named by Ryan (1972) from his Mediterranean cores. The Levantine is 
usually correlated with the Biwa II event (Kawai et al., 1972; Kawai, 1984), as well 
as several others worldwide, e.g. the Mono Lake basin in California (Liddicoat et al., 
1980); Gioia Tauro in southern Italy (Creer et al., 1980; Rampino, 1981); the 
Norwegian-Greenland Sea (Bleil and Gard, 1989); and the Fram Strait (Nowaczyk 
and Baumann, 1992). Most dates range from approximately 280-310ka. The 
Ioannina sequence records a marked polarity episode between 180-187m. Using the 
proposed age model, this event can be dated at 293 .6-301.3ka. It is proposed that this 
be correlated with the Levantine event. 
5.4.9 Biwa III excursion 
This is another polarity episode first recognised from the NW Pacific by Wollin et al. 
(1971) but not formally named. Correlated with a short interval of reversed 
CHRONOLOGY. 91 
inclinations from Lake Biwa in Japan (Kawai, 1972), this brief excursion has 
subsequently become known as the Biwa Ill. Fission track ages on zircons from ash 
layers in the Biwa sections (Kawai et al., 1984) have revised the age estimate for this 
excursion to be around 380ka. Creer et al. (1980) calculate its age to be 400ka from 
their Gioia Tauro marine sequence based on microfossil stratigraphy, whereas Negrini 
et al. (1988) estimate its age as just younger than an ash deposit at Summer Lake, 
Oregon, which is correlated with an ash dated at 360-370ka. By using an updated 
value of 783ka for the Brunhes-Matuyama boundary (Baski et al. , 1992), Glen and 
Coe (1997) recalculated the age initially suggested by Champion et al. (1988) to be 
417±9ka. 
There is no clear polarity episode at loannina with which this excursion could be 
correlated. A single shallow inclination at approximately 212m could represent part 
of the Biwa Ill, although its calculated age (using the pollen-derived age model) 
would be approximately 338.2ka (slightly too young). Another shallow. inclination at 
around 232m might also represent part of this excursion. Its calculated age is 
approximately 414.6ka, more in accordance with the published dates for this 
excursion. It is also possible that the coarse sampling resolution used at loannina has 
'missed' it entirely. Further high resolution analysis may resolve this problem. 
An interesting point to make at this stage is that if the position of MIS-lO in 284 is 
shifted upwards slightly, then the shallow inclination value at approximately 232m 
falls exactly where the Biwa III excursion might be expected. The current position of 
MIS-lO is denoted by a negative oscillation in the carbonate curve at around 215m, 
matching a similar oscillation in the 249 APINAP curve. At around 208m, however, 
there exists a more extreme negative oscillation that is more in accordance with the 
structure of the SPECMAP curve for this period (Imbrie et al., 1984). If this were 
actually the case, then the shallow inclinations at 212m would correspond to an age of 
371.7ka and those at 232m would be 417.6ka. This second value, in particular, is 
much more in accordance with the probable age of the Biwa III excursion 
(particularly that of 417ka obtained by Glen and Coe [1997]), making it a prime 
candidate for further detailed palaeomagnetic analysis. However, since the 
interpretation of either shallow inclination value remains uncertain and in the interests 
of chronological consistency, the pollen-derived stratigraphy for this part of the 
record is retained until better evidence becomes available. 
CHRONOLOGY. 92 
5.4.10 Emperor excursion 
First described from one of the Caribbean cores of Ryan (1972), the Emperor 
excursion has since been identified from several other records, including sequences 
from the East Pacific (Wilson and Hey, 1981), the Norwegian-Greenland Sea (Bleil 
and Gard, 1989) and south-eastern California (Glen and Coe, 1997). Age estimates 
range from 460-490ka. No obvious candidate exists in the Ioannina record. It is 
again possible that the brief nature of this excursion means that it was missed by the 
relatively coarse (1m) sampling interval employed. Further high-resolution analysis 
might reveal this excursion at Ioannina. 
5.4.11 Big Lost excursion 
This was originally described by Champion et al. (1988) from lava flows in the Snake 
River Plain, Idaho, which they dated (using K-Ar techniques) at 565±14ka. A 
correlative of this excursion is thought to be that described by Negrini et al. (1987) 
from sediments from the Humboldt River Canyon in Nevada, which they estimated at 
slightly younger than 595ka (the age derived for the underlying Dibekulewe tephra) . 
More recent work on this tephra (Sarna-Wojcicki et al., 1991) has revised this date to 
61Oka. A polarity episode occurs in the Ioannina record at around 297m. Its age 
(using the pollen-derived age model) is approximately 595.9ka. It is therefore 
proposed that this be correlated with the Big Lost excursion. 
One additional palaeomagnetic parameter worth consideration is that of 
palaeointensity. The intensity of the geomagnetic field is known to fluctuate over 
time (e.g. Jacobs [1994] for detailed discussion) . In particular, a study by Valet and 
Meynadier (1993) confirmed that the geomagnetic field intensity relaxes gradually (in 
an asymmetric saw-tooth fashion) over periods of approximately O.5Ma, recovering 
relatively rapidly (over only a few thousand years) after a full polarity sub-chron. 
There are no definite polarity sub-chrons within the Brunhes; there are, however, the 
series of excursions and events described above. Valet and Meynadier (1993) note 
that the Brunhes palaeointensity record is more complex than for previous magnetic 
chrons, possibly as a result of insufficient compaction of the upper parts of their 
Pacific cores. Nevertheless, a series of sharp intensity drops are present within the 
Brunhes, many of which correspond well with documented polarity excursions. In 
their cores from the western Indian Ocean stretching back 140ka, Meynadier et al. 
(1992) note low palaeointensity values at approximately 115ka and 40ka, which they 
CHRONOLOGY. 93 
recogmse as coincident with the Blake and Laschamps excursions respectively. 
Interestingly, they give little attention to a period of similar low palaeointensity values 
at approximately 65ka, that is also present in the curves discussed by Valet and 
Meynadier (1993). 
It is not currently possible to critically examine the palaeointensity data from the 
loannina record, since to obtain comparable results the intensity values for each level 
must first be normalised (usually against anhysteretic remanent magnetisation [ARM] 
measurements). Due to machine time constraints, ARMs have not been measured for 
the loannina samples. Nevertheless, investigating the possibility that palaeointensity 
variations could help confirm the position of the magnetic polarity episodes seen in 
core 284 (or predict where to look for 'missing' episodes) remains an exciting avenue 
for future work. 
In summary, the magnetostratigraphy (fig. 5.7) proposed for the loannina sequence 
appears to support the pollen-derived age model extremely well. In addition, it also 
helps to confirm the extrapolated chronology below 423ka (the extent of the pollen 
record from core 249). Further careful detailed palaeomagnetic investigations may 
well reveal the 'missing' Emperor excursion and confirm the presence of others (such 
as the Mono Lake or Biwa Ill) that are only hinted at in the current record from 284. 
5.5 AMINO ACID EPIMERIZATION DATA 
Analysis of amino acid ratios from mollusc shells, or from other carbonate fossils, is a 
somewhat controversial relative dating technique (e.g. Goodfriend, 1987). An outline 
description of the method is given by Bowen et al. (1989) and Sykes (1991). It 
essentially relies on the relationship between two different amino acid configurations 
(the building-blocks of proteins) after death of the organism. Only L-isomers are 
present in living organisms, but on death, these decay (racemize) at a steady rate into 
D-isomers, until an equilibrium state is reached. By measuring the ratios of DIL 
amino acids in a sequence of fossils, the rate of racemization can be determined and a 
relative age for each fossil calculated. When anchored to an independent chronology 
based, for example, on radiocarbon or radiometric dating, a series of amino acid ratios 
can be used to establish an aminostratigraphy for the sequence (e.g. Bowen and 
Sykes, 1988; Bowen et al., 1989). 
CHRONOLOGY • 94 
When using fossil mollusc shells, the rate of epimerization is calculated instead (e.g. 
Goodfriend, 1987). This is a similar process to racemization, but involves a slightly 
different molecular reaction that is somewhat easier to measure. The 
alloisoleucinelisoleucine epimer ratio (All) is calculated in this case. It is important 
to note that the rate of epimerization is not only dependent on age, but also on species 
and temperature. Different taxa epimerize at different rates, so monospecific samples 
should be used throughout. Temperature changes can also seriously affect the rate of 
epimerization. An area that has undergone a variable thermal history, or samples 
which have been heated during preparation, will produce unreliable results. Ratios 
from samples obtained from a single thermally stable site are therefore preferable. 
At Ioannina, the intention was to obtain a series of amino acid ratios that could be 
calibrated against an independent time-scale (for example, the magnetostratigraphy, 
the radiocarbon chronology or the pollen-derived chronology inferred from 249). A 
pilot study involving six samples of the most abundant large mollusc (the bivalve 
Dreissena) were taken from various horizons in the core down to approximately 
140m. The samples were air dried at room temperature and received no chemical 
preparation treatment (e.g. hydrogen peroxide during sediment disaggregation). Four 
of these levels were chosen because they coincided with those from which AMS 
radiocarbon dates were also being obtained (see section 5.3). Ultimately, one of these 
samples (no. 284/6, from a depth of 9.73m) turned out to be too small to measure. 
Two other samples were also measured. One consisted of modern specimens of 
Dreissena collected as a control from the lake shore in 1994. The other consisted of 
Dreissena shells obtained from the original Kastritsa cave beach sediments (courtesy 
of G.N. Bailey) (see section 7.3.2). These sediments are associated with radiocarbon 
dates (derived from charcoal) of 21,800±470 yrs BP and 20,200±480 yrs BP (Higgs et 
al., 1967; Bailey et al., 1983a). The amino acid ratios were measured by Dr G.A. 
Goodfriend (University of Washington) using the technique outlined in Goodfriend 
(1987). Overall analytical error was -3%. The results are presented below in table 
5.3, along with the estimated age based on the pollen-derived chronology inferred 
from 249 (section 5.2.2). 
The ratios obtained proved to be stratigraphically inconsistent, probably because of 
poor preservation of the amino acids within the mollusc shells. Measurements were 
taken on a single piece of shell from each level (the same part of the shell), with the 
exception of 28417 (the modern sample), where several shells were bulked together. 
It should be emphasised that this was a pilot investigation; it is possible that amino 
acid preservation in different parts of the shell could give more consistent results. It 
is also possible that the thermal history of the site was unsuitable for 
CHRONOLOGY. 95 
Sample Depth (m) Expected Age All ratio Lab No. 
28417 Modern 0.0088 CB-771 
284/5 26.22 -20ka 0.212 CB-770 
284/4 36.27 -30ka 0.193 CB-769 
284/3 42.83 -40ka 0.196 CB-750 
28412 64.75 -75ka 0.183 CB-763 
28411 140.32 - 200ka 0.217 CB-751 
Kastritsa -21ka 0.245 CB-773 
Table 5.3 Amino acid ratio results. 
preservation, although there are no data available on this. The possibility of 
reworking was also considered, though this was thought to be unlikely in this context. 
The material from the Kastritsa cave may also have been subject to a different thermal 
history, or been subaerially exposed for longer. Again, there are no data available. 
On the basis of this pilot study, therefore, and until further work is carried out, it is 
impossible to establish a coherent aminostratigraphy for the core. 
5.6 URANIUM-SERIES DATING 
Uranium-series (U-series) dating is a technique that relies on knowledge of the 
complex decay series of the radioactive isotopes 238 U, 235U and 232Th to stable lead 
isotopes. A detailed discussion of the technique is beyond the scope of this study, but 
a summary is provided by Smart (1991) and Lowe and Walker (1997) . 
Since calcite contains tiny amounts of uranium, the carbonates in lake sediments can 
be dated using the U-series technique, provided that certain conditions are met. The 
most important is that the carbonate must have been precipitated directly from the 
lake water. By virtue of the fractionation process, primary carbonate will contain no 
initial thorium; any contained within the sample will therefore be a result of 
subsequent radiocative decay of the uranium. In reality, since the sediment sample is 
not entirely carbonate in composition, the non-carbonate fraction of the sample 
CHRONOLOGY. 96 
(which will contain thorium and uranium in unknown ratios) will introduce some 
contamination. However, if the non-carbonate and carbonate fractions are both 
assumed to be homogeneous, and the preparation method is assumed to impose equal 
differential fractionation onto all samples, then any contamination can be corrected 
for by using the isochron technique (Schwarcz and Latham, 1989). If these 
assumptions are true, then the gradient of a straight line (an isochron) fitted through 
the data defines the ratio of the isotopes in the carbonate fraction. For example, the 
gradient of a plot of 23D-rh/232Th against 234U/232Th defines the 230Th/234U ratio of the 
carbonate, essential for calculating the age of the sample. 
A pilot study was instigated by T.C. Atkinson, P.E. Rowe (both University of East 
Anglia) and R.c. Preece (University of Cambridge) to attempt to date the 'dirty' 
carbonates of 284 using U-series isochron techniques. The carbonate composition of 
284 was considerably lower than in previous studies, so a well defined interval of the 
core was selected against which the technique could be tested. The last interglacial 
was chosen, since its position in 284 had been clearly identified by a range of 
techniques, including carbonate stratigraphy, magnetostratigraphy (described above) 
and pollen stratigraphy (P.c. Tzedakis, unpuhl. data) . SEM analysis confirmed that 
the calcite crystals were primary precipitates and not detrital in origin (see above) . 
Sediment samples from the beginning and end of the interglacial were analysed (98.8-
97.8m and 85.8-83 .6m respectively) . The detailed methods employed (involving 
partial leaching of the samples followed by alpha spectrometry to determine U and Th 
levels) and the results obtained are currently unpublished, but a comprehensive 
discussion of the study is in preparation. The main results are briefly summarised 
below. 
The age of the upper boundary of the Eemian was calculated at 114.5±4ka, in 
excellent agreement with the SPECMAP age of 116ka (Martins on et al., 1987). The 
calculation of the lower boundary of the Eemian was less successful, with ages 
calculated at 115-13 8ka, depending on the leachates selected, with the optimal choice 
of age being -132ka. The upper boundary results fell tightly on a straight line, 
defining an extremely clear isochron, but this was not the case with the lower 
boundary, which made age determination far more difficult. The non-carbonate 
fraction of the samples was less homogenous than the upper set of samples, a factor 
that would have caused scattering of the data in this way (T.C. Atkinson, pers. 
comm.). Further work is in progress to try and investigate this possibility and allow 
refinement of the age determination. 
CHRONOLOGY. 97 
In summary, the U-series dating method supports the pollen-derived stratigraphy for 
284 by independently confirming the position of the end of the Eemian in the 
sequence. It is possible that further refinement of this technique will not only allow 
the dating of the beginning of the Eemian with greater precision, but might also be 
used independently to date earlier interglacials. 
5.7TEPHRA 
The use of tephra (fine volcanic ash) deposits as a means of correlating sequences 
across a wide geographical area is well established. The eastern Mediterranean region 
has seen much volcanic activity during the Quaternary, particularly in Italy (the 
Campanian and Roman Provinces) and Greece (the Aegean Arc). The tephra (usually 
consisting of a mixture of pyroclastic ejecta, including glass shards, pumice and lithic 
fragments) is generally distributed over a wide geographical area and has a unique 
geochemical signature. Stratigraphical correlations based on geochemical analysis of 
the fragments are therefore possible. 
Oceanographers have been able to make good use of tephrochronology in the 
Mediterranean (e.g. Keller et aI., 1978; Vinci, 1985). Whilst the technique has been 
used widely in helping to date and correlate terrestrial Italian sequences, such as Lago 
Grande di Monticchio (Narcisi, 1996), there has curiously been very little use of the 
technique on mainland Greece (although Readman et al., 1976, Creer et al., 1981 and 
the two studies discussed below are notable exceptions). This may relate to the 
prevailing westerly winds, meaning that the mainland would receive material mostly 
from the distal Italian volcanic provinces, rather than from the Aegean eruptions. 
This would result in fewer and thinner tephras than would be seen in the Italian 
sequences, which would have been immediately adjacent to the volcanic sources. 
Tephras are therefore likely to be considerably harder to recognise in Greek sites. 
One recent study on a site from the Kalodiki basin in Epirus (less than 50km south of 
Ioannina) has recognised a lOcm thick light-brown to olive coloured tephra 
immediately beneath a peat layer with a minimum radiocarbon age of 31.8±1.2ka BP 
(St. Seymour and Christianis, 1995). Geochemical analysis, stratigraphical thickness 
and context all suggest that it is the Y-5 tephra, a well-known and widespread ash, 
originating from the Campanian Province in western Italy. Dates for this eruption 
derived largely from marine sequences range between 28-35ka. 
CHRONOLOGY • 98 
A careful search of the core material has failed to locate this tephra, even though the 
isopach map showing the thickness of ash fallout from the Y-5 eruption (St. Seymour 
and Christianis, 1995) indicates that the ash layer should be approximately 10cm 
thick. It is possible that the colour of the tephra (noted as light-brown to olive in the 
Kalodiki basin) makes it difficult to see in material that is also predominantly light­
brown to olive in colour (see description in section 4.1.1). In addition, it is likely that 
the ash would float on the surface of the lake for some considerable time (van Andel, 
pers camm.). Winds would blow it to the margins of the lake where it would 
eventually sink and become part of the sediment. Much less material would therefore 
accumulate in deeper water, such as the location of 284. It is possible that 
tephrochronology may offer a means by which the stratigraphy of marginal lake cores 
(usually difficult to date because of the presence of hiatuses) can be better understood. 
One very recent study (Pyle et aI., in press) has examined a 2m thick tephra deposit 
from a raised palje at Morphi in Epirus, approximately 50km south-west of Ioannina. 
By using 40 ArP9 Ar dating techniques on feldspars within the tephra, an eruption age of 
374±7ka has been derived. Pyle et al. (in press) have calculated that the ash was the 
result of the eruption of between 130-400km3 of dense trachytic magma from an 
Italian source located at least 750km away. They have suggested that one possible 
candidate for the source of the this tephra is the Baccano-Sacrofano caldera in the 
Sabatini Volcanic Complex, north of Rome. By comparison, the volume of the 
Campanian eruption responsible for the Y -5 tephra was only 80km3 . 
Despite the 2m thick deposit of tephra at MOl'phi, careful analysis of the Ioannina 
material has so far failed to find any definite trace of this ash layer, either over the 
expected interval in the core predicted by the age model, or elsewhere in the 
sequence. Although the points raised above concerning the dissipation of ash within 
the lake must still be borne in mind, it seems highly unlikely that an eruption of such 
a magnitude would not be recorded in the Ioannina sediments. In view of the extreme 
usefulness of a stratigraphical 'golden spike' in this part of the record, further detailed 
analysis of the sediments is clearly justified. It is possible that a study of the magnetic 
susceptibility of the sediments at much higher resolution may reveal the presence of 
tephra particles in this interval of the core. 
CHRONOLOGY • 99 
5.8 SUMMARY AND PROPOSED AGE MODEL 
A range of dating methods were applied to core 284 in an attempt to derive a 
comprehensive and robust age model. A chronological framework was initially 
provided by correlating the magnetic susceptibility profiles from cores 284 and 249. 
The stratigraphy derived for 249 could then be transferred to 284 because both the 
APINAP pollen record from 249 and the carbonate content from 284 are first-order 
proxies for climatic variability. The chronology for 249 appears to extend back to 
423ka. In order to provide a provisional estimate of the age of 284 beyond this, the 
carbonate curve was compared with the marine oxygen isotope record. On this basis 
it is suggested that 284 extends back into MIS-16 and is therefore over 620ka old. 
A variety of other techniques were then used in order to test and refine this 
provisional age model. Radiocarbon dating of molluscan material from the top 40m 
of the sequence provided a suite of dates that were close or slightly older than the 
predicted ages, but which are likely to have been affected by hard-water error. In the 
absence of any corrections, no detailed chronology could be applied to this part of the 
record, although certain tentative interpretations concerning the position of the LGM 
and beginning of the Holocene were made that could later be tested by other 
techniques (e.g. isotopic analysis). An age-depth plot was presented (fig. 5.4) that 
suggested significantly higher accumulation rates during interglacial periods, possibly 
due to increased precipitation and temperatures. 
Palaeomagnetic analysis of the sediments demonstrated that the core was normally 
magnetised. This suggested that the Brunhes - Matuyama boundary had not been 
reached and that the core was younger than 780ka (Shackleton et al., 1990; Baksi et 
al., 1992). Several discrete magnetic polarity reversals were found within the record 
that were correlated with well-documented magnetic excursions from the Brunhes 
chron. This magnetostratigraphy appeared to offer good support to the proposed age 
model. 
Uranium series dating techniques were applied to carbonates derived from two levels 
in the core that were thought (on the basis of the proposed age model) to be 
equivalent to the beginning and end of the last interglacial period. Dates were 
obtained that supported this interpretation. 
Other dating techniques were less successful. A pilot analysis of amino acid ratios 
from mollusc shells was unable to provide a coherent chronology, most probably 
because of preservation problems. A search for tephra layers within the sequence was 
CHRONOLOGY • 100 
also unsuccessful, most probably because of the difficulties in recognising thin, olive­
coloured ash layers in a core of this extreme length, where most of the sediments are 
also olive-coloured. 
In summary, the age model developed initially by using the dating derived for core 
249 appears to provide a comprehensive and robust chronological framework, 
particularly down to about 420ka. Beyond this, a match with the marine oxygen 
isotope curve and possible magnetostratigraphic support suggests that the sequence 
stretches back over 620ka, into MIS-16. If the age model of Prell et al. (1986) is used 
for the marine isotope stage boundaries (except for the MIS-9/8 boundary, discussed 
in section 8.2), an age-depth plot for 284 can be generated (fig. 5.8). The gently 
steepening curve suggests that accumulation rates gradually increased over time, 
although the average was about O.5rnka-'. The actual values are given in table 5.4. 
Comparison between this age model for 284 and the global magnetostratigraphy 
shows remarkable similarity (fig. 5.9). Implicit in this exercise is the assumption that 
terrestrial and marine systems have a synchronous response to climate change. As 
Tzedakis (1991) points out, although the marine oxygen isotope record is in itself a 
valuable stratigraphic tool, an assessment of leads and lags between marine and 
terrestrial responses to climatic variability can only be achieved when an independent 
terrestrial chronology can be established. 
Marine Depth Age Duration Thickness Accumulation 
Isotope Stage (m) (ka) (ka) (m) Rate (mka-1) 
1 12 17.8 1.48 
2 17.8 12 12 14.2 1.18 
3 32.0 24 35 24.8 0.71 
4 56.8 59 12 7.7 0.64 
5 64.5 71 57 33.1 0.58 
6 97.6 128 58 42.4 0.73 
7 140.0 186 59 23.0 0.39 
8 163.0 245 58 23.4 0.40 
9 179.3 278 61 33.9 0.56 
10 213 .2 339 23 4.6 0.20 
11 217.8 362 61 16.8 0.28 
12 234.6 423 55 18.2 0.33 
13 252.8 478 46 19.6 0.43 
14 272.4 524 41 8.8 0.21 
15 281 .2 565 55 27.0 0.49 
16 308.2 620 
Table 5.4 Age-depth values for core 284. Ages (from Prell et al. , 1986) and depths refer to 
boundaries (e.g. MIS-6/5 boundary is located at 97.6m and has an age of 128ka). Position of MIS-9/8 
boundary follows Tzedakis et at. (1997) (see section 8.2 for details). 
~ 
..c 
Q.. 
Q.) 
Cl 
o ~.~---------------------------------------. 
. 
20 -I~ 
. 
• 40 -I ... 3 
60 -
80 -
100 -
120 -
140 -
160 -
180 -
200 -
220 -
240 -
260 -
280 -
300 -
o 
--. 
. 
. 
100 
5 
~ 
. 
. 
. 
. 
-
•••• 7 
". 
.. 
\,~ 9 
• ....... 11 
..•.. 
•• ... • •• 13 
.......... 
.... 
200 300 400 500 600 700 
Age (ka) 
Fig. 5.8 Age-depth plot based on the age model for 284 proposed in 
section 5.8. Selected marine oxygen isotope stages (interglacials) are 
indicated (see text for details). 
0 
. 
~ 20 
o 
. 
40 • Laschamps 
60 
80 
lOO 
120 
140 
160 ~ 
..c 
180 Q.. Q.) 
Cl 
. 200 
220 
240 
..... 
....•. 
~ ....... 
260 
280 
. r 300 .. 
lOO 200 300 400 500 600 700 
Age (ka) 
Fig. 5.9 Comparison of magnetostratigraphy (outlined in section 5.4) 
with the age-depth plot for 284. Position of proposed magnetic polarity 
episodes are indicated (see text for details). 
Cl ~ 
~ 
~ 
Cl 
Cl 
'<:: 
• 
...... 
o 
6. FAUNAL BIOSTRATIGRAPHY AND 
MODERN ASSEMBLAGES 
6.1 INTRODUCTION 
An essential patt of this study is the analysis of the main faunal elements preserved 
within the core. The reasons for this are twofold. Firstly, by attempting to derive 
useful palaeoenvironmental information, it is hoped that a detailed lake history can be 
reconstructed. Secondly, it is hoped that changes within the faunal assemblages can be 
usefully employed as a biostratigraphical tool to help improve the age model. 
In order to derive meaningful palaeoenvironmental information, an understanding of the 
ecologies of the various taxa is required. To a celtain extent, the study of the modern­
day fauna can help to establish the ecological preferences and tolerances of those 
species cUlTently living in the lake. This is useful as long as these species are also 
present within the fossil record; difficulties naturally at'ise if the species are now extinct. 
In addition, unless any modern faunal study is comprehensive, it can be difficult to 
predict accurately the habitat preferences of poorly studied, rare or endemic species. 
Good biostratigraphical information relies either on a widely applicable and easily 
recognisable succession of taxa rising and declining in impOltance (such as is often 
used in palynological studies), or the appeat'ance/evolutioniextinction of a patticular 
taxon over a wide geographical area. Of course, these only provide a means for relative 
correlation, unless they can be linked to an independently derived chronology. In such 
cases, they then have the potential for absolute biostratigraphical correlation. 
Published studies of the lake fauna from Ioannina, or the Balkans in general, are 
somewhat limited. It is true that much work has been undeltaken on the remarkable 
endemic faunas from Lakes Ohrid and Prespa, both located on the Albanian-Greek­
Macedonian borders (e.g. Stankovic, 1960). Whilst of some relevance, these studies 
are highly site-specific and have mostly been conducted on the modern faunas. It is 
therefore uncertain whether conclusions based on the situation in these lakes can be 
transposed onto other lakes in the region. In addition, it is highly likely that a lake with 
a history as long as Pamvotis will have developed its own suite of endemic species. 
FAUNA AND BJOSTRATIGRAPHY. 103 
This study has concentrated on the molluscan and ostracod faunas from the lake. There 
are several reasons for choosing these groups. Firstly, they represent the best 
preserved and most numerous macrofossils within the core. Secondly, both mollusc an 
and ostracod studies have a long history and their use in both palaeoenvironmental and 
biostratigraphical contexts is well understood (e.g. Delorme, 1969; Carbonel et al., 
1988; Whatley, 1983; Goodfriend, 1992; Mouthon, 1992; Magny et al., 1995). 
Thirdly, the ostracods are being used for isotopic studies (section 3.3.3) and an 
understanding of the fauna involved can greatly assist in interpretation of those results. 
In addition, an attempt has been made to use the amino acids derived from celtain 
molluscs occurring in the core to establish an aminostratigraphy for the sequence 
(section 5.5). 
For reasons previously explained (section 1.4), the last glacial- Holocene and Eemian 
sections of the core have been selected for detailed study. In addition to this fossil 
material, the modern molluscan and ostracod faunas have also been examined. These 
reviews are not intended to be exhaustive, since type material archived in a variety of 
European museum collections has only occasionally been consulted. Nevertheless, it 
has been possible to update taxonomies and to describe species that are poorly known 
outside the Balkans. In addition, this is the first time that many of these species, 
particularly the endemic faunas from the lake, have been illustrated. 
6.2 AQUATIC MOLLUSCAN FAUNA 
The malacology of Greece has been the subject of study for at least the last two 
centuries. A comprehensive bibliography is provided by Butot and Welter-Schultes 
(1994), which covers almost all of the literature available on modern and fossil 
freshwater and marine molluscs from Greece over the period 1758-1994. A brief 
review of the malacological literature dealing exclusively with Ioannina is given here, 
concentrating specifically on the aquatic fauna (since it is this fauna that has been 
recovered from the core and collected from the modern lake during this study). This 
information is summarised in table 6.1. 
The first study of molluscan faunas from Ioannina was published by Mousson (1859), 
who described the material collected by Dr Alexandre Schlaefli on his travels around 
eastern Europe and the Balkans. A total of thirty-eight modern molluscan species were 
collected and identified, of which thirteen were aquatic. 
Species 
Viviparus janinensis (MOUSSON, 1859) 
Valvata cristata MULLER, 1774 
Valvata piscinalis (MULLER, 1774) 
Mercuria confusa (DRAPARNAUD, 1805) 
Bithynia graeca (WESTERLUND, 1879) 
Lymnaea auricularia (L1NNAEUS, 1758) 
Lymnaea peregra (MULLER, 1774) 
Lymnaea stagnalis (L1NNAEUS, 1758) 
Physa acuta DRAPARNAUD, 1805 
Planorbis carinatus MULLER, 1774 
Planorbis planorbis (L1NNAEUS, 1758) 
Gyraulus janinensis (MOUSSON, 1859) 
Planorbarius corneus (L1NNAEUS, 1758) 
Ancylus fluviatilis MULLER, 1774 
Anodonta anatina (L1NNAEUS, 1758) 
Anodonta cygnea (L1NNAEUS, 1758) 
Dreissena (Carinodreissena) cf. stankovici L'VOVA AND STAROBOGATOV, 1982 
Pisidium casertanum (POLl, 1791) 
Pisidium nitidum JENYNS, 1832 
Pisidium spp. 
Sphaerium corneum (L1NNAEUS, 1758) 
Mousson 
(1859) 
+ 
+ 
+ 
+ 
+ 
+ 
+ 
+ 
+ 
+ 
+ 
+ 
+3 
+ 
Table 6.1 Comparison of studies of modern aquatic molluscan fauna from Lake Pamvotis. 
, Likely to have been originally mis-identified as Viviparus contectus (see text). 
2 The single juvenile shell recovered during this study may actually be a species of Ferrissia (see text) . 
3/4 Originally identified as Dreissena polymorpha (see text). 
Dollfus 
(1922) 
+ 
+ 
+ 
+ 
+ 
+ 
+4 
Reischutz and This 
Sattmann (1990) study 
+' + 
+ 
+ 
+ + 
+ 
+ 
+ 
+ 
+ 
+ 
+ ~ 
+2 ~ + ~ + ::t>. 
+ ::t>. 
+ ~ 
+ tl::l 
+ 
...... 
0 
+ 
Co/) 
;;:j 
::t>. 
::::j 
Cl 
:::tl 
::t>. 
'"0 
:J:: 
'"<: 
• 
...... 
0 
.j>.. 
FAUNA AND BIOSTRATIGRAPHY • 105 
Possibly the most detailed investigation of the malacology at Ioannina was carried out 
by Dollfus (1922), who examined material collected by Marius Dalloni in 1918 from a 
site south-west of the village of Katsikas . By comparing the mollusc an biostratigraphy 
and lithologies with that from other sites in the Balkans (e.g. Corinth, the Megara 
Basin), the sediments were assigned an Upper Miocene age. This age estimate was 
subsequently supported by Dalloni (1923). Eight fossil aquatic taxa were identified 
(discussed in section 6.2.4). Dollfus (1922) also listed eight modern aquatic taxa 
collected from the lake by Dalloni. 
In his study of the geological evolution of Epirus and Thessaly, Aubouin (1959) listed 
nine species of aquatic mollusc collected from the Katsikas site (discussed in section 
6.2.4). On the basis of updated faunal comparisons, he thought that the sequence was 
Lower Pliocene in age. Fossil molluscs were again collected from this site by Rocher­
Bizon and identified by Gillet (1962), who recorded five aquatic taxa. He then went on 
to review and compare several other similar faunas from sites in Epirus before 
suggesting an Upper Pliocene age for the deposits. 
Guernet et al. (1977) returned to the Katsikas site some years later and took samples 
that were later analysed for pollen, charophytes, ostracods and molluscs. The ostracod 
fauna is discussed below (section 6.3). Eight molluscan taxa were identified (discussed 
in section 6.2.4). On the basis of the faunal assemblage, they also suggested that the 
age of these deposits was Upper Pliocene. 
Whilst on a recent field tour of Epirus, Reischtitz and Sattmann (1990) visited the 
Ioannina basin and collected thilty molluscan taxa, of which seven were aquatic. 
Finally, four species of modern hydrobiids have been described from Ioannina by 
Schtitt (1962, 1980) and Reischtitz and Sattmann (1990): Semisalsa steindachneri 
(WESTERLUND, 1902); Belgrandiella (Belgrandiella) haesitans (WESTERLUND, 1881); 
Horatia (Neohoratia) epirana SCHUTI, 1962; and Paladilhia (Paladilhiopsis) janinensis 
(SCHUTI, 1962). Their occurrence is limited to the springs and small streams on the 
northern shore of the lake; they have not been found within the lake itself. 
Accordingly, they are not considered further in this study, although a comprehensive 
discussion of the classification of these and other Balkan hydrobiids can be found in 
Kabat and Herschler (1993). 
Species 
Viviparus janinensis (MOUSSON, 1859) 
Valvata piscinalis (MULLER, 1774) 
Bithynia graeca (WESTERLUND,1879) 
Lymnaea auricularia (LINNAEUS, 1758) 
Lymnaea peregra (MULLER, 1774) 
Lymnaea stagnalis (LINNAEUS, 1758) 
Physa acuta DRAPARNAUD, 1805 
Anisus vorticulus (TROSCHEL, 1834) 
Gyraulus albus (MULLER, 1774) 
Gyraulus janinensis (MOUSSON, 1859) 
Planorbarius corneus (LINNAEUS, 1758) 
Ancylus jluviatilis MULLER, 1774 
Anodonta anatina (LINNAEUS, 1758) 
Anodonta cygnea (LINNAEUS, 1758) 
Dreissena (Carinodreissena) cf. stankovici 
L'VOVA AND STAROBOGATOV, 1982 
Pisidium nitidum JENYNS, 1832 
Pisidium spp. 
Habitat 
Little ecological infonnation. Likely to prefer deeper water habitats in hard-water lakes. 
Occurs in ponds and slow running waters, but prefers lakes with abundant aquatic vegetation. 
Prefers medium to hard-water open lake environments (dislikes stagnant water). 
Found in standing freshwater bodies, especially lakes in hard-water areas. 
Found in lakes and slow-flowing rivers, especially in hard-water areas. 
Found in a variety of fresh-water habitats. Prefers hard-water lakes with abundant aquatic vegetation. 
Found in marshy or shallow lake environments with abundant aquatic vegetation. 
Found in a wide variety of freshwater habitats. Prefers lakes with abundant aquatic vegetation. 
Found in a wide variety of freshwater habitats. Prefers lakes with abundant aquatic vegetation. 
Little ecological infonnation. Likely to prefer lakes with abundant aquatic vegetation. 
Prefers freshwater bodies with abundant aquatic vegetation. 
Found in rapid-flowing rivers or lakes with stony substrates and low levels of aquatic vegetation. 
Prefers rivers, canals and lakes with sandy substrates in which to bury itself. 
Prefers rivers, canals and lakes with muddy substrates in which to bury itself. 
Little ecological infonnation. Likely to prefer shallow, high-energy littoral areas in a variety of 
fresh-water and brackish environments. 
Found in lakes, streams and ponds, particularly in hard-water areas. 
Various, depending on species. See text. 
Table 6.2 List of aquatic molluscan fauna found during this study, with summarised ecological infonnation. See text for details. 
Modern I 
Fossil 
M/F 
M/F 
M/F 
M 
F 
M 
M 
F 
F 
M/F 
M 
M ~ M ~ M ::t.: 
M/F ;:t.. 
M/F 
§ 
tl:I 
....... M/F <:) 
V:l 
"'"i 
::ti 
;:t.. 
:::l 
<;) 
::>::I 
;:t.. 
"t:l 
::t:: 
"'<: 
• 
-0 
0\ 
Site 1 
Viviparus janinensis 
Valvata piscinalis 
Bithynia graeca 
Planorbarius comeus 
Lymnaea auricularia 
Lymnaea stagnalis 
Physa acuta 
Dreissena cf. 
stankovici 
Anodonta cygnea 
Site 10 
Valvata piscinalis 
Bithynia graeca 
Gyraulus janinensis 
Dreissena cf. 
stankovici 
Pisidium spp. 
FAUNA AND BJOSTRATIGRAPHY. 107 
Site 6 
Planorbarius comeus 
Gyraulus janinensis 
Pisidium sp. 
Site 7 
Viviparus janinensis 
Pl£morbarius comeus 
Lymnaea stagnalis 
Ancylus fillviatilis 
Site 2 
Viviparus janinensis 
PUmorbarius comeus 
Lynmaea auricularia 
Anodonta anatina 
Lake Pamvotis Site 11 
Site 4 
Lymnaea auricularia 
Lymnaea stagnalis 
Physaacuta 
Valvata piscinalis 
Viviparus janinensis 
PlaJlorbarius comeus 
Site 8 
Gyraulus janinensis 
Bithynia graeca 
Fig. 6.1 Sampling sites for modern aquatic molluscan fauna . See text for details. 
FAUNA AND BIOSTRATIGRAPHY • 108 
6.2.1 Systematic Review of the Aquatic Molluscan Fauna at Ioannina 
A general description of all aquatic mollusc an taxa found at Ioannina (modern and 
fossil) is now given. Their taxonomy is summarised, along with what is known of 
their occurrence, ecology and any other relevant details (summarised in table 6.2). The 
taxonomic nomenclature essentially follows Vaught (1988) unless otherwise indicated. 
The original citation is given for each species followed by a synonymy that refers only 
to usage in relation to material from Ioannina. Many of the species are well-known 
European forms and are thus only briefly described. In considering distribution, 
emphasis is placed on European (and in particular, Balkan) occurrences. The 
systematic review is followed by a separate consideration of the modern and fossil 
assemblages derived from this study and, where appropriate, an assessment made of 
their biostratigraphical and palaeoenvironmental importance. 
Class Gastropoda 
Subclass Prosobranchia 
Family Viviparidae 
Genus Viviparus DEMONTFORT, 1810 
Viviparus janinensis (MOUSSON, 1859) 
Plate 11, figs. A-C 
1859 Paludina injlata VILLA var. janinensis MOUSSON, Viertel}ahrschr. NatLlIj. Gesell. ZUrich, 4, 
p.55. 
1922 Vivipariajaninensis (MOUSSON). Dollfus, Bull. Soc. Gio!. Fr., 22 (4), p.123. 
Material examined: Numerous empty modern shells (see fig. 6.1 for sampling 
localities); fossil material from 284. 
Diagnosis: Shell large (mean height = 54.9mm, mean breadth = 40.1mm, n = 16), 
semi-transparent, broadly conical, olive green - brown in colour. Umbilicus absent. 
Spire tapers to a moderately pointed apex. Surface covered with growth-lines; three 
chestnut-brown bands visible on body whorl, two bands on the others. Sutures 
moderately deep and whorls convex. Whorls 5-6. Body whorl large, swollen, nearly 
75% of the total shell height. Aperture large, ear-shaped,with rounded outer margin. 
Operculum has concentric structure, with distinct nucleus. 
FAUNA AND BIOSTRATIGRAPHY. 109 
Occurrence: Ioannina is the type locality for V. janinensis (Mousson, 1859). It was 
also recorded by Dollfus (1922) as being abundant in the modem lake. Viviparids are 
widely distributed across nOlthem Europe, around the eastern Mediterranean, across 
central Africa, southern Asia, north and eastern Australia and central and eastern North 
America (Prashad, 1928). A large number of endemics have evolved, particularly in 
the Balkans (Prashad, 1928). V. janinensis is endemic to Ioannina and no records of it 
exist from elsewhere. 
Ecology: There are no published details on the ecology of V. janinensis. All of the 
specimens collected from the modern lake were shells recovered from the shore; no 
living specimens were observed. This suggests the preference of viviparids in general 
for offshore freshwater habitats in hard-water areas where the bedrock is dominated by 
carbonates (Macan, 1969; 0kland, 1990). 
Remarks: Reischtitz and Sattmann (1990) record V. contectus (MILLET, 1813) as Palt 
of the modern fauna of the lake. However, adult forms of V. contectus al'e much 
smaller than V janinensis, commonly 26-30mm high and 23-28mm broad (Graham, 
1988). They also taper to a sharply pointed apex and have very deep sutures. It is 
easy, therefore, to see the similarity between adults of V. contectus and juveniles of V. 
janinensis. In view of the abundance of V. janinensis in Lake Pamvotis, it seems likely 
that this represents a mistaken identification by Reischtitz and Sattmann (1990). 
However, this can only be confilmed after examination of their collected material 
(al'chived in the Naturhistorischen Museum, Vienna). 
Family Valvatidae 
Genus Valvata MULLER, 1774 
Valvata cristata MULLER, 1774 
1774 Valvata cristata MULLER, Venn. Hist., ii, p.198. 
1990 Valvata cristata MULLER. Reischiitz and Sattmann, Ann. Naturhist. Mus. Wien, 91B, p.256. 
Remarks: V. cristata has a distribution that ranges across most of Europe and parts of 
northern Asia, preferring freshwater habitats, such as lakes and slow-flowing rivers 
that al'e rich in aquatic vegetation (0kland, 1990). It WaS not found living in Lake 
Pamvotis by either Mousson (1859), Dollfus (1922), or during this study. 
FAUNA AND BIOSTRATIGRAPHY • 110 
Valvata piscinalis (MULLER, 1774) 
1774 Nerita piscinalis MULLER, Verm. Hist., ii, p.I72. 
1859 Valvata piscinalis (MULLER). Mousson, Vierteljahrschr. Naturj Gesell. Ziirich, 4, p.55. 
Material examined: Numerous modern shells, both living and dead (see fig. 6.1 for 
sampling localities); fossil material from 284. 
Occurrence: V. piscinalis is common across most of Europe, as well as in western, 
central and northern Asia (0kland, 1990). The species is known from both modern 
and fossil sites in Greece (e.g. Schtitt, 1987; Sattmann and Reischtitz, 1988; Butot and 
Welter-Schultes, 1994). 
Ecology: Although V. piscinalis can be found in slow running waters and occasionally 
in small ponds, it has a distinct preference for lakes with abundant plant macrophytes 
(0k1and, 1990). This was confirmed at Ioannina, where live specimens were collected 
from shallow littoral areas with muddy substrates and abundant floating broadleaf 
macrophytes (sites 4 and 10, fig. 6.1) . 
Family Hydrobiidae 
Genus Mercuria (BOETERS, 1971) 
Mercuria confusa (FRAUENFELD, 1863) 
1863 Amnicola confusa FRAUENFELD, Verhandl. k. k. zool. bot. Gesell. Wien, xiii, p.1029. 
1863 Bythinia similis DRAPARNAUD, in Reeve and Reeve, Moll. Brit. Is ., p.188. 
1859 Bythinia similis DRAPARNAUD. Mousson, Vierteljahrschr. Natufj Gesell. Ziirich, 4, p.54. 
Remarks: B. similis is a junior synonym of M. confusa. Mousson's (1859) record 
from Ioannina is here listed as such, but his specimens need checking to confirm this 
attribution. M. confusa is widespread around the Mediterranean and can be found in a 
variety of mildly brackish habitats with muddy substrates (Graham, 1988). It was not 
recorded by Dollfus (1922), Reischtitz and Sattmann (1990) or during this study. 
FAUNA AND BIOSTRATICRAPHY. III 
Family Bithyniidae 
Genus Bithynia LEACH in ABEL, 1818 
Bithynia graeca (WESTERLUND, 1879) 
Plate Ill, figs. A-G 
1859 Bythinia troscheli PAASCH, in Mousson, Vierteljahrschr. NatUlf Cesell. Zurich, 4, p.54. 
1879 Bythinia graeca WESTERLUND, in Westerlund and Blanc, Aper. Faune Mal. Crece, p.l37. 
1922 Bithynia boissieri KUSTER. Dollfus, Bull. Soc. Ciol. Fr., 22 (4) , p.123. 
1990 Bithynia graeca (WESTERLUND) . Reischi.itz and Sattmann, Ann. Naturhist. Mus. Wien,91B, 
p.256 . 
Material examined: Occasional empty modern shells (all juvenile) (see fig. 6.1 for 
sampling localities); fossil material from 284, mostly juvenile. 
Diagnosis: Shell broadly conical, relatively slender, opaque, tapering to a pointed apex. 
Height approximately Ilmm, breadth approximately 6.5mm. Surface covered with 
faint growth lines. Sutures moderately deep, whorls convex. Whorls 5-6. Body 
whorl large, swollen, nearly 40% of the shell height. Aperture large, oval, with a 
reflected lip (plate Ill). Calcitic operculum has a concentric stlUcture, with a distinct 
paucispiral nucleus. 
Occurrence: loannina is the type locality for Bithynia graeca (Westerlund and Blanc, 
1879). It has a distribution that extends across northern Greece and around the 
northern Aegean and Anatolian coastal areas (Schtitt, 1987). 
Ecology: Eleutheriadis et al. (1993) recently conducted a study of B. graeca from Lake 
Kerkini, an artificial lake in northern Greece. They noted that it prefers hard or medium 
waters and is most abundant at the barrage of the lake. During the summer it feeds on 
diatoms, algae and pond-weed, whereas during the winter it prefers to be buried in 
mud, feeding on organic detritus. Its apparent dislike for stagnant waters was reflected 
at loannina, where it was only found at the open sampling sites (e.g. sites 1, 8 and 10, 
fig. 6.1), rather than those in backwater areas. 
Remarks: Mousson (1859) described two species of bithyniid in the modern fauna at 
loannina. The smaller he called B. similis, which is not a tlUe bithyniid but a hydrobiid 
(Mercuria confusa, see above). He identified the larger species as B. troscheli PAASCH, 
1842, which is a form closely related to B. leachii SHEPPARD, 1823 (a well-known and 
widespread European species) and regarded by some as conspecific. Dollfus (1922) 
also identified the modern bithyniid in Lake Pamvotis as a form of B. leachii , known as 
FAUNA AND BIOSTRATIGRAPHY • 112 
B. boissieri KOSTER, 1852. Whilst B. leachii is very similar to the bithyniid found at 
Ioannina, it is commonly much smaller (approximately 5mm high and 4.5mm broad), 
has deeper sutures, a larger body whorl and is generally less slender (Graham, 1988). 
It is possible therefore, that Mousson (1859) and Dollfus (1922) only examined 
juvenile forms from Lake Pamvotis. Westerlund and Blanc (1879) considered that the 
bithyniid from Ioannina was sufficiently different from B. leachii to describe it a new 
species, B. graeca. This view was later supported by Schtitt (1987) in his discussion 
and review of Balkan bithyniids. 
Family Lyrnnaeidae 
Genus Lymnaea LAMARCK, 1799 
Lymnaea auricularia (LINNAEUS, 1758) 
1758 Helix auricularia LINNAEUS, Syst. Nat., ed.10, p.774, no.617 . 
1922 Limnea auricularia (LINNAEUS) . Dollfus, Bull. Soc. Geol. Fr., 22 (4), p.123 . 
Material examined: Numerous modern shells, both living and dead (see fig . 6.1 for 
sampling localities); no fossil material was recovered from 284. 
Occurrence: L. auricularia is found throughout Europe, Asia, northern Africa and 
Alaska and has been introduced to the continental United States (0kland, 1990). It is 
widespread in Greece, where it is known from both modern and fossil records (e.g. 
Schtitt et al., 1985; Schtitt, 1987; Sattmann and Reischtitz, 1988; Butot and Welter­
Schultes, 1995). 
Ecology: L. auricularia generally prefers standing freshwater bodies, pruticularly lakes 
in hard-water areas (Macan, 1969; 0kland, 1990). At Ioannina, live specimens were 
particularly abundant at shallow, vegetation-rich littoral sites with muddy substrates 
(e.g. site 4, fig. 6.1). 
Remarks: Interestingly, L. auricularia was not recorded as part of the modern fauna by 
Mousson (1859) or Reischtitz and Sattmann (1990), although it was included by 
Dollfus (1922). 
FAUNA AND BIOSTRATICRAPHY. 113 
Lymnaea peregra (MULLER, 1774) 
1774 Buccinum peregrum MULLER, Verm. Hist. H, p.130. 
1859 Limnaeus vulgaris (MULLER). Mousson, Vierteljahrschr. Natulf, Cesell. Zurich, 4, p.52. 
1990 Lymnaea peregra (MULLER). Reischiitz and Sattmann, Ann. Naturhist. Mus. Wien, 91B , 
p.257 . 
Material examined: Occasional fossil material from 284. 
Occurrence: L. peregra is widespread throughout Europe, northern and middle Asia 
and north-western Africa (0kland, 1990). It is also well-known as a modern species 
from the Balkans, for example from Lake Prespa (Sattmann and Reischiitz,1988) and 
from the fossil record, for example the Lower Quaternary deposits of Greek Macedonia 
(Schiitt, 1987). 
Ecology: L. peregra prefers lakes and slow-flowing rivers rich in aquatic vegetation 
(0kland, 1990). No modern forms were found at Ioannina during this study, although 
since it has previously been recorded as patt of the modern fauna by Mousson (1859) 
and Reischiitz and Sattman (1990), it may be recovered with further detailed sampling. 
Lymnaea stagnalis (L1NNAEUS, 1758) 
1758 Helix stagnalis LINNAEUS, Syst. Nat., ed.lO, p.774, no.612. 
1859 Limnaeus stagnalis (LINNAEUS). Mousson, Vierteljahrschr. Nattllf, Cesell. Zurich, 4, p.52. 
1922 Lil1lnea stagnalis (LINNAEUS). Dollfus, Bull. Soc. Cia/. Fr., 22 (4), p.123. 
Material examined: Numerous modern shells, both living and dead (see fig. 6.1 for 
sampling localities); no fossil material was recovered from 284. 
Occurrence: L. stagna lis has a holarctic distribution, being found throughout Europe, 
Asia, North America and parts of northern Africa. It is well known in Greece from 
both modern and fossil records (e.g. Schiitt et al., 1985; Schiitt, 1987; Sattmann and 
Reischiitz, 1988; Butot and Welter-Schultes, 1995). 
Ecology: L. stagnalis can be found in most freshwater habitiats, but has a preference 
for lakes in hard-water areas with abundant macrophytic vegetation (Macan, 1969; 
FAUNA AND BIOSTRATIGRAPHY • 114 
0kland, 1990). This was confirmed at Ioannina, where live specimens were found to 
be abundant in shallow, vegetation-rich backwater sites (e.g. site 7, fig 6.1). 
Family Physidae 
Genus Physa DRAPARNAuD,1801 
Physa acuta DRAPARNAUD, 1805 
1805 Physa acuta DRAPARNAUD, Hist. Moll. France, p.55, pl.iii, figs. 10 and 11 
Material examined: Several modern shells, both living and dead (see fig. 6.1 for 
sampling localities); no fossil material was recovered from 284. 
Occurrence: P. acuta is generally known as a Mediterranean species (Macan, 1969) and 
can be found throughout the Balkans, for example in Lake Kerkini in northern Greece 
(Eleutheriadis et al., 1993). It is also found in the fossil record from the region, for 
example in Lower Pleistocene deposits from north-eastern Greece (Schtitt, 1987). This 
is the first record of it from Ioannina. 
Ecology: P. acuta is known to favour marshy or vegetation-rich freshwater bodies, 
particularly lakes (Macan, 1969; Eleutheriadis et al., 1993). This was confirmed at 
Ioannina, where live specimens were abundant at site 4, a shallow littoral area with 
abundant broadleaf vegetation. 
Family Planorbidae 
Genus Planorbis GEOFFROY, 1767 
Planorbis carinatus MULLER, 1774 
1774 Planorbis carinatus MULLER, Verm. Hist., ii, p.157 . 
1859 Planorbis carinatus MULLER. Mousson, Vierteljahrschr. Naturf Gesell. Ziirich, 4, p.53. 
1922 Planorbis carinatus MULLER. Dollfus, Bull. Soc. Geol. Fr., 22 (4), p.123. 
Remarks: P. carinatus is common in a range of lakes rich in aquatic vegetation across 
most of Europe and parts of Asia (Meier-Brook, 1983; 0kland, 1990). Although 
recorded by Mousson (1859) and Dollfus (1922) as part of the modern fauna, it was 
not found by either Reischtitz and Sattmann (1990) or this study. 
FAUNA AND BIOSTRATlGRAPHY • 115 
Planorbis planorbis (LINNAEUS, 1758) 
1758 Helix planorbis LINNAEUS, Syst. Nat., ed.l0, p.769, no.578. 
1805 Planorbis marginatus DRAPARNAUD, Hist. Moll. France, pA5, pl.ii, figs. 11 and 12. 
1859 Planorbis marginatus DRAPARNAUD . Mousson, Vie!1eljahrschr. Nattllf Gesell. ZUrich, 4, 
p.53 . 
1990 Planorbis planorbis (LINNAEUS). Reischiitz and Sattmann, Ann. Naturhist. Mus. Wien, 91B, 
p.257 . 
Remarks: P. planorbis is common in a range of lakes and ponds rich in aquatic 
vegetation across most of Europe, western and northern Asia and parts of North Africa 
(Meier-Brook, 1983; 0kland, 1990). Although recorded by Mousson (1859) and 
Reischtitz and Sattmann (1990) it was not found as part of the modern fauna by either 
Dollfus (1922) or this study. 
Genus Anisus STUDER, 1820 
Anisus vorticulus (TROSCHEL, 1834) 
1834 Planorbis vorticulus TROSCHEL, De Limnaeaceis, p.51. 
Material examined: Rare fossil material from 284. 
Occurrence: A. vorticulus has a distribution stretching across central and south-eastern 
Europe, into the western regions of Asia (Meier-Brook, 1983). This is the first record 
of this species from Ioannina. 
Ecology: A. vorticulus can be found in a variety of freshwater habitats, but like many 
of the more common planorbids, has a preference for lakes with abundant macrophytic 
vegetation (0kland, 1990). No modern material was found, making it impossible to 
assess its ecological preferences at Ioannina. 
FAUNA AND BIOSTRATIGRAPHY • 116 
Genus Gyraulus CHARPENTIER, 1837 
Gyraulus albus (MULLER, 1774) 
1774 Planorbis albus MULLER, Verm. Hist. , ii, p.l64. 
Material examined: Occasional fossil material from 284. 
Occurrence: Gyraulus albus is known throughout Europe and in western parts of Asia 
(Meier-Brook, 1983). It is also recorded from numerous fossil and modern sites in 
Greece (e.g. Schtitt et al., 1985; Schtitt, 1987; Sattmann and Reischtitz, 1988). This is 
the first record of this species from Ioannina. 
Ecology: Found in most freshwater habitats, G. albus prefers lakes with abundant 
macrophytic vegetation (0kland, 1990). No modern material was found , making it 
impossible to assess its preferences at Ioannina. 
Gyraulus janinensis (MOUSSON, 1859) 
Plate Ill, figs. H-M 
1859 Planorbis janinensis MOUSSON, Vierteljahrschr. NatUlf Gesell. ZUrich, 4, p.53. 
Material examined: Occasional empty modern shells (see fig. 6.1 for sampling 
localities); fossil material from 284. 
Diagnosis:, Shell discoidal, relatively small (mean breadth = 5mm, mean height = 
1.7mm, n = 5), thin and semi-transparent. Weak traces of spiral sculpture, pruticularly 
on final whorl. Wide, depressed umbilicus. Spire is involute, with each whorl 
overlapping the previous one. Whorls 4-5, rapidly enlarging. Moderately deep 
sutures. Keel distinctive and blunt in comparison to G. albus. Ovate aperture. 
Occurrence: Ioannina is the type locality for G. janinensis. It is unknown elsewhere 
and appears to be endemic. 
Ecology: No ecological information IS available on G.· janinensis. It is likely, 
however, that in common with other similru' planorbids it prefers lake habitats rich in 
aquatic vegetation. It was found at a variety of sites at Ioannina (fig. 6.1), making it 
difficult to assess any specific ecological preferences. 
FAUNA AND BIOSTRATIGRAPHY. 117 
Remarks: Endemic species of Gyraulus are well-known from the Balkans. For 
example, five endemic species have been recorded from Lakes Ohrid and Prespa (e.g. 
Hubendick and Radoman, 1959), although these have all now been assigned to the 
subgenus Carinogyraulus POLlNSKl, 1929 (Meier-Brook, 1983). The morphological 
characteristics of the form at Ioannina clearly permit a taxonomic revision, enabling it to 
be transferred from the genus Planorbis to Gyraulus. 
Genus Planorbarius FRORIEP, 1806 
Planorbarius corneus (LINNAEUS, 1758) 
Plate 11, figs. D, H, I 
1758 Helix cornea L1NNAEUS, Syst. Nat., ed.lO, p.770, no.587. 
1859 Planorbis corneus (L1NNAEUS) . Mousson, Vierteljahrschr. Naturf Gesell. ZUrich, 4, p.52. 
1922 Planorbis corneus (L1NNAEUS). Dollfus, Bull. Soc. Gi o/. Fr., 22 (4), p.123. 
Material examined: Numerous modern shells , both living and dead (see fig. 6.1 for 
sampling localities); no fossil material was recovered from 284. 
Occurrence: P. corneus is common throughout most of Europe (except the north), as 
well as large parts of Asia (0kland, 1990) and the Balkans (e.g. Stankovic, 1960). On 
the Greek mainland, it is widespread except in the Peleponnese, although it has been 
found in Pleistocene deposits from that region (Schtitt et al., 1985). It is also known 
from a range of Quaternary deposits from elsewhere in Greece (e.g. Schtitt, 1987). 
Ecology: P. corneus can be found in lakes and ponds, although it appears to favour 
smaller freshwater bodies rich in macrophytic vegetation (0kland, 1990). This was 
confirmed at Ioannina, where live specimens were abundant, particularly in vegetation­
rich backwater sites (e.g. sites 2, 6 and 7, see fig. 6.1). 
Remarks: The species of P. corneus found at Ioannina is unusually large (adults are 
commonly -40mm in diameter, compared with the -30mm diameter of more typical 
specimens). Schtitt (1987) also discusses similarly large specimens of P. corneus from 
north-eastern Greece and Lake Prespa. He suggests that they should strictly be referred 
to as Planorbarius corneus fOlma grandis DUNKER, 1850. 
FAUNA AND BIOSTRATIGRAPHY. 118 
Family Ancylidae 
Genus Ancylus GEOFFROY, 1767 
Ancylus fiuviatilis MULLER, 1774 
1774 Ancylus fluviatilis MULLER, Venn. Hist ., ii, p.20l. 
1859 Ancylus radio latus, KUSTER. Mousson, Vierteljahrschr. Natwf. Gesell. ZUrich, 4, p.54. 
1990 Ancylus fluviatilis MULLER. Reischtitz and Sattmann, Ann. Naturhist. Mus. Wien, 91B, 
p.257. 
Material examined: A single juvenile shell from site 7 of the modern lake (see fig. 6 .1 
for sampling locality). 
Occurrence: A. fluviatilis has a broad distribution throughout most of central and 
southern Europe (0kland, 1990). It is also well-known from Greece (e.g. Sattmann 
and Reischiitz, 1988). 
Ecology: A. fluviatilis has a preference for rapid-flowing rivers and lake environments 
subject to wave action. It also prefers habitats with stony substrates and minimal 
aquatic vegetation (0kland, 1990). 
Remarks: The single, juvenile shell examined possessed radial striations from the apex 
that were finer than normal for A. fluviatilis. It is possible therefore that it represents a 
species of Ferrissia WALKER, 1903, although without adult specimens, this is impossible 
to verify. Since A. fluviatilis has been recorded from Lake Pamvotis by both Mousson 
(1859) and Reischiitz and Sattmann (1990), this specimen has also been listed as such, 
pending further investigation. 
Class Bivalvia 
Family Unionidae 
Genus Anodonta (LINNAEUS, 1758) 
Anodonta anatina (LINNAEUS, 1758) 
1758 Mytilus anatinus LlNNAEUS, Syst. Nat., ed. lO, p.706, no. 219. 
1859 Anodonta piscinalis (NILSSON) . Mousson, Vierteljahrschr. Natwf. Gesell. ZUrich, 4, p.56. 
1922 Anodonta anatina (LlNNAEUS). Dollfus, Bull. Soc. Gio!. Fr., 22 (4), p.123. 
FAUNA AND BIOSTRATIGRAPHY. 119 
Material examined: Several dead shells from sampling site 2 of the modern lake (fig. 
6.1). It was not present in the fossil fauna from 284. 
Occurrence: A. anatina has a broad palaearctic distribution (Ellis, 1962) and is known 
throughout the Balkans as part of the modern and fossil record (e.g. Butot and Welter­
Schultes, 1994). 
Ecology: A. anatina prefers freshwater habitats such as rivers, canals and lakes with 
sandy substrates in which it can bury itself (Ellis, 1962). At Ioannina it was only found 
at site 2 (fig. 6.1), a relatively shallow backwater area with a muddy substrate. It is 
probable that the dead shells were transported to this site and washed ashore by wave 
action. 
Anodonta cygnea (LINNAEUS, 1758) 
1758 Mytilus cygneus LINNAEUS, Syst. Nat., ed.10, p.706, no. 218. 
1859 Anodonta cellensis (LINNAEUS). Mousson, Vierteljahrschr. Natuif. Gesell. ZUrich, 4, p.56. 
Material examined: A single pair of dead shells from sampling site 1 of the modern lake 
(fig. 6.1). It was not present in the fossil fauna from 284. 
Occurrence: A. cygnea has a broad palaearctic distribution (Ellis, 1962). It is 
widespread throughout the Balkans and is also known from the Quaternaty fossil 
record from Greece (e.g. Schtitt, 1987). 
Ecology: A. cygnea prefers freshwater habitats such as rivers, canals and lakes with 
muddy substrates (Ellis, 1962). At Ioannina it was only found at site 1 (fig. 6.1), a 
relatively high energy beach where it is probable that it was washed up by wave-action. 
Family Dreissenidae 
Genus Dreissena BENEDEN, 1835 
Subgenus Carinodreissena L'VOVA AND STAROBOGATOV, 1982 
Dreissena (Carinodreissena) cf. stankovici L'VOVA AND STAROBOGATOV, 1982 
Plate n, figs. E, F, J 
FAUNA AND BIOSTRATIGRAPHY. 120 
1982 Dreissena (Carillodreissena) stankovici L ' VOVA AND STAROBOGATOV, Zoologicheskij 
Zhurnal, 61 (11), p.1749, fig .I. 
Material examined: Numerous dead shells from the modern lake shore (see fig . 6.1 for 
sampling locations); fossil material from 284; comparative modern material from Lake 
Ohrid (collected by T. Meijer). 
Occurrence: D. stankovici was originally described as an endemic species from Lake 
Ohrid (L'vova and Starobogatov, 1982). 
Ecology: Little is known concerning the ecology of D. stankovici, although Dreissena 
sp. is known to prefer shallow, relatively high energy littoral areas in a wide range of 
fresh and brackish conditions. It often occurs in mussel 'beds' in extremely high 
numbers, attached to the substrate by means of byssal threads. 
Remarks: Dreissena was recorded (as D. polymorpha PALLAS, 1771) as patt of the 
modern fauna at Ioannina by both Mousson (1859) and Dollfus (1922). D. 
polymorpha has a broad distribution as both a fossil and a modern form in western 
Europe, parts of Turkey, parts of western Asia (including the Caspian Sea, Aral Sea 
and the Black Sea) and the Balkans (e.g. Serafimova-Hadzische, 1974; Sattmann and 
Reischtitz, 1988; Schtitt, 1993; Rosenberg and Ludyanskiy, 1994). However, 
examination of the specimens from Ioannina by T. Meijer (RGD, Haat'lem) suggested 
that both the modern and fossil specimens were not D, polymorpha, since they had a 
thread-like carina, rather than the more usual blunt carina seen on D, polymorpha. He 
thought that the Ioannina specimens to be most similar to the endemic Ohrid species D. 
stankovici described by L'vova and Stat'obogatov (1982) (T. Meijer, pers comm.). 
These authors suggested assigning the Lake Ohrid species to a new subgenus, 
Carinodreissena, a recommendation adopted by Rosenberg and Ludyanskiy (1994) in 
their recent nomenclatural review of this genus. 
The Ioannina specimens are very similat' to the Ohrid material both in terms of overall 
morphology and the presence of a thread-like carina. They also possess a septum that 
is shaped like an equilateral triangle, rather than the more irregulat'ly-shaped triangulat' 
septum of D. polymorpha (L'vova and Starobogatov, 1982), However, the Ioannina 
forms are smaller and generally more slender, with an average length of approximately 
23mm (n = 16), as opposed to an average length of 40mm for the Ohrid species 
(L'vova and Starobogatov, 1982). It is therefore possible that they represent a 
completely new species, perhaps endemic to the lake. Clearly, further comparison with 
type material is necessary before any final assessment can be made. 
FAUNA AND BIOSTRATIGRAPHY • 121 
Family Pisidiidae 
Genus Pisidium PFEIFFER, 1821 
Pisidium casertanum (POLl, 1791) 
1791 Cardium casertanum POLl, Test. utr. Silicae, i, ord.II, p.65, pl.xvi, fig. 1. 
1990 Pisidium caseltanum (POLl). Reischiitz and Sattmann, Ann. Naturhist. Mus. Wien, 91B, 
p.258 . 
Remarks: P. casertanum is a very widely distributed, eurytopic species. It can be 
found throughout Europe, parts of Asia, Africa, North America and New Zealand in 
various freshwater habitats at altitudes up to 2,500m and at depths down to 40m (Ellis, 
1962). It was not found as part of the modern fauna by either Mousson (1859), 
Dollfus (1922) or this study. 
Pisidium nitidum JENYNS, 1832 
1832 Pisidium nitidum JENYNS, TraIlS. Camb. Phi/' Soc., iv, p.304, p\.xx, figs. 7 and 8. 
Material examined: Numerous dead shells from the modern lake (see fig. 6.1 for 
sampling sites); fossil material from 284. 
Occurrence: P. nitidum is widespread throughout Europe and parts of North Amel1ca 
(Ellis, 1962). It is also well known from the fossil record, for example from Lower 
Pleistocene deposits in north-eastern Greece (Schiitt, 1987) and the Megalopolis Basin 
(Schiitt, 1985). This is the first record of P. nitidum from Ioannina. 
Ecology: P. nitidum can be found in most lakes, streams, rivers and ponds, though 
prefers harder water areas (Ellis, 1962). At Ioannina it was only found as dead shells 
around the lake edge (fig. 6.1), making more specific conclusions concerning its 
ecology there difficult. 
Pisidium spp. 
Material examined: Several dead shells from the shores of the modern lake (see fig. 6.1 
for sampling sites); fossil material from 284. 
FAUNA AND BIOSTRATIGRAPHY. 122 
Remarks: It is notoriously difficult to distinguish between the different species of 
Pisidium, especially with worn fossil specimens. Several species were found in both 
the modern and fossil fauna at Ioannina as part of this study, of which only P. nitidum 
was identified to species level (above) . No species of Pisidium were found by either 
Mousson (1859) or Dollfus (1922) in their studies of the modern lake fauna, although 
Reischiitz and Sattmann (1990) found P. casertanum (above). Specialist examination 
of both the modern and fossil fauna is clearly necessary. 
Genus Sphaerium SCOPOLI, 1777 
Sphaerium corneum (LINNAEUS, 1758) 
1758 Tellina cornea LINNAEUS, Syst. Nat., ed.lO, p.678, no.57. 
1859 Cyclas cornea (LINNAEUS). Mousson, Vierteljahrschr. Natwf Gesell. ZUrich, 4, p.55. 
Remarks: According to Ellis (1962), S. corneum has a holarctic distribution throughout 
Europe, nOlthern Asia and parts of North America. It is also well known from the 
fossil record, for example from Lower Pleistocene deposits in north-eastern Greece 
(Schiitt, 1987) and the Megal6polis Basin (Schiitt, 1985). It lives up to depths of 30m 
in most kinds of fresh-water habitats and although can tolerate mildly brackish 
conditions, it cannot live in polluted water (Ellis, 1962). This may explain its absence 
from the modem faunallist of Dollfus (1922) and from this study. 
6.2.2 Modern Aquatic Molluscan Fauna 
During September, 1994, water chemistry and sediment samples were taken at eleven 
littoral zone sites around Lake Pamvotis (section 3.2.6). At the same time, a qualitative 
collection of molluscs was made at eight of these sites. The sample site locations are 
shown in fig 6.1 and briefly described in appendix A. A total of 14 taxa were collected: 
Viviparus janinensis 
Valvata piscinalis 
Bithynia graeca 
Lymnaea auricularia 
Lymnaea stagnalis 
Physaacuta 
Gyraulus janinensis 
Gyraulus albus 
Planorbarius corneus 
Anodonta cygnea 
Anodonta anatina 
Dreissena cf. stankovici 
. Pisidium nitidum 
Pisidium spp. 
FAUNA AND BJOSTRATIGRAPHY. 123 
It must be emphasised that the collection techniques employed were neither rigorous nor 
quantitative and the list is likely to be incomplete (see table 6.1 for comparisons with 
other modern faunal studies). Nevertheless, this list of taxa is of value since there are 
several differences to those published previously, in particular, the study by Mousson 
(1859). This could be a reflection of the onset of serious pollution problems, a 
response to the attificial draining of the lake at the end of the last century, or a 
combination of the two. Certainly, these are likely to be the two most impOltant factors 
affecting the lake ecology over such a short period of time. 
Although collection of the modern fauna was not undertaken in a quantitative manner, 
certain interesting patterns do emerge which help to confirm what is known of the 
ecological preferences of certain species. For example, Dreissena was only collected 
from site 1 on the south-western shore, where it dominated the mollusc an fauna. This 
is a shallow-shelving beach subject to relatively high-energy conditions due to the 
prevailing winds. In fact, although not formally sampled, Dreissena could be found as 
dead shell accumulations at several other beach sites along this stretch of lake edge. 
Dreissena is known to occur in abundance in shallow, littoral environments that 
experience reasonably high-energy conditions. All of the other sampling sites 
provided, by comparison, relatively low-energy habitats. This is an impOltant 
ecological observation; a similar concentration of Dreissena in the fossil record can also 
be interpreted as representing a shallow, littoral environment. 
The mollusc an fauna from site 1 also included a wide range of other species, several of 
which are not regarded as littoral. This is unsurprising; prevailing winds will generate 
sufficient wave activity to carry most shells (whether derived from shallow or deep 
waters) and deposit them on the beach. 
Site 4 represents a low-energy environment with a muddy substrate and abundant 
aquatic vegetation. This is a habitat suited to such species as Lymnaea auricularia, L. 
stagnalis and Physa acuta, all of which at'e found here. Similat· environments at'e 
represented at site 7 (where L. stagnalis is found) and site 2 (where L. auricularia is 
found). At site 11, a steep-sided vegetation-free channel cut for irrigation purposes, 
only deeper water species such as Viviparus janinensis and Planorbarius corneus can be 
found. Observations of a similar nature can be made at other sampling sites. 
FAUNA AND BIOSTRATIGRAPHY. 124 
6.2.3 Fossil Aquatic Molluscan Fauna 
Molluscan faunal remains were preserved in core 284 down to a depth of about 145m. 
Throughout most of this interval, preservation was relatively poor, with only fragments 
and opercula being recovered. Nine species were identified: 
Viviparus janinensis 
Valvata piscinalis 
Bithynia graeca 
Lymnaea peregra 
Anisus vorticulus 
Gyraulus albus 
Gyraulus janinensis 
Dreissena cf. stankovici 
Pisidium nitidum 
Pisidium spp. 
The well-established quantitative method for carrying out mollusc an analysis (counting 
the number of apices from a set weight of sediment) was of little use here, mainly 
because of the small amount of sediment available from each level and the low faunal 
abundances. Of the species listed above, the most common were Valvata piscinalis, 
Viviparus janinensis, Bithynia graeca and Dreissena cf. stankovici; the others were only 
found sporadically. Fragments of these molluscs (and bithyniid opercula) were found 
intermittently throughout the top -145m of the core. Unfortunately, the counts of 
picked specimens were too low to permit a statistically valid assessment of the faunal 
structure. Nevertheless, the data still allow a few general observations to be made. 
In general, the occurrence together of the four main species listed above would suggest 
a shallow littoral environment, somewhat like sample site 1 today. Although Viviparus 
and Bithynia in particular prefer non-littoral habitats, their remains would still be found 
in high-energy littoral environments (much as they are today). In these cases, 
Dreissena would be the dominant species; the others would be found in association with 
it, but only in relatively small numbers. The presence of densely packed Dreissena 
beds in the core therefore suggests a true littoral, shallow water environment. This 
situation is seen at occasional discrete intervals in the core, illustrated in fig. 6.2 and 
described below. Elsewhere, Dreissena is either absent or a minor constituent of the 
fauna, suggesting a deeper water/lower energy environment. 
The first consistent record of mollusc fragments (dominantly Bithynia and Valvata) 
occurs at a depth of around 145m. They are seen occasionally in the core material until 
around 125m, at which point they disappear. Fragments again occur above about 85m, 
although initially these are rare. From 70m, they become more common until between 
approximately 45-42m several discrete Dreissena beds occur. Above these, the 
sediment returns to being only occasionally shelly until further Dreissena beds are seen 
FAUNA AND BIOSTRATIGRAPHY • 125 
0 
5 
10 
15 
20 
25 
30 
35 
40 
45 
50 
55 
60 
65 
70 § 
75 
..c: 
0. 
<1.l 80 Cl 
85 
90 
95 
100 
105 
110 
115 
120 
125 
130 
135 
140 
i45 
150 
0 10 20 30 40 50 s D 
Carbonate Content (%) 
Fig. 6.2 Relative lake-level inferred from the frequency of aquatic molluscs (solid/dashed line). S = 
shallower; D = deeper; .,!;. = location of Dreissena beds (see text for details). 
FAUNA AND BIOSTRATIGRAPHY • 126 
at approximately 32m and 26.3m. Occasional fragments are then seen until around 
18.5m, after which they become more common. Between 8.5-7.5m, the lithology 
becomes more clay-rich (see section 4.1.1) and contains many tiny fragments, of which 
most are bithyniid opercula. After 7.5m, the lithology becomes a shelly silt once again, 
with abundant fragments of Bithynia, Valvata, Dreissena and, towards the very top of 
the sequence, Gyraulus janinensis, Anisus vorticulus and Pisidium spp .. 
The occurrence of mollusc an fauna described above clearly has implications for 
interpreting such factors as long-term lake-level fluctuations over time. This is 
discussed further in section 7.3.1. 
The low counts and lack of detail in the fossil assemblages means that the molluscan 
fauna is unsatisfactory for any biostratigraphical purposes. 
6.2.4 Aquatic Molluscan Fauna from Core 256 
Although this study did not consider in any detail the molluscan fauna of core 256 (the 
marginal core drilled to the south-west of 284), a few pilot investigations of the lower 
section were initially undertaken. This lower part of 256 is thought to be older than any 
of 284, since all the sediments have a reversed palaeomagnetic signature (unpubl. data). 
In addition, this is the section of core that reaches bedrock and might therefore be Plio­
Pleistocene in age (the age of the lake is thought to be Plio-Pleistocene - see section 
2.3). Four aquatic species were recovered, all of which are known from Upper 
Pliocene lacustrine deposits sampled from the Katsikas site described above (Dollfus, 
1922; Aubouin, 1959; Gillet, 1963; Guernet et al., 1977). The species recovered from 
256 are listed in table 6.3, along with other taxa known from the Katsikas site. Without 
further detailed investigation, it is not yet possible to make a robust biostratigraphical 
correlation between these deposits and the lower patt of 256. 
Species 
Viviparus clathrata (DESHA YEs) 
Viviparus brusinai (NEUMA YR) 
Viviparus megarensis (FUCHS), 
Theodoxus micans (GAUDRY AND FISCHER) 
Theodoxus licherdopoli STEFANESCU 
Theodoxus quadrifasciatus (BIELZ) 
Pyrgula brusinai TOURNOUER 
Pyrgula eugeniae (BERNADI) 
Melanopsis costata (OLIVIER) 
Melania curvicosta DESHA YES 
Melania rhodensis BUKOWSKI 
Neumayria janinensis (DOLLFUS) 
Bithynia cf. meridionalis (FRAUENFELD) 
Ancylus sp. 
Valvata sp. 
DolIfus 
(1922) 
+ 
+ 
+ 
+ 
+ 
+ 
+ 
Aubouin GilIet Guernet et al. 
(1959) (1963) (1977) 
+ 
+ + 
+ 
+ + 
+ 
+ 
+ + +2 
+ 
+ + + 
+ + + 
+ 
+ + ?3 
+ 
+ 
+ 
Table 6.3 Comparison of aquatic molluscan fauna recovered from Upper Pliocene deposits at Katsikas with those recovered from core 256. 
, Viviparus megarensis may be a synonym of Viviparus brusinai. 
256 
(this study) 
+ 
+ 
+ 
+ 
+ 
2 Description of Pyrgula sp. by Guernet et al. (1977) matches that of Pyrgula brusinai. Their material must be exanllned before a critical assessment can be made. 
3 Guernet et al. (1977) refer to Hydrobia janinensis (DOLLFUS). It is possible that they actually mean Neumayria janinensis (DOLLFUS). Again, their material must be 
examined before a critical assessment can be made 
~ 
c::: ~ 
~ § 
ttl 
...... 
a 
~ 
~ 
~ 
:::l 
Cl 
:::>::l 
~ 
~ 
'""<: 
• 
...... 
N 
-.l 
FAUNA AND BIOSTRATlGRAPHY. 128 
6.3 OSTRACOD FAUNA 
The ostracod faunas from selected parts of the loaninna record were examined as part of 
this study. In particular, assemblages from the Eemian and last glacial - Holocene 
intervals were analysed, along with modem material collected during September, 1992. 
These are discussed in detail below. 
Whilst there exist many broad studies of both modem and fossil ostracods from Greece 
(e.g. Griffiths, 1995a,b), there are very few studies of the ostracod fauna from 
loannina itself. Aubouin (1959) records the presence of Candona sp. at the Katsikas 
site. Guernet et al. (1977) also identify several ostracods from this site: Darwinula 
stevensoni, Candona stupelji, llyocypris bradyi and Cyprinotus sp. (now known as 
Heterocypris sp., see note below) . They also formally describe a new sub-species of 
Cyprideis torosa, as C. torosa janinensis. 
The present study represents the first detailed account of Quaternary and modem 
ostracod faunas from loannina. 
6.3.1 Taxonomy 
As with molluscs, there are certain taxonomic problems inherent in working on 
ostracods from the Balkans. These include the high degree of endemism encountered 
with some genera (e.g. Candona, llyocypris, Leptocythere), as well as the difficulty in 
untangling the often unclear and complex stratigraphical schemes used by some of the 
Croatian, Slovenian, Yugoslavian and Greek authors (Yugoslavia is taken to include 
both Serbia and Montenegro). Fortunately, this task has been made considerably easier 
by Griffiths's (1995a) comprehensive database of European freshwater ostracods, in 
addition to his "biostratigraphic and palaeobiogeographic primer" (Griffiths, 1995b). 
His taxonomic conventions (also outlined in Griffiths and Evans, 1995) are used 
throughout this study. Before an account is given of the modem and fossil ostracod 
fauna found at loannina, special consideration needs to be given to the genus Candona, 
which has specific taxonomic problems. 
Candona s.l. BAIRD, 1845 is one of the broadest and most confusing of all European 
ostracod genera (Griffiths, 1995b). Griffiths (1995a) reports that although Limnofauna 
Europaea records over 150 extant species (LOffler and Danielopol, 1978), many more 
have since been described from a wide range of habitats (including subterranean and 
interstitial environments). Candona s.l. is a hugely speciose group, having a broad 
1 
FAUNA AND BIOSTRATIGRAPHY. 129 
geographical distribution across most continents. Subtle differences in shell 
morphology, polymorphism and, occasionally, sexual dimorphism, widespread 
endemism and poor descriptions have given rise to much taxonomic confusion both 
with extant forms and the vast number of fossil species known since the Tertiary. 
Candona s.l. has long been in need of a comprehensive taxonomic revision and several 
schemes have been proposed. Probably the most widely accepted of these was 
suggested by Danielopol (1978), in which Candona s.s. is nominated as one of six new 
genera. For a comprehensive discussion regarding the 'Candona problem', see 
Griffiths (1995a). 
Unfortunately, despite applying the nomenclature of Danielopol (1978), many 
taxonomic problems remain within Candona s. s.. Petkovski (l969a,b) suggested 
splitting the genus into two broad groups, the 'neglectoids' and the 'candidoids' . 
However, the number of endemics and extinct species (particularly amongst the 
neglectoids) means that this classification is of little use except as a first order 
taxonomic distinction. In the Balkans for example, highly diverse endemic faunas are 
known from Lake Ohrid (e.g. Mikulic, 1961; Petkovski, 1969a), Lake Prespa (e.g. 
Petkovski, 1960), Lake Skadar (Petkovski, 1961) and the Pannonian Basin (e.g. 
Krstic, 1988). In Greece, C. neglecta s.s. and associated neglectoids are recorded at 
several modem and Quaternary sites, such as the Megal6polis Basin (Hiltermann and 
Ltittig, 1969), Patras (Femandez-Gonzalez et al., 1994), Limni Lema (Zangger and 
Malz, 1989) and Ioannina (Guemet et al. , 1977). The neglectoid forms found at 
Ioannina as part of this study are discussed in more detail below. 
6.3.2 Systematic Review of the Ostracod Fauna from Ioannina 
A general description of the fourteen modem and fossil ostracod taxa found at Ioannina 
is now given. Their taxonomy is summarised, along with what is known of their 
occurrence, ecology and any other relevant details. As with the molluscs, the original 
citation is given for each species, followed by a synonymy that refers only to usage in 
relation to material from Ioannina. In considering distribution, emphasis is placed on 
European (and in particular, Balkan) occurrences. The ecological information is 
summarised in table 6.4. This is followed by separate consideration of the modem, last 
glacial - Holocene and Eemian assemblages and, where appropriate, an assessment 
made of their biostratigraphical and palaeoenvironmental importance. For the sake of 
clarity, distinction is made between those fossil species occurring in the Eemian, the last 
glacial- interglacial transition (LGIT, here taken to be the period from the height of the 
last glacial to the beginning of the Holocene) and the Holocene itself. 
Species 
DarwinuLa stevensoni 
(BRADY AND ROBERTSON, 1870) 
Cyprideis torosa 
(JONES, I 850) 
Leptocythere cf. ostrovskensis 
PETKOVSKl AND KEYSER, 1992 
ParaLimnocythere cf. compressa 
(BRADY AND NORMAN, 1889) 
Ilyocypris bradyi 
SARS, 1890 
Ilyocypris sLavonica 
SOKAC AND VAN HARTEN, 1978 
Candona cf. dedeLica 
PETKOVSKl, 1969 
Candona permanenta 
KRSTIC, 1985 
Candona cf. parvuLa 
MIKULlC, 196 I 
Cypria ophtaLmica 
JURlNE, 1820 
Cypridopsis vidua 
MULLER, 1776 
Eucypris virens 
(JURI NE, 1820) 
Heterocypris rotundatus 
(BRONSHTEIN, 1928) 
Prionocypris sp. 
Habitat 
A eurytopic and eurythermal species. Commonly found in permanent lakes and ponds, no deeper than 10m. Prefers soft, 
muddy, plant-rich substrates. 
A eurytopic and euryhaline species. Capable of tolerating a wide range of salinities. Can be found in lagoons, deltas, 
estuaries and oligosaline lakes . Prefers a muddy or sandy substrate. Found no deeper than about 30m. 
I 
Little ecological information. Likely to prefer habitats with silty substrates and abundant macrophytic vegetation. 
Little ecological information. Likely to be a cold water stenothermal species. 
A littoral species that prefers muddy substrates and abundant vegetation. 
Prefers muddy and silty substrates in still or moving freshwater with abundant macrophytic vegetation. 
Little ecological information. Likely to be eurytopic, eurythermal and euryhaline, preferring benthic, sublittoral habitats with 
muddy or sandy substrates that are rich in decaying organic matter. 
No ecological information. Likely to be as for C. cf. dedeLica (above). 
Little ecological information. A benthic, sublittoral species, it probably prefers muddy or sandy substrates that are rich in 
decaying organic matter. 
A hardy, eurytopic species found in permanent and temporary water bodies. An active swimmer, it is often associated with 
submerged vegetation, especially reed beds . 
Generally found in the littoral zone of lakes and ponds. Often associated with algal mats and aquatic vegetation. Intolerant of 
low oxygen levels . 
Found in ephemeral habitats, including seasonally-flooded reed beds on lake margins. Also associated with well-vegetated 
shallow down-stream areas. 
Little ecological information. Likely to be a hardy, euryhaline species. 
Little ecological information. Likely to be a littoral cold water species, preferring muddy substrates and abundant aquatic 
vegetation. 
Occurrence in 284 
M/H ILGH/E 
M 
H/LGH/E 
H 
M/H 
MI H/LGH 
M/H 
M/H/LGH/E 
LG/E 
M/H/LGH/E 
M/H 
H 
H 
LGH 
Table 6.4 List of ostracod fauna found during this study, with summarised ecological infonnation. M = modern; H = Holocene; LGH = Last glacial - Holocene; 
E = Eemian. 
~ 
c:::: 
~ 
~ 
~ 
O:l 
..... 
<:::) 
V:l 
;;5 
~ 
:::l 
Cl 
~ 
"'tl 
::r:: 
'"<: 
• 
VJ 
o 
FAUNA AND BIOSTRATIGRAPHY. 131 
Site 6 
Candona permanenta 
Cypria ophtalmica 
Site 1 
Cyprideis torosa 
/lyocypris slavonica 
Candona pemzanenta 
Site 10 
Darwinula stevensoni 
/lyocypris bradyi 
Candona dedelica 
Cypria ophtalmica 
Site 7 
Candona pemzanenta 
Cypria ophtalmica 
Cypridopsis vidua 
Site 2 
Candona pemzanenta 
~. <,> J",>, eyp,'a aphtalmica 
···:::t.\ 11 
." 
Lake Pamvotis 
Site 4 
'., 
Site 11 
Darwinula stevensoni 
Candona dedelica 
Candona pemzanenta 
Cypria ophtalmica 
Cypridopsis vidua 
Site 8 Darwinula stevensoni 
/lyocypris bradyi 
Candona permanenta 
Cypria ophtalmica 
Cypridopsis vidua 
Candona permal/el/ta 
Fig. 6.3 Sampling sites for modern ostracod fauna. See text for details . 
FAUNA AND BIOSTRATIGRAPHY • 132 
Class Ostracoda 
Sub-class Podocopa 
Superfamily Darwinuloidea 
Family Darwinulidae 
Genus Darwinula BRADY AND ROBERTSON, 1885 
Darwinula stevensoni (BRADY AND ROBERTSON, 1870) 
Plate VII, fig. F 
1870 Polycheles stevensoni BRADY AND ROBERTSON, Ann. Mag. N.H., Series 4, VI, p.25 , pl.YI, 
figs . 1-7 and pl.X, figs. 4-14. 
1977 Darwinula stevensoni (BRADY AND ROBERTSON, 1870). Guernet et al., Geobios, 10 (2), 
p.299 . 
Material examined: Modem valves and carapaces (see fig. 6.3 for sampling localities); 
fossil material from the Eemian, LGIT and Holocene intervals of 284. 
Occurrence: Darwinula stevensoni has a cosmopolitan distribution, being recorded 
from every continent except Australasia and Antarctica. In Europe, it is found south of 
latitude 600 N (Griffiths and Butlin, 1994). There are several Plio-Pleistocene records 
from the Balkans, including Greece (e.g. Hiltermann and Ltittig, 1969). D. stevensoni 
has also been reported from Upper Pliocene deposits at Ioannina (Guernet, et al., 
1977). 
Ecology: Generally, D. stevensoni is both eurytopic and eurythermal and is usually 
found no deeper than about lOm in non-desiccating habitats, such as lakes and ponds 
(McGregor, 1969). Regarded primarily as a littoral species, it is thought to prefer soft, 
muddy, plant-rich substrates. Temperature is not generally limiting, as demonstrated 
by latitudinal and depth distributions. A comprehensive biological and ecological 
review can be found in Griffiths and Butlin (1994). 
Superfamily Cytheroidea 
Family Cytherideidae 
Genus Cyprideis JONES, 1857 
Cyprideis torosa (JONES, 1850) 
Plate IV, figs. A-E 
FAUNA AND BIOSTRATIGRAPHY • 133 
1850 Candona torosa JONES , 1850, Ann. Mag. N.H. , Series 2, VI, p.27, p1.3 , figs . 6a-e. 
1977 Cyprideis torosa (lONES , 1850). Guernet et al., Geobios, 10 (2), p.301, pl.l, figs. 7-14. 
Material examined: Modern valves and carapaces from sampling site 1 (fig. 6.3). 
Occurrence: C. torosa is widespread throughout northern Europe, central Asia, central 
Africa and Mediterranean regions (Athersuch et ai., 1989). There are several Plio­
Pleistocene records from Greece (e.g. Mostafawi, 1986, 1994; Rommelt-Doll, 1990). 
Ecology: C. torosa is a eWhaline and eurytopic species that can tolerate salinities from 
almost freshwater to around 60%0. It can therefore be found inhabiting marginal marine 
environments such as lagoons, estuaries, fjords and deltas, as well as inland ponds and 
lakes. According to Athersuch et al. (1989), it prefers a muddy or sandy substrate and 
can be found at all depths down to about 30m. 
Remarks: The possible relationship between the carapace morphology of C. torosa and 
salinity levels has been much debated (e.g. Kilenyi, 1972; van Harten, 1975). A weak 
relationship between low salinity and the presence of tubercles on the valves has been 
suggested. However, some authors attribute the noding to other environmental 
parameters, such as the open or closed status of the water body (e.g. Bodergat et aI. , 
1991), whilst others explain it as an expression of genetic polymorphism (e.g. Kilenyi, 
1972). Clearly, further work is necessary to resolve the problem. 
Both noded and smooth forms of C. torosa were found in the modern sediment samples 
collected from Ioannina. Some carapaces were found with soft parts still intact; it is 
therefore clear that C. torosa is living in the lake today. No fossil specimens were 
found in the core material, indicating that its arrival and expansion in the lake has been 
very recent. It is possible that its success is related to pollution of the lake from 
agricultural and industrial waste during historical times, although few studies have been 
carried out on the effects that pollutants have on ostracod ecology. C. torosa dominated 
the fauna recovered from site 1, a beach to the north of the city, suggesting that it 
prefers higher-energy environments, with a sandy substrate. 
It is not clear whether the species of C. torosa occurring at Ioannina today is the same 
as the Upper Pliocene sub-species C. torosa janinensis described by Guernet et al. 
(1977). Comparison with the type material is necessary (the published description and 
plates illustrating this sub-species are of insufficient detail to permit a satisfactory 
assessment). 
FAUNA AND BIOSTRATIGRAPHY. 134 
Family Leptocytheridae 
Genus Leptocythere SARS, 1926 
Leptocythere cf. ostrovskensis PETKOVSKl AND KEYSER, 1992 
Plate VI, figs. A-C 
1992 Leptocythere ostrovskensis PETKOVSKI AND KEYSER, Mitt. hamb. zoo!. Mus. Inst., 89, 
p.228, pl.l, figs. 1-4. 
Material examined: Valves and carapaces from the Eemian, LGIT and Holocene 
intervals of 284. 
Occurrence: Extant freshwater leptocytherids are represented by only six species, all of 
them endemics from the Balkans. Three species were described from Lake Oill'id (Klie, 
1939), two from Lake Prespa (Petkovski, 1959) and one from Lake Vegoritis in NW 
Greece (Petkovski and Keyser, 1992). Some fossil records exist (also from the 
Balkans), for example from the Upper Pliocene of the Mega16polis basin in Greece 
(Hiltermann and Llittig, 1969). The form occurring at Ioannina is possibly an 
undescribed endemic species of leptocytherid having the closest affinities with L. 
ostrovskensis, originally described from Lake Vegoritis (Petkovski and Keyser, 1992). 
Examination of the type-material is clearly necessary to enable critical comparisons. 
Ecology: No ecological infOlmation has been published on any of the freshwater 
leptocytherids:' In the Ioannina record it is found with Candona permanenta, /lyocypris 
slavonica, and Prionocypris, suggesting that it prefers habitats that have silty substrates 
and abundant macrophytic vegetation. 
Family Lirnnocytheridae 
Genus Paralimnocythere CARBONNEL, 1965 
Paralimnocythere compressa (BRADY AND NORMAN, 1889) 
Plate VII, fig. C 
1889 Limnicythere inopinata var. compressa BRADY AND NORMAN, Sci. Trans. Roy. Dub!. Soc., 
Series 2,4, p.170, pI. XVII, figs. 18-19. 
Material examined: Rare valves from the Holocene of 284 . . 
FAUNA AND BIOSTRATIGRAPHY. 135 
Occurrence: P. compressa has a broad distribution across Europe in a wide variety of 
Quaternary deposits. It has not been collected alive since the nineteenth century 
(Martens, 1992). It has an extensive fossil record from the Balkans, being recorded 
from Croatia (e.g. Sokac, 1976) and Yugoslavia (e.g. Krstic, 1988). There are no 
other records from Greece. 
Ecology: The absence of modem material means that the ecology of P. compressa is 
unknown (Martens, 1992). Griffiths (l995a) suggests that it may be able to survive in 
seasonal habitats and is probably a cold water stenothermal species. 
Superfamily Cypridoidea 
Family Ilyocyprididae 
Genus /lyocypris BRADY AND NORMAN, 1889 
Ilyocypris bradyi SARS, 1890 
Plate IV, fig. F 
1890 /lyocypris bradyi SARS, Christiana Vid. Selsk. Forh., p.59. 
1977 /lyocypris bradyi SARS. Guernet et aI. , Geobios, 10 (2), p.301, pl.I , fig . IS. 
Material examined: Modem valves (see fig. 6.3 for sampling localities); fossil material 
from the Holocene interval of 284. 
Occurrence: Laffler and Danielopol (1978) record that l. bradyi is recorded throughout 
Europe, except in the northern Balkans and the Crimea. It is also known from North 
America (e.g. DelOlme, 1991). l. bradyi has a comprehensive Quaternary fossil record 
from across Europe (e.g. Griffiths, 1995a,b), including a Holocene record from 
southern Greece (Zangger and Malz, 1989). 
Ecology: Despite its broad distribution, there have been few detailed studies on the 
ecology of I. bradyi. It is thought to be associated with the littoral zone of lakes, as 
well as with springs. At Ioannina, l. bradyi was found in abundance at site 4 (fig. 
6.3), a shallow littoral area of the lake with a muddy substrate and abundant floating 
and submerged aquatic vegetation. 
Remarks: Many ilyocyprid species are polymorphic, which along with a high degree of 
endemism, can make identification of fossil material extemely difficult (modem 
identification is based on genital characteristics). 
FAUNA AND BIOSTRATIGRAPHY • 136 
Jlyocypris slavonica SOKAC AND VAN HARTEN, 1978 
Plate IV, figs. G-H 
1978 /lyocypris slavonica SOKAC AND VAN HARTEN, Geol. vjesnik, 30, p.219, pI. 1 and 2. 
Material examined: Modem valves (see fig. 6.3 for sampling localities); fossil material 
from the LGIT and Eemian intervals of 284. 
Occurrence: Until now, l. slavonica has only been known from the fossil record. It 
has been recorded from the Pleistocene of Croatia (e.g. Sokac, 1976; Sokac and van 
Harten, 1978), the Pleistocene of Yugoslavia (e.g. Krstic, 1985) and the Lower 
Pleistocene of Tegelen, in the Netherlands (Sokac and van Harten, 1978). This is the 
first recorded modern occurrence of the species. 
Ecology: Little is known of the ecology of l. slavonica. The Dutch and Croatian 
material (Sokac and van Harten, 1978) was found in sediments interpreted as being 
silty or muddy channel fills. Morphologically, 1. slavonica is most similar to Ilyocypris 
monstrifica (NORMAN, 1862), which occurs throughout north-western Europe. This is a 
warm-stenothermal flow-tolerant species that lives amongst submerged vegetation on 
muddy substrates. In the absence of other information, it would seem likely that I. 
slavonica has similar ecological preferences. 
Family Candonidae 
Genus Candona BAIRD, 1845 
Candona cf. dedelica PETKOVSKI, 1969 
Plate V, figs. E-F 
1969 Candona dedelica PETKOVSKI, Acta Musei Macedonici Scientiarum Naturalhun, 11 (5), p.85 , 
figs . 6-9 and 14-17. 
Material examined: Modem valves (see fig. 6.3 for sampling localities); fossil material 
from the Holocene interval of 284. 
Occurrence: First described by Petkovski (1969a) from the Lake Ohrid, C. dedelica 
has also been described from the fossil record (Middle Pleistocene?) as C. cf. dedelica 
from the Milijevo Valley, near Beograd in Serbia (Krstic, et al., 1981). There are no 
other records from Greece. 
FAUNA AND BIOSTRATlGRAPHY. 137 
Ecology: Not surprisingly, there is very little infOlmation on the ecology of C. 
dedelica. As a typical neglectoid (see above), it is likely to be epibenthic, moving 
around on and just within the sediment surface, feeding on decaying organic material 
(Griffiths, 1995a). It is probably eurytopic, able to withstand a broad range of 
temperatures and salinities. A few valves were found in the modem sediment samples 
from Ioannina, from sites 10 and 11 (fig. 6.3). Site 10, located south-east of Ioannina 
itself, was a shallow littoral site with a muddy substrate and abundant aquatic 
vegetation. Site 11 was a drainage channel at the eastem extent of the lake cut through 
the reed beds to a pumping station (for agricultural irrigation purposes). It had steep, 
muddy sides, some floating aquatic vegetation and the water had a distinct algal 
covenng. 
Candona permanenta KRSTIC, 1985 
Plate V, figs. A-D 
1985 Candona permanenta KRSTIC, Radovi Geoillstituta, 18, p.197, p\.1, figs. 1-3. 
Material examined: Modem valves and carapaces (see fig. 6.3 for sampling localities); 
fossil material from the Eemian, LGIT and Holocene intervals of 284. 
Occurrence: C. permanenta was first described by Krstic (1985) from Lower-Middle 
Pleistocene deposits at Vojvodina in the Pannonian Basin, Yugoslavia. It has since 
been recorded from other Lower and Middle Pleistocene deposits from the same basin 
(e.g. Krstic, 1993; Krstic and Schomikov, 1993). This is the first record from Greece 
and the only modem record of this species. 
Ecology: No ecological data exist for C. permanenta. As a typical neglectoid, the same 
ecological assumptions can be made as for C. dedelica. C. permanenta was found at 
Ioannina throughout core 284. Modem sediment samples (from almost all sampling 
sites) yielded carapaces with decaying soft parts, making it highly likely that C. 
permanenta lives in the lake today. 
Remarks: In their study of the Upper Pliocene fauna from the Katsikas site, Guemet et 
al. (1977) note the occurrence of Candona stupelji KRSTIC, 1974. First described from 
the village of Stupelj in Slovenia (Krstic, 1974), this form has remarkable 
morphological similarities with C. permanenta. It is possible that Guemet et al. (1977) 
FAUNA AND BIOSTRATIGRAPHY. 138 
rnis-identified their species, but this IS impossible to ascertain without critical 
comparison of material. 
Candona cf. parvula MIKULIC, 1961 
Plate V, figs . G-H 
1961 Candona parvula MIKULIC, Bulletin du Museum d'Histoire Naturelle de Belgrade, B17 , p .98 , 
pl.1 , fig . lO. 
Material examined: Fossil material from the Eemian and LGIT intervals of 284. 
Occurrence: C. parvula was first described as an endemic species from Lake Ohrid by 
Mikulic (1961). Morphological similarities suggest that the Ioannina form is 
synonymous with C. parvula, although as the published descriptions and illustrations 
are poor, this can only be critical assessed by comparison with the type material. 
Ecology: Little is known of the ecology of C. parvula. Stankovic (1960) reports that it 
is a sublittoral benthic species, with a very limited distribution in littoral and profundal 
habitats. As a neglectoid, it is assumed that it also requires the presence of decaying 
organic matter within the substrate, on which it feeds. 
Remarks: C. cf. parvula abruptly disappears from the record in core 284 at a depth of 
23 .99m (the significance of which is discussed below). 
Genus Cypria (ZENKER, 1854) 
Cypria ophtalmica JURINE, 1820 
Plate VII, fig. E 
1820 Monoculus ophtalmicus JURINE, Hist. Mon ., p.l78 , p1.19, figs. 16-17. 
Material examined: Modem valves and carapaces (see fig. 6.3 for sampling localities); 
fossil material from the Eemian, LGIT and Holocene intervals of 284. 
Occurrence: Sometimes also referred to as C. ophthalmica, this species is known from 
a wide variety of sites and is common in north-western Europe (e.g. Devoto, 1965; 
FAUNA AND BJOSTRATIGRAPHY. 139 
Henderson, 1990) and in the Balkans (e.g. Krstic, 1988). Griffiths (1995a) notes that 
its relatively uncommon occurrence in Quaternary deposits is probably due to its 
avoidance of flowing waters and to the fragility of the shell. 
Ecology: C. ophtalmica occupies a wide range of freshwater habitats and is often 
regarded either as a pioneer species or an indicator of stressed environmental 
conditions, when present in a restricted fauna with other hardy species (Griffiths, 
1995a). It is able to tolerate a wide range of variation in pH, temperature, anoxia and 
salinity (e.g. Delorme, 1991), is found in both permanent and temporary water bodies 
and is often associated with submerged vegetation (particularly reed beds). It is an 
active swimmer, but is also capable of burrowing into soft, muddy substrates (Griffiths 
and Martin, 1993). 
Family Cyprididae 
Genus Cypridopsis BRADY, 1867 
Cypridopsis vidua (MULLER, 1776) 
Plate VII, figs. A-B 
1776 Cypris vidua MULLER, Z. Dan. Prod., p.199 
Material examined: Modem valves and carapaces (see fig. 6.3 for sampling localities); 
fossil material from the Holocene interval of 284. 
Occurrence: C. vidua is a very common species with a wide distribution across Europe 
and the Balkans in a variety of Quaternary deposits (although curiously no record exists 
from Greece) . 
Ecology: Griffiths (1995a) notes that C. vidua has an apparent preference for the 
littoral zone of eutrophic lakes and ponds. An active swimmer, it is also capable of 
burrowing in soft substrates. C. vidua is often associated with algal mats and aquatic 
vegetation such as Potamogeton (Tabacchi and Matmonier, 1994). It is also known to 
be intolerant of low oxygen levels (Danielopol, 1991). 
FAUNA AND BIOSTRATIGRAPHY • 140 
Genus Eucypris (vAvRA,1891) 
Eucypris virens (JURINE, 1820) 
Plate VII, fig. D 
1820 Monoculus virens, JURINE, Hist. Mon., p.174, pU8, figs. 15-16. 
Material examined: Rare fossil material from the Holocene interval of 284. 
Occurrence: E. virens is found throughout most of Europe and has a comprehensive 
Quaternary record (e.g. Laffler and Danielopol, 1978; Griffiths, 1995a,b). The only 
known record from Greece is a Lower Pleistocene sub-species, E. virens latissima 
ALM, 1914, from the Mega16polis Basin (Hiltermann and Ltittig, 1969). 
Ecology: Griffiths (1995a) notes that E. virens tends to occur in ephemeral habitats, 
including seasonally-flooded Phragmites reed beds around lake margins and in well 
vegetated, shallow downstream regions of streams. 
Remarks: Although not recorded from the modem fauna at Ioannina as part of this 
study, it is possible that further detailed sampling (particularly in the extensive reed 
beds) would recover this species. 
Genus Heterocypris CLAUS, 1892 
Heterocypris rotundatus (BRONSHTEIN, 1928) 
Plate VI, fig. D-E 
1928 Cyprinotus rotundatus BRONSHTEIN, Rab. Sev.-Kavk. Gidrobiol. St., 2 (2-3), p.90, 112, p1.8 , 
figs. 16-21. 
Material examined: Fossil material from the Holocene interval of 284. 
Occurrence: The genus Heterocypris has a very broad distribution and is known from 
almost every continent. Many species are recorded from the Balkans and circum­
Mediterranean zone (e.g. Laffler and Danielopol, 1978). H. rotundatus is known from 
Eernian deposits in Germany (Fuhrmann and Pietrzeniuk, 1990) and from Plio­
Pleistocene deposits on the Greek island of Kos (Mostafawi, 1988). 
FAUNA AND BIOSTRATlGRAPHY. 141 
Ecology: The specific ecological preferences of H. rotundatus are not well-known, but 
Heterocypris sp. are generally considered to be hardy and capable of tolerating a broad 
range of salinities. 
Remarks: In their study of the Upper Pliocene fauna from the Katsikas site, Guemet et 
al. (1977) note the occurrence of Cyprinotus sp. (currently known as Heterocypris 
sp.). It is possible that the form collected by Guemet et al. (1977) is H. rotundatus, but 
this is impossible to ascertain without critical comparison of material. 
Genus Prionocypris BRADY AND NORMAN, 1896 
Prionocypris sp. 
Plate VI, figs . F-G 
Material examined: Fossil material from the LGIT interval of 284. 
Occurrence: This is generally a poorly-known genus, represented by two species, P. 
zenkeri (CHYZER AND TOTH, 1858) and P. serrata (NORMAN, 1862). The latter is the 
nominate species, since the true status of P. zenkeri remains uncertain (e.g. Danielopol 
and McKenzie, 1977; Griffiths, 1995a). While both occur across Europe in a range of 
Pleistocene and Holocene deposits (many of which require further confirmation given 
the uncertainty over the species), only P. zenkeri has been recorded from Greece 
(Hiltermann and LUttig, 1969). 
Ecology: Given the confusion over the taxonomy, ecological data are difficult to 
ascertain. In core 284, Prionocypris occufd with L. cf. ostrovskensis, 1. slavonica, 
D. stevensoni and C. permanenta, suggesting that it is likely to be a littoral, cold water 
species that has a preference for muddy substrates and aquatic vegetation. 
Other remarks: Although Prionocypris sp. only occurs at a single level in the core 
(23 .30m), it constitutes 20% of the ostracod fauna at this level. This is somewhat 
puzzling and is unlikely to be explained satisfactorily without further ecological 
information being available. 
FAUNA AND BIOSTRATIGRAPHY • 142 
6.3.3 Modern Ostracod Fauna 
During September, 1994, eight sediment samples were collected from the littoral zone 
of Lake Pamvotis (section 3.2.6). The sample site locations are shown in fig 6.3 and 
briefly described in appendix A. Although no live specimens were collected, many 
carapaces still contained the remnants of soft parts, indicating that the animals still live 
in the lake today. Due to the qualitative nature of the sampling technique employed, it is 
possible that the following list of modern species is incomplete. A total of nine species 
were recovered: 
Darwinula stevensoni 
Cyprideis torosa 
/lyocypris bradyi 
/lyocypris slavonica 
Candona cf. dedelica 
Candona permanenta 
Cypria ophtalmica 
Cypridopsis vidua 
Eucypris virens 
As previously described (section 4.2.2), modern Lake Pamvotis is a relatively shallow 
« lOm), freshwater, eutrophic water body with a range of clay, silt and sandy 
substrates. These factors are reflected by the modern ostracod fauna, most of which 
have ecological preferences for muddy or silty substrates and for the presence of 
abundant aquatic vegetation. In addition, all of the species present are able to tolerate 
fresh - oligosaline conditions. 
The range of microhabitats within the lake is also reflected in part by the ostracod fauna 
found at those sites. For example, sites 4 and 7 are low-energy, shallow littoral areas 
with dense macrophytic vegetation (see fig. 6.3 and appendix A). The fauna recovered 
from these sites include Cypria ophtalmica and Cypridopsis vidua, both of which 
favour littoral areas with muddy substrates and macrophytic vegetation. Sites 2, 6 and 
11 are located along the north-eastern shore, where there are abundant reed beds. 
Cypria ophtalmica is also often associated with reed beds and is a major constituent of 
the fauna at these sites. 
Site 1 (a shallow-dipping beach north of Ioannina city) is interesting, as prevailing 
winds ensure that it experiences relatively high-energy conditions compared with other 
sampling locations. The fauna at site 1 is dominated by Cyprideis torosa, a hardy, 
eurytopic and euryhaline species which represents 82% of the ostracod fauna at this 
location. The valves are robust and well calcified, enabling it to tolerate a higher energy 
environment than other, more delicate genera. It occurs with /lyocypris slavonica 
(thought to be able to tolerate moving water) and Candona pennanenta, a eurytopic 
FAUNA AND BIOSTRATIGRAPHY • 143 
species that generally dominates both the modern and fossil fauna. It is interesting that 
Cyprideis torosa is found only at this site. One likely explanation is that although it is a 
eurytopic species, all the other sampling sites experience considerably lower energy 
conditions, which may be unsuitable for it. Other species might therefore have a 
competitive advantage in these low-energy, vegetation-rich environments with muddy 
rather than silty or sandy substrates. 
6.3.4 Taphonomy 
Before discussing the fossil ostracod faunas, some aspects of their taphonomy must 
first be considered. The fossil Candona assemblages all showed a distinct bias in the 
preservation of growth stages. In particular, there were relatively few adult valves and 
carapaces recovered relative to other instal's. A histogram plotting the absolute number 
of picked instars of C. permanenta recovered from the last glacial - Holocene section of 
the core is shown in fig. 6.4 (n = 60). It shows a definite skew towards the A-4 / A-5 
instal's, with the number of adult / A-I valves recovered being the smallest group. This 
graph bears a remarkable resemblance to the 'type-C' distribution pattern described by 
Whatley (1983), which he interprets as being characteristic of a low-energy death 
assemblage. He suggests that the low number of larger valves is due to some degree of 
weak current action 'winnowing' out the heavier valves. 
If weak current activity was the cause of this preservational bias in the candonids (the 
largest ostracods present), then a more balanced distribution of instal's of the smallest 
species might be expected, since the current velocity would presumably be sufficient to 
carry all or most of their growth stages. In fact, most of the smaller species (e.g . 
Cypria ophtaimica, Leptocythere cf. ostrovskensis) also show a preservational bias, 
although this time it is almost entirely skewed towards the adult valves; the smallest 
instars are hardly represented. This is most probably an artifact caused by the smallest 
size of sieve used in preparing the sediments (l50llm), which may have lost the smaller 
instal's of these particular species. Clearly, a reappraisal of the instar distribution of the 
smallest species would require resieving the sediment samples using finer-mesh sieves. 
It is not possible to compare these results with modern instar distributions, as the 
original sampling method was not rigorously quantitative. In addition, many of the 
higher-energy sample sites (e.g. the beach at site 1) would already possess a skewed 
distribution pattern. Further sampling, preferably of lake-bottom sediments across the 
entire lake, would be useful in determining not only instar distribution, but also benthic 
community distributions. 
FAUNA AND BIOSTRATIGRAPHY. 144 
2000~--------------------------------~ 
1500 
"0 
<.) 
..>:: 
u 
.s.. 
V> 
<.) 
;> 
C;; 
;> 1000 
"--0 
0 
I':: 
<.) 
S 
'0 
V> 
.0 
--< 500 
0 
C<) If) ... 
-< -< -< 
<.) 
:;a 
--
8 ~ N "<t 
"5 
-< -< 
V> 
"0 "0 
--< I':: 
'" \0 
-< 
Fig. 6.4 Instar distribution of Candona permanenta picked from the last glacial - Holocene interval 
of core 284. 
The evidence for current activity therefore suggests that the ostracod fauna in core 284 
is representative of a death assemblage. This means that the association of species 
found in the fossil record might not necessarily reflect the association of these species 
when they were alive. This obviously has serious implications for the interpretation of 
the faunal counts in the core. 
In general, the recovery of specimens has been poor; counts have consistently been 
below the 350 valves required from each level to make a statistically robust assessment 
of faunal composition. In addition, the ecological information available for each species 
is not particularly precise or detailed, limiting palaeoenvironmental inferences to very 
general statements. It is therefore unlikely that the added complication of what is likely 
to have been a relatively small degree of transport will have a significant effect on the 
final interpretations. Only if the faunal counts had been consistently high would 
potentially detailed interpretations be affected by this problem. Although all of the 
above points must be borne in mind when analysing the fossil data, a useful first order 
FAUNA AND BIOSTRATIGRAPHY. 145 
assessment of faunal structure and succession through the selected parts of the core can 
still be achieved. 
6.3.5 Fossil Ostracod Fauna: Last Glacial - Holocene 
The section of core 284 that includes the last glacial - Holocene (2S-1.4m) was sampled 
for ostracods at 20cm intervals, as described in section 3.2.7. A total of 12 species 
were recovered: 
Darwinula stevensoni 
Leptocythere cl ostrovskensis 
Paralimnocythere cf. compressa 
/lyocypris bradyi 
/lyocypris slavonica 
Candona cf. dedelica 
Candona permanenta 
Cypria ophtalmica 
Cypridopsis vidua 
Eucypris virens 
Heterocypris rotundatus 
Prionocypris sp. 
The results of the faunal counts are plotted in fig. 6.S, presented as valves per gramme 
of dry sediment against depth. The plotting package used, psimpoll 2.2S (written in C 
by K.D. Bennett, Department of Plant Sciences, University of Cambridge), was 
originally developed for presenting pollen data. However, many of the features (such 
as its ability to statistically calculate a zonation scheme) can equally be applied to faunal 
data. 
Five distinct faunal zones are recognised. After an initial description of each zone, the 
data are discussed. 
10-1 2S.00-24.0Sm 
V ruiable values of Candona permanenta, occasional Darrvinula stevensoni and 
Leptocythere cf. ostrovskensis. End of zone marked by abrupt disappearance of 
Candona cf. parvula. 
10-2 24.0S-16.70m 
Highly vru'iable and oscillating values of C. permanenta. Appearance of /lyocypris 
slavonica. Single appearance (at 23.30m) of Prionocypris sp.. Variable values of L. 
cf. ostrovskensis and D. stevensoni, although the former . is more common at the 
beginning of the zone, the latter more common towards the end. Rare OCCUlTence of 
Cypria ophtalmica, particularly at the end of the zone. Between approximately 23.0-
Ioannina 284 Last Glacial - Holocene O'tracod Fauna 
Depth 
m 
4 
10 
11 
12 
13 
14 
15 
16 
17 
IS 
19 
20 
21 
22 
23 
24 
25 
~,~ 
tot /' 
s'~ ;.;.'~ 
o,S j,.' ~ )..i' 
, ~ ~ ~ F . 
<Q<l.'I 
Ij' ~i~~ 
~o~ 
&.<I.~ 
~,r:..r> 
,,~,o 
~,r, f) 
0"'<1. ~..,o 
·ss<l.· 
cif 
0\' 0 ",~ ~~l ,~' .i~ ,0'''' ~~..,' 
", s' ~~ &.~ \s <,s '<J' 
i$' . ~~ 0<1. to~ 0<1.' cI<I. <,s .,/' 
.,t 
cI~ 
'1l ~~ 
r"''O 
-.1''"' 
® 169S:l:45 
C?~ 
~o~~ 
G~~ 
toO ~;,o ~ / ~ ~ ~ , ~ ~ ~ & if & # ~ 
~,,, 
<,/0"'<1. ~' ~~' 
g o __________ __ ____ ______________________ __ _______ _  -- - -- - - ---- --------t---------t- -------- 1 ---- -.-. -~- ------- -f----- ---- [o-- - -----I------. __ ~ __ _______ '" ________ ... _____ ___ _ 
® 2465:1:45 
® 5370±50 
® 7505:1:60 
® l00S0±70 
. "'''''''' p mmmummummmumu 
® 2076O±230 [ n _______ n _ _ n ___ nnnn __ ____ • 
~ ~~~~~~~~~~~ 
o 10 15 20 o o o o o o o o o o o o 
Values expressed as valves per g 
Fig. 6.5 Ostracod fauna from the last glacial- Holocene interval of core 284. See text for zonation scheme details. 
~ 
,se; 
~l 
00 &. 
10-5 
10-4 
10-3 
10-2 
10-1 
~ 
c:::: ~ 
~ 
~ 
O:l 
...... 
C) 
Cl:> 
;;5 
~ 
::j 
c;J 
~ 
~ 
'"tl 
::r: 
"<: 
• 
-.j:>. 
0\ 
FAUNA AND BIOSTRATlGRAPHY. 147 
22.0m, all taxa experience a decline followed by a recovery, highlighted particularly in 
Candona permanenta and 1. slavonica. 
10-3 16.70-1S.2Sm 
Sharp decline of Candona permanenta to very low values at the beginning of the zone, 
recovering slightly by the end of the zone. The only other species represented is a 
single level with D. stevensoni. 
10-4 lS .2S-S.S0m 
Lengthy zone featuring oscillations in the values of C. pelmanenta, with a distinct peak 
at approximately lOm and a noticeable trough between approximately 7.S-6m. Other 
species present (sproadically) are L cf. ostrovskensis, D. stevensoni and Cypria 
ophtalmica. 
10-5 S.SO-1.30m 
Final zone, commencing with a very sharp peak in Candona permanenta, which then 
declines to a low level again. Cypria ophtalmica increases in value through the zone. 
Candona cf. dedelica and D. stevensoni are present sporadically. Several species 
appear at the end of the zone: Cypridopsis vidua, Heterocypris rotundatus, Jlyocypris 
bradyi, Eucypris virens and Paralimnocythere cf. compressa. 
With the limited, ecological information available for these species, even first order 
interpretation of the faunal succession is difficult. That there is a definite succession is 
not in doubt; different species disappear and appear over the interval and there is a 
definite suggestion of higher order variability in the data. The dominance of the fauna 
throughout by C. permanenta would indicate that the site was probably situated in a 
sublittoral, possibly profundal location for much of the time. The low abundances of 
the littoral species, coupled with the preservational bias of the growth stages (discussed 
in section 6.3.4, above) also suggests that the site was in relatively deeper, rather than 
shallower, water depths. This confirms the lithological and strati graphical information 
obtained by IGME (section 2.3) and their interpretation that core 284 was taken from 
the deepest of the sub-basins. The main points of each of the faunal zones (fig. 6.S) are 
now discussed in turn. 
The end of zone 10-1 witnesses the abrupt disappearance of Candona cf. parvula, 
coinciding with what has been interpreted as the end of the LGM (section 7.S.3). Since 
this species is present below this point into at least into MIS-6 (see section 6.3.6, 
below), its disappearance at the end of the LGM is a valuable biostratigraphical marker. 
The cause of this extinction event is not clear, since very little is known of its ecological 
FAUNA AND BIOSTRATIGRAPHY. 148 
tolerances. As it is also present in the Eemian, it is obviously a reasonably eurythermal 
species. However, it is possible that conditions became too severe at the height of the 
last glacial and either the temperature became too low, or reduced lake levels (see 
discussion of Dreissena beds, section 6.2.3) ensured that its prefelTed sublittoral 
benthic habitat became restricted. 
Despite the oscillations of C. permanenta throughout zone 10-2, five other species show 
a reasonably clear succession. At the beginning, l. slavonica and Prionocypris sp. 
appear, only to be replaced by greater dominance of L. cf. ostrovskensis and D. 
stevensoni. With the exception of a very brief appearance in the middle, Cypria 
ophtalmica only appears at the very end of the zone. Although this part of the record 
has been interpreted as including the Lateglacial interstadial (chapter 5), lack of precise 
ecological information makes a detailed interpretation difficult. However, in the 
broadest sense, the succession could represent the recovery and deepening of the lake 
following the LGM, with littoral species such as I. slavonica giving way to slightly 
deeper water species such as L. cf. ostrovskensis and D. stevensoni. 
10-3 witnesses a sharp decline in C. permanenta, with a minimumin the middle of the 
zone, at around 16m. A possible interpretation for this part of the record is that it 
reflects the Younger Dryas (section 5.8), thought in the eastern Mediterranean to be an 
arid, cold phase (e.g. Tzedakis, 1993; Rossignol-Strick, 1995). Productivity is likely 
to have been low during this time, reflected by the decline in the population of C. 
permanenta. It is not possible from these data to determine whether this was 
accompanied by a lake-level fall. 
10-4 sees a recovery by C. permanenta, although it oscillates markedly throughout this 
zone and experiences a pronounced crash between -7.5-6m. This minima coincides 
with a distinct litho logical change to a shelly calcareous silt, which could well reflect a 
shallowing of the lake. Other species are present only sporadically throughout the 
zone. This part of the record has been interpreted as the first part of the Holocene 
(section 5.8). 
10-5 begins with a marked increase in C. pennanenta, which then declines towards the 
end of the zone. Cypria ophtalmica also experiences a rise at the beginning of the zone 
and gains in importance. As a hardy eurytopic littoral species, C. ophtalmica is often 
successful in stressed environments. This zone coincides with another change in 
lithology (to a shelly, clayey silt), which could be a reflection of the continuing 
shallowing of the lake. The occurrence of other littoral species such as Eucypris virens 
and Cypridopsis vidua towards the end of the zone, supports this interpretation. 
FAUNA AND BIOSTRATlGRAPHY. 149 
Interestingly there is no record of Cyprideis torosa, another hardy eurytopic species 
which currently dominates the ostracod fauna at site 1 (a high-energy shallow beach). 
It is possible that despite the shall owing of the lake, the site of 284 never experienced 
sufficiently high-energy conditions. It is also possible that it has only been able to 
thrive in the lake in recent historical times, perhaps due to a change in nutrient status 
caused by the effects of pollution. 
In summary, therefore, it seems that the ostracod fauna through this part of the core 
records the recovery of the lake following the LGM (and the return of the core site to 
deeper waters), followed by a gradual shallowing of the lake during the Holocene. 
This interpretation is discussed further in section 7.5.3, when it is compared with the 
palaeoenvironmental information provided by the mollusc an fauna. 
6.3.6 Fossil Ostracod Fauna: Eemian 
The section of core 284 that includes the Eemian (l00-83 .2m) was sampled for 
ostracods every 20cm, as described in section 3.2.7. A total of 6 species were 
recovered: 
Darwinula stevensoni 
Leptocythere cf. ostrovskensis 
Jlyocypris slavonica 
Candona permanenta 
Candona cf. parvula 
Cypria ophtalmica 
In general, the fauna was relatively sparse and poorly preserved compared with that 
from the last glacial - Holocene. The results of the faunal counts are again presented as 
valves per gramme of dry sediment against depth (fig. 6.6). Eight distinct faunal zones 
are recognised. After an initial description of each zone, the data are discussed. 
le-l 100.0-98.30m 
Very sparse zone, with only occasional valves of Candona pennanenta or Leptocythere 
cf. ostrovskensis. 
le-2 98.30-96.10m 
Increase in value of C. permanenta at the beginning of the zone, followed by a decline 
in values (from -4 valves/g to -0.2 valves/g) at around 97m, followed by an increase in 
values again at the end of the zone. C. cf. parvula and L. cf. ostrovskensis are both 
recorded at the very beginning of the zone. 
Ioannina 284 Eemian Ostracod Fauna 
Depth 
m 
84 
85 
86 
87 
88 
89 
90 
91 
92 
93 
94 
95 
96 
97 
98 
99 
lOO 
\~ 
,<:-
~<:' 
<,,' ~' ,~ 
~o' 
cl' 
~ 
0 
~o~ 
cl' 
.3>~ 
~<' (;.,.~ 
I I 
2 4 6 0 2 
Values expressed as valves per g 
.3>~ 
,j;$" 
\)~ 
i:-' <:-~o 
,< 
~\' 
,.i' 
,0Cl' 
V~ 
(;.,. 
~,~ 
.~ 
,,0 
o~ .~~ (;-~' 
~ 0"-~ ... o-
C;1~ 
~~~ 
o o o 
Fig. 6.6 Ostracod fauna from the Eemian interval of core 284. See text for zonation scheme details . 
10-8 
10-7 
10-5 
10-4 
10-3 
10-2 
10-1 
~ 
~ 
c::: 
s;: 
~ 
~ 
tl:i 
...... 
a 
V:l 
;;5 
~ 
:::l 
CJ 
~ 
"tI 
::t: 
""<: 
• 
Vl 
o 
FAUNA AND BIOSTRATIGRAPHY • 151 
lea 3 96.1O-90.10m 
Lengthy zone with a velY poor fauna. Low values of C. permanenta and a record of C. 
cf. parvula from a single level. 
le-4 90.1O-87.70m 
Sudden rise in C. permanenta at the beginning of the zone. Present at relatively high 
values throughout. Values of C. cf. parvula and L. cf. ostrovskensis increase towards 
the end of the zone. 
le-S 87.70-84.90m 
Oscillating values of C. permanenta. C. cf. parvula peaks at beginning and end of the 
zone; absent in the middle. Oscillations in values of L. cf. ostrovskensis, broadly 
matching those in C. permanenta. Single occurrence of Cypria ophtalmica at the 
beginning of the zone and of Darwinula stevensoni at the end of the zone. 
le-6 84.90-84.70m 
Zone spanning a single, particularly rich sample at 84.80m. VelY high values of 
Candona permanenta and C. cf. parvula. L. cf. ostrovskensis also recorded. 
le-7 84.70-84.10m 
Very low values of C. pennanenta at beginning of the zone, increasing towards the end 
of the zone. Low values of L. cf. ostrovskensis at the beginning and the end of the 
zone. C. cf. parvula and D. stevensoni both occur towards the end of the zone. 
le-8 84.1O-83 .20m 
Low values of C. permanenta, C. cf. parvula and L. cf. ostrovskensis. 
Because of the relative sparseness of the fauna, interpretations are even more difficult 
than for the last glacial- Holocene interval. In general, the lack of littoral species and 
the dominance of the fauna once again by Candona pennanenta would suggest that the 
site was located in the sub-littoral or profundal zone of the lake. The interpretations 
suggested below are later compared with other proxies through this interval (section 
7.5.1), in particular the pollen record. 
The first zone, Ie-1 , contains a very sparse fauna indeed, which makes any 
palaeoenvironmental interpretation almost impossible. This part of the record has been 
interpreted as representing the velY end of the penultimate glaciation (end of MIS-6). It 
is possible that the glacial conditions meant that productivity in the lake was low, 
accounting for the lack of material. 
FAUNA AND BIOSTRATIGRAPHY • 152 
C. permanenta rises steeply during zone le-2, although frequencies fall at -96.80m but 
recover again soon after. This part of the record is thought to represent the beginning 
of the Eemian (section 7.5.1). The carbonate and ostracod (5 180 records both oscillate 
sharply at the beginning of the Eemian, described as an abrupt 'Younger Dryas-type' 
cooling event before the full interglacial commences. The minima in C. permanenta 
coincides with this abrupt oscillation, probably representing a drop in productivity 
during this cooler period. 
le-3 is a lengthy zone, with only low values of the two Candona taxa represented. This 
part of the record has been interpreted as representing the first part of the Eemian, 
thought to be the warmest part of the interglacial (e.g. Tzedakis, 1991). The absence of 
ostracod fauna is puzzling, since warm and wet conditions should have promoted 
increased aquatic productivity. It is possible that the lake was relatively deep during 
this time (i.e. too deep for the Candona fauna), since it is known from studies in Lake 
Ohrid (e.g. Stankovic, 1960) that C. parvula prefers epibenthic rather than true benthic 
habitats. This may also be true of C. permanenta, although once again, the lack of 
precise ecological information for these species makes any interpretation very difficult. 
The beginning of le-4 sees a resurgence of C. pennanenta and the appearance of L. cf. 
ostrovskensis towards the end of the zone. If the assumptions concerning the 
ecological preferences of C. pennanenta made above ar'e correct, then this zone might 
represent a loweri~ig of the lake level (reduced temperatures and/or precipitation?) to 
return the site of 284 to a sub-littoral/profundal habitat once more. 
Zone le-5 sees a slight lowering in the average values of C. permanenta, but there is a 
marked rise in L. cf. ostrovskensis and C. parvula. The frequencies of all three taxa 
fall at -86m. These oscillations could again be due to fluctuations in lake level or 
productivity. The minima coincides with drops in the carbonate, 8180 and 813C records. 
This might represent a brief cool period within the main interglacial, during which 
productivity levels were reduced. 
le-6 is a narrow zone consisting of one particularly rich sample, with a very abundant 
Candona fauna. It is likely that this represents a brief time of optimal sublittoral 
conditions with plenty of decaying organic matter on which to feed. 
le-7 is a return to similar conditions seen in le-5, with moderate levels of candonids and 
a few other taxa poorly represented. 
FAUNA AND BIOSTRATIGRAPHY • 153 
le-8 sees a return to the situation seen in le-I, with low Candona counts and the 
occasional L. cf. ostrovskensis. This part of the record has been interpreted as the end 
of the Eemian and the beginning of the stadial MIS-5d. The sudden resumption of 
much cooler and drier conditions can also be clearly seen in the carbonate and isotopic 
records (section 7.5.1). 
In summary, therefore, it seems that the ostracod fauna through this pmt of the core 
records the behaviour of the lake through the Eemian. Starting with a sharp and 
oscillatory inferred rise in lake level, relatively deep water levels were maintained until 
the beginning of zone le-4, when the beginning of a slight drop is inferred. For the rest 
of the Eemian, levels appear to have fluctuated until the end of zone le-7, when there 
was a return to the lower lake levels seen at the very beginning of the interval. This 
interpretation is discussed further in section 7.5.1, when it is compm·ed with the 
palaeoenvironmental information provided by other proxies. 
6.3.7 Other Intervals of 284 
This study has only considered in detail the modern, last glacial - Holocene and Eemian 
ostracod faunas, representing only -16% of the total core length. Time did not permit a 
comprehensive examination of other parts of the record, but several pilot levels were 
analysed from each of the intervals interpreted as glacials and interglacials (chapter 5) . 
Although unsatisfactory from a both a biostratigraphical and a faunal stmcture point of 
view, it is still of use in estimating the overall range of some of the species involved. 
For example, Candona permanenta occurs throughout the entire core, whereas C. cf. 
parvula only appears in the record between 170-140m (i.e. during MIS-7). Other 
species, such as /lyocypris slavonica and Leptocythere cf. ostrovskensis, occur 
sporadically and in very low abundances throughout the core. However, the main 
observation is that the ostracod fauna is dominated throughout by the candonids and in 
some cases, such as the sample analysed from 260.74m, the abundances are very high. 
This would suggest that the core site was chiefly located in a sublittoral or even 
profundal setting for much of the time, confirming previous assumptions based on 
lithology and stratigraphy (section 2.3). Clearly, the potential for fmther high 
resolution faunal analyses of this material is considerable. 
FAUNA AND BIOSTRATIGRAPHY. 154 
6.4 SUMMARY 
This chapter has discussed the analysis of the modern, last glacial - Holocene and 
Eemian mollusc an and ostracod faunas from Ioannina. The taxonomy, occurrence and 
ecology of twenty-one aquatic mollusc and fourteen ostracod taxa were reviewed and 
their relevance to the Ioannina sequence discussed. Some of the problems associated 
with endemic taxa were also outlined. This represents the first study of Quaternary 
mollusc an faunas and the only comprehensive study of the ostracod faunas (both 
modern and fossil) from Lake Pamvotis. It also represents the first time that poorly­
known endemic faunas from the lake have been properly described and illustrated. 
Examination of both the molluscs and ostracods in 284 provided only low faunal 
counts, meaning that the usual statistically derived analysis methods could not be used. 
However, it was still possible to derive useful palaeoenvironmental information. For 
example, the presence of Dreissena beds in the core material helped to pinpoint periods 
of low lake level (palticularly at the LGM). The succession of various ostracod taxa 
were also used to determine lake level variation. 
In assessing the biostratigraphical importance of the faunas in the core, it was shown 
that the low frequencies of the molluscs in particular precludes their use for these 
purposes. Although a similar situation existed for the ostracods, the appal'ently ablUpt 
disappearance of Candona cf. parvula at a level in the core interpreted as being the LGM 
may constitute a useful biostratigraphical marker. Clearly, there is a need for this to be 
replicated in other cores before it can be generally applied. The faunas from the core al'e 
more useful in helping to interpret palaeoenvironmental conditions than as 
biostratigraphical tools. 
7. INTERPRETATION AND SYNTHESIS OF 
RESULrrS 
7.1 INTRODUCTION 
In prevIOus chapters, results have been presented individually, without detailed 
discussion. In this chapter, the interpretation of those results is considered in full and 
their stratigraphical importance assessed using the age model developed in chapter 5 . 
Where appropriate, comparisons are made with other data obtained as part of this study. 
A synthesis of all the results is then presented in order to derive a comprehensive 
palaeoenvironmental history of the lake basin over the last 600ka, with special attention 
given to those intervals of the record that were singled out for detailed analysis. The 
broader significance of the Ioannina 284 sequence is discussed in chapter 8. 
7.2 SEDIMENT ANALYSES 
The physical and chemical composItion of the core material has been extensively 
analysed. The se quence is thought to be continous because of the lack of sharp 
lithological breaks or sedimentological indicators of hiatus or lake emergence (e.g . 
evaporites, soil or peat horizons, oxidised layers) . The dominantly clayey silt lithology 
with relatively low «30%) carbonate levels is typical of profundal sediments from 
eutrophic mid-latitude temperate hard-water lakes (Dean, 1981). 
In section 3.3.1 the four main constituents of sediments from lakes such as Pamvotis 
were listed, namely detrital clastic material, biogenic silica, carbonate minerals and 
organic matter. SEM analysis demonstrated the low allogenic input to the borehole site 
from detrital material, although the possibility of limited aeolian input of matelial 
derived from the deserts of North Africa was considered (section 2.3). SEM analysis 
also confilmed the high level of biogenic silica present in the sediment (largely 
consisting of diatomaceous material and a siliceous cement binding individual grains 
together). Jones and Bowser (1978) consider organically-derived silica (i.e. due to 
diatoms) to be the most significant authigenic (endogenic sensu stricto) contributor of 
silica in lake systems. Regarding the final two sediment constituents, it has often been 
suggested that organic matter and/or carbonate levels can be used as proxies for lake 
INTERPRETATIONS AND SYNTHESIS. 156 
productivity. To assess the validity of this statement (and to interpret the curves 
illustrated in figs. 4.4 and 4.5), the origins of each are now discussed in some detail. 
7.2.1 Carbonate Content 
Carbonate minerals within lake sediments can be derived from a variety of sources. 
Kelts and Hsii (1978) recognise four main processes: they can be allogenic in origin, 
having been eroded from the catchment; they can be bioc1astic, consisting of debris 
from calcareous plants and animals; they can be produced by primary inorganic 
precipitation (induced by either physical or biogenic factors); or they can be derived 
from early diagenetic reactions. Dean (1981) notes that most carbonate in temperate 
hard-water lake sediments is either inorganically precipitated or biogenically-induced. 
Greater seasonal temperature variations in the littoral zone of a lake ensure that the rate 
of inorganic carbonate precipitation is higher than in profundal regions. Detailed 
discussion of the chemistry governing the precipitation of carbonates in lakes is beyond 
the scope of this study, but Wetzel (1975), Kelts and Hsii (1978) and Dean (1981) all 
give comprehensive reviews. Essentially, assimilation of CO2 from the water by the 
photosynthetic withdrawal of CO2 by abundant algae and aquatic macrophytes increases 
the pH, resulting in higher rates of biogenic ally-induced carbonate formation 
(demonstrated, for example, by calcareous enclUstation of plants). The proportion of 
detrital (allogenic) material and of bioc1astic carbonate (debris from ostracods, 
molluscs, charophytes and the like) is also significant. Littoral zone sediments are 
therefore generally characterised by a higher total percentage of carbonate than those 
derived from the profundal zone. 
By comparison, carbonate production in the profundal zone is generally dominated by 
inorganic and, in particular, biogenically-induced precipitation (largely due to the 
photosynthetic activity of phytoplankton). Bioc1astic carbonate is much less important 
and detrital contributions are also usually minor, since most allogenic material remains 
trapped in the littoral zone (Dean, 1981). However, detrital material can sometimes be 
deposited in the profundal zone, particularly if it is fine-grained and is canied as 
suspended load by inflowing rivers (Kelts and Hsii, 1978). When this is the case, 
allogenic carbonates can be almost indistinguishable from primary carbonate material. 
Another possibility is for littoral material to be transported to deeper levels by turbidity 
CUlTents triggered either by the excess load carried by flooding rivers (Kelts and Hsii, 
1978), or by earthquakes. Here again, distinguishing between allogenic and primary 
carbonates can be almost impossible. Neither of these possibilities are a major problem 
INTERPRETATIONS AND SYNTHESIS • 157 
at Ioannina. The core site was apparently profundal for most of its history (section 2.3) 
and the lake has no fluvial input, limiting the possibility of suspended material reaching 
the centre. Although Epirus is a tectonic ally active region, turbidity currents triggered 
by flood events can also be largely ruled out, as they usually operate only where the 
lake is moderately or steep-sided (e.g. Ludlam, 1974). This is not the case at Ioannina, 
where gradients are gentle. Turbidity current deposits can also be recognised by 
distinctive bedforms (e.g. graded bedding), differences in colour to the normal 
sediment and the presence of large quantities of (usually fragmented) littoral bioc1astic 
and organic material. These features are absent in core 284. Detrital carbonate material 
is therefore expected to be an extremely minor component of the core sediments, an 
assumption borne out by SEM analysis. 
As mentioned above, the carbonate content of sediments is often correlated with 
primary lake productivity. Increased aquatic productivity (caused, for example, by an 
increase in temperature) will result in enhanced rates of photosynthesis by aquatic 
macrophytes, so increasing the amount of biogenic ally precipitated carbonate. In 
addition, raised temperatures may also encourage inorganic precipitation of carbonate 
from the water column. In simple lake systems, a strong linkage between productivity 
and carbonate content can often be demonstrated. However, this is not always the case. 
Care must be taken to avoid indiscriminately interpreting carbonate content from lake 
records as a proxy for aquatic productivity without first acquiring a detailed 
understanding of how each individual lake system operates. 
Decay of organic material within the sediment lowers the pH and generates CO2, which 
can lead to carbonate dissolution. Diagenetic effects such as this can be severe; in these 
cases, the amount of carbonate preserved within the sediment bears little relation to that 
produced by primary lake productivity (e.g. Dean, 1981). At Ioannina, however, SEM 
analysis of authigenic carbonates (particularly ostracod carapaces) shows no evidence 
for carbonate dissolution. High (>2) Mg/Ca ratios of the pore waters can also promote 
diagenetic alteration of primary carbonates (Muller et aI., 1972). As outlined in section 
4.2.2, the Mg/Ca ratio of the lake waters at Ioannina is low (0.37), confirmed by the 
precipitation of platey hexagonal calcite crystals, seen under the SEM. This suggests 
that pore-water fluids would have a similarly low Mg/Ca ratio. It is therefore assumed 
that any diagenetic alteration of carbonates within the sediment are negligible. 
The other possible contributor of calcite to the sediments of core 284 is from bioc1astic 
sources. For most of the history of the sequence, the absence of molluscan remains 
(dominantly littoral or sub-littoral in habitat preferences) would support the 
stratigraphical evidence that the core was located in a deep sub-basin. The dominance 
INTERPRETATIONS AND SYNTHESIS • 158 
of the fauna below this level by profundal benthic ostracods (Candona permanenta) also 
supports this assumption. Molluscan fragments first occur in any frequency in the 
sediment above about 145m, suggesting that the lake-level began to drop. A reduction 
in lake-level (for whatever reason) would have resulted in the setting of 284 becoming 
progressively more littoral. This is discussed at length in section 7.3.1, below. 
Despite careful avoidance of obvious shell fragments when sampling, increased 
bioclastic input would still affect the level of measured carbonate content in the core. 
This can clearly be seen at depths such as 26m and 32m (fig. 5.3), where the lake 
shallowed sufficiently for shell-beds of Dreissena to develop. Relatively high carbonate 
content values were recorded at these levels, even though obvious shelly material was 
removed from the sediment sample prior to measurement. 
It is therefore proposed that below 145m the carbonate curve can be used as a proxy for 
aquatic productivity with reasonable confidence. Above this depth, however, any 
interpretation of the carbonate curve should be regarded with caution, as values are 
likely to be biased by bioclastic input. 
7.2.2 Organic Matter Content 
Dean (1981) outlines two main sources of organic matter in temperate hard-water lake 
sediments. It may be allogenic, derived chiefly from catchment vegetation, or it may be 
autochthonous, derived either from vegetation in the littoral zone, or from pelagic 
phytoplankton and zooplankton. After detailed consideration of a series of lakes from 
the English Lake District, GOl'ham et al. (1974) concluded that most organic material in 
eutrophic lakes is authochthonous in origin and that allogenic material only becomes 
significant in the most oligotrophic water bodies. However, it is well established that 
primaly production of organic matter "is not necessarily related to the relative amount of 
organic matter that ultimately becomes part of the sediment" (Dean, 1981: p.225). Most 
organic material (allochthonous or autochthonous) that accumulates in the littoral zone is 
subject to destruction by benthic respiration and aerobic decay, particularly in eutrophic 
lakes where productivity is always relatively high. Even in the water column, 
phytoplankton and zooplankton debris is subject to aerobic decay within the epilimnion 
and to anaerobic decay within the hypolimnion and sediment. It is worth noting that 
even though dissolved inorganic carbon (DIC) is the major reservoir of carbon in a lake 
system, by a factor of at least 10 over pruticulate matter (Wetzel, 1975), it contributes 
very little to the proportion of organic carbon eventually stored in the sediment. As a 
result, only about 7% of the organic carbon produced in the lake actually becomes 
preserved in the sediment (Wetzel, 1975). 
INTERPRETATIONS AND SYNTHESIS • 159 
Because of the significant amount of organic carbon destroyed in both the water column 
and sediment, there is often only a very weak conelation between primary lake 
productivity and the amount of organic matter finally preserved in the sediment (Dean, 
1981). Care is therefore required in the interpretation of organic matter content of lake 
sediments. This also explains the often weak correlation between carbonate content (in 
lakes where this is strongly related to productivity) and organic matter content. At 
Ioannina, even though carbonate is thought to be a reasonable indicator of productivity 
below 145m (outlined above), the covariance between the two curves is reflected by a 
low r-value of 0.26. A comprehensive interpretation of the total organic matter curve in 
this case is therefore difficult and the points outlined above must be borne in mind. 
Accordingly, when discussed as part of the overall synthesis of the core data (section 
7.5), it is refened to in only the broadest terms. 
7.2.3 Magnetic Susceptibility 
Although magnetic susceptibility measurements were taken at a relatively coarse 
sampling interval along the entire core (approximately every 1m), the results were 
nevertheless sufficient to pennit conelation with core 249 (section 5.1). In addition, 
the results can be used to derive meaningful palaeoenvironmental information. A 
reasonable first-order anti-correlation with the carbonate and organic matter curves (fig. 
7 .1) suggests that there was higher erosion and inwash from the catchment during 
glacial phases. This was possibly due to a poorer degree of soil surface stability 
originating from reduced vegetational cover. Nevertheless, higher order variability in 
the magnetic susceptibility curve indicates that other factors, for example changing 
erosional and weathering regimes, also affected the concentration and rate of magnetic 
mineral inwash into the lake. 
In section 4.1.2 it was suggested that the magnetic carrier in the intervals displaying 
very high susceptibility values had characteristics similar to magnetite/maghaemite. It is 
important to consider the origin of this canier. As mentioned in section 3 .2.2, 
pedogenic enhancement of soils can occur even on bedrock that lacks primary magnetic 
minerals, such as limestones (e.g. Longworth et al., 1979; Dearing, 1979; Flower et 
al., 1984). Studies of the magnetic enhancement of topsoils have demonstrated that the 
fine-grained (sub-micron) ferrimagnetic iron oxide minerals formed in this way are 
often either thought to be magnetite (Fe30 4) or maghaemite (yFe20 3) (e.g. Le Borgne, 
1955; Longworth et al., 1979). The pe do genic evolution of these minerals is poorly 
understood, although Mullins (1977), Longworth et al. (1979) and Dearing et al. 
0.0 
10.0 
20.0 
30.0 
40.0 
50.0 
60.0 
70.0 
80.0 
90.0 
100.0 
110.0 
120.0 
§ 130.0 140.0 
..c 
0.. 150.0 0) 
Cl 160.0 
170.0 
180.0 
190.0 
200.0 
210.0 
220.0 
230.0 
240.0 
250.0 
260.0 
270.0 
280.0 
290.0 
300.0 
310.0 
::e 
f f f 
0 5 10 15 
Organic Matter Content (%) 
f 
20 0 10 20 30 40 
Carbonate Content (%) 
50 -1 0 2 3 
Magnetic Susceptibility 
(xlo-7m3kg- 1) 
4 5 
Fig. 7.1 Comparison of organic matter and carbonate contents with magnetic susceptibility. Zones of anti-correlation are shown (discussed in text). 
~ 
~ 
t;j 
:::>;, 
'"tl 
:::>;, 
tll 
~ 
:::l 
<::) 
~ 
~ 
~ 
V) 
~ 
5l 
~ 
• 
...... 
0\ 
o 
INTERPRETATIONS AND SYNTHESIS. 161 
(1985) all suggest several possible mechanisms, including burning and gradual redox 
processes occurring under nOlmal soil forming conditions. 
In her study of sediment cores from the Kopais Basin in central Greece, AlIen (1986) 
carried out detailed magnetic analyses and determined the carrier to be dominantly 
magnetiteimaghaemite in the sediments younger than about 12.5ka BP and dominantly 
haematite in the sediments older than this. She also acknowledged the probable 
contribution from airborne dust (coated in what was probably haematite) derived from 
the Sahara. The difference in magnetic mineralogy in the cores was explained by 
advocating some of the conclusions of Tite and Linnington's (1975) study on the 
effects of climate on the magnetic susceptibility of soils. She suggested that 
'MeditelTanean' climatic conditions re-established at the end of the Lateglacial and 
during the Holocene would have favoured the reduction (under humid winter 
conditions) and subsequent reoxidation (under arid summer conditions) of minerals 
such as haematite and goethite into magnetiteimaghaemite. 
The soils in the Ioannina catchment are rich in the iron oxide minerals goethite and 
haematite (section 2.3). On the basis of the existing evidence therefore, it would seem 
reasonable to assume that the magnetic carrier (at least in the intervals of core displaying 
high susceptibility values) is magnetiteimaghaemite. There is therefore a clear potential 
for further detailed rock magnetic work, not only to investigate the nature of the calTier 
in the other levels of the core, but also to establish the exact source of these minerals. 
For example, work carTied out at Rhode River in Maryland, USA linked the magnetic 
characteristics of soils and sediment samples to enable sources within the catchment to 
be distinguished (Oldfield, 1983; Dearing et al. , 1985; Yu and Oldfield, 1989). 
Measurement of the magnetic char'acteristics of a variety of catchment lithologies in the 
Ioannina basin (including soils, flysch and limestone units), coupled with a detailed 
magnetic analysis of the core material at higher resolution may enable sediment sources 
to be identified and conclusions drawn as to changing erosional and drainage regimes. 
7.3 FAUNAL ANALYSES 
7.3.1 Lake-Levels 
In chapter 6, the use of faunal indicators as a relative measure of lake-level was 
discussed. This was also referred to when discussing the carbonate content of the 
sediment (section 7.2.1, above). Essentially, the faunal data obtained during this study 
INTERPRETATIONS AND SYNTHESIS. 162 
offer a means for inferring broadscale lake-level variation. Stratigraphical information 
(discussed in section 2.3) suggests that for most of its history, core 284 was located in 
a deep sub-basin and therefore had a profundal setting. The absence of molluscan 
remains for most of the sequence (largely preserved in littoral or sub-littoral sediments) 
and the presence of benthic, profundal ostracod shells supports this interpretation. 
However, it would appear that during the penultimate glaciation (MIS-6), the lake-level 
began gradually to fall. Molluscan shells are preserved in the sediment above about 
14Sm, indicating that the location of bore hole 284 was located in increasingly shallower 
water. This may be a result of the severe conditions at the end of the penultimate 
glaciation, thought to be one of the most extensive glacial phases of the Pleistocene 
(e.g. van Andel and Tzedakis, 1996). The pollen record from Tenaghi Philippon (Smit 
and Wijmstra, 1970) indicates that during this time the climatic conditions in northern 
Greece were likely to have been cold and arid. The presence of vegetational refugia in 
sheltered Balkan locations, such as Ioannina, would suggest that conditions were 
perhaps not equally as severe everywhere (Bennett et al., 1991; Tzedakis, 1993). 
Nevertheless, Lake Pamvotis is known to have been sensitive to variations in PIE 
(discussed in section 7.4, below). Conditions of low precipitation and relatively high 
evaporation during the penultimate glaciation may have forced the lake-level to fall to 
the point where aquatic mollusc fragments could become preserved in the sediment at 
the site of core 284. 
The subsequent mcrease in PIE during the Eemian may account for the lack of 
molluscan fauna through that interval. The reappearance of shell fragments at around 
7Sm, during the cold stadial MIS-Sb, suggests the re-establishment of low PIE 
conditions. These must have persisted until the height of the last glacial, since an 
increasing proportion of shelly material culminated in very shallow lake-levels, 
recorded in the core by the presence of beds of Dreissena, a shallow-littoral freshwater 
mussel. A schematic lake-level curve derived from faunal evidence has previously been 
illustrated (fig. 6.2). Although details of Eemian and LateglaciallHolocene lake-level 
variations are discussed further in section 7.S, one specific example of the use of faunal 
data for this purpose is now given. 
7.3.2 The Kastritsa Problem 
Near the village of Kastritsa in the south of the Ioannina basin, extensive archaeological 
work has been canied out on an Upper Palaeolithic rock-shelter at the base of the 
Kastrelaki limestone ridge (Higgs and Vita-Finzi, 1966; Higgs et al., 1967; Vita-Finzi, 
INTERPRETATIONS AND SYNTHESIS • 163 
1978; Bailey et al ., 1983a,b). Some of the sediments found within the cave were 
interpreted as beach deposits (approximately 6m above current lake-level) on the basis 
of the graded pebble bed lithology (Higgs, et al., 1967). Hearths "within the beach 
deposits" were recorded and two radiocarbon determinations on charcoal yielded dates 
of 20,800±810 yrs BP (1-2466) and 20,200±480 yrs BP (1-2468). This has been used 
extensively as evidence for high lake-levels in the basin at the time of the LGM (e.g. 
Higgs and Vita-Finzi, 1966; Higgs et al., 1967; Vita-Finzi, 1978; Bailey et al., 
1983a,b; Roberts, 1982, 1983; Prentice et al., 1992; Harrison and Digerfeldt, 1993). 
However, this conflicts with the pollen record from 10annina and elsewhere in the 
northern Mediterranean, which on the basis of Artemisia and chenopod dominated 
pollen assemblages, indicate an arid and cold climate at this time (e.g. Bonatti, 1966; 
Smit and Wijmstra, 1970; Pons and Reille, 1988; Tzedakis, 1994). 
It has always been assumed that the original deposits were interpreted correctly as beach 
sediments. Certainly, the graded water-worn pebble beds are reasonably good 
evidence; as Bailey et al. (l983a) point out, the distribution, size and shape of the 
pebbles are consistent with deposits forming on beaches of the modern day lake. 
Further examination of some of the original sediments (courtesy of G.N. Bailey) 
revealed a monospecific faunal content. Dreissena is a shallow-water bivalve that often 
occurs in large numbers (it appears in shell-beds in the lake today). Examination of the 
fine sieved fraction of these sediments (for example, to look for ostracod evidence) has 
unfortunately not been possible. The veracity of the dates must also be considered. 
The published sections (Higgs et al., 1967, fig. 10, p.23; Bailey et al., 1983a, figs. 7 
and 8, pp. 27 and 28) show the hearths actually underlying the 'beach' sediments and 
overlying a red, stony deposit that contains charcoal and a lithic industry. Without 
further three-dimensional details of the excavation, the question arises as to the 
stratigraphical integrity of placing the hearths with the overlying, rather than the 
underlying sediments. 
Several authors have tried to resolve the situation by invoking climatic arguments. 
Bailey et al. (1983a, b) proposed a scenario in which lower evaporation rates 
accompanied reduced temperatures. Prentice et al. (1992) used water-balance and 
biome climatic models to reconstruct the palaeoenvironment at Ioannina during the 
LGM. They concluded that if the jetstream was shifted southwards during winter (as 
several atmospheric models have predicted), then enhanced precipitation during severe 
winters would result, which, coupled with summer drought, would explain the 
presence of semi-arid steppic vegetation at times of high lake-level. Harrison and 
Digelfeldt (1993) also presented the Ioannina data as their primruy evidence for high 
lake-levels during the LGM in the Mediterranean. Out of eleven sites quoted, only 
INTERPRETATIONS AND SYNTHESIS. 164 
Ioannina, Xinias (Greece) and Padul (Spain) had records which stretched back to the 
LGM. Examination of Global Circulation Models (GCMs) for the region led them to 
suggest that cooler and cloudier summers, coupled with extremely dry and cold 
winters, would be sufficient to account for the apparent paradox. 
As Tzedakis (1994) points out, what many of these authors have seemingly failed to 
take into account is the tectonic setting of the site. As mentioned in section 2.2, Epirus 
has been a region of tectonic compression since the beginning of the Pleistocene 
(Clews, 1989). Tzedakis (1994) refers to unpublished IGME stratigraphical data 
gleaned from borehole studies, which shows the depth of the top lignite band 
shallowing markedly from the lake shore towards Kastritsa (from 229-40m). This 
suggests uplift of the south-eastern part of the basin, an observation that is ill 
accordance with the terraces seen outcropping to the east (IGSR-IPF, 1966). In 
addition, King and Bailey (1985) record the presence of up arched fluvial deposits at the 
southern end of the Kastrelaki ridge. They attribute this deformation to uplift of part or 
all of the ridge (although they do not try to estimate the degree or rate of any 
movement). 
Significant new mollusc an evidence has been presented as part of this study for low 
lake-levels during the LGM (section 6.2.3). Beds of Dreissena occur in core 284 at this 
time, which (as explained previously) can be used as an indicator for shallow, littoral 
environments. Borehole 284 is thought to have been drilled through one of the deepest 
sub-basins of the lake (section 2.3). It is not possible for shallow lake-levels at this 
central site to have been coeval with high lake-levels at the margins. It is entirely likely 
therefore, that the elevated beach deposits in the Kastritsa cave have been subject to 
uplift and so cannot be used as a reliable former lake-level indicator without a correction 
for tectonic movements first being applied. This may have important consequences for 
the Mediterranean palaeolake-Ievel studies referred to above. 
7.4 CHEMICAL ANALYSES 
7.4.1 Modern Waters 
The results of the modern water sampling (section 4.2.2) indicate that Pamvotis is a 
typical freshwater, eutrophic lake, with a chemistiy dominated by calcium (Ca2+) and 
bicarbonate (HC03} The individual sampling sites were located around the periphery 
INTERPRETATIONS AND SYNTHESIS. 165 
of the lake (fig.3.1). In the following discussion, figs. 4.8 and 4.9 should also be 
consulted. 
Samples 11, 3 and 7 display slightly lower 813C values, indicative of plant or soil­
derived dissolved inorganic carbon (DIC). These samples were taken from sites that 
were located in backwater areas with abundant plant macrophytes and decaying plant 
matter, where lower 813C values would be expected. Site 6 was located adjacent to a 
freshwater outflow leading from a spring to the north-west of the lake. Its lower 8D, 
813C and 8180 values and consequent position near to the meteoric water line (MWL) 
are indicative of non-evaporated, relatively fresh water. The substantially higher 
chloride and sulphate content of site 6 waters can be explained as being derived from 
pollution originating from the agricultural small-holdings currently lining the water­
course leading from spring to lake. 
The average 8 I3C value of the waters was around -3.5%0, somewhat lower than might 
be predicted for present-day waters in equilibrium with atmospheric CO2, The HC03-
CO2 (gas) isotopic fractionation is around +8.7%0 (at 18°C) and atmospheric CO2 has a 
813C value close to -8%0. This means that waters in equilibrium with present-day 
atmospheric CO2 would be expected to have a 8 I3C value of approximately +0.7%0, not 
-3.5%0. The depletion seen in the Ioannina samples is possibly because the lake waters 
still possess a remnant of the groundwater-derived bicarbonate, or, more likely in view 
of the eutrophic nature of the lake, it is because they still contain some 12C-rich carbon 
derived from organic matter decay (T.H.E. Heaton, pers. comm.). In other words, 
these figures suggest that the modern lake waters are not in full equilibrium with 
atmospheric CO2, The implications of the strong covariance between the 8180 and 813C 
values are discussed in section 7.4.4, below. 
7.4.2 Sediment Isotope Geochemistry 
Stable oxygen and carbon isotopic measurements were carried out on bulk sediment 
carbonate at 20cm intervals between 100-83m, interpreted as including the entire 
Eemian period (section 5.8). In order that a balanced interpretation of the results 
(fig.4.2.1) might be attempted, the possible range of controls on stable isotopes in a 
lake system is now considered. 
INTERPRETATIONS AND SYNTHESIS • 166 
(a) Oxygen 
The 180/160 value of calcite is a function of: (i) the temperature of calcite precipitation, 
i.e. water temperature; and (ii) the 180/160 of the lake water, which may be controlled 
by the isotopic signature of the precipitation (which may in turn be controlled by such 
factors as air temperature and moisture source), the PIE ratio, the isotopic signature of 
any dissolved carbonates from the catchment and any groundwater inputs. The origin 
of the bulk calcite in the profundal sediments that make up 284 is chiefly a result of 
biogenic precipitation in surface waters (discussed in section 7.2.1, above). 
If the 8180 signal recorded in the carbonates over this interval was dominantly 
controlled by water temperature, then the curve would be expected to become markedly 
more positive during the warm interglacial period. This is because of the temperature 
dependency of the fractionation process (e.g. Faure, 1986). In fact, a marked negative 
oscillation is seen across the Eemian (fig. 4.7), which is more consistent with a PIE 
control on 8180. 
An increase in the PIE ratio introduces more 160 into the system (in the form of H 2160), 
ensuring that 8180 decreases. Obviously, the reverse is tme during times of low PIE 
ratio. Pollen data derived from core 249 indicate that the Eemian was relatively warm 
and wet in comparison to the previous glacial period (Tzedakis, 1991), an interpretation 
confirmed by unpublished pollen data from core 284 (see section 7.5, below). In 
addition, the numerical values recorded by the 8180 curve are within the limits of what 
might be expected for carbonate precipitating in equilibrium at Ioannina. Using present­
day climatic data (section 2.4), a typical range of 8180 values for precipitating 
carbonates is calculated to be between about -4.0 to -8.0%0. Assuming that the range of 
temperatures and isotopic values of precipitation during the last interglacial was 
reasonably similar to that of today, the mean 8180 value of the bulk carbonates of 
around -5%0 falls comfortably within the predicted range. Since the other factors which 
contribute to the 180/160 of the lake water (i.e. isotopic composition of precipitation and 
ground water) are likely to be relatively constant factors in comparison, it seems 
reasonable that PIE is the dominant control on 8180 values through this interval. 
(b) Carbon 
Interpreting the carbon isotope signal through this interval is not as straightforward as 
for the oxygen isotope record, as there are many more potential influences on the 8l3e 
values. These include: (i) the surface water productivity (i.e. uptake of 12e by 
INTERPRETATIONS AND SYNTHESIS. 167 
photosynthesis); (ii) respiration/decay of organic matter (i.e. release of 12C at depth); 
(iii) the isotopic composition of inflowing waters and precipitation; and (iv) the degree 
of CO2 exchange with the atmosphere (McKenzie, 1985). These will now be 
considered in turn. 
(i) Suiface Water Productivity. Photosynthesising phytoplankton in the surface waters 
of the lake preferentially incorporate 12C from the dissolved inorganic carbon (DIC) 
pool, leaving the DIC enriched in 13C (McKenzie, 1985). As noted in section 7.2.1, 
this removal of 12C from the DIC produces a disequilibrium in the bicarbonate­
carbonate system, raising the pH and allowing calcite to precipitate with an enhanced 
()13C composition. Carbonates precipitated in surface waters (or epilimnion) will 
therefore record an increase in aquatic productivity as an increase in ()13c. 
(ii) RespirationlDecay. Organic (plant) material living in the surface waters eventually 
dies and sinks. At depth, this organic matter decays, releasing 12C-enriched CO2 back 
into the water, reducing the ()13C value of the DIe. Since Pamvotis is a shallow lake, 
this process is likely to be a relatively minor, though reasonably constant, factor 
(discussed further below). Carbonates precipitated at the bottom of the lake (or in the 
hypolimnion) will therefore record an increase in aquatic productivity as a decrease in 
()13C. 
(iii) Input. Since there is no fluvial input to the lake, only the isotopic composition of 
precipitation and groundwater need to be considered. The lake is fed by springs at the 
base of a karstic aquifer system; the dissolved carbonates in the groundwater are 
therefore likely to remain at fairly constant levels over long periods of time and so can 
largely be discounted. The contribution to DIC from precipitation is also likely to be a 
negligible factor. 
(iv) Exchange With Atmospheric CO2 , A longer residence time of lake water will 
permit more gaseous exchange, resulting in more positive values of () l3c. The 
converse applies if the residence time is decreased. 
If groundwater, precipitation input and respiration/decay are all likely to be reasonably 
constant factors affecting the ()13C of the lake system, then either aquatic productivity or 
CO2 exchange remain the prime candidates for the dominant controlling influence on 
()13e. As explained above, an increase in surface water productivity produces a more 
positive ()13C signal in carbonates precipitated in surface waters. During the Eemian, 
increased temperatures and precipitation would have encouraged increased levels of 
aquatic productivity (broadly reflected in the carbonate curve over this interval [section 
INTERPRETATIONS AND SYNTHESIS. 168 
4.2.1]). However, the trend in ol3C over this interval is markedly negative, which 
would rule out a controlling influence by surface water productivity. 
If the carbonates being measured had been precipitated on the surface of (or within) the 
lake sediments, then a negative shift in ol3C would reflect increased respiration/decay, 
which is linked directly to surface water productivity (as described above). However, 
this is unlikely because of the physio-chemical factors affecting calcite saturation. 
Primary carbonates tend to be precipitated in surface waters (rather than at depth) 
because calcite saturation is linked to the solubility of CO2, which is temperature and 
pressure dependent (Dean, 1981). Increasing temperature or decreasing pressure 
decreases the solubility of CO2, the removal of which leads to calcite precipitation. In 
addition, the solubility of calcite itself is also temperature and pressure dependent (Kelts 
and Hsti, 1978). Other factors influencing calcite saturation and precipitation are 
discussed in Kelts and Hsti (1978) and Dean (1981). On the other hand, precipitation 
due to the oversaturation associated with loss of CO2 or bicarbonate during 
photosynthesis (i.e. biogenic carbonates) can occur anywhere within the photic zone. 
However, if conditions during the last interglacial were in any way similar to those 
prevalent today, thick algal material in the surface waters would have made the water 
very turbid, limiting the effective depth of this zone (see modern sampling data in 
appendix A). The negative trend in ol3C across this interval is therefore unlikely to be 
due to the carbonates precipitating on the bottom of the lake. 
Consideration must therefore be given to the exchange of CO2 with the atmosphere. 
Longer residence times will allow the lake waters to develop an evaporative signal, 
resulting in raised ol3C values. If the residence time is reduced, the waters have less 
opportunity to exchange with the atmosphere and so cannot develop such a strong 
evaporative signal. During the Eemian, the 0180 curve indicates that it was a time of 
high PIE relative to the preceding glacial phase (also supported by pollen data). This 
would result in a relatively more negative ol3C signal than during, for example, the 
previous glacial, when according to the 0180 record, evaporation rather than 
precipitation was more important. 
This interpretation can be supported by considering the magnitude of the isotopic values 
concerned. The isotopic fractionation between the 013C of bicarbonate and that of 
atmospheric CO2 is around +9%0 at 18°C. If Eemian CO2 levels are assumed to be 
between -7 to -6%0 (modern day values are approximately -8%0), then carbonate 
precipitating in equilibrium with the atmosphere would have a 013C value of between +2 
to +3%0. This is close to the values seen at the beginning of the sampled interval 
INTERPRETATIONS AND SYNTHESIS • 169 
(interpreted as MIS-6) and at the end (interpreted as MIS-5d). These two periods are 
known to have been cold and relatively arid, which suggests low levels of input to the 
lake by precipitation and from springs (i.e. a low PIE). Since the ()13C of the carbonate 
precipitating during these times was approximately in equilibrium with the atmosphere, 
this confirms longer residence times. Values of ()13C for the bulk of the Eemian were 
much lower however, averaging around -1%0. This suggests much shorter residence 
times (i.e. less opportunity to exchange with the atmosphere), in accordance with the 
high PIE indicated by the ()ISO record. 
Modern values of ()13C average around -4%0 whereas Eemian ()13C values average 
around -1 %0. This is a substantial difference if present day conditions are to be used as 
a basis for comparisons. The discrepancy can partly be explained by the fact that in the 
Eemian, lake-level was undoubtedly higher than in today's artificially-drained lake. 
Molluscan evidence presented above (section 7.3.1) clearly shows that the lake began to 
shallow after the end of the Eemian and never regained its former depth, even during 
the Holocene (this assertion is also supported by lithological evidence). A deeper lake 
would permit relatively longer residence times than in a shallower lake, explaining why 
the Eemian ()13C values are more positive than modern values. 
7.4.3 Ostracod Shell Chemistry 
Stable oxygen and carbon isotopic measurements were carried out on shells of the 
ostracod Candona pennanenta at 20cm intervals over the top 25m of the core. Before 
any interpretation of the isotopic data can be suggested, the primary influences 
controlling the ()ISO and ()13C of the ostracods must be discussed. 
(a) Oxygen 
The controls on ()ISO in ostracod shell calcite in a shallow lake are effectively the same 
as for primary precipitated calcite (section 7.4.2, above), once biological factors such as 
'vital effects' are taken into account (discussed below). Essentially, these controls are 
the temperature of calcite precipitation (i.e. water temperature) and the ()I SO of the lake 
water (which may in turn be influenced by the PIE ratio as well as the isotopic 
composition of precipitation and inflowing groundwaters). Following the same 
arguments outlined for primary carbonates, the ()ISO of carbonates formed in a shallow 
lake, such as Pamvotis, is most likely to be influenced by water temperature and the PIE 
ratio, particularly as the thermocline is not very pronounced (Overbeck, 1980). Other 
INTERPRETATIONS ' AND SYNTHESIS. 170 
factors, such as the isotopic composition of precipitation and groundwater, are likely to 
remain relatively constant. 
The main control on ()ISO for the primary precipitated carbonates in Lake Pamvotis was 
thought to be the PIE ratio. If this factor is also the most important for the top 25m of 
sediment, then more positive ()ISO values denote a low PIE ratio. The ()ISO curve 
shown in fig. 4.10 shows more positive values at the base of the profile (-25m), at 
around 16m and again at about 10m, suggesting that these periods were relatively dry. 
This corresponds well with dry phases suggested by the age model and discussed at 
length in section 7.5, below. 
This interpretation also makes sense from the point of view of the magnitude of the 
()ISO values. As explained above (section 7.4.2), using present-day climatic data, the 
range of ()ISO values for calcite forming in the lake is around -4.0%0 to -8%0, with the 
highest value of calcite precipitating from non-evaporated rain likely to be about -2.5%0 
(T.H.E. Heaton, pers. comm.). However, this value will become more positive if the 
waters have undergone evaporation, as is the case with the modern waters (average 
()ISO is around -1 %0). In addition, since ostracods do not always fractionate isotopes in 
equilibrium with their host waters, they often display species-specific offsets known as 
'vital effects' (e.g. von Grafenstein et al., 1992). Xia et al. (l997b) recently used 
laboratory cultures of Candona rawsoni (a North American species) and estimated this 
offset to be around +1%0 (depending upon temperature). On the other hand, from an 
extensive (unpublished) collection of field data, U. von Grafenstein (pers. comm.) 
estimates that a more realistic estimate of offset for a variety of Candona species from 
central European lakes to be around +2%0. The ()ISO values obtained from this study 
are therefore well within the expected range (assuming that the waters have undergone a 
degree of evaporation), even making suitable allowances for the fact that modern 
climatic data were used. Once again, PIE appears to be the dominant influence on the 
()ISO signal. 
(b) Carbon 
The ()I3C signal from ostracod shell calcite can sometimes be extremely difficult to 
interpret. Whilst ostracods are subject to similar controls on () 13C as primary 
precipitated carbonates (section 7.4.2, above), they have the 'added complication of 
living in a variety of different microhabitats and might therefore be subject to different 
isotopic influences. For example, in deep lakes, benthic epifaunal species may record 
different isotopic signatures to littoral or swimming species. The release of '2C-rich 
INTERPRETATIONS AND SYNTHESIS. l71 
CO2 back into the DIC from respiration/decay of organic matter at depth will reduce the 
relative 8I3C value of the benthic ostracod shell calcite. On the other hand, littoral or 
swimming species in the epilimnion may record an enhanced 8 I3C value as 
photosynthesising phytoplankton take up 12C-rich CO2 from the DIe. Even if two 
species have shallow water ecological preferences, if one prefers muddy backwater 
areas amongst stagnant and decaying vegetation, whilst another prefers sandy and 
vegetation-free habitats, then the former will probably record a lower 813C signal than 
the latter, due to a local enhancement of 12C within the DIe. 
The ecology of Candona permanenta, the species used for isotopic determinations in 
this study, is discussed in section 6.3.2. It is likely to be a sublittoral epibenthic 
species, moving around over the substrate feeding on decaying organic matter in parts 
of the lake beyond the littoral zone. From the range of controls influencing 8 I3C 
composition, the most important is probably aquatic productivity, reflected in variations 
in respiration/decay of organic material at depth. This was regarded as a relatively 
minor factor when primary precipitated carbonates were considered (see above). 
However, it is likely to assume more importance when dealing with a benthic ostracod 
species that precipitates its shell carbonate whilst living on the floor of the lake. 
If 8 I3C is primarily controlled by respiration/decay, the curve might be expected to 
covary with 81 80, since increased precipitation (at a time when temperatures are also 
known to have increased) would result in increased surface water productivity (and 
hence increased decay at depth). In fact, for the lower part of the sequence (up to 
-ISm) the two curves only covary moderately (r = 0.31 for the smoothed data), 
indicating that perhaps respiration/decay was not such a dominant factor through this 
interval and other influences (such as variations in input from groundwaters or 
exchange with atmospheric CO2) had increased importance. However, covariance is 
stronger through the upper part of the sequence (r = 0.75 for the smoothed data), a 
period interpreted as the Holocene (section 5.8). This may indicate that 
respiration/decay became the dominant factor controlling 813C through this interval. 
For primary precipitated carbonate, it was thought that 8 13C was mainly controlled by 
exchange with atmospheric CO2, This resulted in a more positive values during cold 
stages, where less precipitation and longer residence times would allow the waters to 
develop an evaporated 813C signal. This is a trend that would be indistinguishable from 
respiration/decay, since drier climatic phases would result in longer residence times 
(more positive 813C values) and a lowering of aquatic productivity (less material to 
decay at depth, so more positive 813C values). In deeper lakes, benthic ostracods might 
be expected to respond more to respiration/decay rather than exchange with atmospheric 
INTERPRETATIONS AND SYNTHESIS. 172 
CO2 (which would have more of an effect in the surface waters). In a shallow lake like 
Pamvotis, however, the signals are likely to be indistinguishable unless independent 
indications exists to support the influence of one control over another. Whilst 
mollusc an and lithological evidence (sections 7.3.1 and 4.1.1) suggest that the lake 
became progressively shallower over this interval, there is no independent evidence to 
suggest that productivity levels varied. Carbonate is thought to be an unreliable proxy 
for aquatic productivity over this interval (section 7.2.1). In addition, although the 
upper levels of the core are richer in faunal material than lower intervals, this may 
simply be a reflection of the deposition site being located in progressively shallower 
water, rather than an indication that productivity of the lake as a whole was increasing. 
As a result, it is difficult to determine exactly which control (if any) is the more 
dominant throughout this interval of core. As this particular ostracod is an epibenthic 
sublittoral species, it seems probable that the controlling factor is most likely to be 
respiration/decay, but this is difficult to justify. These interpretations are discussed 
further in section 7.5. 
One final point to make concerning the ostracod isotope results is that of seasonality. 
Studies have shown that most species of Candona moult only at certain times of the 
year when conditions are optimal (e.g. Xia et al., 1997a). By means of extensive field 
sampling in Lake Ammersee in Bavaria, von Grafenstein (1992) has found that many 
species of Candona display a frequency maximum of adults during autumn and early 
winter. This means that the isotope signals derived from these species will carry a 
seasonal bias. Central European climate is different to that of the Eastern Mediterranean 
and little is known about the preferences and tolerances of those ostracod species found 
only in Balkan lakes. It is possible that the data from Ioannina also shows an 
autumn/early winter bias. However, without detailed ecological and population studies 
on C. permanenta, it is difficult to support such an assertion. 
7.4.4 Open or Closed Lake System? 
Hydrologically closed lake basins are known to be sensitive to climatic variability, since 
their volume and chemistry are influenced primarily by PIE. Lake Pamvotis is a 
shallow lake with no pelmanent outflow, suggesting that it may be a closed system. If 
this were the case throughout its history, then much of the chemical and isotopic 
variability noted in the record can be interpreted as resulting primarily from climatic 
INTERPRETATIONS AND SYNTHESIS. 173 
influences. However, as discussed in section 2.5, the lake is part of a karstic aquifer 
system underlying the Mitsikeli ridge. In addition, the blocked sink-holes (katavothrae) 
in the floor of the lake can become temporarily cleared (for example, by tectonic 
activity), inducing a lake-level drop that is not a result of climatic influences. 
The chemical signature of lake waters can sometimes be indicative of the 'open' or 
'closed' status of a lake basin. Talbot (1990) suggests that a covariance of r> 0.7 
between 8180 and 813C of primary lacustrine carbonates indicates that they were 
precipitated in a hydrologic ally closed system. However, he also points out that several 
open lakes have covariances close to those of closed lakes. These include Lobsigensee 
in Switzerland (r = 0.65) (Siegenthaler and Eicher, 1986) and Little Lake in Ontario, 
Canada (r = 0.51) (Turner et al., 1983). Talbot (1990) argues that this is probably 
because, like closed lake systems, they have relatively long residence times. In 
addition, von Grafenstein (pers. comm.) advocates extreme caution when assessing the 
significance of covariant stable isotopic data from lake systems. His detailed 
(unpublished) 8180 and 813C measurements from Lake Ammersee in Bavaria show a 
strong covariance (r> 0.9), but the basin is demonstrably open. He attributes these 
results to the importance of PIE for this particular lake system and stresses the danger in 
the automatic classification of open/closed lakes based solely on isotopic evidence. The 
chemical results obtained from the modem Ioannina lake waters and fossil carbonates 
will now be considered in turn. 
(a) Modern 
The covariance between the 8180 and 8 13C values of the modem lake waters (fig. 4.9) 
is reasonably strong (r = 0.91), suggesting that it may be effectively closed. The 
evaporative trajectOlY of the Ioannina waters also helps to support this suggestion (fig. 
4.8). Its gradient of 4.8 is close to the value of 5 that Faure (1986) considers indicative 
of closed lake basins. 
(b) Last glacial - Holocene 
The isotopic signals from ostracods were measured over this interval (section 4.2.3). 
As explained previously (section 7.4.3), ostracod shell calcite precipitated in shallow 
lakes behaves like primarily precipitated calcite, once biological factors such as 'vital 
effects' have been accounted for. Overall covariance between values of 8180 and 813C 
derived from ostracod shells over this interval is only moderate (r = 0.54, smoothed 
INTERPRETATIONS AND SYNTHESIS • l74 
data). However, covariance varies between different parts of the record (fig. 4.11). 
Below ISm the curves covary with an r-value of 0.31; above ISm, the r-value improves 
to 0.75. This may be a reflection of the changing influence of other factors on the 
isotopic composition of the waters from which the ostracods build their shells, for 
example the changing trophic status of the lake in response to climatic variation. A 
more likely alternative is that it represents an increase in organic productivity caused by 
lake-level fluctuations (discussed in section 7.5, below). 
(c) Eemian 
An isotopic signal derived from bulk sediment carbonate was measured over this 
interval (section 4.2.1). The 0180 and oI3e values covaried reasonably well (r = 0.58). 
Again, the fact that ol3e does not covary even more closely with 0180 probably 
indicates that other hydrological factors were also significant, though to a lesser degree. 
(d) Discussion 
Until a detailed hydrological study is carried out on the modem lake, any interpretation 
of the data has to be viewed with a degree of caution. It is difficult to classify the 
hydrological status' of the modem Lake Pamvotis as either entirely open or entirely 
closed, since in reality, it probably falls somewhere in between. In general terms, 
however, the isotopic values show that the surface water has a residence time sufficient 
enough to permit some evaporation, which may be broadly indicative of 'closed' status 
(agreeing with the blocked katavothrae hypothesis previously outlined). 
In palaeolake terms, it is again difficult to assess clearly the hydrological status of the 
lake. Although the covariances between 0180 and ol3e values do not unequivocably 
fall into the closed-basin category of Talbot (1990), they are reasonably close. 
However, they celtainly support the probability of long residence times and the 
important influence of PIE on lake-level indicated by the isotopic curves for both the 
Last glacial - Holocene and Eemian intervals. 
In summruy, the available data therefore suggest that Lake Pamvotis can be considered 
in the broadest telIDS to be a closed hydrological system. Modem and fossil isotopic 
results indicate long residence times and the importance of PIE on lake-level variability. 
Nevertheless, other hydrological factors ru'e also likely to be important and further 
INTERPRETATIONS AND SYNTHESIS • 175 
detailed work is necessary (particularly on understanding the complexities of the 
modem lake system) before they can be adequately accounted for. 
7.5 SYNTHESIS 
This section represents an attempt to link together the data presented in the previous 
chapters and, bearing in mind some of the interpretations discussed above, derive a 
coherent and comprehensive palaeoenvironmental history of the lake. Particular 
attention is paid to those intervals of the core studied in higher resolution. 
Core 284 does not extend back to bedrock, making it impossible to derive any 
conclusions concerning the origins of the lake. The 284 record begins somewhere in 
MIS-16, approximately 600ka ago. The borehole site was situated in part of a deep 
(perhaps the deepest) sub-basin within the lake. Ostracod faunal evidence (and lack of 
molluscan shell material) suggests that the site was probably profundal and remained so 
for much of its history, although it is difficult to estimate an absolute value for the water 
depth. 
Sediments accumulated reasonably rapidly on the floor of the lake, a process that was 
aided by the continual subsidence of the sub-basin. The sediments consisted of 
primarily precipitated calcite, siliceous material (largely diatomaceous), aeolian material 
(derived originally from the deserts of North Africa) and clay minerals, ultimately 
derived from the weathering of the catchment (particularly, the nearby flysch deposits). 
Pedogenic enhancement of the catchment soils (on both limestone and flysch units) 
provided magnetic minerals that were eventually washed into the lake and incorporated 
into the sediments. The productivity of the lake appears to have fluctuated over time in 
response to climatic variation, in tandem with the variability in vegetation communities 
within the catchment (Tzedakis, 1991 , 1993, 1994) (see also the discussion of 
leads/lags, below). Warmer, interglacial periods were characterised by high PIE, a 
forest succession and increased primary calcite precipitation linked to a rise in aquatic 
productivity. Glacial periods saw low PIE, a corresponding expansion of open 
vegetation communities and lower levels of calcite precipitation. 
Extended severe climatic conditions during the penultimate glaciation (MIS-6) led to a 
gradual lowering of the lake-level. Aquatic molluscs, which are nOlmally mainly 
preserved in littoral or sub-littoral sediments, appeared in the record as the site of the 
284 borehole became shallower. 
INTERPRETATIONS AND SYNTHESIS. 176 
7.5.1 Eemian 
According to the age model, the interval 83-100m is thought to include the whole of the 
Eemian. The beginning and the end of the last interglacial in core 284 is well 
constrained both palaeomagnetically and by means of U-series determinations (chapter 
5). The vegetational history from core 249 suggests that the last interglacial period 
(designated as the Metsovon interglacial at Ioannina) was characterised by a significant 
increase in open forest taxa, dominantly Quercus, along with Ulmus, Zelkova and Olea 
(Tzedakis, 1991). This was succeeded by a Carpinus-dominated forest, in which Abies 
also became important at higher altitudes. A second expansion of Quercus-dominated 
forest followed, with the final phase of this period being marked by the appearance of 
Pinus and Betula, accompanied by a slight decrease in forest cover. This was 
interpreted by Tzedakis (1991) as indicative of warm, 'Mediterranean' conditions, 
followed by an increase in moisture availability. 
Unpublished high resolution pollen analysis at 20cm intervals through this period of 
284 (appendix C, courtesy of P.e. Tzedakis) confilms this vegetational succession. 
Quercus frequencies increase markedly just below 96m, followed by an expansion of 
Mediterranean evergreen elements. Carpinus values reach a maximum at about 93m. 
The other main features of the Metsovon interglacial are present (such as the second 
increase in Quercus and Pinus, with Betula achieving more importance towards the end 
of the interglacial), but with considerably more detail. This is now discussed in 
conjunction with the other proxies analysed through this interval. 
A comparison of the stable isotope measurements, carbonate and organic matter 
content, magnetic susceptibility and selected pollen data is illustrated in fig.7 .2. A 
summary pollen diagram for this interval is presented in appendix e. As in section 
4.2.1, the plot is split into arbitrary zones for the purposes of description. The 
contribution that this sequence offers to the debate concerning climatic variability within 
the Eemian is discussed further in section 8.5.1. 
100-98.5m The end of the penultimate glaciation appears to have been a time of 
unstable climate. Several indicators show distinct variations across this interval. The 
values of (5 180 oscillate around -3.5%0, although the overall trend is progressively 
negative. This suggests that the PIE ratio during this time was not entirely stable and 
that the arid (low PIE) conditions at the end of the glacial. phase were gradually 
becoming wetter. Over the same interval, (5I3C values also vary in the same way, with 
a positive oscillation occurring at the same time as in the oxygen record (-99.5m). This 
suggests that the residence time of the lake waters was probably changing in response 
g 
..c 
Q.. 
Q,) 
Cl 
Organic Matter Content (%) oI3 C (%0) AP (%) 
o 5 10 15 20 -2 -I 0 I 2 3 4 o 20 40 60 80 100 
83 ] .~ I·S 7 ~ :;; ~ "-<:; )' [ 
84  ; ~ 
85 
86 
87 
88 
89 
90 
91 
92 
93 
94 
95 
96 
97 
98 
99 
100 
101 
102 > ~ \ ,~ ,< 
o 10 20 30 
Carbonate Content (%) 
40 
-I 0 2 3 
Magnetic susceptibility 
(x!()7 m3kg- l ) 
4 -7 -6 -5 -4 
8, 80 (%0) 
-3 -2 o 500000 
Pollen Concentration 
(grains crrr3) 
Fig. 7.2 Summary plot covering Eemian section of core. Pollen concentration and AP% curves from Tzedakis (unpubl. data) . See text for zonation scheme. 
83 
84 
85 
86 
87 
88 
89 
90 
91 
92 
93 
94 
95 
96 
97 
98 
99 
100 
101 
102 
g 
..c 
Q.. 
Q,) 
Cl 
~ 
~ 
:::tl ;g 
~ 
~ 
::j 
a 
~ 
~ 
~ 
V:l 
~ 
~ ~ 
• 
-.l 
-.l 
INTERPRETATIONS AND SYNTHESIS. 178 
to the variablility in PIE. Total carbonate content also varies in the same way as the 
isotopes, with a matching positive oscillation, perhaps indicative of variable 
productivity levels. The vegetation record (appendix C) shows a small increase in 
arboreal pollen (especially Quercus, Carpinus and Corylus) and a slight decline in 
grasses and Artemisia. This confirms that the open vegetation conditions prevalent 
during the penultimate glaciation were coming to an end. 
98.5-96.0m This zone is characterised by marked variability in all indicators. Both 
the oxygen and carbon isotope values oscillate sharply, beginning with a major negative 
excursion, followed by a positive excursion of almost the same amplitude and then a 
slight negative shift at the very end of the zone. In fact, the 81BO values undergo a 
subdued second oscillation at the end of the zone, which is not reflected velY well in the 
8 I3C record. However, it is seen (as an inverse relationship) in the total carbonate 
curve, which itself records the whole oscillation. These data suggest that an initial 
increase in PIE at the beginning of the interval suddenly undergoes at least one, or 
possibly two oscillations before the end of the zone. This variability at the beginning of 
the Eemian has been noted in other high-resolution sequences, for example, the 
un smoothed curve of marine core V19-30 (Shackleton et al., 1983), a Danish 
foraminiferal and isotopic record (Seidenkrantz, 1993) and Norwegian speleothem 
records (Lauritzen, 1990; 1995). The oscillation has now been termed the Zeifen­
Kattegat Oscillation (ZKO) after the warm Zeifen interstadial (first recognised from 
Zeifen in southern Germany) and the Kattegat stadial (first recognised from the Anholt 
IT borehole from Denmark) (Seidenkrantz et al., 1996). The ZKO also manifests itself 
in the vegetation record from 284 over this interval, with a resurgence of grasses and a 
peak in the alga Pediastrum (appendix C). 
96.0-90.90m Values of 8I3c remain at a fairly constant level throughout (around 
-0.8%0), whereas values of 81BO gradually decrease by approximately 0.3%0. 
Carbonate content is slightly more variable, averaging around 20%, whereas magnetic 
susceptibility is reduced across the entire interval. These data suggest that this was a 
relatively stable period, with conditions of high PIE and productivity being maintained 
throughout. The complete absence of molluscan fragments and the low abundances of 
ostracods suggests relatively deep lake-levels. This zone is interpreted as the 
establishment of full, stable Eemian conditions, a conclusion supported by the pollen 
signal (appendix C). The APINAP curve stabilises and the vegetational succession is 
characterised by the rise in Quercus (deciduous and evergreen), Carpinus, Osttya, 
Abies, Alnus and other taxa at the expense of Artemisia, grasses and chenopods. 
INTERPRETATIONS AND SYNTHESIS. 179 
90.90-86.10m A fall in 8 13C at the beginning of the zone is mirrored by a fall in 
8180, albeit slightly delayed. Values of 8 13C rise slightly during the rest of the interval, 
only falling slightly at the velY end of the zone. Values of 8180 remain fairly constant 
(around -5%0), although the very end of the zone sees a fall (again, slightly lagging 
behind the 8 13C). Both curves record a distinct drop at 88m (but just over a single 
point). The fact that this is not obviously reflected in any of the other indicators 
suggests that it is a spurious point, probably related to errors derived from preparation 
or measurement procedures. Carbonate content varies markedly over the entire interval, 
with a noticeable positive oscillation towards the end of the zone. Magnetic 
susceptibility increases marginally at the beginning of this interval. 
These data suggest that this was a period of increasing instability. The decrease in 8180 
suggests slightly wetter conditions (higher PIE), athough it is accompanied by an 
increase in 8 13C, suggesting that residence times were increasing. This apparently 
anomalous situation can be explained if seasonal factors (for example, wetter winters 
and drier summers) became more important, possibly accompanied by a slight drop in 
temperatures. Faunal evidence suggests that the lake-level was shallowing gradually 
(an increase in sub-littoral Candona ostracod species is seen), also confirmed by the 
increase in 8 13c. The oscillations of the carbonate curve are interesting, particularly as 
the magnitude of the variablity is not reflected as clearly in most of the other indicators. 
However, a drop in carbonate content does coincide with the decrease in 813C at the end 
of the zone. This probably indicates a slightly higher lake-level, rendering the site of 
284 once more too deep for many of the benthic ostracods that contribute to the overall 
carbonate content of the sediment. The pollen record also suggests that conditions were 
becoming slightly cooler and more variable over this interval, patticularly noticeable in 
the APINAP curve (fig. 7.2 and appendix C). Artemisia, grasses and chenopods all 
increase gradually and Betula reappears. Carpinus, Ulmus and Ostrya on the other 
hand, all decline. 
86.10-83.0m This final zone sees a return to the extreme variability seen at the 
beginning of the sampled interval. Above 84m, marked (though irregular) increases are 
seen in values of 813C and 8 180 to values compat'able to those recorded at the end of 
MIS-6. This is accompanied by a rise in magnetic susceptibility and distinct crashes in 
the carbonate content and the APINAP curve. Data suggest an inegular resumption of 
cool and dry stadial conditions, confirmed in the pollen record by oscillations in the 
APINAP values (fig. 7.2 and appendix C). 
INTERPRETATIONS AND SYNTHESIS. 180 
7.5.2 MIS-5d - Last Glacial Maximum 
The period stretching from the end of the Eemian (approximately 116ka) to the LGM 
(approximately 20ka) has not been studied in detail during this study. Nevertheless, 
some meaningful comments can still be made. In common with evidence derived from 
long European pollen sequences, ice cores and the marine oxygen isotope record (all 
discussed at length in chapter 8), the Ioannina sequence displays distinct variability over 
this interval (see also the review by van Andel and Tzedakis, 1996). The carbonate 
curve oscillates several times (between approximately 5-10%) suggesting variable levels 
of productivity, although this is likely to be influenced by the re-appearance of 
molluscan material in the sediment above about 70m. The molluscan faunal evidence 
itself suggests an oscillating lake-level, but with an overall trend towards 
shall owing. Above about 45m, occasional beds of the freshwater mussel Dreissena 
appear in the record, indicating very shallow lake-levels (probably in the order of a few 
metres). These culminate in substantial shell beds at around 26m, coinciding with the 
part of the record interpreted by the age model as representing the LGM. Conditions at 
Ioannina during this time inferred from pollen data were thought to have been distinctly 
cold and arid, with average summer and winter temperatures as much as 10°C lower 
than present-day (Bottema, 1974; Vita-Finzi, 1978). The Ioannina 249 pollen record 
suggests a desert-steppe environment, the vegetation being dominated by open, 
discontinuous grasses and Artemisia (Tzedakis, 1991, 1994). 
7.5.3 Last Glacial - Holocene Transition 
Detailed isotopic and faunal analyses were carried out over the top 25m of the core, 
interpreted by the age model to extend from the end of the last glacial period, through 
the Lateglacial and into the Holocene. Isotopic measurements were derived from 
ostracod shells at 20cm intervals, carbonate and organic content was measured every 
lOcm and magnetic susceptibility was measured every 1m (fig 7.3). Radiocarbon 
determinations (section 5.3) also helped to constrain the chronology for this section of 
core. 
25.0-16.80m Values of carbonate content and organic matter both increase slightly 
above about 23.50m, after which the organic matter remains at a relatively constant 
level, whereas the carbonate content fluctuates and begins to rise at the end of the zone. 
Values of 0180 undergo a gradual oscillation, decreasing from values of around 0%0 at 
24m to a minimum of around -2.5%0 at 20m, followed by an increase to about -1 %0 at 
the end of the zone. Values of ol3C follow a similar trend, although they also display a 
E 
.c 
C. 
<l.l 
Cl 
o t'1t.ti!.:.6.t '!.t 1 
2 
3 
4 
5 
6 
7 
8 
9 
10 
11 
12 
13 
14 
15 
16 
17 
18 
19 
20 
21 
22 
23 
24 
25 
..... 
.... . .... 
......... 
@ 1,480±45 / 
1,915±45 
iiiiilil!iiiily'~"5J'10f5d 
.. . . ..... 
... . ..... 
. ....... . 
... ... . . . 
......... 
......... 
. . ....... 
.. . ...... 
. .. . ..... 
... ...... 
......... 
......... 
@ 7,505±60 
::::::::::::::::: ,. Q;)'lO':08 Of70' 
?!\!\/!\/' @ 15,330±140 
.. ....... 
......... 
, . . ..... . 
::»» @ 20,760±230 
......... 
. ........ 
o 
o 10 20 30 
Carbonate content (0/0) 
Organic Matter (%) 
5 10 15 20 
40 -5 -4 -3 -2 
8 13 C (%0) 
-3 -2 -1 
-I o 
818 0 (%0) 
o 
-J 
2 3 , 4 
0 2 
Magnetic Susceptibility (S I units) 
Fig. 7.3 Summary plot for top 25m of core 284. Zonation explained in text. Raw/smoothed isotope data vs V -PDB. Units for magnetic susceptibility: x 10-7 m3kg- I, 
0 
2 
3 
4 
5 
6 
7 
8 
9 
10 E 
~ 
1 I .c C. 
<l.l 
12 Cl 
13 
14 
15 ~ 
16 t;l ::.;, 
17 "'i:l ::.;, 
t'l'j 
18 ~ 
19 :::l 
0 
20 ~ 
21 ::t.. 
22 ~ 
23 ~ 
24 ~ 
::tj 
25 ?;; 
3 ...... ~ 
• 
00 
INTERPRETATIONS AND SYNTHESIS. 182 
marked negative oscillation at about 22m, where values drop from around -2.5%0 to 
almost -5%0, before recovering again. Magnetic susceptibility values remain 
approximately constant across this interval. 
Data suggest that in general terms, the lake-level was recovering after its low-stand at 
the LGM. Conditions became wetter (higher PIE) and slightly warmer, although the 
isotopic evidence suggests that aridity (low PIE) became impOltant again towards the 
end of the zone. The increase in lake-level is also reflected by the ostracod fauna, 
which shows a succession of species throughout the interval that have a preference for 
increasingly deeper water habitats. The anomalous negative excursion of the (5I3C 
record at about 22m is interesting, particularly as it is not accompanied by any 
noticeable change in the (5 180 record. It is possible that this reflects a period of 
increased organic productivity as rising lake-levels ensured that the site of core 284 
became sub-littoral once more (it is assumed that sub-littoral rather than deeper water 
depths would provide the most suitable conditions for this). From their examination of 
a series of eastern Mediterranean pollen records, van Zeist and Bottema (1982) infer 
that this was a time of fluctuating humidity, marked at Ioannina by an increase in 
precipitation to around 18ka BP, after which it began to decrease again. Until the 
radiocarbon chronology for this part of the Ioannina 284 record is properly corrected 
for reservoir effects (section 5.3), it is not possible to interpolate an age for the point at 
which the (5 180 curve achieves its minimum value over this zone (reflecting maximum 
PIE). Nevertheless" with caution it would appear that the reduction in PIE occurs at 
approximately the right level in the 284 sequence to support this interpretation. 
The increase in carbonate content at the end of the zone possibly reflects an increase in 
productivity at the beginning of the Lateglacial interstadial (l4-10ka BP). Again, this 
would agree with van Zeist and Bottema (1982), who suggest that this period was 
characterised by markedly increased humidity levels, implying increased temperatures, 
coupled with increased evaporation and precipitation. Preliminary pollen data from this 
part of 284 supports this interpretation, since Quercus and Corylus begin to increase in 
frequency at the same time as Artemisia, grasses and chenopods all begin to decline 
(P.e. Tzedakis, unpubl. data). 
16.80-1S.0m Carbonate content continues to increase, although experiences a 
negative oscillation at 16m. Values of organic matter maintain a constant level for most 
of the interval, before falling suddenly at the end of the zone. The (5 180 record shows a 
sharp increase across the interval (to a maximum of +3%0), before declining again at the 
end of the zone. By contrast, the (513C record shows a steady decrease throughout. 
Values in magnetic susceptibility experience a slight reduction. Isotopic resolution 
INTERPRETATIONS AND SYNTHESIS. 183 
across this zone drops relative to the rest of the sequence due to low ostracod 
abundances (they experience a minimum between 16-15m) and the loss of several 
samples due to machine failure. 
Data suggest that this was a period of low PIE, indicating cool, arid conditions. The 
lake-level would have reduced, allowing the site of core 284 to become sub-littoral once 
more. This explains the apparently anomalous behaviour of the ol3C, since although 
lake-levels were falling, water depths became more suitable for aquatic productivity, 
recorded as a decrease in ol3C (due to more decay at depth). The ostracod abundances 
show minimum values across this interval, apparently disagreeing with this 
interpretation. However, examination of the faunal data shows that overall preservation 
across this interval was poor and that only valves of Candona pennanenta were 
recovered. This is thought to be a deeper-water benthic species that would become 
stressed in a habitat of rapidly shallowing water depth and a increased levels of aquatic 
vegetation. 
It has been suggested (section 5.8) that this zone may include the Younger Dryas (YD), 
known in the eastern Meditenanean to have been a brief, arid cold phase between 11-
lOka BP (e.g. Rossignol-Strick, 1995). As yet, no unequivocal record of this stadial 
exists from Greece and without the appropriate corrections to the radiocarbon 
chronology of 284 (section 5.3), it is not possible to confidently ascribe the oscillation 
evident from the isotopic record to the YD. In addition, the preliminary pollen record 
from 284 is not cunently of sufficient resolution to detell'nine its possible existence. 
Nevertheless, as the maximum extent of this arid, cool event in 284 occurs at 15.20m, 
just below the radiocarbon age of lO,080±70 yrs BP at 15.lOm, it is likely to be the 
prime candidate. 
lS.0-9.0m The beginning of this zone is characterised by a crash in values of ol3C, 
0180 and carbonate content. In contrast, the percentage of organic matter in the 
sediment increases sharply and magnetic susceptibility values also display a slight 
increase. For the rest of the zone, values of ol3C and 0 180 increase steadily despite 
several oscillations (most noticeably at 13m and lOm). Organic matter content remains 
at around 15% until towards the end of the zone, when values begin to decline. 
Carbonate content values remain low for some time, not achieving maximum values 
(-30%) until approximately 11.5m. Initial magnetic susceptibility values show a slight 
increase, then decline to a minimum at approximately 11.5m, before gradually 
increasing again towards the end of the zone. 
INTERPRETATIONS AND SYNTHESIS. 184 
The data suggest that PIE increased rapidly at the beginning of the zone, causing the 
lake-level to rise and (possibly) washing increased levels of nutrients into the lake 
(reflected by enhanced levels of organic matter in the sediment). The response of the 
carbonate curve is interesting in that it initially crashes rather than increases, as might be 
expected. This may represent a period when the rapidly increasing lake-level ensured 
that the site of 284 became too deep for most aquatic mollusc an or ostracod fauna. 
Analyses reveal only a few molluscan fragments and a sparse Candona pennanenta 
ostracod fauna between approximately 15-12m, indicative of deeper water conditions. 
As time progressed, lake-levels began to drop gradually, indicated by the positive 
isotopic trends and the recovery of the carbonate content, retuming the site of 284 to 
optimum conditions for molluscan and ostracod faunas. This situation is also reflected 
in the isotopic trends; from approximately 11-9m, values of 0180 continue to increase, 
whilst at the same time, values of 013C undergo a negative oscillation. This may 
represent a similar situation to that occurring in the preceeding zone, in which organic 
productivity continues to be maintained at the same time that PIE reduces over a 
relatively short-lived interval. This may also explain a similar oscillation in the isotope 
curves at around 13m. 
The age model suggests that this zone represents the early part of the Holocene, 
agreeing with the interpretation above. In northem Greece, the beginning of this period 
was characterised by an increase in both precipitation and temperature (e.g. van Zeist 
and Bottema, 1982), reflected in the vegetational records by a rapid expansion of 
Quercus woodland, accompanied by increased frequencies of Ulmus, Tilia and Corylus 
and followed at Ioannina by a slightly later rise in Ostrya (e.g. Bottema, 1974; 
Tzedakis, 1994). Pollen records from the eastem Mediterranean also record a distinct, 
brief, cool event over this period, variously dated between 8-7.5ka BP and 
characterised by the abrupt disappearance of Pistacia (see review by Rossignol-Strick, 
1995). It is also likely that this event is equivalent to the (global?) short-lived cool and 
dry event detected in several ice core records and other marine and telTestrial sequences 
that has been dated at around 7.5ka BP (Alley et al., 1997). This may be reflected at 
Ioannina by the oscillation in the isotopic curves at approximately 13m. Interpreted as a 
brief, arid event (see above), it appears to match the requirements in telms of duration 
and stratigraphical position. However, as has been pointed out previously, until the 
radiocarbon stratigraphy has been fully resolved in telms of reservoir effects, any 
potential correlation must be treated with caution. 
It is possible that the upper oscillation in the isotopes (at 9.5m) represents a similar 
cooling event that is slightly longer and more severe in nature than the one at 13m. 
Although it does not appear correspond with any clear event known from other Balkan 
INTERPRETATIONS AND SYNTHESIS. 184 
The data suggest that PIE increased rapidly at the beginning of the zone, causing the 
lake-level to rise and (possibly) washing increased levels of nutrients into the lake 
(reflected by enhanced levels of organic matter in the sediment). The response of the 
carbonate curve is interesting in that it initially crashes rather than increases, as might be 
expected. This may represent a period when the rapidly increasing lake-level ensured 
that the site of 284 became too deep for most aquatic mollusc an or ostracod fauna. 
Analyses reveal only a few molluscan fragments and a sparse Candona permanenta 
ostracod fauna between approximately 15-12m, indicative of deeper water conditions. 
As time progressed, lake-levels began to drop gradually, indicated by the positive 
isotopic trends and the recovery of the carbonate content, returning the site of 284 to 
optimum conditions for molluscan and ostracod faunas. This situation is also reflected 
in the isotopic trends; from approximately 11-9m, values of 8180 continue to increase, 
whilst at the same time, values of 813C undergo a negative oscillation. This may 
represent a similar situation to that occurring in the preceeding zone, in which organic 
productivity continues to be maintained at the same time that PIE reduces over a 
relatively short-lived interval. This may also explain a similar oscillation in the isotope 
curves at around 13m. 
The age model suggests that this zone represents the early part of the Holocene, 
agreeing with the interpretation above. In northern Greece, the beginning of this period 
was characterised by an increase in both precipitation and temperature (e.g. van Zeist 
and Bottema, 1982), reflected in the vegetational records by a rapid expansion of 
Quercus woodland, accompanied by increased frequencies of Ulmus, Tilia and Corylus 
and followed at Ioannina by a slightly later rise in Ostrya (e.g. Bottema, 1974; 
Tzedakis, 1994). Pollen records from the eastern Meditenanean also record a distinct, 
brief, cool event over this period, variously dated between 8-7.5ka BP and 
characterised by the abrupt disappearance of Pistacia (see review by Rossignol-Strick, 
1995). It is also likely that this event is equivalent to the (global?) short-lived cool and 
dry event detected in several ice core records and other marine and tenestrial sequences 
that has been dated at around 7.5ka BP (Alley et al., 1997). This may be reflected at 
Ioannina by the oscillation in the isotopic curves at approximately 13m. Interpreted as a 
brief, arid event (see above), it appears to match the requirements in terms of duration 
and stratigraphical position. However, as has been pointed out previously, until the 
radiocarbon stratigraphy has been fully resolved in terms of reservoir effects, any 
potential correlation must be treated with caution. 
It is possible that the upper oscillation in the isotopes (at 9.5m) represents a similar 
cooling event that is slightly longer and more severe in nature than the one at 13m. 
Although it does not appear correspond with any clear event known from other Balkan 
INTERPRETATIONS AND SYNTHESIS. 185 
records, Huntley and Prentice (1993) suggest that changes in the seasonality of the 
climate were occurring at around 6ka BP, reflected by significantly cooler winters and 
warmer summers and represented in the pollen record by the spread of taxa such as 
Cmpinus orientalis/Ostrya (e.g. Bottema, 1967, 1974; van Zeist and Bottema, 1982; 
Willis, 1994). The variable seasonality in climate might be responsible for the overall 
decrease in PIE seen at Ioannina (more positive values of ( 180), at the same time as an 
apparent increase in productivity (more negative values of o13C). Another possibility is 
that this oscillation is related to anthropogenic impact, since Neolithic agriculture is 
thought to have become established in the Balkans between 8-6ka BP (e.g. Whittle, 
1985). Higgs et al. (1967) note that taxa such as Carpinus orientalis and Ostrya 
carpinifolia can be observed today in areas of recent deforestation and suggest that their 
appearance in Bottema's (1967) pollen record may be linked with the mTival of 
Neolithic pastoralists and/or cultivators. They also cite the occurrence of a Neolithic 
site at the northern end of the basin as support for this viewpoint. However, widescale 
disturbance of the catchment is not generally thought to have begun until around 5-
4.5ka BP (discussed in more detail below) and the oscillation at this level is 
accompanied by relatively low magnetic susceptibility values (rather than the higher 
values that might be expected if landscape clem'ance was being undertaken). The nature 
of this event is therefore unclear from the data currently available and may only be 
resolved once detailed pollen analysis across this interval has been completed. 
9.0-0m This final zone begins with mm'ked negative oscillations in values of o13C, 
0180, carbonate and organic matter contents. Magnetic susceptibility on the other hand, 
oscillates in a positive fashion. The duration of this event (between 8.5-7 .5m) 
coincides with the first noticeable change in lithology of the core, from the typical olive­
green clayey silt of the vast majority of the sequence, to a dark greyish-brown 
calcareous silt that is rich in tiny shell fragments (particularly Bithynia opercula). 
Values of all indicators return to near-previous levels immediately after the oscillation, 
although from about 6m the values of carbonate and organic mattter content decline 
markedly, whilst values of o13C, 0180 and magnetic susceptibility all increase. 
The data can be interpreted as reflecting anthropogenic influence within the catchment. 
As mentioned above, landscape clearance is thought to have become widespread in the 
region from around 5-4.5ka (e.g. Bottema, 1982), which would be in approximate 
accordance with the existing radiocarbon chronology, even after making suitable 
allowances for reservoir effects. The increase in magnetic susceptibility (pmticulm'ly 
after about 6m), the oscillations in the isotopic, carbonate and organic matter curves and 
the changes in lithology all suggest increased inwash (coupled with a possible 
shallowing of the lake), probably due to soil degradation and instability occurring as a 
INTERPRETATIONS AND SYNTHESIS. 186 
result of landscape clearance. The pollen record from Ioannina over this period records 
an increase of the evergreen oak (Quercus coccifera), along with the re-appearance and 
increase of Fagus, Abies, Olea, Vitis and Ericaceae (Bottema, 1967, 1974). Higgs et 
al. (1967) interpret this as evidence for the clearance and agricultural exploitation of the 
landscape by man. 
7.5.4 Response Times 
One of the key assumptions in presenting a synthesis of this kind, is that proxies such 
as vegetational, faunal and chemical records all respond to climatic changes in a quasi­
synchronous fashion. This is largely due to the discontinuous nature of the sampling, 
which constrains the time period represented by each slice of sediment analysed and, of 
course, the time period represented by the intervening slice of sediment that is not 
analysed. This is complicated by varying accumulation rates. For example, a 1cm 
thick sample from a glacial interval will represent a much longer period of time than a 
lcm thick sample from an interglacial interval. With coarser sampling resolution, 
natural systems that normally respond at different rates will appear to be effectively 
synchronous (for example, compare the response times of beetle faunas with that of 
vegetation). Only progressively finer resolution can resolve some of the leads and lags 
between different proxies, although there are obvious limits to the resolution 
achievable, depending on the accumulation rates and thicknesses of sediment in 
individual sequences. 
Detailed consideration of the response times of different natural systems to climatic 
change is beyond the scope of this report. An excellent review is provided by Wright 
(1984), who discusses the dynamics of physical systems (including oxygen isotope 
ratios derived from ice records, the terrestrial glacial record and lakes) and biological 
systems (including vegetation, tree rings, insects and veltebrate faunas) and their 
sensitivity to climatic variablility. Nevertheless, several broad observations can still be 
made concerning the data derived during this study. 
The sediment accumulation rates in the Ioannina 284 sequence were generally very high 
in comparison to other records (for example, most marine sequences), providing the 
opportunity for high resolution comparison of response times. The average 
accumulation rate over the entire length of the core was approximately O.5mka-', 
although it was typically higher during interglacial periods arid lower during glacial 
periods (sections 5.3 and 5.8). During the Eemian, for example, the accumulation rate 
was around double the average. The isotopic signals of bulk carbonates were measured 
INTERPRETATIONS AND SYNTHESIS. 187 
evelY 10cm, corresponding to a resolution of approximately one sample every 100 
years. The carbonate and organic matter contents were also sampled at the same 
resolution. From fig. 7.2, it would appear that the ol3C record is more sensitive than 
the 0180 record, as indicated by the swifter decline in values at around 91m and again at 
around 86m. The 013C record is thought to be controlled by residence time of the lake 
water and the opportunity to exchange with atmospheric CO2, which is turn linked to 
the PIE ratio that controls the 0180 record. Since the carbonates precipitate in the 
surface waters of the lake, they are likely to be much more sensitive to the variablity in 
time available to exchange with atmospheric CO2, than changes in the overall PIE ratio 
for the catchment. 
The pollen record for most of the Eemian (P.e. Tzedakis, unpuhl. data) was sampled at 
approximately 20cm intervals (half the resolution of the isotopes), making direct 
comparisons between response times difficult. Further detailed sampling over clUcial 
intervals will therefore be necessary before any balanced assessment can be made, 
although even then there are other factors that must be taken into account. For example, 
a pollen spectlUm represents the complex interplay of a multi-variate vegetation system 
with its own internal dynamics and feedback mechanisms, whereas an isotope curve 
represents a single variant. Comparison of the isotopic record with a simple APINAP 
curve is likely to be unsatisfactory on all but the most simplistic level, so other criteria 
must also be considered, such as primary succession and the occurrence of pioneer 
species (e.g. Wright, 1984; Tzedakis, 1991). 
Similar comparison of response times between proxies is more difficult for the top 25m 
of 284. The resolution of the isotopic signals (derived from ostracod shell calcite) is 
coarser (20cm intervals), corresponding to one sample approximately evelY 160 years . 
In addition, the primary controls governing the ol3C and carbonate curves in particular 
are much more complex than during the Eemian. Nevertheless, the high accumulation 
rates during this and other interglacial periods of 284 suggest that the sequence has 
great potential for future high resolution analysis of the sensitivity of proxies to climatic 
change. 
7.6 SUMMARY 
This chapter has attempted to interpret, synthesise and discuss the results obtained and 
presented previously. Sediment analyses were dicussed initially. The origins and 
controls on carbonate production within a lake such as Pamvotis were outlined and it 
1 
INTERPRETATIONS AND SYNTHESIS. 188 
was concluded that in the case of the 284 sequence, the carbonate fraction of the 
sediment was dominantly authigenic when lake-levels were relatively deep. Only when 
lake-levels shallowed did bioclastic carbonate material become a problem. Therefore, it 
was suggested that the carbonate curve would probably be a reasonable proxy for lake 
productivity below a depth of around 145m; above this level, the intermittant 
appearance of molluscan shell fragments indicated a shallower lake, rendering any 
interpretation of the carbonate curve difficult. 
The origin and controls on organic matter production and eventual preservation in the 
sediment was also discussed. It was pointed out that since a significant amount of 
organic matter produced in the lake system was destroyed by a variety of processes in 
the water column and sediment, that any correlation with primary lake productivity was 
likely to be very weak. In view of this, caution was advised regarding any 
interpretation of all but the broadest variations in the organic matter content of the 
sediment. 
Magnetic susceptibility profiles were also considered. It was concluded that the curve 
derived from the Ioannina sequence was probably a reasonable indicator of inwash 
from the eroded catchment, reflecting broad climatic changes. The origins of the 
magnetic carrier (maghaemite/magnetite) through those intervals of the core that 
displayed extremely high susceptibility values was thought to be related to pedogenic 
enhancement of the catchment soils. More data are required before it is known whether 
this carrier is responsible for the magnetic susceptibility signal in the rest of the 
sequence. 
Analyses of the faunal content of the core were then discussed. The use of molluscan 
and ostracod material as lake-level indicators was considered to provide meaningful 
relative estimates on the status of the lake at different times. An example was cited of 
the usefulness of this technique with respect to the 'Kastritsa cave problem' , where 
molluscan evidence from this study indicated low lake-levels during the LGM, in 
agreement with pollen records from the region. The reason for the intepretation of high 
lake-levels from the Kastritsa data in previous studies was attributed to workers not 
accounting for tectonic movements in the basin since the LGM. 
This was followed by detailed discussion of the chemical analyses of the core material. 
A classification of the modern lake waters from chemical measurements was proposed, 
concluding that Pamvotis is cUlTently a eutrophic, fresh-water temperate hard-water 
lake, with waters that have had an opportunity to develop an evaporated isotopic 
signature. The controls on the stable oxygen and carbon isotopes derived from bulk 
INTERPRETATIONS AND SYNTHESIS. 189 
carbonates were then outlined. Interpretations of the data from the lOO-83m interval of 
core demonstrated that ()ISO was being dominantly controlled by variations in PIE, 
whereas () 13C was being dominantly controlled by variations in the residence time of the 
lake waters. A similar situation was thought to be operating in the top 25m of the core, 
where isotopes had been measured from ostracod shell calcite. Here, whereas the ()I SO 
curve was again being controlled by variations in PIE, the interpretation of the () 13C 
curve was more difficult, probably due to the shallow lake-levels over this interval and 
biological considerations of using ostracods as a source of calcite, rather than bulk 
material. It was thought that () 13C was probably dominated by the influence of 
respiration/decay at depth (linked to surface water productivity), but that residence time 
I the opportunity to exchange with atmospheric CO2 may also have played a significant 
role. 
This section concluded with a discussion on the open/closed status of the lake in the 
past and the present -day. It was demonstrated that the lake has been an effectively 
closed system, probably for much of its life. However, it was pointed out that caution 
should be exercised in basing any conclusions on the status of lakes on purely chemical 
evidence without hydrological factors being taken into account. 
The final part of the chapter provided a detailed sythesis of all data, particularly through 
those parts of the sequence studied at high resolution, namely the Eemian and the last 
glacial - Holocene transition. There appeared to be good correspondance between the 
Ioannina record and other climatic proxies from the region (particularly pollen records). 
The evidence for climatic instability during the Eemian was presented (this is discussed 
further in chapter 8). In addition, the possibility that the Younger Dryas has been 
recorded at Ioannina was also discussed, although until the radiocarbon chronology is 
corrected for reservoir effects and further pollen and isotopic data become available (all 
ongoing), then any correlation remains extremely tentative. 
A discussion of response times of different proxies to climatic change concluded the 
chapter, with the suggestion that ()I3C may be more sensitive to climatic change than 
()ISO, probably related to the controls on each of these indicators. Detailed comparison 
of the response times between other proxies was considered difficult at this stage, 
although the potential of the sequence for future high-resolution investigations of this 
kind was highlighted. 
8. COMPARISON OF IOANNINA 284 WITH 
OTHER LONG RECORDS 
8.1 INTRODUCTION 
The Ioannina 284 sequence provides a continuous record of climatic change over 
multiple glacial- interglacial cycles, possibly extending back over 600,000 years . The 
purpose of this chapter is to assess briefly the relative importance of the sequence with 
respect to some of the other major records that span multiple climatic cycles. These 
include the long European terrestrial records, the marine oxygen isotope record and the 
Greenland ice cores. This review is not intended to be exhaustive, but it is hoped that 
the main similarities and differences can be highlighted and correlations proposed where 
appropriate. 
8.2 TERRESTRIAL SEQUENCES 
Several well-known long European sequences exist which extend back over more than 
one climatic cycle (fig 1.1), at least two of which have already been mentioned 
(loannina 249 and Tenaghi Philippon). Although some studies have concentrated on 
climatic proxies, such as insect remains and mineral magnetic variability (both of which 
are discussed below), most are palynological. There are no long European sequences 
that have been extensively studied with respect to isotopic data, whether derived from 
bulk carbonates or biological material. 
Analysis of the way in which vegetation communities change in an area through time is 
the principal method by which Quaternary palaeoenvironments are reconstructed from 
the terrestrial record. The chronology for 284 has relied extensively on the pollen­
derived stratigraphy from core 249 (section 5.8). Since it has been possible to correlate 
both cores via their magnetic susceptibility profiles, the comparisons that Tzedakis 
(1991, 1994) made between 249 and other long European pollen sequences can, with 
some caution, now be applied to 284. 
COMPARISON WITH OTHER RECORDS. 191 
Tzedakis (1991, 1994) demonstrated that Ioannina 249 provided a record of changing 
vegetation communities in the region that extended back over 420ka. After discussing a 
rigorous strategy by which long pollen sequences could be compared, he was able to 
correlate the 249 record with Tenaghi Philippon from northeastern Greece (Wijmstra, 
1969; Wijmstra and Smit, 1976; van der Wiel and Wijmstra, 1987a, 1987b); Valle di 
Castiglione in central Italy (e.g. Follieri et al., 1988; Narcisi et al., 1992); and the 
French sequences of Grande Pile (Woill ard , 1978; de Beaulieu and Reille, 1992) and 
Les Echets (e.g. de Beaulieu and Reille, 1984, 1989). The correlation scheme was later 
refined by reference to the marine oxygen isotope record and extended to include the 
latest and longest sequences from Lac du Bouchet (Reille and de Beaulieu, 1990; de 
Beaulieu and Reille, 1995) and Praclaux (Reille and de Beaulieu, 1995), both from the 
Massif Central region of France (Tzedakis et al., 1997). These sequences are all now 
briefly described in turn, before suggesting a basis on which tentative correlation with 
284 can be made. 
(a) Tenaghi Philippon. Located in part of the tectonic ally controlled Drama Basin 
in northeastern Greece, Tenaghi Philippon is the longest terrestrial palynological record 
from Europe. The pollen stratigraphy from two cores suggest that the sequence extends 
back into the Lower Pleistocene (Wijmstra, 1969; Wijmstra and Smit, 1976; van der 
Wiel and Wijmstra, 1987a, 1987b). Chronology is provided by a series of radiocarbon 
dates for the upper prut of the sequence (Wijmstra, 1969) and a series of 
palaeomagnetic measurements (unpublished, but referred to by van del' Wiel and 
Wijmstra, 1987). The sequence and its age model have already been described in detail 
(section 5.2.1). 
(b) Valle di Castiglione. This 250ka sequence is derived from an aItificially 
drained volcanic maaI' crater lake, 20km east of Rome in central Italy (Follieri et al., 
1988; Narcisi et aI., 1992). The chronology is provided by a series of 21 radiocarbon 
dates from the upper part of the sequence and by counting annual laminations (Alessio 
et aI., 1986; Magri, 1989). 
(c) Grande Pile and Les Echets. These two extensively studied sequences from 
central France provide detailed palaeoclimatic records extending back at least 140ka 
(Woillard, 1978; de Beaulieu and Reille, 1984, 1989, 1992). Grande Pile is now a peat 
bog in the southern Vosges Mountains; Les Echets is a mire, located 15km northeast of 
Lyon, in the southwestern part of the Dombes Plateau. The chronology is provided by 
a series of radiocarbon dates and on correlations of the APINAP curve with the marine 
oxygen isotope record (e.g. Woillard, 1978; Woillard and Mook, 1982; Guiot et al., 
1989). Generally accepted correlations have been proposed between these two 
COMPARISON WITH OTHER RECORDS. 192 
sequences and Tenaghi Philippon (Woillard, 1978) and Valle di Castiglione (Follieri et 
al., 1988). 
(d) Lac du Bouchet and Praclaux. These two sequences are derived from cores 
taken from two adjacent volcanic craters in the Vel ay Mountains, in the Massif Central 
region of France. The records extend back over 400ka, with chronological control 
being provided by radiocarbon dating and correlation of the APINAP curve with the 
marine oxygen isotope record and other terrestrial sequences (e.g. de Beaulieu and 
Reille, 1995; Reille and de Beaulieu, 1990, 1995) . • 
It should be emphasised once again that the Ioannina 284 sequence does not yet have 
have detailed pollen data available. As explained in section 5.2.2, the correlation with 
249 was based initially on matching magnetic susceptibility profiles, followed by 
comparison of the APINAP curve of 249 with the carbonate content curve of 284 (both 
considered to be first-order proxies for climatic variability). Where limited pollen data 
are available for 284 (for example, across the Eemian interval), then correspondence 
between the two sequences has been good (P.e. Tzedakis, pers. comm.). Correlation 
between 284 and any other terrestrial sequence must therefore be treated as highly 
tentative until detailed pollen-based comparisons are possible. In addition, it must be 
remembered that the 249 record only extends back approximately 420ka, whereas 284 
and Tenaghi Philippon both extend back further. Since the original comparisons 
between various sites (Tzedakis 1991, 1994; Tzedakis et al., 1997) were attempted on 
the basis of matching pollen assemblages for glacial and interglacial phases (rather than 
simple 'wiggle matching'), it would be unwise to extend the correlation between 284 
and Tenaghi Philippon without further pollen data from 284. 
A suggested scheme of correlations is proposed in table 8.1. The comparison with the 
marine oxygen isotope stratigraphy (including the designation of MIS-'9a') is discussed 
in section 8.3, below. Tzedakis (1991, 1994) fOlmally proposed a series of local 
names to describe interglacials and interstadials from the 249 record. These have been 
retained here; local names from each of the other cited pollen sequences have also been 
used. It should be stressed that the correlations between 284 and the other sequences 
will remain tentative until such time as detailed pollen data for 284 becomes available. 
These are by no means the only long pollen sequences from Europe. Other well-known 
examples include Padul in southern Spain (e.g. Florschutz et al., 1971; Pons and 
Reille, 1988), and Lago Grande di Monticchio in southern Italy (e.g. Watts et al., 
1996a,b). Comparison of the 284 record with these and other long European pollen 
Ioannina 284 
(depth, m) 
0-15.0 
64.7-66.2 
78 .1-80.5 
81.2-81.9 
83.9-96.6 
139.9-143.9 
149.9-152.2 
157.2-163.2 
179.3-190.7 
196.2-205.2 (?) 
207.5-213.1 (?) 
217.8-234.6 (7) 
252.8-272.4 (?) 
281.2-308.2 (7) 
Ioannina 249 
Holocene 
Vikos 
Perama 
Thyamis 
Metsovon 
1N-26 
Zitsa 
1N-23a 
Katara 
1N-17 
Pamvotis 
Dodoni I and II 
Grande Pile / Valle di 
Les Echets Castiglione 
Holocene Holocene 
St Gennain II VdC-14 
St Gennain le VdC-12c 
St Gennain la VdC-12a 
Eemian VdC-lO 
Romaill 
Roma II 
Romal 
Lac du Bouchet / 
Praclaux 
Holocene 
St Geneys II 
St Geneys le 
St Geneys la 
Ribains 
Le Bouchet 3 
Le Bouchet 2 
Le Bouchet 1 
Amargiers 
Ussel 
Landos 
Prac1aux and Jagonas 
Tenaghi 
Philippon 
Holocene 
Elevtheroupolis 
Drama 
Doxaton 
Pangaion 
H2-3 Symvolon 
HI Symvolon 
Strymon 
Kavalla 
Krimenes 
Litochoris 
Lekanis 
Marine Oxygen 
Isotope Stage 
5a 
5c 
5c 
5e 
7a 
7c 
7e 
'9a' 
9c 
ge 
11 
13 
15 
Table 8.1 Proposed correlation scheme between loannina 284 and other records. Local stage names for wann periods (interglacials and interstadials) defined at each of the 
terrestrial sites. Use of MIS- '9a' follows Tzedakis et al. (1997). See text for full references and details. (?) denotes uncertain correlation. Table adapted from Tzedakis (1991, 
1994) and Tzedakis et al., 1997. 
Cl 
Cl 
2i:: 
~ 
::tI (;:; 
~ 
~ 
5l 
Cl 
5l ~ 
::tI 
tl:I 
Cl 
Cl 
::tI 
i::::I 
V:! 
• 
...... 
\0 
W 
COMPARISON WITH OTHER RECORDS. 194 
sequences has not been attempted due to the current lack of detailed palynological data. 
Some of these long terrestrial sequences have also been studied with respect to climatic 
proxies other than pollen. Examples include the study of beetle remains from Grande 
Pile (e.g. Guiot et al., 1993; Ponel, 1995) and detailed mineral magnetic studies from 
Monticchio (Creel' and Morris, 1996) and Lac du Bouchet (Williams et al, 1996). No 
work of this nature has yet been catTied out at Ioannina, making direct correlations 
difficult. 
8.3 MARINE RECORD 
The record of global ice volume vat'iability, derived from the oxygen isotopic signal of 
marine foraminifera (tve marine oxygen isotope record) has long been used as the 
yardstick against which most Quatemat-y chronologies are ultimately compared. The 
technique by which the record has been derived is so well established that it requires 
little discussion here, although a good summary is provided by Lowe and Walker 
(1997). The standard reference sequence is often taken to be the orbitally-tuned 780ka 
stacked and averaged record known as the SPECMAP curve (Imbrie et al., 1984). 
Subsequent high-resolution work has provided other orbitally-tuned chronostratigraphic 
timescales, for example, covering the last 300ka (Mattinson et al., 1987) and the 
Middle and Early Pleistocene (e.g. Ruddiman et al., 1986, 1989; Raymo et al., 1989; 
Shackleton et al., 1990). 
Reasonably good correspondence exists between the mat'ine and terrestrial records for 
the last 150ka or so (e.g. Shackleton, 1969; Heusser and Shackleton, 1979; Kukla and 
Briskin, 1983; Mangerud, 1989). Tzedakis et al. (1997) recently tried to extend these 
correlations back to 500ka. They compat'ed four terrestrial records that have reasonably 
well defined chronologies (Ioannina 249, Valle di Castiglione, Tenaghi Philippon and 
Lac du BouchetJPraclaux) with the SPECMAP mat'ine record, using glacial-to-deglacial 
transitions as tie points. Although good agreement between all of the sequences was 
demonstrated, differences in the relative magnitude of events between vegetational 
changes and global ice volume vat'iations became apparent. For example, pollen 
evidence from all the terrestrial sites indicated the presence of substantial forest 
development in MIS-8e. Its similarity with other deglacial substages (e.g. 5a, 5c) was 
noted and it was suggested that MIS-8e should in fact be considered patt of the MIS-9 
interglacial complex. This interval was accordingly re-designated MIS-'9a'. They 
concluded by emphasising the need for a greater understanding of controls on 
. 
-
COMPARISON WITH OTHER RECORDS. 195 
sedimentation rates in basins with different geological settings, before a tighter degree 
of correlation over such a timescale would be possible. 
It is not yet possible to correlate directly the Ioannina 284 record with the marme 
record. Once again, it is only via the correlation with Ioannina 249 and using its pollen­
derived stratigraphy that any suggested comparisons with the marine record are 
possible. Interestingly, Tzedakis (1991, 1994) followed Shackleton (1989) by initially 
correlating the 249 record with a composite physical marine sediment sequence from the 
Pacific, rather than with the SPECMAP curve (discussed in section 5.2.1). However, 
when the 249 record was subsequently included in the 500ka comparison of terrestrial 
and matine sequences (Tzedakis et al., 1997), the correlation was adjusted relative to 
the SPECMAP curve. 
A suggested scheme by which the Ioannina 284 record can be linked with the marine 
isotope record, using the 249 sequence as a 'stepping stone', is outlined in table 8. 1. 
The protocol of Tzedakis et al. (1997) is followed in adopting MIS- '9a'. Once again, it 
must be emphasised that all correlations are tentative, particulat'ly those beyond the 
extent of 249, since these comparisons are based solely on sparse palaeomagnetic 
evidence and similarity of climatic proxy curves ('wiggle matching'). This was 
discussed in detail in section 5.8 and has been illustrated in fig. 5.7. 
8.4 ICE CORES 
A series of ice cores recently drilled in Antarctica and Greenland have demonstrated 
their potential for recording high-resolution climatic signals (e.g. louzel et al., 1990; 
GRIP members, 1992; Grootes et aI., 1993). In patticular, the European Greenland 
Ice-core Project (GRIP) drilled at the ice divide at Summit has provided a detailed 
3000m record that stretches back over the last 250ka. For the top 2,750m 
(corresponding to an age of 103ka BP) it correlates almost exactly with the American 
Greenland Ice Sheet Project core (GISP2), drilled 28km west of Summit (Grootes et 
al., 1993). Below this depth however, there exist marked differences, particulat'ly 
through the Eemian (MIS-5e) section, where a series of short frequency oscillations 
have been atttributed to rapid climate fluctuations (GRIP members, 1992). This is 
discussed in more detail in section 8.5, below. However, both cores record a series of 
rapid climate fluctuations through the last glacial period (e.g. Dansgaard et al., 1993; 
Bond et al., 1993; Andrews et al., 1994), many of which are also recorded in the 
COMPARISON WITH OTHER RECORDS. 196 
Antarctic Vostok core (e.g. Bender et al., 1994; Jouzel et al ., 1990; Jouzel, 1994). 
This is also discussed further in section 8.5, below. 
The ice cores were studied at high-resolution with respect to a range of different 
proxies. For example, the stable isotopic signature of the ice has been related to 
temperature changes (e.g. Johnsen et aI., 1989; Dansgaard et al., 1993; Grootes et al., 
1993); the composition of air bubbles trapped within the ice have been linked to levels 
of atmospheric greenhouse gases such as carbon dioxide and methane (e.g. Dansgaard, 
et aI., 1993); and changes in the electrical conductivity of the ice are thought to reflect 
variations in alkaline dust content, often indicative of cold periods (e.g. Taylor et al. , 
1993). 
Despite broadscale agreement on a first -order basis between ice core, marme and 
terrestrial records (e.g. Jouzel, et al., 1990; Walker, 1995), robust correlations between 
the three are more difficult to achieve, principally because of the lack of a common 
timescale (e.g. Lowe and Walker, 1997). The ice core chronologies are largely derived 
by counting annual layers, whereas the chronologies for the upper parts of the marine 
and terrestrial records are largely based on the radiocarbon timescale. One approach is 
to convert 'ice-layer years' into d4C years' , such as Walker's (1995) comparison of 
climatic trends in terrestrial, marine and ice core data over the last glacial - interglacial 
transition in Europe. Lowe et al. (1995) adopted the reverse approach. They converted 
,
14C years ' to 'ice-layer years' to enable a close comparison to be made between 
palaeotemperature data derived from beetle assemblages with the snow accumulation 
I 
record from the GISP2 core, also over the last glacial - Holocene transition. Yet 
another approach has been to use tephrochronology, for example, to accurately link ash 
layers found in terrestrial and marine sequences around Iceland and in the North Sea to 
ash layers in the GRIP core (e.g. Gronvold et al., 1995; Haflidason et al., 1995). 
Clearly, any attempt to correlate directly the upper part of 284 with the ice core records 
can only be made when the radiocarbon stratigraphy has been properly corrected for 
reservoir effects (section 5.3) . Nevertheless, it is worth noting the broadscale first­
order similarity between the carbonate curve from 284 and ice core records such as 
GRIP, particularly during well constrained intervals further down in the sequence. One 
such interval is the Eemian, precisely located by a combination of magnetostratigraphy, 
U -series dating and pollen stratigraphy. This is discussed further in section 8.5. 1, 
below. Continued high-resolution work on the chronology ·of 284 may eventually 
permit additional correlations with the ice core records. 
COMPARISON WITH OTHER RECORDS. 197 
8.5 RAPID CLIMATE FLUCTUATIONS 
Arguably, the most important outcome of the ice core data has been to highlight the 
evidence for high frequency climatic variablilty, particularly over the last 130ka or so. 
This has not only prompted further detailed analysis of the ice core records, but has 
encouraged marine and tenestrial workers to re-examine existing data and to undertake 
new studies at high-resolution. Some of these rapid climatic fluctuations are now 
considered with respect to the Ioannina 284 record, since the unusually high 
accumulation rates from this sequence mean that it has considerable potential for high­
resolution analyses. Once again, although this discussion is not intended to be 
exhaustive, it is hoped that the potential of sites such as Ioannina for work of this nature 
can be emphasised. 
8.5.1 The Eemian 
The last interglacial (known in Europe as the Eemian) is thought to begin at around 128-
130ka. In the marine record, the end of the penultimate glaciation and the beginning of 
the Eemian is mat·ked by a sudden shift to more positive oxygen isotope values (from 
MIS-6 to MIS-5e), at what has become known as Termination 11 (van Donk, 1976; 
Broecker, 1984). The correlation between MIS-5e in the marine record and the Eemian 
in the terrestrial record has long been accepted (Shackleton, 1969), although this has 
recently been questioned by some authors (Kukla et al., 1997). Evidence for climatic 
instability at the very beginning of the Eemian (the Zeifen-Kattegat Oscillation, ZKO) 
has already been mentioned (section 7.5.1). This 'Younger Dryas-type' vat-iability has 
not only been noted in the type sites for the ZKO (Seidenkrantz et al., 1996), but also in 
other high-resolution sequences, such as the unsmoothed curve of marine core V19-30 
(Shackleton et al. , 1983), a Danish foraminiferal and isotopic record (Seidenkrantz, 
1993) and in Norwegian speleothems (Lauritzen, 1990, 1994). It has already been 
suggested that the carbonate and isotopic curves from 284 record the ZKO (section 
7.5.1). 
There is also some evidence from the ice core records not only for the ZKO, but also 
for other more dramatic periods of climatic instability during the Eemian. For example, 
a series of short frequency oscillations have been recorded in the GRIP core, which 
have been attributed to rapid climate fluctuations (GRIP members, 1992). More than 
once, temperatures reconstructed from the oxygen isotope signal of the ice appeat· to 
have plummeted to mid-glacial levels and only recovered on time scales ranging from 
COMPARISON WITH OTHER RECORDS. 198 
-70 to several thousand years. The GRIP workers divided the Eernian into three 
distinct warm phases (5e1, 5e3 and 5e5) and two cold episodes (5e2 and 5e4). 
However, these oscillations are not present in the GISP2 core (e.g. Grootes et aI., 
1993; Johnsen et al., 1995). The discrepancy is thought to be due to stratigraphical 
disturbances in at least one of the cores, since below a depth of 2,848m in the GRIP 
core and 2,679m in the GISP2 core, many of the ice layers exhibit tilts of between 10-
20° (Boulton, 1993; Alley et aI., 1993; Johnsen et al., 1995). The oscillations are not 
apparently recorded in the Antarctic Vostok core (GRIP members, 1993; Jouzel, 1994; 
Johnsen et al., 1995). 
With the exception of the ZKO, which is gradually achieving broader recognition as a 
real event, the question of the validity of the intra-Eemian climatic oscillations suggested 
by the GRIP ice core data is still unanswered (e.g. Peel, 1995). Despite the efforts of 
marine and terrestrial workers to find further evidence for this instability in a wide range 
of high-resolution sequences (e.g. Field et al., 1994; McManus et al., 1994; 
Seidenkrantz and Knudsen, 1994; Thouveny et al., 1994; Tzedakis et al., 1994; 
Seidenkrantz et al., 1995; Linsley, 1996; Litt et al., 1996; Zhisheng and Porter, 1997), 
no common consensus has yet been reached. Nevertheless, evidence appears to be 
gradually mounting for some degree of intra-Eemian instability, although whether it 
was to the same extent as the GRIP record suggests, remains to be seen. Possibly the 
most convincing evidence to date is for a mid-Eemian cooling event, noted from a 
variety of high-resolution marine records, such as from the Norwegian Sea (e.g . 
Cortijo et al., 1994). Maslin and Tzedakis (1996) suggest that this event (which lasted 
for a few hundred years, around 122ka) can also be detected in terrestrial records such 
as Lac du Bouchet (e.g. Thouveny et al., 1994), where it corresponds to an increase in 
herbaceous vegetation. 
The position of the Eemian in the Ioannina 284 record is well constrained by U-series 
dating, magnetostratigraphy and pollen stratigraphy. Full interglacial conditions are 
represented by a sediment thickness of approximately 13m, suggesting a high 
accumulation rate in excess of 1mka-'. Carbonate contents determined at lOcm intervals 
over this section of the core reveal significant oscillations that bear remarkable similarity 
to the 8180 curve from the GRIP ice-core (fig. 8.1), at least in its upper section (i.e. 
5el-5e3). Although both 8180 and carbonate content can often be considered as 
proxies for climate change, caution has been suggested in the interpretation of the 
carbonate curve from Ioannina (discussed in section 7.2.1). It would therefore be 
unwise to suggest a correlation with the GRIP record, especially since these marked 
oscillations are not reflected in either the isotope or pollen records. The question of 
c;-
C 
Q) 
0/) 
<t: 
GRIP record Ioannina 284 record 
--- ~ 105.0 --I -- Sc Sc r 82 
5d 
110.0 ~ ~ ~ [ 84 5d ? 86 ~ 5e! .......... 115.0 1 
5e2 ~ 88 5e3 120.0 ] 5e 90 5e4 
125.0 <- I- 92 
94 
130.0 1 0::=::: 5e5 ~ ~ ..., ..:a:: 96 
135.0 
-
ZKO 
""1. 
:: ----.. t 9' f 6 140.0 ~? 100 
0 5 10 15 20 25 30 
-40.00 -37.50 -35 .00 -32.50 -30.00 
8180 (%0) Carbonate Content (%) 
Fig. 8.1 Comparison between the GRIP oxygen isotope record (relative to SMOW) with the Ioannina 284 carbonate content record. 
No direct correlation is proposed (see text). ZKO = possible position of Zeifen-Kattegat Oscillation. Isotope stages attributed to the 
GRIP record are shown; those suggested for Ioannina 284 are also illustrated. 
§ 
..c 
i5.. 
Q) 
Cl 
Cl 
Cl 
~ 
:::tl 
C;; 
~ 
~ 
~ 
Cl 
~ ~ 
:::tl 
~ 
Cl 
Cl 
:::tl 
tl 
Vl 
• 
1.0 
1.0 
COMPARISON WITH OTHER RECORDS • 200 
what is controlling the carbonate content over this interval has also been addressed in 
section 7.2.1, where it was suggested that the variable contribution of bioclastic 
material to the sediment (which in turn was linked to lake-level changes) might be partly 
or wholly responsible. 
Despite the lack of obvious and immediate correlation with the GRIP record, the 
isotopic and pollen records do display significant variability within the Eemian 
(described in section 7.5.1) . After the ZKO at the beginning of the interval, marked 
changes occur at around 91m and 86m, which appear to be related to variations in the 
level of moisture and temperature and are indicative of the mid-Eemian onset of unstable 
climatic conditions. The shifts noted in the isotopic records seem to represent phase 
changes rather than brief events. Further high-resolution analysis of the Eemian section 
of 284 is therefore continuing (including diatom analysis), to try and better understand 
the complex palaeoenvironmental history that has already been suggested by a range of 
prOXIes. 
8.5.2 The Montaigu Event 
The Montaigu Event is a cool episode occurring in the early part of MIS-5c, equivalent 
to the St Gelmain I or Br0rup Stadial in European stratigraphical telms (Woillard, 1978; 
Reille et al ., 1992). In many European pollen sequences (e.g. Tenaghi Philippon, 
loannina 249, Valle di Castiglione, Lac du Bouchet, Les Echets, Grande Pile, Padul) , it 
is recorded as a brief period of open vegetation. It does not appear in the SPECMAP 
marine record, a fact that Tzedakis et al. (1997) partly attribute to the way in which the 
averaged stack was created and smoothed. A marked fluctuation also appears within 
the MIS-5c interval of both the GRIP and GISP2 ice cores, suggesting that it may be 
the same brief, cool oscillation. It is not known if the Montaigu Event was global in 
extent, although Reille et al. (1992) suggest that it may have been recorded from Clear 
Lake in California (Adam et al., 1981). 
A brief fluctuation occurs in the carbonate record from loannina 284 during the interval 
interpreted as being MIS-5c (fig. 5.7). Preliminary pollen analysis suggests that this is 
characterised by an increase in open vegetation (P.e. Tzedakis, unpubl. data). 
However, it is too early to propose a formal correlation with the Montaigu Event, 
especially in view of the comments concerning the interpretation of the carbonate curve 
(section 7.2.1) and the incomplete nature of the pollen data. Further investigations are 
continuing. 
COMPARISON WITH OTHER RECORDS. 201 
8.5.3 The Last Glacial 
Known as the WeichselianIWurmian in northern and central Europe and spanning MIS-
4 to 2 in the marine record, the last glacial is thought to have begun around 74ka. A 
detailed account of this complex period is beyond the scope of this report, but a 
comprehensive summary of the terrestrial record can be found in van Andel and 
Tzedakis (1996) and of the marine and ice core records in Lowe and Walker (1997). 
The European terrestrial record shows a dominantly cold, open steppe-tundra 
environment, periodically punctuated by a series of interstadials (e.g. Oerel, Glinde, 
Moershoofd, Hengelo, Denekamp). The ice core sequences during MIS-3 in particular 
record a series of brief (millennial- to century-scale) climatic oscillations known as 
Dansgaard-Oeschger cycles (Johnsen et al., 1992). When grouped together as 'Bond 
cycles', after Bond et al. (1993), they display a characteristic saw-tooth pattern of 
increasing cooling, terminating in a massive iceberg discharge event from the 
Laurentide and/or Fennoscandian ice sheets, known as a Heinrich Event (e.g. Heinrich, 
1988; Bond et al., 1992; Bond and Lotti, 1995; Dowdeswell et al., 1995; Fronval et 
al., 1995). Marine sequences record these events as layers of ice-rafted detritus, 
identifiable as increases in lithic fragments (sometimes carbonate-rich sediments), 
increases in magnetic susceptibility and a decrease in the frequency of foraminifera. 
Sequences from different parts of the Pacific Ocean suggest that these ice-rafting events 
were circumpolar' (e.g. Thunell and Mortyn, 1995; Kotilainen and Shackleton, 1995). 
It is thought that the cyclicity of ice sheet growth and collapse is linked to complex 
interactions between the atmosphere, oceanic circulation and ice sheet dynamics (e.g. 
MacAyeal, 1993). 
Some European terrestrial pollen records suggest variablity on similar frequencies to the 
oscillations present in the ice core records (e.g. Broecker, et al., 1988). However, as 
van Andel and Tzedakis (1996) and Tzedakis et al. (1997) point out, much higher 
sampling resolutions and tighter chronological controls are necessary before detailed 
correlations can be attempted. The same is tlUe for the Ioannina 284 record. 
Oscillations in the carbonate curve suggest climatic variability at frequencies similar to 
those displayed in the ice core records (fig. 5.7). However, the comments concerning 
the interpretation of the carbonate curve over this interval must be borne in mind 
(referred to above). In addition, the current lack of any pollen or isotopic data for this 
interval makes any comparisons difficult. Detailed work is continuing on this patt of 
the record. 
COMPARISON WITH OTHER RECORDS • 202 
8.5.4 The Last Glacial - Holocene Transition 
In terrestrial sequences, the last glacial - Holocene transition is recorded as a complex 
two-step deglaciation. The first step is represented by the warming of the Lateglacial 
interstadial from around 15ka BP. According to fOlmal European chronstratigraphic 
nomenclature (Mangerud et al ., 1974), this includes two warm chronozones (B0lling, 
13-12ka BP; and Aller0d, 11.8-11ka BP) and two cold chronozones (Older Dryas, 12-
11.8ka BP; and Younger Dryas, ll-lOka BP). A summary of further stratigraphical 
nomenclature can be found in Lowe and Walker (1997). The ice core records over this 
period display remarkably close correspondence with the reconstructed temperature 
curves from terrestrial records (e.g. Dansgaard et al., 1989; Kapsner et al. , 1995; Lowe 
et al., 1995; Walker, 1995). In the marine isotope record, this interval is marked by 
Termination I, at the MIS-2 to 1 boundary. High-resolution marine sequences also 
faithfully record the two-step deglaciation seen in the terrestrial records, for example the 
sea-surface temperature reconstructions from diatoms in the southeast Norwegian Sea 
(Kot; Karpuz and Jansen, 1992), or detailed planktonic formaniniferal records, also 
from the Norwegian Sea (Lehman and Keigwin, 1992). 
The transition from the last glacial to the present interglacial is represented in core 284 
by approximately the top 25m of sediment (fig. 7.3) . The proxies from this interval 
have been interpreted in section 7.5.3. Because of low ostracod abundances and 
measurement problems (several samples were lost due to machine failure), the 
resolution of the isotopic curves was not as high across the 18-15m interval as 
intended. This interval has been interpreted as possibly including the Lateglacial 
interstadial and Younger Dryas events. The relatively coarse resolution means that it 
has not been possible to distinguish between B011ing/Older Dry as/ Aller0d, although the 
maximum value of the oxygen isotope curve at around 15.1m may well represent the 
Younger Dryas. 
Another oscillation of interest is that occurring in the isotopic curves at around 13m. 
This has been interpreted as a brief cool, arid event, possibly equivalent to the (global?) 
event recorded in high-resolution terrestrial, marine and ice core sequences and dated at 
around 7.5ka BP (e.g. Alley et al ., 1997). This has already been discussed in section 
7 .5.3. 
High-resolution isotopic determinations from bulk carbonate samples across the entire 
25m interval are currently being undertaken. It is hoped that in conjunction with the 
corrected radiocarbon chronology and ongoing detailed pollen analyses, they will allow 
greater resolution of this interval in the 284 record and permit firm correlations with the 
established European terrestrial stratigraphy to be suggested. 
COMPARISON WITH OTHER RECORDS • 203 
8.6 SUMMARY 
This chapter has reviewed the importance of the Ioannina 284 sequence with respect to 
other well-known high-resolution records of climatic change. Where appropriate, 
possible correlations have been suggested. 
Initially, the European telTestrial record was discussed. Because of the direct 
cOlTelation with the pollen record of 249, this second sequence was used as a 'stepping 
stone' to permit tentative comparisons with several other long European records, 
including Tenaghi Philippon, Valle di Castiglione, Lac du BouchetfPraclaux and 
Grande PilelLes Echets. 
The marine oxygen isotope record was then briefly reviewed. Once again, by using the 
conelations originally established for 249, tentative comparisons with 284 were 
suggested. This was not thought possible when the ice core records were considered. 
Although there are many similarities between the climatic variability recorded in the 
Ioannina 284 sequence and that recorded in the ice cores, no means of direct cOITelation 
is possible. Nevertheless, the similarity of well-constrained parts of 284 with the GRIP 
record were noted (for example, the Eemian). 
Lastly, the potential of Ioannina 284 for detailed analysis of high frequency climatic 
fluctuations was discussed. The evidence for climatic variability in the Eemian from ice 
core and other records was first addressed. The 284 sequence clearly records the 
instability at the end of the penultimate glaciationibeginning of the Eemian (the so-called 
Zeifen-Kattegat Oscillation) in a range of proxies. Intra-Eemian instability was also 
evident in the 284 record, but although this did not appear to correspond with the 
excursion detected in the GRIP ice core record, it may have been related to a mid­
Eemian cooling event noted from certain high-resolution marine and terrestrial 
sequences. 
Other high frequency events were also discussed. Some evidence exists in the 284 
record for the Montaigu Event, a brief cooling episode detected in several European 
terrestrial records and probably the Greenland ice cores. However, the 284 record 
needs further detailed work before any fOlmal correlation can be proposed. 
The similarity of the carbonate record from 284 and the high frequency climatic 
oscillations evident from ice core and marine records through the last glacial (and MIS-3 
in particular) was noted. However, until the nature of the carbonate curve in 284 
COMPARISON WITH OTHER RECORDS • 204 
through this interval is fully understood and other proxy evidence is obtained (such as 
pollen and isotopic data), no detailed comparisons are possible. 
Finally, the last glacial - Holocene transition was discussed. Despite detailed isotopic 
measurements across this interval, the individual chronozones of the Lateglacial 
interstadial were not distinguishable. It was thought that this was probably due to 
poorer resolution of measurements through this section of core. Neveltheless, evidence 
for the position of a possible Younger Dlyas event was outlined, as well as a brief 
period of cooling known from around 7.5ka BP. It was emphasised that with 
improved pollen stratigraphy, additional isotopic measurements and a corrected 
radiocarbon timescale (all currently ongoing), the position of these and other high 
frequency events could be accurately determined. 
9. CONCLUSIONS 
This study has demonstrated that the Ioannina 284 sequence represents a continuous, 
319m sedimentary record of climatic change. Detailed palaeomagnetic measurements, a 
series of radiocarbon dates, U-series deterrninations and the direct coupling of the 
record with the adjacent 249 pollen sequence using magnetic susceptibility 
comparisons, have all enabled a strong chronological framework to be established from 
which an age model has been derived. This indicates that the 284 sequence extends 
back into the Middle Pleistocene, possibly in excess of 600ka (back to MIS-16) and 
records changes from glacial to interglacial mode, providing a basis for correlation with 
other established European sequences. 
Clearly, the age model developed for the sequence plays a crucial prut in the 
interpretation of the vru'ious proxy data. The upper part of the record (O-40m) has been 
constrained by a series of 12 AMS radiocru'bon dates derived from shell material. 
These are therefore likely to be slightly too old, because of reservoir effects. Work is 
currently in progress to assess the magnitude of any age anomaly by attempting to date 
pollen concentrates from the same levels that the shell-derived 14C dates were obtained. 
The age model for the entire sequence has principally been established from a 
combination of palaeomagnetic analyses and the correlation, via magnetic susceptibility 
profiles, of the carbonate curve from 284 with the pollen record of 249 (which has its 
own independently derived chronology). Although further high-resolution work is 
required, the magnetostratigraphy from 284 is already one of the longest and most 
comprehensive records of polru'ity excursions during the Brunhes chron obtained from 
a European terrestrial site. 
The original aims of this project were broadly based, in that once a chronological 
framework had been established, the intention was to attempt a reconstruction of the 
climatic history of the Ioannina basin. This was to be achieved by using a variety of 
multi-proxy techniques, including physical and geochernical analyses of the fauna and 
the sediment. These were intended to complement the existing palynological data from 
the adjacent 249 record, in order to provide a synthesis of palaeoenvironmental 
changes. The extreme length of the record meant that a sensible research strategy had 
first to be established if meaningful results were to be derived within the time available. 
Therefore, whilst appropriate blanket techniques have been applied to the entire 
sequence, more detailed analytical procedures, such as isotopic studies, have been 
CONCLUSIONS • 206 
restricted to selected parts of the record, namely the last glacial - Holocene transition 
and the last interglacial. 
Detailed analyses of these intervals have revealed the occurrence of climatic instability, 
which can be related to evidence from other high-resolution terrestrial, marine and ice 
core sequences. For example, the Zeifen-Kattegat Oscillation, occurring before the 
onset of the full Eemian, is readily reflected by all indicators. The Ioannina record is 
probably the most detailed terrestrial record of this event from Europe. Intra-Eemian 
climatic instability has also been detected, although it is not comparable in extent or 
structure to the instability recorded in the GRIP ice core record. At the last glacial -
Holocene transition, isotopic evidence appears to indicate the position and extent of the 
Lateglacial interstadial and the Younger Dryas, but further stratigraphical and 
palynological work is necessary before this can be fOlmally confirmed. A brief, arid 
phase has also been detected during the first half of the Holocene, which may represent 
a widespread cooling event occurring at around 7.5ka BP. 
As an integral part of this project, the chemistry of the lake system has been analysed in 
considerable detail. The importance of precipitation/evaporation in controlling the 
isotopic evolution of this effectively closed, shallow freshwater lake over its lifetime is 
readily apparent, particularly in the stable oxygen isotope data. It is also indirectly 
reflected in the stable carbon isotope data, which are largely controlled by a combination 
of residence time and surface water productivity. An understanding of the isotopic 
signature of the modern lake system has helped to interpret the fossil data and has 
enabled palaeoenvironmental and palaeoclimatic inferences to be made. The impOltance 
of the carbonate record has also been highlighted. Below a depth of about 145m, it has 
been suggested that the carbonate content of the sediment acts as a good proxy for 
palaeoproductivity (although above this level, bioclastic input confuses the signal), 
allowing it to be used as a climatic indicator in the derivation of the age model. 
A faunal study of the molluscs and ostracods has been undeltaken on both the modern 
lake and the core material. This was to investigate the possibility of using faunal data 
for biostratigraphical and palaeoecological purposes and also to provide a detailed 
summary of the fauna from the site. The biostratigraphical potential of the record was 
found to be limited, principally because of the sparseness of the fauna and the long 
ranges of some of the ostracods. However, this study has provided the opportunity not 
only to carry out the most comprehensive study of molluscan fauna from the lake to 
date, but has also enabled the first detailed study of the ostracod fauna to be undertaken. 
Several endemic species of mollusc and ostracod have been described and illustrated, 
some for the first time. 
CONCLUSIONS • 207 
The faunal data have also provided crucial palaeoecological and lake-level information 
and play a major part in the final synthesis of the lake history. For example, shell-beds 
of the bivalve Dreissena at levels in the core corresponding to the Last Glacial 
Maximum (LGM) are clear indicators of low lake-levels at this time. Although this 
agrees with regional pollen data, which suggest cool and arid conditions, it conflicts 
with geomorphological evidence derived from the Palaeolithic cave site of Kastritsa. 
Beach deposits within the cave, dated to the LGM, have been used as evidence for 
relatively high lake-levels at this time. It is suggested that subsequent tectonic uplift of 
the cave site, previously unaccounted for, is responsible for this apparent discrepancy. 
Another of the principal aims of the project was to undeltake mUlti-proxy investigations 
that did not specifically involve palynological analyses. This was because a reasonably 
detailed pollen sequence (249) was already available, located less than lkm from the 
284 site. However, it has been clearly demonstrated that the palaeoecological and 
stratigraphical information provided by both the 249 record and new, high-resolution 
pollen studies from 284 (P.e. Tzedakis, unpuhl. data), when coupled with other proxy 
data sampled at similar intervals (particularly the isotopic and faunal analyses), provides 
a very powerful tool for palaeoenvironmental reconstruction. 
The 284 sequence has also been compared with several other long European tenestrial 
records and a series of correlations suggested. In terms of stratigraphical length, only 
Tenaghi Philippon from northeastern Greece is longer than 284, although the resolution 
of the record is generally poorer and the details of the stratigraphy can be questioned in 
light of new analytical and dating techniques (e.g. Rossignol-Strick, 1995). It has also 
been possible to suggest tentative correlations with the marine oxygen isotope record. 
Because of the high accumulation rates and the overall length of the sequence, Ioannina 
284 offers a unique opportunity for high-resolution analyses of key parts of the Middle 
and Late Quaternary record from Europe. Some of this potential has already been 
realised, although it is a sobering thought that less than 16% of the record has been 
analysed in detail during this study. Clearly, many questions can be addressed by the 
application of fmther high-resolution palynological, isotopic and other analytical 
techniques (discussed below), possibly as part of a co-ordinated larger-scale project. 
For example, there is the opportunity to analyse critically the structure of at least three 
other well-defined interglacial periods, a luxury that most other terrestrial sequences 
simply do not possess. One potential target for future investigation is MIS-11, thought 
to be possibly the warmest interglacial of the mid-Quaternary, despite relatively low 
levels of insolation (e.g. Burckle, 1993). Howard (1997) raises the possibility that this 
CONCLUSIONS • 208 
orbital geometry may make MIS-11 a better analogue for our own interglacial than MIS-
5e (the Eemian), often regarded as a close analogue for Holocene conditions. 
New analytical techniques that could be used to further investigate the sequence might 
include C-H-N measurements of the organic fraction of the sediments, as well as an 
assessment of the proportion of C/C4 plant biomass (known to be linked to 
environmental conditions). Diatom analyses through key intervals of the record and 
more comprehensive appraisal of the sediment geochemistry will also help in deriving a 
more detailed palaeoenvironmental history of the lake basin. Further analysis of the 
magnetic susceptibility record (along with an assessment of the magnetic characteristics 
of a variety of catchment lithologies) may permit sediment sources to be identified 
(discussed previously in section 7.2.3). X-ray diffraction measurements might also 
help to identify and categorise magnetic carriers and provide further detailed infOlmation 
on the variability of sediment composition. 
Increasing chronological resolution and improving the independent justification of the 
age model should also be two of the main priorities for future work. It should be 
possible to apply the U-series technique that has proven so successful in dating the 
carbonates from the Eemian interval of the sequence, to earlier interglacials (T.C. 
Atkinson, pers. camm.). In addition, there are clear advantages in investigating the 
possibility of deriving a tephrochronology from the site. As outlined in section 5.7, 
this technique has enormous potential for providing crucial, detailed stratigraphical 
information for parts of the record beyond the reach of other, more conventional dating 
methods. The recent discoveries of substantial tephra deposits of suitable age in Epirus 
(e.g. St Seymour and Christianis, 1995; Pyle et al., in press) have served to highlight 
the possibilities for future work of this nature on the Ioannina sequence. 
This project has clearly highlighted the benefits of working at high-resolution on long, 
continuous terrestrial sequences and adopting a multi-proxy approach to analytical 
techniques. The record represents a comprehensive and detailed archive of 
palaeoenvironmental infOlmation that has the potential to become one of the most 
extensive and important terrestrial sequences in Europe. 
REFERENCES 
Adam, D.P., Sims, I.D. and Throckmorton, C.K. (1981) 130,000-yr continuous pollen record from 
Clear Lake, Lake County, California. Geology 9, 373-377. 
Albanis, T.A. , Pomonis, P.J. and Sdoukos, A.T. (1986) Seasonal fluctuations of organochlorine 
and triazine pesticides in the aquatic system of the Ioannina Basin (Greece). The Science of the 
Total Environment 58, 243-253. 
Alessio, M., Allegri, L., Bella, F., Calderoni, G., Cortesi, e., Dai Pra, G., De Rita, D., Esu, D., 
Follieri, M., Improta, S., Magri, D., Narcisi , B. , Petrone, V. and Sadori , L. (1986) 14C 
dating, geochemical features, faunistic and pollen analyses of the uppermost lOm core from 
Valle di Castiglione (Rome, Italy) . Geologica ROlllana 25, 287-308. 
AlIen, H.D. (1986) Late Quaternary of the Kopais Basin, Greece: Sedimentary and Environmental 
HistOlY. Ph.D Thesis (unpublished), University of Cambridge. 
AlIen, S.E. (1974) Chemical Analysis of Ecological Materials . Blackwell Scientific Publications, 
Oxford. 
Alley, R.B ., Gow, A.J., Johnsen, S.J., Kipfstuhl,1., Meese, D.A. and Thorsteinsson, T. (1993) 
Comparison of deep ice cores. Nature 373, 393-394. 
Alley, R.B., Mayewski , P.A., Sowers, T., Stuiver, M., Taylor, K.C. and Clark, P.U. (1997) 
Holocene climatic instability : a prominent, widespread event 8200 yr ago. Geology 25 (6), 
483-486. 
Anagnostidis, K. and Economou-Amilli, A. (1980) Limnological studies on Lake Pamvotis 
(Ioannina), Greece. Archiv fill' Hydrobiologie 89 (3), 313-342. 
Andrews, J.T. , Erlenkeuser, H., Tedesco, K., Aksu, A.E. and Jull, A.J.T. (1994) Late Quaternary 
(stage 2 and 3) meltwater and Heinrich events, northwest Labrador Sea. Quaternary Research 
41 , 26-34. 
Aschmann, H. (1973) Distribution and peculiarity of Mediterranean ecosystems. In: di Castri, F. 
and Mooney, H.A. (eds.) Mediterranean-type Ecosystems. Springer-VerJag, Berlin. 11-19. 
Athersuch, 1., Horne, D.1. and Whittaker, J.E. 
Synopsis of the British Fauna (N.S.) 43. 
Oegstgeest, Netherlands. 
(1989) Marine and Brackish Water Ostracods. 
Universal Book Services / Dr W Backhuys, 
Atkinson, T.e., Briffa, K.R. and Coope, G.R. (1987) Seasonal temperatures in Britain during the 
last 22,000 years, reconstructed using beetle remains. Nature 325, 587-592. 
Aubouin, I . (1959) Contribution a I'etude geologique de la Grece septentrionale: les confins de l'Epire 
at de la Thessalie. Annales Geologiques des Pays Helteniques 10, 1-525. 
Bailey, G.N., Carter, P.L., Gamble, e.S. and Higgs, H.P. (1983a) Asprochaliko and Kastritsa: 
further investigations of Palaeolithic settlement and economy in Epirus (North-West Greece). 
Proceedings of the Prehistoric Society 49, 15-42. 
Bailey, G.N., Carter, P.L., Gamble, C.S. and Higgs, H.P. (1983b) Epirus revisited: seasonality 
and inter-site variation in the Upper Palaeolithic of north-west Greece. In : Bailey , G.N. (ed.) 
Hunter-Gatherer Economy in PrehistOl)'.' A European Perspective. Cambridge University 
Press, Cambridge. 64-78 . 
Baksi, A.K., Hsu, V., McWilliams, M.O. and Farrar, E. (1992) 4°ArP9Ar dating of the Brunhes­
Matuyama geomagnetic field reversal. Science 256, 356-357. 
REFERENCES. 210 
Barbetti , M. and McElhinny, M. (1972) Evidence of a geomagnetic excursion 30,000 yr BP. Nature 
239, 327-330. 
Barbetti, M. and McElhinny, M. (1976) The Lake Mungo geomagnetic excursion. Philosophical 
Transactions of the Royal Society of London B281, 515-542. 
Barry, RG. and Chorley, RJ. (1992) Atmosphere, Weather and Climate (6th Edition). Routledge, 
London. 
Benda, L., van der Meulen, AJ., Meyer, K.-1. and van de Weerd, A. (1987) Biostratigraphic 
correlations in the Eastern Mediterranean Neogene. Newsletters on Stratigraphy 17 (3), 129-
141. 
Bender, M., Sowers, T., Dickson, M.-L., Orchardo, 1. , Grootes, P., Mayewski, P.A. and Meese, 
D.A. (1994) Climate correlations between Greenland and Antarctica during the past 100,000 
years . Nature 372, 663-666. 
Bengtsson, L. and Enell, M. (1986) Chemical Analysis. In: Berglund, B.E. (ed.) Handbook of 
Holocene Palaeoecology and Palaeohydrology. John Wiley and Sons Ltd., Chichester. 423-
451. 
Bennett, K.D., Tzedakis, P.e. and Willis, K.J. (1991) Quaternary refugia of north European trees . 
Journal of Biogeography 18, 103-115. 
Bleil, U. and Gard, G. (1989) Chronology and correlation of Quaternary magnetostratigraphy and 
nannofossil biostratigraphy in Norwegian-Greenland Sea sediments. Geologische Rundschau 
78, 1173-1187. 
Bodergat, A.-M., Rio, M. and Andreani, A.-M. (1991) Composition chimique et ornamentation cb 
Cyprideis torosa (Crustacea, Ostracoda) dans le domaine paralique. Oceanologica Acta 14 (5) , 
505-514. 
Bonatti, E. (1966) North Mediterranean climate during the last Wiirm glaciation. Nature 209 , 984-
985 . 
Bond, G. and Lotti, R. (1995) Iceberg discharges into the North Atlantic on millennial timescales 
during the last glaciation . Science 267, 1,005-1,010. 
Bond, G, Heinrich, H., Broecker, W., Labeyrie, L. , McManus, 1. , Andrews, 1., Huon, S., 
Jantschik, R, Clasen, S., Simet, c., Tedesco, K., Klas , M., Bonani, G. and Ivy, S . 
(1992) Evidence for massive discharges of icebergs into the North Atlantic ocean during the 
last glacial period. Nature 360, 245-249. 
Bond, G., Broecker, W. , Johnsen, S.J., McManus, J., Labeyrie, L. , Jouzel , J. and Bonani, G. 
(1993) Correlations between climate records from North Atlantic sediments and Greenland ice. 
Nature 365, 143-147. 
Bonhomrnet, N. and Babkine, 1. (1966) Sur une direction an om ale du champ magnetique terrestre au 
cours du Quaternaire recent. Comptes Rendus de l'Academie des Sciences 262, 919-921. 
Boomer, 1. (1993) Palaeoenvironmental indicators from the Late Holocene and contemporai'y 
Ostracoda of the Aral Sea. Palaeogeography, Palaeoclimatology, Palaeoecology 103, 141-
153 . 
Bottema, S. (1967) A Late Quaternary pollen diagram from Ioannina, north-western Greece. 
Appendix in: Higgs, E.S., Vita-Finzi, c., Harris, D.R. and Fagg, A.E. The climate, 
environment and industries of Stone Age Greece: Part Ill. Proceedings of the Prehistoric 
Society 33, 26-29. 
Bottema, S. (1974) Late Quatemmy Vegetation History of Northu;estem Greece Ph.D. thesis , 
Rijksuniversiteit te Groningen . 
REFERENCES. 211 
Bottema, S. (1982) Palynological investigations in Greece with special reference to pollen as an 
indicator of human activity . Palaeohistoria 24, 257-289. 
Boulton, G.S . (1993) Two cores are better than one. Nature 366, 507-508. 
Bowen, D.Q. and Sykes, G.A. (1988) Correlations of marine events and glaciations on the northeast 
Atlantic margin. Philosophical Transactions of the Royal Society of London B318, 619-
635 . 
Bowen, D.Q., Hughes, S. , Sykes, G.A. and Miller, G.H. (1989) Land - sea correlations in the 
Pleistocene based on isoleucine epimerization in non-marine molluscs. Nature 316 , 251-
253. 
Broecker, W.S. (1984) Terminations. Ill: Berger, A. , Imbrie, J., Hays, J., Kukla, G. and 
Saltzman, B. (eds.) Milankovitch and Climate. Understanding the Response to Astronomical 
Forcing D. Reidel Publishing Co., Dordrecht, Netherlands. 687-698. 
Broecker, W.S., Andree, M., Bonani, G., Wolfi, W., Oeschger, H. and Klas, M. (1988) Can the 
Greenland climatic jumps be identified in records from ocean and land? QuaternaJ)' Research 
30, 1-6. 
Broussoulis, J., Ioakim, c., Kolovos, G. and Papanikos, D. (in prep.) The Ioannina Basin: 
geological and palaeoenvironmental evolution in Quaternary and historical times. 
Brown, T.A., Nelson, D.E., Mathewes, R.W., Vogel, J.S. and Southon, J.R. (1989) Radiocarbon 
dating of pollen by accelerator mass spectrometry. QuaternaJ)' Research 32, 205-212. 
Brunhes, B. (1906) Recherches sur le directions d'aimantation des roches volcaniques. Journal of 
Physics 5, 705-724. 
Burckle, L.H. (1993) Late Quaternary interglacial stages warmer than present. QuaternaJ)' Science 
Reviews 12, 825-831 . 
Butot, L.J.M. and Welter-Schultes, F.W. (1994) Bibliography of the mollusc fauna of Greece. 
Schriften zur Malakozoologie 7, 1-160. 
Butzer, K.W. (1980) Holocene alluvial sequences: problems of dating and correlation . In: 
Cullingford, R.A. , Davidson, D.A. and Lewin, J. (eds.) Timescales in Geomorphology. 
John Wiley, Chichester. 131-142. 
Carbonel, P., Colin, J.-P ., Danielopol , D.L. , Uiffler, H. and Neustrueva, I. (1988) Paleoecology 
of limnic ostracodes: a review of some major topics. Palaeogeography, Palaeoclimatology, 
Palaeoecology 62, 413-461. 
Champion, D.E. , Lanphere, M.A. and Kuntz, M.A. (1988) Evidence for a new geomagnetic 
reversal from lava flows in Idaho: discussion of short polarity reversals in the Brunhes and late 
Matuyama polarity chrons. Journal of Geophysical Research 93 (B 10), 11,667-11,680. 
Chivas, A.R. , De Deckker, P. , Cali, J.A., Chapman, A., Kiss, E. and Shelley, J.M.G . (1993) 
Coupled stable-isotope and trace-element measurements of lacustrine carbonates as 
palaeoclimatic indicators. In: Swart, P.K., McKenzie, J.A. and Lohman, K.C. (eds.) 
Geophysical Monograph 78: Climate change in continental isotopic records. American 
Geophysical Union, Washington D.C. 113-121. 
Chivas, A.R. , De Deckker, P. and Shelley, J.M.G . (1983) Magnesium, strontium and barium 
partitioning in nonmarine ostracod shells and their use in palaeoenvironmental reconstructions 
- a preliminary study. In: Maddocks, R.F. (ed.) Applications of Ostracoda. The University 
of Houston, Houston. 238-249. 
Chivas, A.R., De Deckker, P. and Shelley, J.M.G. (1986) Magnesium content of non-marine 
ostracod shells: a new palaeosalinometer and palaeothermometer. Palaeogeography, 
Palaeoclimatology, Palaeoecology 54, 43-61. 
REFERENCES • 212 
Clews, lE. (1989) Structural controls on basin evolution: Neogene to Quaternary of the Ionian zone, 
Western Greece. lournal of the Geological Society 146, 447-457. 
Collinson, D.W. (1983) Methods in Rock Magnetism and Palaeol11agnetism - Techniques and 
Instrumentation. Chapman and Hall, London. 
Conispolatis, N. , Panagos, A., Perissoratis, C. and Varnavas, S. (1986) Geological and 
sedimentological patterns in the Lake Pamvotis (Ioannina), NW. Greece. Annales 
Geologiques des Pays Helleniques 33, 269-285. 
Coope, G.R., Gibbard, P.L., Hall, A.R., Preece, R.c., Robinson, J.E. and Sutcliffe, A.J. (in 
press) Climatic and environmental reconstructions based on fossil assemblages from Middle 
Devensian (Weichselian) deposits of the River Thames at South Kensington, central London. 
QuatemQl)l Science Reviews. 
Cortijo, E., Duplessy, J.-c., Labeyrie, L., Leclaire, H., Duprat, l and van Weering, T.C.E. 
(1994) Eemian cooling in the Norwegian Sea and North Atlantic Ocean preceding continental 
ice-sheet growth. Nature 372, 446-449. 
Craig, H. (1957) Isotopic standards for carbon and oxygen and correction factors for mass­
spectrometric analysis of carbon dioxide. Geochimica et Cosl11ochimicaActa 12, 133-149. 
Creer, K.M. and Morris, A. (1996) Proxy-climate and geomagnetic palaeointensity records extending 
back to ca. 75,000 BP derived from sediments cored from Lago Grande di Monticchio, 
southern Italy. Quaternary Science Reviews 15, 167-188. 
Creer, K.M., Readman, P.W. and Jacobs, A.M. (1980) Palaeomagnetic and palaeontological dating 
of a section at Gioia Tauro, Italy: identification of the Blake event. Earth and PlanetGl), 
Science Letters 50, 289-300. 
Creer, K.M., Readman, P.W. and Papamarinopoulos, S. (1981) Geomagnetic secular variations in 
Greece through the last 6000 years obtained from lake sediment studies. Geophysical lournal 
of the Royal Astronomical Society 66, 193-219. 
Curtis, J.H. and Hodell, D.A. (1993) An isotopic and trace element study of ostracods from Lake 
Miragoane, Haiti: a 10,500 year record of paleosalinity and paleotemperature changes in the 
Caribbean. In: Swart, P.K., McKenzie, J.A. and Lohman, K.c. (eds.) Geophysical 
Monograph 78: Climate change in continental isotopic records. American Geophysical 
Union, Washington D.C. 135-152. 
Dalloni, M. (1923) Contribution a I'Etude des terrains Tertiares de la Thessalie et de I'Epire. Bulletin 
de la Societe Geologique Franr,;ais 23 (4), 284-294. 
Danielopol, D.L. (1978) Uber Herkunft und Morphlogie der Siisswasser-hypogiiischen Candoninae 
(Crustacea, Ostracoda). Sitzungsberichten der Osterreichischer Akadamie der Wissenschajten, 
Mathematische-Ilaturwissenschajtliche klasse (Abteilung 1) 187, 1-162. 
Danielopol, D.L. (1991) Spatial-distribution and dispersal of interstitial Crustacea In alluvial 
sediments of a backwater of the Danube at Vienna. Stygologia 6 (2), 97-110. 
Danielopol, D.L. and McKenzie, K.G. (1977) Psychrodromus gen. n. (Crustacea, Ostracoda), with 
redescription of the cypridid genera Prionocypris and llyodrol1lus. Zoologica Scripta 6, 301-
322. 
Dansgaard, W., White, lW.C. and Johnsen, SJ. (1989) The abrupt termination of the Younger 
Dryas climate event. Nature 339, 532-534. 
Dansgaard, W., Johnsen, SJ., Clausen, H.B., Dahl-Jensen, D., Gundestrup, N.S., Hammer, C.U., 
Hvidberg, C.S., Steffensen, J.P., Sveinbjornsdottir, S.E., Jouzel, l and Bond, G. (1993) 
Evidence for general instability of past climate from a 250-kyr ice-core record. Nature 364, 
218-220. 
REFERENCES. 213 
David, P. (1904) Sur la stabilite de la direction d'aimantation dans quelques roches volcaniques. 
Comptes Rendus de l'Academie des Sciences 138, 41-42. 
Dean, W.E. (1974) Determination of carbonate and organic matter in calcareous sediments and 
sedimentaIY rocks by loss on ignition: comparison with other methods. Journal of 
Sedimentary Petrology 44, 242-248. 
Dean, W.E. (1981) Carbonate minerals and organic matter in sediments of modern north temperate 
hard-water lakes. Society of Economic Palaeontologists and Mineralogists: Special 
Publication 31, 213-231. 
Dearing, lA. (1979) Application of magnetic measurements to studies of particulate flux in lake­
watershed systems. Ph.D. Thesis (unpublished), University of Liverpool. 
Dearing, lA (1983) Changing patterns of sediment accumulation in a small lake in Scania, southern 
Sweden. Hydrobiologia 103, 59-64. 
Dearing, lA. (1986) Core correlation and sediment influx. In: Berglund, B.E. (ed.) Handbook of 
Holocene Palaeoecology and Palaeohydrology. John Wiley and Sons Ltd. , Chichester. 247-
270. 
Dearing, lA. , Elner, J.K. and Happey-Wood, CM. (1981) Recent sediment flux and erosional 
processes in a Welsh upland lake catchment based on magnetic susceptibility measurements. 
Quaternmy Research 16, 356-372. 
Dearing, J.A, Maher, B.A and Oldfield, F. (1985) Geomorphological linkages between soils and 
sediments: the role of magnetic measurements. In : Richards, K.S., Arnett, R.R. and Ellis, 
S. (eds.) Geol1lorphology and Soils George Allen and Unwin, London. 245-266. 
de Beaulieu, l-L. and Reille, M. (1984) A long upper Pleistocene pollen record from Les Echets, 
near Lyon, France. Boreas 13, 111-132. 
de Beaulieu, l-L. and Reille, M. (1989) The transition from temperate phases to stadials in the long 
Upper Pleistocene sequence from Les Echets (France). Palaeogeography, Palaeoclimatology, 
Palaeoecology 72, l47-159. 
de Beaulieu, l-L. and Reille, M. (1992) The last climatic cycle at La Grande Pile (Vosges, France): 
a new pollen profile. Quaternmy Science Reviews 11, 431-438 . 
de Beaulieu, l-L. and Reille, M. (l995) Pollen records from the Velay craters: a review and 
correlation of the Holsteinian Interglacial with isotopic stage 11. Mededelingen, Rijks 
Geologische Dienst 52, 59-70. 
de Beaulieu, l-L., Andrieu, V., Ponel, P. and Reille, M. (1994) The Weichselian Late-glacial in 
southwestern Europe (Iberian Peninsula, Pyrenees, Massif Central, northern Appenines). 
Journal of Quaternary Science 9, 10 I-I 07. 
De Deckker, P. and Forester, R.M. (1988) The use of ostracodes to reconstruct continental 
palaeoenvironmental records. In: De Deckker, P., Colin, lP. and Peypouquet, J.P. (eds.) 
Ostracoda in the Earth Sciences. Elsevier, Amsterdam. 176-199. 
De Deckker, P., Chivas, A, Shelley, lM.G. and Torgensen, T. (1988) Ostracode shell chemistry: 
a new palaeoenvironmental indicator applied to a regressive / transgressive record from the 
Gulf of Carpentaria, Australia. Palaeogeography, Palaeoclimatology, Palaeoecology 66, 
231-241. 
Deevey, E.S., Gross, M.S. , Hutchinson, G.E. and Kraybill , H.L. (1954) The natural 14C contents 
of materials from hard-water lakes. Proceedings of the National Academy of Sciences of the 
USA 40, 285-288 . 
Delesse, A. (1849) Sur le magnetisme polaire dans les mineraux et dans le roches. Annales ck 
Chimie et de Physique 25, 194-209. 
REFERENCES. 214 
Delorme, L.D. (1969) Ostracodes as Quaternary paleoecological indicators. Canadian Journal of EQ/1h 
Sciences 6, 1,471-1,476. 
Delorme, L.D. (1991) Ostracoda. In: Thorp, J.H. and Covich, A.P. (eds.) Ecology and 
Classification of North American Freshwater Invel1ebrates Academic Press Inc., San Diego. 
691-722. 
Denham, CR. (1976) Blake polarity episode in two cores from the greater Antilles outer ridge, Italy: 
identification of the Blake event. Earth and PlanetalY Science Letters 29, 422-434. 
Denham, CR. and Cox, A. (1971) Evidence that the Laschamps polarity event did not occur 
13,300-30,400 years ago. Earth and Planetary Science Letters 13, 181-190. 
Denham, CR., Anderson, R.F. and Bacon, M.P. (1977) Paleomagnetism and radiochemical age 
estimates for Late Brunhes polarity episodes. EQ/1h and PlanetalY Science Letters 35, 384-
397. 
Denton, G.H. and Hughes, TJ . (1981) The Last Great Ice Sheets. Wiley, New York. 
Dettman, D.L., Smith, AJ., Rea, D.K., Moore Jr., T.C and Lohmann, K.C. (1995) Glacial 
meltwater in Lake Huron during early postglacial time as inferred from single-valve analysis of 
oxygen isotopes in ostracodes. Quaternary Research 43, 297-310 
Devoto, G. (1965) Lacustrine Pliocene in the Lower Liri Valley . Geologica Romana 4, 291-368. 
Dollfus, G.F. (1922) Faune malacologique du Miocene superieur de Janina en Epire. Bulletin de la 
Societe Geologique Franrais 22 (4), 101-123. 
Dowdeswell, J.A., Maslin, M.A., Andrews, J.T. and McCave, LN. (1995) Iceberg production, 
debris rafting and the extent and thickness of Heinrich layers (H-I, H-2) in North Atlantic 
sediments. Geology 23 (4), 301-304. 
Eicher, U. and Siegenthaler, U. (1984) Stable isotopes in lake marl and mollusc shells from 
Lobsigensee (Swiss Plateau) . Revue de Paleobiologie 2, 217-220. 
Eleutheriadis, N., Hatziioannou, M. and Lazaridou-Dimitriadou, M. (1993) Contribution to the 
study of the biology of Bithynia graeca (Prosobranchia Gastropoda) and Physa aeuta 
(Pulmonata Gastropoda) in Lake Kerkini (North Greece). Biologia Gallo-hellenica 20 (1), 
267-272. 
Ellis, A.E. (1978) British Freshwater Bivalve Mollusca. Synopsis of the British Fauna (N.s.) 11 . 
Academic Press, London. 
Engstrom, D.R. and Wright, H.E. (1984) Chemical stratigraphy of lake sediments as a record of 
environmental change In: Haworth, E.Y. and Lund, J.W.G. (eds.) Lake Sediments and 
Environmental HistOl)'. Leicester University Press, Leicester. 11-67. 
Ericson, D.B., Ewing, M., Wollin , G.A. and Heezen, B.C (1961) Atlantic deep sea sedimentary 
cores. Geological Society of America Bulletin 72, 193-286. 
Evin, J. (1990) Validity of the radiocarbon dates beyond 35 ,000 years B.P. Palaeogeography, 
Palaeoclimatology, Palaeoecology 80, 71-78. 
Faure, G. (1986) Principles of Isotope Geology: Second Edition. John Wiley and Sons, Ltd., New 
York. 
Fels, E. (1957) Der IOlinnina-See in Greichenland. Stuttgart Geographische Studieren 69, 247-252. 
Fernandez-Gonzalez, M., Frydas, D., Guernet, C. and Mathieu, R . . (1994) Foraminiferes et 
ostracodes du Pliopleistocene de le region de Patras (Grece). InterSt stratigraphique et 
paleogeographique. Revista EspQ/iola de Micropaleontologia 26, 89-108. 
REFERENCES • 215 
Flower, RJ., Dearing, J.A. and Nawas, R. (1984) Sediment supply and accumulation in a small 
Moroccan lake: an historic perspective. Hydrobiologia 112, 81-92. 
Florschutz, F., Menendez Amor, J. and Wijmstra, T.A. (1971) Palynology of a thick Quatemary 
succession in southern Spain . Palaeogeography, Palaeoclil11atology, Palaeoecology 10, 
233-264. 
Folk, R.L. (1974) The natural history of crystalline calcium carbonate: effect of magnesium content 
and salinity. Journal of Sedimental}' Petrology 44 (1), 40-53. 
Follieri, M. , Magri, D. and Sadori, L. (1988) 250,000-year pollen record from Valle di Castiglione 
(Roma). Pollen Spores 30, 329-356. 
Fontes, J.c., Coque, R., Dever, L., Filly, A. and Mamou, A. (1983) Paleohydrologie isotopique 
de l'Oued el Akarit (Sud Tunisie) au Pleistocene superieur et a l'Holocene. Palaeogeography, 
Palaeoclimatology, Palaeoecology 43, 41-62. 
Frakes, L.A., Francis, J.E. and Syktus, J.I. (1992) Climate modes of the Phanerozoic. Cambridge 
University Press, Cambridge. 
Fronval, T ., Jansen, E., Bloemendal, J. and Johnsen, S. (1995) Oceanic evidence for coherent 
fluctuations in Fennoscandian and Laurentide ice sheets on millenium timescales . Nature 
374, 443-446. 
Fuhrmann, R. and Pietrzeniuk, E. (1990) Die Aussage der Ostrakodenfauna zum 
Sedimentationsablauf im Interglazialbecken, zur klimatischen Entwicklung und zur 
stratigraphischen Stellung des Interglazials von Neumark-Nord (Geiseltal) . In: Mania, D. , 
Thomae, M. , Litt, T. and Weber, D. (eds.) Neumark-Grobern. Beitriige zur Jagd des 
Mittlepaliiolithischen Menschen . Veroffenentlichungen der Landesmuseum Vorgeschichtlich 
Halle 43, 161-166. 
Gasse, F., Fontes, J .c., Plaziat, J.c., Cat·bonel, P., Maczmarska, I. , De Deckker, P. , Sou lie­
Marsche, I. , Callot, Y. and Dupeuple, P.A. (1987) Biological remains , geochemistry and 
stable isotopes for the reconstruction of environmental and hydrological changes in the 
Holocene lakes from North Sahara. Palaeogeography, Palaeoclimatology, Palaeoecology 60, 
1-46. 
Gat, J.R. and Carmi, I. (1970) Evolution of the isotopic composition of atmospheric waters in the 
Mediterranean Sea area. Journal of Geophysical Research 75 (15), 3039-3048. 
Gillet, S. (1962) Remarques sur des Gastropodes de quelques gisements du Pliocene et du Quatemaire 
d'Epire. Praktika Akadamiae Athens 37, 260-273. 
Gillet, S. (1963) Nouvelles donnees sur le gisement villafranchien de Nea-Corinthos. Praktika 
Akadamiae Athens 38, 400-417. 
Glen, J.M. and Coe, R.S. (1997) Paleomagnetism and magnetic susceptibility of Pleistocene 
sediments from drill hole OL-92, Owens Lake, California. In: Smith, G.I. and Bischoff, 
J.L. (eds.) An 800,OOO-Year Paleoclimatic Recordfrom Core DL-92, Dwens Lake, Southeast 
California Geological Society of America Special Paper 317, Boulder Colorado. 67-78 . 
Gomez, B. (1987) The alluvial terraces and fills of the Lower Vasilikos Valley, in the vicinity of 
Kalavasos, Cyprus. Transactions of the Institute of British Geographers, New Series 12, 
345-359. 
Goodfriend, G.A. (1987) Evaluation of amino-acid racemization / epimerization dating using 
radiocarbon-dated fossil land snails. Geology 15, 698-700. 
Goodfriend, G.A. (1992) The use of land snail shells in palaeoenvitonmental reconstruction. 
Quaternal)' Science Reviews 11, 665-685. 
REFERENCES. 216 
Gorham, E., Lund, J.W.G., Sanger, J.E. and Dean, W.E. (1974) Some relationships between algal 
standing crop, water chemistry and sediment chemistry in the English lakes. Lil1lllology and 
Oceanography 19, 601-617. 
Graham, A. (1988) Molluscs: Prosobranch and Pyramidellid Gastropods. Synopsis of the British 
Fauna (N.S.) 2. Universal Book Services / Dr W Backhuys, Oegstgeest, Netherlands. 
Griffiths, H.!. (1995a) The application of freshwater ostracods to the study of Late Quatel71G1), 
palaeoenvironl1lents in north-western Europe. Ph.D thesis (unpublished), University of 
Cardiff. 
Griffiths, H.!. (1995b) European Quaternary freshwater Ostracoda: a biostratigraphic and 
palaeobiogeographic primer. Scopolia (Ljubljana) 34, 1-168. 
Griffiths, H.I. and Butlin, R.K. (1994) Darwillula stevensoni: a review of the biology of a persistant 
parthenogen. In: Horne, DJ. and Martens, K. (eds.) The Evolutionary Ecology of 
Reproductive Modes in Non-marine Ostracoda: University of Greenwich Press, London. 27-
36. 
Griffiths, H.!. and Evans, J.G. (1995) An annotated check-list of British Pleistocene, Holocene and 
modern freshwater ostracods. Journal of Micropalaeontology 14, 59-65 . 
Griffiths, H.!. and Martin, D.S. (1993) The spatial distribution of benthic ostracods in the profundal 
zone of Loch Ness. The Scottish Naturalist 105, 137-147. 
GRIP members (1993) Climate instability during the last interglacial period recorded in the GRIP ice 
core. Nature 364, 203-207. 
Gronvold, K., Oskarsson, N., Johnsen, S.J., Clausen, H.B., Hammer, C.U., Bond, G. and Bard, 
E. (1995) Ash layers from Iceland in the Greenland GRIP ice core correlated with oceanic and 
land sediments. Earth and Planetw)' Science Letters 135, 149-155. 
Grootes, P.M., Stuiver, M., White, J.W.c., Johnsen, S. and Jouzel, J. (1993) Comparison of 
oxygen isotope records from the GISP2 and GRIP Greenland ice cores. Nature 366, 552-
554. 
Grove, A.T. (1997) Classics in physical geography revisited : 'The Mediterranean Valleys '. Progress 
in Physical Geography 21 (2), 251-256. 
Guernet, C., Sauvage, J. and Soulie-Marsche,!. (1977) Le Levantin de Joanina (Epire, Grece): 
observations stratigraphiques et paleontologiques. Geobios 10 (2), 297-309. 
Guiot, J., de Beaulieu, J.-L., Cheddadi, R., David, F., Ponel, P. and Reille, M. (1993) The 
climate in western Europe during the last glacial/interglacial cycle derived from pollen and 
insect remains . Palaeogeography, Palaeoclimatology, Palaeoecology 103, 73-93 . 
Guiot, J., de Beaulieu, J.-L. and Reille, M. (1989) A 140,000-year continental climate 
reconstruction from two European pollen records. Nature 338, 309-313. 
Haflidason, H., Sejrup, H.P., Kristensen, D.K. and Johnsen, S. (1995) Coupled response of the 
late glacial climatic shifts of northwest Europe in Greenland ice cores: evidence from the 
northern North Sea. Geology 23 (12), 1,059-1,062. 
Harkness, D.D. (1975) The role of the archaeologist in C-14 measurement. In: Watkins, T. (ed.) 
Radiocarbon: Calibration and PrehistOl),. Edinburgh University Press, Edinburgh. 128-135. 
Harrison, S.P. and Digerfeldt, G. (1993) European lakes as palaeohydrological and palaeoclimatic 
indicators. Quaternw), Science Reviews 12, 233-248. 
Harvey, A.M. and Wells, S.G. (1987) Response of Quaternary fluvial systems to differential 
epeirogenic uplift: Aguas and Feos river systems, south-east Spain. Geology 15, 689-693. 
REFERENCES. 217 
Heaton, T.H.E., Holmes, lA. and Bridgwater, N.D. (1995) Carbon and oxygen isotope variations 
among lacustrine ostracods: implications for their use in palaeoclimatic studies. The Holocene 
5, 428-434. 
Heinrich, H. (1988) Origin and consequences of cyclic ice rafting in the northeast Atlantic Ocean 
during the past 130,000 years . Quaternary Research 29, 143-152. 
Henderson, P.A. (1990) Freshwater Ostracods: Keys and Notes for the Identification of the Species. 
Synopsis of the British Fauna (N.S.) 42. Universal Book Services / Dr W. Backhuys, 
Oegstgeest, Netherlands. 
HelTero-Bervera, E., Helsley, C.E., Hammond, R. and Chitwood, L.A. (1989) A possible 
lacustrine palaeomagnetic record of the Blake magnetic polarity episode from Pringle Falls, 
Oregon, USA. Physics of the Earth and Planetary Interiors 56, 112-113. 
Hen'ero-Bervera, E., Helsley, CE., Sarna-Wojcicki, A.M ., Lajoie, K.R., Meyer, CE., 
McWilliams, M.O., Negrini, R.M., Turrin, B.D., Donnelly-Nolan, J.M. and Liddicoat, 
J.C (1994) Age and correlation of a paleomagnetic episode in the western United States by 
4°Ar/39 Ar dating and tephrochronology: The Jamaica, Blake, or a new polarity episode? Journal 
of Geophysical Research 99 (B 12), 24,091-24,103. 
Heusser, L. and Shackleton, N.l (1979) Direct marine-continental correlation: 150,000-year oxygen 
isotope-pollen record from the North Pacific. Science 204, 837-839. 
Higgs, E.S . and Vita-Finzi, C. (1966) The climate, environment and industries of Stone Age 
Greece: Part Il. Proceedings of the Prehistoric Society 32, 1-29. 
Higgs, E.S., Vita-Finzi, C, Harris, D.R. and Fagg, A.E. (1967) The climate, environment and 
industries of Stone Age Greece: Part Ill. Proceedings of the Prehistoric Society 33, 1-29. 
Hiltermann, H. and Ltittig, G. (1969) Biofazies und Palaolimnologie der pliozanen und pleistozanen 
Seen im Megalopolis-Becken (Peloponnes). Mitteilungen derlnternationalen Vereinigung for 
Theoretische und Angewandte Limnologie 17, 306-314. 
Holmes, lA. (1992) Trace-element chemistry of non-marine ostracod shells: a preliminary evaluation 
of cleaning methods. Journal of Micropalaeontology 11, 36. 
Holmes, lA. (1996) Trace-element and stable-isotope geochemistry of non-marine ostracod shells in 
Quaternary palaeoenvironmental reconstruction. Journal of Paleolimnology 15, 223-235. 
Holmes, J.A., Hales, P. and Street-Perrott, F.A. (1992) Trace element chemistry of nonmarine 
ostracodes as a means of palaeolimnological reconstruction: an example from the Quaternary 
of Kashmir, northern India. Chemical Geology 95, 177-186. 
Howard, W.R. (1997) A warm future in the past. Nature 388, 418-419. 
Hubendick, B. (1972) The European fresh-water limpets (Ancylidae and Acroloxidae). Informations ck 
la Societe Belge de Malacologie 1 (8), 109-128. 
Hubendick, B. and Radoman, P. (1959) Studies on the Gyraulus species of Lake Ochrid. 
Morphology. Arkiv for Zoologi, Series 2 (Stockholm) 12 (16), 223-243. 
Huntley, B. and Prentice, LC (1993) Holocene vegetation and climates of Europe. In: Wright, 
H.E., Jr., Kutzbach, J.E., Webb, T., Ill, Ruddiman, W.F., Street-Perrot, F.A. and 
Bartlein, P.J. (eds.) Global Climates Since the Last Glacial Maximum. University of 
Minnesota Press, Minneapolis. 136-169. 
Imbrie, J. , Hays, lD., Martinson, D.G., McIntyre, A., Mix, A.C, Morley, J.1., Pisias, N.G ., 
Prell , W.L. and Shackleton, N.1. (1984) The orbital theory of Pleistocene climate: support 
from a revised chronology of the marine 8 180 record. In: Berger,' A., Imbrie, J., Hays, J., 
Kukla, G. and Saltzman, B. (eds.) Milankovitch and Climate. Understanding the Response 
to Astronomical Forcing. D. Reidel Publishing Co., Dordrecht, Netherlands. 269-305. 
REFERENCES • 218 
Institute of Geology and Subsurface Research (IGSR) and Institut Fran<;:ais du Petrole (IFP) (1966) 
Etude Geologique de l'Epire. Edition Technipress, Paris . 
Jackson , J .A. (1994) Active tectonics of the Aegean region . Annual Review of Earth and PlanetGl), 
Sciences 22, 239-271. 
Jacobs, J.A. (1994) Reversals of the Earth's Magnetic Field (2nd Edition). Cambridge University 
Press, Cambridge. 
Jarvis, I. and Jarvis, K.E. (1992) Plasma spectrometry in the Earth Sciences : techniques, 
applications and future trends . Chemical Geology 95, 1-33. 
Johnsen , S.1., Dansgaard, W. and White, J.W.c. (1989) The origin of Arctic precipitation under 
present and glacial conditions. Tellus 41B, 452-468 . 
Johnsen , S.1., Clausen, H.B., Dansgaard, W. , Fuhrer, K., Gundestrup, N., 
Iversen, P ., Jouzel, J., Stauffer, B. and Steffensen, J.P . (1992) 
interstadials recorded in a new Greenland ice core. Nature 359, 311-313. 
Hammer, C.U., 
Irregular glacial 
Johnsen, S.1., Clausen, H.B., Dansgaard, W., Gundestrup, N.S. , Hammer, C.V . and Tauber, H. 
(1995) The Eem stable isotope record along the GRIP ice core and its interpretation. 
Quaternary Research 43, 117-124. 
Johnson, T.c., Halfman, J.D. and Showers, W.1. (1991) Paleoclimate of the past 4,000 years at 
Lake Turkana, Kenya, based on the isotopic composition of authigenic calcite. 
Palaeogeography, Palaeoclimatology, Palaeoecology 85, 189-198. 
Jones, B.F. and Bowser, C.J. (1978) The mineralogy and related chemistry of lake sediments. In: 
Lerman, A. (ed.) Lakes: ChemistlY, Geology, Physics. Springer-Verlag, New York. 179-
236. 
Jouzel, J. (1994) Ice cores north and south. Nature 372, 612-613 . 
Jouzel, J ., Petit, J .R. and Raynaud, D. (1990) Palaeoclimatic information from ice cores: the 
Vostok record. Transactions of the Royal Society of Edinburgh: Earth Sciences 81, 349-
355. 
Kabat, A.R and Herschler, R (1993) The Prosobranch snail family Hydrobiidae (Gastropoda: 
Rissooidea) : review of classification and supraspecific taxa . Smithsonian Contributions to 
Zoology 547, 1-94. 
Kalogeropoulos, N., Karayannis, M.I., Vassilaki-Grimani , M . and Grimanis, A .P. (1994) 
Application of trace element determination and multivariate statistics to a pollution study of 
Lake Pamvotis, NW Greece. Fresnius Environmental Bulletin 3, 187-192. 
Kapsner, W .R , Alley, RB ., Shuman, C.A., Anandakrishnan, S. and Grootes, P.M. (1995) 
Dominant influence of atmospheric circulation on snow accumulation in Greenland over the 
past 18,000 years. Nature 373, 52-54. 
Kan'ow, P.F. and Anderson, T.W. (1975) Palynological study of lake sediment profiles from south­
western New Brunswick: Discussion. Canadian Journal of Earth Sciences 12, 1,808-1 ,8 12. 
KaITOW, P.F., Warner, B.G. and Fritz, P. (1984) Corry Bog, Pennsylvania: a case study of the 
radiocarbon dating of mar\. QuaternQ/Y Research 21, 326-336. 
Kawai, N . (1984) Palaeomagnetic study of the Lake Biwa sediments. In: Horie, S. (ed.) Lake Biwa. 
W.Junk, Dordrecht, Netherlands. 399-416. 
Kawai , N. , Yaskawa, K., Nakajima, T., Torii, M. and Horie, S. (1972) Oscillating geomagnetic 
field with a recUlTing reversal discovered from Lake Biwa. Proceedings of the Japan Academy 
48, 186-190. 
REFERENCES. 219 
Keller,1., Ryan, W.B.F., Ninkovitch, D. and Altherr, R. (1978) Explosive volcanic activity in the 
Mediterranean over the past 200,000 yrs as recorded in deep-sea sediments. Geological Society 
of America Bulletin 89, 591-604. 
Kelts, K. and Hsti, KJ. (1978) Freshwater carbonate sedimentation. In: Lennan, A. (ed.) Lakes: 
Chemistl)', Geology, Physics. Springer-Verlag, New York. 295-323. 
Kelts, K. and Talbot, M.R. (1989) Lacustrine carbonates as geochemical archives of environmental 
change and biotic - abiotic interactions. In: Tilzer, M.M. and Serruya, C. (eds.) 
Ecological Structure and Function in Large Lakes. Science and Technology Publishers, 
Madison, Wisconsin. 290-317. 
Kilenyi, T .!. (1972) Transient and balanced genetic polymorphism as an explanation of variable 
noding in the ostracode Cyprideis torosa. Micropalaeontology 18, 47-63. 
King, G. and Bailey, G. (1985) The palaeoenvironment of some archaeological sites in Greece: the 
influence of accumulated uplift in a seismically active region. Proceedings of the Prehistoric 
Society 51 , 273-282. 
King, G., Bailey, G. and Sturdy, D. (1994) Active tectonics and human survival strategies. ]oumal 
of Geophysical Research 99 (B 10), 20,063-20,078 . 
King, G., Sturdy, D. and Whitney, 1. (1993) The landscape geometry and active tectonics of 
northwest Greece. Geological Society of America Bulletin 105, 137-161. 
Klie, W. (1939) Studien uber Ostracoden aus dem Ohridsee: H. Limnocytherinae und Cytherinae. 
Archiv fiir Hydrobiologie 35, 631-646. 
Ko<; Karpuz, N. and Jansen, E. (1992) A high-resolution diatom record of the last deglaciation from 
the SE Norwegian Sea: documentation of rapid climatic changes. Paleoceanography 7, 499-
520. 
Kotilainen, A.T. and Shackleton, NJ. (1995) Rapid climate variability in the north Pacific Ocean 
during the past 95,000 years. Nature 377, 323-326. 
Krstic, N. (1974) Pliocenski ostrakodi Metohije, 2: Pont lokalnosti Stupelj. Bulletin du Museum 
d'Histoire Naturelle (Beograd) A29, 73-79. 
Krstic, N. (1985) New specia of the Quaternary ostracodes from Vojvodina. Radovi Geoinstituta 
(Beograd) 18, 197-206. 
Krstic, N. (1988) Some Quaternary ostracods of the Pannonian Basin with a review of a few 
neglectoida. In : T. Hanai, N . Ikeya and K. Ishizaki (eds.) Evolutionary Biology of 
Ostracoda: Kodansha Ltd, Tokyo. 1,063-1 ,072. 
Krstic, N. (1993) Several rare ostracode species of Backa Quaternary. Radovi Geoinstituta (Beograd) 
28, 173-184. 
Krstic, N. and Schornikov, E.!. (1993) Scordiscia, a new genus of Limnocytheridae. In: McKenzie, 
K.G. and Jones, P.J. (eds.) Ostracoda in the Earth and Life Sciences. A.A. Balkema, 
Rotterdam. 249-257. 
Krstic, N., Isakovic, 0., Batalovic, K. and Coric, J. (1981) Prilog poznavanju kvartarnikh 
sedimenata Mirijeva u Beogradu. Radovi Geoinstituta (Beograd) 1981, 161-174. 
Kukla, G. and Briskin, M. (1983) The age of the 4/5 isotopic stage boundary on land and in the 
oceans. Palaeogeography, Palaeoclil1latology, Palaeoecology 42, 35-45. 
Kukla, G. and Zijderveld, 1.D.A. (1977) Magnetostratigraphic pitfalls .. Nature 266, 774-776. 
Kukla, G., McManus, J.F., Rousseau, D.-D. and Chuine,!' (1997) How long and how stable was 
the last interglacial? Quaternary Science Reviews 16 (6), 605-612 
REFERENCES • 220 
Lajoie, K.R., Liddicoat, J.e. and Robinson, S.W. (1980) Refinement of the chronology and 
palaeomagnetic record at Mono Lake, California. EOS (Transactions of the American 
Geophysical Union) 61 (17), 215 . 
Lamb, H.F., Gasse, F., Benkaddour, A., El Hamouti, N., van der Kaars, S., Perkins, W.T., 
Pearce, NJ. and Roberts, e.N. (1995) Relation between century-scale Holocene arid 
intervals in tropical and temperate zones. Nature 373, 134-137. 
Lauritzen, S.-E. (1990) Uranium series dating of speleothems: a glacial chronology for Nordland, 
Norway, for the last 600ka. Striae 34, 127-132. 
Lauritzen, S.-E. (1995) High-resolution palaeotemperature proxy record for the last interglaciation 
based on Norwegian speleothems. Quaternw), Research 43, 133-146. 
Le Borgne, E. (1955) Abnonnal magnetic susceptibility of the topsoil. Annales de Geophysique 11, 
399-419. 
Lehman, SJ. and Keigwin, L.D. (1992) Sudden changes in North Atlantic circulation during the last 
deglaciation. Nature 356, 757-762. 
Le Pichon, X., Chamot-Rooke, N., Lallemant, S., Noomen, R. and Veis, G. (1997) Geodetic 
determination of the kinematics of Central Greece with respect to Europe: implications for 
eastern Mediterranean tectonics. Journal of Geophysical Research 100, 12,675-12,690. 
Levi, S. and Karlin, R. (1989) A sixty thousand year paleomagnetic record from Gulf of California 
sediments: secular variation, late Quaternary excursions and geomagnetic implications. EO/th 
and Planetar), Science Letters 92, 219-233 . 
Levi, S., Audunsson, H., Duncan, R.A., Kristjansson, L., Gillot, P.-Y . and Jakobsson, S.P . 
(1990) Late Pleistocene geomagnetic excursion in Icelandic lavas: confirmation of the 
Laschamp excursion. Earth and Planetary Science Letters 96, 443-457. 
Lewin, J., Macklin, M.G. and Woodward, J.e. (1991) Late Quaternary fluvial sedimentation in the 
Voidomatis Basin, Epirus, northwest Greece. Quaternw), Research 35, 103-115. 
Liddicoat, J.e. (1990) Aborted reversal of the palaeomagnetic field in the Brunhes Normal Chron in 
east-central California. Geophysical Journal International 102, 747-752. 
Liddicoat, J.e., Lajoie, K.R., Bailey, R.A., Sarna-Wojcicki, A.M., Russell, R.C. and Woodward, 
M. (1980) Reversal of the paleomagnetic field in Brunhes-age lacustrine sediments in Long 
Valley and Mono basin, California. EOS (Transactions of the American Geophysical Union) 
61 (17), 215 . 
Liddicoat, le., Lajoie, K.R. and Sarna-Wojcicki, A.M. (1982) Detection and dating of the Mono 
Lake excursion in the Lake Lahontan Sehoo Formation, Carson Sink, Nevada. EOS 
(Transactions of the American Geophysical Union) 63 (45), 920. 
Linsley, B.K. (1996) Oxygen-isotope record of sea level and climate variations in the Sulu Sea over 
the past 150,000 years. Nature 380, 234-237. 
Lister, G.S. (1988) A 15,000 year isotopic record from Lake Zurich of deglaciation and climate 
change in Switzerland. Quaternw), Research 29, 129-141. 
Lister, G.S., Kelts, K. , Zao, e.K., Yu, J.-Q. and Niessen, F. (1991) Lake Quinghai , China: 
closed-basin lake levels and the oxygen isotope record for ostracoda since the latest 
Pleistocene. Palaeogeography, Palaeoclimatology, Palaeoecology 84, 141-162. 
Litt, T., Junge, F.W. and Bottger, T. (1996) Climate during the Eemian in north-central Europe - a 
critical review of the palaeobotanical and stable isotope data from central Gelmany. 
Vegetation HistOl), alld Archaeobotany 5, 247-256. 
Laffler, H.H. and Danielopol, D.L. (1978) Ostracoda. In: Illies, l (ed.) Limnofauna Europaea. 
Gustav Fischer Verlag, Stuttgart. 196-208. 
REFERENCES • 221 
Longworth, G., Becker, L.W., Thompson, R., Oldfield, F., Dearing, J.A. and Rummery, T.A. 
(1979) Mossbauer effect and magnetic studies of secondary iron oxides in soils. lournalof 
Soil Science 30, 93-110. 
L~vlie, R. (1989) Palaeomagnetic stratigraphy: a correlation method. Quaternary> International 1, 
129-149. 
Lowe, JJ. and Walker, M.J.C. (1997) Reconstructing Quaternmy Environments (2nd Edition). 
Longman Group Ltd., Harlow. 
Lowe, J.J., Ammann, B., Birks, H.H., Bjorck, S. , Coope, G.R., Cwynar, L.e., de Beaulieu, 1.­
L. , Mott, R.J., Peteet, D.M. and Walker, M.J.C. (1994) Climatic changes in areas 
adjacent to the North Atlantic during the last glacial - interglacial transition (l4-9ka BP). 
lournal of Quaternmy Science 9, 185-198. 
Lowe, J.J. , Coope, G.R., Sheldrick, C., Harkness, D.D. and Walker, M.J.C. (1995) Direct 
comparison of UK temperatures and Greenland snow accumulation rates. lournal of 
Quaternaty Science 10, 175-180. 
Ludlam, S.D. (1974) Fayetteville Green Lake, New York. 6. The role of turbidity currents in lake 
sedimentation. Limnology and Oceanography 19, 656-664. 
L'vova, A.A. and Starobogatov, Y.I. (1982) Novyj vid drejsseny (Bivalvia, Dreissenida) IZ 
Okhridskogo ozera. Zoologicheskij Zhurnal (Moskva) 61 (11), 1749-1752. 
Macan, T.T. (1969) Scientific Publication No.]3. A Key to the British Fresh- and Brackish-Water 
Gastropods. Freshwater Biological Association, Ambleside. 
MacAyeal , D.R. (1993) Binge/purge oscillations of the Laurentide ice sheet as a cause of the North 
Atlantic's Heinrich Events. Paleoceanography 8 (6), 775-784. 
Mackereth, FJ.H. (1966) Some chemical observations on postglacial lake sediments. Philosophical 
Transactions of the Royal Society of London B250, 165-213. 
MacLeod, D.A. (1980) The origin of the red Mediterranean soils in Epirus, Greece. lournal of Soil 
Science 31, 125-136. 
MacLeod, D.A. and Vita-Finzi , e. (1982) Environment and provenance in the development of recent 
alluvial deposits in Epirus, NW Greece. Earth Swface Processes and Landforms 7, 29-43. 
Magny, 1., Mouthon,1. and Ruffaldi, P. (1995) Late Holocene level fluctuations of the Lake islay 
in Jura, France: sediment and mollusc evidence and climatic implications. lournal of 
Paleolimnology 13,219-229. 
Magri, D. (1989) Interpreting long-term exponential growth of plant populations in a 250,OOO-year 
pollen record from Valle di Castiglione (Roma). New Phytologist 112, 123-128. 
Manabe, K.-1. (1977) Reversed magnetozone in the Late Pleistocene sediments from the Pacific coast 
of Odaka, northeastern Japan. Quaternary Research 7 372-379. 
Mangerud, 1. (1989) Correlation of the Eemian and the Weichselian with deep sea oxygen isotope 
stratigraphy. Quaternmy International 3/4, 1-4. 
Mangerud, J., Andersen, S.T., Berglund, B.E. and Donner, J.J. (1974) Quaternary stratigraphy of 
Norden: a proposal for terminology and classification. Boreas 3, 109-127. 
Marinos, P. (1974) Hydrogeological Study of the Spring Krias of Ioannina. Unpublished report. 
Hydroresearch Co., Athens. 1-143. 
Martens, K. (1992) A reassessment of Paralimnocythere CARBONNEL, 1965 (Crustacea, Ostracoda, 
Limnocytherinae), with a description of a new genus and two new species. Bulletin van het 
Koninklijk Belgisch Instituut voor Natuurwetenschappen, Biologie 62, 125-168. 
REFERENCES • 222 
Martinson, D.G., Pisias , N.G., Hays, J.D., Imbrie, J., Moore, T.e., Jr. and Shackleton, N.J. 
(1987) Age dating and the orbital theory of the Ice Ages: development of a high-resolution 0 
to 300,000-year chronostratigraphy. Quatemmy Research 27 1-29. 
Maslin, M. and Tzedakis, P.e. (1996) Sultry last interglacial gets sudden chill. EOS (Transactions of 
the American Geophysical Union) 77 (37), 353-354. 
McGregor, D.L. (1969) The reproductive potential, life history and parasitism of the freshwater 
ostracod Darwinula stevensoni. In: Neale, J.W. (ed.) The Taxonomy, Morphology and 
Ecology of Recent Ostracoda Oliver and Boyd, Edinburgh. 194-221. 
McKenzie, J.A. (1985) Carbon isotopes and productivity in the lacustrine and marine environment. 
In: Stumm, W. (ed.) Chemical Processes in Lakes. John Wiley and Sons Inc., New York. 
99-118. 
McManus, J.F., Bond, G.e., Broecker, W.S., Johnsen, S., Labeyrie, L. and Higgins, S. (1994) 
High-resolution climate records from the North Atlantic during the last interglacial. Nature 
371, 326-329. 
Meier-Brook, e. (1983) Taxonomic studies on Gyraulus (Gastropoda: Planorbidae). Malacologia 24 
1-113. 
Melloni, M. (1853) Du magnetisme des roches. Comptes Rendus de l'Academie des Sciences 37, 
966-968. 
Meynadier, L., Valet, J.-P., Weeks, R., Shackleton, N.J. and Hagee, V.L. (1992) Relative 
geomagnetic intensity of the field during the last l40ka. Earth and Planetmy Science Letters 
114, 39-57. 
Mikulic, F. (1961) Nove Candona vrste iz Ohridskog Jezera. Bulletin du Museum d'Histoire 
Naturelle (Beograd) B17, 87-107. 
Mbrner, N.-A. (1977) The Gothenburg Magnetic Excursion. Quaternmy Research 7, 413-427. 
Mbrner, N.-A. and Lanser, J.P. (1974) Gothenburg Magnetic 'Flip'. Nature 251, 408-409. 
Mbrner, N.-A., Lanser, J.P. and Hospers, J. (1971) Late Weichselian palaeomagnetic reversal. 
Nature 234, 173-174. 
Mostafawi, N. (1986) Pleistozane Ostracoden aus der Nikolaos-Formation von Ost-Kos, Greichenland. 
Senckenbergiana Lethaea 67, 275-303. 
Mostafawi, N. (1988) SiH3wasser-Ostracoden aus dem Plio-Pleistozan del' Inseln Kos (Gl'iechenland). 
Meynialla 40, 175-193. 
Mostafawi, N. (1994) SuBwasser-Ostracoden aus dem Ober-Pliozan von N-Eubba (Greichenland). 
Neues lahrbuch fur Geologie und Paltiontologie Monatsheften 5, 309-319. 
Mousson, A. (1859) Coquilles Terrestres et Fluviatiles . Recueilles dans l'Orient par M. le Dr 
Alexandre Schlaefli. Vierteljahrsschrift del' Natwforschenden Gesellschaft in Zurich 4, 37-
71. 
Mouthon, J. (1992) Snail and bivalve population analyses in relation to physio-chemical quality of 
lakes in eastern France. Hydrobiologia 245, 147-156. 
Muller, G., Irion, G. and Fbrstner, U. (1972) Formation and diagenesis of inorganic Ca-Mg 
carbonates in the lacustrine environment. Natunvissenschaften 59, 158-164. 
Mullins, e.E. (1977) Magnetic susceptibility of the soil and its significance in soil science. lournal 
of Soil Science 28, 223-246. 
Munsell Soil Color Charts (1975) Munsell Col or, Baltimore, Maryland . 
REFERENCES. 223 
Narcisi, B. (1996) Tephrochronology of a Late Quaternary lacustrine record from the Monticchio Maar 
(Vulture Volcano, southern Italy). Quaternal), Science Reviews 15, 155-165. 
Narcisi, B., Anselmi, B., Catalano, F., Dai Pra, G. and Magri, G. (1992) Lithostratigraphy of the 
250,000 year record of lacustrine sediments from the Valle di Castiglione crater, Roma. 
Quaternary Science Reviews 11, 353-362. 
Negrini, R.M., Davis, lO. and Verosub, KL. (1984) Mono Lake geomagnetic excursion found at 
Summer Lake, Oregon. Geology 12, 643-646. 
Negrini, R.M., Erbes, D.B ., Roberts, A.P. , Verosub, KL., Sarna-Wojcicki, AM. and Meyer, 
C.E. (1994) Repeating waveform initiated by a 180-190ka geomagnetic excursion in western 
North America: implications for field behavior during polarity transitions and subsequent 
secular variation. Journal of Geophysical Research 99 (BI2), 24,105-24,119. 
Negrini, R.M., Verosub, K.L. and Davis, J.O. (1987) Long-telm nongeocentric axial dipole 
directions and a geomagnetic excursion from the Middle Pleistocene sediments of the 
Humboldt River Canyon, Pershing County, Nevada. Journal of Geophysical Research 92 
(B 10), 10,617-10,627. 
Negrini , R.M., Verosub, K.L. and Davis, lO. (1988) The middle to late Pleistocene geomagnetic 
field recorded in fine-grained sediments from Summer Lake, Oregon and Double Hot Springs, 
Nevada, USA Earth and Planetary Science Letters 87, 173-192. 
Nowaczyk, N.R. and Baumann, M. (1992) Combined high resolution magnetostratigraphy and 
nannofossil biostratigraphy for late Quaternary Arctic Ocean sediments. Deep Sea Research 
39, 567-601. 
Nowaczyk, N.R., Fredrichs, T.W., Eisenhauer, A and Gard, G. (1994) Magnetostratigraphic data 
from late Quaternary sediments from the Yelmak Plateau, Arctic Ocean: evidence for four 
geomagnetic polarity events within the last 170Ka of the Brunhes Chron . Geophysical 
Journal International 117, 453-471. 
0kland, l (1990) Lakes and Snails. Environment and Gastropoda in 1,500 Norwegian Lakes. 
Universal Book Services / Dr W. Backhuys, Oegstgeest. 
Oldfield, F. (1983) The role of magnetic studies in palaeohydrology. In: Gregory, KJ. (ed.) 
Background to Palaeohydrology John Wiley and Sons Ltd., Chichester. 141-165. 
Olssen, LU. (1986) Radiometric dating. In: Berglund, B.E. (ed.) Handbook of Holocene 
Palaeoecology and Palaeohydrology. John Wiley and Sons Ltd., Chichester. 273-312. 
Overbeck, l (1980) Appendix to Anagnostidis, K and Economou-Amilli, A Limnological studies 
on Lake Pamvotis (Ioannina), Greece. Archiv fUr Hydrobiologie 89 (3), 342. 
Palacios-Fest, M.R., Cohen, A.S . and Anad6n, P. (1994) Use of ostracodes as paleoenvironmental 
tools in the interpretation of ancient lacustrine records. Revista Espaiiola de Paleontologia 9 
(2), 145-164. 
Peel, D.A. (1995) Profiles of the past. Nature 378, 234-235. 
Pennington, W., Haworth, E.Y., Bonny, AP. and Lishman, J.P. (1972) Lake sediments in 
northern Scotland. Philosophical Transactions of the Royal Society of London B 264, 191-
294. 
Perrier, R., Potier, L., Savoyat, E., Koukouzas, K and Monopolis, D. (1967) Geological Map of 
Greece: loannina Sheet, 1:50,000. Institute for Geology and Subsurface Research, Athens. 
Petkovski, T.K (1959) Beitrag zur Kenntnis der Ostracoden-Fauna Jugoslawiens (V). Istanbul 
University Fen Fakiiltesi Hidrobiologi (Series B) 4, 159-155. 
Petkovski, T.K. (1960) Zur Kenntnis der Crustaceen des Skandar- (Scutari-) Sees. Acta Musei 
Macedonici Scientiarum Naturalium (Skopje) 8 (2), 29-52. 
REFERENCES • 224 
Petkovski, T.K. (1961) Zur Kenntnis del' Crustaceen des Prespasees. Fragmenta Balcanica Musei 
Macedonici Scientarium Naturalium (Skopje) 3 (15), 117-131. 
Petkovski, T.K. (1969a) Einige neue und Bemerkenswerte Candoninae aus dem Ohl'idsee und einigen 
anderen Fundorten in Europa. Acta Musei Macedonici Scientiarum Naturalhtln (Skopje) 11 
(5), 81-111. 
Petkovski, T.K. (1969b) Uber die Notwendigkeit einer revision del' siisswasser-ostracoden Europea. 
In: Neale, lW. (ed.) The Taxonomy, MOIphology and Ecology of Recent Ostracoda:. 
Oliver & Boyd, Edinburgh. 76-81. 
Petkovski, T.K. and Keyser, D. (1992) Leptocythere ostrovskensis sp. n. (Crustacea, Ostracoda, 
Cytheracea) aus dem See Vegoritis (Ostrovsko Ezero) in NW Greichenland. Mitteilungen aus 
den Hamburgischen Zoologischen Museum und Institut 89, 227-237. 
Pilcher, J.R. (1991) Radiocarbon dating. In: Smart, P.L. and Frances, P.D. (eds.) Quatemmy 
Dating Methods - A User's Guide. Quaternary Research Association, London. 16-36. 
Piper, DJ.W., Panagos, A.G. and Pe, G.G. (1978) Conglomeratic Miocene flysch , western Greece. 
lournal of Sedimentaty Petrology 48 (I), 117-126. 
Ponel, P. (1995) Rissian, Eemian and Wiirmian Coleoptera assemblages from La Grande Pile 
(Vosges, France). Palaeogeography, Palaeoclimatology, Palaeoecology 114, 1-41. 
Pons, A. and Reille, M. (1988) The Holocene and Upper Pleistocene pollen record from Padul 
(Grenada, Spain): a new study. Palaeogeography, Palaeoclimatology, Palaeoecology 66 , 
243-263 . 
Pope, K.O. and van Andel, T.H. (1984) Late Quaternary alluviation and soil formation in the 
southern Argolid: its history, causes and archaeological implications. loumal of 
Archaeological Science 11, 281-306. 
Prashad, B. (1929) Recent and fossil Viviparidae. A study in distribution, evolution and 
palaeogeography. Memoirs of the Indian Museum 8 (4), 153-252. 
Prell, W.L., Imbrie, J., Martinson, D.G., Morley, IJ., Pisias, N.G., Shackleton, NJ. and 
Streeter, H.F. (1986) Graphic con'elation of oxygen isotope stratigraphy application to the 
Late Quaternary . Palaeoceanography 1 (2), 137-162. 
Prentice, Le., Guiot, J . and Harrison, S.P. (1992) Mediterranean vegetation, lake levels and 
palaeoC\imate at the Last Glacial Maximum. Nature 360, 658-660. 
Pyle, D.M., van Andel, T.H., Paschos, P. and van den Bogaard, P. (in press) A Middle Pleistocene 
volcanic ash layer from Epirus, Greece. Quaternaty Research. 
Rampino, M.R. (1981) Revised age estimates of Brunhes palaeomagnetic events: support for a link 
between geomagnetism and eccentricity. Geophysical Research Letters 8 (10), 1,047-1,050. 
Raymo, M.E., Ruddiman, W.F., Backman, l, Clement, B.M. and Martinson, D.G. (1989) Late 
Pleistocene evolution of the northern hemisphere ice sheets and North Atlantic deep water 
circulation. Paleoceanography 4, 413-446. 
Readman, P.W., Coey, lM.D., Mosser, e. and Weber, F. (1976) Analysis of some lake 
sediments from Greece. Journal de Physique: Colloque C6-37 (12), 845-848. 
RegneU, J. (1992) Preparing pollen concentrates for AMS dating - a methodological study from a 
hard-water lake in southern Sweden. Boreas 21, 373-377. 
Reille, M. and de Beaulieu, J.-L. (1990) Pollen analysis of a long Upper Pleistocene sequence in a 
Velay maar (Massif Central, France). Palaeogeography, Palaeoclimatology, Palaeoecology 
80, 35-48. 
REFERENCES. 225 
Reille, M. and de Beaulieu, J.-L. (1995) Long Pleistocene pollen records from the Praclaux Crater, 
south-central France. QuaternGlY Research 44, 205-2] 5. 
Reille, M., Guiot, J. and de Beaulieu, J.-L. (1992) The Montaigu Event: an abrupt climatic change 
during the early Wiilm in Europe. In: Kukla, G.J. and Went, E. (eds.) Start of a Glacial. 
NATO ASI Series, Vol. 13, Springer Verlag, Berlin. 85-95. 
Reischiitz, P.L. and Sattmann, H. (1990) Beitrage zur Molluskenfauna des Epirus , 11. Annalen des 
Naturhistorischen Museums in Wien B91, 253-272. 
Roberts, N. (1982) Lake levels as indicators of Near Eastern palaeoclimates. In: Bintliff, J.L. and 
van Zeist, W. (eds.) Palaeoclimates, Palaeoenvironments alld Human Communities ill the 
Eastern Mediterranean Region in Later PrehistOlY. British Archaeological Reports 
International Series 133, 235-267. 
Roberts, N. (1983) Age, palaeoenvironments and climatic significance of Late Pleistocene Konya 
Lake, Turkey. QuaternGlY Research 19, 154-171. 
Robinson, C. (1994) Lago Grande Monticchio, southern Italy: a long record of environmental change 
illustrated by sediment geochemistry. Chemical Geology 118, 235-254. 
Rommelt-Doll, J. (1990) Ostrakoden aus neogenen und pleistozanen Sediment-Zyklen am Isthmus 
von Korinth. Senckenbergialla Lethaea 70, 171-197. 
Rosenberg, G. and Ludyanskiy, M.L. (1994) A nomenclatural review of Dreissella (Bivalvia: 
Dreissenidae), with identification of the quagga mussel as Dreissena bugensis. Canadian 
Journal of Fisheries and Aquatic Sciences 51, 1,474-1,484. 
Rossignol-Strick, M. (1995) Sea-land correlation of pollen records in the eastern Mediterranean for the 
glacial-interglacial transition: biostratigraphy versus radiometric timescale. QuaternaJ)' 
Science Reviews 14, 893-915. 
Ruddiman, W.F., McIntyre, A.F. and Raymo, M.E. (1986) Matuyama 4] ,000-year cycle: North 
Atlantic Ocean and northern hemisphere ice sheets. EaJth and PlanetGlY Science Letters 80, 
]17-129. 
Ryan, W.B .F. (1972) Stratigraphy of late Quaternary sediments in the Eastern Mediterranean. In: 
Stanley, D.J. (ed.) The Mediterranean Sea: A Natural Sedimentation LaboratOlY Dowden, 
Hutchison and Ross, Stroudsberg, Pennsylvania. 149-169. 
Sarna-Wojcicki, A.M., Lajoie, K.R., Meyer, C.E., Adam, D.P. and Rieck, H.J. (1991) 
Tephrochronological cOlTelation of upper Neogene sediments along the Pacific margin , 
conterminous United States. In : Morrison, R. (ed.) The Geology of North America. 
Volume K2, NOIl-Glacial QuaternaJY of the United States. Geological Society of America, 
Boulder, Colorado. 117-140. 
Sattmann, H. and Reischiitz, P.L. (1988) Beitrage zur Molluskenfauna Nordgreichenlands. Allnalen 
des Naturhistorischen Museums in Wiell B90, 203-214. 
Schelske, c.L. and Hodell, D.A. (1991) Recent changes in productivity and climate of Lake Ontario 
detected by isotopic analysis ofsediments. Limllology and Oceanography 36 (5), 961-975. 
Schiitt, H. (1962) Neue SiiBwasser-Prosobranchier Greichenlands. Archiv fill' Molluskenkullde 91 
(4), 157-166. 
Schiitt, H. (1980) Zur Kenntnis greichischer Hydrobiiden. Archiv fill' Molluskenkullde 110 (4), 
115-149. 
Schiitt, H. (1987) Limnische Mollusken aus alterem Quartar Makedoniens. Zoologische 
Mededelillgell 61 (8), J 13-121. 
Schiitt, H. (1993) Die Gattung Dreissena im Quartar Anatoliens (Bivalvia Eumellibranchiata: 
Dreissenacea). Archiv fill' Molluskenkunde 122, 323-333. 
REFERENCES • 226 
Schiitt, H., Velitzelos, E. and Kaouras, G. (1985) Die QuartiiImollusken von Megalopolis 
(Greichenland). Neues jahrbuchfiir Geologie und Paliiontologie Abhandlungen 170 (2), 183-
204. 
Schwalb, A., Lister, G.S. and Kelts, K. (1994) Ostracode carbonate 0180- and o13C-signatures of 
hydrological and climatic changes affecting Lake Neuchatel, Switzerland, since the latest 
Pleistocene. journal of Paleolimnology 11, 3-17. 
Schwarcz, H.P. and Eyles, N. (1991) Laurentide ice sheet extent inferred from stable isotopic 
composition (0, C) of ostracodes at Toronto, Canada. Quaternary Research 35, 305-320. 
Schwarcz, H.P. and Latham, A.G. (1989) Dirty calcites. 1. Uranium series dating of contaminated 
calcites using leachates alone. Chemical Geology (Isotopes Geosciences Section) 80, 35-43. 
Seidenkrantz, M.S. (1993) Benthic foraminiferal and stable isotope evidence for a 'Younger Dryas-
style' cold spell at the Saalian - Eemian transition, Denmark. Palaeogeography, 
Palaeoclimatology, Palaeoecology 102, 13-120. 
Seidenkrantz, M.S. and Knudsen, K.L. (1994) Marine high resolution records of the last interglacial 
in northwest Europe: a review. Geographie Physique et Quaternaire 48 (2), 157-168. 
Seidenkrantz, M.S ., Kristensen, P. and Knudsen, K.L. (1995) Marine evidence for climatic 
instability during the last interglacial in shelf records from northwest Europe. journal of 
Quaternmy Science 10 (1), 77-82. 
Seidenkrantz, M.S., Bornmalm, L., Johnsen, S.J., Knudsen, K.L., Kuijpers, A., Lauritzen, S.-E., 
Leroy, S.A.G., Mergeai,1., Schweger, C. and van Vliet-Lanoe, B. (1996) Two-step 
deglaciation at the Oxygen Isotope Stage 6/5e transition: the Zeifen-Kattegat Climate 
Oscillation. Quaternary Science Reviews 15 (1), 63-75. 
Serafimova-Hadzische, J. (1974) Zooplankton in some lakes of Aegean lake zone. Fragmenta 
Balcanica (Skopje) 9 (17), 165-168. 
Shackleton, N.J . (1969) The last interglacial in the marine and terrestrial records. Proceedings of the 
Royal Society of London B174, 135-154. 
Shackleton, N.J. (1987) Oxygen isotopes, ice volumes and sea level. Quaternary Science Reviews 
6, 183-190. 
Shackleton, N.J. (1989) The Plio-Pleistocene ocean: stable isotope history. In: Rose, J. and 
Schliichter, C. (eds.) Quatemary Type Sections: Imagination or Reality? Balkema, 
Rotterdam. 11-24. 
Shackleton, N.J. and Opdyke, N.D. (1973) Oxygen isotope and palaeomagnetic stratigraphy of 
equatorial Pacific core V28-238: oxygen isotope temperatures and ice volumes on a 105 year 
and 106 year scale. Quaternmy Research 3, 39-55. 
Shackleton, N.J. and Opdyke, N.D. (1976) Oxygen isotope and paleomagnetic stratigraphy of 
equatorial Pacific core V28-238, Late Pliocene to latest Pleistocene. In: Cline, RM. and 
Hays, J.D. (eds.) Investigation of Late Quaternary Paleoceanography and Paleoclimatology 
Geological Society of America Memoir 145. 449-464. 
Shackleton, N.J., Berger, A. and Peltier, W.R (1990) An alternative astronomical calibration of the 
lower Pleistocene timescale based on ODP Site 677. Transactions of the Royal Society of 
Edinburgh: Earth Sciences 81, 251-261. 
Shackleton, N.J., Imbrie, J. and Hall, M.A. (1983) Oxygen and carbon isotope record of East 
Pacific core VI9-30: implications for the formation of deep water in the late Pleistocene North 
Atlantic. Earth and Planetary Science Letters 65, 233-244. 
Shotton, F.W. (1972) An example of hard-water error in radiocarbon dating of vegetable matter. 
Nature 240, 460-461. 
\1 
REFERENCES • 227 
Siegenthaler, U. and Eicher, U. (1986) Stable oxygen and carbon isotope analyses. In : Berglund, 
B.E. (ed.) Handbook of Holocene Palaeoecology and Palaeohydrology John Wiley and Sons 
Ltd., Chichester. 407-422. 
Smart, P.L. (1991) Uranium series dating. In: Smart, P.L. and Frances, P .D. (eds.) QuatemGl), 
Dating Methods - A User's Guide. Quaternary Research Association, London. 128-160. 
Smit, A. and Wijmstra, T .A. (1970) Application of transmission electron microscope analysis to 
the reconstruction of former vegetation. Pollen et Spores 13, 615-621. 
Smith, J.D. and Foster, J.H. (1969) Geomagnetic reversal in Brunhes nOlmal polarity epoch. 
Science 163, 565-567. 
Sokac, A. (1976) Pleistocenska fauna Ostrakoda iz nekih busotina u istocnoj Slavoniji (sjeverna 
Hrvatska) . Geoloski Vjesnik (Zagreb) 29, 159-172. 
Sokac, A. and van Harten , D. (1978) /lyocypris slavonica n. sp. (Crustacea, Ostracoda) from the 
Pleistocene of Vincovci (Slavonija, north Croatia) and of Tegelen (the Netherlands). Geoloski 
Vjesnik (Zagreb) 30, 219-223. 
Stankovic, S. (1960) The Balkan Lake Ohrid and its living world. Monographiae Biologicae 9, 1-
357. 
St. Seymour, K. and Christianis, K. (1995) Correlation of a tephra layer in Western Greece with a 
Late Pleistocene eruption in the Campanian Province of Italy. QuatenlGl), Research 43 , 46-
54 . 
Stuiver, M. (1970) Oxygen and carbon isotope ratios of fresh water carbonates as climatic indicators. 
Journal of Geophysical Research 75, 5,247-5,257. 
Sykes, G. (1991) Amino Acid Dating. In : Smart, P.L. and Frances, P.D. (eds.) Quatel71G1), 
Dating Methods - A User's Guide. Quaternary Research Association, London. 161-176. 
Tabacchi, E. and Marmonier, P. (1994) Dynamics of the interstitial ostracod assemblage of a pond 
in the Adour alluvial plain . Archiv fitr Hydrobiologie 131 (3), 321-340. 
Talbot, M.R (1990) A review of the palaeohydrological interpretation of carbon and oxygen isotopic 
ratios in primary lacustrine carbonates. Chemical Geology (Isotope Geoscience Section) 80, 
261-279. 
Tarling, D.H. (1983) Palaeomagnetism. Chapman and Hall , London. 
Taylor, K.C., Lamorey, G.W., Doyle, G.A. , Alley, R.B., Grootes, P.M., Mayewski, P .A., 
White, J.W.C. and Bm·low, L.K. (1993) The 'flickering switch' of late Pleistocene climate 
change. Nature 361, 432-436. 
Thompson , R. (1973) Palaeolimnology and palaeomagnetism. Nature 242, 182-184. 
Thompson, R, Battarbee, RW. , O'Sullivan, P.E. and Oldfield, F. (1975) Magnetic susceptibility 
of lake sediments. Limnology and Oceanography 20, 687-698. 
Thouveny, N., Creer, K.M. and Blunk,1. (1990) Extension of the Lac du Bouchet palaeomagnetic 
record over the last 120,000 years . Earth and PlanetGl), Science Letters 97, 140-16l. 
Thouveny, N., Beaulieu, J.-L. , Bonifay, E., Creer, K.M., Guiot, 1. , kole, M ., Johnsen, S., 
Jouzel , 1., Reille, M., Williams, T. and Williamson, D. (1994) Climate variations in 
Europe over the past 140 kyr deduced from rock magnetism. Nature 371, 503-506. 
Thunell, R.e. and Mortyn, P.G. (1995) Glacial climate instability in the northeast Pacific Ocean. 
Nature 376, 504-506. 
Tite, M.S. and Linnington, R.E. (1975) Effect of climate on the magnetic susceptibility of soils. 
Nature 256, 565-566. 
REFERENCES • 228 
Tric, E., Laj, C., Valet, J.-P. , Tucholka, P. Paterne, M. and Guichard, F. (1991) The Blake 
geomagnetic event: transition geometry, dynamical characteristics and geomagnetic 
significance. Earth and Planetary Science Letters 102, 1-l3. 
Truesdell , A.H. and Jones, B.F. (1974) WATEQ, a computer program for calculating chemical 
equilibria of natural waters. U.S. Geological Survey lournal of Research 2, 233-248. 
Turner, lV., Fritz, P., Karrow, P.F. and Warner, B.G. (1983) Isotopic and geochemical 
composition of marl lake waters and implications for radiocarbon dating of marl lake 
sediments. Canadian lournal of Earth Sciences 20, 599-615. 
Turpen, lB. and Angell, R.W. (1971) Aspects of molting and calcification In the ostracode 
Heterocypris . Biological Bulletin 140, 331-338. 
Tzedakis, P.e. (1991) Vegetation dynamics in northwest Greece in response to QuaternalY climatic 
cycles. Ph.D thesis (unpublished), University of Cambridge. 
Tzedakis, P.C. (1993) Long-term tree populations in northwest Greece through multiple Quaternary 
climatic cycles. Nature 364, 437-440. 
Tzedakis, P.e. (1994) Vegetation change through glacial- interglacial cycles: a long pollen sequence 
perspective. Philosophical Transactions of the Royal Society B345, 403-432. 
Tzedakis, P.e. , Bennett, K.D. and Magri, D. (1994) Climate and the pollen record. Nature 370, 
513. 
Tzedakis, P.e. , Andrieu, V., de Beaulieu, J.-L., Crowhurst, S., Follieri, M., Hooghiemstra, H., 
Magri, D., Reille, M., Sadori, L., Shackleton, N.J . and Wijmstra, T.A. (1997) 
Comparison of teITestrial and marine records of changing climate of the last 500,000 years. 
Earth alld Planetary Science Letters 150, 171-176. 
Valet, l-P. and Meynadier, L. (1993) Geomagnetic field intensity and reversals during the past four 
million years. Nature 366, 234-238. 
Vallentyne, J.R. (1963) Isolation of pyrite spherules from Recent sediments. Limnology and 
Oceanography 8, 16-30. 
van Andel, T.H. (in press) Paleosols, red beds and the Old Stone Age in Greece. Geoarchaeology. 
van Andel , T.H. and Tzedakis, P.e. (1996) Palaeolithic landscapes of Europe and environs, 150,000-
25,000 years ago: an overview. Quaternary Science Reviews 15 (5/6), 481-500. 
van der Wiel , A.M. and Wijmstra, T.A. (1987a) Palynology of the lower part (78-120m) of the core 
Tenaghi Philippon 11, Middle Pleistocene of Macedonia, Greece. Review of Palaeobotany and 
Palynology 52, 73-88. 
van der Wiel, A.M. and Wijmstra, T.A. (1987b) Palynology of the 112.8-197.8m interval of the 
core Tenaghi Philippon Ill, Middle Pleistocene of Macedonia. Review of Palaeobotany C07d 
Palynology 52, 89-117. 
van Donk, J. (1976) An 180 record of the North Atlantic Ocean for the entire Pleistocene. Geological 
Society of America Memoir 145, 147-164. 
van Hat"ten, D. (1975) Size and environmental salinity in the modern euryhaline ostracod Cyprideis 
torosa (Jones, 1850), a biometrical study. Palaeogeography, Palaeoclimatology, Palaeoecology 
17, 35-48. 
van Zeist, W. and Bottema, S. (1982) Vegetational history of the eastern Mediterranean and the Near 
East during the last 20,000 years. In: Bintliff, lL. and van Zeist, W. (eds.) 
Palaeoclil1lates, Palaeoenvironments and the Human Communities in the Eastern 
Mediterranean Region in Later Prehistory. British Archaeological Reports International Series 
133, 277-321. 
REFERENCES. 229 
Verosub, K.L. (1982) Geomagnetic excursions: a critical assessment of the evidence as recorded in 
sediments of the Brunhes Epoch. Philosophical Transactions of the Royal Society of London 
A306, 161-168. 
Vinci, A. (1985) Distribution and chemical composition of tephra layers from Eastern Meditenanean 
abyssal sediments. Marine Geology 64, 143-155. 
Vita-Finzi, C. (1969) The Mediterranean Valleys: Geological Changes in Historical Times . 
Cambridge University Press , Cambridge. 
Vita-Finzi , C. (1978) Archaeological Sites in their Setting. Thames and Hudson Ltd., London . 
Vaught, K.c. (1989) A Classification of the Living Mollusca. American Malacologists, Inc., 
Melbourne, USA. 
von Grafenstein, U., Erlenkeuzer, H., Kleinmann, A., Mi.iller, 1. and Trimborn, P. (1994) High­
frequency climatic oscillations during the last deglaciation as revealed by oxygen-isotope 
records of benthic organisms (Ammersee, southern Germany). Journal of Paleolimnology 
11, 349-357. 
von Grafenstein, U., ErIenkeuzer, H., Mi.iller, J. and Kleinmann-Eisenmann, A. (1992) Oxygen 
isotope records of benthic ostracods in Bavarian lake sediments. Reconstruction of Late and 
Post Glacial air temperatures. Naturwissenschaften 79, 145-152. 
von Grafenstein, U., ErIenkeuzer, H., Mi.iller, J., Trimborn, P. and Alefs, 1. (1996) A 200 year 
mid-European air temperature record preserved in lake sediments: an extension of the 3180 p-air 
temperature relation into the past. Geochimica et Cosmochimica Acta 60 (21) , 4,025-
4,036. 
Wagstaff, J.M. (1981) Buried assumptions: some problems in the interpretation of the 'Younger Fill' 
raised by recent data from Greece. Journal of Archaeological Science 8, 247-264. 
Walker, M .J.C. (1995) Climatic changes in Europe during the last glacial - interglacial transition. 
QuaternaJ)' International 28, 63-76. 
Watts, W.A., Alien, J.R.M. and Huntley, B. (1996a) Vegetation history and palaeocIimate of the 
last glacial period at Lago Grande di Monticchio, southern Italy. QuaternaJY Science Reviews 
15, 133-153. 
Watts, W.A. , AlIen,1.R.M., Huntley, B. and Fritz, S.C. (I 996b) Vegetation history and climate 
of the last 15,000 years at Laghi di Monticchio, southern Italy. Quatel71aJY Science Reviews 
15, 113-132. 
Westerlund, c.A. and Blanc, H. (1879) Aperpt sur la Fauna Malacologique de la Grece inclus l'Epire 
et la Thessalie. Coquilles Extramarines. Naples. 
Westgate, 1. (1988) Isothermal plateau fission-track age of the late Pleistocene Old Crow tephra, 
Alaska. Geophysical Research Letters 15, 376-379. 
Wetzel, R.G. (1975) Limnology. W.B. Saunders and Co., Philadelphia. 
Whatley, R. (1983) The application of Ostracoda to palaeoenvironmental analysis. In: Maddocks, 
R.F. (ed.) Applications of Ostracoda. Department of Geosciences, University of Houston, 
Houston. 51-77 . 
Whittle, A. (1985) Neolithic Europe: A Survey. Cambridge University Press, Cambridge. 
Wigley, T.M.L. and Farmer, G. (1982) Climate of the eastern Meditenanean and Near East. In: 
Bintliff, 1.L. and van Zeist, W. (eds.) Palaeoclil11ates, Palaeoellvironments and the Human 
Communities in the Eastern Mediterranean Region in Later PrehistOl)'. British Archaeological 
Reports International Series 133, 3-37. 
REFERENCES • 230 
Wijmstra, T.A. (1969) Palynology of the first 30 metres of a 120m deep section in northern Greece. 
Acta Botanica Neerlandica 18 (4), 511-527. 
Wijmstra, T.A. and Smit, A. (1976) Palynology of the middle part (30-78 metres) of the 120m deep 
section in northern Greece (Macedonia). Acta Botanica Neerlandica 25 (4), 297-312. 
Wijmstra, T.A. and Groenhart, M.C. (J 983) Record of 700,000 years vegetational history in eastern 
Macedonia (Greece). Revista de la Academia Colol1lbiana de Ciencias Exactas, Ffsicas y 
Naturales 15 (58), 87-98. 
Williams, T., Thouveny, N. and Creer, K.M. (1996) Palaeoclimatic significance of the 300ka 
mineral magnetic record from the sediments of Lac du Bouchet, France. Quatemmy Science 
Reviews 15, 223-235. 
Willis, KJ. (1994) The vegetational history of the Balkans. Quaternmy Science Reviews 13 769-
788 . 
Wilson, D.S. and Hey, RN. (1981) The Galapagos axial magnetic anomaly: evidence for the 
Emperor event within the Brunhes and for a two-layer magnetic source. Geophysical Research 
Letters 8, 1,051-1,054. 
Wiltshire, P.E.J., Edwards, K.J. and Bond, S. (1994) Microbially-derived metallic sulphide 
spherules, pollen and the waterlogging of archaeological sites. AASP Contributions Series 
29, 207-221. 
Woillard, G.M. (1978) Grande Pile peat bog: a continuous pollen record for the last 140,000 years. 
Quaternmy Research 9, 1-21. 
Woillard, G.M. and Mook, W.G. (1982) Carbon-14 dates at Grande Pile: correlation of land and sea 
chronologies. Science 215 (8), 159-161. 
Wollin, G., Ericson, D.B., Ryan , W.B.F. and Foster, 1.H. (1971) Magnetism of the Earth and 
climatic changes. Earth and Planetary Science Letters 12, 175-183. 
Wright, H.E., Jr. (1984) Sensitivity and response time of natural systems to climatic change in the 
Late Quaternary . Quaternmy Science Reviews 3, 91-131. 
Xia,1., Engstrom, D.R and Ito, E. (1997a) Geochemistry of ostracode calcite: Part 2. The effects of 
water chemistry and seasonal temperature variation on Candona rawsoni. Geochimica et 
Cosl1lochimica Acta 61 (2), 383-391. 
Xia, J. , Ito , E. and Engstrom, D.R (1993) Oxygen isotope composition of lacustrine ostracodes as 
a paleoclimate indicator. EOS (Transactions of the American Geophysical Union) 74, 365. 
Xia, J., Ito, E. and Engstrom, D.R (1997b) Geochemistry of ostracode calcite: Part 1. An 
experimental determination of oxygen isotope fractionation. Geochimica et Cosmochimica 
Acta 61 (2), 377-382. 
Yu, L. and Oldfield, F. (1989) A multivariate mixing model for identifying sediment source from 
magnetic measurements. Quatemary Research 32, 168-181. 
Zangger, A. and Malz, H. (1989) Late Pleistocene, Holocene and Recent ostracods from the Gulf of 
Argos, Corinth. Courier Forschungsinstituts Sellckellberg 113, 159-175. 
Zagwijn, W.H. (1985) An outline of the Quaternary stratigraphy of the Netherlands. Geologie en 
Mijnbouw 64 (1), 17-24. 
Zhisheng, A. and Porter, S.c. (1997) Millennial-scale climatic oscillations during the last 
interglaciation in central China. Geology 25 (7), 603-606. 
Zhu, RX., Zhou, L. -P., Laj, c., Mazaud, A. and Ding, Z.L. (1994) The Blake geomagnetic 
polarity episode recorded in the Chinese loess. Geophysical Research Letters 21 (8), 697-
700. 
'1: 
I 
I 
I 
PLATES 
PLATE I 
A Black octohedral iron oxide grains (probably magnetite) from 180.20m, 
responsible for high levels of magnetic susceptibility at this level. 
B 
c 
D 
Pinus pollen grain preserved by surface coating of silica. 
Typical aggregated sediment grain, from 89.02m depth. 
Close-up of the aggregated grain illustrated in (C), to show mica-like 
hexagonal structure of carbonate particles, indicative of rapid precipitation in 
low-Mg waters. 
E Framboidal iron sulphide nodule (probably pyrite). 
I 
I 
I 
I 
I 

A 
B 
c 
D, H, I 
E, F, J 
PLATE 11 
Adult specimen of Viviparus janinensis. 
Juvenile specimen of Viviparus janinensis. 
Operculum of adult Viviparus janinensis. 
Views of adult Planorbarius comeus. 
Views of Dreissena (Carinodreissena) cf. stankovici. Carina clearly 
visible on fig. E. 
11 
, I 
c 
1 cm 
B 
F 
E 
H 
1 cm 
1 cm 
PLATE 11 
PLATE III 
A-C Juvenile specimens of Bithynia graeca. 
D Damaged adult specimen of Bithynia graeca. 
E Operculum from adult Bithynia graeca. 
F-G Detail of reflected lip from adult Bithynia graeca. 
H-L Views of adult and juvenile specimens of Gyraulus janinensis. 
M Detail of spiral sculpture from Gyraulus janinensis. 
.j 
-------

A-E 
F 
G-H 
PLATE IV 
External, internal and dorsal views of adult and juvenile specimens 
of Cyprideis torosa. 
External view of adult specimen of /lyocypris bradyi. 
External and internal views of /lyocypris slavonica. 

A-B 
C-D 
E-F 
G-H 
PLATE V 
External and internal views of adult female specimen of Candona 
permanenta. 
External and internal VIews of adult male specImen of Candona 
permanenta. 
External and internal VIews of adult specImens of Candona cf. 
dedelica. 
External and internal VIews of adult specImens of Candona cf. 
parvula. 

A-C 
D-E 
F-G 
PLATE VI 
External, internal and dorsal views of adult specImens of 
Leptocythere cf. ostrovskensis. 
External and internal views of specimens of Heterocypris 
rotundatus. 
External and internal views of adult specimens of Prionocypris sp. 

PLATE VII 
A-B External and dorsal views of adult specimens of Cypridopsis vidua. 
c External view of adult specimen of Paralimnocythere cf. compressa. 
D External view of adult specimen of Eucypris virens. 
E External view of adult specimen of Cypria ophtalmica. 
F External view of adult specimen of Darwinula stevensoni. 

APPENDICES 
APPENDIX A 
SAMPLE SITE DETAILS 
Unless otherwise stated, the weather at sampling points was clear and hot both prior to 
and at the time of sampling. Turbidity was measured using a l00mrn white disc, 
noting the depth at which it could no longer be seen. Sample site locations are 
illustrated in fig. 3.1. 
SITE 1 Limnopoula, adjacent to KEKOP centre 
Sampled: 13/9/94 Time: 08.10 
Shallow, rocky shoreline 
Some sparse aquatic broadleaf vegetation. Algal covering on water surface. Rubbish 
and debris from human activity. 
Water depth at sample point: 
Turbidity: 
Water temperature: 
pH: 
Dissolved oxygen: 
Alkalinity: 
Conductivity : 
0.23m 
0.30m 
12.2°C 
7.53 
6.1mg/l 
120mg/l 
332JlS 
SITE 2 NE shore of lake in 'backwater' area 
Sampled: 13/9/94 Time: 18.45 
Grassy banked steeply dipping shoreline adjacent to substantial reed-beds. Used by 
cattle, goats, sheep etc. for drinking and grazing. 
Tenestrial vegetation dominantly grass, a few shrubs. Aquatic vegetation included 
floating broadleaf, some reeds, algal covering. 
Water depth at sample point: 
Turbidity: 
Water temperature: 
pH: 
Dissolved oxygen: 
Alkalinity: 
Conductivity: 
0.28m 
0.18m 
28.3 °C 
7.48 
6.3mg/l 
120mg/l 
359JlS 
APPENDIX A 
SITE 3 NW of Site 2 along lake road, approx 500m 
Sampled: 15/9/94 Time: 09.15 
Marshy, muddy reed-beds. Used by cattle, sheep goats, etc. as drinking place. 
Terrestrial vegetation limited to grass. Aquatic vegetation dominantly reeds and 
grasses, substantial algal slime. 
Water depth at sample point: 
Turbidity: 
Water temperature: 
pH: 
Dissolved oxygen: 
Alkalinity: 
Conductivity: 
0.18m 
0.22m 
21.TC 
7.77 
0.8mg/l 
224mg/1 
44511S 
SITE 4 NNW of Katsikas at lake edge 
Sampled: 15/9/94 Time: 10.55 
Low-lying grass area, surrounded by reed-beds. 
Terrestrial vegetation limited to grass. Aquatic vegetation dominantly floating broadleaf 
(lily), amidst some grass and reeds. 
Water depth at sample point: 
Turbidity: 
Water temperature: 
pH: 
Dissolved oxygen: 
Alkalinity: 
Conductivity : 
0.15m 
0.13m 
19.TC 
7.42 
9.5mg/l 
212mg/l 
33411S 
SITE 5 NW of Katsikas, sandy shore of channel to pumping station 
Sampled: 15/9/94 Time: 18.45 
Sandy shoreline, fairly steeply dipping. 
Ten'estrial vegetation entirely short grass. Minimal aquatic vegetation, a little floating 
algae (likely channel cleared regularly). Used by cattle etc. Much human rubbish and 
debris nearby. 
Water depth at sample point: 
Turbidity: 
Water temperature: 
pH: 
Dissolved oxygen: 
Alkalinity: 
Conductivity: 
0.22m 
O.13m 
20.9°C 
7.53 
11.1mg/l 
149mg/1 
32311S 
• I 
'. 
APPENDIX A 
SITE 6 Between Arnfithea and Perama on grassy 'breakwater' road, close to inlet 
stream 
Sampled: 16/9/94 Time: 17.55 
Steeply dipping shore from artificial roadway, grassy and reed-covered. 
Terrestrial vegetation dominantly grass. 
vegetation. 
Water depth at sample point: 0.18m 
0.22m 
13.3°C 
7.64 
1.lmg/l 
217mg/l 
679~S 
Turbidity: 
Water temperature: 
pH: 
Dissolved oxygen: 
Alkalinity: 
Conductivity: 
SITE 7 Nissi Island, southern side 
Sampled: 17/9/94 Time: 09.35 
Reeds and grasses dominate aquatic 
Shallow, rocky shoreline encroaching onto muddy flats fringed by extensive reed-beds. 
Much aquatic broadleaf vegetation. Algal covering on water surface. Rubbish and 
debris from human activity. Used by sheep etc. 
Water depth at sample point: 
Turbidity: 
Water temperature: 
pH: 
Dissolved oxygen: 
Alkalinity: 
Conducti vity: 
SITE 8 Nissi Island, SE tip 
0.18m 
0.24m 
15.5°C 
7.83 
1.5mg/l 
295mg/l 
603~S 
Sampled: 17/9/94 Time: 11.35 
Shallow, rocky shoreline 
Little terrestrial or aquatic vegetation. Thick green algal covering on water surface. 
Rubbish and debris from human activity. 
Water depth at sample point: 
Turbidity: 
Water temperature: 
pH: 
Dissolved oxygen: 
Alkalinity: 
Conductivity: 
0.24m 
0.03m 
21.5°C 
7.53 
6.4mg/l 
127mg/l 
332~S 
SITE 9 North of Island, on disused car-ferry platform 
Sampled: 17/9/94 Time: 19.25 
Metal platform in channel cut through reed-beds. 
Some floating broadleaf; reeds and grasses adjacent. 
Water depth at sample point: 0.15m 
0.28m 
24.1°C 
7.72 
9.3mg/1 
135mg/1 
3341lS 
Turbidity: 
Water temperature: 
pH: 
Dissolved oxygen: 
Alkalinity: 
Conductivity: 
SITE 10 Shoreline SE of Ioannina town. 
Sampled: 18/9/94 Time: 10.50 
Shallow, flat-lying mud / grassy area. 
APPENDIX A 
Abundant floating broadleaf (lily). Thick algal covering on water surface. Rubbish 
and debris from human activity. 
Water depth at sample point: 0.24m 
0.08m 
13.0°C 
8.09 
12.9mg/1 
151mg/1 
340llS 
Turbidity: 
Water temperature: 
pH: 
Dissolved oxygen: 
Alkalinity: 
Conductivity: 
SITE 11 Eastern extent of lake, channel cut through reeds to pumping station. 
Sampled: 19/9/94 Time: 10.25 
Steep muddy shore. 
Some sparse floating broadleaf. Algal covering to water surface. 
Water depth at sample point: 0.17m 
0.15m 
23.1 °C 
7.54 
1O.2mg/l 
172mg/1 
3371lS 
Turbidity: 
Water temperature: 
pH: 
Dissolved oxygen: 
Alkalinity: 
Conducti vity: 
.' ," -' , . 
Parameter SITE SITE SITE SITE SITE SITE SITE SITE SITE SITE SITE 
(mg/l) 1 2 3 4 5 6 7 8 9 10 11 
Zn 0.008 0.011 0.003 b.d. 0.001 b.d. 0.001 0.005 b.d. b.d. 0.001 
Pb b.d. 0.054 b.d. 0.035 b.d. 0.001 0.003 0.012 0.011 0.038 0.047 
Cd 0.006 0.007 0 b.d. b.d. b.d. b.d. 0.002 0.001 0.003 0.004 
Si 4.872 7.36 8.895 5.482 5.893 4.554 10.42 5.296 5.874 5.123 5.672 
Mn 0.011 0.096 0.092 0.028 0.026 0.015 0.097 0.015 0.069 0.032 0.015 
Fe b.d. 0.028 0.613 b.d. b.d. 0.012 1.421 b.d. 0.063 b.d. b.d. 
Mg 10.37 10.38 10.39 9.989 10.06 5.387 14.21 10.11 10.3 10.01 10.14 
Al 0.029 0.037 0.002 0.026 0.024 0.026 0.041 0.032 0.015 0.04 0.024 
V 0.009 0.01 0.003 0.004 0.007 0.014 0.008 0.016 0.014 O.oI1 0.017 
Ca 38.76 44.02 52.78 40 39.81 70.37 90.7 40.2 41.23 39 39.57 
Cu 0.002 0.007 0.009 0.008 0.007 0.005 0.008 0.004 0.009 0.003 0.009 
Sr 0.214 0.226 0.245 0.212 0.216 0.324 0.4 16 0.219 0.219 0.212 0.218 
Na 19.77 20.01 20.72 19.74 19.67 57.93 25.26 19.08 19.33 18.94 19.44 
K 3.492 3.431 3.072 3.366 3.505 0.602 1.638 3.461 3.444 3.479 3.279 
Nitrate 0.8 0 .9 1.6 1.4 1.7 0.6 -0.3 1.5 1.3 1.8 1.9 
Sulphate 0.76 1.02 1.27 1.27 0.95 28.86 7.44 2.03 0.95 1.59 2.61 
Silica 10.4 15.8 19.0 12.5 12.6 9.7 22.3 11.3 12.57 11.0 12.1 
Phosphate 8.1 0.7 0.7 1.6 1.3 0.6 0.8 2 1.1 1.3 1.1 
Chloride 29 23 27 21 28 100 37 24 23 23 26 
Alkalinity 146.5 146.5 273.4 258 .7 181.8 264.8 360.0 155.0 164.8 184.3 209.9 
. Dissolved Oxygen 6.1 6.3 0.8 9.5 11.1 1.1 1.5 6.4 9.3 12.9 10.2 
a Tempe¥ure (QC) 15.6 15.8 16.9 15.8 15.8 15.6 17.7 15.6 18.5 15.6 15.5 
pH 7 .53 7.48 7.77 7.42 7.53 7.64 7.83 7.53 7.72 8.09 7.54 
Conductivity V"S) 332 359 445 334 323 679 603 332 334 340 337 
Salinity (%0) 0.268 0.266 0.41 0.369 0.299 0.539 0.561 0.269 0.278 0.294 0.326 
Charge Balance Error (%) 2.94 11.26 7.55 11.96 1.05 8.43 1.46 5.16 5.09 0.94 6.05 
Appendix B. Modern water chemistry data. Salinity calculated from Total Dissolved Solids (TDS). b.d. = below detection. Mg/Ca ratio = 0.34. 
lOAN 284 A\' 
' 0" 0<:' 
Depth 
m 
84 
85 
86 
87 
88 
89 
90 
91 
92 
93 
94 
95 
96 
97 
98 
99 
100 
~"" .~0 " )"~' <?~" 
,::," 
'0,,0 
?:Jbv~ 
.j'l> If"'" 
«/0,,0 
0<:- .,~"" 
,'li (j'l>~ ~"" 
'li'Ci, 1 ~v ~ « ~ 
,," \'1> 0'1> ",,'I> ,," 
eI" 'I> ,'I>G .~4,~ "". ~v cA'I> ',<:' 
OV 0,0 <?,,, «'<:'! 0'(:' {.'$>'I> Cp<!i o"~ cl '0" ~' ~ ·l 
L..-J ~ W I ,I I, I , I , I , L L.... w w L~ L.... w w L..-L I , I , I , I ,I ~_ I , I ' I ' I , I , I , I , I , I , I '.-.J 
o 0 10 20 0 0 10 20 30 40 50 0 0 0 0 0 0 0 0 10 0 10 20 30 40 0 10 20 0 10 20 30 40 50 60 70 80 90 
% 
Appendix C Summary pollen diagram of the Eemian interval of corel284 (p.C. Tzedakis, unpubl. data). 
84 
85 
86 
87 
88 
89 
90 
91 
92 
93 
94 
95 
96 
97 
98 
99 
100