The Nile Basin: Quaternary Geology, Geomorphology and Prehistoric Environments 110717919X, 9781107179196

Table of contents :
Contents
Preface
Acknowledgements
1 The Nile Basin: An Introduction
1.1 Introduction
1.2 Early Speculation about the Nile
1.3 Unique Attributes of the Nile
1.4 Aims and Structure of This Volume
2 Evolution of the Nile Basin
2.1 Introduction: How Old Is the Nile?
2.2 Ethiopian Uplift and Volcanism
2.3 Erosion of the Ethiopian Nile Headwaters
2.4 Tectonic History of Lake Victoria and the Ugandan Nile Headwaters
2.5 Tectonic and Structural Control of the Nile and Its Tributaries
2.6 Volume of the Nile Cone
2.7 Conclusion
3 Climate and Hydrology
3.1 Introduction
3.2 Climates of the Nile Basin
3.3 Nile Hydrology and Nile Floods
3.4 Conclusion
4 Geology and Soils
4.1 Introduction
4.2 Geology
4.3 Soils
4.4 Conclusion
5 Vegetation, Land Use and Human Impact
5.1 Introduction
5.2 Natural Vegetation Zones
5.3 Current Land Use
5.4 Human Impact on the Natural Vegetation and Soils
5.5 Controlling the Floods: Dams, Reservoirs and Disease
5.6 Conclusion
6 The Ethiopian Highlands
6.1 Introduction
6.2 Cenozoic Uplift and Volcanism
6.3 Cenozoic Erosion: The Blue Nile and Tekezze Gorges
6.4 Miocene and Pliocene Environments in Ethiopia
6.5 Quaternary Environments
6.6 The Late Pleistocene Blue Nile
6.7 The Early Holocene Blue Nile
6.8 Conclusion
7 The Ugandan Lake Plateau
7.1 Introduction
7.2 Cenozoic Disruption of Drainage
7.3 Origin of the Ugandan Lakes
7.4 Late Quaternary Fluctuations of Lakes Victoria and Albert
7.5 Late Quaternary Fluctuations of Lake Challa
7.6 The ‘African Humid Period’
7.7 Kilimanjaro Holocene Ice Core Records
7.8 Conclusion
8 The Sudd Swamps and the White Nile
8.1 Introduction
8.2 The Sudd
8.3 The White Nile
8.4 White Nile Islands
8.5 Prehistoric Occupation of the White Nile Valley
8.6 Conclusion
9 Lake Turkana and Overflow into the Sobat
9.1 Introduction
9.2 Lake Turkana
9.3 Quaternary Sediments in the Lower Omo Valley
9.4 Overflow of Lake Turkana into the White Nile
9.5 Conclusion
10 The Khor Abu Habl Fan and the Desert Dunes of Kordofan and Darfur
10.1 Introduction
10.2 The Umm Ruwaba Formation and the Khor Abu Habl Fan
10.3 Desert Dunes and Their Environmental Significance
10.4 The Desert Dunes of Kordofan and Darfur
10.5 Freshwater Mollusca and Holocene Lakes
10.6 Conclusion
11 The Gezira Alluvial Fan and Blue Nile Palaeochannels
11.1 Introduction
11.2 Age and Origin of the Gezira
11.3 Blue Nile Palaeochannels
11.4 Source-Bordering Dunes
11.5 Prehistoric Occupation Sites
11.6 Conclusion
12 The Atbara
12.1 Introduction
12.2 Cold Climate Landforms and Glaciation in the Semien Highlands
12.3 Denudation Rates in the Tekezze Basin
12.4 Quaternary Alluvial Formations in the Atbara Valley
12.5 Holocene Environments
12.6 Quaternary Fossils and Prehistoric Artefacts
12.7 Conclusion
13 Jebel Marra Volcano
13.1 Introduction
13.2 Geological History of Jebel Marra
13.3 Flora of Jebel Marra and Its Significance
13.4 Piedmont Sediments
13.5 Deriba Crater Lakes and Late Pleistocene High Lake Levels
13.6 Pleistocene and Holocene Erosion and Sedimentation
13.7 Conclusion
14 The Desert Nile
14.1 Introduction
14.2 Deciphering Nile Alluvial History
14.3 Pleistocene Erosion and Sedimentation in Southern Egypt
14.4 Late Quaternary Depositional Environments in Northern Sudan
14.5 Meta-analysis of the Desert Nile Holocene Fluvial Archive
14.6 Conclusion
15 West of the Nile: The Western Desert of Egypt and the Eastern Sahara – Part 1
15.1 Introduction
15.2 Early Exploration
15.3 Wadi Howar and Adjacent Areas
15.4 The Darb el Arba’in Desert: Oyo, El Atrun and Selima Oasis
15.5 Conclusion
16 West of the Nile: The Western Desert of Egypt and the Eastern Sahara – Part 2
16.1 Introduction
16.2 Dakhla and Kharga Oases
16.3 The Gilf Kebir, Jebel ‘Uweinat, Jebel Arkenu and Environs
16.4 Bir Sahara, Bir Tarfawi and the Tushka Lakes
16.5 Saharan Groundwater Recharge during the Quaternary
16.6 Late Quaternary Environments in the Sahara: Implications and Cautions
16.7 Conclusion
17 The Fayum
17.1 Introduction
17.2 Origin of the Fayum Depression
17.3 Holocene Lake Fluctuations in the Fayum
17.4 Epi-Palaeolithic/Mesolithic and Neolithic Settlement in the Fayum
17.5 Conclusion
18 The Red Sea Hills
18.1 Introduction
18.2 Origin and Evolution of the Red Sea Hills
18.3 Pleistocene Rivers Flowing from the Red Sea Hills
18.4 Pleistocene and Holocene Spring Tufas and Their Climatic Significance
18.5 Mesolithic and Neolithic Occupation in the Red Sea Hills
18.6 A Wetter Climate in the Red Sea Hills 2,000 Years Ago
18.7 Conclusion
19 The Sinai Peninsula
19.1 Introduction
19.2 Origin and Evolution of the Sinai Peninsula
19.3 Periglacial Landforms in the Sinai Mountains
19.4 Tufa Deposits in the Sinai Peninsula and Their Climatic Significance
19.5 Late Pleistocene Valley-Fills of the Sinai Peninsula
19.6 Desert Dunes of the Sinai Peninsula and Adjacent Northern Negev Desert
19.7 Prehistoric Occupation in the Sinai Peninsula
19.8 Conclusion
20 The Nile Delta
20.1 Introduction
20.2 Origin and Evolution of the Nile Delta
20.3 Holocene History of Maryut Lagoon, Western Nile Delta
20.4 Variations in Nile Delta Sediment Provenance
20.5 Holocene Fluctuations in Nile Delta Sedimentation
20.6 Holocene Variations in Nile Delta Subsidence
20.7 Human Occupation of the Nile Delta
20.8 Conclusion
21 The Nile Cone
21.1 Introduction
21.2 Age and Volume of the Nile Cone
21.3 Analysis of Marine Sediment Cores from the Nile Cone
21.4 Nile Floods and Sapropel Formation
21.5 Conclusion
22 Origins of Plant and Animal Domestication in the Nile Basin
22.1 Introduction
22.2 Some General Considerations
22.3 The Transition from Mesolithic to Neolithic in the Fayum and Main Nile Valley
22.4 The Transition from Mesolithic to Neolithic in the Eastern Sahara
22.5 The Transition from Mesolithic to Neolithic in Central and Eastern Sudan
22.6 The Transition from Mesolithic to Neolithic in Ethiopia and East Africa
22.8 Conclusion
23 Epilogue: ‘Out of Africa’
23.1 Introduction
23.2 Quaternary Environments in North and East Africa
23.3 Quaternary Environments in Eurasia
23.4 Movement of Homo erectus/Homo ergaster Out of Africa
23.5 Movement of Homo sapiens Out of Africa
23.6 Conclusion
References
Index

Citation preview

TH E NIL E B AS IN

The Nile Basin contains a record of human activities spanning the last million years. However, the interactions between prehistoric humans and environmental changes in this area are complex and often poorly understood. This comprehensive book explains in clear, non-technical terms how prehistoric environments can be reconstructed, with examples drawn from every part of the Nile Basin. Adopting a source-to-sink approach, the book integrates events in the Nile headwaters with the record from marine sediment cores in the Nile Delta and offshore. It provides a detailed record of past environmental changes throughout the Nile Basin and concludes with a review of the causes and consequences of plant and animal domestication in this region and of the various prehistoric migrations out of Africa into Eurasia and beyond. A comprehensive overview, this book is ideal for researchers in geomorphology, climatology and archaeology. m a rt i n w i l l i am s is Adjunct Professor in Earth Sciences at the University of Adelaide, Australia. He has worked with archaeologists in the Sahara, Nile Valley and Ethiopia, and has written more than 200 research papers and a dozen books, including Climatic Change in Deserts (2014), A Land Between Two Niles (with Donald Adamson, 1982) and The Sahara and the Nile (with Hugues Faure, 1980). He received the Farouk El-Baz Award for Desert Research from the Geological Society of America in 2008.

THE NILE BASIN Quaternary Geology, Geomorphology and Prehistoric Environments MARTIN WILLIAMS The University of Adelaide

University Printing House, Cambridge CB2 8BS, United Kingdom One Liberty Plaza, 20th Floor, New York, NY 10006, USA 477 Williamstown Road, Port Melbourne, VIC 3207, Australia 314–321, 3rd Floor, Plot 3, Splendor Forum, Jasola District Centre, New Delhi – 110025, India 79 Anson Road, #06–04/06, Singapore 079906 Cambridge University Press is part of the University of Cambridge. It furthers the University’s mission by disseminating knowledge in the pursuit of education, learning, and research at the highest international levels of excellence. www.cambridge.org Information on this title: www.cambridge.org/9781107179196 DOI: 10.1017/9781316831885 © Martin Williams 2019 This publication is in copyright. Subject to statutory exception and to the provisions of relevant collective licensing agreements, no reproduction of any part may take place without the written permission of Cambridge University Press. First published 2019 Printed in the United Kingdom by TJ International Ltd. Padstow Cornwall A catalogue record for this publication is available from the British Library. Library of Congress Cataloging-in-Publication Data Names: Williams, M. A. J., author. Title: The Nile Basin : quaternary geology, geomorphology and prehistoric environments / Martin Williams. Description: Cambridge, United Kingdom ; New York, NY : Cambridge University Press, 2019. | Includes bibliographical references and index. Identifiers: LCCN 2018027622 | ISBN 9781107179196 (Hardback) Subjects: LCSH: Geology – Nile River Watershed. | Geomorphology – Nile River Watershed. | Nile River Watershed – Antiquities. | Geology, Stratigraphic – Quaternary. Classification: LCC QE328 .W55 2019 | DDC 556.2–dc23 LC record available at https://lccn.loc.gov/2018027622 ISBN 978-1-107-17919-6 Hardback Cambridge University Press has no responsibility for the persistence or accuracy of URLs for external or third-party internet websites referred to in this publication and does not guarantee that any content on such websites is, or will remain, accurate or appropriate.

For Frances, who shares my love of wild, remote and rugged places

Contents

Preface Acknowledgements

page xiii xv

1 The Nile Basin: An Introduction 1.1 Introduction 1.2 Early Speculation about the Nile 1.3 Unique Attributes of the Nile 1.4 Aims and Structure of This Volume 2 Evolution of the Nile Basin 2.1 Introduction: How Old Is the Nile? 2.2 Ethiopian Uplift and Volcanism 2.3 Erosion of the Ethiopian Nile Headwaters 2.4 Tectonic History of Lake Victoria and the Ugandan Nile Headwaters 2.5 Tectonic and Structural Control of the Nile and Its Tributaries 2.6 Volume of the Nile Cone 2.7 Conclusion 3 Climate and Hydrology 3.1 Introduction 3.2 Climates of the Nile Basin 3.3 Nile Hydrology and Nile Floods 3.4 Conclusion 4 Geology and Soils 4.1 Introduction 4.2 Geology 4.3 Soils 4.4 Conclusion 5 Vegetation, Land Use and Human Impact 5.1 Introduction 5.2 Natural Vegetation Zones 5.3 Current Land Use

1 1 3 5 5 8 8 9 11 14 15 16 17 19 19 19 26 32 33 33 35 47 58 59 59 60 66

vii

viii

Contents

6

7

8

9

10

5.4 Human Impact on the Natural Vegetation and Soils 5.5 Controlling the Floods: Dams, Reservoirs and Disease 5.6 Conclusion The Ethiopian Highlands 6.1 Introduction 6.2 Cenozoic Uplift and Volcanism 6.3 Cenozoic Erosion: The Blue Nile and Tekezze Gorges 6.4 Miocene and Pliocene Environments in Ethiopia 6.5 Quaternary Environments 6.6 The Late Pleistocene Blue Nile 6.7 The Early Holocene Blue Nile 6.8 Conclusion The Ugandan Lake Plateau 7.1 Introduction 7.2 Cenozoic Disruption of Drainage 7.3 Origin of the Ugandan Lakes 7.4 Late Quaternary Fluctuations of Lakes Victoria and Albert 7.5 Late Quaternary Fluctuations of Lake Challa 7.6 The ‘African Humid Period’ 7.7 Kilimanjaro Holocene Ice Core Records 7.8 Conclusion The Sudd Swamps and the White Nile 8.1 Introduction 8.2 The Sudd 8.3 The White Nile 8.4 White Nile Islands 8.5 Prehistoric Occupation of the White Nile Valley 8.6 Conclusion Lake Turkana and Overflow into the Sobat 9.1 Introduction 9.2 Lake Turkana 9.3 Quaternary Sediments in the Lower Omo Valley 9.4 Overflow of Lake Turkana into the White Nile 9.5 Conclusion The Khor Abu Habl Fan and the Desert Dunes of Kordofan and Darfur 10.1 Introduction 10.2 The Umm Ruwaba Formation and the Khor Abu Habl Fan 10.3 Desert Dunes and Their Environmental Significance 10.4 The Desert Dunes of Kordofan and Darfur 10.5 Freshwater Mollusca and Holocene Lakes 10.6 Conclusion

70 77 79 81 81 81 83 84 85 93 93 96 97 97 97 100 101 104 105 106 106 107 107 108 111 118 119 124 127 127 129 129 130 131 132 132 132 136 139 142 142

Contents

11 The Gezira Alluvial Fan and Blue Nile Palaeochannels 11.1 Introduction 11.2 Age and Origin of the Gezira 11.3 Blue Nile Palaeochannels 11.4 Source-Bordering Dunes 11.5 Prehistoric Occupation Sites 11.6 Conclusion 12 The Atbara 12.1 Introduction 12.2 Cold Climate Landforms and Glaciation in the Semien Highlands 12.3 Denudation Rates in the Tekezze Basin 12.4 Quaternary Alluvial Formations in the Atbara Valley 12.5 Holocene Environments 12.6 Quaternary Fossils and Prehistoric Artefacts 12.7 Conclusion 13 Jebel Marra Volcano 13.1 Introduction 13.2 Geological History of Jebel Marra 13.3 Flora of Jebel Marra and Its Significance 13.4 Piedmont Sediments 13.5 Deriba Crater Lakes and Late Pleistocene High Lake Levels 13.6 Pleistocene and Holocene Erosion and Sedimentation 13.7 Conclusion 14 The Desert Nile 14.1 Introduction 14.2 Deciphering Nile Alluvial History 14.3 Pleistocene Erosion and Sedimentation in Southern Egypt 14.4 Late Quaternary Depositional Environments in Northern Sudan 14.5 Meta-analysis of the Desert Nile Holocene Fluvial Archive 14.6 Conclusion 15 West of the Nile: The Western Desert of Egypt and the Eastern Sahara – Part 1 15.1 Introduction 15.2 Early Exploration 15.3 Wadi Howar and Adjacent Areas 15.4 The Darb el Arba’in Desert: Oyo, El Atrun and Selima Oasis 15.5 Conclusion

ix

143 143 145 148 158 160 163 164 164 166 171 172 173 174 174 176 176 178 180 181 187 192 194 196 196 198 200 202 210 210 211 211 212 213 222 226

x

Contents

16 West of the Nile: The Western Desert of Egypt and the Eastern Sahara – Part 2 16.1 Introduction 16.2 Dakhla and Kharga Oases 16.3 The Gilf Kebir, Jebel ‘Uweinat, Jebel Arkenu and Environs 16.4 Bir Sahara, Bir Tarfawi and the Tushka Lakes 16.5 Saharan Groundwater Recharge during the Quaternary 16.6 Late Quaternary Environments in the Sahara: Implications and Cautions 16.7 Conclusion 17 The Fayum 17.1 Introduction 17.2 Origin of the Fayum Depression 17.3 Holocene Lake Fluctuations in the Fayum 17.4 Epi-Palaeolithic/Mesolithic and Neolithic Settlement in the Fayum 17.5 Conclusion 18 The Red Sea Hills 18.1 Introduction 18.2 Origin and Evolution of the Red Sea Hills 18.3 Pleistocene Rivers Flowing from the Red Sea Hills 18.4 Pleistocene and Holocene Spring Tufas and Their Climatic Significance 18.5 Mesolithic and Neolithic Occupation in the Red Sea Hills 18.6 A Wetter Climate in the Red Sea Hills 2,000 Years Ago 18.7 Conclusion 19 The Sinai Peninsula 19.1 Introduction 19.2 Origin and Evolution of the Sinai Peninsula 19.3 Periglacial Landforms in the Sinai Mountains 19.4 Tufa Deposits in the Sinai Peninsula and Their Climatic Significance 19.5 Late Pleistocene Valley-Fills of the Sinai Peninsula 19.6 Desert Dunes of the Sinai Peninsula and Adjacent Northern Negev Desert 19.7 Prehistoric Occupation in the Sinai Peninsula 19.8 Conclusion 20 The Nile Delta 20.1 Introduction 20.2 Origin and Evolution of the Nile Delta 20.3 Holocene History of Maryut Lagoon, Western Nile Delta

227 227 228 229 239 242 244 246 248 248 250 251 252 255 257 257 257 259 260 261 261 265 267 267 267 269 270 270 274 276 277 278 278 278 281

Contents

20.4 Variations in Nile Delta Sediment Provenance 20.5 Holocene Fluctuations in Nile Delta Sedimentation 20.6 Holocene Variations in Nile Delta Subsidence 20.7 Human Occupation of the Nile Delta 20.8 Conclusion 21 The Nile Cone 21.1 Introduction 21.2 Age and Volume of the Nile Cone 21.3 Analysis of Marine Sediment Cores from the Nile Cone 21.4 Nile Floods and Sapropel Formation 21.5 Conclusion 22 Origins of Plant and Animal Domestication in the Nile Basin 22.1 Introduction 22.2 Some General Considerations 22.3 The Transition from Mesolithic to Neolithic in the Fayum and Main Nile Valley 22.4 The Transition from Mesolithic to Neolithic in the Eastern Sahara 22.5 The Transition from Mesolithic to Neolithic in Central and Eastern Sudan 22.6 The Transition from Mesolithic to Neolithic in Ethiopia and East Africa 22.8 Conclusion 23 Epilogue: ‘Out of Africa’ 23.1 Introduction 23.2 Quaternary Environments in North and East Africa 23.3 Quaternary Environments in Eurasia 23.4 Movement of Homo erectus/Homo ergaster Out of Africa 23.5 Movement of Homo sapiens Out of Africa 23.6 Conclusion References Index

xi

284 286 287 288 290 291 291 291 293 294 300 301 301 303 306 309 311 317 320 322 322 323 324 326 328 332 334 394

Preface

The Nile Basin has been a reliable haven for prehistoric human groups for more than a million years. Early, Middle and Late Stone Age artefacts can be seen scattered throughout the Nile Basin, including in areas that are now waterless and inhospitable for all but the hardiest of present-day human communities. Another feature of the Nile Basin is the abundant evidence that the climate has been very much wetter than today on innumerable occasions in the past. All of this prompts us to ask what caused these dramatic changes in climate. The Nile Basin covers the northeast quadrant of Africa and contains a generous slice of the climatic history of the Earth. It falls under the influence of three major climate systems. In the far north the westerly winds that blow across the eastern Mediterranean in winter bring sporadic rain today to northern Egypt. At intervals in the recent past the influence of these winter rains extended much farther south, bringing precipitation to the Red Sea Hills in the east and, possibly, to the great sandstone plateau of the Gilf Kebir in the west. In the equatorial south of the Basin the seasonal migrations of the Intertropical Convergence Zone (ITCZ) bring summer rain to the centre of the Basin and to the valleys of the Blue and White Nile Rivers in central Sudan. Here again, there is strong evidence that the influence of the ITCZ once extended much farther north, well into Nubia and the now hyperarid eastern Sahara adjoining the Nile Valley. The most recent northward excursion of the ITCZ was during the Early to Middle Holocene, when groups of Mesolithic people made a living from fishing, hunting and gathering wild plant foods. By about 8,000–7,000 years ago we see the inception of plant and animal domestication in the Nile Valley, several thousand years after its adoption by Neolithic communities in the Fertile Crescent of the Levant and Anatolia. Why was the onset of the Neolithic so late in the Nile Basin compared to farther north? The African summer monsoon is the third main climate system. The summer floods in the Nile that so intrigued the great Greek traveller and historian Herodotus (484–425 BC) some 2,500 years ago depend on the African summer monsoon in the Ethiopian headwaters of the Nile. During the Early Holocene, the summer monsoon was considerably stronger than today so that the Blue Nile and Atbara Rivers contributed far more water and sediment to the main Nile than they do today. The result was widespread flooding in northern Sudan and

xiii

xiv

Preface

Egypt, and the progressive build-up of the Nile flood plain and Nile Delta. These alluvial sediments, when appropriately deciphered, can tell us a great deal not only about both the flood history of the Nile during the last 15,000 and more years, but also about the timing and significance of local contributions to the Nile sediment load from local wadi systems and, later, as the climate became drier, from wind-blown sand and dust. There is also good evidence that the regional climate has become progressively drier during the last several hundred thousand years. The eastern Sahara was studded with large lakes, including in the region between Bir Sahara and Bir Tarfawi in the presently arid desert west of Aswan at intervals during the Early and Middle Stone Age. The valley of the White Nile was filled with a vast lake during the Last Interglacial. This lake had attained an elevation of 386 m by 110,000 years ago and was more than 500 km long from north to south and up to 80 km wide. A second big lake that refilled the White Nile Valley with the abrupt return of the summer monsoon 14,500 years ago attained an elevation of only 382 m, and although still vast, was much smaller than the Last Interglacial mega-lake. Whether or not these lakes facilitated northward movement through the Nile Valley remains an open question. The aim of this book is to explore the issues mentioned here in appropriate detail and to seek answers to some of the questions raised. The approach adopted is geographical rather than chronological, allowing the evidence from each of the fifteen or so major physical regions in the Nile Basin to be assessed according to the local geological and geomorphic contexts. Our focus is on the reconstruction of past environmental changes in the Nile Basin, thereby allowing archaeologists to see their work in a more rounded context. We conclude with a review of the uniquely important contributions that marine sediment cores recovered from the submarine Nile Cone have provided to our understanding of Nile flood history and changing patterns of sediment sources and delivery. Some chapters in this book are quite long and detailed, others less so. By way of defence I can do no better than quote from the introduction to L. C. Beadle (1974), The Inland Waters of Tropical Africa: An Introduction to Tropical Limnology: ‘Though I have made some effort to maintain a balance, I cannot pretend to have avoided giving relatively more prominence to certain subjects and regional studies than a dispassionate reader might think they deserve. I can reply that it is better for a student to hear about subjects of which the author has a direct knowledge and in which he is especially interested. It is impossible to disguise the fact that I know more from direct experience about eastern than about western tropical Africa.’ In my own case, I know a great deal more at first-hand about the Nile and its tributaries in Ethiopia and the Sudan than I do about the Nile in Egypt and the White Nile headwaters in Uganda, although I have travelled and worked throughout much of the Nile Basin. I leave the last word to Geoffrey Blainey (1966) in his preface to The Tyranny of Distance: ‘I found I had ended up with a kind of history . . . not a comprehensive history, but then every history of every country is a mirror of the author’s own interests and therefore selective rather than comprehensive.’

Acknowledgements

Many friends and colleagues have helped over the years in enlarging my appreciation of the Nile Basin and its people. I was introduced to the Nile in October 1962, after joining Hunting Technical Services to work as a soil surveyor in the Sudan as part of the Roseires Dam Project. My job was to map soils in the lower Blue and White Nile Valleys and I was fortunate to work under the experienced eye of Colin Mitchell, who spoke fluent Arabic and had worked many years in the Sudan. During 1963–64, I led a nomadic life as a reconnaissance soil surveyor in the lower White Nile Valley. I retain ineffable memories of the unfailing courtesy and generosity of the people of the central Sudan, whether villagers or nomads, and the hospitality freely offered by Sayed Idris Habbani, Umda of Hashaba, by the late Sayed Omar Mustafa, Umda of Esh Shawal, and the late Sayed el Hadi Abdl Rahman el Fadi el Mahdi, who so generously made available his private rest-house on Aba Island on the White Nile during Ramadan in 1964. During leave from Sudan local guides helped me to explore the Nile Valley in Egypt, visiting Luxor and the Valley of the Kings before the Aswan High Dam was completed and certain monuments had to be moved. In late 1969 and early 1970 I was able to explore some of the Ugandan Lake Plateau and visit Lake Victoria. The late Professor Bill Bishop advised me on questions of Ugandan tectonic and sedimentary geology. During later visits to Sudan in the 1970s and 1980s, friends and colleagues old and new at the University of Khartoum shared their knowledge of and enthusiasm for the Nile. Dr Ekhlas Abd el Bari and Professor Mohamed Obeid Mubarak of the Botany Department; Professor John Vail, Professor Ismail el Boushi, Dr Yassin Abdel Salaam and Dr Salah el Raba’a of the Geology Department; and Dr Asim el Moghraby of the Hydrobiology Research Unit were at all times pillars of support, as was my great friend Dr Donald Adamson, who shared my passion for the Nile and proved an inspiring and indefatigable field companion during the 1970s and early 1980s. Dr El Sammani Abdalla Yacoub and Dr Ali Mohayad Bannaga of the National Council for Research helped to facilitate my research and brought me into contact with Dr Hassan Hag Abdalla of the Gezira Research Station at Wad Medani, and most helpfully with Sayed Abdel Latif Widatalla, Sayid Yusif Sulieman, Sayed Abdulla Hassan Ishag and Sayed el Tayeb M. Saeed of the Sudan Geological Survey. The Khartoum staff of Hunting Technical Services (HTS) and Sir Murdoch Macdonald and Partners (MMP) were invariably helpful, xv

xvi

Acknowledgements

for which I thank them warmly, as also Mr Vernon Robertson (HTS), Mr Ian Matthews (MMP) and Mr Martin Adams (HTS), all of whom encouraged my studies of the Nile. In December 1971 Dr Bill Morton and Dr Getaneh Assefa (Geology, Addis Ababa University) invited me to join a student field excursion to the Blue Nile gorge, after which I explored the glaciated mountains around Ras Dashan in the Semien Highlands of Ethiopia. In January 1972 Sayed Awad Medani showed me the shell beds at Wadi Mansurab west of the White Nile, while Tony and Pat Harris and Professor John Cloudsley Thompson introduced me to the rock carvings at Jebel Qeili in the Butana Desert. Professor J. Desmond Clark and his archaeological team from the University of California, Berkeley, joined us in early 1973. We were able to excavate and date several Mesolithic, Neolithic and Iron Age sites in the White Nile Valley and at Jebel Moya. Sayed Negm-ed-Din Sherif, Director of Antiquities, smoothed our way and visited our excavations. Desmond and I later worked on Early, Middle and Late Stone Age sites in Ethiopia and the Afar Desert. During January 1976 my future wife Frances Dakin and the late Dr Bill Morton, both from the Geology Department of the University of Addis Ababa, accompanied us in a rapid survey up the Blue Nile and an intensive survey of the piedmont deposits around Jebel Marra volcano directed by Dr David Parry. In December of that year Dr John Gowlett, archaeologist at the University of Khartoum, accompanied us on a survey along the west bank of the White Nile. I am most grateful to Professor Fred Wendorf (Southern Methodist University, Dallas, Texas) for inviting me to take part in the 1987 fieldwork at Bir Tarfawi and Bir Sahara in the Western Desert of Egypt, and to Dr Bahay Issawi, Dr Hani Hamroush and Ms Angelika Hamroush for help, hospitality and stimulating discussion while in Cairo. During the field sampling programme I profited greatly from the wide experience and sage advice of Professor Fred Wendorf, Professor Romuald Schild and Dr Achilles Gautier. More recently, Abdallah Sami Zaki Bakri has guided my understanding of Egyptian geomorphology. Between 2005 and 2012 I worked with local and international teams throughout central and northern Sudan. Special thanks go to Dr Abdelrazig Ahmed and Dr Yusif Elsamani (respectively, past and present directors of the Geological Research Authority of the Sudan), Dr Yasin Abdl Salaam, formerly University of Khartoum, Professor Osman et Tom, former Director of Soils Research at Wad Medani, and Neil Munro, for their patience, wisdom and guidance during my more recent visits and their sterling help with fieldwork. I also give warm thanks to Dr Donatella Usai and her husband Dr Sandro Salvatori for inviting me to join their Italian archaeological team excavating at El Khiday west of the lower White Nile in November 2011 and to Professor Matthieu Honegger for inviting me to work with him and his Swiss archaeological team at Kerma in northern Sudan during January 2012. I am especially grateful to the members of my research team who were funded by my Australia Research Council grant to study the history of floods and droughts in the Nile Basin over the past 30,000 years: Professor Mike Talbot, University of Bergen; Professors

Acknowledgements

xvii

Geoff Duller and Mark Macklin, Aberystwyth University; and Professor Jamie Woodward, University of Manchester. During 2006 Dr Timothy Barrows and Dr David Haberlah helped in mapping the White Nile mega-lake, and in 2009 Tim joined us in the Semien Highlands of Ethiopia to collect samples of glacial moraine for surface exposure dating. Over the years I have benefited greatly from the enlightened financial support of the Australian Research Council and the US National Science Foundation. My final debt is to the librarians and staff of the Australian National University Menzies Library, the National Library of Australia, the Cambridge University Library, the Khartoum University Library, Macquarie University Library, Monash University Library, the Barr Smith Library, University of Adelaide, the CNRS Quaternary Geology library at Meudon-Bellevue and the CNRS CEREGE library at the Europôle de l’Arbois near Aix-en-Provence, France. As always, friends and colleagues from around the world have sent me books and reprints of their work relating to the Nile Basin, which is one of the joys of belonging to the international community of scholars. I thank Professor Emi Ito for her sage advice on the subtleties of isotope geochemistry and Professor Peter Mühlhäusler for clarifying some arcane aspects of linguistics. Over the years Karl Butzer, Dan Livingstone, Mike Talbot, Claudio Vita-Finzi, Fred Wendorf, Romuald Schild, Desmond Clark and many others assisted with wise advice and welcome publications. I remain in their debt and treasure the memories of my interactions with them. I thank Dr Matt Lloyd for sustained support, Mariela Valdez-Cordero at Cambridge University Press, New York, for her patient and wise editorial advice; Theresa Kornak for copy-editing; Zoe Pruce and Sunantha Ramamoorthy for managing the production of this book. Finally, and by no means least, I thank my wife, Frances, for drawing the figures with her customary elegance and clarity, and for those memorable field trips to the Ethiopian Rift Valley; Jebel Marra volcano, Darfur, Sudan; and the glaciated Semien Highlands of Ethiopia.

1 The Nile Basin: An Introduction

Ex Africa semper aliquid novi (“There is always something new coming out of Africa”) After Pliny the Elder (AD 23–79), who adapted it from Aristotle (384–322 BC)

1.1 Introduction The following comments relate to the hydrologic status of the Nile before construction of the dams that today regulate the flow of water and sediment in its vast drainage basin (see Chapter 3). The reason for this is simple. Only by understanding the hydrologic behaviour of the unregulated Nile River and its tributaries can we begin to make the journey back in time and reconstruct the pattern and tempo of past hydro-climatic changes. The Nile is the longest river in the world (6,853 km) and has the third largest drainage basin (3.3 million km2), after the Amazon (7.2 million km2) and the Orinoco (3.8 million km2) (Williams et al., 1998, Table 8.1; Woodward et al., 2015a). Three main tributaries (and their associated tributaries) provide water and sediment to the main Nile: (a) the White Nile, (b) the Blue Nile and (c) the Atbara (Fig. 1.1). The White Nile rises in the equatorial uplands of Uganda, Rwanda and Burundi, and flows from Lake Victoria into Lake Kyoga and then into Lake Albert, after which it flows across a vast low-angle alluvial fan into the Sudd swamps of South Sudan and thence on to meet the Blue Nile at Khartoum. Fed by numerous small tributaries in its mountainous upper reaches, the headwaters of the Blue Nile flow down from the highly dissected Ethiopian Highlands into Lake Tana, which contributes about 9% of the total Blue Nile discharge (Hurst, 1952; Shahin, 1985; Sutcliffe and Parks, 1999; Sutcliffe, 2009). Below Lake Tana, the Blue Nile is entrenched in a gorge nearly 2 km deep, into which flow major tributaries such as the Jema, Guder, Didessa, Dabus and Beles Rivers. These tributaries provide much of its total discharge and sediment load. The Blue Nile then emerges from the Ethiopian uplands and flows across the Gezira alluvial fan to its confluence with the White Nile. The main Nile or ‘Desert Nile’ flows north from Khartoum for 320 km, at which point it receives water and sediment seasonally from another Ethiopian tributary, the Tekezze-Atbara River, and then flows north for a further 1

2

The Nile Basin: An Introduction 20 oE

30°E

40°E

MEDITERRANEAN SEA IV

cataract marshes

30o N

N

EGYPT

Ni l

LIBYA

e

RE

I

Aswan

D

II

Laqiya

SE

El Qaab III Basin

Kerma Seleim

Dongola Basin

Wa d el M e

d

IV

ar i How

VI

lik

V Atb ar

ER

a

W a

CHAD

A

20°N

KHARTOUM

IT R

EA

El Khiday Semien Mts

Nile

10°N

Lake Tana

e

White

SUDAN

Blu

Nile

Jebel Marra

So ba t

SOUTH

ETHIOPIA SUDAN

SUDD O

hr Ba

CENTRAL AFRICAN REPUBLIC

el Jebel

UGANDA Lake Albert

Lake Edward

0°N 250 km

Figure 1.1 The Nile Basin.

500

DA AN RW

0

o

Lake Turkana

KENYA

Lake Victoria Mt Kilimanjaro Lake Challa

SOMALIA

CONGO

m

1.2 Early Speculation about the Nile

3

2,689 km across the now waterless desert country of northern Sudan and Egypt, finally to reach the Mediterranean. The Nile Basin contains an unparalleled record of how prehistoric human societies adapted to regional climatic events and to changes in the Nile flood regime, from the time of Early Stone Age/Lower Palaeolithic hunter-gatherers to the Neolithic food-producing economies based on plant and animal domestication. The lower Nile Valley was one of the cradles of urban civilisation, totally dependent on floods from the upper Nile, just as Egypt is today, albeit moderated through the Aswan High Dam reservoir. The inhabitants of the arid lands of northern Sudan and Egypt owe their very existence to the Nile. By the year 2020 more than 300 million people will depend on its waters for their livelihood, so that a clear understanding of present land use and the impact of climate change on Nile flooding is essential for any rational and long-term future planning (Swain, 1997; Ayoub, 1999; Mohamed et al., 2005; Williams, 2009b). One way to appreciate the possible impact of future climate change on the Nile is to investigate how the Nile has responded to past changes in global and regional climate. Within the Nile Basin there is widespread evidence of past environmental fluctuations in the form of landforms and sediments from now defunct rivers and lakes as well as fossil remains of plants and animals. Much of this evidence is very well preserved, thanks in large part to the aridity prevalent today across the northern half of the basin. As we shall see, the Nile Basin contains a generous slice of the climatic history of northeast Africa within the confines of its vast catchment of 3.3 million km2. Much of this volume deals with events that took place during the Quaternary Period. The Quaternary is the most recent of all the named geological periods and epochs of the past 4.6 billion years (4.6 × 109 years) of Earth history, and was a time of geologically rapid and frequent changes in climate. It was also the time when prehistoric humans emerged, initially in Africa, later spreading to every continent except Antarctica. The Quaternary Period covers the last 2.6 million years (2.6 Ma) and includes the Pleistocene Epoch (2.6 Ma to 12 ka) and the Holocene Epoch (12 ka to the present). Ma denotes a million years and ka denotes a thousand years. The Pleistocene is divided into three time units: Lower Pleistocene (2.6 Ma to 780 ka), Middle Pleistocene (780 to 125 ka) and Upper Pleistocene (125 to 12 ka) (Gibbard et al., 2010). Where a precise and accurate chronology is lacking and use is made of fossils or prehistoric stone tool assemblages to specify a general age, lowercase lower, middle and upper are used to alert the reader to the lack of precise chronology. Walker et al. (2012) have proposed on a provisional basis that the Holocene be divided into three: Early Holocene (12 to 8.2 ka), Middle Holocene (8.2 to 4.2 ka) and Late Holocene (4.2 ka to present). We follow their suggestion where there is adequate age control, but otherwise use lowercase for early, middle and late Holocene.

1.2 Early Speculation about the Nile Herodotus (ca. 485–425 BC) deserves pride of place among the scholars and explorers who have long sought to understand the Nile. He was an indefatigable traveller, a shrewd listener

4

The Nile Basin: An Introduction

and a highly observant man. It was Herodotus who concluded that ‘Egypt is a gift of the river’. Some authors have translated this as ‘the gift of the Nile’ but the word Nile is a later corruption of the original Egyptian term for ‘the river’. (For a scholarly review of this topic, see Griffiths, 1966.) Herodotus was very intrigued by the Nile (Herodotus, trans. 1954) and by the people and monuments of Egypt, commenting that the Egyptian climate was ‘peculiar to that country and the Nile different in its behaviour from other rivers elsewhere’ (Herodotus, trans. 1954, p. 115). On a day’s journey by boat offshore from the coast of the Nile Delta, in the eastern Mediterranean, he found that casting a weighted line into the sea revealed a muddy bottom at a depth of eleven fathoms (20 m), ‘which shows how far out the silt from the river extends’ (op. cit., p. 104). His voyage over 1,100 km upstream led him to agree with the priests, whom he had questioned at length and often sceptically, that ‘the greater part of the country I have described has been built up by silt from the Nile’ (op. cit., p. 105). He also concluded that an arm of the Nile near the coast could become silted up within 20 to 10 thousand years, preferring the shorter estimate. He displayed a truly geological sense of time. However, what really puzzled him was the flood regime of the Nile: ‘What I particularly wished to know was why the water begins to rise at the summer solstice, continues to do so for a hundred days, and then falls again at the end of that period, so that it remains low throughout the winter until the summer solstice comes round again in the following year’ (op. cit., p. 109). He then mentioned three possible explanations, all seemingly logical, but none in fact correct. He was dismissive of the first two as ‘not worth dwelling upon, beyond a bare mention of what they are’. The first was that the summer winds caused the river to rise by checking the flow of the current towards the sea. Since the summer winds had little effect on other, smaller rivers, he rejected this notion. He rejected the second explanation of the Nile flowing from the ocean and then around the world as entirely fanciful. He considered the third explanation more plausible but utterly far-fetched, with the Nile said to derive its water from melting snow, a suggestion he castigated as ‘obviously, this view is worthless’ (op. cit., pp. 109–110). Of the three, the third hypothesis was closer to the mark than he realised. Of special interest was his discovery (op. cit., p. 108) that during the reign of Moeris (some 900 years before his visit) ‘the whole area below Memphis used to be flooded when the river rose only twelve feet’ (4 m) but that during the time that he was there the river never flooded until it had risen 24 feet (8 m), indicating that the land adjoining the Nile was rising or aggrading. (The ancient city of Memphis is located 24 km [15 miles] south of present-day Cairo.) Napoleon’s engineers and later observers were to confirm this progressive build-up of the Nile’s flood plain (Bell, 1970). Herodotus also commented that he had seen shells on the hills near the coast and ‘noticed how salt exudes from the soil to such an extent that it affects even the pyramids’ (op. cit., p. 106), which is perhaps the first written record of salt weathering. Whether the shells he saw were carried there by humans and were historic or prehistoric shell middens is not made clear but does seem possible. In addition to Herodotus, many other scholars have engaged in speculation about the Nile (see reviews by Harrington, 1967 and Biswas, 1970). For example, Aristotle (384–322 BC)

1.4 Aims and Structure of This Volume

5

and Eratosthenes of Alexandria (276–194/192 BC) correctly attributed the Nile summer floods in Egypt to seasonal rainfall in the Nile headwaters (wherever those headwaters might be). Eratosthenes was to use the length of the sun’s midday shadow at two widely separated locations (Alexandria and modern Aswan) on the Nile of known distance apart to calculate the circumference of the earth, which he did with astonishing accuracy. Leonardo da Vinci (1452–1519 AD) considered that the Nile headwaters flowed from the Mountains of the Moon into three great lakes located at an elevation of about 2,000 m. This conjecture seems remarkably accurate as far as the Ugandan headwaters of the White Nile are concerned. In 1613 the Portuguese Jesuit Pedro Paez described the source of the Blue Nile upstream from Lake Tana in Ethiopia, a discovery scoffed at by the great Scottish explorer James Bruce of Kinnaird (Bruce, 1790) but confirmed a hundred years later by the Portuguese Jesuit Hieronimo Lobo, who stated unequivocally that the floods in Egypt were a result of rainfall in Ethiopia (Moorehead, 1972; Rzóska, 1976).

1.3 Unique Attributes of the Nile The Nile spans 35° of latitude (3°S to 32°N). As a result, the Nile Basin (Fig. 1.1) embraces a very wide range of climates (equatorial, monsoonal, seasonally wet tropical, semi-arid, arid, hyper-arid, Mediterranean), equivalent to a Southern Hemisphere climatic transect from equatorial Indonesia and Papua New Guinea to the winter rainfall regions of southern Australia. Roughly a third of the Nile Basin (1,070,000 km2) is presently devoid of perennial rivers (Woodward et al., 2015a). The White Nile provides much of the low season flow to the Nile but very little sediment. The Blue Nile and Atbara together provide most of the flood flow and much of the sediment. There could hardly be a greater contrast than that between the Blue Nile and the White Nile. This distinction was pithily (and feelingly) encapsulated by Sir Samuel Baker (1866), who described the Blue Nile as ‘a mountain stream, rising and falling rapidly’ while the sluggish White Nile flowed through ‘a land of malaria, marshes, mosquitoes, misery’.

1.4 Aims and Structure of This Volume As the title of this volume indicates, the primary aim of this work is to review what is currently known about the Quaternary geology, geomorphology and prehistoric environments of the Nile Basin. The geographical focus will be the Nile Basin, but wherever relevant we will also discuss the evidence from adjoining regions such as the Sahara and Middle East, and most especially the eastern Mediterranean. Chapter 2 sets the scene with a concise overview of the evolution of the Nile Basin. This is followed in the next three chapters with succinct accounts of the climate and hydrology of the Nile Basin, including historic floods and droughts (Chapter 3), its geology and soils (Chapter 4) and its vegetation and current land use (Chapter 5), including the thorny problem of flood control, dams, reservoirs and disease. Figure 1.2 shows the geographical location of each regional chapter.

6

The Nile Basin: An Introduction 20 oE

30°E MEDITERRANEAN SEA

40°E

21 20 30o N

marshes

19

CAIRO

17 N

EGYPT

18 D RE

16

e Nil

LIBYA

Aswan

16

ar How

15

Atb ar lik

KHARTOUM

Wa d el

Me

di

Dongola

13

ER IT R

a

W a

CHAD

A SE

14

20°N

11

12

EA

Jebel Marra

White Nile

8

SUDAN

ETHIOPIA

SUDD O

el Jebel

9 UGANDA

Lake Albert

Lake Edward

0°N km

500

DA AN RW

250

Figure 1.2 Location of the regional chapters.

m

o

Lake Turkana

KENYA

7

Lake Victoria Mt Kilimanjaro

TANZANIA

Lake Challa

SOMALIA

CONGO

0

6

ba t

hr Ba

CENTRAL AFRICAN REPUBLIC

Nile

So

SOUTH

Lake Tana

e

10°N

Blu

10

SUDAN

1.4 Aims and Structure of This Volume

7

The physical regions within the Nile Basin are considered in some detail in Chapters 6 to 20, proceeding from the headwaters of the Blue and White Nile (Chapters 6 and 7), along the White Nile (Chapter 8) and its tributaries (Chapters 9 and 10) to the lower Blue Nile (Chapter 11) and the Atbara River (Chapter 12). We then discuss the western borders of the Nile Basin, visiting Jebel Marra volcano and its environs (Chapter 13) as well as the now defunct river systems of Wadi Howar, Wad el Melik and Wadi Muqadam. Given their very great importance in prehistory, four presently arid regions are reviewed in some detail: Nubia, the Butana Desert and the Desert Nile in Egypt (Chapter 14) and the Western Desert of Egypt (Chapters 15 and 16) and the Fayum (Chapter 17). Equally significant in prehistory are the Red Sea Hills (Chapter 18) and the Sinai Desert (Chapter 19). A recurrent theme in all of the chapters mentioned so far is the persistent and complex set of interactions between river and desert in prehistory. Chapters 20 and 21 discuss the Nile Delta and the Nile Cone, respectively. The evidence from sediment cores from both the Delta and the submarine Nile Cone provides a wealth of information about past changes in Nile sediment flux, river discharge and changes in Nile sediment sources. Chapter 22 discusses the complex question of plant and animal domestication in the Nile Basin and reviews some of the models that have been proposed to account for the origin and spread of plant and animal domestication across the Nile Basin. The final chapter (Chapters 23) concludes with an attempt to unravel the nature and timing of prehistoric migrations to and from the Nile Basin and the various ‘Out of Africa’ scenarios that have been proposed.

2 Evolution of the Nile Basin

The Soudan is joined to Egypt by the Nile, as a diver is connected with the surface by his air pipe. Without it there is only suffocation. Aut Nilus, aut nihil! Winston S. Churchill (1874–1965), The River War (1899)

2.1 Introduction: How Old Is the Nile? Churchill’s play on words aut Nilus, aut nihil is a whimsical corruption of the old Latin tag aut Caesar, aut nihil, and may be rendered very roughly as ‘without the Nile there would be nothing’. It is a succinct epigram designed to underscore the importance of the Nile River to the lands through which it flows. We return to this theme in Chapters 3 to 5. The aim of this chapter is to provide a concise overview of the evolution of the Nile Basin in order to set the scene for Chapters 6 to 21. Talbot and Williams (2009) have pointed out that ‘the Cenozoic evolution of the Nile Basin reflects a complex interaction between tectonic, volcanic and climatic events’ (p. 37). As we shall see, it is not possible to have a clear appreciation of Nile history without first understanding the tectonic and volcanic events that have shaped the Nile Basin and that have in their turn influenced the climatic and hydrologic history of the Nile and its tributaries. The White Nile joined the main Nile only comparatively recently in geological terms, most likely no earlier than about 0.5 Ma ago (Talbot and Williams, 2009). In contrast, the inception of the Blue Nile and Tekezze/Atbara Rivers goes back to almost 30 Ma ago (Pik et al., 2008; Fielding et al., 2016, 2018), following the final eruption of the Trap Series basalts in Ethiopia and the associated uplift of what became the Ethiopian Highlands (see Chapter 6). Erosion into and through the basalts has removed roughly 100,000 ± 50,000 km3 of rock from the headwaters of the Blue Nile and Tekezze/Atbara Rivers (McDougall et al., 1975), creating huge gorges in which we see exposed the Precambrian bedrock. The volume of the Nile Cone is the same order of magnitude as the volume of rock eroded from Ethiopia, providing initial circumstantial evidence that the main Nile may well have been connected, at least intermittently, with its Ethiopian headwaters since Oligocene times some 30 Ma ago (McDougall et al., 1975; Williams and Williams, 1980). More recent work has confirmed that the Nile Cone has been

8

2.2 Ethiopian Uplift and Volcanism

9

in receipt of sediment from the Ethiopian Highlands for at least 30 million years (Fielding et al., 2016, 2018), and that former rivers that eroded into the Red Sea Hills have also contributed significantly to the amount of sediment deposited in the Nile Cone (Macgregor, 2012). During the Messinian Salinity Crisis (Hsü et al., 1977; Hsü, 1983), now precisely dated to between 5.96 Ma and 5.33 Ma (Cosentino et al., 2013), the Mediterranean dried out and refilled perhaps a dozen times and up to 1 km of halite and other evaporites accumulated on the floor of what became the Mediterranean salt desert (Hsü et al., 1977). During times of peak desiccation, the base level of the late Miocene Nile was lowered to the floor of this salt desert and the ancestral Nile eroded vertically to depths of –2.5 km north of Cairo, –0.8 km at Assiut and –170 m at Aswan, located some 1,200 km upstream of the present Delta (Chumakov, 1967; Said, 1993, p. 38). Said (1993) called this late Miocene Nile canyon the ‘Eonile’ canyon. This canyon became an estuary during the Pliocene and eventually filled with sediment, some of which came from the Nile and some from previously active rivers flowing westwards from the Red Sea Hills and eastwards from the presently arid Western Desert of Egypt. These rivers seldom flow today. Said (1993, p. 39) referred to the Nile River that flowed into the estuary and eventually succeeded it as the ‘Paleonile’.

2.2 Ethiopian Uplift and Volcanism Over the past fifty years and more, many geologists have worked to establish a detailed chronology of the Ethiopian Trap Series basalts (Merla, 1963; Mohr, 1968; McDougall et al., 1975; Hofmann et al., 1997; Chorowicz, 2005; Pik et al., 2003, 2008; Gani et al., 2007; Prave et al., 2016). The Trap Series were erupted over a prolonged interval between 45 and 15 Ma but the bulk of the eruption took place between 31 and 29 Ma. Hofmann et al. (1997) considered that eruption of the Trap Series basalts had occurred about 30 Ma ago and was largely accomplished within about a million years. The apatite helium ages obtained by Pik et al. (2003, 2008) to test models of landscape evolution in Ethiopia revealed that there had been a partial resetting of pre-existing basement rock ages because of burial of the basement rocks beneath a thick cover of Trap Series flood basalts roughly 30 Ma ago. From this they concluded that erosion of the Blue Nile gorge was underway as early as 29–25 Ma ago, vindicating the earlier conclusions of McDougall et al. (1975). Chorowicz (2005) proposed that a mantle plume began to form 30 Ma ago in the region around present-day Lake Tana (Fig. 2.1), resulting in domed uplift and faulting in three main directions that converged on the present lake. More recent work by Prave et al. (2016) has confirmed that the uppermost basalts in the uplands around Lake Tana date back to about 30 Ma, but these authors also consider that there had been a supereruption and that a very large volcanic caldera formed in the region now partially occupied by Lake Tana.

10

Evolution of the Nile Basin

Figure 2.1 Lake Tana, Ethiopia. (Photo: Frances Williams.)

Figure 2.2 The Blue Nile Tisisat Falls downstream of Lake Tana, Ethiopia.

2.3 Erosion of Ethiopian Nile Headwaters

11

2.3 Erosion of the Ethiopian Nile Headwaters James Bruce of Kinnaird was greatly taken with the once spectacular Tisisat Falls (Fig. 2.2) near the outlet of Lake Tana (Bruce, 1790). However, the grandeur of the falls has been greatly curtailed since 2011 as a result of diversion of Blue Nile water and creation of a vast reservoir at Bahar Dar on the southern shore of the lake to provide hydroelectric power for the region. Immediately downstream of the falls and before construction of the dam the entire flow of the Blue Nile emerging from Lake Tana was confined into a narrow canyon spanned by an ancient stone bridge known locally as the Portuguese Bridge (Fig. 2.3). From this point on the Blue Nile flows for 150 km across mid-Tertiary basalts before cutting down into horizontally bedded Mesozoic sedimentary rocks more than 1 km thick to eventually reach the Precambrian basement rocks about 330 km downstream from Lake Tana. In the process of entrenching its channel the Blue Nile has cut a very deep and impressive gorge (Fig. 2.4). About 280 km below Lake Tana the main road between Addis Ababa and Debre Markos crosses the gorge at 10°05 0 N, 38°10 0 E. In the sides of the gorge Trap Series basalts 250 m thick overlie at least 1,150 m of Mesozoic sedimentary rocks (limestone, marl, shale, gypsum and sandstone). The base of the honey-coloured Adigrat Sandstone (Fig. 2.5) which crops out at the bottom of the gorge is not exposed at this location. The gorge is here 1,400 m deep, 20 km wide and incised into a plateau 2,400 m in elevation (McDougall et al., 1975). The Tekezze River has cut an equally impressive gorge to the north and west of the glaciated Semien Highlands, which themselves lie to the north of the Blue Nile headwaters (Fig. 2.6). McDougall et al. (1975) estimated that the volume of rock eroded from the Tekezze headwaters above 1,000 m elevation amounts to 31,000 km3 from a catchment area above that elevation of 85,000 km2. Corresponding values for the Blue Nile headwaters are 71,000 km3 and 190,000 km2. The sum of area and volume for the combined catchments is therefore 275,000 km2 and 102,000 km3. The mean rate of denudation can be obtained using the simple formula V/AT, where V is volume in m3, A is area in km2 and T is time or age in years. If we divide the total volume of rock eroded by the product of the age of the uppermost basalts (30 Ma) into which the rivers are cut and the total catchment area we obtain a mean rate of denudation for the Blue Nile and Tekezze/Atbara headwaters of 12 m3 per km/yr, which is equivalent to a mean annual rate of surface lowering of 0.012 mm/yr. This is the sort of erosion rate we might expect for lowland regions with a dense rainforest cover and not one typical of upland regions prone to rapid uplift (Gibbs, 1967; Douglas, 1967, 1969; Milliman and Meade, 1983; Milliman, 1997). A possible explanation is that the Ethiopian highlands were subject to intermittent uplift separated by long periods of tectonic stability and that for much of this time the climate was humid and the forest and grassland cover relatively dense. Gani et al. (2007) used a digital elevation model to reconstruct the original topography prior to erosion of the Blue Nile gorge. They then plotted the potassium–argon ages of volcanic rocks that now form erosional remnants to determine the history of uplift and

12

Evolution of the Nile Basin

Figure 2.3 Blue Nile channel entrenched into basalt downstream of the Tisisat Falls.

Figure 2.4 Blue Nile gorge about 280 km downstream of Lake Tana.

2.3 Erosion of Ethiopian Nile Headwaters

13

Figure 2.5 Adigrat Sandstone exposed near the bottom of the Blue Nile gorge.

associated vertical erosion, concluding that the Blue Nile and its tributaries had eroded 93,200 km3 of rock from the Ethiopian plateau, and that uplift had taken place in three main stages (29–10 Ma, 10–6 Ma and 6–0 Ma), with two pulses of more rapid erosion towards 10 Ma and 6 Ma. These results accord with those of Merla (1963), who had suggested much earlier that uplift of the Ethiopian Highlands had taken place in two stages, one in the Miocene and one in the Pliocene–Pleistocene, with development of the Blue Nile and other deep gorges during times of accelerated uplift. Ismail and Abdelsalam (2012) suggested that uplift and incision were episodic in the eastern sector of the NW Ethiopian Highlands but may have been more continuous towards the west, away from the influence of rift-flank

14

Evolution of the Nile Basin

Figure 2.6 Semien Highlands from which flows the Tekezze River.

uplift (ca. 11 Ma) and shield volcano formation (ca. 22 Ma). More recently, Sembroni et al. (2016) have argued on theoretical grounds for continuous uplift and a constant rate of fluvial incision in the Blue Nile and Tekezze headwaters. Any model is no better than the assumptions upon which it is based and a different set of initial postulates would have yielded a different set of results, so that the issue of continuous versus episodic uplift is best considered unresolved. The estimate by Gani et al. (2007) that the Blue Nile and its headwater tributaries had eroded 93,200 km3 of rock is similar to the earlier estimate by McDougall et al. (1975) of about 71,000 km3.

2.4 Tectonic History of Lake Victoria and the Ugandan Nile Headwaters Before the inception of rifting in Uganda and creation of the Western Rift, the general orientation of the rivers was from the more elevated lands in the east into the present Congo/ Zaïre Basin lying to the west. Rifting began in the northern sector of the Western Rift during the Miocene about 8–9 Ma ago (Talbot and Williams, 2009), with uplift along the rift margins. The rifting had several consequences (see Chapter 7). One consequence was segmentation and diversion of the earlier drainage system, including ponding of water in what became Lake Kyoga with its characteristic crenulated and ‘barbed arrow’ pattern (Talbot and Williams, 2009, p. 46). Another consequence was the formation of large lakes

2.5 Tectonic and Structural Control of the Nile

15

within the floor of the Western Rift, such as the very large ‘Paleolake Obweruka’ identified and mapped by Van Damme and Pickford (2003). Lakes Albert, Edward and George are but the latest in a succession of still poorly known Rift Valley lakes. A third and important consequence of the earth movements associated with the rifting was the formation of Lake Victoria roughly 0.5 Ma ago. Tilting of the floor of Lake Victoria has exposed a series of as yet imprecisely dated but most likely mid- to late Pleistocene lake sediments along the western margin of the lake. The drainage connections between Lake Victoria, Lake Kyoga and Lake Albert were probably established by the Middle Pleistocene, with overflow from the Uganda plateau via a series of gorges and cascades and across a set of large, gently sloping alluvial fans into the great Sudd swamps of South Sudan, described in detail in Chapter 8.

2.5 Tectonic and Structural Control of the Nile and Its Tributaries Even a casual glance at a map of the Nile Basin (see Chapter 1, Fig. 1.1) will reveal some remarkable linear features, with major Nile tributaries flowing parallel to each other, such as the Atbara and Blue Nile, or the Rahad and the Dinder. In addition, certain reaches of the Nile are very straight. Perhaps the most striking feature of all is the great bend in the Nile, which begins about midway between the 5th and 4th cataracts (Abdelsalam, 2018). Downstream of its confluence with the Atbara the Nile flows NNW for nearly 200 km, then turns abruptly SW for about 200 km, flows NNW past the 3rd cataract for a further 300 km, and then alters its flow to the NE to enter Lake Nasser below the present Aswan High Dam situated on the rocks of the 1st cataract and then flows NNW again. These linear reaches of the Nile reflect the influence of geological features such as faults, dykes, bands of structural weakness within the bedrock, contacts between hard and soft rocks or the metamorphic grain preserved within the Precambrian and younger ‘Basement Complex’ rocks. A useful but very general term for such features is the term ‘lineament’, which simply refers to any linear feature considered to be of natural or structural origin. Analysis of satellite imagery and air photos (Mohr, 1974; Adamson and F. Williams, 1980; Adamson and Williams, 1987; Adamson et al., 1993; Avni et al., 2012) has shown that within the Nile Basin the observed lineaments are oriented in two major directions, one aligned NE–ENE, or parallel to the alignment of the Gulf of Aden and the other aligned SE–SSE, coinciding with the orientation of the Red Sea. A third NNE orientation is also evident in places, notably in Ethiopia, where it coincides with the orientation of the Ethiopian Rift Valley (F. Williams et al., 2004; F. Williams, 2016). The Jebel Marra volcanic massif near the margin of the Nile Basin in western Darfur Province, Sudan, lies at the intersection of two major lineaments (see Chapter 13, Fig. 13.1). One runs from SE to NW, from the Aswa wrench fault in South Sudan across the Sahara to the Mediterranean Sea in northern Libya. The other runs from the Cameroon Mountains in the SW across the Bayuda Desert to the Red Sea Hills in the NE (see Chapter 13).

16

Evolution of the Nile Basin

Although some of the faults within the Nile Basin have been active as recently as the late Pleistocene and Holocene, particularly in zones of active rifting such as Ethiopia and parts of Uganda, many of the lineaments date back to the Archaean-Proterozoic or to the 700–550 Ma East African Orogeny, and have been reactivated during subsequent intervals of tectonic activity (Adamson et al., 1993; Avni et al., 2012; Abdelsalam, 2018). A buried Rift Valley in South Sudan may date back to the Oligocene or earlier and has long acted as a sediment sink. Unpublished drilling and seismic studies by Chevron in the early 1970s revealed that the Cenozoic sediments were at least 11 km thick in this region.

2.6 Volume of the Nile Cone The Nile Cone is one of the world’s great submarine deltas, equal in size and volume to those of the Amazon, Ganga and Brahmaputra. Oil company geologists have carried out a great deal of detailed drilling and geophysical surveys in the Nile Cone and adjacent areas in the eastern Mediterranean. This work has involved comprehensive studies of the microfossils, trace element geochemistry and clay mineral composition of the sediments recovered in the drill cores. Unfortunately for science, much of this work is considered ‘commercial in confidence’, has not been published in peer-reviewed journals or even in publically available reports, and so is not widely available to help us answer key questions relating to the age and development of the Nile Cone. However, the oceanographic research vessels of a number of countries have carried out a detailed program of drilling in the eastern Mediterranean and this has provided useful insights into changes in Nile water and sediment discharge during the past few hundred thousand years, although many of the records are much younger (see Chapter 21). A few Nile Cone sediment cores extend back to the Early Pleistocene and Pliocene and a very few to even earlier. The account that follows is therefore best considered as a work in progress rather than a definitive review. Using gravity anomaly data, Harrison (1955) estimated that the Nile Cone was up to 320 km wide, attaining depths of –2.8 km to the west and –1.6 km in the east. He calculated that the volume of the submerged Nile Cone would amount to 95,000 km3 if there had been no crustal sag, but concluded that 220,000 km3 was a more realistic estimate, taking into account Pliocene and Pleistocene subsidence caused by the weight of overlying water and sediment. Emery et al. (1966) inferred a volume of 140,000 km3 from Harrison’s contour map. Seismic analyses by Wong and Zarudski (1969) led them to propose a volume for the submerged Nile Cone amounting to between 100,000 km3 and 200,000 km3. More recent work indicates a volume slightly in excess of 500,000 km3 (Macgregor, 2012). Rossignol (1961, 1962) found Ethiopian mountain forest pollen in early Pleistocene Nile Cone sediments, and Emel’yanov (1972) found a high proportion of Ethiopian volcanic minerals in early Pleistocene Nile Cone sediments, providing further support for Rossignol’s findings. These data pointed to a minimum age of about 2.5 Ma for the

2.7 Conclusion

17

hydrologic link between the Ethiopian headwaters and the Nile Cone. Owing to the fragmentary nature of the alluvial record in Egypt, early workers such as Butzer and Hansen (1968) had been unable to find convincing evidence of a link between Ethiopia and the Nile in Egypt much before about 50 ka. Hassan (1977) studied the heavy minerals in alluvial sediments in the Egyptian Nile Valley and came up with a minimum age of Middle Pleistocene for the drainage link with Ethiopia. All of this early work has since been superseded by the detailed isotopic and trace element studies by Fielding et al. (2016, 2018) which have demonstrated conclusively that the Nile Cone is at least 30 million years old and has been in receipt of sediment from the Ethiopian volcanic uplands since that time. What remains unclear is how sporadic this connection may have been.

2.7 Conclusion Tectonic, volcanic and climatic events all played a role in the Cenozoic evolution of the Nile Basin. The tectonic influence is evident in the long history of reactivation of major lineaments within and beyond the present-day Nile Basin, as well as in the rectilinear character of the main Nile channel between the 5th and 4th cataracts. Some of these lineaments date back to the Archaean and Proterozoic; others originated (or were re-activated) during the 700–550 Ma East African Orogeny. Many of the faults associated with these ancient lineaments have been active during the Pleistocene and Holocene, notably in the Ethiopian and Afar Rifts and in the Lake Victoria catchment in Uganda. The lineaments follow three major alignments: NE–ENE, parallel to the Gulf of Aden); SE–SSE, parallel to the Red Sea; and NNE, parallel to part of the Ethiopian Rift Valley. The uppermost lava flows through which the headwaters of the Abbai/Blue Nile and Tekezze/Atbara are incised date back to the Oligocene and were extruded some 30 Ma ago. Both river systems have since eroded over 100,000 km3 of rock form their upper catchments, most probably in response to episodic and possibly accelerating uplift in the Ethiopian Highlands. The Nile Cone in the eastern Mediterranean has a volume of 500,000 km3, a significant part of which came from the rocks eroded from the Ethiopian Highlands and transported downstream by the ancestral Blue Nile and Atbara rivers during the past 30 Ma. Pollen grains derived from the mountain forests of Ethiopia and heavy minerals consistent with an Ethiopian volcanic provenance have been recovered from late Pliocene–early Pliocene sediments obtained from the Nile Cone. Isotopic and trace element analyses of Nile Cone sediments have now confirmed that the Ethiopian-Nile Cone connection dates back to 30 Ma, when there was a pulse of widespread eruption of basaltic lavas across what are now the Ethiopian Highlands. The ancestral Blue Nile and Tekezze Rivers began to carve their spectacular gorges in response to the tectonic uplift associated with this volcanism. We do not yet know whether the link to the Mediterranean was intermittent or continuous.

18

Evolution of the Nile Basin

In contrast to the antiquity of the Blue Nile and Atbara River systems, the White Nile only joined the main Nile about 0.5 Ma ago. The inception of Lake Victoria in Uganda and its connection to Lake Albert and the buried Rift Valley in South Sudan is a delayed response to the diversion and segmentation of the Oligocene–Early Miocene drainage system that once flowed westwards across Uganda into the Congo/ Zaïre Basin and was disrupted by the onset of late Miocene rifting in western Uganda.

3 Climate and Hydrology

What I particularly wished to know was why the water begins to rise at the summer solstice, continues to do so for a hundred days, and then falls again at the end of that period, so that it remains low throughout the winter until the summer solstice comes round again in the following year. Herodotus (ca. 485–425 BC), The Histories (1954, p. 109)

3.1 Introduction Herodotus (ca. 485–425 BC) was puzzled by the hundred days of annual Nile floods during the time of the summer solstice when no rain fell in Egypt, correctly interpreting the cause as heavy precipitation in the Nile headwaters during this time of year. The precise causes of the Nile summer floods have intrigued travellers and scholars since at least the time of Herodotus nearly 2,500 years ago. The aims of this chapter are to consider why the Nile summer floods occur and to indicate very briefly how they have varied in geologically recent and historic times.

3.2 Climates of the Nile Basin The Nile flows through a number of different climatic regions (Fig. 3.1). In the equatorial highlands of Uganda, Rwanda and Burundi, the climate is hot and wet throughout the year, with two distinctly wetter seasons at the time of year when the Sun is directly overhead. The subtropical Ethiopian Highlands have a more seasonal climatic regime, with one main wet season during the time of the summer monsoon (June–September) and a shorter wet season earlier in the year (March–April). The northward passage of the Intertropical Convergence Zone (ITCZ: see Section 3.2.1) across South Sudan and the southern half of Sudan ushers in the summer wet season in that region, with mean annual precipitation diminishing progressively northwards from nearly 1,000 mm near Juba (4.8°N, 31.6°E) and 800 mm near Malakal (9.5°N, 31.7°E) on the White Nile, to less than 200 mm in the latitude of Khartoum (15.5°N, 32.5°E) at the confluence of the Blue and the White Nile (Fig. 3.1) (Griffiths, 1972a). South of Khartoum daytime temperatures are hot throughout the year,

19

20

Climate and Hydrology 20 oE

30°E

40°E

MEDITERRANEAN SEA

30o N

EGYPT

LIBYA

RE D 10

A SE

50

CHAD

20E 0

R I

200

TR

300 A E

300

50

10

20°N

J

600 900

S

SUDAN

600 900 10°N SOUTH ETHIOPIA CENTRAL AFRICAN REPUBLIC

SUDAN

900

UGANDA KENYA

SOMALIA

900

CONGO

0°N 250 km

500

DA AN RW

0

K

Figure 3.1 Rainfall (mm) in the Nile Basin. Compiled from data in Williams et al. (1982, Fig. 7.11), Gasse et al. (2008), Nile Basin Initiative (2012, p. 30) and Williams (2014, Fig. 18.1).

3.2 Climates of the Nile Basin

21

but in the desert region north of Khartoum winter temperatures can be very cold and even fall below freezing during the night in the Nubian and Butana Deserts. The eastern Sahara is very hot in summer, with shade temperatures at times higher than 50°C in a land of minimal shade, while the winters are true desert winters: cold and often windy at night and warm during the day. It is dry throughout the year except for rare intense and highly localised rainstorms every few decades. The northern littoral region of the Nile Basin falls within the Mediterranean climatic zone, with long hot summers and cool winters. Winter rains tend to be sporadic in time and space, and in some years the Mediterranean coast of Egypt remains dry all year round. In years when the ITCZ migrates well south of the equator, the rainbearing westerly air masses may extend well inland across the Nile Delta. We will now consider the processes responsible for the differing climatic regimes that prevail across the Nile Basin.

3.2.1 The Hadley Circulation and Seasonal Migrations of the Intertropical Convergence Zone The Sun is most directly overhead for much of the year in equatorial latitudes. As a result, these are the zones that receive the highest incoming short-wave solar radiation. The land and the sea in these low latitudes absorb the radiation from the Sun, become warm and heat the air above them. The air moves upwards as a result of convection to heights of about 12–15 km above the Earth’s surface. As the air rises it expands and cools. Water vapour within the rising air columns condenses once the relative humidity attains saturation (100% relative humidity) and intense convectional downpours occur, usually in the late afternoon after prolonged solar radiation and the progressive formation of large convectional clouds. As the moist tropical air moves upwards over the equator and releases its excess moisture, the air aloft forms two separate plumes. One (the northern plume) moves northwards and the other (the southern plume) moves southward (Fig. 3.2). As these plumes approach about 15–20°N and 15–20°S (Fig. 3.3) they become progressively cooler and denser and begin to subside, bringing dry subsiding air over the tropical deserts (see Section 3.2.5). To compensate for the upward convectional movement of moist equatorial air there is a near surface return flow of air from tropical to equatorial latitudes, generating the trade winds well known to sailors in centuries past. The Intertropical Convergence Zone (ITCZ) is where the two air masses converge. Because of the rotation of the Earth the trade winds blow from NE to SW north of the equator and from SE to NW south of the equator (Fig. 3.3). Owing to the tilt of the Earth’s axis, the overhead Sun moves north in the boreal (northern) summer and south in the austral (southern) summer. The locus of maximum solar heating also moves north towards the Tropic of Cancer during the boreal summer and southwards towards the Tropic of Capricorn during the austral summer. As a result, the ITCZ also moves northwards during the boreal summer and southwards during the austral summer. This circulation system is known as the Hadley Circulation (see Diaz and Bradley, 2004), although as Webster

22

Climate and Hydrology

Figure 3.2 Schematic cross section of the global atmospheric circulation, showing location of the Hadley cells. (After Williams, 2014, Fig. 2.1.)

(2004a) points out, a more fitting name would be the Halley–Hadley Circulation, because it was first described by Sir Edmund Halley in 1686 and refined half a century later by Sir George Hadley in 1735. It is described here in very elementary terms – the reality is vastly more complex (Webster, 2004a) and still only partly understood.

3.2.2 The African Summer Monsoon Seafaring people in the Indian Ocean and surrounding seas have been aware of the seasonal wind changes in this region for well over 2,000 years. The word monsoon probably derives from the Arabic word mawsim and the Urdu and Hindi word mawsam for season. Early Dutch and Portuguese navigators had similar words for these seasonal changes in wind direction (and associated changes in seasonal rainfall), encapsulated today in the terms summer and winter monsoon. A monsoon is a subset of the Hadley Circulation (Webster, 2004a, 2004b) and results from differential heating of the land and sea in subtropical and tropical latitudes. For example, as peninsular India heats up during the boreal early summer months, the intense low pressure generated over the subcontinent results in a change in wind direction. The winter monsoon winds blow away from the land; the summer monsoon winds blow across the warm ocean surface towards the even warmer land surface, leading

3.2 Climates of the Nile Basin

23

Figure 3.3 Seasonal migrations of global wind systems and of the ITCZ. (After Williams, 2014, Fig. 2.2.)

to the intense convectional downpours that herald the onset of the summer monsoon. In Africa, the situation is somewhat more complex (Fig. 3.4). The East African summer monsoon receives moisture from the Indian Ocean and brings heavy rainfall to the Ethiopian Highlands between June and September. Further to the west and south, the West African monsoon receives much of its moisture from the South Atlantic. The Congo Air Boundary (CAB) is a zone of wind convergence and separates the two monsoonal systems (Nicholson, 1996; Gasse and Roberts, 2004; Gasse et al. 2008). The location of the CAB shifts with the seasons and has also varied from its present mean position during the late Quaternary (Costa et al., 2014).

24

Climate and Hydrology

Figure 3.4 Surface winds and frontal locations (a) during July–August and (b) during December. (Modified from Nicholson 1996 and Gasse et al., 2008). CAB, Congo Air Boundary; ITCZ, Intertropical Convergence Zone; NEM, northerly East African monsoon. In section: NHC, northern Hadley Cell; SHC, southern Hadley Cell.

The domain of the summer monsoon is similar to that of the El Niño Southern Oscillation (ENSO) influence (Fig. 3.5) first recognised by Sir Gilbert Walker (Walker, 1924; Williams and Balling, 1996; Williams, 2014). This has profound implications today for the people living in this vast region. During years when a weak summer monsoon coincides with a strongly negative Southern Oscillation Index (SOI), the result is severe and prolonged drought, as in 1982 and 1997; conversely, during years of a stronger than average summer monsoon synchronous with a strongly positive SOI, the result is extreme and widespread floods, as in 1998–2001 and 2010–11. The SOI is a measure of the atmospheric pressure difference between Darwin and Tahiti. Over the past 200 and more years, ENSO events have had a significant influence on the incidence of floods and droughts in some of the more heavily populated agricultural regions of India, China, Indonesia, Australia, Papua New Guinea as well as the Nile Basin (Diaz and Markgraf, 1992; Williams and Balling, 1996; Diaz and Bradley, 2004; Diaz and Markgraf, 2004) (Fig. 3.5). Severe droughts (or floods) in each of these regions coincide with strongly negative (or positive) values of the SOI.

3.2.3 Seasonal Dust Storms One feature of the Nile Valley well known to the present-day inhabitants of Egypt and north-central Sudan is the occurrence of the annual dust storms that turn day into night and make life difficult for humans and animals during these times. A dust storm is ‘a rapidly moving mass of air containing large amounts of dry, opaque particles which reduce visibility to less than 1,000 m’ (Griffiths and Soliman, 1972, p. 93). These authors distinguish three main types of dust storm in northern and central Sudan and southern Egypt: (a) an instability type related to the advancing ITCZ or early monsoon

3.2 Climates of the Nile Basin

25

Figure 3.5 Region influenced by the summer monsoon and the two key regions of the Southern Oscillation. During El Niño or ENSO events (negative SOI), surface atmospheric pressure is above normal in the stippled area and below normal in the hatched area. The opposite pattern prevails during La Niña or anti-ENSO events (positive SOI). For example, when surface atmospheric pressure is below average off the coast of Peru, it is above average in the area encompassing India, southern China, East Africa, Indonesia and Australia, and drought is common in these localities. The mapped areas (showing statistically significant correlations greater than +0.4 or –0.4, at the 95% confidence level) reflect the difference between annual surface atmospheric pressure at Jakarta, Indonesia and global mean sea level pressure. (Modified from Whetton et al., 1990; Williams et al., 2006a and Williams, 2014.)

(July–September), (b) a pressure gradient type caused by southerly winds (July–September) and (c) a pressure gradient type caused by northerly winds (February–May). The local term haboob (Arabic: to blow) is used loosely for all three types but Griffiths and Soliman (1971) contend that it should strictly refer only to type (a). Haboobs can be up to 25 km wide and up to 1 km high, moving at about 55 km per hour, and are responsible for mobilising and depositing significant volumes of fine silt and clay (Kendrew, 1957, plate 7; 1961, p. 71), a characteristic that led early geologists to conclude that the clay plains of central Sudan were formed of windblown dust, somewhat akin to the loess plains of northern Europe (see Chapter 11).

3.2.4 The Mediterranean Winter Rains Both the northern and southern extremities of Africa come within the ambit of the midlatitude westerly winds, which bring precipitation to these regions during their respective

26

Climate and Hydrology

winter seasons (Griffiths, 1972b; Rognon and Williams, 1977; Gasse and Roberts, 2004; Gasse et al., 2008; Nicholson, 2011; Williams, 2014). The Atlas Mountains in north-west Africa receive significant winter precipitation, the northern coast of Libya receives relatively little precipitation except on the high plateaux such as Jebel al Akhdar (‘the Green Mountain’) and the low-lying coastal zones of northern Egypt receive virtually no precipitation in winter except during occasional outbursts of heavy convectional rain moving north across the Red Sea. There is evidence from fossil pollen grains indicative of the former vegetation that on occasions during the Holocene winter rains were common well south of the present Mediterranean coast, a topic we discuss in Chapters 17 to 19.

3.2.5 Saharan Aridity The Sahara is the largest desert on Earth as well as one of the driest and oldest (Williams, 2014). Three main factors contribute to this aridity. First and most important, the Sahara is located astride the Tropic of Cancer and beneath the descending arm of the boreal (northern) Hadley Cell (Figs. 3.2–3.4). The dry air from this cell becomes compressed and warmer as it descends, lowering its relative humidity and enhancing its desiccating capacity. The second factor is the sheer size of the Sahara, 5,000 km from west to east, 2,000 km from south to north. As a result, moist air masses shed their surplus water along the margins of the Sahara and reach dew point in the Sahara only when confronted with high mountain barriers such as Jebel Marra, the Hoggar, Tibesti, Jebel ‘Uweinat and the Aïr (Fig. 3.6). In winter, the Sahara becomes a zone of high atmospheric pressure, with low-level winds blowing outwards in a clockwise fashion from the central and subsidiary anticyclones. In summer the reverse occurs, but the northward movement of tropical air masses is limited to about 20°N (Fig. 3.4), by which time they have released much of their precipitation. A third factor contributing to Saharan aridity and identified in the early 1970s by Flohn is location at the distal end of the subtropical easterly jet stream rising over the Tibetan Plateau during the summer monsoon season (Rognon and Williams, 1977; Flohn, 1980). The dry subsiding air beneath this jet stream has helped to accentuate aridity across Pakistan, peninsular Arabia, Somalia and the Sahara from at least Mid-Miocene times onwards (Williams, 2014).

3.3 Nile Hydrology and Nile Floods The Nile Basin consists of four very distinct hydrological systems: the Ugandan headwaters and the White Nile river basin, the Ethiopian headwaters and the Abbai–Blue Nile river basin, the Ethiopian headwaters and the Tekezze–Atbara river basin, and the main or Saharan or Desert Nile (see Fig. 1.1). The Blue Nile rises in the volcanic uplands of Ethiopia (Chapter 6) and flows through a gorge nearly 2 km deep and 35 km wide, from which it has eroded roughly 70,000 km3 of rock from a catchment area above an elevation of 1,000 m of 190,000 km2, before it emerges from the highlands onto a vast alluvial fan known as the Gezira (Chapter 11).

Figure 3.6 Saharan uplands. Note that the Jebel Marra volcanic massif is situated close to the geographical heart of North Africa and lies roughly 1,500 km from the Atlantic and Mediterranean coasts. The A is Adrar Bous, a ring-complex east of the northern Aïr, which was occupied intermittently over at least the last 0.5 Ma, from Early Stone Ages times onwards.

28

Climate and Hydrology

It then flows across this fan through the semi-arid plains of the central Sudan to join the White Nile at Khartoum. The Tekezze headwaters of the Atbara river basin above 1,000 m elevation cover an area of 85,000 km2 and have eroded over 30,000 km3 of rock during the past 30 Ma. In strong contrast to the highly seasonal Blue Nile and Atbara Rivers, the White Nile emerges from the equatorial lake plateau of Uganda (Chapter 7) and disappears into the extensive swamps of South Sudan (Chapter 8), from where it emerges as a river of nearly constant flow throughout the year, but with a greatly diminished sediment load amounting to only a few million tonnes of silt and clay carried to the main Nile each year. The main Nile begins its long journey to the sea at the confluence of the Blue and White Nile Rivers in Khartoum. It then flows north for 320 km before it is joined from the east by the Atbara (Chapter 12), after which it flows north across the now waterless desert of northern Sudan and Egypt (Chapters 14–16) to reach the Mediterranean 2,689 km below its junction with the Atbara. A number of major wadi systems join the main Nile from uplands to the west of the Nile such as the now defunct Wadi Muqadam, Wadi Howar and Wad el Melik drainage basins (Chapter 15), as well as from the uplands to the east (Chapter 18).

3.3.1 Present-Day Hydrology The Nile Basin covers a total area of 3,310,000 km2, of which roughly a third (1,070,000 km2) is presently devoid of perennial rivers. As noted in Chapter 1, three major tributary basins supply water and sediment to the main Nile: the White Nile, the Blue Nile and the Atbara River. Of these, the White Nile drainage basin covers 1,730,000 km2 or just over half (52%) of the total Nile Basin area. Despite its vast catchment area, the White Nile provides only 28 ± 3 km3 (32%) of the total mean annual Nile discharge of 88 ± 5 km3 and an even smaller contribution (3 ± 2%), amounting to about 7 million tonnes (Mt) of the Nile’s total annual sediment load of 230 ± 20 Mt (Garzanti et al., 2006, 2015; Woodward et al., 2015a, Table 1). The Blue Nile joins the White Nile at Khartoum to form the main Nile. Although the Blue Nile has a relatively small total catchment area of only 330,000 km2, its annual discharge amounts to 48 ± 10 km3, or 54% of the total Nile discharge, and its sediment load amounts to 140 ± 20 Mt, or 61% of the total Nile sediment load. The Atbara River, which joins the main Nile 320 km downstream of Khartoum, occupies a drainage basin with a total area of 180,000 km2, and contributes 12 ± 5 km3 (14%) of Nile discharge and 82 ± 10 Mt (36%) of the Nile’s sediment load. There is some complexity concealed beneath these average figures. In particular, we should be aware that the White Nile provides 83% of Nile discharge during the month of lowest flow and so is responsible for maintaining perennial flow in the Nile during drought years in Ethiopia (Hurst, 1952; Williams et al., 1982). The two major Ethiopian tributaries of the Nile (the Blue Nile and the Tekezze–Atbara River) provide, respectively, 68% and 22% of the peak flow (Williams et al., 1982) and, as noted earlier, 61% (140 ± 20 Mt) and 36% (82 ± 10 Mt) of the total annual sediment load of 230 ± 20 Mt (Garzanti et al., 2006,

3.3 Nile Hydrology and Nile Floods

29

2015). North of Khartoum the main Nile flows through the eastern Sahara Desert north into the Mediterranean Sea and receives no further perennial inflow downstream of the Atbara confluence.

3.3.2 Late Quaternary Nile Floods and Droughts Marine sediment cores collected from the floor of the eastern Mediterranean show a repetitive depositional sequence of alternating dark sediments rich in organic matter, known as sapropels, and calcareous muds with a significant content of Saharan wind-blown dust (Rossignol-Strick et al., 1982). The sapropel units are thought to reflect accumulation in anoxic bottom waters during times of enhanced freshwater flow into the Mediterranean from now inactive Saharan rivers and from the Nile (Freydier et al., 2001; Scrivner et al., 2004). The influence of the summer monsoon over northern Africa was apparently stronger, and Nile floods more extreme, during intervals of sapropel accumulation (Rossignol-Strick, 1985, 1999; Larrasoaña et al., 2003; Revel et al., 2010, 2015), and there is some limited evidence of enhanced winter rainfall over northern Africa during certain phases of sapropel accumulation. Many of these inferences about former Nile floods are based on indirect and often very circumstantial evidence, as are the inferences about variations in winter rainfall and in the summer monsoon regime. There is therefore a need to test the marine depositional models and derived climate models (Claussen et al., 1998, 1999; Chylek et al., 2001) against a well-dated set of terrestrial archives capable of providing independent insights into the climatic changes in the northeast quadrant of Africa. In an attempt to achieve this goal, several teams of earth scientists have worked to obtain high-resolution records of late Quaternary hydrological and climatic changes from fluvial, eolian and lacustrine sedimentary sequences and landforms in the Nile Basin (Williams et al., 2000, 2003, 2006b; 2010, 2015a, 2015b; Woodward et al., 2001, 2007, 2015a, 2015b; Welsby et al., 2002; Williams, 2009a, 2012a, 2012b, 2013; Macklin et al., 2012, 2013, 2015). We discuss these newly acquired data in Chapters 8, 11 and 14. Evidence from the Nile Basin also contributes to current debates on the environmental impact of extreme climatic events (Cullen et al., 2000; Weiss, 2000). For example, we know that the Younger Dryas cold climate event that prevailed in northern Europe and Greenland between 12.7 and 11.5 ka had a major impact on tropical African climate (Swezey, 2001; Gasse et al., 2008). The period of low Nile flow that caused the demise of the Old Kingdom in Egypt 4,200 years ago may likewise have been caused by an abrupt climatic event (Krom et al., 2002). Were a similar disruption to White Nile flow to occur today, even for a decade, the effects would be catastrophic. There is some suggestion in the White Nile flood record covering the past 15,000 years of a 1,500-year periodicity (Williams and Nottage, 2006; Williams, 2009a). Prasad et al. (2004) found a 1,500-year cycle in the laminations of Pleistocene Lake Lisan in the Dead Sea Rift. Turney et al. (2004) carried out spectral analysis of high-resolution peat humification data from Lynch’s Crater in tropical northeast Australia and concluded that there was

30

Climate and Hydrology

a 1,490- year cycle evident in the peat record. Finally, Moy et al. (2002) found a 1,470-year cyclicity in ENSO events in Holocene lake sediments in southern Ecuador. Taken together, these data point to a global climatic control over flood events in the Nile Basin.

3.3.3 Holocene Flow from Local Desert Wadis A series of now dry river channels and former lakes are scattered west of the main Nile and lower White Nile in what is today the arid eastern Sahara (see Chapters 8, 14, 15 and 16). Analysis of the sediments, subfossil aquatic snail shells, and stable oxygen and carbon isotopic composition of subfossil aquatic shells has demonstrated that conditions in this region were significantly wetter between about 10 ka and 7.5 ka (Williams et al., 1974, 2015b; Ayliffe et al. 1996; Williams and Jacobsen, 2011). Studies of fossil pollen grains in lake deposits at Oyo in the far northwest of the Sudan (Ritchie et al., 1985), and of lacustrine and fluviatile deposits in northern Sudan (Pachur and Kröpelin 1987; Ritchie and Haynes, 1987; Pachur and Hoelzmann 1991) and at Selima Oasis (Haynes et al., 1989), have confirmed the widespread nature of this Early to Middle Holocene wet phase during which there was widespread human occupation of what is now desert (Wendorf and Schild, 1976; Wendorf et al., 1993; Haynes, 2001; Schild and Wendorf, 2001; Nicholl 2001, 2003; Linstädter and Kröpelin, 2004; Smith et al., 2004; Hoelzmann et al., 2004; Gasse and Roberts, 2004; Kuper and Kröpelin, 2006; Williams et al., 2010). These moist intervals also coincide with times of high flood levels on the lower Blue and White Nile Rivers in central Sudan (Williams, 2009a; Williams et al., 2015b), which is to be expected, for the summer monsoon controls the amount of water flowing down the Nile as well as summer rainfall in the Sudan.

3.3.4 Historic and Recent Nile Floods As the Nile has the longest well-documented annual flood record (~1,000 years) of any river on Earth (Hassan, 1981), it can offer valuable high-resolution information about the annual incidence of past floods and droughts that complements the much shorter historic records available from big rivers elsewhere in the world. Precise information relating to times of high and low Nile flow has been derived from studying the archival record of Nile floods engraved upon the various stone flood gauges or ‘Nilometers’ which show the maximum level attained by consecutive annual floods. Analysis of the Nile floods engraved in the Roda gauge for the period AD 622–1470 shows some well-defined cycles of high and low flows (Fraedrich et al., 1997; De Putter et al., 1998). Flow in the Nile was often low between AD 950 and 1250, coinciding in time with the Mediaeval Warm Period (MWP) in Europe. A five-year periodicity is also apparent during intervals when the Nile flows were both higher and lower, and most likely reflects the influence of ENSO events, which are now well documented in the Nile records (Whetton et al., 1990, 1996; Whetton and Rutherfurd, 1996; Ortlieb, 2004). Fraedrich et al. (1997) and De Putter et al. (1998)

3.3 Nile Hydrology and Nile Floods

31

identified eight episodes characterised by abrupt changes in lowest and highest Nile flood levels, which appeared to show a 35- to 45-year periodicity. De Putter et al. (1998) also discerned a 75.9-year cycle from records of high flood levels during AD 950–1250. This may be similar to the roughly 90-year periodicity evident in the Lake Lisan laminations in the Dead Sea Rift (Prasad et al, 2004). The increase in frequency of high Nile floods after about AD 1250 may itself be linked to the increase in the strength of the Southwest Asian monsoon inferred from records of Globigerina bulloides collected from marine sediment cores from the Arabian Sea (Anderson et al., 2002). The causes behind each of these cycles remain obscure. The synoptic conditions associated with more recent floods and droughts in the Nile Valley have been investigated in some detail. Consider, for instance, the year 1999, which was an unusually wet year in the Nile Basin, with very high flow in the Nile and its tributaries, as well as severe and widespread flooding in China, India, Indonesia and much of Australia. All of these regions are subject to the influence of ENSO events, but other factors also intervened. The year 1999 was a La Niña year (Nicholson and Selato, 2000), with very positive values of the SOI. The July rains came early in central Sudan and were very heavy, persisting during August and September, causing severe floods across the Gezira Irrigation Area bounded by the Blue and White Nile Rivers, leading to a major canal breaching its banks and overflowing across the cultivated fields. Further west, where a series of dunes and swales occupy 160,000 ha between Hashaba and Jebel Aulia along the White Nile east bank (see Chapters 8 and 11), the depressions between the dunes filled with water and remained wet for many weeks, similar in many respects to the Early Holocene environments in this region that provided the Mesolithic hunting, gathering and fishing people with their livelihood (Williams and Nottage, 2006; Williams et al., 2015b). On 16 August 1999, Khartoum received as much rain in six hours as it usually received in a single year. These floods were the result of intense and highly localised downpours, unlike the 1998 floods, when prolonged heavy rain in the Ethiopian Highlands led to above average flow in the Blue Nile and Atbara Rivers, with flooding downstream in central and northern Sudan. The 1999 floods need to be seen in context. The mean annual rainfall at Khartoum amounted to 164 mm (Sudan Meteorological Service, no date) during the relatively wet thirty-year period 1931–60, and in the fifty-seven-year period 1900–57 it ranged from 48 mm to 380 mm, but only in five years did it exceed 250 mm (Rzóska, 1961). Hulme and Trilsbach (1989) concluded that the 200-mm storm of 4–5 August 1988 was ‘unprecedented in its magnitude’ and estimated that it had a return period of between 400 and 700 years. The storm of 16 August 1999 was equally exceptional. The synoptic conditions associated with the extreme rainfall events in 1999 are reasonably well understood (UNOCHA 1999a, 1999b) and are akin to those observed in other unusually wet years. The synoptic patterns in such years are usually associated with a warm equatorial Indian Ocean, a strong summer monsoon over both Africa and India, a northward shift of the ITCZ earlier and further north than usual, and the presence of deep, well developed westerlies accompanied by a strong Tropical Easterly Jet that allows more

32

Climate and Hydrology

moisture transport into Africa from the South Atlantic via the Congo Basin, leading to very heavy precipitation in the Ethiopian uplands and central Sudan (Camberlin, 1997; Camberlin et al., 2001; Mo et al., 2001; Osman and Shamseldin, 2002; Vizy and Cook, 2003; Williams and Nottage, 2006).

3.4 Conclusion The Nile Basin extends across a wide range of climatic zones. In the far south the equatorial highlands of Uganda and Burundi are wet for much of the year while in the north the Nile flows through a hot and waterless subtropical desert. Between these two extremes the climate is seasonally wet and tropical, with the onset of the northern or boreal summer wet season coinciding with the northward passage of the ITCZ. Much of this vast region becomes dry once more during the boreal winter, during which the ITCZ migrates south of the equator, bringing renewed rain to the great lakes that feed into the Ugandan headwaters of the White Nile. The northern coastal fringes of the Nile Basin receive rare and sporadic rain during the winter, when the westerly air masses in the Mediterranean have shifted further south. On occasion in the past they reached hundreds of kilometres inland, enabling plants, animals and humans to occupy now barren regions of the Red Sea Hills east of the Nile and the Gilf Kebir west of the Nile. The highly seasonal flood regime of the Nile was re-established during the terminal Pleistocene some 14.5 ka ago, when, after a long arid interval, Lakes Albert and Victoria in Uganda overflowed once more into the White Nile and Lake Tana in the Ethiopian Highlands overflowed again into the Blue Nile. The hydrologic history of the Nile is one of repeated wet and dry phases and times of high and low flow, at time scales ranging from millennial to centennial to decadal. Human groups have occupied the Nile Basin for well over a million years, and have had to adapt to these climatic vicissitudes and hydrologic fluctuations, either by altering their social, economic and technical attributes, sometimes involving modifying their environment to harvest and use Nile water more effectively, or by moving to more habitable areas. Later chapters expand on these themes.

4 Geology and Soils

I have noticed, too . . . that the soil of Egypt does not resemble that of the neighbouring country of Arabia, or of Libya, or even of Syria . . . but is black and friable as one would expect of an alluvial soil formed of the silt brought down by the river from Ethiopia. The soil of Libya is, as we know, reddish and sandy, while in Arabia and Syria it has a larger proportion of stone and clay. Herodotus (ca. 485–425 BC), The Histories (1954, p. 106)

4.1 Introduction This chapter provides a brief outline of the geology, geomorphology and soils of the Nile Basin. We pay special attention to the scope and limitations of using evidence from Quaternary geology, landforms and soils to reconstruct prehistoric environments and changes in local and regional climate. Table 4.1 is a summary of the types of evidence from the Nile Basin that have been used to reconstruct prehistoric environments as well as to determine past fluctuations in Nile discharge and sediment transport and deposition. Williams (2014, chapters 7 to 17) provides a detailed evaluation of each of the types of evidence listed in Table 4.1. Implicit in this table is that the age of each proxy needs to be determined by at least one and preferably by two or more independent dating techniques. Williams (2014, chapter 6) provides a review of dating methods. The most common dating techniques relevant to the present volume are radiocarbon (14C) dating for organic samples younger than about 40–50,000 years; luminescence dating for sediments that contain quartz or feldspar, with a time range back to a few hundred thousand years and, in ideal conditions, back to nearly a million years; and uranium-series dating of calcareous tufas and travertines, with a possible age range of at least several hundred thousand years. Other dating methods that have been used in the Nile Basin include potassium–argon or argon–argon dating of certain crystals within volcanic rocks and tephra (ash) beds; palaeomagnetic dating of magnetic reversals in rocks or sediments that contain magnetic minerals; and amino acid racemisation dating of shell or wood samples.

33

34

Geology and Soils

Methods of relative rather than ‘absolute’ dating include the use of fossils, whether plants or animals, marine or continental, macroscopic or microscopic. On land pollen spectra have been widely used to reconstruct past changes in vegetation but are often rare or badly oxidised in arid environments (Lézine and Casanova, 1989; Lézine et al., 1990, 2011). Pollen preserved in marine sediments also offer useful records of changes in the vegetation on land (Leroy and Dupont, 1994, 1997). Fossil vertebrate remains that have been independently dated in other localities can provide a useful chronology but often only at a coarse resolution, which is still better than guessing (Vrba et al., 1995). Marine sediment cores are usually dated using changes in the presence or absence of certain microfossil species together with analysis of changes in the ratio of the stable isotopes of oxygen, often supplemented by carbon isotope analysis. Terrestrial pollen grains are also found in marine sediments such as in the Nile Cone. Quaternary marine sediments across the world’s oceans now provide a relatively fine resolution record of variations in global ice volume, deep and shallow ocean water temperature and more local changes in sea water salinity (Lisiecki and Raymo, 2005; Lisiecki and Raymo, 2007; Railsback et al., 2015), enabling individual marine isotope stages (MIS) to be placed in a numerical age sequence and related to past glacial–interglacial cycles and associated shorter climatic cycles. So, for example, MIS 1 extends from present to 14 ka, and MIS 2 from 14 ka to 29 ka (Table 4.2). The even numbers refer to the colder glacial or stadial climatic intervals and the uneven numbers to the warmer interglacial or interstadial climatic intervals (Gibbard and Cohen, 2008; Railsback et al., 2015).

Table 4.1 Evidence used to reconstruct prehistoric environments in the Nile Basin Proxy data source

Variable measured

Quaternary geology and geomorphology Soils and fossil soils

Soil type, particle size, soil chemistry; isotopic composition of soil carbonate Lakes and lake sediments Lake shore level(s); facies changes; mineralogical composition; geochemistry Eolian sediments: desert dust, dunes, sand plains Mineralogical composition; surface texture; geochemistry; isotopic composition Speleothems, tufas Stable isotopic composition; geochemistry Mountain glaciers Terminal positions Glacial deposits (e.g., moraines) and features of Equilibrium snowline of cirques glacial erosion (e.g., cirques) Periglacial features Distribution and elevation Biology Fossil pollen and spores; plant macrofossils and Type, relative abundance and/or absolute microfossils’ concentrations; distribution Vertebrate fossils; invertebrate fossils: mollusca; Type, assemblage, abundance; trace element ostracods; diatoms; insects geochemistry; stable isotopes

4.2 Geology

35

Table 4.1 (cont.) Proxy data source

Variable measured

Modern population distributions

Refugia: relict plant populations

Marine sediments Foraminifera; sapropels; other sediments

Abundance, assemblage, trace-element geochemistry, oxygen and carbon isotopic composition

Archaeology Written records; plant remains; animal remains, Stone tool assemblages (provenance, geochemistry, cultural affinities, distribution); associated including hominins; cemeteries; rock art; fauna and flora; skeletal remains (distribution, hearths, dwellings, workshops; artefacts: bone, morphology); teeth (wear patterns, isotopic stone, wood, shell, leather composition) Adapted from Williams (2014).

Table 4.2 Marine Isotope Stages (MIS) 1 to 20 (0–790 ka) MIS 1: 0–14 MIS 2: 14–29 MIS 3: 29–57 MIS 4: 57–71 MIS 5: 71–130

MIS 6: 130–191 MIS 7: 191–243 MIS 8: 243–300 MIS 9: 300–337 MIS 10: 337–374

MIS 11: 374–424 MIS 12: 424–478 MIS 13: 478–533 MIS 14: 533–563 MIS 15: 563–621

MIS 16: 621–659 MIS 17: 659–676 MIS 18: 676–712 MIS 19: 712–761 MIS 20: 761–790

Even MIS numbers are cold phases and uneven MIS numbers are warm phases. Compiled from data in Williams et al. (1998), Walker (2005), Lisiecki and Raymo (2005, 2007) and Gibbard and Cohen (2008).

4.2 Geology 4.2.1 The Precambrian Legacy We use the term Precambrian somewhat loosely to refer to rocks that are older than 540 Ma, after which complex living organisms became more widespread. The age of the Earth dates back to 4.6 billion (4.6 × 109) years ago, but we will confine our attention here only to rocks in and around what is now the Nile Basin and which date back to the last billion years or so, for which the evidence is modestly well dated and reasonably intensively studied (Vail, 1976, 1983, 1985; Asrat et al., 2001; Collins and Pisarevsky, 2005; Fritz et al., 2013), as is that of the sediments derived from these rocks (Padoan et al., 2011; Garzanti et al., 2015; Fielding et al., 2016, 2018). Early geologists working in Uganda, Ethiopia, Sudan and Egypt mapped and described a series of often highly eroded igneous and metamorphic rocks

36

Geology and Soils

Northern

contine nts

D

N

G

O

West African craton

Amazon craton

W

A

Sahara craton

N

A Azania

Aus t cratralian on

Equator

Indian craton Congo craton East Antarctica craton

Kalahari craton

Present continents

Craton

South pole

Figure 4.1 Gondwana at about 420 Ma ago. (Adapted from F. M. Williams, 2016, Fig. 6.1.)

overlain with marked unconformity by younger sedimentary rocks (Andrew, 1948; Whiteman, 1971; Vail, 1974, 1978; see also the recent review by F. M. Williams, 2016). Because these ancient rocks formed a foundation or basement beneath the younger relatively horizontal sedimentary formations, they became widely known as the basement rocks or the Basement Complex. Early workers in the 1960s inferred that the Basement Complex rocks had last been folded, faulted and metamorphosed during a major phase of mountain building termed the PanAfrican Orogeny of 550–500 Ma, during the assembly of Gondwana, a vast supercontinent whose amalgamation was completed by 420 Ma. Gondwana (Fig. 4.1) was made up of a series of smaller coalescent landmasses, which later split off at different times from the parent supercontinent to become Africa, South America, India, Australia and Antarctica (Fig. 4.2). Once the Basement Complex rocks had been more rigorously dated and more thoroughly analysed and mapped, it became clear that the concept of a Pan-African Orogeny was oversimplified and no longer useful. We now refer to the East African Orogen (EAO) for the zone of former mountain building that extended from Arabia and Egypt in the north to Mozambique and Madagascar in the south. The style and tempo of mountain building and metamorphism was quite different in different parts of the EAO, with accretion of former slivers of both continental crust and ocean crust (Fritz et al., 2013). For the sake of simplicity and convenience, geologists distinguish what is loosely termed the

4.2 Geology

r th

A

E (Eur urasia ope and Asia )

ca

ge

No

ri me

37

Rid

Equator

tlantic

Antarctica

Au str ali

a

A Mid -

South America

C ar l sberg Ridge

India

Africa

Spreading axis

Figure 4.2 Gondwana at about 60 Ma ago. (Adapted from F. M. Williams, 2016, Fig. 7.1.)

Arabian–Nubian Shield in the north and the Mozambique Belt (MB) in the south. The southern sector of the Nile Basin includes small but significant outcrops of MB rocks, and the northern two-thirds of the Nile Basin are underlain by the so-called Nubian Shield rocks (Fig. 4.3). The East African Orogeny lasted from about 850 Ma to about 550 Ma (Fritz et al., 2013) and consisted of a number of orogenic episodes and associated metamorphism, separated by several phases of erosion and intrusion of coarse-grained igneous rocks such as granite, which were in turn subject to intense heat and pressure and metamorphosed to gneisses and other rocks, depending on their original composition. Padoan et al. (2011) used variations in the strontium and neodymium isotopic composition of sediments in the Nile Basin to identify the bedrock sources of the alluvium. Fielding et al. (2016, 2018) expanded on this work by identifying the uranium–lead and hafnium isotope ages of zircons within Nile sediments throughout the Basin. In addition to defining in some detail the present-day sources of Nile alluvium, their work has also helped to refine our understanding of the different stages of development of the Arabian Nubian Shield (840–780 and 700–600 Ma), and the timing of the Gondwana supercontinent assembly (650–500 Ma). The rocks formed during the EAO are important for several reasons. They are host to many precious and semi-precious metals and minerals that have been sought after from prehistoric times onwards. In the more humid tropical regions of the southern Nile Basin the

20 oE

30°E

40°E Quaternary eolian sands

MEDITERRANEAN SEA

Cenozoic volcanic rocks Cenozoic sedimentary rocks

30o N

Mesozoic sedimentary rocks Precambrian igneous and metamorphic rocks

EGYPT

LIBYA

D

RE SE A

20°N CHAD

ER IT R

E

A

J

S SUDAN

10°N SOUTH ETHIOPIA CENTRAL AFRICAN REPUBLIC

SUDAN

UGANDA KENYA

SOMALIA

CONGO

0°N 250 km

500

DA AN RW

0

K TANZANIA

Figure 4.3 Geology of the Nile Basin. (Compiled by the author with data from Russell, 1962: 1: 4,000,000 Geology Map; The Atlas of Africa, 1973: Geology Map, p. 29; (Vail, 1978: 1:2,000,000 Geological Maps, North Sheet and South Sheet; Bureau de Recherches Géologiques et Minières, 1981: 1: 2,000,000 Geological Map of the Sudan; Vail, 1985; Robertson Research and Geological Research Authority of the Sudan, 1988; Geological Research Authority of the Sudan, 1995; Schlüter, 1997; Geological Research Authority of the Sudan, 2004: 1:2,000,000 Geological Map of the Sudan; Fritz et al., 2013; and F. M. Williams, 2016: Fig. 4.2.)

4.2 Geology

39

Basement Complex rocks have been weathered to form moderately fertile soils that support (or once supported) a luxuriant cover of forest and woodland. They have also played a major role in the geomorphic history of the Nile. Many faults, shear zones and other tectonic lineaments formed during the Precambrian to Cambrian East African Orogeny have been re-activated at intervals during the past 500 million years (Roden et al., 2011). Buried rift valleys in South Sudan are now filled with 10–15 km of sediments eroded from the mountains of southern Ethiopia that arose during the East African Orogeny. Ancient lineaments such as the Aswa wrench fault (see Chapter 13, Fig. 13.1) control the orientation of drainage in South Sudan. In northern Sudan and Egypt, the linear pattern of many Nile reaches is a direct reflection of the influence that these ancient structural controls exercise on Nile drainage patterns today, a topic we discussed in Chapter 2.

4.2.2 Northward Movement of the African Lithospheric Plate It is convenient to begin this account with the break-up of Gondwana, which led to the separation of India–Australia–Antarctica from Africa some 180 Ma ago, followed by the splitting away of South America from Africa about 130 Ma ago. Africa then began to drift slowly northwards towards Eurasia, following a clockwise trajectory during the past 75 Ma (Habicht, 1979; Smith et al., 1981). In northern Nigeria and the Aïr Mountains of Niger there is a series of mid-Palaeozoic igneous intrusions and ring complexes that are older in the north and younger in the south, suggesting that they may have formed over one or more hotspots as the African plate drifted slowly northwards (Bowden et al., 1976). Another series of ‘Younger Granite’ igneous intrusions and ring complexes are aligned diagonally from SW to NE from the Cameroun Mountains of West Africa across the southern Sahara through southern Libya, northern Sudan and northern Eritrea to the Red Sea Hills. Ring complexes consist of a central granite core flanked by a series of concentric igneous rocks ranging from dolerites to gabbros. Differential erosion of the less resistant rocks means that concentric ridges of resistant rocks rise above annular- or crescent-shaped valleys flanked by alluvial fans and ephemeral stream channels and sometimes host now dry lakes. Such localities were attractive to prehistoric humans, as shown by the very long record of prehistoric human occupation at Adrar Bous ring complex in the Ténéré Desert of Niger (see Chapter 3, Fig. 3.6), extending from Early Stone Age to proto-historic times (Clark et al., 2008a, 2008b, 2008c; Gifford-Gonzalez, 2008). The magnificent galleries of polychrome paintings of humans and herds of domestic cattle at Jebel ‘Uweinat ring complex, which straddles the border between Sudan, Egypt and Libya, are striking evidence of how important these mountains were to Neolithic desert pastoralists. Jebel ‘Uweinat also has a reliable supply of excellent fresh water in the form of several underground springs, and so has served as an occasional refuge for small bands of indigenous Tibu nomads early last century. Jebel Arkenu is a smaller ring complex 40 km northwest of ‘Uweinat, and was also visited by Neolithic and later herders who left behind some vivid

40

Geology and Soils

rock paintings (Williams and Hall, 1965). It is possible that some of the ring complexes in northern Sudan could also have been places of prehistoric refuge from the surrounding desert during times of somewhat wetter climate. The western sector of North Africa moved into the dry tropical latitudes of the Northern Hemisphere during the late Mesozoic and early Cenozoic, resulting in widespread formation of evaporite deposits in Tunisia and Algeria (Coque, 1962; Conrad, 1969), to be followed by Egypt and northern Sudan during Miocene times onwards. The inception of aridity in the Sahara stems from these times, and was interspersed by prolonged intervals when the prevailing climate was less arid, allowing Mediterranean plants to migrate south into the central and southern Sahara along long defunct waterways, and plants from the wet tropics to migrate northwards, where they now form relict populations in the mountain refugia of the Hoggar, Tibesti, Jebel Marra and the Aïr (see Chapter 3, Fig. 3.6) (Quézel and Martinez, 1958–59, 1962; Quézel, 1962; Beucher, 1971; Wickens, 1976a, 1976b; Ozenda, 1977; Maley, 1980, 2004, 2010).

4.2.3 Palaeozoic and Mesozoic Sedimentation During the long intervals following episodes of mountain building, such as from about 450 Ma onwards, much of North Africa and the region now occupied by the Nile Basin was subject to prolonged intervals of erosion and continental sedimentation, leading to widespread accumulation of sandstones such as the Mesozoic Nubian Sandstone Formation. These long intervals of erosion by former river systems were interspersed with occasional periods of intense cold and widespread glaciation, with the Cambro-Ordovician glaciations leaving a legacy of glacial sediments and glacially scoured rock pavements across Gondwana. The Ordovician glacial pavements in the Hoggar region of the central Sahara have only recently been exhumed from their sedimentary cover and are so well preserved that they resemble the most recent glacial landscapes of the late Quaternary (Beuf et al., 1971). Not all parts of the region underwent erosion at the same time. On occasion the oceans flooded some of the areas at lower elevations or which were subject to sustained tectonic subsidence. Much of the Sahara and the Horn of Africa, especially Somalia and the Ogaden region of Ethiopia (F. M. Williams, 2016) lay beneath the sea at intervals during the Cretaceous (Faure, 1962a; Furon, 1963). Marine Cretaceous sediments are found today at elevations approaching 2,000 m in the northern Aïr massif, indicating faulting and uplift after the seas receded (Faure, 1959, 1962b).

4.2.4 Cenozoic Tectonism, Volcanism and Sedimentation The Cenozoic Era spans the last 65 million years of geological time. It marks the proliferation of land mammals following the demise of the dinosaurs at the end of the Mesozoic Era. As the African lithospheric plate moved north and began to impinge on Eurasia, portions of the crust became deformed and uplifted during the Hercynian Orogeny, which gave rise to

4.2 Geology

41

the Atlas Mountains and the Alps. Portions of the previously stable Saharan shield also experienced uplift and associated volcanic activity, as in Jebel Marra in western Sudan and the Hoggar and Tibesti in the central Sahara. During the Palaeocene and Eocene great tracts of what is now the Sahara were forested and subject to a hot wet equatorial climatic regime. Miocene uplift and climatic desiccation led to erosion of the deep weathering mantle and exhumation of the irregular weathering front (Dresch, 1959; Greigert and Pougnet, 1967; Thorp, 1969; Williams, 1984). The origin of the Sahara as a desert probably stems from about this time (Williams et al., 1987; Williams, 2014). However, it would be misleading to think that the Miocene Sahara was as arid as it is today. During the late Miocene, a series of big rivers flowed from what is now the Chad Basin northwards, carving wide drainage channels in the southern Libyan Desert between Tibesti volcano to the west and three Nubian Sandstone plateaux located just to the east. Griffin (1999, 2002, 2006, 2011) called these the Sahabi Rivers and inferred that during the time they flowed north across the present Libyan Desert to the southern coast of the Mediterranean, the climate in their upper catchment was relatively humid. This conclusion is consistent with the geochemical, faunal and floral evidence reported by Moussa et al. (2016) of a mosaic of wetlands, savanna grasslands and woodlands in the Chad Basin during the late Miocene. The late Miocene Mediterranean was a salt desert during what has been called the Messinian Salinity Crisis (Hsü et al., 1977; Hsü, 1983) dated to between 5.96 Ma and 5.33 Ma (Cosentino et al., 2013). Desiccation of the Mediterranean may have been a consequence of tectonic movements but seems more likely to have been caused by fluctuating sea levels associated with the waxing and waning of ice caps in West Antarctica (Williams et al., 1998; Williams, 2014). During times of lower glacial sea levels, the shallow sill at the western end of the Mediterranean (the Straits of Gibraltar) would act as a dam and prevent free flow of Atlantic water into the Mediterranean Basin, and corresponding outflow of Mediterranean waters to the Atlantic. Given present rates of evaporation across the Mediterranean, a few short centuries would suffice to dry out much of the sea. Successive inflow and desiccation resulted in the accumulation of halite and other evaporites on the floor of the basin, eventually leading to the precipitation of a layer of salt up to about 1 km thick on the floor of the Mediterranean Basin. Repeated drying out of the Mediterranean Sea implies repeated lowering of river base level with concomitant deep fluvial entrenchment and, in the case of the Nile, formation of the deep fluvial canyon described in Chapter 2. It is likely that other big rivers also cut deep canyons at this time. Tectonic movements were far from synchronous across the Nile Basin and adjoining areas. In Ethiopia uplift and associated widespread volcanic activity during the last 30 Ma seems to be related to the presence of a hotspot or intensely hot mantle plume beneath what is now Arabia and Ethiopia. Uplift led to crustal fracturing. Faults developed progressively along what is now the Red Sea, perhaps as early as 25 Ma, with sea-floor spreading commencing some 5 Ma ago. The Ethiopian and Afar Rifts began to open 15–10 Ma ago, but did not acquire their present form until during the late Pliocene and early Pleistocene. Uplift of what became the Ethiopian Highlands (and headwaters of the Blue

42

Geology and Soils

Nile and Atbara Rivers) was accompanied by episodic subsidence in the Afar Rift to the east and in the lowlands of South Sudan in the west. Both of these localities became significant depocentres or zones of sediment accumulation, with 10–15 km of oil-bearing sediments accumulating in South Sudan beneath what is now the White Nile valley. In the arid Afar Rift there are sporadic sedimentary remains of once large lakes, as well as the fossil-bearing Pliocene and Quaternary deposits with their well-known hominin fossils. Volcanic activity has persisted until the present day and, together with sporadic earthquakes, remains as much of a potential hazard for the modern inhabitants of Ethiopia as it must have been for their prehistoric forebears. There are also scattered volcanic hills in the arid deserts of Nubia and the Butana. The columnar basalts often appear very fresh and angular – perhaps because of minimal chemical weathering in this arid environment, but perhaps also because the eruptions were quite recent. On the slopes and summit of one such basalt hill rising above the dissected Nubian Sandstone tablelands 50 km east of Kerma on the Nile right bank in northern Sudan there are scattered Late Stone Age/Mesolithic and younger stone flakes. The previously horizontal Nubian Sandstone beds through which the basaltic lavas were extruded have been tilted around the margins of the volcano to angles of 15–20°. Sandford (1933) had earlier described explosion craters and the denuded stumps of volcanoes southwest and northeast of Jebel ‘Uweinat on the border between Sudan, Egypt and Libya, noting that ‘both types were encircled by steeply tilted, hardened, and locally fused Nubian sandstone’ (Sandford, 1933, p. 47). Despite their very fresh appearance he concluded that ‘these explosion craters seem to be far older than they at first appear’ (Sandford, 1933, p. 49) but offered no suggestion as to their age. He commented briefly on the basaltic lavas and ‘volcanoes’ in the Bayuda desert between Berber and Merowe, which likewise intrude through the Nubian Sandstone and so are younger than Upper Cretaceous age.

4.2.5 Quaternary Climatic Fluctuations and Long-Term Desiccation The Quaternary Period began 2.58 Ma ago and continues to this day. It consists of two Epochs of unequal duration: the Pleistocene (2.58 Ma to 11.7 ka) and the Holocene (11.7 ka to present) (Cohen et al., 2013). There is much talk of defining a new Epoch – the Anthropocene – starting at a yet to be decided date, but most probably about the middle of last century (Waters et al., 2016). For present purposes, we will consider that we are still in the Holocene. The Quaternary is characterised by rapid fluctuations in global temperature and global ice volume, reflected in multiple glacial–interglacial cycles. These cycles were modulated by three astronomical cycles (Fig. 4.4), all of which control the amount of solar radiation received in the upper atmosphere, which in turn controls the temperature at the Earth’s surface (Williams et al., 1998; Williams, 2014). The relative influence of these three cycles on global climate has varied through time and may be discerned from the record of marine

4.2 Geology

43

Figure 4.4 Orbital fluctuations. (After Williams, 2014: Fig. 3.6.)

sediment cores. It is useful to discuss each in turn since they have all exerted an important influence upon the Nile Basin. Earth follows an elliptical path around the sun each year. At present, when Earth is closest to the Sun (termed the perihelion), the distance between Earth and Sun is 147.1 million km. When farthest from the sun (termed the aphelion) the distance is 152.1 million km. The shape of this elliptical path changes over time, being sometimes more and sometimes less elliptical. This cyclical change is known as orbital eccentricity, lasts 96,600 years, and leads to a 3.5% variation in solar radiation received in the outer atmosphere. The tilt of Earth’s axis – at present 23° 27 0 – also varies, from a maximum tilt of 24° 36 0 to a minimum tilt of 21° 59 0 . When the inclination is greatest, summers in higher latitudes tend to be hotter and winters colder. Times of minimum tilt are times of mild winters and mild summers. This obliquity cycle lasts 41,000 years and exerts a powerful control over seasonality. The third cycle is the precessional cycle and over the last million years may have varied in length from 16.3 ka to 25.8 ka, but over the past 10 million years is generally considered to have varied between 23 ka and 19 ka. This cycle reflects the changing season of the year when Earth is closest to the Sun and is controlled by the direction in which the spin axis of Earth points in space. In the late Pliocene, the dominant cycles evident in marine sediment cores are the 19 ka and 23 ka precessional cycles. The 41 ka obliquity cycle then became dominant until about 0.7 Ma, after which the 100 ka orbital eccentricity cycle became dominant (Lisiecki and

44

Geology and Soils

Raymo, 2005; Clark et al., 2006; Lisiecki and Raymo, 2007). In terms of direct impact on climate, the last 0.7 Ma were subject to high-amplitude but low-frequency fluctuations, while the preceding time was subject to lower amplitude but higher frequency fluctuations. In short, glacial–interglacial cycles lasted about 100 ka after about 0.7 Ma ago, but were shorter and less severe before then. During the last 0.7 Ma, each glacial cycle was characterised by a long, slow, saw-toothed build-up to extreme cold, maximum ice volume and minimum sea level, followed by a short, rapid warming and a brief interglacial interval of warmer and often wetter climate within the intertropical zone. Our present interglacial effectively began at the start of the Holocene 11,700 years ago, although there were brief intervals of warmer global climate in the preceding few thousand years before then. The Quaternary was characterised by the accumulation of large ice caps over North America and NW Europe roughly 2.6 Ma ago, followed by the waxing and waning of these ice caps (Williams et al., 1998). The repercussions included global sea level fluctuating in relation to how much water had been abstracted from the oceans to feed the growing ice caps, and how much glacial melt-water was released into the oceans once the ice melted. As one might expect, the glacial intervals were times when temperatures across the continents were generally colder than today, as were the sea surface temperatures. Lower sea surface temperatures were associated with reduced evaporation from the tropical oceans, leading to a reduction in precipitation over land, at least within the intertropical regions (Galloway, 1965a; Williams, 2014). Tropical lakes dwindled in size, some becoming more saline or drying out altogether (Tillet, 1997; Gasse et al., 2008; Williams, 2014). Hitherto perennial rivers became more seasonal or even ephemeral. Reduced precipitation and lower glacial temperatures caused the tropical rain forests to shrink and become fragmented (Flenley, 1979; Anhuf et al., 2006), with corresponding habitat fragmentation and isolation of some of the less mobile mammals, birds and insects (Moreau, 1963). In the seasonally wet tropics once vegetated and stable dunes became bare and mobile, as along the southern margin of the Sahara (Grove, 1958; Grove and Warren, 1968; Talbot, 1980). With stronger and gustier winds and a reduced plant cover it was easier to mobilise and transport fine soil and sediment particles, and dust storms were also more frequent (McGee et al., 2010). With a return to warmer wetter conditions, there was a corresponding expansion of the tropical and montane rain forest and a stabilisation of previously active sand plains and dune fields. Lake levels rose once more and rivers became less seasonal. Plants and animals expanded their habitat into previously arid or semi-arid areas, and parts of the southern Sahara became covered in savanna woodland and grassland, most recently during the first half of the Holocene (Williams, 2008). During the past 5,000 years conditions became arid once more. Many small lakes that were scattered across the Sahara dried out completely, as regional water tables sank well below the surface, and once perennial or seasonal streams and rivers ceased to flow (Hoelzmann et al., 2004; Williams, 2014). The former river channels in the eastern Sahara detected by side-scanning radar are good examples of such now defunct drainage systems (McCauley et al., 1982, 1986; Breed et al., 1987). The oldest of these may date back to Oligocene time while the youngest were still active during the

4.2 Geology

45

Lower Palaeolithic/Early Stone Age, being associated with Acheulian hand axes and other bifacial tools (McCauley et al., 1986). The seasonal dust storms such as the haboob storms in the Nile Valley and the Harmattan dust storms in West Africa became more active, the latter attracting the attention of Dobson (1781) and the young Charles Darwin (1846) as well as many later workers (McTainsh, 1980, 1984, 1987; McTainsh and Walker, 1982; Washington et al., 2006). Active sand dunes and sand plains cover about one fifth of the Sahara today but during drier climatic phases dunes were active more than 500 km south of their present limits in West Africa and the Sudan (Fig. 4.5) (Grove and Warren, 1968; Talbot, 1980; Swezey, 2001). Wind-blown sands are present in many of the Neogene well sections examined by Michel Servant in the Chad Basin (Servant, 1973; Servant and Servant-Vildary, 1980). There is a close association between former rivers, sand dunes and closed sedimentary basins in North Africa (Williams, 2015), and the same is true of the Nile Basin. Quaternary and older rivers flowing from the upland drainage divides between topographic depressions ferried gravel, sand and finer particles across the hill slopes and down to the valley floors. The finer material (silt and clay) provided the source material for desert dust, much of which was blown out to sea (Morales, 1979; Schütz et al., 1981) or even as far afield as the Amazon Basin (Swap et al., 1992; Prenni et al., 2009). The resistant quartz sand grains at the distal end of desert river channels were soon reworked by wind and in due course formed desert dunes of great symmetry and multiple forms (Cooke et al., 1993; Warren, 2013). There is always a type of geomorphic tug-of-war between wind and water in arid areas. Sometimes the rivers flow strongly and erode the dunes. At other times wind action gains the upper hand and former stream channels are buried beneath wind-blown sand. Given enough time, wind-blown sand will gradually obliterate former lakes and ephemeral stream channels, as in the case of the concealed river channels revealed by side-scanning radar to the west of the Egyptian Nile (Breed et al., 1987). Both the White Nile and the main Nile show signs of temporary channel blockage by wind-blown sand during drier intervals in the late Quaternary (Butzer and Hansen, 1968; Williams et al., 2010). Shifting wind belts during the late Quaternary fashioned linear dune systems, which today show slightly different orientations and degrees of soil development, as is clearly evident in the Qoz west of the lower White Nile in Kordofan Province (Warren, 1970). (Qoz is a general Arabic term for sand dune or sand plain). Some of the linear dunes west of the Nile are many hundreds of km long, and the remarkable Qoz Abu Dulu dune extends N–S for more than 500 km and, in common with many dunes in this area, appears to be a polygenic dune (Williams et al., 2010, 2015a). In contrast to North America and NW Europe, Quaternary glaciations were never a major factor in fashioning the landscapes of the Nile Basin. There were small ice caps and valley glaciers in the Ethiopian Highlands, notably in the Bale Mountains of southern Ethiopia and in the Semien Mountains close to the headwaters of the Blue Nile and Atbara Rivers. In Uganda, the glaciers that are still present on Mount Ruwenzori expanded and Mount Kivu (Livingstone, 1975, 1980) was also glaciated during what is called the Last Glacial Maximum (LGM) some 21,000 ± 2,000 years ago (Mix et al., 2001). In the Sinai Desert of

Figure 4.5 Map showing the limits of active and fixed dunes in and beyond the Sahara. The present-day limit of active dunes is bounded by the 150-mm isohyet. Fixed dunes extend up to 500 km south of the Sahara, locally into areas that now receive 1000 mm of mean annual rainfall. (Modified from Vail, 1978; Grove, 1980; Mainguet et al., 1980; and Talbot, 1980.)

4.3 Soils

47

Egypt at least one of the high mountains shows signs of having been glaciated in geologically recent times (Messerli and Winiger, 1980) and the high Atlas Mountains in northwest Africa had a more extensive cover of snow and ice during the LGM (Messerli, 1972; Messerli et al., 1980; Messerli and Winiger, 1992). The Hoggar Mountains in the central Sahara also show signs of several phases of late Quaternary glacial erosion and deposition (Rognon, 1967). It seems unlikely that Jebel Marra near the western edge of the Nile Basin was ever covered in ice, but frost shattering during times when temperatures were up to 8–12°C lower than today may have been quite widespread across the more arid parts of the Nile Basin. Frost shattering was probably responsible for some of the highly angular blocky rubble seen by the author on the summit of a small basalt volcano located at 19°42.101 0 N and 30°40.379 0 E in the desert of northern Sudan. Frost shattering and downslope movement of angular rubble were certainly active in the Semien Mountains during the LGM, as discussed in Chapter 7.

4.3 Soils Figure 4.6 shows the main types of soil found within the Nile Basin and Table 4.3 lists their main characteristics. A soil may be defined very simply as weathered rock or sediment that has been altered from its original state as a result of biological processes resulting from plant, animal and microbial activity. Soils can be discontinuous or can form more or less continuous mantles across the landscape. Soil scientists have traditionally identified five main factors that are responsible for soil formation: parent material, topography, biological activity, climate and time (Jenny, 1941; Kubiëna, 1950; United States Department of Agriculture, 1951; Mohr and Van Baren, 1959; Mohr et al., 1972; Duchaufour, 1978; Paton, 1978; Paton et al., 1995; Birkeland, 1999; Retallack, 2001; Williams, 2014). Parent material and topography have the greatest influence on soil characteristics during the early stages of soil development. Paton (1978, p. 96) termed these two factors the

Table 4.3 Major soil groups in the Nile Basin Soil group

Characteristics

Andosols Arenosols Ferralsols Fluvisols Lithosols Regosols Vertisols

Soils formed in volcanic ash, with abundant volcanic glass (Andisols) Sandy soils with minimal texture contrast and weak or no soil horizons (Entisols) Soils rich in iron or aluminium (Ultisols) Young soils in alluvial deposits still showing signs of alluvial stratification (Entisols) Shallow soils, composed mainly of unweathered rock fragments Soils formed on deep unconsolidated recently deposited sands or alluvium (Entisols) Heavy dark churning clay soils with deep vertical cracking in the dry season. Contain abundant swelling clay minerals (notably, montmorillonite). Variable salinity and alkalinity (Vertisols)

Based on author’s observations with names adapted from World Reference Base for Soil Resources (2010).

48

Geology and Soils 20 oE

30°E

40°E Arenosols Vertisols Andosols

30o N

Fluvisols Vertisols and Entisols Regosols

EGYPT

LIBYA

Fluvisols Ferrasols and Regosols

20°N CHAD

ER IT R

E

A

J

S SUDAN

10°N SOUTH ETHIOPIA CENTRAL AFRICAN REPUBLIC

SUDAN

UGANDA KENYA

SOMALIA

CONGO

0°N 250 km

500

A AND RW

0

K TANZANIA

Figure 4.6 Soils of the Nile Basin. (Compiled by the author with data from Greene, 1948; Russell, 1962: 1: 4,000,000 Soils Map; Mitchell, 1984, Fig. 4.3; Buursink, 1971; Blockhuis, 1993; Nile Basin Initiative, 2012; and author’s unpublished observations.)

4.3 Soils

49

‘passive factors’ of soil formation. As time progresses, climate and biological activity (Paton’s ‘active factors’) exert an increasingly important influence on soil development. Climate refers to soil climate, which is ultimately controlled by the local climatic regime, and includes both the soil moisture regime and the soil temperature and evaporation regimes. Soil moisture is controlled by soil permeability and by the soil infiltration capacity. Different rock types, whether igneous, metamorphic or sedimentary rocks, exert quite different influences on the physical and chemical characteristics of immature soils. As a result, it is possible to distinguish quite easily between young soils developed on the most common types of rock such as granite, basalt, sandstone, siltstone and limestone, discussed in Section 4.3.1. A long-acknowledged and distinctive feature of all soils worldwide is that they show some degree of vertical differentiation and are arranged in roughly horizontal or gently sloping soil layers traditionally known as soil horizons (Joffe, 1949; Kubiëna, 1950; United States Department of Agriculture, 1951; Mohr and van Baren, 1959; Soil Survey Staff, 1960; Mohr et al., 1972; Buol et al., 1973; Duchaufour, 1978; Paton et al., 1995; Birkeland, 1999; Retallack, 2001; Soil Survey Staff, 2010; Williams, 2014). The near-surface soil layers tend to be darker in colour because more enriched in organic matter derived from the plants growing in the soil. This topsoil is termed the ‘A horizon’ and will become thicker and darker over time as the plant cover becomes well established. Beneath the ‘A horizon’ is a layer of soil usually with a higher clay content than the topsoil and usually with brighter colours if the site is well drained; this layer is called the ‘B horizon’. Beneath the ‘B horizon’ is the parent sediment or ‘C horizon’, which is generally altered from its initial state but still retains signs of its depositional origin. If the soil is developed directly over bedrock, the ‘B horizon’ will overlie what is termed the ‘R horizon’. The role of the soil fauna has long been recognised. Darwin (1881) was the first to quantify in a systematic fashion the soil-moving abilities of earthworms. Seven years later, Drummond (1888) argued very persuasively that in terms of their soil moving capacity, termites were the tropical analogues of earthworms in more temperate climatic regions. Many fossil soils in arid North Africa show evidence of former termite activity (Conrad, 1959), with implications for soil classification. Not all soils in tropical areas fall neatly into the ABC system of soil horizons recognised by soil scientists. For instance, soil profiles developed over weathered granite in tropical Africa and tropical northern Australia are a result of the activities of the soil fauna – in this case termites – and have been described as having M, S and W horizons (Nye, 1954, 1955; Watson, 1961, 1962, 1964; Williams, 1968a, 1978). Many other soils also reflect the influence of the soil fauna, especially from ants, earthworms and burrowing mammals (Paton, 1978; Johnson, 1993; Paton et al., 1995). The M horizon is a migratory mineral layer of mainly quartz coarse sand and very fine quartz granules up to 3 mm in size, and often up to 50 cm thick. The M horizon overlies the S horizon or stone layer, which may be enriched towards the base in clay-sized particles. The stones consist of mostly quartz fragments over about 3 mm in size derived from quartz veins within the weathered granite. The W horizon refers to the granite weathering profile which grades down into fresh rock (Fig. 4.7). Such soils have formed as a result of the mining activities of mound-building worker termites, which travel down their

Geology and Soils

W1

migratory

soil horizons

50

1 2 3 4 5

weathered granite

W2

6

7 8 9

W3

10 11 12

Figure 4.7 Schematic profile of soil developed on weathered granite. (Based on the author’s unpublished observations.) 1. Sand over quartz stone layer. 2. Gradual irregular boundary. 3. Stony red clay loam. 4. Diffuse wavy boundary. 5. Undisturbed quartz or pegmatite veins within weathered granite. 6. Reddish-brown or yellowish-red sandy clay loam. 7. Whitish-grey granite remnant. 8. Clear smooth boundary. 9. Clay-enriched (illuviated) surface of W3 zone. 10. White or grey coarse sandy loam. 11. Diffuse irregular boundary. Fresh granite.

subterranean galleries until they reach the moist weathering front where they forage for silt-, clay- and sand-sized particles with which to construct their mounds during the wet season. The sand grains act as bricks and the wet mud as mortar. Once abandoned, the mounds are eroded by rainstorms and soon break down into their constituent particles. The finer particles are rapidly washed downslope by rain splash (Williams, 1969) and overland flow, leaving the sands and fine quartz granules behind to form the M horizon. The long-term effect of the termite mining activities at the weathering front is to leave a concentrated subsurface residue of any quartz particles too big for the worker termites to carry, usually about 2–3 mm. This residual layer comprises the S horizon or stone layer and may contain mineral particles such as gold.

4.3 Soils

51

Higher etchplain Upl

ift

A

B

Duricrust

Inselberg

Lower etchplain

Pediment

C

B

Basal weathering front

C

Active slopewa

sh

Weathered mantle Bedrock

Figure 4.8 Deep weathering of granite rocks and exhumation of the weathering front, southern Nile Basin. The exhumed deep weathering front is termed an etchplain. The duricrust is an iron-enriched caprock. An inselberg is a hilly erosional remnant. A pediment is a gently sloping rock-cut surface at the base of a plateau or an inselberg. The basal weathering front is the transitional zone between fresh and deeply weathered bedrock (weathered mantle). (Based on the author’s unpublished observations.)

4.3.1 Soils Formed on Weathered Bedrock and Associated Colluvium Roughly half of the soils found within the Nile Basin have been formed in situ on weathered bedrock and are termed sedentary soils. Colluvial sediments derived from the adjacent weathered bedrock are often found next to areas of sedentary soil and may support skeletal or immature soils. Colluvial materials can range from very coarse-grained to fine-grained, and consist of sediments that have moved downslope under the influence of gravity, either rapidly as with debris-flow and rock-fall deposits or slowly as with fine sediments subject to soil creep and other forms of slow mass movement. As noted earlier, the physical and chemical properties of recently formed sedentary soils closely reflect the nature of the parent rock from which they are derived. For instance, soils developed on weathered granites are often slightly acidic and usually consist of a topsoil of quartz sand over a more clay-rich sandy subsoil over weathered rock in which some or all of the original rock fabric is still visible. In those seasonally wet tropical areas where moundbuilding termites are active, the granite-derived soils can consist of a three-layered soil profile, with a topsoil of coarse quartz sand over a stone layer of angular quartz over the

52

Geology and Soils

Figure 4.9 Granite inselberg, central Sudan.

weathered parent rock (Fig. 4.7). Such soils are not particularly fertile and will tend to support a cover of sparse trees and tall grasses. Around the periphery of many granite hills in the Sudan there is often an aureole of sandy clay, which retains moisture during the dry season and supports a much denser band of vegetation (Ruxton, 1958). The reason for this is that mechanical eluviation by subsurface lateral flow (Kirkby and Chorley, 1967) moves the finer clay and silt particles down slope and through the sandy topsoil until they reach the bottom of the slope, at which point the hydraulic gradient diminishes rapidly and the fine particles are deposited within the voids around the coarser sand grains. In the hot wet tropical regions in the south of the Nile Basin, the granites are usually deeply weathered, and have often undergone one or more later phases of vertical incision and valley widening, resulting in the exhumation of the irregular former weathering front (Fig. 4.8). The landscapes formed in this way consist of low granite tors and tall granite hills or inselbergs, rising above gently sloping plains and wide alluvial flats (Fig. 4.9). As one might expect, soils formed recently on quartz sandstone consist of a fairly uniform layer of quartz sand that lies directly over the parent sandstone. In the drier parts of the Nile Basin, these sandy soils may also contain lenses of gravel deposited by local ephemeral stream channels. In Nubia and southern Egypt, the Nubian Sandstone covers a huge area and is a major regional aquifer, so that some modest cultivation is possible despite the lack of plant nutrients in these sandy soils. What nutrients there are have often come from wind-blown dust, which frequently consists of calcareous clay particles that are

4.3 Soils

53

subsequently washed into the sandy soils during occasional rainstorms or during wetter intervals in the recent past. In the far south of the Nile Basin in localities where the climate is perennially wet or where there is a prolonged wet season, the sandstone is often highly weathered and the soils formed above it contain abundant concretions of iron-cemented sand. These red ironstone soils have a variety of local names and are often termed laterites or ferralsols. [The term laterite has been used to describe such a huge variety of soil materials of quite different origins (Paton and Williams, 1972) that the discerning reader had best declare a closed season on the use of this term.] Certain fossil soils developed over Nubian Sandstone in arid northern Sudan have the vesicular honeycomb structure characteristic of soils once inhabited by termites. In order to survive, termites need to forage for grass or wood, depending on the species, and in order to construct their galleries need access to moist clay or silty clay, all of which presuppose a climate somewhat less arid than at present in what is today known as Nubia (Williams, 2012b). Efforts to obtain optically stimulated luminescence (OSL) ages for these soils have proven difficult to interpret, yielding ages ranging between late Pleistocene and middle Holocene (ca. 52–6.5 ka), which may simply indicate that termites were last active in this now arid environment early in the Holocene. We discuss this further in Chapter 14. Beds of siltstone, shale or mudstone are sometimes sandwiched between more massive beds of sandstone, and being easier to erode than sandstone often form relatively flat depressions or valleys, depending on the thickness of the siltstone beds. The soils developed on siltstone are often quite variable and may be overlain by a transported layer of colluvial sandstone, but are in all cases fine-grained and usually richer in plant nutrients than the sandstone soils. In parts of Nubia local villagers quarry the buff and purple siltstones and use them to coat the walls of their mud-brick houses. Some of the siltstone soils contain gypsum, and modest quantities of gypsum can improve the permeability of heavy clay soils. In the drier northern regions of the Nile Basin and in places along the main Nile valley there are localised outcrops of limestone. The soils developed on limestone vary considerably, depending on whether the parent rocks consist of muddy limestone, which weathers to calcareous clay, or very pure limestones, which give rise to thin and highly calcareous soils unless there is a significant contribution from wind-blown dust. Provided that there is a protective capping of desert pavement, the dust layer beneath the surface lag gravel can become thicker over time and can eventually form well-developed soils. (A desert pavement is a more or less continuous thin surface layer of stones that protects the underlying soil or sediment from erosion.) By far the most fertile soils in the Nile Basin are those derived from volcanic rocks. The Ethiopian headwaters of the Blue Nile and the Atbara rise amidst the basaltic lava flows of the Ethiopian Highlands, as do the headwaters of the Sobat, which carries a suspension load of mud from Ethiopia to the White Nile, and the headwaters of the Rahad and Dinder Rivers, which today flow into the lower Blue Nile but may once have flowed directly across the alluvial plains of central Sudan to join the White Nile directly. We discuss these alluvial soils in Section 4.3.2; our concern here is with soils formed directly from basalt. In welldrained sites the basaltic soils are well aerated and the iron content sufficiently well oxidised to give rise to vivid red clay soils with the dominant clay minerals kaolinite and

54

Geology and Soils

illite. In less well drained sites the dominant clay mineral is smectite (montmorillonite) and the soils are dark grey or dark grey brown cracking clays or vertisols, which expand when they are wet and shrink and crack as they dry out. These soils are widespread in the Ethiopian Highlands and are easily eroded during the intense rainstorms of the summer monsoon, bringing silt and clay to the lowlands of central Sudan and the flood plains of the Blue Nile, Atbara and main Nile, to which we now turn.

4.3.2 Alluvial Soils At intervals throughout the Quaternary, and most recently during the last 15,000 years of the terminal Pleistocene and Holocene (Talbot et al., 2000), the Ethiopian tributaries of the Nile have ferried a suspension load of silt and clay to the main Nile during the course of the Ethiopian summer monsoon season. The relatively narrow flood plain of the Nile has been built up and periodically eroded throughout this time, leaving a series of alluvial terraces as witness to these alternating episodes of alluvial sedimentation or aggradation, and fluvial incision or degradation (Hull, 1896; Butzer and Hansen, 1968; Butzer, 1980). The soils that have subsequently developed on these alluvial silts and clays (Table 4.3) reflect their combined volcanic and alluvial inheritance, and today consist of the dark cracking clays or vertisols (Fig. 4.10) that are such a characteristic feature of the arable lands of the Egyptian Nile Valley (Butzer and Hansen, 1968; Said, 1981, 1993; Embabi, 2004) and of the Gezira plain between the Blue and White Nile rivers in central Sudan (Tothill, 1946a, 1948; Sir Alexander Gibb & Partners, 1954; Jewitt, 1955; Beinroth, 1966; Drover, 1966; De Vos and Virgo, 1969, Buursink, 1971; Fadl, 1971; Gunn, 1982; Williams et al., 1982; Blockhuis, 1993; Williams et al., 2015a). These soils are often selfmulching in that they develop a granular surface layer (Fig. 4.10), allowing them to retain moisture during the dry season. They generally have a moderate to high content of clay-sized particles (< 2 μ) and a correspondingly high cation exchange capacity. Although mostly nonsaline and non-alkaline, in some sites such as shallow depressions or former fluvial embayments or may’a that were once flooded but have since dried out, the alluvial clays can be both saline and alkaline and may have a massive surface crust. Such soils occur sporadically along the lower White Nile valley, and reflect the flood history of that river, with its very gentle flood gradient of 1:100,000 (Williams, 1968b). One still unresolved question is the origin of the clays at higher elevations between the Blue and White Nile Rivers (Sir Alexander Gibb & Partners, 1954; Gunn, 1982; Williams et al., 1982), as well as the widespread clay plains situated east of the Blue Nile (Bunting and Lea, 1962; Ruxton and Berry, 1978). Are these clays ancient alluvial clays deposited by the Blue Nile when it once flowed at much higher elevations on leaving the Ethiopian Highlands? Or are they clays formed in situ by weathering of the underlying bedrock and later moved downslope by variable amounts through processes of slow mass movement? Certainly, there is good field evidence that some of these clays do indeed appear to be part of a deep weathering profile grading down into the underlying bedrock (Hunting Technical Services, 1964), but we cannot rule out an alluvial origin for at least some of the higher-level clays located west of the present Blue Nile

4.3 Soils

55

Figure 4.10 Cracking clay soil, Blue Nile Valley, Sudan.

between the latitudes of Roseires (11.8°N, 34.4°E) and Sennar (13.4°N, 33.2°E) and extending out to the White Nile Valley above an elevation of 386 m.

4.3.3 Dune Soils Soils that are developed on sand dunes initially have few obvious soil horizons and support meagre plant cover. However, as time progresses, the previously active dunes become stabilised by vegetation and trap calcareous desert dust particles which are then washed into the porous sand to form a clay-rich substrate. This process can be quite rapid: a matter of decades rather than centuries, depending on the dust flux (Williams, 2014). Analysis of the strontium and neodymium content of such soils shows that dust storms are the main contributors of the clay and calcium carbonate (Woodward et al., 2015a). Close inspection of many vegetated dunes shows multiple clay bands running parallel to the original bedding. Once these bands coalesce the outcome is a topsoil of quartz sand over the clay-rich subsoil in which calcium carbonate concretions are often very abundant. Such concretions are at first soft and quite diffuse, but over time become hard and may form cylindrical coatings (‘rhizocretions’) around tree roots (Williams, 1968c; Amit et al., 2007; Williams, 2014; Williams et al., 2015a). At a later stage the individual concretions may become cemented into calcrete, a very hard, massive and impermeable layer of calcium carbonate (Williams, 2014). Calcretes are found near the base of some of the dunes east of the lower White Nile at depths of 1.5–2.0 m and are associated with former

Figure 4.11 Dune soil catena east of the lower White Nile, central Sudan. (Modified from Williams, 1968c), showing polygenetic and truncated soil profiles.)

4.3 Soils

57

ponds between dunes (Fig. 4.11) or the now buried margins of former White Nile lakes (Williams, 1968c; Williams et al., 1982) discussed in Chapters 8 and 11.

4.3.4 Desert Soils, Polygenic Soils and Fossil Soils Desert soils in the Nile Basin tend to be shallow, with limited horizon development, and are often both saline and alkaline. They are easily eroded unless protected by a surface layer of gravel (‘desert pavement’) or a hard surface crust. In some instances, the desert pavement may overlie one or more resistant bands of finer sediment formed during earlier intense and very localised desert rainstorms. These sediment layers are often finely laminated, reflecting deposition from overland flow. In addition, a thin layer of fine silt and clay particles may form an impermeable crust, promoting further runoff; such crusts are usually short lived. These sediment layers may also become cemented with silica, iron or calcium carbonate to form ‘hardpans’. Soils developed on stable land surfaces in arid areas may over time develop distinct soil profiles indicative of past fluctuations in precipitation (Amit et al., 2007, 2010). The depth to gypsum within the soil may be a guide to former precipitation (Retallack and Huang, 2010). Both the amount and depth of soil carbonate within a given dryland soil may also be indicative in a very general sense of the climatic conditions under which it formed (Williams, 2014, pp. 276–278). The problem here is a variant on the well-known geomorphic conundrum of magnitude versus frequency. Is it the events of low frequency but high magnitude that do most of the work? Or is it the events of moderate magnitude and moderate frequency that are the most effective agents of erosion? Do the red paleosols developed along the wadi channels eroded in Nubian Sandstones in the desert on either side of the main Nile (Williams et al., 2010; Williams, 2012b) indicate a long interval of only slightly wetter climate or a shorter interval of much wetter climate? To find convincing answers to these questions we need to supplement any insights obtained from fossil soils with other independent lines of enquiry. We also need to buttress our field and laboratory observations of soil structure, texture and chemistry with additional microscopic analysis of soil fabric and mineralogy (Brewer, 1964; Zerboni, 2005, 2008; Williams et al., 2015b). Another interpretative problem may arise in the case of polygenic soil profiles. For example, in the Nubian Nile Valley of northern Sudan east of Kerma, the eastern margin of the alluvial flood plain is situated about 10–15 km from the present Nile channel. The flood plains soils are cracking clays or vertisols formed on alluvial clays laid down during the course of Holocene Nile floods, as is evident from the lenses of sub-fossil aquatic snail shells scattered throughout the parent alluvium. In some places, there are stacked layers of three or more vertisols, indicative of discrete episodes of flood plain accretion followed by cessation of flooding and the subaerial development of the cracking clay soil profiles. These are examples of polygenic soil profiles, in which soil-forming processes may continue to operate within soils of different ages. Towards the edge of the Holocene flood plain well sections reveal an alternation of alluvial sands and dark grey-brown cracking clays. In this instance, the sands are derived from local ephemeral stream channels flowing westwards from the dissected Nubian Sandstone

58

Geology and Soils

escarpment, so that we have an inter-digitation of local wadi sands and fine gravels with distal Nile flood plain clays. The wadi sands were laid down during flash floods or intervals when the local climate was wetter than today. These sands are quite distinct from alluvial sands and gravels deposited by the Nile during times of extreme floods, which have a very different heavy mineral content, and often contain shells of the freshwater Nile mussel (Unio sp.) and the Nile oyster (Etheria elliptica), often with both valves closed and in primary context (see also Chapter 11).

4.3.5 Soils Formed on Volcanic Ash Soils developed on volcanic ash occupy relatively small areas within the Nile Basin, including the region around Jebel Marra volcano in western Darfur, described in Chapter 13. Despite their limited extent, they are important for several reasons. First, they give rise to very fertile soils rich in potassium and other trace elements. As a consequence, they often support a dense cover of trees, herbs and grasses, providing food and shelter to mammals large and small, all of which would have been attractive for prehistoric hunters and gatherers. The soils formed on volcanic ash are not only fertile but also easy to work with simple hoes and digging sticks, and so would have been attractive to early farmers, as the terraced slopes of Jebel Marra bear witness. Finally, the ash beds would have buried and in some cases preserved some of the prehistoric plants, as in the case of the late Early to early Middle Pleistocene oil palm leaf fossils found in several piedmont localities 50 km SSW of Jebel Marra caldera (Wickens, 1975a, 1976a; Williams et al., 1980; Philibert et al., 2010).

4.4 Conclusion Precambrian igneous and metamorphic rocks form the underlying basement across the Nile Basin. These ancient rocks are often concealed beneath a thick mantle of mostly Palaeozoic, Mesozoic and Cenozoic sedimentary rock formations. The highlands of Ethiopia owe their origin and current elevation to uplift and volcanic activity during the last 30 Ma. The silts and clays deposited along the flood plain of the Nile in Egypt and Sudan are derived primarily from weathered volcanic rocks eroded and transported by rivers in the headwaters of the Blue Nile and Atbara drainage basins. Smaller volcanic centres are scattered across the desert region adjoining the main Nile. In the far west of the Nile Basin, Jebel Marra is a presently dormant volcano with fumaroles and hot springs that last erupted during the Holocene. It may have begun to form during the Miocene, and was intermittently active during the early to middle Pleistocene. The soils of the Nile Basin range from deep red loams developed on highly weathered rocks in the tropical south to weakly developed sandy soils formed on dunes and weathered Nubian Sandstone in the arid north. The alluvial soils flanking the Nile and its tributaries have long proved attractive to human settlement, and played a seminal role in the inception of agriculture in this region, as outlined in later chapters.

5 Vegetation, Land Use and Human Impact

Seen close, the destruction was incredible. The place looked like a First World War battleground. Everywhere the ground was thick with hot ash, smoking debris, charred stumps, and partially burned tree trunks, lying about willy-nilly. Some of the felled trees had been sixty or more feet [20 m] in height, and now they lay burned on the ground. I could hardly believe that a few men, with simple, blunt iron axes had felled so many huge trees: Cedar, olive, hagenia, Podocarpus, euphorbia and many others. This year alone, Bogale and his sons and followers had cut and burned about three square kilometres of forest. Clive Nicol, From the Roof of Africa (1971, p. 262)

5.1 Introduction The aim of this chapter is to review some of the key factors that influence the nature and distribution of the natural vegetation cover in the Nile Basin. After reviewing the influence of what may be considered first-order climatic factors such as precipitation, temperature and evaporation, we take note of the subtler influences of local topography and soil type, which at a more local scale often over-ride the regional influence of purely climatic factors. The present natural vegetation in the Nile Basin also reflects past migrations of plants from the Mediterranean plant domain in northwest Africa, the African tropics in the south, and the drier parts of Eurasia extending through Arabia and Iran as far as India. We also consider how human activities in the Nile Basin are all too often out of harmony with the natural world and have had a huge and deleterious impact upon the original plant cover. The opening quotation is a vivid illustration of how much damage a few peasant farmers can wreak upon the natural vegetation using simple tools and fire. The use of fire in Africa extends back a million years, and burning off the dead grass each year before sowing is an age-old practice among farmers in many parts of the Nile Basin, especially in Sudan and South Sudan, where it is known as the harīq system of cultivation (Burnett, 1948, p. 292). The impact on the natural plant cover is even greater where mechanised agriculture is concerned. During years of below average rainfall – common enough events in the drier two thirds of the Nile Basin – the effects of drought exacerbate the effects of overgrazing, so 59

60

Vegetation, Land Use and Human Impact

that a wave of desertification and accelerated soil erosion is set in motion (Obeid et al., 1982; Wickens, 1982; Kassas and Batanouny, 1984; Kassas, 1995a; Williams and Balling, 1996; Williams, 2014, chapter 24). It is easy to blame humans for land degradation in arid and semi-arid areas, but we need to remember that erosion is a natural process and that ‘natural desertification’ is quite widespread and owes nothing to human mismanagement, although humans can (and often do) unwittingly accelerate these natural processes. Finally, we conclude with an evaluation of both the benefits and the damaging environmental and health consequences of large dams built to control Nile floods and provide a reliable supply of water for irrigated agriculture.

5.2 Natural Vegetation Zones One of the most intriguing aspects of the present-day vegetation in the Nile Basin and adjacent regions is its diversity. Some of this diversity reflects the extreme climatic gradients that run from south to north across the Basin (see Chapter 3). But some of the variation is also a result of repeated plant migrations from across a huge area to the east, south and northwest. Quézel (1962, 1997), Ozenda (1977), Wickens (1976a, 1976b), Maley (1980, 1981, 1996) and many others have drawn attention to the admixture of both tropical African, Indo-Iranian and Mediterranean floral elements growing today on the mountains of the Sahara, including Jebel Marra volcanic massif on the western margin of the present Nile Basin. The great Saharan polymath Théodore Monod noted that there were also relict populations of animals still surviving within some of the Saharan mountains, having followed former vegetation corridors linking the rain forests of West Africa with uplands such as the Aïr (Monod, 1963). These corridors of riparian vegetation had developed during wetter climatic intervals when now ephemeral or defunct river channels experienced perennial flow. Moreau (1963) explained the present disjunct distribution of the montane avifauna in east and central Africa by invoking lowered temperatures and isolation of the high-altitude forest during past glacial episodes. The entire question of inheritance, plant and animal refugia, isolation, genetic change and adaptation to new conditions is a huge and complex one. Suffice to say here that the concept of an unchanging Quaternary equatorial and tropical rain forest has long been discredited as a growing body of evidence from fossil pollen spectra reveals a dynamic response of the forest to changing climatic conditions during the Quaternary (Bonnefille, 1972, 1976b, 1980; Flenley, 1979; Hamilton, 1982; Maley, 1996; Bonnefille et al., 2004; Anhuf et al., 2006; Gasse et al., 2008).

5.2.1 Influence of Precipitation and Temperature Precipitation decreases more or less regularly from well over 1,500 mm a year in the southern part of the Nile Basin to less than 5 mm a year in the northern deserts of Egypt and Sudan. The latter figure is somewhat meaningless, because it conceals a more

5.2 Natural Vegetation Zones 20 oE

61

30°E

40°E

MEDITERRANEAN SEA

Desert Semi-desert Low rainfall savanna

30o N

Flood regions High rainfall savanna

EGYPT

LIBYA

Rain forest and montane forest

RE D A SE

20°N CHAD

ER IT R EA

J

S

SUDAN

10°N SOUTH ETHIOPIA CENTRAL AFRICAN REPUBLIC

SUDAN

UGANDA KENYA

SOMALIA

CONGO

0°N 250 km

500

DA AN RW

0

K

Figure 5.1 Vegetation zones in the Nile Basin. (Compiled by the author with data from Andrews, 1948: Fig. 1; Rattray, 1960: 1:10,000,000 Vegetation Map; Russell, 1962: 1:4,000,000 Vegetation Map; Schaller and Kuls, 1972: 1:4,000,000 Vegetation Map; Wickens, 1982; Nile Basin Initiative (2012); and the author’s unpublished observations.)

62

Vegetation, Land Use and Human Impact

extreme reality: years or decades with no rain at all, interspersed with rare episodes of intense and sometimes catastrophic rainstorms and floods. Even in the seasonally wet tropical regions of the Nile Basin, there is considerable year-to-year variability, with intervals several decades or more in duration with below average rainfall, followed by long intervals of above average rainfall. The natural vegetation has adapted to this variable rainfall regime by developing a whole series of very effective survival strategies that range from the purely physiological to the selection of favourable sites that minimise the impact of high rates of evaporation and allow access to reliable sources of soil water (Kassas and Batanouny, 1984; Wickens, 1984; Williams, 2014, chapter 4). Figure 5.1 shows very clearly that climate exerts a first-order control over the distribution of major vegetation zones within the Nile Basin, with equatorial and tropical rain forest in the region encompassing the White Nile headwaters and the Ugandan Lake Plateau in the south, savanna woodland and grassland in much of the centre (Rattray, 1960) and desert steppe in the arid north (State of the Nile Basin, 2012). The very distinctive thickets of Euphorbia candelabra growing around Erkowit, a mist oasis in the Red Sea Hills near Port Sudan, are a striking consequence of how even relatively low elevations (just over 1,000 m) can trigger enough precipitation – in this case fog rising from the nearby Red Sea – to allow somewhat atypical plants like Euphorbia candelabra to grow (Fig. 5.2). Euphorbia candelabra is also found scattered across the rocky semi-arid plains of Tigray in northern Ethiopia (Fig. 5.3). In addition to the strong latitudinal influence exerted by precipitation and temperaturecontrolled evaporation upon vegetation zones, topography and elevation also exert a strong influence on vegetation in mountainous regions of the Nile Basin. The adiabatic lapse rate in the Ethiopian Highlands amounts to a decline in temperature of roughly 0.65°C for every 100 m of increased elevation (Williams et al., 1978). The elevation of the Ethiopian capital Addis Ababa is 2,355 m, and the highest summit in Ethiopia is Ras Dashan in the Semien Mountains at 4,550 m elevation. Applying this lapse rate would suggest that the average temperature on top of Ras Dashan would be about 14°C colder than at the elevation of Addis Ababa. Needless to say, both diurnal and seasonal temperatures fluctuate considerably at these and other localities, but it remains true that sites at high elevation are colder than sites in the lowlands, with profound consequences for native plants and agriculture. At extreme elevations, the plants need to be able to withstand the diurnal frost cycles, and temperature controls the upper tree line (or timber line) in the high mountains, which coincides very approximately with the 10°C isotherm for the warmest month. During colder glacial intervals, temperatures were 4–8°C colder than today in the Ethiopian Highlands, so that the upper limit of the tree line was about 1,000 m lower than it is today (see Chapter 7). Temperature is of course not the only factor that controls vegetation zones in the mountains of the Nile Basin. Precipitation also tends to increase with elevation. This is because as warm moist air rises, it cools adiabatically, reaches dew point (100% relative humidity) and the water vapour condenses to liquid form. The resulting drops coalesce and reach sufficient mass that they overcome any frictional resistance and fall as raindrops,

5.2 Natural Vegetation Zones

63

reaching terminal velocities of 5–8 m/sec on impact with the ground or tree canopy. (If the ground surface is bare and unprotected, the impact of falling raindrops can cause significant rainsplash erosion, which acts as a precursor to erosion from overland flow.) The converse to rainfall increasing with increasing elevation applies too, so that precipitation will tend to decrease as elevation decreases. Put simply, the lowland plains are hot and dry, the higher elevations are cold and wet, and the intermediate elevations are moist and warm. Ethiopian peasant farmers have recognised five major agro-climatic zones for at least the past 400 years. According to the account written by the Portuguese Jesuit Manoel de Almeida during 1628 and 1646 and published posthumously, the five zones are known as follows in Amharic: (1) Choquê (high and extremely cold country), (2) Deqâ (high and perpetually cold country); (3) Oinadegâ (high but temperate country without an excess of cold or heat; (4) Collâ (very hot lowlands); and (5) Baraqhâ (extremely hot desert) (Almeida, 1628–1646). In a footnote to the 1954 Hakluyt highly abridged edition and translation of Almeida’s work, the climatic zones (now mysteriously reduced to four) are rendered in an editorial footnote as barahã (the desert), quallã (the hot lowlands), waynã degã (the temperate highlands, literally ‘highlands of the grape’) and degã (the cold highlands). It is worth remembering that during the time of Almeida’s travels in this region Ethiopia was in the grip of the Little Ice Age, in common with the rest of the world (Lamb, 1970, 1972, 1977; Grove, 1988, 2004; Vita-Finzi, 2008), so that it would have been even colder at higher elevations than it is today. Furthermore, droughts were longer and more frequent at this time in southern Kenya (Verschuren et al., 2000), indicating a weaker summer monsoon over East Africa, including Ethiopia. The Ethiopian lowlands were therefore most likely even drier than they are today. Vegetation in the hot dry lowlands ranges from tall grass savanna with scattered acacias down to desert steppe as rainfall diminishes and temperatures increase. The relatively temperate intermediate zone mostly supports acacia woodland and greatly modified grassland, with thickets of a limited number of exotic eucalyptus species in and around towns, churches and villages (Schaller and Kuls, 1972). The highlands used to support a moderately dense cover of tall mountain forest (Podocarpus, Juniper, bamboo, Hagenia) but much of this has been removed during the past fifty years and more, except in sanctuaries close to churches, such is the hunger for arable land.

5.2.2 Influence of Local Topography Although regional climatic gradients in temperature and precipitation control the major vegetation zones in the Nile Basin, other factors also play an important role at a more local level. Kassas and Batanouny (1984) have emphasised how local desert wadi systems concentrate sporadic precipitation and associated runoff in the desert areas of Egypt and Sudan, enabling trees and shrubs to grow in otherwise desolate areas largely devoid of vegetation. In northern Sudan, the dissected margins of the Nubian Sandstone plateaux east

64

Vegetation, Land Use and Human Impact

Figure 5.2 Euphorbia candelabra on the hills near Erkowit mist oasis in the Red Sea Hills, Sudan.

Figure 5.3 Euphorbia candelabra on the rocky semi-arid plains of Tigray in northern Ethiopia. (Photo: Frances Williams.)

5.2 Natural Vegetation Zones

65

of the Nile Valley concentrate sufficient occasional runoff that the low-angle alluvial fans and shallow radiating ephemeral stream channels between plateau and flood plain are covered in scattered acacia trees and grass tussocks, in contrast to the bare rocky plateaux from which they emerge (Williams, 2012b). The deep ravines along the sheltered flanks of Jebel Marra volcanic massif are difficult to access and contain a relict flora of gallery forest surviving from a time of wetter climate, most probably early Holocene (Wickens, 1976a, 1982).

5.2.3 Influence of Soil Type Although we might expect that clay soils, with their high cation-exchange capacity, greater innate fertility and much higher water holding capacity would support a more luxuriant vegetation than sandy soils, the opposite is true in the vast semi-arid stretches of the Nile Basin. In an elegant study entitled Distribution of tree species in the Sudan in relation to rainfall and soil texture, and now considered a classic among plant ecologists working in arid areas, Smith (1949) demonstrated that any given species of acacia growing on sand needs only about two-thirds of the annual precipitation of the same species growing on clay (Fig. 5.4). He was careful to exclude sites that were not level, so that there could be no question of surface water directly after rain being either concentrated locally or shed by overland flow (runoff). Williams et al. (1982, p. 129) commented further on this relationship between rainfall, soil texture and the distribution of acacia trees in what was then the Sudan: ‘During the 30 year period 1931–1960 rainfall in the Sudan ranged from over 1400 mm in the south to less than 10 mm in the north . . . Since this range far exceeds the likely span of Holocene rainfall fluctuations in the Gezira . . . [see Chapter 11 in this volume], the relationship deduced by Smith probably also held good in central Sudan during the past 11 000 or so years.’ The reason why sandy soils are able to support tree growth in semi-arid areas more effectively than heavy clay soils is most likely because soil water in sand is held under far less tension than in clay, and so is more readily available to the plant roots, especially during the dry season. Wickens (1982, p. 42, Fig. 3.6) took this factor into account in his hypothetical reconstruction of the vegetation of the Sudan during the early Holocene. Adjacent to the lower White Nile there are broad shallow embayments and backswamps, many of them still prone to flooding even in years of relatively low White Nile discharge. These topographic depressions are known locally as maya’a and are often mantled with heavy alkaline clay soils. The maya’a soils support a different array of plants to those growing on the nearby clay plains. On the heavier, alkaline clays we find Acacia seyal, Euphorbia aegyptiaca, Ipomoea cordofana and Sporobolus helveolus. On the lighter but slightly more elevated seasonally flooded soils Fagonia cretica, Chloris virgata, Panicum turgidum, Bergia suffruticosa and Setaria species are the dominant association (Hunting Technical Services Limited, 1964, p. 17).

66

Vegetation, Land Use and Human Impact

Figure 5.4 Influence of rainfall and soil texture on two acacia species, Sudan. (After Smith, 1949; Williams et al., 1982; and Williams, 2014, Fig. 4.4.)

5.3 Current Land Use 5.3.1 Cattle Keepers and Camel Herders In his thoughtful book on Pastoralism in Africa: Origins and Development Ecology, Andrew Smith states that ‘Pastoralism is the most effective means of exploiting African grassland environments within a traditional economy’ (Smith, 1992, p. 10). He was referring in particular to grasslands in areas of low and erratic rainfall avoided by most cultivators. He also emphasised that pastoralism has been practised in Africa for several thousand years at least, although the nature and timing of its origins are still not fully understood. The inception of cattle domestication in the Nile Basin and adjacent Sahara remains a hotly debated topic (Smith, 1992; Clark et al., 2008c). On present evidence, it seems that domesticated cattle appear in the archaeological record towards 8–7 ka in the Egyptian Nile Valley and adjacent desert (Honegger and Williams, 2015) and somewhat later in central Sudan and Ethiopia (Smith, 1992; Clark et al., 2008c). In the central Sahara at the prehistoric site of Adrar Bous in the Ténéré Desert of Niger, the complete skeleton of a short-horned domesticated Neolithic cow

5.3 Current Land Use

67

Bos brachyceros, was recovered whole from what appeared to have been a former pond or swamp (Williams, 2008). Collagen from a bone sample gave an uncorrected conventional radiocarbon age of 5,760 ± 500 14C years BP (UCLA-1658) and in 2004 enamel from one of the molars gave a calibrated AMS 14C age of 4145 ±45 BP (CAMS-103518) (Clark et al., 2008c). As we shall see in later chapters, during the first half of the Holocene the climate throughout the Nile Basin was significantly wetter than it is today, albeit with drier interludes, allowing pastoral nomads to graze their cattle in areas that are presently too dry to provide much pasture except after localised rainstorms, enabling the dormant seeds of hardy desert grasses to germinate. The gizzu or seasonal camel grazing in northwest Sudan consists mostly of nissa (Aristida ciliata) and dirim (Indigofera arenaria), both of which seem to be preferred when in a desiccated state (Bacon, 1948, pp. 398–399). The long and close interaction between cattle and humans is also reflected in burial rituals involving cattle skulls (Dubosson, 2011), a ritual still evident today among the Hamar people of the Omo delta region in southern Ethiopia. The Dinka, Shilluk and Nuer Nilotic people of South Sudan are major cattle herding tribes, as are the Baggara cattle herders of western Sudan. (The Arabic word for cow is ba´qara.) The Baggara migrate northwards during the rainy season, both to find fresh and nutritious grazing brought on by the rains, and to escape the biting Tabanidae horse flies that can make life so vexatious for man and beast. During the migration the women and babies ride on the backs of oxen, often shaded by tent-like structures. The Neolithic rock paintings of the Tassili sandstone plateaux in Saharan Algeria recorded by Henri Lhote show very similar scenes of women riding on the backs of oxen (e.g., Museen der Stadt Köln, 1978, p. 421, plate 7). In the drier parts of Ethiopia cattle are the main source of wealth and prestige for the pastoral nomads, although they may have larger herds of sheep and goats, and cattle raiding remains common. In the wetter wooded regions of the Nile Basin, where the bite of tsetse flies (Glossinidae) can cause Trypanosomiasis in cattle and sleeping sickness in humans, rearing cattle is more difficult, so that cattle herders tend to prefer the savanna grasslands and the desert margins. Dogs, with their far longer history of domestication than cattle, appear often on the Neolithic rock paintings of cattle and humans at Jebel ‘Uweinat (on the border between Sudan, Egypt and Libya) as well as elsewhere in the Sahara (Muzzolini, 1995; Coulson and Campbell, 2001). We need also to bear in mind that cattle rearing is not necessarily an exclusive activity: the Nuer and other pastoral tribes in South Sudan cultivate crops on the higher ground and obtain much of their protein from fish caught near the end of the dry season. Many pastoral groups are also opportunistic hunters of small animals and collectors of wild plant food, the latter activity being especially important in times of drought and famine. The Hyksos brought the horse to Egypt shortly before 1650 BC, and horses were used in warfare during the Persian invasion of Egypt led by Cambyses. Camels were present at least sporadically in the Sahara during about the second millennium BC, and in the northern Nile Valley by 2.6 ka, but it was not until the Arab invasions from about 400 AD onwards that camels fully adapted to desert conditions become numerous and widespread. Camel rearing

68

Vegetation, Land Use and Human Impact

is now an important industry in Sudan and each year great herds of camels make the journey north to Egypt where they are slaughtered for meat. The Kababish camel herders in northern Sudan depend on their herds for transport, milk, occasional consumption of meat, and hair to make tents and rugs, and payment of dowries. Racing camels are highly prized and obtain a good price in Egypt and the Middle East as well as at regional markets in Sudan. The Shukriya pastoral nomads in east-central Sudan graze huge herds of camels on the vast clay plains east of the Blue Nile, Dinder and Rahad Rivers. The mid-height na’al grass (Cymbopogon nervatus) provides excellent grazing around the granite hills of Jebel Fau, and the camels’ milk is so rich that the author once used it to make a simple cream cheese. Scattered amidst the na’al grasslands one can find sporadic patches of domestic sorghum or dura (Sorghum vulgare), sown opportunistically and harvested in similar fashion.

5.3.2 Rain-Fed Cultivation Away from the Nile in the vast semi-arid areas that make up about half of the Nile Basin, the farmers rely on the seasonal summer rains to grow their crops. Sorghum (dura) is widely grown on the clay soils and millet (dukhn: Pennisetum typhoideum) on the sandy soils, together with a variety of vegetables including onions, tomatoes, green vegetables, okra or bamia (Hibiscus esculentus) supplemented by groundnuts, sesame and, in some favoured localities, fruit trees. At least five distinct species of sorghum are regularly cultivated in central Sudan – a degree of speciation reflecting some three thousand years of sorghum cultivation in the lower White Nile Valley (Clark and Stemler, 1975). Considerable efforts are often made to conserve both water and soil. The Fur cultivators and their ancestors have practised terracing on the slopes of Jebel Marra for several thousand years. Hafirs or excavated earth ponds designed to trap runoff are common throughout the Sudan. The Arabic term ‘hafir’ comes from ‘hafir’ meaning to dig (Robertson, 1950). Smaller trenches are simply called ‘hufra’. On the clay soils of central Sudan, the farmers maintain a dense network of earth bunds (terūs) to concentrate rainwater in the fields being used that season for crops. It is not unknown for water to be diverted from small outlet canals (mostly at night and sometimes illicitly) to irrigate small plots of onions and tomatoes. During the early 1960s before the severe and widespread Sahel drought that began in 1968, local farmers in the western Gezira informed the author that on average they might expect about four good years in every ten. A good year is one in which the peasant farmers have enough dura to meet their normal needs.

5.3.3 Floodwater Cultivation from the Nile Given this level of uncertainty in regard to the timing and amount of seasonal rainfall, one can see the attraction of a ready access to the more reliable floodwaters of the Nile, or, rather, to the moist silts exposed once the Nile waters had receded from the flood plain. French authors refer to this type of cultivation as the décrue method of farming, with the

5.3 Current Land Use

69

word décrue simply meaning a fall in flood level. This method of cultivation is still in use today in many parts of West Africa as well as in the Nile Valley and has probably been used by Neolithic farmers since the time when a regular flood regime was established along the Nile from about the middle Holocene onwards. Before then, a seasonal sufficiency of wild grass seeds, tubers and other plant foods, allied to abundant game and fish, did not require the extra labour of sowing, harvesting and storage. Once the early Neolithic cultivators along the Nile Valley had mastered the use of an efficient sickle for harvesting the mutant heads of wild cereal grasses in which the zone of abscission at the base of the inflorescence did not function in such a way as to shed the ripening grains throughout the growing season, they were well on the way to starting the whole process of cereal plant domestication, as Ann Stemler (1980) has shown. The second fairly obvious prerequisite is the use of rodent-proof pots in which to store enough of the harvested mutant heads of grain to allow the grains to be sown the following season.

5.3.4 Irrigated Cultivation from Canals As the level of the Nile dropped during the dry season, a variety of ingenious ways of raising water were developed along the banks of the Nile, banks that are often quite steep. Narrow erosional gullies forming a type of badland topography known as kerrib are common along the Nile and its tributaries. It is possible that such gullies, with some modest refashioning, could have been used to divert floodwaters towards the low-lying and often marshy back swamps along the distal margins of the flood plains. In any event, the notion of digging canals for irrigation purposes was soon adopted by the ruling élite in Egypt, where a centralised form of government, reasonably secure food supplies and an abundant labour force allowed such major construction works to be implemented and maintained. A stone engraving dating back to more than 5,000 years ago shows the process of canal excavation in action, under the stern gaze of a forceful ruler (Butzer, 1976). Early irrigated farming had to rely on using the rising floodwaters of the Nile to flow along the developing network of canals constructed close to the river and its anabranches. The canals allowed Nile flood waters to be distributed more widely and more efficiently, but did not allow dry season storage of water, which only began with the construction of large dams from the very late nineteenth century onwards. Sir William Willcocks designed the first Aswan dam, which was built between 1898 and 1902. By 1912 the dam was heightened under the supervision of the Scottish civil engineer Sir Murdoch MacDonald, who also designed the Sennar Dam on the Blue Nile and the early stages of the Jebel Aulia dam on the lower White Nile 35 km south of Khartoum. Further heightening of the Aswan dam took place between 1960 and 1970, when the Aswan High Dam was finally completed, the reservoir reaching capacity by 1976. The main purpose of these and the many other dams built along the Nile drainage system is to store the normally abundant Nile flood water for use in dry years and during the dry season, thus allowing large-scale irrigated agriculture to proceed with a high degree of water security. A subsidiary but still important benefit is to curtail the damage done by wild Nile floods –

70

Vegetation, Land Use and Human Impact

something that is never fully reliable, as the catastrophic floods in Sudan during July 2016 were all too sadly to demonstrate. A final benefit is the generation of hydroelectric power. One of the more successful large-scale irrigation projects in the world is the Gezira Irrigation Scheme (GIS) located between the Blue and White Nile Rivers. This began with some small-scale experimental projects devoted to growing high-quality longstaple cotton for export. From the start, the concern was to integrate the local farming population into the project, which would allow crops to be raised on a rotational basis while also producing cotton as a high value cash crop. Gaitskell (1959) provides an informed and dispassionate account of the early history of the GIS. In its heyday in the 1960s, when world prices for long-staple cotton were high, the Gezira Irrigation Area was responsible for providing about two-thirds of the export revenue of the Sudan, grown on barely 1% of the total land area. The canals are gravity fed, and water is now stored in the reservoirs behind both the Sennar Dam, completed in 1925, and the more recent Roseires Dam, built between 1961 and 1966, and like many dams also designed to provide electricity. Section 5.5 considers some of the less beneficial repercussions of building large dams and reservoirs.

5.3.5 Forestry The people within the Nile Basin are better known for causing fairly widespread deforestation (see Section 5.4.1) than for fostering large-scale afforestation projects. Small forestry plots are becoming more common, and often involve a mixture of local and introduced trees, especially eucalyptus (Goda et al., 2004), but also including the neem tree (Azadirachta indica) from India and a member of the mahogany family, the oil from whose leaves deters mosquitoes and so can help with malaria control. There are well over 600 species of eucalypts growing in Australia, adapted to a wide range of climates and soils, so foresters in the Nile Basin have plenty of choice. One possible problem with growing eucalyptus trees is that they operate as very effective groundwater pumps, and so can lower the local water table below the reach of plants with shallow roots. To offset this, they grow rapidly, provide useful wood for fuel and construction, and can help stabilise steep and otherwise bare slopes potentially prone to erosion.

5.4 Human Impact on the Natural Vegetation and Soils 5.4.1 Deforestation Deforestation in the Nile Basin probably began during the Neolithic when the use of hatchets with stone heads ushered in more effective ways of clearing forest than simply resorting to burning. These hatchets consisted of a wooden haft to which was attached a stone axe head sharpened by grinding, an operation that takes about 30–40 minutes, depending on the type of rock chosen, and results in a more convex cutting edge than with a steel axe head. A reasonably fit adult using a sharpened stone hatchet can fell a tree of

5.4 Human Impact on Vegetation and Soils

71

20–30 cm girth at eye height within 30–60 minutes, although the cut will be wider than when using a bronze, iron or steel axe blade. The pace of clearing accelerated during the course of the Iron Age, as the human population increased and more land was needed to grow crops. Pollen spectra from Uganda indicate more widespread clearing and burning during the last 3,000 years (Hamilton, 1982). Charcoal production for fuel has caused extensive removal of forest around the White Nile headwaters in Uganda and Burundi and of savanna woodland in Ethiopia during the past fifty years (Schaller and Kuls, 1972; State of the Nile Basin, 2012). Some authors have argued that the amount and rate of deforestation in Ethiopia and other parts of Africa has been grossly exaggerated and that cycles of historic clearing and regrowth are the norm rather than the postulated irreversible loss of forest (Leach and Mearns, 1996; McCann, 1999). This conclusion may be valid for some of the less densely populated regions with deep fertile soil and high annual precipitation, but does not square with the evidence from many other parts of Ethiopia, including critical areas in the headwaters of the Atbara and Blue Nile (Hurni, 1999; Nyssen et al., 2004). The changes in forest and woodland are equally evident in the semi-arid plains of central Sudan. The natural vegetation in this region is transitional between the semi-desert scrub to the north and the acacia savanna and seasonally flooded tall grasslands to the south (Obeid et al., 1982). This region was once thickly wooded and free from much human interference. Muriel (1901) noted the extensive stands of forest on the east bank of the White Nile from Jebel Aulia southwards and earlier still Baker (1866) referred to the ‘dark forests’ around Ed Dueim on the west bank. There are no such forests left today. In the southern Gezira, the dense Cadaba rotundifolia–Acacia mellifera woodland shown on maps drawn up in the 1920s had given way to low Acacia nubica scrub with a sparse undercover of Aristida grasses mapped by the author in 1962–63 (Hunting Technical Services Limited, 1964, p. 15). In central Sudan over the past fifty years the author has observed that trees like Capparis decidua have been cut down from the dunes flanking the lower White Nile to provide timber for buildings and a durable superstructure for village wells. In more recent years even the ‘toothbrush tree’ Salvadora persica has not been spared, although some modest replanting is now under way. The prolonged unrest and internecine warfare in South Sudan has delayed any long-term prospects for forestry in this vast region, as is sadly also the case in Sudan’s Darfur Province. As happens all too often, unresolved social and political issues tend to stymie intelligent and ecologically sustainable use of natural resources.

5.4.2 Overgrazing Pastoral nomads are fully aware that it is not in their best interests to allow their animals to take more from the land than it can replenish, so that herd size is adjusted according to available fodder and ease of herd management. Provided a large enough area of land is potentially available for grazing, with watering points and wells located within easy walking distance, overgrazing is unlikely to occur. However, during years of prolonged

72

Vegetation, Land Use and Human Impact A

B

C

D

Vegetated

Bare

Figure 5.5 Schematic diagram showing progressive stages in the conversion of vegetated land to bare degraded land as a result of overgrazing and drought.

drought, unrest or draconian government land tenure policies, neither water nor grazing may be freely available, so that migration into the lands occupied by cultivators or in dire cases to the outskirts of towns may ensue. This can lead to localised destruction of the plant cover (Fig. 5.5). Another cause of overgrazing occurs when governments fence off large areas of traditional grazing land to create a national park, or wildlife sanctuary, or development project such as fields of irrigated cotton, sugarcane or similar cash crops. The almost inevitable result is increased grazing pressure on the now much curtailed land available for grazing. Bare soil and daily dry season dust storms are often the final outcome (Fig. 5.6). More imaginative and more sustainable policies are those allowing multiple use of the land, with the pastoralists being given a genuine role in park and wildlife management (Western, 2002). Where population pressure from humans and their herds has led to loss of topsoil, groves of the utterly unpalatable ‘ushar bush or Dead Sea apple (Calotropis procera) soon colonise the degraded land and are one of the early warning signs of loss of soil fertility and reduced soil moisture storage capacity. Jackals will nibble the otherwise toxic globular fruit as an emetic. Goats will not touch it. The milky white sap is poisonous and can also cause blindness. The soils developed on sand (locally known as qōz or gōz) are especially vulnerable to erosion by wind and water (Jewitt, 1950; Jewitt and Manton, 1954; El Tahir et al., 2004). Once the loose sandy topsoil has been eroded, leaving the underlying clay-enriched subsoil

5.4 Human Impact on Vegetation and Soils

73

Figure 5.6 Dust storm, Ethiopian Rift. The bare soil results from overgrazing.

exposed as in the Qoz Dango of western Darfur during the Sahel drought (Parry and Wickens, 1981), the less permeable subsoil becomes susceptible to gully erosion, a phenomenon widespread today along the southern margins of the Sahara in areas receiving in excess of 400–500 mm of precipitation annually (Daveau, 1965; Talbot and Williams, 1978, 1979). We elaborate on this issue in Section 5.4.4.

5.4.3 Salt Accumulation Large areas of the Nile Basin experience seasonal drought and in the northern third of the basin aridity prevails for most of the year. Evaporation exceeds precipitation for much or all of the year in the northern two-thirds of the region, so that the soils will tend to accumulate salt unless the salt is leached out during irrigation. If the irrigation canals are poorly maintained and prone to leakage, water from them will gradually accumulate in lowlying areas in the landscape, leading to waterlogging. During dry years or dry seasons the water will evaporate, and a saltpan will eventually form. The salt will kill all but the most salt-tolerant of plants. Salt can also accumulate in shallow depressions without the need for any human intervention. When the White Nile flooded much of its lower valley about 15–14 ka ago, a wide shallow lake developed during the wet season (see Chapter 8). A reduction in discharge ultimately caused the lake to recede, and former flood plain

74

Vegetation, Land Use and Human Impact

embayments became shallow wetlands, which later dried out to form saline depressions. These depressions or maya’a were subsequently buried beneath at least a metre of nonsaline alluvial clay (Williams, 1968b). In northern Sudan over 250,000 hectares of land are affected by salinity and sodicity (Mustafa and Saeed, 2004). Nevertheless, we need to bear in mind also that many of the soils in the drier parts of the Nile Basin are naturally saline and alkaline. To separate natural salt accumulation in soils from that triggered by human action is not always easy.

5.4.4 Desertification and Dune Reactivation A great deal has been written about the causes and consequences of desertification and the most appropriate remedies for this insidious form of land degradation in and beyond the Nile Basin (Obeid et al., 1982; Granger, 1990; Warren and Khogali, 1992; Mainguet, 1994; Thomas and Middleton, 1994; Kassas, 1995; Williams and Balling, 1996; Barakat and Hegazy, 1997; El Wakeel, 2004; Mustafa and Mahdi, 2004; Mustafa and Saeed, 2004; UNESCO, 2007; Williams, 2009b; Williams, 2014, chapters 24 and 26). Desertification has been defined as ‘a change to a more desertic condition’, which begs the question as to what is deemed ‘desertic’. The currently accepted international definition is not very helpful either. It states that desertification is ‘land degradation in arid, semi-arid and dry sub-humid areas resulting from various factors including climatic variations and human activities’ (UNEP, 1992a). Before June 1992 desertification was defined as ‘land degradation in arid, semi-arid and dry sub-humid areas resulting mainly from adverse human impact’ (UNEP 1992b). The Sahel drought that began in 1968 and lasted on and off for several decades was responsible for the change in attitude, an attitude all too ready to blame impoverished farmers and nomadic pastoralists for the damage to plants and land caused largely or entirely by lack of rain. Desertification has remained an elusive concept for several reasons. One is that it is often very hard to distinguish between the various impacts on the landscape brought about by natural processes as opposed to human activities. Another is that desertification involves a whole suite of processes – ones that affect soil, groundwater, plant cover, dune mobility and even microclimate. It seems more useful and certainly more informative to consider the observed consequences (Table 5.1) and then to work back to the probable causes – causes which encapsulate a myriad of intertwining social, economic, and political factors as well as purely physical processes of erosion (Table 5.2). A hotly debated issue is whether or not human activities can actually cause drought. The celebrated meteorologist Jules Charney and his colleagues proposed that overgrazing could alter the surface albedo (reflectivity) enough to increase the amount of long wave radiation reflected back to the atmosphere from the ground surface (Charney, 1975; Charney et al., 1975). This would lead to cooling of the ground surface and so reduce convection. Reduced convection would in turn mean less convectional rain. Less rain would mean a reduced grass cover, leading to a vicious spiral of diminishing precipitation

5.4 Human Impact on Vegetation and Soils

75

Table 5.1 Consequences of desertification • Accelerated erosion by wind and water • A decline in soil structural stability, with an ensuing increase in surface crusting and surface runoff, and a reduction in soil infiltration capacity and soil moisture storage

• An increase in the flow variability of dryland rivers and streams • A decline in soil organic matter and nutrient status with an attendant decline in crop and fodder • • • • • •

yields, and, in extreme cases, social disruption, famine and human and livestock migration Salt accumulation in the surface horizons of dryland soils An increase in the salt content of previously freshwater lakes, wetlands and rivers Replacement of forest or woodland by secondary savanna grassland or scrub A reduction in species diversity and plant biomass in dryland ecosystems An increase in dust particles An increase in carbon particles and trace gases from biomass burning

After Williams (2014, chapter 24, p. 484).

Table 5.2 Possible causes and consequences of desertification Causal factor

Consequences

Direct land use Overcultivation (shorter fallows; mechanised farming)

Physical processes affected Decline in soil structure and soil permeability; depletion of soil nutrients and soil organic matter; increased susceptibility to erosion; compaction of soil; sand dune mobilisation Overgrazing Loss of biodiversity and biomass; increased wind and water erosion; soil compaction from trampling; increased runoff; sand dune mobilisation Mismanagement of irrigated lands Waterlogging and salinisation of soil; lower crop yields; possible sedimentation of water reservoirs Deforestation Promotes artificial establishment of savanna (burning to clear land; fuel and fodder collection) vegetation; loss of soil-stabilising vegetation; exposed and eroded soil increases soil aridity; more frequent dust storms; sand dune mobilisation Exclusion of fire Promotes growth of unpalatable woody shrubs at the expense of herbage Indirect government policies Failed population planning policies Irrigation subsidies

May cause over-exploitative land use practices Increases need for food cultivation, and so may lead to over-exploitation of marginal land Exacerbates flooding and salinisation

76

Vegetation, Land Use and Human Impact

Table 5.2 (cont.) Causal factor

Consequences

Settlement policies/land tenure

Forces settlement of nomads, promotes concentrated use of land, which often exceeds the local carrying capacity Although often beneficial, can aggravate the problem by attracting increased livestock and human populations or increasing risk from salinisation; possible lowering of groundwater table below dams and around boreholes; silting up of reservoirs; waterlogging; promotes largescale commercial activity with little local benefit; flooding may displace people and perpetuate cycles of poverty Displaces subsistence cropping; pushes local people into marginal areas to survive; promotes less resilient monocultures; fosters expansion and intensification of land use Incentive to crop on marginal lands Forcing grazing or cultivation to levels beyond land capacity Valuable resources, both human and financial, are expended on war at the expense of environmental management and the needs of the people; large-scale migration with resultant increased pressure on receiving areas

Improved infrastructure (e.g., roads, large-scale dams, canals, boreholes)

Promotion of cash crops and push towards national and international markets

Price increases on agricultural produce High interest rates War

Natural Extreme drought

Extreme flooding

Ecological fragility

Decreased vegetation cover and land more vulnerable to soil erosion. Creates an environment that exacerbates overexploitation. Loss of arable land, houses and infrastructure; displacement of people; increased land use pressure on receiving areas Impact of land-use practices (will also depend on resilience of environment)

Modified from Darkoh (1989, 1999); Williams et al. (1995); Williams (2002, 2014).

and further reduction in plant cover. They called this hypothesis a ‘biogeophysical model’ of drought (Table 5.3). There seems little doubt that intense local overgrazing can causes changes in surface temperature, pressure, wind flow, gustiness and dust storms (Williams and Balling, 1996) (Figure 5.6). However, it is highly unlikely that this mechanism could

5.5 Controlling the Floods

77

Table 5.3 The ‘biogeophysical’ model of drought 1. 2. 3. 4. 5. 6. 7. 8.

Overgrazing reduces vegetation cover. Reduced plant cover increases albedo. Increased albedo decreases surface net radiation. Decreased surface net radiation results in surface cooling. Surface cooling promotes subsidence of air aloft. Subsidence decreases convection and cloud formation. Reduced convectional instability leads to less precipitation. Additional drying in the Sahel region leads to regional climatic desertification, which positively feeds back to 1. 9. Atmospheric general circulation models show that an albedo increase from 14% to 35% north of the Intertropical Convergence Zone (ITCZ) results in a southward shift of a few degrees in the ITCZ. 10. Rainfall in the Sahel region is thus decreased in the model by 40% during the rainy season. Proposed by Charney (1975) and Charney et al. (1975, 1977), modified from Williams and Balling (1996, p. 33) and Williams (2014, Table 23.5). This model may account for local droughts and dust storms, but does not explain the onset and end of major regional droughts.

cause regional drought. The onset of droughts over the Nile Basin and elsewhere in Africa is linked to global changes in sea surface temperature (Williams and Balling, 1996) as well as associated changes in atmospheric circulation modulated by ENSO events (see Chapter 3). The Charney model cannot explain the synchronous onset and termination of major droughts in the two hemispheres, nor can it account for the incidence of droughts across parts of Africa well before the increase in human populations and their herds (Verschuren et al., 2000). One very visible effect of desertification in the Nile Basin is the reactivation of previously vegetated and stable dunes. Dunes on either side of the lower White Nile have been reactivated during the last few decades and in places are advancing across former alluvial plains and seasonal stream channels (Williams et al., 2010; Williams et al., 2015a). Movement of once stable dunes in Kordofan Province has prompted the relevant authorities to explore and implement appropriate measures to halt sand movement and re-establish a stable plant cover (El Tahir et al., 2004). The marekh bush (Leptadenia pyrotechnica) has proven very effective in arresting sand flow, although if not carefully monitored it can take over from other plants.

5.5 Controlling the Floods: Dams, Reservoirs and Disease Construction of large dams (Fig. 5.7) is an expensive task that requires careful planning and a considerable investment of time and resources. The expected long-term benefits include flood management, water storage for irrigation, and electric power generation. These benefits need to be adequate to justify the infrastructure costs. In an ideal world, the benefits

78

Vegetation, Land Use and Human Impact 20 oE

30°E

40°E

MEDITERRANEAN SEA

ZIFTA

IDFINA

30o N

N

CAIRO

IDELTA EGYPT

ASYUT LIBYA

HAMMADI ASWAN LOW

RE D

ASWAN HIGH

Nil e

Lake Nasser

A SE

MEROWE

20°N CHAD

Atb ar

SUDAN

a

KHARTOUM

ERITREA

JEBEL AULIA

KHASHM EL GIRBA

White Nile

SENNAR ROSEIRES RENAISSANCE

10°N

So

SOUTH

Blu

ba t

e Nile ETHIOPIA

JONGLEI CANAL CENTRAL AFRICAN REPUBLIC

SUDAN

O

hr Ba

o

el Jebel Lake Albert

KENYA

OWEN FALLS

250 km

500

DA AN RW

0°N 0

Lake Edward

Lake Victoria

TANZANIA

Figure 5.7 Location of major dams in the Nile Basin.

SOMALIA

Lake Turkana

UGANDA

CONGO

m

5.6 Conclusion

79

should far outweigh any negative impacts, which is one of the reasons for conducting environmental impact assessments (EIAs). When the Aswan High Dam was first mooted, alarm bells resounded among archaeologists from around the world. There was a very real concern expressed that unique and irreplaceable temples and other priceless monuments to Egypt’s rich cultural history would be submerged beneath the proposed reservoir. UNESCO funded a major programme of salvage archaeology and careful excavation and transport of the major historic edifices to secure and higher elevations well beyond the reach of rising waters in the huge reservoir above the dam. Less obvious to many at that time but clear to a few prescient observers, including the renowned Egyptian plant ecologist and conservationist Professor Mohamed Kassas, were a multitude of other undesirable side effects (Kassas, 1972). These included trapping of Nile silts within the reservoir and concomitant coastal erosion along the delta (Nielsen, 1973), salt accumulation in soils downstream of the reservoir, the spread of water-borne diseases such as filariasis and schistosomiasis (Williams and Balling, 1996, pp. 137–140), the need for additional fertiliser to compensate for the loss of the annual silt, erosion of bridges and culverts downstream of the dam, and changes in Mediterranean sea water chemistry and current flow (Morcos and Messieh, 1973). A further important consequence is displacement and resettlement of the people whose homes and land were submerged beneath the rising waters of the reservoir (Fahim, 1973; Dafalla, 1975). On a more positive note, during the Sahel drought in the late 1960s and 1970s, when many thousands of people died of starvation along the southern margins of the Sahara and in Ethiopia, no single person in Egypt is recorded to have died of hunger.

5.6 Conclusion Natural vegetation zones in the Nile Basin are primarily determined by climate, especially the ratio of precipitation to evaporation, and the seasonal incidence and year-to-year variations in rainfall. At a more local scale, topography exerts a powerful control over the altitudinal zonation of trees and shrubs in mountainous areas in which temperatures become colder with increasing elevation, although precipitation tends to increase. In extreme situations, the tree line is controlled by temperature and is broadly coincident with the 10°C isotherm for the warmest month. Soils also have an important local influence. In semi-arid areas of the Nile Basin, trees growing on sandy soils only require about twothirds of the annual rainfall needed by trees growing on clay soils in the same latitude. (We need to consider this somewhat counter-intuitive finding when seeking to reconstruct past changes in tree cover, because these may reflect changes in sedimentation rather than climate.) Humans have had a long history of interaction with plants and animals in the Nile Basin, not always to the benefit of all concerned. Cattle herding began about 7–8,000 years ago in Nubia, but camel herding only became common during the last 2,000 years. Cultivation of rain-fed crops such as sorghum and millet in central Sudan and wheat and barley further north, is widely practised today. In the absence of a reliable source of Nile water, whether from irrigation canals or directly from

80

Vegetation, Land Use and Human Impact

the river, reliance on local precipitation can be precarious, especially during times of drought. Forestry is something of a Cinderella industry in much of the Basin, which is a concern, given the evidence for prolonged and widespread deforestation throughout the region. Desertification is often caused by overgrazing and by salt build-up in irrigated lands. Dams and reservoirs help to guarantee a reliable supply of water; the downside is the rapid spread of water-borne diseases such as onchocerciasis (river blindness), filariasis (spread by small black flies and mosquitoes) and schistosomiasis (parasitic liver flukes) into areas previously devoid of these scourges. Malaria will also proliferate where stagnant water allows mosquito larvae to breed with impunity.

6 The Ethiopian Highlands

The river had been considerably increased by rains, and fell in one sheet of water . . . with a force and noise that was truly terrible . . . A thick fume, or haze, covered the fall all around, and hung over the course of the stream both above and below. James Bruce of Kinnaird (1730–1794), Travels to Discover the Source of the Nile (1790, p. 95)

6.1 Introduction The headwaters of the Atbara/Tekezze and Blue Nile/Abbai rise in the Ethiopian Highlands (Fig. 6.1). These two rivers provide the bulk of the Nile sediment load and most of the Nile flood discharge. The Blue Nile provides 68% of the peak flow and 61% of the annual sediment load (140 ± 20 million tonnes) (Garzanti et al., 2006). The Atbara provides 22% of the peak flow and 35% of the annual sediment load (82 ± 10 million tonnes). Since the total annual Nile sediment load amounts to 230 ± 20 million tonnes (Garzanti et al., 2006), the importance of these two rivers is clear. To understand the reasons for past fluctuations in Nile discharge and sediment transport, we need to understand the history of the Ethiopian headwaters of the Nile. Any attempt to answer the question ‘How old is the Nile?’ must also begin in the headwaters.

6.2 Cenozoic Uplift and Volcanism The Afro-Arabian dome (Bowen and Jux, 1987) developed as a result of slow and prolonged crustal doming during the Oligocene. The elliptical dome was roughly 1,500 km wide and extended across Ethiopia and Yemen with its 3,000 km long axis oriented SSW–NNE and its centre above the Afar plume (Avni et al., 2012). Uplift caused sporadic reactivation of preexisting fractures, shear zones and other geological structures often simply termed lineaments. Many of these lineaments developed during the 550 Ma East African Orogenic events (Talbot and Williams, 2009; Fritz et al., 2013) and some extend back well into the Precambrian. The angular pattern of the main Nile reflects the structural control exerted by these features (Adamson and F. Williams, 1980; Adamson et al., 1993). The uplift also initiated a long

81

82

The Ethiopian Highlands Atba

to main Nile

ra

SUDAN

ERITREA

N Tek ezze

ba At

14°N

ra

Ra d ha

Ras Dashen 4543m

Din

Semien Highlands

L Tana

de

12°N

r Bl ue

le

Nile

Didessa

Awas h

Blue

10°N

Ni

ETHIOPIA

Ba ro

8°N

0 36°E

38°E

100 km

200

40°E

Figure 6.1 The Ethiopian headwaters of the Nile. (Shaded area above 200m).

episode of regional denudation, estimated to have lasted 6–10 million years, during which a major erosion surface developed, termed the ‘Oligocene regional truncation surface’ by Avni et al. (2012). Uplift accelerated in the late Oligocene and early Miocene, leading to rifts developing by 25 Ma in what became the Red Sea and Gulf of Aden (Corti et al., 2009), disrupting the original Afro-Arabian lithospheric plate and leading to the formation of what became two separate plates – the African plate and the Arabian plate (see Chapter 2). In Ethiopia uplift of what became the Ethiopian highlands was accompanied by widespread volcanic activity and the extrusion of numerous basalt flows, often termed the ‘Trap Series’. Hofmann et al. (1997) considered that most of these flood basalts were erupted within an interval of about 1 million years roughly 30 Ma ago. Prave et al. (2016) have recently obtained uranium–lead zircon ages of 31.1 Ma and 30.8 Ma for felsic ignimbrites and rhyolites bordering Lake Tana, which they interpret as fault blocks resulting from a super-eruption and associated caldera collapse at that time. They also obtained a single 40Ar/39Ar age of 0.033 ± 0.005 Ma (33 ka) for localised scoriaceous

6.3 Cenozoic erosion

83

basalts ‘that form the present-day dam and outflow of the lake’ (Prave, 2016, p. 2). A 90-m sediment core recovered from the lake has a 14C age of 17 ka at a depth of 10.2 m, prompting Marshall et al. (2011) to infer an extrapolated age for the base of the core of at least 150 ka. As a general rule, the sediments in a lake will be no older than the dam that forms the lake. Acting on the principle that a single swallow does not a summer make, we simply note that there is an unresolved discrepancy between the putative 33 ka age claimed for the dam and the >150 ka age proposed for sediment within the lake. Both ages should be viewed with caution.

6.3 Cenozoic Erosion: The Blue Nile and Tekezze Gorges The Tekezze and Blue Nile gorges are two of the most impressive in Africa. The Blue Nile gorge extends from its inception below Lake Tana to the western edge of the Ethiopian escarpment, a distance of roughly 350 km. In places the canyon is 20–30 km wide and 1,500 m deep, and cuts through Cenozoic basalts, Mesozoic and Palaeozoic sandstones, limestones, marls and gypsum down to the Precambrian basement rocks. McDougall et al. (1975) obtained the first potassiumargon ages (27–23 Ma) for the uppermost basalt flows in the gorge. They estimated that a total volume of rock amounting to 100,000 ± 50,000 km3 had been eroded from the two gorges from a combined drainage area of 275,000 km2. Dividing volume by the product of area and time gives a mean annual denudation rate of 15 ± 7.5 km3 per km2, or 12 ± 6 km3 per km2 if the more widely accepted age of 30 Ma is used, which is a very low rate of erosion for an area of high relief and active uplift, but reasonably common for lowland tropical rainforest regions, and points to long intervals of tectonic stability interspersed with shorter phases of rapid uplift. Of interest in this context is the recovery of abundant pollen grains characteristic of tropical lowland rainforests from lignites intercalated between basalt flows that were initially thought to be Miocene in age (Yemane et al., 1985) but now considered to be Oligocene and closer to 30 Ma (Kappelmann et al., 2003; Abbate, 2014). The basalts are now found near Gondar at 2,000 m elevation, suggesting a mean rate of uplift of 0.07 mm/yr during that time. On the south side of the Blue Nile gorge investigated by McDougall et al. (1975) the upper portion of the section consists of three basalt flows each several tens of metres thick extending laterally for more than 5 km. Talus obscures the lower part of the sequence but there are at least two more thick basalt flows. A dark cracking clay, or vertisol, up to 3 m thick lies beneath the second flow from the top and contains carbonised tree trunks and plant leaves of probable mid-Tertiary age. Diatomites are visible in places beneath the upper flow, indicating the presence of former lakes (Williams and Williams, 1980). As we saw in Chapter 2, the volume of the Nile Delta and much larger submerged Nile Cone amounts to about 550,000 km3 (Macgregor, 2012). This is substantially more than the estimate used by McDougall et al. (1975), who considered that the Nile Cone had a probable volume of 150,000 ± 50,000 km3 after taking into account the differences in bulk density (Nile Cone sediments: 1.5 g per cm3; Ethiopian bedrock: 2.8 g per cm3) and allowing for 30% losses of material in solution. The broad equivalence in volume between the amount of bedrock eroded from Ethiopia and the amount of sediment stored in the Nile Delta and Nile Cone led McDougall et al. (1975) to conclude that erosion in the Ethiopian headwaters of

84

The Ethiopian Highlands

the Blue Nile/Abbai and Atbara/Tekezze Rivers was the primary source of sediment in the Nile Cone, prompting the question as to when such erosion began. Later work has shown that much of the sediment in the Nile Cone is also derived from erosion of the Red Sea Hills (Macgregor, 2012). In addition, the detailed isotopic and geochemical analyses of Nile Cone sediments by Fielding et al. (2016, 2018) have demonstrated that there was a fluvial connection between Ethiopia and the main Nile some 30 million years ago. Pik et al. (2003, 2008) used thermo-chronology to evaluate models of Ethiopian landscape evolution. The results of their apatite helium ages were consistent with burial of the basement rocks beneath a thick cover of Trap Series flood basalts 30 million years ago. They concluded that erosion of the Blue Nile (and Tekezze) gorges began 29–25 Ma ago, confirming the pioneering work of McDougall et al. (1975). Pik et al. (2003) also argued for continuous uplift and continuous fluvial erosion since that time. Later work by Gani et al. (2007) involved compiling all existing potassium–argon ages for volcanic rocks in the northwest Ethiopian Highlands that are now erosional remnants and then using a digital elevation model to reconstruct the prevailing topography before erosion of the Blue Nile gorge. They estimated that at least 93,200 km3 of rock had been eroded from the Ethiopian Plateau since 29 Ma, and concluded that uplift had taken place in three main phases (29–10 Ma, 10–6 Ma and 6–0 Ma) with uplift and erosion accelerating at 10 Ma and at 6 Ma. Ismail and Abdel Salaam (2012) analysed river long profiles and slope gradients in the upper Tekezze and upper Blue Nile catchments, concluding that incision had been episodic since it began soon after 30 Ma, with an increase in incision at ca. 22 Ma and ca.11 Ma. They did note, however, that incision was more uniform towards the west, away from the influence of rift flank uplift. Present-day annual erosion rates calculated from the volume of sediment carried by the Nile amount to 120–240 m3 per km2, which most likely reflect the impact of historic deforestation and cultivation on steep slopes in the Ethiopian headwaters (Williams, 2009b). The estimated volume of sediment stored in the Gezira Fan between the Blue and White Nile, in the Atbara Fan and along the main Nile, amounts to about 1,800 km3, 800 km3 and 600–100 km3, respectively (McDougall et al., 1975). Compared to the total volume of rock eroded from Ethiopia these are trivial quantities, and indicate that the Nile has long been a very efficient conveyor of sediment from its Ethiopian headwaters to the eastern Mediterranean.

6.4 Miocene and Pliocene Environments in Ethiopia Apart from some sketchy information relating to episodic uplift and erosion, very little is known about the environmental history of the Ethiopian headwaters of the Nile during the long interval from Miocene (23–5.3 Ma) to Pliocene (5.3–2.6 Ma). We therefore need to draw on what evidence is available from other areas in Ethiopia such as the Awash Valley and the Southeast Highlands. The headwaters of the Awash lie 100–150 km southeast of the Blue Nile gorge. Tiercelin (1981) has mapped remnants of large Miocene lakes along the margins of the present Afar Rift. A deep freshwater lake occupied the Middle Awash Valley intermittently during the early Pliocene until a major eruption and associated seismic activity breached the dam that

6.5 Quaternary Environments

85

was impounding this early Pliocene lake 4 Ma ago (Williams et al., 1986). The level of the lake fluctuated from full and fresh to shallow and swampy (Adamson and Williams, 1987). The Middle Awash Valley is well known internationally for the fossil hominids retrieved from sediments within the Valley. Considerable effort has gone into reconstructing the environments in which these hominids lived. Abundant vertebrate fossils, less abundant plant remains (pollen, phytoliths), fossil soils, analysis of stable carbon and oxygen isotopic composition of pedogenic carbonate nodules have all helped to shed some light on the habitats occupied by the Australopithecines, their predecessors and their successors (Taïeb, 1974; WoldeGabriel et al., 1994; Kalb, 1995; Cerling et al., 1997, 2010, 2011; Barboni et al., 1999; WoldeGabriel et al., 2001; Quade et al., 2004; Wynn et al., 2006; WoldeGabriel et al., 2009). What has emerged from this work is a portrait of a mosaic pattern of woodland, grassland and dense riparian forest vegetation very different from the present desert scrub away from the Awash River, and confirmation of a far less arid climate than today during the late Miocene and Pliocene (Bonnefille, 1972; Bonnefille et al., 2004). On the upland plains of Gadeb at 2,300 m elevation in the Southeast Highlands, a lava flow dammed the ancestral Webi Shebeli River 2.7 Ma ago and created a lake that persisted until 2.35 Ma (Williams et al., 1979; Eberz et al., 1988). Pollen within the lake sediments indicated a regional fall in temperature around 2.35 Ma ago (Bonnefille, 1983), heralding the start of the Quaternary ice ages (Williams et al., 1998). The diatoms within Pliocene Lake Gadeb show progressively drier conditions during the lifetime of the lake (Gasse, 1980). The late Cenozoic in Ethiopia was a time of progressive cooling and desiccation. Evidence from the fossil remains of both plants and animals indicates wetter climatic conditions during both the Miocene and Pliocene. A drop in regional temperature at about 2.35 Ma reflects the impact of the Quaternary glacial–interglacial cycles with their associated oscillations in precipitation and temperature.

6.5 Quaternary Environments 6.5.1. Quaternary Lake Fluctuations Provided one can clearly identify the lake shorelines, lakes are potentially among the best archives for carrying out water balance calculations and for reconstructing past hydrological budgets (Williams, 2014, pp. 192–193). The pioneering work on lakes in Ethiopia was that of Françoise Gasse, who worked primarily on the Quaternary lake fluctuations in the Afar Rift, and of Alayne Street, who worked on lakes in the Ethiopian Rift Valley (Gasse, 1975; Gasse and Street, 1978; Street, 1979). Gasse used diatoms in particular to determine past changes in water depth, salinity, alkalinity and temperature (Gasse, 1976), and over the years developed a set of transfer functions applicable to lakes across Africa (Gasse et al., 1995, 1997). The Afar lakes presented some interesting problems. Some fluctuated in response to changes in runoff, some to changes in groundwater levels and some to a mixture of both. Although some of the initial radiocarbon dating is no longer considered reliable, Gasse (1975, 1976) was able to show that the Afar lakes had fluctuated widely

86

The Ethiopian Highlands

during the late Quaternary, but all revealed that the Last Glacial Maximum (LGM) (21 ± 2 ka) was a time of extreme aridity during which lakes dried out or became highly saline. Unravelling the history and causes of older (Pliocene–Pleistocene) lake fluctuations has proven more complex in this region of active volcanoes and frequent earthquakes and often tells us more about tectonic and volcanic activity than about past changes in climate (Gasse et al., 1980a, 1980b; Gasse, 1990). A finer resolution record is available for Holocene lake fluctuations in the Afar and Ethiopian Rift, and East Africa more generally, with the first half of the Holocene appearing generally wetter than the last 4,500 years (Butzer et al., 1972; Delibrias et al., 1973; Fontes et al., 1973; Fontes and Pouchan, 1975; Williams et al., 1977, 1981; Abell and Williams, 1989; Gasse and Fontes, 1989; Gasse, 2000a, 2000b; Chalié and Gasse, 2002; Garcin, 2006; Garcin et al., 2006a, 2006b, 2007). More recent work on sediment cores recovered from the submerged bed of Lake Tana has provided critical new information that is directly relevant to understanding changes in the late Quaternary history of the Blue Nile (Lamb et al., 2007; Marshall et al., 2011; Costa et al., 2014). The Blue Nile (or its longest tributary the Little Abbai) has its source in springs at Sakala, now known as Gish Abbai, at an elevation of 2,897 m (Hurst and Philips, 1931; Hurst, 1952) and then flows north for roughly 100 km before reaching Lake Tana at an elevation of 1,829 m. Lake Tana is a broad shallow lake 78 km long, 67 km wide and 14 m deep. On emerging from the lake, the river (or Great Abbai) flows southeast for 30 km until the Tisisat Falls that so impressed James Bruce more than two centuries ago (see quotation at the start of this chapter). The river then drops 50 m and enters a narrow canyon cut into the basalt, deepening its bed progressively over a distance of about 330 km until the canyon is 1,500 m deep and 30 km wide. Below Lake Tana a number of other major tributaries such as the Didessa contribute a considerable volume of sediment and water to the Blue Nile. Lamb et al. (2007) demonstrated that Lake Tana had been dry during the LGM, with the corollary that flow in the Blue Nile would have been less and even more seasonal at that time. Overflow resumed soon after 17 ka and was well under way by 15–14 ka. Using variations in the titanium content in sediment cores from Lake Tana, Marshall et al. (2011) inferred that the lake was low at 13–12.5 ka, 8.4 ka, 7.5 ka and especially at 4.2 ka, with an overall reduction in discharge after 6.8 ka. The intervals in between these regression phases were times of higher lake level. These results are in good general accord with those of Blanchet et al. (2013, 2015) from the distal end of the Nile. Sediments in a 6 m-long core retrieved from a depth of 700 m on the Nile deep-sea fan provide an indirect measure of Holocene fluctuations in Blue Nile discharge and sediment yield. Using bulk element composition and strontium–neodymium sediment ratios, Cécile Blanchet and her colleagues inferred relatively high runoff and sediment input from the Blue Nile in the Early (9.5–7.3 ka) and Late Holocene, when spring insolation was higher, with a reduction in runoff and sediment between 8 ka and 4 ka, when autumn insolation was high. They also noted evidence of aridity during 8.5–7.3 ka and 4.5–3.7 ka. They concluded that this latter arid phase marked the end of the so-called African Humid Period (AHP). Costa et al. (2014) used the deuterium content of leaf waxes in sediment cores from Lake Tana to argue that the onset of wet conditions in the area surrounding the lake began several

6.5 Quaternary Environments

87

thousand years earlier than the onset of wet conditions at Lake Challa in Tanzania. They therefore rejected the notion that the postglacial onset of wet conditions (and the start of the AHP) were synchronous in East Africa. We return to this issue in Chapter 7.

6.5.2 Glacial and Periglacial Landforms and Sediments A number of high mountains in Ethiopia show evidence of recent glaciation. These include the Semien, Arussi and Bale Mountains (Fig. 6.1) as well as certain more isolated peaks (Potter, 1976; Hastenrath, 1974, 1977; Hurni, 1982; Osmaston et al., 2005; Williams et al., 2015a). Air photos were initially used to map the extent of glaciation (Potter, 1976; Hastenrath, 1977), a process that can lead to errors in interpretation. In the case of the Semien Mountains, Hans Hurni (1982) also carried out sustained and detailed fieldwork. His maps combine careful field observations with detailed topographic mapping and show the location of both glacial cirques and moraines with great accuracy. Evidence of glaciation includes both erosional and depositional landforms. The erosional landforms include glacial valleys, glacial cirques and nivation hollows, and glacially striated rock pavements. Depositional landforms include lateral, terminal and ground moraines, all of which can be directly dated using cosmogenic nuclides such as 10Be and 36Cl. According to Osmaston et al. (2005), a central ice cap at least 30 km2 in area covered the Bale Mountains during the last glacial, when at least 180 km2 of the massif underwent glaciation. On Mount Badda (4,170 m) an ice cap is estimated to have covered 140 km2. The Semien Highlands (Fig. 6.1) appear to have been covered by small cirque glaciers only, at least during the latest glaciation. For snow to persist and eventually form glaciers and ice caps, two things are essential: adequate precipitation in the form of snow, and cold temperatures in both winter and summer. Gasse et al. (1980a) noted that lake levels in the Ethiopian and Afar Rifts were uniformly low during the LGM and suggested that peak ice cover probably occurred towards 30 ka rather than during the cold and dry LGM, when precipitation would have been restricted. Cosmogenic nuclide 36Cl ages obtained from geologically recent moraines on Mount Kenya and Mount Kilimanjaro cluster around 30 ka (Mark and Osmaston, 2008) and so appear to support this hypothesis. However, the radiocarbon ages obtained by Osmaston et al. (2005) for organic remains immediately postdating glacial landforms in the Bale Mountains point to LGM glaciation, as do the 36Cl ages obtained recently from moraines on two of the highest peaks in the Semien Mountains (Williams et al., 2015a). These ages are discussed in detail in Chapter 12. Deglaciation seems to have occurred by about 13–14 ka on Mount Badda. Alayne Street collected a peat core 3 m long from a small cirque basin on Mount Badda at 4,040 m elevation. Analysis of the pollen and diatoms in the core and 14C ages for the basal 10 cm show that the ice had gone from this site by 13.5 ka (Hamilton, 1977; Gasse, 1978; Gasse et al., 1980a). Because the accumulation of snow depends on the two independent variables of temperature and precipitation it is often hard to distinguish which of the two is the more important. Did snow fall and persist because temperatures were particularly low or because precipitation

88

The Ethiopian Highlands

in the form of snow was especially high? To answer this question, we need an independent measure of past temperature changes, for which we can use the evidence from periglacial deposits. A number of distinctive landforms reflect the influence of periglacial freeze-thaw processes, and it is sometimes easy to confuse periglacial deposits with debris flows and angular colluvium, as Hurni (1982) has noted. A number of characteristics are diagnostic of true periglacial deposits when seen in the field. The periglacial deposits are generally poorly sorted, with the long axes of the angular clasts within them oriented downslope. They become thicker and more widespread upslope, and tend to be finer at depth and coarser near the surface. In the Semien Highlands there is a clear altitudinal distribution of presently active periglacial landforms, with frost shattering active today only above 4,250–4,300 m elevation (Hastenrath, 1974; Williams et al., 1978). However, frost-shattered angular rubble can be seen at elevations down to 3,100–3,750 m (Hastenrath, 1974; Williams et al., 1978), indicating colder temperatures when the rubble mantles were last active. Freezing temperatures would have been needed to shatter the bedrock at these lower elevations. Using the presentday mean temperature lapse rate of 0.6°C/100 m measured for the uplands of East Africa, including Ethiopia, Williams et al. (1978) calculated that temperatures were between 4°C and 8°C lower when the angular rubble was laid down. As a very rough rule of thumb, the lower limit of periglacial solifluction deposits coincides with the upper limit of the tree line, which in turn corresponds to the 10°C isotherm for the warmest month. At the time that the angular rubble was accumulating along the hill slopes in the Semien Highlands, the tree line would have been about 600–1200 m lower than it is today.

6.5.3 Quaternary Changes in Plant Cover In contrast to other parts of the world such as the Bolivian Altiplano and the maar lakes of tropical northeast Queensland in Australia, there are no pollen-bearing sites in Ethiopia with long continuous records of past vegetation changes. The evidence that does exist is fragmentary and not always precisely dated. Present-day vegetation in the Ethiopian uplands is zoned according to altitude (Schaller and Kuls, 1972, map 3), and would have responded to past changes in temperature linked to the Quaternary glacial–interglacial cycles. At lower elevations precipitation becomes more important than temperature as a control on the type of vegetation. Changes in precipitation were also linked to the glacial–interglacial cycles, with times of peak aridity seeming to coincide with times of maximum cold, at least during the late Quaternary. We can speculate that during the interglacial phases, when the climate was warm and wet, plant cover would have been at a maximum at all elevations. During times of glacial aridity, plant cover would become more xeric and sparser in the lowlands, and the different vegetation zones in the uplands would be at lower elevations than they are today. One example will suffice to show the sort of information that can be gleaned from even less than ideal sediments. The prehistoric site of Melka Kontouré lies some 150 km from the headwaters of the Awash River and is located atop the western escarpment of the Ethiopian Rift, at an elevation of about 2,000 m. The clearly stratified archaeological sediments contain

6.5 Quaternary Environments

89

a wealth of stone tools and fossil bones of Lower and Middle Pleistocene age (Chavaillon, 1976) and also contain fossil pollen (Bonnefille, 1976). The site is within easy walking distance of the river, which was no doubt fordable during some or all of the time it was occupied – the word melka being Amharic for ford. The pollen evidence indicates a cooler, wetter climate during the Oldowan occupation around 1.75–1.6 Ma, with abundant arboreal pollen of the montane trees Podocarpus and Juniperus although more than half the pollen grains were of grass pollen. A drier climate set in during the Middle Pleistocene, with Acacia and Combretum savanna replacing the previous montane forest and grassland. A period of increasing aridity ensued, which gave way about 0.7 Ma to a less arid climate and an Acacia savanna, much as is present there today, albeit in relict form owing to moderately intense land use. None of this evidence is particularly well dated and the age estimates have large error terms. Nevertheless, there is a picture emerging of fluctuating temperatures and precipitation superimposed upon progressively drier conditions during the course of the Pleistocene.

6.5.4 Quaternary Erosion, Weathering and Soil Formation Elevation exerts a major control on precipitation, temperature and evaporation in Ethiopia. Figure 6.2 shows how precipitation increases steadily as we ascend from Gewani (618 m elevation; mean annual precipitation [P] is 410 mm; mean annual evaporation [E] is 1,055 mm) in the southern Afar Rift through Metahara (947 m; P is 558 mm; E is 983 mm) in the Ethiopian Rift and then through Wonji (1,618 m; P is 789 mm; E is 855 mm) on the Rift escarpment to Addis Ababa (2,355 m; P is 1,089 mm; E is 736 mm) in the Ethiopian Highlands. Temperature likewise decreases with elevation, as does evaporation. Precipitation exceeds evaporation above about 1,700 m elevation, and is exceeded by evaporation below that elevation. During times when the temperature was much colder, such as during the LGM at 21 ± 2 ka ago, the altitudinal zone in which precipitation (P) was equal to evaporation (E) would have been much lower. If annual rainfall was reduced, the elevation at which P = E would have been higher (Fig. 6.2). At elevations around 1,700 m where precipitation and evaporation are evenly balanced rates of weathering, erosion and soil formation are likely to be in equilibrium. Where E>P, chemical weathering and soil formation will be limited by the seasonal availability of soil water, and at high elevations physical weathering will tend to be more important than chemical weathering, and plant growth and soil formation will both be temperature limited.We can now use these general propositions to develop a simple model of how Quaternary fluctuations in temperature and precipitation might have influenced how the Ethiopian headwaters of the Nile responded to past changes in climate, plant cover, erosion, weathering and soil formation.

6.5.5 Quaternary Fluctuations in River Flow and Sediment Transport Figure 6.3 is a schematic model of the factors that control dissolved load in river and Figure 6.4 illustrates the factors that control sediment load in rivers. Dissolved load depends

90

The Ethiopian Highlands Present-day precipitation Present-day evapotranspiration

A

A

A

A

Elevation (m)

2000

W

M

1000

M

WW

W

M

M

A

Addis Ababa

W Wonji G

G

G

G

Hypothetical precipitation at 18,000 BP Hypothetical evapotranspiration at 18,000 BP

M

Metahara

G

Gewani

0 500

1000 Precipitation / Evapotranspiration (mm)

1500

Figure 6.2 Changes in precipitation and evaporation with elevation: east-central Ethiopia (adapted from UN, 1971, p. 43 and Williams and Adamson, 1980, Fig. 12.5) and hypothetical situation at the LGM (21 ka) (after Williams and Adamson, 1980, Fig. 12.5).

on rock type, runoff, soil and vegetation cover and temperature. Sediment load depends on certain key factors such as rates of tectonic uplift, rock type, slope angle and slope length, rainfall amount, seasonality and intensity and vegetation cover. On steep slopes, rapid mass movement in the form of landslides and debris flows will be a major contributor of sediment to river channels. In the Blue Nile gorge today major landslips are very evident (the previous bridge across the river was grouted in an immense landslide and began to crack underneath: Fig. 6.5a and 6.5b) and large blocks of sandstone fall sporadically from the vertical cliffs of the Adigrat Sandstone near the bottom of the gorge (Fig. 6.6) as well as blocks of basalt from the basalt cliffs near the top of the gorge. Slower forms of mass movement such as soil creep and surface rock creep are proportional to the sine of the angle of slope, regardless of whether movement is triggered by freeze–thaw processes or by rain splash erosion and runoff. Raindrop impact is proportional to the total momentum of any given rainstorm, with momentum determined by drop size and terminal velocity (Williams, 1969). On bare slopes soil loss from rain splash erosion can be up to forty times greater per unit momentum than on well-vegetated grassy slopes. Trees offer little protection if the surface beneath them is bare. Erosion by overland flow will tend to increase with slope length and gradient unless the soil surface is highly permeable and/or covered in a dense plant litter. Hurst (1952) pointed out that much of the upper catchment of the Blue Nile, Rahad, Dinder and other Ethiopian tributaries of the Nile lies between 1,835 m and 3,060 m in elevation, placing the headwaters somewhat below the lower limit of periglacial solifluction

6.5 Quaternary Environments Tectonic History

Climate

Relief

Rainfall

Rock type

Soil

Basin form

Plant cover

91

Runoff

Erosion

Sediment load

Figure 6.3 Factors controlling dissolved load in rivers. (After Williams and Balling, 1996, p. 95.)

Temperature

Rock type

Weathering

Rainfall

Evapotranspiration

Plant cover

Soil

Groundwater

Runoff

Discharge

Dissolved solids

Figure 6.4 Factors controlling sediment load in rivers. (After Williams and Balling, 1996, p. 94.)

during the late Pleistocene. During cold and dry glacial and stadial intervals, with intense mechanical weathering, reduced plant cover and a more seasonal flow regime, the headwaters would receive an abundant supply of poorly sorted and often angular coarse debris

92 (a)

The Ethiopian Highlands (b)

Figure 6.5 The former bridge viaduct across the Blue Nile gorge (a) caused by movement of the very large landslide in which the viaduct foundations were grouted and (b) seen from underneath showing cracking

Figure 6.6 Impact of falling blocks of Adigrat Sandstone near the bottom of the Blue Nile gorge at the police post on the north side of the Blue Nile bridge. The damaged drums are 44 gallons/200 litres petrol drums.

6.7 The Early Holocene Blue Nile

93

and so would operate as bed-load streams, with wide, shallow unstable channels transporting and depositing coarse sand and gravel in their distal reaches. Conversely, during warm and wet interglacial and interstadial phases, with a replenished plant cover, active chemical weathering and soil formation, the slopes would furnish much finer sediment to the rivers in the form of silt and clay with minor amounts of fine and medium sand. The stream channels would function as suspension load channels in their lower reaches, with relatively deep and narrow channels, stable banks and a tendency to meander. They would also have a much higher load of dissolved solids than during times of cold dry climate in the headwaters. We are now in a position to apply these general principles to provide depositional models for the Late Pleistocene and Early Holocene Blue Nile.

6.6 The Late Pleistocene Blue Nile During the very late Pleistocene, between about 25 ka and 17 ka, lake levels throughout Ethiopia were very low and many lakes dried out altogether. In the Ethiopian Highlands, small ice caps and cirque glaciers occupied the highest ground until about 15 ka. Temperatures in the Semien Highlands were 4–8°C lower than today during the Last Glacial Maximum, and the tree line was correspondingly about 600–1200 m lower than it is today. Periglacial processes, including frost shattering of exposed bedrock (‘gelifraction’) and slow downslope movement of slope mantles under the influence of freeze–thaw processes and gravity (‘periglacial solifluction’) were active down to about 3,000 m elevation in the Semien Highlands. We can therefore envisage that the late Pleistocene Blue Nile was a highly seasonal river, with a reduced annual discharge, but capable of transporting a bed-load of coarse sand and gravel out onto its alluvial fan in the Gezira region of central Sudan (see Chapter 11) and even as far as northern Sudan and southern Egypt, where gravel alluvial terraces crop out sporadically above the Holocene clay flood plains (see Chapter 14). On reaching the lowland plains of central Sudan the Blue Nile formed a series of distributary channels that radiated across the Gezira alluvial fan. During the dry winter months, strong winds deflated the channel sands and point-bars, and a suite of source-bordering dunes developed along the distal reaches of the Blue Nile distributary channels. The flood regime of the Blue Nile at that time was probably akin to that of the modern Atbara River, which dries out in its lower reaches during the dry winter months (or did until the Khashm el-Girba dam was completed in 1964). Nevertheless, the Atbara, which has onethird the annual discharge of the White Nile (10 km3 against 33 km3), still carries twenty times more sediment to the main Nile during the summer floods than does the White Nile.

6.7 The Early Holocene Blue Nile Williams and Adamson (1980) noted that as post-glacial temperatures and precipitation increased in Ethiopia, montane forest re-occupied the former periglacial zones in the

94

The Ethiopian Highlands

Table 6.1a The Blue Nile at 21 ± 3 ka • • • • • • • • • • • •

Glacial cooling and aridity throughout Ethiopia. Reduced summer rainfall and weaker summer monsoon. Small cirque glaciers above 4,000 m in the Semien Mountains. Periglacial limits 600–1,200 m lower in the Semien Mountains. Winters 4–8°C cooler than today. Treeline 600–1,200 m lower than today in the Semien Mountains. Slopes unstable down to 3,000 m, with periglacial solifluction mobilising abundant coarse debris. Savanna-desert ecotone higher. Reduced annual discharge but high summer peak flows. Blue Nile distributary channels radiate across the Gezira alluvial fan. Distributary channels transport and deposit sand and gravel. Dry season deflation of sandy point-bars leads to formation of source-bordering dunes at distal end of distributary channels. • Flow to the main Nile highly seasonal with much reduced winter discharge.

Table 6.1b The Blue Nile at 12 ± 3 ka • • • • • • •

Postglacial warming and stronger summer monsoon. Longer wet season and increased summer rainfall. Warmer winters and cirque glaciers melt and disappear after 15 ka. Slopes vegetated and stable well above 3,000 m. Lowland savanna replaces semi-desert scrub along the lower Blue Nile valley. Volcanic tuffs and basalts weather to form clay soils in the uplands. Seasonal erosion of these clays soils provides the Blue Nile with an abundant suspension load of silt and clay. • Higher annual discharge, enhanced base flow, attenuated flood peaks but Blue Nile still highly seasonal. • Perennial flow and prolonged widespread flooding in central Sudan by sinuous and straight distributary channels. • High seasonal discharge into the main Nile with seasonally high suspension load. After Williams (2014), p. 183 and Williams et al. (2015a), p. 101.

Ethiopian Highlands. Pollen analysis has since confirmed these inferences (Umer et al., 2007). Chemical weathering became active once more and red kaolinitic and black montmorillonitic clay soils developed on the basalts and volcanic tuffs in the Ethiopian uplands. Today, at an elevation close to 3,600 m in the Semien Mountains, dark colluvial-alluvial silty clay loams up to 1.5 m thick either lie above or abut against angular blocks of frostshattered basalt. Higher still, at elevations above 4,000 m, highly organic loams and

6.7 The Early Holocene Blue Nile

95

Figure 6.7 (a) Depositional model for the late Pleistocene Blue Nile. (Adapted from Williams and Adamson, 1980, Fig. 12.4a). (b) Depositional model for the early Holocene Blue Nile. (Adapted from Williams and Adamson, 1980, Fig. 12.4b.)

peaty loams over a metre thick occupy ill-drained hollows on the upper flanks of the highest peaks. One such bog site was drilled by the writer on 30 November 2009 on Mount Mesareriya (13°12.958 0 N, 38°13.1900 E; elevation: 4,161 m) and yielded 14C ages of 292 ± 25 BP at 0–5 cm depth (WK-30890); 620 ± 25 BP at 50–60 cm depth (Wk-30891) and 639 ± 25 BP at 100–110 cm depth (Wk-30892). The important point is that even at high elevations finegrained slope mantles are common, and would have been from Early Holocene times onwards. The Holocene depositional model proposed by Williams and Adamson (1980, pp. 289–290) for the Blue Nile ‘is that of a seasonal river of high summer discharge carrying a sizeable suspended load of clay and silt together with a still significant bed-load of sand. Seasonal deposition in the swampy plains of the central Sudan would depend on the extent

96

The Ethiopian Highlands

and duration of flooding, and on the trapping action of the vegetation growing along the Blue Nile flood-plain south of Khartoum’. Figure 6.7 and Table 6.1 summarise this model.

6.8 Conclusion Initial incision of the Abbai/Blue Nile and Tekezze/Atbara Rivers began soon after extrusion of the Trap Series basalts some 30 Ma ago. Uplift of the Ethiopian Highlands was episodic, with an acceleration of uplift from terminal Miocene/early Pliocene times onwards. The volume of rock eroded from the present-day Tekezze and Blue Nile gorges amounts to 100,000 ± 50,000 km3 and provided a significant component of the total volume of the Nile Cone during the past 30 Ma. At the present time, the Atbara provides 22% of the peak flow and 35% of the annual sediment load reaching the Nile, while the Blue Nile provides 68% of the peak flow and 61% of the annual sediment load. Limited and fragmentary evidence from the Middle Awash Valley in the Afar Rift is consistent with wetter conditions in the Ethiopia uplands during the very late Miocene and Pliocene. Diatom and pollen data from late Pliocene–early Pleistocene Lake Gadeb in the Southeast Highlands indicate progressively drier and colder climatic conditions between 2.7 Ma and 2.35 Ma. Pollen retrieved from Melka Kontouré prehistoric site show a change towards drier conditions from about the Middle Pleistocene, although here too the evidence is fragmentary and not rigorously dated. The combined evidence from well-dated late Pleistocene and Holocene lake level fluctuations and less well dated late Pleistocene glacial and periglacial landforms is consistent with glacial aridity, with a cold and dry Last Glacial Maximum (21 ± 2 ka) during which lakes were low or dry and the uplands above 4,000 m supported modest cirque glaciers and small ice caps. Soon after 17 ka, and especially from about 15 ka until about 4.5 ka, the climate became warmer and wetter throughout Ethiopia. These climatic changes were reflected in the flow regime and type of sediment carried by the Blue Nile. The late glacial Blue Nile was a highly seasonal bed-load river that ferried impressive amounts of sand and gravel across its alluvial fan in its distal reaches (see Chapter 11). The Early and Middle Holocene Blue Nile was very different. It was a less seasonal suspension-load river carrying mostly silt and clay, with minor amounts of sand. Unlike the wide and shallow late glacial river, its channel was deep and narrow, with stable banks of clay and a sinuous channel pattern in the plains of central Sudan. These climatically controlled oscillations in river dynamics reflect the influence of glacial–interglacial cycles operating at time scales of 100 ka and 20 ka, which ultimately determined changes in precipitation, temperature, weathering, plant cover, soil formation and hill slope erosion.

7 The Ugandan Lake Plateau

The whole pent-up volume of water dashes out of a ravine like a burst water-main; it is really more of an explosion of water than a fall, and it can exert a curious mesmerism in the mind if one stands there and watches for a while. The pattern of thundering water is endlessly repeated yet never for two seconds quite the same. Alan Moorehead, The White Nile (1971, p. 105) (On visiting Murchison Falls and the Victoria Nile in Uganda)

7.1 Introduction The great lakes of Uganda and the permanent snows of the Ruwenzori Massif are the actual counterparts of the mythical Mountains of the Moon, which were long believed to be the source of the Nile. To the extent that runoff from Ruwenzori contributes water throughout the year to the lakes from which flows the Bahr el Jebel, which feeds into the White Nile, the enduring legend of the Mountains of the Moon has a strong basis in fact. The lakes owe their origin to late Cenozoic rifting, drainage reversal and sequential formation of the western branch of the East African Rift. Lake Victoria is but the latest in a series of drainage impoundments and is still subject to tectonic tilting. Present evidence, as yet not precisely dated, suggests an age of less than 0.5 Ma for Lake Victoria, which could imply a similarly young age for the White Nile.

7.2 Cenozoic Disruption of Drainage The pre-Miocene drainage in Uganda (Fig. 7.1) appears to have flowed from the more elevated regions in the east westwards across what is now the Ugandan lake plateau and on into the Congo Basin (Cooke, 1958; Goudie, 1985; Talbot and Williams, 2009). This wellintegrated drainage system was severely disrupted by the uplift and rifting processes that gave rise to the East African Rift in this region. The rift developed two separate arms flanking what are now the central plateau and the Lake Victoria basin (Fig. 7.2). Lakes Albert and Edward occupy the western arm of the rift in the north. The Semliki River flows

97

98

The Ugandan Lake Plateau

Uba

ngi

Co nti nen tal Divide

?

Congo

n Co

go

Atl ant

0

km

100

ic

Late Mesozoic to Miocene N ile

Uba

ngi

Lake Kyoga

Congo

0°

0° Lake Victoria

Co

Western Rift

o ng

Atl

Continental Divide

ant ic

Pleistocene to present

Lake Tanganyika

Figure 7.1 Hypothetical drainage pattern before and after the onset of rifting in Uganda. (Adapted from Talbot and Williams, 2009, Fig. 2.)

south in the central part of the western arm of the rift, which is occupied further south by Lake Tanganyika and Lake Malawi (Fig. 7.2). The Kenyan lakes Magadi, Naivasha, Elmenteita and Nakuru occupy the eastern arm of the Rift Valley, which continues north via lakes Bogoria, Baringo and Turkana to merge into the main Ethiopian Rift Valley (Fig. 7.2). Talbot and Williams (2009) have reviewed in some detail the complex history of the earth movements that led to the present drainage pattern and distribution of lakes. These tectonic movements remain active to this day, and the Lake Victoria basin, for example, has been tilted progressively down towards the east, leading to exposure of late Pleistocene lake sediments along its western margin.

7.2 Cenozoic Disruption of Drainage

99

Nile

Omo

Southern extremity of Main Ethiopian RIft

White Nile

L. Albert

ana

Albert Nile

u rk L. T

N

L.Chew Bahir

L. Kyoga

Semliki R.

0°

Mt. Elgon Lake Victoria

EASTERN

RI FT

L. Edward

RIFT

Ruwenzori Mts

Virunga volcanoes

L. Eyasi

Mt. Kenya L.Nakuru L.Elmenteita L.Naivasha L.Magadi L.Natron Mt. Kilimanjaro L.Manyara

OCEAN

Lake

WEST

ERN

L. Kivu

L.Baringo L.Bogoria

n Ta ga ny

L. Rukwa

ika INDIAN

L. Mweru

10°S Lake

L. Bangweulu

Malawi

0

km 30°E

500 40°E

Figure 7.2 The East African Rift and its western and eastern branches. (Adapted from Beadle, 1974, Fig. 3.4 and 14.1; Westcott et al., 1996, Fig. 2; and Johnson, 1996, Fig. 1.)

100

The Ugandan Lake Plateau

7.3 Origin of the Ugandan Lakes The Lake Albert basin contains a considerable thickness (ca. 5 km) of late Miocene and Pliocene sediments. Prior to the inception of Lake Albert, a much larger lake – ‘Paleolake Obweruka’ – occupied much of the western arm of the Rift and was a precursor to late Miocene Lake Albert (Van Damme and Pickford, 2003; Van Damme and Van Bocxlaer, 2009) (Fig. 7.3). The overflow from Lake Albert is essential in maintaining perennial flow in the White Nile. Lake Victoria (Fig. 7.4) is a much younger lake and at present provides about 90% of the water that flows via Lake Kioga and the northern edge of Lake Albert to the Bahr el Jebel and the White Nile. We have no clear idea as to when this link was forged. Lake Victoria is a comparatively young lake, despite being the largest lake in Africa in terms of area. It is also somewhat unusual in that it creates its own climate, with at least half of the precipitation over the lake coming from water evaporated from it (Sutcliffe, 2009). The maximum thickness of sediment within Lake Victoria amounts to only about 60 m (Johnson et al., 1996). Extrapolating dated rates of Holocene and late Pleistocene sedimentation would give an approximate age of 0.5 Ma for Lake Victoria (Johnson et al., 1996,

Marsh Rift margin

N

Palaeolake Obweruka

Palaeolake Manonga Palaeolake Tanganyika

200 km

Figure 7.3 ‘Paleolake Obweruka’ as inferred by Van Damme and Pickford (2003). (Adapted from Talbot and Williams, 2009, Fig. 3.)

7.4 Fluctuations of Lakes Victoria and Albert Albert Nile

101

Murchison

2°N

Falls Lake Albert Lake Kyoga

Mt Elgon

Victoria Nile

Katonga R.

KAMPALA

0°

Lake Kag

era

R.

Victoria

2°S

N

100 km 32°E

34°E

Figure 7.4 Lake Victoria Basin, East Africa. (Adapted from Beadle, 1974, Fig. 14.1; Livingstone, 1980, Fig. 14.2; Westcott et al., 1996, Fig. 2; and Meyer et al., 1996, Fig. 1.)

Fig. 4), which is broadly comparable to the luminescence age of 0.4 Ma obtained for finely laminated green lake clays at 5 m depth at the site of Esh Shawal in the lower White Nile Valley (Williams et al., 2003), discussed in Chapter 8. These lake clays may reflect early overflow of Lake Victoria into the upper White Nile.

7.4 Late Quaternary Fluctuations of Lakes Victoria and Albert Large cross-beds exposed in a sand quarry near Ed Dueim on the west bank of the lower White Nile show that around 30,000 years ago the White Nile river was transporting a bed-

102

The Ugandan Lake Plateau

load of sand under conditions of very high-energy flow (Williams et al., 2010). However, by 20,000 years ago its flow had dwindled to a mere trickle. The reason is clear. During the Last Glacial Maximum (LGM: 21 ± 2 ka: Mix et al., 2001), when global ice volume was at a maximum, Lakes Victoria, Albert and Edward in Uganda, which had until then been mostly full and overflowing, began to dry out altogether or at least fell to very low levels and no longer flowed into the White Nile (Beuning et al., 1997b; Lærdal et al., 2002; Stager and Johnson, 2000, 2008; Stager et al., 1986, 2002; Stager and Johnson, 2008). Soils developed on the exposed lake floor sediments (Williams et al., 2006). Even the Sahara Desert was drier and more extensive at that time, and desert dunes reached as far south as latitude 12°N (Grove, 1980; Mainguet et al., 1980; Talbot, 1980). Flow resumed from Lakes Victoria and Albert by 14.5 ka (Williams et al., 2006; Talbot and Williams, 2009; Williams et al., 2015b), an event reflected in cross-bedded fluvial sands in the lower White Nile Valley, dated by Optically Stimulated Luminescence (OSL) to 13.3 ± 0.9 ka (Williams et al., 2010). The abrupt return of the summer monsoon at 14.5 ka was evident not only in the overflow from the Ugandan headwaters of the White Nile (Williams et al., 2006) but also in the overflow from Lake Tana in the Ethiopian headwaters of the Blue Nile discussed in Chapter 6 (Lamb et al., 2007; Marshall et al., 2011). During this time, the summer monsoon was sporadically more intense than it is today across a broad swathe of tropical Africa (Williams et al., 2006; Lamb et al., 2007; Gasse et al., 2008; Williams, 2009; Lézine et al., 2011). Lake Chad had also expanded at this time (Grove and Pullan, 1963; Servant, 1973; Servant and Servant, 1980), although not as far as it had during the last interglacial (Armitage et al., 2007). This was also the time when a very large lake (Lake Megafezzan) covered much of southwest Libya 100–110 ka ago (Armitage et al., 2007), and the lower White Nile valley was filled by the 386-m lake (see Chapter 8). The question of just when the present hydrologic regime of the Nile became reestablished has provoked some debate. Beuning et al. (1997a) analysed the 18O/16O ratios in sediment cellulose from Lake Victoria and concluded that Lake Victoria had remained a closed basin until 7.2 ka. This conclusion runs counter to a large body of work in the Nile valley and its tributaries that point to an actively flooding Nile by 14–15 ka (Butzer and Hansen, 1968; Butzer, 1980; Adamson et al., 1980, 1982; Williams and Adamson, 1980, 1982). Pollen and glaciological studies from the Ugandan highlands clearly show a postglacial rise in temperature and precipitation after 14–15 ka, consistent with diatom and sedimentary evidence of overflow from Lakes Victoria and Albert into the upper White Nile in southern Sudan (Livingstone, 1962, 1967, 1975, 1980). In order to resolve this issue, Talbot et al. (2000) used strontium isotopes as tracers to determine when Lakes Victoria and Albert overflowed into the upper White Nile. They analysed the strontium isotope ratio (87Sr/86Sr) preserved in freshwater gastropod shells collected from Blue and White Nile sediments ranging in age from terminal Pleistocene to present day, and compared the values obtained with the strontium isotope ratios from the Ugandan lakes (Fig. 7.5). The results confirmed that Lake Victoria was overflowing into the upper White Nile by about 14.5 ka, which was also when the present-day integrated Nile drainage network became reestablished.

7.4 Fluctuations of Lakes Victoria and Albert

103

Figure 7.5 Strontium isotopic composition of Blue and White Nile waters and of lakes in the White Nile headwaters. The numbers are radiocarbon ages of dated late Quaternary samples from sites along the Blue and White Nile Valleys. (After Talbot et al., 2000, Fig. 2 and Williams, 2014, Fig. 7.1.)

Cockerton et al. (2015) analysed changes in the stable isotopes of oxygen and silicon in samples collected from sediment cores obtained from Lakes Victoria and Edward, together with changes in lipid biomarkers diagnostic of aquatic and terrestrial plants. They used these data to reconstruct the hydrological history of Lake Victoria over the past 21 ka and of Lake Edward over the past 15 ka. These lakes are major contributors to discharge in the upper White Nile. The results confirm lake history reconstructions for Lakes Victoria and Edward obtained by previous workers (Beuning et al., 1997b; Lærdal et al., 2002; Stager and Johnson, 2000, 2008; Stager et al., 1986, 2002; Stager and Johnson, 2008). The diatom record from Pilkington Bay in Lake Victoria (Stager et al., 2003) shows evidence of high rainfall at 8.8–8.3 ka, becoming more seasonal thereafter, with sharp drops at about 8.2 and 5.7 ka, and sharp century-scale increases in rainfall at about 8.5, 5.8 and 4 ka. The present climatic regime was established after 2.7 ka and there was an interlude of major droughts between about 1200 and 600 ka BP. The combined evidence from all these studies shows that the climate in the region surrounding Lake Victoria has fluctuated considerably during both the late Pleistocene and Holocene – fluctuations that will have affected the White Nile downstream (see Chapter 8).

104

The Ugandan Lake Plateau

In an attempt to obtain a longer record of Quaternary climatic fluctuations in the Lake Victoria Basin, Tryon et al. (2015) investigated a number of Middle and Late Stone Age archaeological sites located around the present shoreline of the lake. Using a combination of faunal and stable carbon isotope analysis, buttressed by a few OSL ages, they concluded that repeated phases of lake desiccation and expansion would have exerted a strong push– pull influence on human societies dependent on hunting large herbivores in drier times when fish and aquatic foods were no longer available. They concurred with earlier workers that Lake Victoria was completely dry and soils developed on the lake floor during 17–16 ka and 15–14 ka. They also inferred that the climate was drier than present between about 100 ka and 35 ka, with widespread C4 grasslands growing within and around the basin. This conclusion is based on a very incomplete chronologic framework and also appears to contradict the evidence obtained from Lake Challa, which we now discuss.

7.5 Late Quaternary Fluctuations of Lake Challa Despite being reasonably well dated for the Holocene and terminal Pleistocene, the history of the lakes that feed in to the White Nile is not known with any real precision and not for any great length of time. It is therefore useful to consider the record of hydrologic changes near the White Nile headwaters from another independent source of information, namely, Lake Challa, where the chronology is tightly controlled and inferences about changes in water balance are based on both seismic reflection data and on analysis of changes in the lipid chemistry of soil bacteria and aquatic organisms. Lake Challa (3.3°S; 37.7°E) is a small crater lake on the lower eastern slope of Mt. Kilimanjaro and has what is probably the best-dated late Quaternary record of any lake in East Africa (Verschuren et al., 2009; Moernaut et al., 2010). The chronology of lake level fluctuations is based on 164 AMS 14C ages for bulk organic carbon samples, corrected for lakecarbon reservoir ages that vary with depth from ca. 200 to ca. 450 years (Verschuren et al., 2009). The sedimentary sequence dated by 14C extends from about 20.5 ka to present. Lake levels were interpreted as low (shown by a capital L) during 20.5–14.5 ka (L6), 12.9–12.0 ka (L5), ca. 8.0–6.7 ka (L4), ca. 5.9–4.7 ka (L3), ca. 3.6–3.0 ka (L2) and ca. 0.7–0.6 ka (L1). The intervals between these times of low lake level were periods of relatively high level. The floor of the lake is overlain by roughly 210 m of sediments. Using seismic reflection data, Moernaut et al. (2010) were able to identify a number of intervals when the lake level was low during the past 140 ka. They used the age model obtained by Verschuren et al. (2009) for the upper part of the lacustrine sequence to devise a plausible age sequence for the past 140 ka. The driest interval in this time occurred prior to ca. 128 ka, and was followed by extremely dry conditions between ca. 114 and ca. 97 ka. There then ensued a long interval of about 75,000 years (ca. 97 ka to 20.5 ka) during which the climate was moist and fairly stable. The Last Glacial Maximum was dry but not as dry as the drought preceding the penultimate glaciation (>128 ka).

7.6 The ‘African Humid Period’

105

Lake Challa thus provides an independent record of possible hydrologic fluctuations near the White Nile headwaters. It was high between 10.5 and 8.5 ka, when the White Nile floods were also very high (Williams et al., 2010; Williams et al., 2015a, 2015b) and there was a large freshwater lake in northwest Sudan (Hoelzmann et al., 2000). The four intervals when Lake Challa was low during the Holocene (8.0–6.7 ka, 5.9–4.7 ka, 3.6–3.0 ka, 0.7–0.6 ka) also appear to be coeval with times of low White Nile flow. The penultimate dry phase evident at Lake Challa may have been part of a wider regional drought, reflected in the sudden decrease in precipitation over the southern Dead Sea at ca. 3.9 ka (Frumkin, 2009), but more evidence is needed to confirm or refute this suggestion. 7.6 The ‘African Humid Period’ De Menocal et al. (2000) analysed the Saharan desert dust content in a single marine sediment core retrieved off the coast of Mauritania. They found low concentrations of dust between 14.8 ka and 5.5 ka and concluded that this was evidence of a widespread episode of wetter than present climate across North Africa, which they termed the ‘African Humid Period’ (AHP). That this phrase has since been widely adopted reflects an uncritical enthusiasm for snappy shorthand labels rather than any rigorous evaluation of the regional climatic data. During the alleged AHP some parts of Africa were indeed wetter than they are today, other parts were drier and others were much the same as they are at present (Gasse et al., 2008). In addition, the detailed late Quaternary record from Lake Challa showing changes in the ratio of precipitation to evaporation reveals three major phases of low lake level (L5, L4 and L3) during the putative AHP. Furthermore, the record of fluctuating late Quaternary White Nile flood levels reviewed in Chapter 8 reveals considerable hydrologic variability during the so-called AHP. Also in question is when the AHP is considered to have begun. Tierney and deMenocal (2013) used the evidence from a single marine sediment core (P178–15P) in the Gulf of Aden to propose that the onset and close of the African Humid period were synchronous across northern, eastern and equatorial Africa, and Otto-Bliesner et al. (2014) used results from a transient climate model to argue for synchronous changes in precipitation across northern and southeast equatorial Africa during the last deglaciation. An opposing view is that of Costa et al. (2014), who analysed the deuterium content of leaf waxes in sediment cores from Lake Tana and found that the onset of wet conditions around the lake began several thousand years after the onset of wet conditions at Lake Challa, also determined from leaf wax deuterium values. Buttressed by the evidence from other lake studies across the region, they rejected the conjecture by Tierney and deMenocal (2014) that the start and close of the AHP were indeed synchronous across northern, eastern and equatorial Africa. Until this matter is finally resolved, it would seem best to avoid using the term ‘African Humid Period’– a term which begs a lot of questions. A recent thorough review by Holmes and Hoelzmann (2017) entitled ‘The Late Pleistocene-Holocene African Humid Period as evident in lakes’ concluded that in the presently arid and semi-arid areas of North Africa the climate was indeed wetter from about

106

The Ugandan Lake Plateau

14.7 ka (the start of the Bølling-Allerød interstadial in Greenland), with a return to drier conditions during the Younger Dryas stadial (12.8–11.6 ka). The wet interval finally ended towards 5–4 ka, but was characterised by a time-transgressive response from north to south and from east to west.

7.7 Kilimanjaro Holocene Ice Core Records The evidence obtained by Thompson et al. (2002) from ice cores collected from the summit of Mt Kilimanjaro provide additional insights into Holocene climatic fluctuations in this region. Six ice cores spanning the last 11,700 years revealed that there were three distinct periods of abrupt climatic change and much reduced precipitation at about 8.3 ka, 5.2 ka and 4 ka. The chronology is not as solidly based as that obtained from Lake Challa, but within the limits of dating errors these three dry intervals seem to coincide with times when Lake Challa was low. Interpreting the oxygen isotope record from the ice cores is not easy. Thompson et al. (2002) considered that times of marked ∂18O depletion indicated substantial cooling (e.g., during 6.5–5.2 ka), but Gasse (2002) cautioned that such depletion in this region could better be interpreted as evidence of intensely heavy precipitation, citing in support the isotopic work by Barker et al. (2011) on Mt Kilimanjaro. Present information about the isotopic composition of precipitation in this region does not allow us to choose between either of these two apparently conflicting hypotheses.

7.8 Conclusion The Last Glacial Maximum (23–19 ka) was a time of widespread aridity across much of equatorial and tropical northern Africa. Lake levels were low in Uganda at this time and soils developed on the exposed lake floors of Lake Victoria and Lake Albert. The summer monsoon strengthened fairly suddenly soon after 14,500 years ago, lake levels rose once more, and water flowed out of Lakes Victoria and Albert into the Bahr el Jebel and thence into the White Nile, which flowed in turn into the main Nile at Khartoum. Lake Challa, a small crater lake on the south flank of Mt. Kilimanjaro, has an exceptionally well dated record of late Quaternary high and low lake levels, with very low levels evident at 20.5– 14.5, 12.9–12.0, ca. 8.0–6.7, ca. 5.9–4.7, ca. 3.6–3.0 and ca. 0.7–0.6 ka. Ice cores recovered from the summit ice cap on Mt. Kilimanjaro indicate abrupt climate change and much reduced precipitation at about 8.3, 5.2 and 4 ka. The early to mid-Holocene was thus not uniformly warm and wet, but was interspersed by frequent intervals during which the climate was drier and possibly cooler than before. The causes of these millennial scale fluctuations are not yet known.

8 The Sudd Swamps and the White Nile

Jan 22d. – The luxuries of the country as usual – malaria, marshes, mosquitoes, misery; as far as the eye can reach, vast treeless marshes, perfectly lifeless. Sir Samuel Baker (1821–1893), The Albert N’Yanza. Great Basin of the Nile and Explorations of the Nile Sources (1866)

8.1 Introduction The headwaters of the White Nile originate in the highlands of Burundi and flow into the great lakes of Uganda before passing through a narrow gorge with rapids to emerge into the swampy lowlands of tropical South Sudan (Fig. 8.1). When the White Nile, locally known as the Bahr el Jebel (Arabic: ‘River of the Mountain’), first enters the vast Sudd swamps in South Sudan it has a total annual discharge slightly in excess of 23 km3. By the time it emerges from the Sudd about half of its total discharge has been lost to seepage and evapotranspiration, bringing it down to about 10 km3 (Hurst and Phillips, 1931; Hurst, 1952; Williams et al., 1982; Shahin, 1985). The Bahr el Ghazal joins the White Nile north of the Sudd and contributes an extra 3 km3 of water to the river. The Sobat flowing from Ethiopia joins the White Nile above Malakal and brings an additional 10 km3 of water, bringing the total annual White Nile flow back to roughly 23 km3, but by then the water chemistry of the White Nile has changed considerably (Bishai, 1962; Talling, 1957, 1976; Williams and Adamson, 1973). The lower White Nile has a number of attributes that at first sight appear anomalous (Williams et al., 2000; Williams et al., 2006). The river has a remarkably gentle flood gradient amounting to 1 cm/km (1:100,000), and yet the channel pattern is braided rather than meandering, with numerous large islands that form gigantic mid-channel bars. The soils bordering the river are locally highly saline at depths below about 1.5 m. Buried fossil shell beds can be found in places at depths of about a metre east of the present White Nile channel. These shell beds extend up to 5 km east of the present channel banks and occur at elevations up to 4.5 m above the unregulated maximum flood level of the White Nile. Finally, in a number of localities sands underlie the surface alluvial clays, and are sometimes cemented with calcium carbonate to form massive calcrete layers with the

107

108

The Sudd Swamps and the White Nile 20°E

30°E

Ed Dueim

White

Jebel Marra

hr

el Ara b

SOUTH

Melut Malakal

Ghazal

el

So

B.

10°N

ile e N Blu Nile

Esh Shawal

SUDAN

Ba

40°E

KHARTOUM

Semien Mts

Lake Tana

ETHIOPIA

ba t

SUDAN

N

National boundary

CONGO

Marshes 0

250

500

Bor

UGANDA

o

O

r el Jebel Bah

5°N

m

SUDD

Lake Turkana

Lake Albert

km

KENYA

0°N Lake Edward

Lake Victoria

Lake Challa

BU RU

ND

I

RWANDA

Mt Kilimanjaro

Figure 8.1 The White Nile Basin.

consistence of concrete. Any convincing account of the recent geological history of the lower White Nile needs to be able to explain each of these phenomena.

8.2 The Sudd 8.2.1 General Characteristics of the Sudd The geographical name Sudd seems to be a corruption of the Arabic word sadd denoting an obstruction or a barrier, which it most certainly was to the Roman legionaries sent by the

8.2 The Sudd

109

Emperor Nero in about AD 60 to locate the sources of the Nile. On their return to Rome the two centurions in charge of the exploration party reported that impenetrable swamps had finally defeated them (Jackson, 1957). They described two hills that were surrounded by the swamps. These two hills, which appear to be the two low granite hills known today as Jebel Ahmed Agha, immediately east of the White Nile at latitude 11°N, now lie 200 km north of the swamps, suggesting that the climate may have been wetter here 2,000 years ago (Mawson and Williams, 1984). In fact, as we shall see later, this was indeed a time of regionally wetter climate. The present-day Sudd covers a notional area of about 57,000 km2 (22,000 square miles), but during the height of the rainy season the swamps can extend to over 130,000 km2, flooding both adjacent woodland and grasslands or toich (Thompson, 1976). The swamps begin about the latitude of Bor (6°13 0 N) and cover the very gently sloping flood plains of three main rivers: the Bahr el Jebel, flowing from Uganda, the Bahr el Zeraf (a major anabranch of the Bahr el Jebel) and the Bahr el Ghazal, flowing in from the west, close to the divide between the Chad Basin and the White Nile Basin (Fig. 8.1). There has long been speculation that the Sudd occupies the bed of an ancient lake (Lombardini, 1865a, 1865b), but this seems highly unlikely. The overall downstream gradient of the White Nile as it flows through the Sudd amounts to about 1:70,000, which is steeper than the flood level of the lower White Nile. Scattered through the Sudd are a series of sandy ridges, such as the Duk ridge, which appear to have an alluvial origin (Wells, 1921; Hurst and Phillips, 1938; Jonglei Investigation Team, 1955). The overall geomorphic appearance of the Sudd is that of a very low-angle alluvial fan (Williams et al., 1982). The high rainfall, high temperatures and dense aquatic vegetation are collectively responsible for the present morphology of the swamps, which are fed by perennial river flow and regular precipitation during the wet season. Deep drilling has shown that beneath the Sudd there are more than 10 km of sediments, which appear to have been derived from the Ethiopian highlands at intervals during the Cenozoic (Williams and Williams, 1980). A possible buried rift appears to extend beneath the region now occupied by the Sudd (Adamson and Williams, 1980; Salama, 1987, 1994; Adamson et al., 1993). Prolonged subsidence of this tectonic basin from Mesozoic times onwards would account for the considerable depth of sediments beneath the present swamps.

8.2.2 Role of the Ugandan Swamps and of the Sudd in Regulating White Nile Flow In the Blue Nile, the ratio of discharge in the wettest month to that in the driest month is 40:1 but in the White Nile it is only 5:2, with the result that in the main Nile the ratio is reduced to 16:1. As Thompson (1976) pointed out, the upland swamps of Uganda are a blessing in disguise because they help to damp down the flood peaks of the White Nile and sustain perennial flow in that river. This is crucial for the health of the main Nile because the White Nile provides the Nile with 83% of its low season flow (Hurst, 1952). The Sudd also exerts

110

The Sudd Swamps and the White Nile

a major buffering role on the White Nile. Without this moderating influence, the White Nile would be a far more seasonal river, as was the case on occasion during the Pleistocene at times when the Sudd had dried out (Williams et al., 2010).

8.2.3 Role of the Sudd as a Biogeochemical Filter Although, as we have seen, roughly half of the water in the Bahr el Jebel that enters the Sudd is lost in evapotranspiration in the swamps, the total salt concentration in water leaving the Sudd is virtually the same as that entering it (Talling, 1957; Bishai, 1962). The swamps and their underlying sediments are therefore absorbing and retaining about half of all the salts dissolved in the water flowing into the Sudd. Talling (1957) considered that the Sudd was functioning as a giant filter or exchange system. Both Talling (1957) and Bishai (1962) demonstrated that the swamps had a considerable influence on the chemical composition of the water, which showed a progressive decrease in dissolved oxygen and a progressive increase in dissolved carbon dioxide caused by the growth and decay of organisms under partially anaerobic conditions. Dissolved iron showed an increase and sulphur a marked decrease, related to reducing conditions and biological processes within the swamp waters. The pH decreased from 8 to just over 7, phosphate levels decreased and ammonia nitrogen levels increased. The sediment deposited beneath the swamps had a percentage salt content of 0.1 to 0.54, rising in places to as high as 1.19% (Jonglei Investigation Team, 1955). Salt content tended to increase with organic matter content. Sustained unrest in this region over the past half century has meant that it has not been possible to continue a regular program of water quality monitoring, but there is no reason to assume that the analyses by Talling (1957) and Bishai (1962) do not remain valid today.

8.2.4 The Jonglei Canal: To Drain or Not to Drain? There has long been a plan to drain the Sudd by cutting a canal (the Jonglei Canal) through the swamp to allow free flow of water across it. The justification has been expressed as more water for the north and more land for the south, but the likely adverse impacts of such a project may have been underestimated. The potential repercussions of draining the Sudd can be summarised quite succinctly. One foreseeable effect will be an increase in the seasonal amplitude of White Nile discharge downstream, leading to accelerated bank erosion and damage to the inlet pumps of the private pump schemes along both banks of the lower White Nile, which even today are lacking in spare parts. Another will be a change in the water chemistry of the lower White Nile, with possible consequences for irrigation. The third possible effect might be to release substantial amounts of salt to both the Nile and adjacent soils. The project has begun, but has been halted by the current unrest in South Sudan.

8.3 The White Nile

111

8.3 The White Nile 8.3.1 Lake Sudd: A History of Speculation The most enduring speculation relating to the White Nile is that a vast lake once occupied the lower White Nile valley – a view rejected by Lyons, who noted the lack of supporting geological evidence (Lyons, 1906). The Italian hydrologist Lombardini (1865a, 1865b) was the first to suggest that the White Nile flowed across the bed of a vast former lake, thus accounting for its very gentle gradient, an idea endorsed by the American geologist Lawson (1927), who had travelled in this region. The experienced Nile Valley surveyor and geographer Dr John Ball (1939) proposed that Lawson’s Lake Sudd attained an elevation of 400 m and reached as far north as Sabaloka, which operated as a dam. Both Lawson and Ball considered that the Gezira clays had been laid down in their Lake Sudd. This hypothesis was firmly refuted when J. D. Tothill (1946, 1948) convincingly demonstrated that the Gezira clays were alluvial, and not lacustrine, having been deposited as flood plain clays by former Blue Nile distributary channels under a previously wetter climate in conditions similar to the seasonally flooded toich grasslands of South Sudan (Williams, 1966).

8.3.2 The White Nile’s Last Interglacial Mega-Lake During soil surveys in central and southern Sudan in 1951–53 carried out on behalf of Sir Alexander Gibb & Partners, Mr R. H. Gunn identified a major break of slope east of the White Nile at an elevation of 386 m (Sir Alexander Gibb & Partners, 1954; Gunn, 1982). He considered that the break of slope was erosional and marked the boundary between soils developed in situ on weathered bedrock above that elevation and alluvial soils laid down by the White Nile when it once flowed at a higher level. Later work (Hunting Technical Services Limited, 1965) based on a more extensive set of air photos revealed an undulating band of sand and gravel east of the White Nile between Rabak and Renk (Fig. 8.2) coinciding with the 386 m break of slope and very reminiscent of the cuspate shoreline of a former lake. Williams et al. (2003) used satellite imagery and levelled survey data to map the shoreline in more detail and were able to reconstruct the margins of a very large lake that had occupied the lower White Nile Valley at some time in the Quaternary (Fig. 8.2). Initial attempts to date the 386 m lake shoreline using Optically Stimulated Luminescence (OSL) relied on indirect evidence from what now appear to be much older lake sediments (Williams et al., 2003) and so were not successful in providing an age for the 386 m lake shoreline. However, Williams et al. (2003) did manage to show that the finely laminated green lacustrine clays at a depth of 5.0 m in a trench dug near Esh Shawal were about 400,000 years old (400 ka), synchronous with Marine Isotope Stage 11 (425–375 ka), which was a prolonged and warm interglacial. Later work on the 386-m beach involved dating alluvial fan deposits overlying the beach sediments (Williams, 2010), yielding terminal Pleistocene minimum ages only.

112

The Sudd Swamps and the White Nile 31ºE

32ºE

33ºE

34ºE

Khartoum Jebel Aulia

15ºN e Blu e Nil

GEZIRA (Blue Nile alluvial fan)

Medani

KORDOFAN Managil

14ºN

MANAGIL RIDGE Kawa

Ed Dueim linear dunes

Esh Shawal

K. Kile

Kosti

13ºN

ikis

Khor Abu Habl alluvial fan Jebelein Keri Kera K. A bu H

12ºN NUBA

aram

Renk K. D

MOUNTAINS

11ºN W

te hi

olei b

le Ni K. Bibban

Kaka Melut

Paloich

10ºN Kodok

9ºN

ebel B.el J B.e l Ze raf

ha z al L.No

K. W ol Sob at R iver

Malakal

B.e lG

SUDD

MACHAR MARCHES 0

50 km

Figure 8.2 The Last Interglacial White Nile Mega-lake. (After Williams et al., 2003, Fig. 1 and Williams, 2009a, Fig. 2.)

8.3 The White Nile

113

Barrows et al. (2013) were finally able to date the beach sands directly. The cosmogenic nuclide 10Be ages obtained indicated that the lake had reached its maximum level (386 m) by about 110,000 years ago, after which it receded and finally dried out. This lake occupied an area of at least 45,000 km2, comparable to the largest freshwater lakes on Earth today. It extended more than 650 km from south to north and up to 80 km from east to west (Fig. 8.2). The last interglacial was a period of much enhanced humidity across North Africa (Causse et al., 1988; Wendorf et al., 1993; Osborne et al., 2008), and quite large lakes were common across the Sahara at this time (Armitage et al., 2007; Maxwell et al., 2010).

8.3.3 Late Pleistocene Desiccation of the White Nile The White Nile experienced a short interval of very high floods 27,000 years ago during which it transported large quantities of medium sand at least as far as Ed Dueim (Fig. 8.1), where very large cross-beds are clearly visible in a sand quarry (Williams et al., 2010). The presence of such large cross-beds points to very high energy flow, and the transport of a sandy bed-load in the lower White Nile valley is incompatible with the existence of the Sudd swamps upstream. It seems probable that after an interval of reduced precipitation over Uganda and South Sudan, which led to desiccation of the swamps, the climate was becoming wetter, resulting in a return to active flow in the White Nile unimpeded by the swamps. Sediment cores taken from near the centre of Lake Albert in Uganda show that the lake had become dry on two occasions in the late Pleistocene, with the development of soils on the lake floor muds. These two intervals of desiccation and soil formation have calibrated AMS radiocarbon (AMS 14C) ages of 21.15–18.25 ka and 16.7–15 ka (Williams et al., 2006, Fig. 9b). Lake Victoria was also dry at this time, and the Last Glacial Maximum (LGM: 21 ± 2 ka) was a time of widespread intertropical aridity and low lake levels across equatorial and tropical Africa (Butzer et al., 1972; Williams, 1975; Gasse et al., 2008). In the case of the White Nile, the progressive Late Pleistocene desiccation had two other effects. As the area of the Sudd swamps diminished and as desert dunes lost their former stabilising plant cover and became active once more, the White Nile began to transport fine and medium sand rather than clay. Some of the sand was blown in from the dunes west of the river and some was brought down from upstream. Local groundwater rich in dissolved calcium bicarbonate percolated through the sandy alluvium, eventually releasing the calcium carbonate from solution and cementing the parent sands. The amount of carbonate within the sands ranges from a few per cent to over 50%, and in consistence from soft and friable to extremely hard, with the degree of induration depending on the amount of carbonate present. Analysis of the stable oxygen isotopic composition of microcrystalline calcite and dolomite precipitated at depths between 4.8 and 1.3 m in the 6 m trench ES82/1 near Esh Shawal (Fig. 8.3) shows a progressive hydrological isolation of the site, which culminated in a strong evaporative event at 30.5–29 ka (calibrated), evident in the ∂18O enrichments.

114

The Sudd Swamps and the White Nile

There was a weaker evaporative event before 15–13 ka, after which the system returned to its present fresh water hydrology (Williams et al., 2006, Fig. 8a). The carbon isotopes show a similar but more scattered pattern. 8.3.4 Sand Dunes East of the White Nile between 14°15 0 N and 15°15 0 N Sand dunes occupy 160,000 ha east of the lower White Nile Valley between Hashaba in the south and Jebel Aulia in the north (Williams, 1968a; Williams and Adamson, 1973). These dunes originated as source-bordering dunes derived from Blue Nile distributary channels that once flowed across the Gezira between the Blue Nile and the White Nile, and are discussed in detail in Chapter 11.

8.3.5 Return of the Summer Monsoon and Inception of the 382 m White Nile Lake Lake Albert refilled and flow resumed into the White Nile from about 15 ka onwards, with a brief halt at 4.2 ka (Williams et al., 2006). Analysis of the strontium isotopic composition of aquatic snail shells dated by conventional and by AMS 14C confirmed that the present hydrological regime of the Nile was re-established in the terminal Pleistocene by 14.5 ka (Talbot et al., 2000). With the return of the summer monsoon and the high seasonal floods in both the Blue Nile and the White Nile, a large lake once again filled much of the lower White Nile Valley, this time to an elevation of 382 m. Tothill (1948, p. 134) observed that the sand dunes at Hashaba were not true desert dunes but the remains of lakeside dunes. He discovered a strandline near the top of the dune at 382.14 m elevation behind the Inspector’s house at Hashaba containing water worn shells of Cleopatra and Melanoides ‘representing dead adults washed up from the river or lake bed by wave action.’ He concluded that ‘this strand line probably marks the winter shore-line of this lake at the time of its greatest development’. I visited this site in February 1963 but unfortunately any remains of the 382m shoreline had long been obliterated by recent human activity. The 382-m lake was up to 25 km wide and at least 400 km long, reaching south of Jebelein (Fig. 8.2). Once the lake began to recede, it left behind the scattered remains of freshwater gastropod shells, which were later buried beneath a metre of White Nile alluvial clay (Fig. 8.3). The shells have not undergone any post-depositional diagenetic changes and have yielded calibrated conventional and AMS radiocarbon ages bracketed between 14.5 and 13 ka (Williams, 2009a). As the lake dried out some of the receding waters were trapped in broad shallow depressions or maya’a, where they became increasingly saline (Williams, 1968b). During later floods the saline maya’a sediments became buried beneath a mantle of non-saline alluvial clay a metre or more in thickness. The land between Jebel Tomat and Esh Shawal is criss-crossed with a network of rills and drainage channels formed during the latest stages of flood plain abandonment in this area (Fig. 8.4). The twin granite peaks of Jebel Tomat rise about 30 m above the plain and the northern peak provides an excellent view of the channels from its summit.

2 378.

381.2 Telegraph line

Line of transect

13o35'N

13o40'N

?

(a)

?

?

?

1 km

Position of borehole

Position of trench

B

378

380

382

East

Sub-fossil freshwater mollusc shells

Quartz or Basement Complex pebbles

Carbonate nodules

Carbonate evaporite deposits

Well sorted pale yellow quartz sands

Olive green lacustrine clays and loams

Transitional stony dark clays and olive clays and loams

Dark grey and grey brown clays

Laminated silts and clays

Figure 8.3 Buried shell beds east of Esh Shawal. (After Williams, 2009a, Fig. 5.)

374

2 2. 38

374

2 3. 38

376

5km

386.2

376

ES82/2

ES82/1

380.2

378

ES3

A

379.2

380

382

Telegraph line

.2 379

384.2

West

A

B

32o45'E

2 385.

(b)

Esh Shawal

.

378.2

ILE WHITE N

32o40'E

388.2

Elevation (m)

387.2

116

The Sudd Swamps and the White Nile 32°40´E

13°45´N

13°45´N

Kawa

389 .2

.2 3 77

377.2

38

13°40´N

38 6.2

37 9. 2

2 2. 38 .2 1 38 0 . 2 38

37 7. 2

.2 385

38 3.2

378 37 .2 7. 2

N

4.2 38

13°40´N

8.2 38 7.2

Karari

Jebel Tomat Esh Shawal

13°35´N

378.2

13°35´E

.2 379

37 9 .2

Telegraph line Shallow linear depression

Guli Sandy island or former island Main open water channels in swamps At or below present high water level of J. Aulia dam

13°30´N 0 32°40´E

km

5

Contour interval 1.0 m

Figure 8.4 Holocene drainage channels mapped between Esh Shawal and Jebel Tomat, east of the White Nile. The channels were mapped from air photos and satellite imagery. (After Williams, 2009a, Fig. 7a.)

The stable carbon and oxygen isotopic composition of the shells of three subfossil aquatic snail species collected from 100 to 110 cm depth in trench ES82/2 (Fig. 8.3) were analysed (Williams et al., 2000, Table 2 and Fig. 4). The snails investigated were Melanoides tuberculata, Lymnaea natalensis and Biomphalaria sudanica, and lived towards the end of a period of exceptionally high river level during which large areas adjacent to the main river were flooded for many months of the year and connected to the

8.3 The White Nile

117

river. It was also the time when shelly beaches were forming as the 382 m White Nile Lake began to recede soon after 14–13 ka. The ∂18O enrichment evident in the shells is consistent with evaporative conditions and may be related to inception of the Younger Dryas arid event (12.8–11.5 ka), during which Lake Victoria also experienced a regression. A question that puzzled Ball (1939) and many others concerns the nature of the dam responsible for impounding any former lake in the White Nile Valley, which is why he suggested the Sabaloka cataract as a likely candidate. In fact, the explanation is much more straightforward, and was recognised more than a century ago by Sir William Willcocks (1904), who had an unrivalled knowledge of Nile Basin hydrology. Willcocks observed that when in flood the waters of the Blue Nile travel 300 km up the White Nile (Willcocks, 1904, p. 42), which takes on the appearance of a ‘pulsating lake’ and not a river. Once the Blue Nile discharge had fallen from about 11,000 to 6,000 m3/sec, the ponded waters of the White Nile were released and moved north into the main Nile at a rate somewhat in excess of 6,000 m3/sec. Willcocks (1904, p. 43) considered that the White Nile in flood was essentially akin to a flood reservoir. Acting on the principle that the present is the key to the past, we can assume that during times of very high peak floods in the Blue Nile, such as during the last interglacial and again during the early Holocene, ponding by the Blue Nile would have created very large seasonal lakes for many hundreds of kilometres upstream of the Blue and White Nile confluence. We can explain both the 386 m lake and the 382 m lake by this simple mechanism, without the need to invoke a dam of rock, sediment, vegetation or dunes.

8.3.6 Late Pleistocene and Holocene White Nile Floods A reconnaissance survey by the author of the soils along the east bank of the White Nile between Khartoum and Rabak, based on 440 boreholes and 126 soil pits dug to depths of 2 m or more (Hunting Technical Services, 1964) revealed that the physical and chemical properties of the soils reflected the depositional history of the Nile rather than the presentday climate (Williams, 1968a, 1968b). These insights prompted a detailed and systematic program to date the alluvial sediments in the lower White Nile Valley and establish a late Quaternary flood history for the river (Williams and Adamson, 1980, 1982). The alluvial chronology was based on radiocarbon dating of shells of freshwater gastropods and bivalves (such as Melanoides tuberculata, Lymnaea natalensis, Cleopatra bulimoides, Biomphalaria pfeifferi, Bulinus truncatus and Corbicula fluminensis) and Nile oyster (Etheria elliptica). The shells were tested for any recrystallisation using X-ray diffraction. The shells of living snails collected from the White Nile were also dated to check for any reservoir effects. This work demonstrated that White Nile flood levels were high at 14.7–13.1 ka, 9.7–9.0 ka, 7.9–7.6 ka, 6.3 ka and 3.2–2.8 ka (Williams et al., 1982; Williams, 2009a). Later work based on scores of additional trenches and boreholes dated using both AMS 14C and OSL showed that the elevation attained by successive White Nile

118

The Sudd Swamps and the White Nile

floods had diminished progressively during the course of the Holocene (Williams et al., 2010, 2015b).

8.3.7 Middle Holocene Incision of the White Nile The Blue Nile not only controls seasonal ponding of floodwaters in the White Nile but also acts as the local base level of that river. During intervals when the Blue Nile channel was incising, the White Nile also cut down and the effects of this channel entrenchment gradually propagated upstream, leading to abandonment of its previous flood plain. The most recent phase of Blue Nile incision seems to have begun about 8,000 years ago, and has resulted in a progressive decline in White Nile high flood levels from that time onwards. Superimposed upon the long-term entrenchment of the White Nile channel were episodes of higher flood levels, so that some care is needed in interpreting the alluvial record.

8.4 White Nile Islands Between its junction with the Sobat and the Melut bend (Fig. 8.1), the White Nile flood gradient is as low as 1:50,000, diminishing further downstream to only 1:100,000. There are six large islands in the stretch between Melut and Kosti, each about 30 km long. The main channel abuts the islands on one side, and one or more minor channels border the islands on the other side. Apart from Umm Garr Island, which is shaped like a pear, the islands are all long and narrow. There are also many smaller islands in the White Nile. Berry (1962) considered that the big, elongated islands were formed during an unspecified time when the White Nile discharge was ten times greater than at present. He based his discharge estimate on the empirical relationship between meander wavelength and bank-full discharge inferred from studies of much smaller and far steeper gradient rivers in recently glaciated Europe and North America. (For details of the method, see Leopold et al., 1964.) He assumed that the islands are equivalent to riffles in a pool-and-riffle channel sequence, in which deep (pool) and shallow (riffle) reaches alternate, and that the distance between riffles is equivalent to that found in meandering rivers. Both assumptions are questionable. McEvedy (1992) investigated a number of White Nile islands using air photo mosaics. She found that there was a continuum in the size of islands, many of which are being modified by present-day fluvial erosion and sedimentation. She also observed that there were abundant former islands on the flood plain that had been abandoned by the present main river channel. Finally, the larger islands often appeared to be composite features, formed sequentially, and therefore polygenic. Given the complex late Quaternary history of the White Nile, it is likely that the islands formed at different times and have experienced a number of phases of erosion and sediment accumulation (Williams et al., 2000). Figure 8.5 shows in schematic form the various stages in the formation of a polygenic island in the lower White Nile.

8.5 Prehistoric Occupation of White Nile Valley

119

Figure 8.5 Hypothetical stages in the formation of a polygenic island in the lower White Nile.

8.5 Prehistoric Occupation of the White Nile Valley 8.5.1 Palaeolithic Sites Sporadic Lower Palaeolithic (Acheulian) bifacially worked tools (hand axes and cleavers) made of silicified Nubian Sandstone can be seen on the gravel-strewn surface of Nubian Sandstone outcrops west of the lower White Nile, but none of these occurrences have as yet been systematically mapped and studied. They all appear to be surface finds and to be lag deposits formed by runoff erosion and deflation of once coherent beds of sediment that have now vanished. The lack of any visible near-surface stratigraphy suggests that they will not be easy to put into any meaningful environmental context. They most probably range in age between 1.5 and 0.5 Ma. Arkell (1949a) recorded the presence of some 200 similar artefacts recovered from a coarse gravel unit overlying weathered Nubian Sandstone exposed in the banks of Khor Abu Anga, an ephemeral left-bank tributary which joins the Nile near

120

The Sudd Swamps and the White Nile

Omdurman, 1 km below the confluence of the Blue and White Nile. The presence of the artefacts within a bed of alluvial gravel suggests that they are not in primary context but have been washed in from upslope or upstream. Arkell’s analysis of the artefacts showed that sixty-one were Early Acheulean type hand axes and seventy-nine were Late Acheulian type hand axes (Arkell, 1949a, p. 10). During a reconnaissance survey of the southwest Libyan Desert in 1957, Arkell (1964) found Late and Developed Acheulian hand axes made of ferruginous silcrete near Wanyanga Kebir (Arkell, 1964, pp. 2–3, 10–11). Wanyanga Kebir is located midway between Ennedi and Tibesti and is the site of a permanent but saline lake today. This lake was fresh during the Early Holocene and no doubt on many occasions before then. Roughly 1,000 km further west, at Adrar Bous in the Ténéré Desert of Niger, Clark et al. (2008) also described Late Acheulean assemblages excavated in situ from alluvial clays in this presently arid Saharan mountain. Because the Acheulean spans a time range of nearly a million years, environmental generalisations are unwise, but it is clear that on a number of occasions during the Pleistocene small groups of ancestral humans armed with an Early Stone Age tool kit were able to roam across the central and southern Sahara at least as far as the lower White Nile valley in their quest for plant and animal foods. The climate at these times must therefore have been less arid in this vast region than it is today. We return to this issue in Chapters 13 and 16. The apparent absence of Middle Palaeolithic/Middle Stone Age (MSA) stone tools in the White Nile Valley is curious because they are common at a number of sites further west in the southern and central Sahara as well as in the piedmont deposits around Jebel Marra (Chapter 13). It seems probable that as pre-Holocene deposits in the White Nile Valley are investigated more closely, evidence of MSA occupation will be found. Support for this suggestion comes from the discovery by Haaland (1984) of an eroded surface scatter of MSA stone tools near Rabak at a site located 3 km east of the White Nile and 3.5 km above its Holocene flood plain. She considered the quartzite scrapers and points to be Mousterian. No ages are available.

8.5.2 Mesolithic, Neolithic and Iron Age Sites Except for military duties during 1940–44, A.J. Arkell was Commissioner for Archaeology and Anthropology in the Sudan Government from 1938 to 1948. In this time, he conducted a series of major excavations (Arkell, 1949b, 1953), which provided the groundwork for subsequent archaeological research in the Sudan. He carried out extensive excavations of a Mesolithic or ‘Early Khartoum’ site near the central railway station at Khartoum (Arkell, 1949b). His prescient forecast that the age of the ‘Early Khartoum’ tradition was likely to prove to be about 8,000 years old was remarkably accurate, and was vindicated by the ages of 8–9 ka obtained for the Mesolithic sites at Tagra and Shabona (Adamson et al., 1974; Adamson et al., 1982; Clark, 1989) (Fig. 8.6). Arkell (1949b) concluded from the abundant swamp fauna and the presence of Celtis integrifolia seeds at the site that the climate at

8.5 Prehistoric Occupation of White Nile Valley

Khor Abu A

nga

121

KHARTOUM

J.Baroka

N J.Aulia UE BL

W

H

E NIL

IT E

39 0

15°N

38 5

b

NILE

Wadi Mansu ra

380

El Khiday

41 0

40 0

Shabona

14°N

Ed Dueim

Tagra MANAGIL RIDGE

Esh Shawal J.Tomat 0

50 km

Contour interval : 5m 32°E

33°E

Figure 8.6 Location of excavated prehistoric sites in the lower White Nile Valley.

Khartoum during the Mesolithic was similar to that in South Sudan today, with a longer wet season and a mean annual precipitation at least three times higher than at Khartoum today, that is, >500 mm as opposed to 175 mm. He also found evidence of Nile flood deposits 10 m above present-day maximum flood levels. His overall conclusion was that the climate at the confluence of the Blue and White Nile was very much wetter during the Mesolithic (i.e., 9,000–8,000 years ago) with Nile floods some 10 m higher at that time. The site of Tagra (Fig. 8.6) east of the White Nile yielded fragments of two barbed bone points characteristic of the Mesolithic Early Khartoum tradition, as well as fragments of small bones of mammal and fish bones, including catfish spines (Adamson et al., 1974). These remains were sandwiched between two layers of unbroken aquatic shells. The upper

122

The Sudd Swamps and the White Nile

shell bed at 1.2–1.4 m depth had a 14C age of 8,130 ± 225 years BP (calibrated age 9,060 ± 300 years BP); the lower shell bed at 1.9–1.6 m depth had a 14C age of 8,700 ± 350 years BP (calibrated age 9,750 ± 440 years BP) (Adamson et al., 1982; Williams, 2009a). Shabona lies 8 km east of the present White Nile near the northern end of a former shallow embayment of the river. It is located on a sand dune that rises 2–3 m above the surrounding clay plain. Occupation debris is scattered across at least 50,000 m2 of the dune surface, and consisted of abundant quartz microliths, sandstone grindstones, pottery tempered with quartz or grass, and occasional barbed bone points, all of which recall the occupation debris at the Early Khartoum Mesolithic sites excavated by Arkell (1949b). A striking feature of this site is the abundance of Pila wernei shells in shallow pits, which showed no signs of burning; they may have been boiled in pots before being eaten. Pila is an amphibious snail, with lungs and gills, and burrows into mud at the end of the wet season. They would have been easy to collect in the swamps and ponds during the rainy season, especially when the White Nile flooded the embayment immediately south of the site. Wild grasses such as Digitaria, Panicum turgidum, Echinochloa esculentis and the prickly Cenchrus biflorus would have been ripe and available for harvesting at the end of the rainy season. To this day, Cenchrus biflorus is a widely used famine food during times of major drought along the southern margins of the Sahara. In addition to Pila, the fauna at Shabona included abundant fish and tortoises, Varanus niloticus (the large monitor lizard), Mahelya sp. (snake), Crocodylus niloticus and Hippopotamus amphibius. Savanna bovids were also common, including Redunca redunca (reedbuck), Kobus kob (kob), Syncerus caffer (Cape buffalo), Loxodonta africana (African elephant) and Phacochoerus aethiopicus (warthog). The Pila shells yielded two 14C ages: 7,050 ± 120 years BP (calibrated age 7,880 ± 120 years BP) and 7,130 ± 40 years BP (calibrated age 8,382 ± 236 years BP), and human bone gave a 14C age of 7,470 ± 240 years BP (calibrated age 8,382 ± 236 years BP). During August–September 1999 exceptionally heavy rains fell in central Sudan (Williams and Nottage, 2006). Satellite imagery from before and during this extreme event revealed that the swales between the dunes east of the lower White Nile filled with water, and so provided a possible modern analogue for the Early Holocene wet season geography of this area. For more than a decade, Daniella Usai and her husband Sandro Salvatori with their team of archaeologists and geologists have been painstakingly excavating Mesolithic and Neolithic sites in the vicinity of El Khiday village west of the White Nile (Fig. 8.6). The sites are located on low sand and gravel ridges that originally formed as point-bars or mid-channel bars in the terminal Pleistocene White Nile some 14,500 years ago (Williams et al., 2015b). The sites revealed so far cover parts of the Early and Middle periods of the Mesolithic between 9 ka and 8.5 ka, followed by a gap of about 700–1000 years before the Late Mesolithic, one site of which has a date of 7.4–7.3 ka (Salvatori et al., 2011, 2014; Salvatori, 2012). One unique object recovered from the most ancient layers (9–8.7 ka) is a painted pebble with a representation of a boat (Usai and Salvatori, 2007). Another remarkable discovery from this locality is the use of salt to preserve fish during the Middle Mesolithic (Maritan et al., 2018).

8.5 Prehistoric Occupation of White Nile Valley

123

The Neolithic pottery is similar to that excavated at the Early Neolithic site of Shaheinab on the main Nile north of Khartoum (Arkell, 1953), with four 14C ages between 6.5 ka and 6 ka. Fish are the dominant item in the fauna, especially Clariidae (clariid catfish). The fish were usually more than 40 cm long, suggesting that people were probably fishing at the start of the flood season, when spawning Clariidae are easy to catch in shallow waters (Van Neer, 2004). The remaining fauna include other fish species that live in marshes and open water habitats, Trionychidae (softshell turtles) and C. niloticus (crocodile), together with herbivores characteristic of a savanna environment (Williams et al., 2015b). More convincing evidence that the Early Holocene was significantly wetter in this area comes from two other nearby localities: the shell-bearing pans near Wadi Mansurab and the Jebel Baroka former wetland located about 20 km north of Wad Mansurab (Fig. 8.6). A series of shallow clay pans occur 10–12 km west of the White Nile and 1–6 km north of ephemeral Wadi Mansurab at elevations of 400–420 m, and so well above any possible White Nile floods. The pans contain abundant shells of terrestrial, semi-aquatic and aquatic shells with calibrated AMS 14C ages between 9.9 and 7.6 ka, with 11 out of 14 ages between 9.0 and 8.4 ka (Williams and Jacobsen, 2011). These ages are similar to those obtained from the Mesolithic barbed bone harpoon sites of Tagra and Shabona. The oxygen isotopic composition of shells from four species of aquatic snails (Melanoides tuberculata, Cleopatra bulimoides, Biomphalaria sudanica and Lymnaea natalensis) collected from the pans shows highly variable and strongly depleted oxygen isotope values (see Chapter 4), suggesting a distant (possibly South Atlantic) precipitation source, with local runoff from a seasonal rainfall regime with considerable variability from year to year (Ayliffe et al., 1996). In addition, the abundant shells of the large land snail Limicolaria flammata are consistent with a mean annual rainfall perhaps three times that of today (Williams et al., 1974). Limicolaria flammata today inhabits the acacia-tall grass region where the annual rainfall is at least 450–500 mm and generally about 450–800 mm (Tothill, 1946, pp. 155–159; Andrew, 1948, pp. 40–41; Tothill, 1948, pp. 138–139; Bunting and Lea, 1962). The present-day annual rainfall at Wadi Mansurab is 150–200 mm. Equally persuasive evidence that the Early Holocene was significantly wetter in this region than it is today comes from the Jebel Baroka wetland (Williams et al., 2015b, Fig. 6), which once formed a swamp or seasonal lake extending 30 km from west to east and up to 4 km in width. Sediment samples at 40–50 cm depth had OSL ages of 10 ka. Two radiocarbon and two OSL ages from the upper 10 cm showed that the wetland was still receiving local runoff between 8.5 and 6 ka, after which it began to dry out. The shells within the upper 10 cm of sediment were mainly Pila wernei, Lymnaea natalensis and Limicolaria flammata, suggesting a seasonally flooded depression surrounded by acacia-tall grass savanna, with a mean annual rainfall of at least 400 mm (Williams et al., 2015b).

124

The Sudd Swamps and the White Nile

8.5.3 Sponge Spicule Pottery of the White Nile Valley A unique form of pottery tempered with sponge was first recognised by Dr D.A. Adamson in January 1972 at Jebel Tomat, a site with evidence of early domestication of sorghum (Clark and Stemler, 1975). This pottery contains abundant organic and siliceous particles of the swamp-dwelling sponge Eunapiens nitens Carter (Penny and Racek, 1968) and occurs at a number of Iron Age occupation sites dated between about 3.5 and 1.5 ka in the White Nile Valley (Adamson et al., 1987). The sponge pottery is common on both sides of the White Nile near the river between latitudes 12°N and 15°N, over a distance of some 400 km. It also occurs at Meroe on the main Nile and as far south as Jebel Nyijwaat (9° 18 0 N, 31°´08 0 E), 50 km southwest of the confluence of the Sobat and White Nile. The siliceous megascleres form the temper, but abundant gemmules and gemmoscleres occur within the pottery. The sponges were collected, crushed and deliberately incorporated during pot making, as is done by Amazonian potters today (Linné, 1965) and show no sign of the abrasion or fragmentation to be expected of alluvial clays containing transported fragments of sponge. The spicules or cylindrical siliceous rods would have acted very much as do the steel reinforcing rods used to strengthen concrete today. A fine silt or clay paste was used to make the pottery, which was fired at low temperatures, evident in the lack of distortion in the spicules. The pottery is 4 to 9 mm thick, in contrast to the much thicker pottery made without using sponge as temper. The pots were made as deep, steep-sided bowls with straight or slightly everted rims, often decorated with chevron and other patterns made by incision, rouletting and stamping. The pottery contained little or no sand and no dung or plant material. Little is known about the people who practised this swamp-based technology, but the stylistic elements of the pottery are common elsewhere in Sudan and East Africa (Adamson et al., 1987).

8.6 Conclusion The White Nile flows from the great lakes of Uganda through the vast Sudd swamps of South Sudan, where much of its sediment load is filtered out, so that it contributes very little sediment to the main Nile. The Sudd swamps in South Sudan are among the biggest freshwater swamps on Earth, and operate as a gigantic physical and biochemical filter for the White Nile. These swamps fluctuate considerably in size between the wet and dry seasons in any given year as well as from year to year. The Bahr el Ghazal flowing from the Congo–Nile divide in the west provides a modest amount of discharge to the Bahr el Jebel/ White Nile. Owing to very high rates of evapotranspiration from the Sudd, about half of the initial amount of water coming into the Sudd is lost to evaporation and seepage, which also modifies the chemical composition of the water leaving the swamps. Contrary to a long standing but erroneous view, the Sudd was not a former lake – the gradient is too steep – but in reality it occupies a vast low-angle alluvial fan. There was indeed a former lake, in fact there were several successive lakes at different times during the middle and late Quaternary, but they were located further north, in the lower White Nile Valley.

8.6 Conclusion

125

Although the overall catchment area of the White Nile is huge, amounting to just over half of the entire area of the Nile Basin, its contribution to Nile sediment load is at present minimal, because the Sudd swamps have efficiently filtered out all but the finest sediment. It was not always thus. During times of glacial aridity, when the Ugandan lakes dried out, and flow into the White Nile effectively ceased apart from a minor trickle from the Bahr el Ghazal, the Sudd swamps dried out. Once the summer monsoon resumed, bringing renewed overflow into the White Nile from the Ugandan lakes, there was a brief interval of high White Nile discharge and coarse sediment transport. The remarkably gentle flood gradient of the lower White Nile – equating to 1 cm per km or 1:100,000 – is easy to explain. During times of exceptionally high Blue Nile flow when the summer monsoon was at its strongest, as it was during the last interglacial some 120 ka ago, and again during the terminal Pleistocene and early Holocene 15–12 ka ago, the strong Blue Nile flow caused water in the White Nile to back up for hundreds of kilometres, creating a vast seasonal lake in which fine mud accumulated. However, it is during the dry season that the White Nile comes into its own, contributing nearly 85% to the low season flow of the Nile. Furthermore, during prolonged drought years in the Ethiopian headwaters of the Nile (i.e., the Blue Nile and the Atbara Rivers), it is water from the White Nile that helps to maintain perennial flow in the main Nile. On occasions in the past when overflow from the Ugandan lakes had ceased, flow in the White Nile most probably dwindled to a trickle and the main Nile dried up into a series of pools for many months of the year, such as during the cold dry and windy Last Glacial Maximum 23,000 to 19,000 years (23–19 ka) ago. Limited data suggests that a lake may have occupied the White Nile Valley at about 400 ka, possibly during Marine Isotope Stage 11, which was a very long interglacial. A very large lake did occupy the White Nile valley during the Last Interglacial until 110 ka, after which it receded. This mega-lake rose to an elevation of 386 m and was at least 650 km long from south to north and up to 80 km wide from east to west, with an area of at least 45,000 km2, comparable to that of the largest lakes on Earth today. Following a long dry interval during the very late Pleistocene, culminating in the drying out of the Sudd swamps in South Sudan and Lakes Victoria and Albert in Uganda, the African summer monsoon became stronger and a wetter climate prevailed across the Nile Basin during the Early and Middle Holocene until about 4.5–4.2 ka, when widespread desiccation set in. A large lake reoccupied the lower White Nile valley between about 14.5 and 13 ka. This lake peaked at 382 m elevation, at which time it was up to 25 km wide and 400 km long from south to north. High White Nile flood levels are evident at 14.7–13.1 ka, 9.7–9.0 ka, 7.9–7.6 ka, 6.3 ka and again at 3.2–2.8 ka. Incision by the White Nile began soon after 8 ka, leading to a progressive reduction in flood levels during the course of the Holocene. Sporadic Lower Palaeolithic/Early Stone Age artefacts (Early and Late Acheulean hand axes and cleavers in particular) occur west of the lower White Nile until slightly north of the confluence of the Blue Nile and White Nile. Almost nothing can be said about the environmental conditions during this long interval of time beyond noting that the presence

126

The Sudd Swamps and the White Nile

of similar Acheulean assemblages in the southern and central Sahara would seem to indicate that plant and animals must have been present and hence the prevailing climate must have been less arid than it is today in that region. Middle Palaeolithic/Middle Stone Age artefacts have seldom been recorded from the White Nile Valley, most probably because pre-Holocene deposits have not so far been systematically excavated in this area. Mesolithic and Neolithic sites are common along both sides of the lower White Nile, the former dating back to 9–8.5 ka and the latter to 6.5–6 ka. Fish were a major component of both the Mesolithic and Neolithic diets, together with semi-aquatic snails (especially Pila wernei), turtles, crocodiles, hippos, savanna herbivores and wild grass seeds. The climate during Mesolithic times was wetter than today, with a longer wet season and a mean annual rainfall three times that of today, amounting to at least 400–500 mm as opposed to 175 mm.

9 Lake Turkana and Overflow into the Sobat

Western Ethiopia and northern Kenya are lands such as this, where the sense of undisturbed wilderness can be appreciated on the craggy precipices of uninhabited mountains, along the white sands that rim the Jade Sea, Rudolf [Turkana], or the endless green forests that seem to swallow the brown river waters cascading from highland bastions. Each landscape is different, each has its own esthetic spell, and each is rich in history, waiting to be deciphered and told. Karl W. Butzer, Recent History of an Ethiopian Delta: The Omo River and the Level of Lake Rudolf (1971)

9.1 Introduction The region around Lake Turkana and the lower Omo Valley attracted considerable international attention during the late 1960s and early 1970s as a result of the abundant and welldated hominin and other vertebrate fossils that were being recovered from the sediments in that region. Three rival teams from France, Kenya and the USA were busy excavating their respective concessions until interest shifted to the more abundant and much older fossil record from the Middle Awash Valley in the Afar Rift. Coppens et al. (1976) have summarised the results of this early work in what was then known as the Lake Rudolf Basin. Several points of interest emerged from this and later work. Cerling et al. (1977) and Cerling (1979) analysed the oxygen isotopic composition of pedogenic and groundwater carbonates across East Africa as well as in the Turkana Basin. They concluded that a sharp increase in the heavier isotope of oxygen (∂18O) meant that precipitation had decreased suddenly around Lake Turkana roughly 2.0–1.8 Ma. Later work by deMenocal (2004) on the African faunal record found evidence of step-like changes in aridity and climatic variability followed by more open habitats at about 2.8 Ma, 1.7 Ma and 1.0 Ma, all of which appeared to be synchronous with the onset and intensification of high-latitude glacial cycles. The 2.0–1.8 Ma step-function change in climate and habitat coincided with the emergence of Homo erectus/Homo ergaster at about 1.8 Ma. However, the faunal evidence often has major gaps, making it prone to misinterpretation. This is well illustrated by Williamson’s (1982) detailed analysis of molluscan biostratigraphy of the hominin127

128

Lake Turkana and Overflow into the Sobat

bearing deposits at Koobi Fora in the north Kenya Turkana Basin. When he compared the high-resolution molluscan biostratigraphy to the stratigraphy inferred solely form the fossil vertebrate evidence, Williamson (1982) found major miscorrelations between several local sections. Once again, it is critical not to rely on any single line of evidence when reconstructing past changes in climate and hydrology. This chapter reviews some of the evidence for Quaternary climatic fluctuations in the Lake Turkana Basin, including times when the lake overflowed to the north across a now dry channel to eventually flow into the Pibor, Sobat and White Nile Rivers. 38°E

ve

r

36°E

Go

ETHIOPIA

Kibi

sh

er

River

SOUTH SUDAN

River

Mago R .

6°N

Omo

Riv

jeb

Omo

Ri

N

8°N

ETHIOPIA

4°N

Lake

KENYA

Ke

rio

2°N

River

Turkwel Riv er

Turkana

UGANDA

Mt. Elgon

0

100 km

0°

Figure 9.1 Lake Turkana Basin. (Compiled from data in Butzer, 1971a, Fig. 1.2; Atlas of Africa (1973); and Johnson and Malala, 2009, Fig. 1.)

9.2 Lake Turkana

129

9.2 Lake Turkana Lake Turkana (previously known as Lake Rudolf) occupies a tectonic depression situated between the Kenya Rift to the south and the Ethiopian Rift to the northeast. The present lake overlies some 3.5 km of sediments that are up to 4.3 Ma old (Johnson and Malala, 2009). During the past 3 Ma, conditions in what is now Lake Turkana have oscillated between lacustrine and fluvial environments (Feibel, 1988, cited by Westcott et al., 1996). Based on the presence of early Pleistocene (ca.1.9–1.3 Ma) fossil stingray spines in lacustrine and fluvial sediments, Feibel (1994) considered that Lake Turkana overflowed southeast into the Indian Ocean until at least 1.9 Ma, after which tectonic movements altered the direction of overflow and opened up an intermittent passage to the Nile (Johnson and Malala, 2009). The present-day Lake Turkana (Fig. 9.1) is 265 km long, has a mean width of 30 km, a mean depth of 35 m and a maximum depth of about 115–120 m (Beadle, 1974; Cerling, 1996; Johnson, 1996; Muchane, 1996; Johnson and Malala, 2009). Its drainage basin covers an area of about 146,000 km2. The Omo River is the major (and perennial) contributor of water and sediment to the lake. It rises some 500 km to the north in the Ethiopian Highlands close to the upper Blue Nile interfluve before flowing south into Lake Turkana. Lesser amounts of water and sediment come from the seasonal Turkwel River in northern Kenya, which flows for a few months each year. On occasion during the late Quaternary there were intervals of much higher discharge from the Omo River. These wetter episodes, of which the most recent was during the first half of the Holocene, led to overflow of Lake Turkana across its present divide near Sanderson’s Gulf into the Pibor River and thence across the Lotigipi Swamp into the Sobat River and so into the White Nile (Butzer, 1971a; Beadle, 1974; Adamson and F. Williams, 1980, pp. 232–233; Harvey and Grove, 1982; Johnson, 1996) (Fig. 9.2). The addition of the Omo and Lake Turkana Basins would have increased the catchment area of the Nile Basin by 146,000 km2.

9.3 Quaternary Sediments in the Lower Omo Valley As far as the Nile is concerned, the most informative and most reliably dated set of deposits in the lower Omo Valley are those classified as belonging to the Kibish Formation (Brown and Fuller, 2008; Brown et al., 2012). McDougall et al. (2008) found evidence of rapid deposition of three members of the Kibish Formation bracketed by volcanic ash layers for which they obtained precise Ar/Ar ages. They found that Member 1 (196 ka) was coeval with sapropel S7 (195 ka) in the eastern Mediterranean. Member III (104 ka) they equated with sapropel S4 (102 ka) and Member IV (9.5–3.3 ka) with sapropel S1 (13.7–8.9 and, possibly, 5.0 ka). Sapropels are highly organic sediments that formed on the floor of the Mediterranean during times of exceptionally high freshwater discharge into the sea and so are good indicators of times of very high Nile floods. McDougall et al. (2008) concluded that these three intervals of rapid sedimentation in the Omo Delta were times of stronger summer monsoon, with Lake Turkana 20–40 m deeper and overflowing into the White Nile via the Pibor and Sobat

130

Lake Turkana and Overflow into the Sobat

Rivers. Because the headwaters of the Omo and Blue Nile share a common divide, it will be no surprise to find that these were also times of very high flow in the Blue Nile, reviewed in Chapter 11. Brown and Fuller (2008) also equated Member 4 of the Kibish Formation to an early Holocene interval of very high flow in the lower Omo and overflow of Lake Turkana into the White Nile via the Sobat. Finally, it is worth noting that some of the oldest presently known anatomically modern human fossils (Homo sapiens) have been recovered in the lower Omo Valley in southern Ethiopia in sediments dated between 195 ka and 104 ka, with the earlier date considered most likely (McDougall et al., 2005, 2008; Fleagle et al., 2008; Brown et al., 2012).

9.4 Overflow of Lake Turkana into the White Nile Both the fish fauna in Lake Turkana and the freshwater snail fauna show a significant overlap with those in the Nile (Beadle, 1974; Van Damme, 1984; Van Damme and Van Bocxlaer, 2009), pointing to past connections between the two basins. Butzer (1971a) identified a series of former lake strandlines northwest and northeast of the present lake located at elevations above the present overflow level of the lake and obtained a series of 14C ages from gastropod shells within the beach ridges. Harvey and Grove (1982) followed and mapped parts of the overflow channel on the ground (Fig. 9.2) and obtained some additional 14C ages. In contrast to Butzer (1971a), who considered that there had been three Holocene phases of very high lake level, Harvey and Grove (1982) were confident in identifying two intervals only when the Holocene lake was high and overflowed, a conclusion cautiously endorsed by Johnson and Malala (2009). The lake was high and overflowing between 11.5 and 7.8 ka (calibrated 14 C ages) and again between 7.4 and 4.7 ka (Johnson and Malala, 2009, Fig. 8). The lake level then dropped from an elevation of about 455 m to about 420 m elevation. There is a single uncalibrated 14C age of 3,250 years BP during which overflow may have occurred once more (Butzer, 1971a), but this date contradicts the diatom evidence from a sediment core analysed by Johnson and Malala (2009), who concluded that ‘the date is erroneous, and that the last time of Lake Turkana overflow to the Nile was around 4,300 to 4,700 [cal 14C years BP], based on the diatom record . . . and dated beach deposits, respectively’ (Johnson and Malala, 2009, p. 299). There is some evidence of early episodes of late Quaternary overflow, but the ages are close to the limit of reliable radiocarbon dating and not especially convincing. The fluvio-deltaic deposits of the Kibish Formation, discussed in Section 9.4, provide some evidence of times of earlier periods of very high discharge into Lake Turkana from the Omo River. Harvey and Grove (1982) used aneroid barometer readings to determine that the gradient of the overflow channel over a distance of 200 km amounted to only 1:4000. They observed that the width of the channel ranged from about 11 km to about 200 m. The floor of the channel consisted of black cracking clay showing no signs of incision. At intervals along the channel there were what they termed pre-ceramic and ceramic Late Stone Age sites of middle Holocene age. They also considered that although the channel had been intermittently in use during earlier times in the Quaternary, the early and middle Holocene link with the Nile was sporadic, with only occasional overflow.

9.5 Conclusion 30°E

35°E

SUDAN

Lake Tana

Nile

N

131

Whi

te

10°N

So ba tR .

rR .

Ba l Jebel

0

hr e

5°N

Overflow channel

m

o Pib

SOUTH SUDAN

ETHIOPIA

o

R.

O

Sandersons Gulf 250 km

500

Lotigipi Swamps KENYA

Lake Turkana

Figure 9.2 Approximate location of the early Holocene overflow channel from Lake Turkana to the White Nile. (Compiled from data in Butzer, 1971a, Fig. 1.5; Adamson and F. Williams, 1980, Fig. 10.5; Harvey and Grove, 1982, Fig. 2; and Johnson and Malala, 2009.)

9.5 Conclusion Lake Turkana lies at the distal end of the Omo River Basin and fluctuates in response to changes in discharge from that river. On a number of occasions during the Quaternary when precipitation in the Ethiopian Highlands was significantly higher than today, leading to very high discharge in the Omo River, the level of Lake Turkana rose until it overflowed into Sanderson’s Gulf. It then flowed via the Kangen River into the Pibor River and across the Lotigipi Swamp to join the Sobat River, which in turn flowed into the White Nile near Malakal. The Sobat River itself has its headwaters in the Southwest Highlands of Ethiopia before its somewhat circuitous journey to the White Nile, where it contributes significantly to the discharge of that river. Its contribution to White Nile flow would have been even greater during times when Lake Turkana overflowed. At present two great rivers that rise in the Ethiopian Highlands (the Blue Nile and Atbara) together provide 97% of the total annual sediment load of the Nile as well as 90% of the peak summer flood discharge. The contribution of the White Nile to low-season Nile flow in particular would have been even greater on the occasions when Lake Turkana overflowed.

10 The Khor Abu Habl Fan and the Desert Dunes of Kordofan and Darfur

In this new dune-field the sandy hollows were littered not only with rough stones but with a profusion both of flaked implements and grinders left at some unknown date by a large and long-continued settlement of men. This opens up the possibility that . . . there may have been periods when a slightly more frequent rainfall produced enough grazing among the great dune-fields to support a nomad population for many months of the year. Ralph A. Bagnold (1896–1990), Libyan Sands: Travel in a Dead World (1935, reprinted 1987, pp. 236 and p. 253)

10.1 Introduction In this chapter, we consider how the Khor Abu Habl alluvial fan (Figs. 10.1 and 10.2) and the linear dunes of Kordofan and Darfur reflect past changes in climate and how they may provide some insights into past shifts in rainfall belts during the late Quaternary. By way of context, we examine some of the interactions between fluvial and eolian systems in North Africa and evaluate some of the limitations of using dunes to infer past changes in climate.

10.2 The Umm Ruwaba Formation and the Khor Abu Habl Fan 10.2.1 The Umm Ruwaba Formation The Khor Abu Habl is a seasonal river 378 km long that flows across a gently sloping surface between the Nuba Mountains and the White Nile Valley. Unconsolidated clays, sands and gravels up to several hundred metres in thickness lie beneath this surface. The age of these mostly alluvial sediments is not known with any degree of precision. They overlie Upper Cretaceous Nubian Sandstones and are most likely Cenozoic in age. Andrew and Karkanis (1945) considered that they were Tertiary and Pleistocene in age and designated then as the ‘Umm Ruwaba Formation’. Andrew (1948, p. 104) expressed his frustration at being unable to specify an age any more precise than younger than mid-Tertiary and older than the Kordofan desert dunes: ‘. . .The immediate significance of the deposits is much

132

10.2 Umm Ruwaba Formation and Khor Abu Habl Fan

133

Nile

KHARTOUM

White

Land > 1000m a.s.l. Land > 500m a.s.l.

N

Khor Abu Habl fan

Ed Dueim

14°N

100 m El Obeid

alla

Tendelti Umm

Rahad

al

Gh

ad W

Kosti

Ruwaba

i

Jebelein

1451m Dilling

M

O

UN

1023m

Nile

TA I

NS

12°N

Wh

UT SO

10°N

H

SU

DA

SU

N

DA

N

Talawdi

ite

NU

BA

Kadugli

Bahr el Arab

el hr a B Bahr el Jebel 30°E

Ghazal

Malakal Bahr el Zaraf

32°E

Figure 10.1 The Nuba Mountains and the Khor Abu Habl fan (shown schematically).

reduced by the absence of evidence of age; they are younger than the mid-Tertiary peneplain, and possibly extend well into the Pleistocene. They are older than the Kordofan (“qoz”) sands and dark clays of the plain.’ Whiteman (1971) and Vail (1978) considered that they were most likely of similar age to the Gezira Formation (see Chapter 11) and Ruxton and Berry (1978) considered that they simply belonged under the general category of ‘Older Alluvium’ with no age specified. The sediments of the Umm Ruwaba Formation appear to have been deposited as a series of alluvial fans over an irregular erosional surface and were probably laid down at a time

134

Khor Abu Habl Fan and Desert Dunes of Kordofan and Darfur

Rocky hills (jebels) and pediments Recent alluvium: clay, silt,sand Low dunes “Qoz”: sand sheets and alluvial clay

13.5°N

Seif dunes, sandsheets and old fans Alluvial fan: clays and sandy clays

e hit W

Colluvium from Nuba Mts.

N

Outwash fans from Nuba Mts.

Kosti

Linear dunes.

Tendelti 13°N M TS

Khor Abu Habl

Umm Ruwaba NUB

A

Nile

Khor Abu Habl fan

12.5°N

0

50 km

31.5°E

32°E

32.5°E

Figure 10.2 The Khor Abu Habl fan. The inactive northern sector of the fan is concealed beneath wind-blown dunes. The southern sector, shown by horizontal lines, is intermittently active today. (Modified from Gunn, 1982, Fig. 6.4.)

when the late Cenozoic climates of North Africa were becoming increasingly unstable and alternating between humid and arid conditions. In this context, it is worth noting that Andrew (1948, p. 104) observed that the minerals (biotite, feldspar) within the sandy beds were ‘undecayed’. El Boushi and Abdel Salaam (1982) also observed that the feldspar minerals in the sands at the base of the Gezira Formation were likewise unweathered. Independent evidence confirms that the late Cenozoic climates over North Africa have oscillated between arid and less arid. The microfossil assemblages and eolian dust in marine

10.2 Umm Ruwaba Formation and Khor Abu Habl Fan

135

sediment cores collected off the west coast of the Sahara show episodes of cold dry climate at 24–20, 18–14, 13–9.5, 7.5–5.3 and 3.2–1.9 Ma and from 0.73 Ma onwards (Sarnthein et al., 1982). During the intervening phases (20–18, 14–13, 9.5–7.5, 5.3–3.2 and 1.9–0.73 Ma) the climate was less arid, with periods of intense river discharge to the ocean. These climatic fluctuations were superimposed upon a long-term trend towards climatic desiccation caused by the northward shift of the African tectonic plate into drier tropical latitudes (see Chapter 4). The Umm Ruwaba Formation was built up by rivers flowing down from the major uplands of Kordofan and Darfur, especially the Nuba Mountains in the south of this region and, from Miocene times onwards, Jebel Marra volcanic massif in the north. These rivers deposited extensive large low-angle alluvial fan deposits across the landscape. The Khor Abu Habl fan is but the latest in a series of aggrading alluvial fans and provides us with a model for how the Cenozoic sediments in this region might have accumulated.

10.2.2 The Khor Abu Habl Fan The Khor Abu Habl rises near Dilling to the north of the Nuba Mountains and flows across a vast alluvial fan (Figs. 10.1 and 10.2) of radius 150 km, which abuts against the White Nile between latitudes 12°20 0 N and 13°30 0 N, a distance of roughly 160 km from north to south. With a total area of about 112,500 km2 it would certainly qualify as a mega-fan (Leier et al., 2005). In its 140-km reach between Dilling and Rahad it has a gradient of 128.6 cm/ km (Andrew, 1948, p. 116). Between Rahad and Umm Ruwaba (73 km) its gradient decreases to 64.4 cm/km and decreases once more to 43.1 cm/km in the 72 km between Umm Ruwaba and Tendelti, which is where the well-defined wadi channel ends. From Tendelti to Kosti (93 km) the mean gradient amounts to 37.6 cm/km, but the trace of the channel is very indistinct and often covered by wind-blown sand. During exceptionally wet years some water reaches almost as far as the White Nile, where it disappears between the dunes. During the early Holocene, there was most likely a regular seasonal influx of water and sediment into the White Nile valley from the northern Nuba Mountains via the Khor Abu Habl and its distributaries. Gunn (1982) observed that the eastern margin of Khor Abu Habl fan ended quite sharply at the 382 m contour, which is the level attained by the White Nile during the early Holocene (Chapter 8). This would suggest that it has not been especially active since then. Drainage in the northern half of the fan is now defunct and mostly covered by sand sheets and vegetated linear dunes (Gunn, 1982, Fig. 6.4). Drainage is more active in the southern sector although here too there are extensive patches of sandy soils, possibly relating to a time when the dune field was active much farther to the south. Linear dunes extend south of the upper reaches of the Khor Abu Habl. Coarse sandy alluvium transported north from the Nuba Mountains provides sand for the dunes and sediment for the distributary channels that radiate across the fan. The northern dunes close to the river show evidence of periodic activity during drier intervals in the Holocene (Williams et al., 2010).

136

Khor Abu Habl Fan and Desert Dunes of Kordofan and Darfur

10.2.3 Interactions between Dunes and Distributary Channels We know remarkably little about the alluvial history of the Khor Abu Habl fan, the age of the channels that flow across the fan and whether or not these channels are still active. To provide some initial age estimates for the distributary channels, Williams et al. (2010) examined three sites in the distal reaches of the fan. One site (13°08 0 N, 32°31 0 E) was a shallow well dug in a seasonally flooded depression between two stable and vegetated dunes. Three freshwater shell samples and one charcoal sample collected from depths between 75 cm and 1.7 m had calibrated AMS 14C ages between modern and 200 years BP. A second site located 3 km further north was a well 4.75 m deep. Three shells of the amphibious mollusca Pila wernei and Lanistes carinatus were collected from moist grey sandy clay that had just been excavated from the well. They yielded calibrated AMS 14 C ages 1.0, 1.1 and 1.7 ka, with very small error terms. The third alluvial site was located near the fan apex at 13°00 0 N, 31°50 0 E, where a trench 1.5 m deep was dug next to a seasonal channel. The channel was lined with at least 2.1 m of clay and was covered with a 10–15 cm mantle of loose sand. The trench revealed a sequence of wind-blown sands, alluvial fine and coarse sands, and clayey sands. At 1.5 m depth, a fragment of green bottle glass with a probable maximum age of fifty years showed that the channel was still active, although previously stable dunes were now encroaching on the channel which had become much narrower by this time (Williams et al., 2010).

10.3 Desert Dunes and Their Environmental Significance 10.3.1 Factors That Control Dune Movement Before discussing the dunes (‘qōz’) of Kordofan and Darfur, it will be helpful to consider dunes in their wider context. Figure 4.5 in Chapter 4 shows the limits of active and fixed dunes in and beyond the Sahara. The present-day limit of active dunes is bounded very roughly by the 150 mm isohyet. Fixed dunes extend up to 500 km south of the Sahara, locally into areas that now receive up to 1000 mm of mean annual rainfall. Dune formation and movement reflect the interaction of three main factors: sand supply, wind velocity and vegetation cover. Bagnold (1941) tested his observations of dune pattern and form in the Libyan Desert with wind tunnel experiments conducted at home in which he examined the relationship between movement of sand grains and wind velocity. He found that the saltation of individual sand grains took place at a threshold wind velocity of 4 m/sec. Other workers have obtained threshold values of 6 m/sec (Wasson et al., 1983), from which we can conclude that sand movement is likely to occur once the wind speed is between about 4 and 6 m/sec. Bagnold also observed an exponential increase in the amount of sand movement as wind velocity increased. Dune mobilisation is not in itself a sign of aridity, however. Rather, as pointed out above, it reflects a complex interaction between sand supply, wind velocity and vegetation cover (Tchakerian, 1995, 2009). In the Namib and Kalahari Deserts of southern Africa, Stone and Thomas (2008) drew upon a large number of luminescence ages to infer that the dunes were

10.3 Desert Dunes and Their Significance

137

close to their movement threshold throughout much of the late Quaternary, and that a slight reduction in plant cover would have been sufficient to initiate dune movement. (A luminescence age gives the last time a grain of quartz sand within the dune was exposed to sunlight immediately before becoming buried beneath a further influx of sand.) In the coastal stretches of the southern Negev Desert in Israel the linear dunes were mobile during the humid late Pleistocene and not, as one might have expected, during times of peak aridity. The reasons for this apparent anomaly were an abundant sand supply at this time associated with stronger onshore winds caused by stormy winter cyclones blowing in from the eastern Mediterranean (Roskin et al., 2011a, 2011b). Chase (2009) had earlier concluded that in the case of the southern African dunes the influence of wind velocity could outweigh that of aridity in causing dune movement.

10.3.2 Interactions between Eolian and Fluvial Systems Many workers have drawn attention to the orientation of linear dunes in the Sahara (Wilson, 1971; Mainguet and Canon, 1976; Vail, 1978; Grove, 1980; Mainguet et al., 1980; Cooke et al., 1993; Tchakerian, 1995; Warren, 2013). These dunes are aligned parallel to the dominant sand-transporting winds and follow an elliptical pattern consistent with the winds blowing from the high pressure or anticyclonic cell located across the central Sahara. This anticlockwise pattern of dune movement is interrupted wherever sand-moving winds meet a major or minor topographic obstruction, in which case the dunes are deflected around the mountain or inselberg, as shown on Figure 13.7 in Chapter 13. Examples in or close to the Nile Basin include Jebel Marra (Williams et al., 1980), and Jebel ‘Uweinat and Jebel Arkenu in the southeast Libyan Desert (Williams and Hall, 1965). What is less apparent is that the major sand seas or ergs in fact occupy sedimentary depressions, providing an obvious clue to the question of where the sand came from (Williams, 2014). Rivers flowing from the desert uplands into lowland valleys and closed depressions were the ultimate source of all the sand within the dune systems. There is a constant interplay between erosion of dunes by large actively flowing rivers and obstruction of drainage from dune barriers during times of reduced river flow and greater dune movement (Liu and Coulthard, 2014; Williams, 2015). As climatic desiccation progressed during the course of the late Cenozoic, the once perennial Saharan rivers became seasonal, then ephemeral and in some cases dried out altogether, such as the Miocene Sahabi Rivers that once flowed from northern Chad across the Libyan Desert and into the Mediterranean (Griffin, 1999, 2002, 2006, 2011). Other Saharan rivers were active during wet interglacial phases (Armitage et al., 2007; Drake et al., 2011; Coulthard et al., 2013), allowing further transport of alluvial sand to be fashioned into dunes during the ensuing arid phases. Even during quite arid climatic phases, ephemeral desert streams, which typically have very high rates of bed load transport (Laronne and Reid, 1993), would still be able to contribute sandy material to dunes in their distal reaches.

138

Khor Abu Habl Fan and Desert Dunes of Kordofan and Darfur Dune sand Incised channel

Soil Reworked dune sand Incised channel

Fixed dune

Fan

Figure 10.3 Cross section through an alluvial fan flanking an eroded dune at Janjari, Niger, southern Sahara. (Modified from Talbot and Williams, 1979, Fig. 7.)

10.3.3 Preservation of Dunes Dunes tend to be quite ephemeral features of the landscape. During wetter climatic phases the dunes become vegetated and trap any wind-blown dust. As the fine silt and clay dust particles are washed down into the sands, soils will develop and help to further stabilise the former dune surface. During intense sporadic downpours, gullies develop along the flanks of the now stable dunes and deposit sediment further downstream in the form of small alluvial fans (Talbot and Williams, 1978, 1979). Soils develop in turn on these fans, so that in cross section they reveal alternating layers of reworked sand and weakly developed soil horizons (Fig. 10.3). The final outcome of the combined impact of rill and gully erosion, slow and rapid forms of mass movement, and erosion by raindrop impact and overland flow is the creation of a very gently undulating sand plain on which it is hard to discern much trace of the former dunes. The end result of dune degradation under a more humid climatic regime is a stable reddish sandy surface with a surface crust up to 5 cm thick that is often bound by cyanophytes and fungi. A weakly developed soil is usually found beneath the surface crust. Talbot (1985) has described these surfaces, which are in part erosional and in part depositional in origin, as ‘major bounding surfaces’. Such highly stable surfaces are very widespread along the southern margins of the Sahara, and are evidence of a major change in climate from arid to humid. Based on his experience studying desert dunes in Tunisia, Swezey (2003) concluded that for dunes to be preserved, rapid burial beneath coastal, lacustrine or alluvial deposits is essential. In his review of the published ages of dunes in the Sahara, Swezey (2001) found that many of the Saharan dunes once thought to have last been mobile during the cold, dry and windy Last Glacial Maximum (21 ± 2 ka) were in fact reactivated during the cold Younger Dryas stadial (12.8–11.5 ka), which was also a time of regional aridity across

10.4 Desert Dunes of Kordofan and Darfur

139

much of inter-tropical Africa (Roberts et al., 1993; Gasse, 2000b) as well as further south (Abell and Plug, 2000). Another situation in which the record of past dune activity can be preserved is when soils develop quite rapidly on the surface of the dunes during less arid intervals but ones in which rain storms scavenge calcareous dust from the atmosphere and deposit it onto the porous and sparsely vegetated dune surface. The calcium carbonate dissolved in the slightly acidic rain (essentially a weak solution of carbonic acid) penetrates the soil as far as the wetting front, where the calcium carbonate is precipitated out of solution. Soft aggregates of carbonate eventually become hard concretions and finally coalesce into a band of massive and highly resistant calcrete. In the Rajasthan Desert of India near the now saline Lake Didwana, Singhvi et al. (2010) have obtained a series of luminescence ages from a 200-ka polygenic dune profile revealed in a trench dug to a depth of 18.3 m. Twelve episodes of dune sand accretion and twelve calcrete and/or buried soil horizons are evident in this section. The duration of each wet–dry cycle amounted to roughly 20,000 years, indicating a strong precessional influence on the Indian summer monsoon. With these caveats in mind, it is time to consider the dunes of Kordofan

10.4 The Desert Dunes of Kordofan and Darfur 10.4.1 Previous Work In a benchmark paper entitled ‘Quaternary landforms and climate on the south side of the Sahara’ Grove and Warren (1968) used the air photos, maps and publications then available to construct a detailed map showing the former lake shorelines and the orientation of the now vegetated and inactive dunes along the southern margin of the Sahara. They also reviewed the radiocarbon ages obtained on freshwater shells associated with former lake shorelines and identified four main climatic phases. These were (a) a Middle and Late Pleistocene arid period (or periods) during which desert dunes were active over 500 km south of their present limits; (b) a very wet phase during which the dunes were vegetated and stabilised; (c) a brief dry phase leading to further dune activity; and (d) a final wet phase, associated with expansion of Lake Chad and prehistoric human occupation of Wadi Howar in northern Sudan. Another pioneering study this time focused specifically on the Holocene and older climates of the Sudan is that of Warren (1970). He investigated the dominant sandmoving wind directions and the alignments of the vegetated dunes of Kordofan Province west of the White Nile. From the alignments of several generations of these now fixed linear dunes (which he called the Low Qoz and the High Qoz), he postulated a sequence of late Quaternary shifts in the wind and rainfall zones. In the absence of any dates, Warren tentatively considered the first and most arid phase to be late Pleistocene or older. He inferred a 450-km shift of both wind and rainfall belts to the south of their present position during this time. Sand dunes were thought to have been active at this time as far south as latitude 10°N in Qoz Dango and Qoz Salsigo in southern Darfur (Warren, 1970), a figure

140

Khor Abu Habl Fan and Desert Dunes of Kordofan and Darfur

later amended by Parry and Wickens (1981) to 9°30 0 N. Parry and Wickens (1981) investigated these two localities in some detail and concluded that they had originally been part of a larger dune field before becoming separated by fluvial erosion during a humid climatic phase. Both Bagnold (1933) and Sandford (1953) had observed much earlier that the Low Qoz in northern Sudan had been dissected by stream action. Parry and Wickens (1981) also noted that the sands of which the dunes were composed were derived from local sediments in both areas. In their exploratory journey from El Obeid to Jebel Marra, Lynes and Campbell Smith (1921, p. 209) described crossing ‘rolling sandy hills among stretches of level plains, in which the sands are sometimes pebbly and hard and clad more with scrub and grass than with bush’. It therefore seems highly improbable that the ultimate source of the dunes of Kordofan and Darfur originated in northern Libya, in the Sirte Basin and northern Cyrenaica, as suggested by Vail (1978, p. 31). A local source from weathered Nubian Sandstone and Basement Complex rocks with reworking by local streams seems a sufficient explanation. Warren (1970) regarded the wet phase that followed this arid interval as equivalent to the terminal Pleistocene–early Holocene (15–8 ka) wet phase dated elsewhere along the southern margins of the Sahara. Warren estimated that during this time there was a shift of the climatic and vegetation belts to roughly 250 km north of their present position, a conclusion accepted by Wickens (1975, 1982) in his reconstruction of the hypothetical early Holocene vegetation of the Sudan. Wickens (1982) pointed out that such a shift would deliver nearly 500 mm of rain as far north as the latitude of Khartoum, which presently receives about 175 mm of rain a year. He considered that this shift in the rainfall belts would account for the now isolated presence of forest species such as Trema orientalis, Casearia barteri and Polyscias fulva in the gallery forests of Jebel Marra (Wickens, 1976a, 1976b). The second arid phase was brief and only involved a 200-km southward shift of the wind and rainfall belts. The final moist phase Warren equated with the moist Neolithic phase dated to 7–5 ka in Chad. The reported presence of domestic cattle in the Gilf Kebir and at Jebel ‘Uweinat in the far southeast of the Libyan Desert (El Baz et al., 1980) prompted Wickens to revise his earlier estimate of a northward shift of the early to mid-Holocene climatic belts from 250 km to 500 km (Parry and Wickens, 1981, pp. 310–311). We revisit these estimates in Chapters 15 and 16. The report of the Soil Conservation Committee presented by Whyte (1951) had earlier speculated that during the late Quaternary there was a minimum northward shift of the present-day 200 mm isohyet of about 550 km. This estimate was derived from the reported presence of Neolithic artefacts and subfossil freshwater shells scattered across wide areas of the Libyan Desert (e.g., Sandford, 1933a; Shaw, 1936). We return to this topic in Chapter 16. Wickens (1982, p. 25) considered that the Conservation Committee had ‘obviously underestimated the significant response that can be obtained from quite small increases in annual precipitation. A northerly shift of the 200 mm isohyet by approximately 250 km would include the Wadi Howar area and even be sufficient to support life away from the natural drainage lines. Today permanent settlement with cultivation under similar rainfall conditions exist in northern Kordofan’.

10.4 Desert Dunes of Kordofan and Darfur

141

Wickens (1982), citing Shaw (1936), commented that Wadi Howar had enough water and vegetation at that time (i.e., the 1930s) to support for at least some months of the year such animals as oryx, addax, ril, gazelle, giraffe, ostrich, hyena, jackal, fox, red hussar monkey, ant-bear, porcupine, lion and hunting dogs. No doubt many of these animals have since been made locally extinct by occasional hunting parties armed with rifles.

10.4.2 Holocene Dune Activity To check whether or not the now stable dunes in the distal sector of the Khor Abu Habl fan had been active earlier in the Holocene, a trench was dug on the summit of a well-vegetated dune to a depth of 1.9 m. Three samples of quartz sand collected from depths of 0.53, 1.03 and 1.66 m had Optically Stimulated Luminescence (OSL) ages of 2.9 ± 0.5, 4.8 ± 0.9 and 6.6 ± 0.9 ka, respectively. These ages are progressively older with depth and are consistent with the dune stratigraphy. The three ages are broadly similar to the first three of the four periods of low Holocene lake levels (8.0–6.7 ka, 5.9–4.7 ka, 3.6–3.0 ka and 0.7–0.6 ka) dated by Verschuren et al. (2009) at Lake Challa, already discussed in Chapter 7. They are also consistent with three of the six periods of rapid Holocene climate change recognised by Mayewski et al. (2004), during which polar cooling was coeval with tropical aridity. Rainforest decreased significantly in southern Cameroon at 2.5–2.4 ka, most likely caused by a southward shift of the Intertropical Convergence Zone (ITCZ) at this time (Ngomanda et al., 2009). There was also a sharp decrease in precipitation along the southern Dead Sea at 3.9 ka (Frumkin, 2009), so that the youngest dated phase of dune accretion on the Khor Abu Habl fan may reflect a more widespread period of aridity, but until more sites have been studied and dated, further speculation is not warranted.

10.4.3 Modern Dune Activity The dune or qoz soils of Darfur and Kordofan consist of deep reddish sands with about 90% of their particles between 0.01 and 1.0 mm in diameter. Under ideal conditions they are cultivated for 4 years with a 12-year fallow under Acacia senegal, the main source of gum arabic (Jewitt, 1950). However, as Jewitt and Manton (1954) noted more than half a century ago, population increase and in particular overpopulation close to permanent sources of water has led to a reduction in the length of fallow and a reduction in fertility and soil structural stability, making the soils very liable to wind erosion. Dunes have recently become active in a number of localities in Kordofan (Mohamed et al., 1982). Soils left for 30 years under acacia forest have a higher pH and a greater exchangeable calcium content than soils cultivated for 30 years. The forest soils, when cultivated for groundnuts and sesame, show much higher yields than soils cultivated away from forests. Soil exhaustion and the reactivation of once stable dunes in Kordofan Province are now seen as a major problem and have stimulated research into how best to halt sand movement and re-establish a stable plant cover (Mohamed et al., 1982; El Tahir et al., 2004).

142

Khor Abu Habl Fan and Desert Dunes of Kordofan and Darfur

10.5 Freshwater Mollusca and Holocene Lakes Andrew (1948, p. 107) commented that ‘during the early stages of the “qōz” period, lakes were more common than at present; there are deposits of freshwater limestone and diatombeds interbedded with the lowest “qōz” sands in many parts of Darfur and northern Kordofan’. Whiteman (1971, p. 139) also refers to this statement, but unfortunately neither author provides any specific local evidence beyond vague and quite irrelevant references to Chad and Tibesti. We therefore know virtually nothing about the early stages of ponded drainage within the dune fields of western Sudan. The Holocene record is better documented. Huckriede and Venzlaff (1962) refer to a pluvialzeitliche Molluskenfauna (‘snail fauna dating to a pluvial period’) in Kordofan. In December 1976, Dr W. H. Morton collected subfossil gastropod shells for the author from a former lake near En Nahud in Kordofan (SUA-606: N12°44 0 , E 28°26 0 ) for which we obtained a 14C age of 6,800 ± 110 years BP (calibrated age 7,650 ± 90 years BP) (Williams and Adamson, 1980; Williams, 2009). Former ponds west of the lower White Nile near Wadi Mansurab have calibrated 14C ages between 9.9 and 7.6 ka (Williams and Jacobsen, 2011), and a now dry lake west of the main Nile in northern Sudan was full and fresh between 9.5 and 7.5 ka (Williams et al., 2010). This lake was fed from overflow from the Nile during times of very high flood level. Scattered along the former lake margins there are sporadic Mesolithic and Neolithic artefacts. It thus appears that the most recent wet phase of any significance in this region lasted at least intermittently from about 10 ka until about 7.5 ka.

10.6 Conclusion The Khor Abu Habl fan is a vast low-angle alluvial fan of radius 150 km and area roughly 112,500 km2. Running from north to south across the fan surface are at least two generations of linear desert dunes. The earliest distinguishable phase of dune activity involved a southward shift of the rainfall belt in this region by about 450 km. This interval probably spanned both the Last Glacial Maximum (21 ± 2 ka) and the ensuing cold Younger Dryas stadial (12.8–11.5 ka), during which the regional climate was cold, dry and windy. A wetter phase followed, most likely between about 15 ka and 8 ka, with a northward shift of the rainfall belts by roughly 250 km. Small lakes were present in the inter-dune hollows and Mesolithic and Neolithic sites are common in presently arid areas. A later phase of aridity led to dune reactivation and a possible southward shift of the rainfall belts by about 200 km. Dunes were intermittently active throughout the Holocene, with identified dune activity and sand accretion at 2.9 ± 0.5, 4.8 ± 0.9 and 6.6 ± 0.9 ka. These dated intervals of sand movement were coeval with low stands of Lake Challa (see Chapter 7) and with three periods of widespread polar cooling and tropical aridity.

11 The Gezira Alluvial Fan and Blue Nile Palaeochannels

Gezira was a land of mirage. At dawn in winter the horizon stood up like a pink cliff circling a giant hollow in which a curious refraction of light disclosed villages and fields beyond the range of normal sight. Arthur Gaitskell (1900–1985), Gezira: A Story of Development in the Sudan (1959, p. 26)

11.1 Introduction The Gezira (Arabic for island or peninsula) is a wedge-shaped tract of land bounded by the Blue Nile to the east, the White Nile to the west and the Manaqil Ridge to the south. The southern boundary is somewhat arbitrary and is often taken as the Sennar–Kosti railway line, which for simplicity we also use here (Fig. 11.1). The northern apex of the wedge is the confluence of the Blue and White Nile at Khartoum, the capital of Sudan. The surface of the Gezira is mantled in clay. A series of now defunct Blue Nile distributary channels radiate across the Gezira from the Blue Nile, indicating that the Gezira is a low-angle alluvial fan (Fig. 11.1). This fan has a radius of about 230 km and covers an area of roughly 25,000 km2, making it one of the world’s few mega-fans (Leier et al., 2005). If we also include the area covered by the Manaqil Ridge north of the Sennar–Kosti railway line, which amounts to about 5,000 km2, then the total area of the Gezira is just over 30,000 km2. Since the 1920s, irrigated cotton has been grown on the Gezira, and in the 1950s and 1960s high quality long-staple cotton grown on barely 1% of the land area of the country provided well over half of the export revenue of the Sudan. Because of its agricultural and economic importance, the Gezira has long attracted the attention of irrigation engineers, agronomists and soil scientists. To this day, it remains the agricultural heartland of the Sudan. Less well known, because often highly disturbed, are scattered signs of Holocene prehistoric occupation, including the site of Jebel Moya in the far south of the Gezira, which was excavated before the First World War by the eccentric and wealthy philanthropist Sir Henry Wellcome.

143

144

Gezira Alluvial Fan and Blue Nile Palaeochannels

KHARTOUM

LEGEND Well defined palaeochannel

380

Poorly defined palaeochannel Masoudia El Masid

J.Aulia

Meander cutoff

39 0

15°N

H

Branko

Artimeili

El Geteina

W

E NI L

NILE

UE BL

Volcanic ash

N

E IT

Hasaheisa Wad es Said

Naima

0 40

Wad Medani

Wad ez Zaki Managil

Hashaba

14°N

d ha Ra

0 41 Ed Dueim MANAGIL

Di nd er

RIDGE Kawa Esh Shawal

Sennar

Jebel Tomat

Jebel Moya

50

0 Kosti 13°N

Singa

km

Contour interval : 5m 33°E

34°E

Figure 11.1 Late Quaternary Blue Nile palaeochannels on the Gezira alluvial fan, central Sudan, showing location of dated sites. (After Williams, 2009a, Fig. 3a and Williams et al., 2015a, Fig. 1a.)

11.2 Age and Origin of the Gezira

145

11.2 Age and Origin of the Gezira The physical and chemical properties of the Gezira clay have been widely studied (Jewitt, 1955; Beinroth, 1966; Buursink, 1971; Blockhuis, 1993) but it was the pioneering work of J. D. Tothill (1946a, 1948) that first shed real light on the age and origin of the Gezira clay. Previous workers had proposed that the Gezira clay had developed either by weathering of the underlying bedrock or else had formed from wind-blown desert dust blown in from the north during the haboob season. Grabham, (1917, 1934, p. 19) argued nearly a century ago that the Gezira clay was in fact loess, that is, wind-blown dust. Dust storms or haboobs occur in the far north of Sudan at least once a year and become more frequent further south, with Khartoum having an average of 18–20 per year (Griffiths and Soliman, 1972, p. 93). Our concern here is with the type (c) haboob coming from the north (see Chapter 3) in the latter half of the dry season between February and May. It is hard to estimate the total volume of dust blown across the Gezira by these haboobs but they do deposit a great deal of silt and very fine sand. The silt content (defined here as mineral particles 2–20 µm in diameter) in the upper horizon of all soils in the 300 km long tract of land between the White Nile and the western margin of the Gezira decreases progressively from north to south (see particle size data in Hunting Technical Services Limited, 1964). It is therefore likely that desert dust from the north does provide some material to the soils of the Gezira and may well have provided more during drier and windier climatic episodes. Wells dug across the Manaqil Ridge and across the clay plains of central Sudan more generally (Gibb, 1954; Ruxton and Berry, 1978; Gunn, 1982) show that the clay soils grade into deeply weathered bedrock. The presence in this area of scattered inselbergs or erosional remnants of Basement Complex igneous and metamorphic rocks, especially granites and gneisses, points to exhumation of an irregular deep weathering front during the Cenozoic and possibly earlier. There may well have been several phases of landscape stability and deep weathering followed by erosion and stripping of the weathering front. It is therefore highly likely that the soils of the Manaqil Ridge are predominantly sedentary soils formed by bedrock weathering in situ, although some soil movement has probably occurred under the influence of gravity and slope-wash. Ruxton and Berry (1978) consider that the sedentary clays of the central Sudan are upper Pliocene to Middle Pleistocene in age, but that the clays along the Blue Nile are younger and are alluvial. Tothill (1946a, 1948) was the first to demonstrate conclusively that the Gezira clay flanking the Manaqil Ridge is alluvial in origin. Systematic collection and identification of shells found in a series of soil pits dug to depths of 6–10 feet (1.8–3 m) revealed the presence of aquatic, semi-aquatic and terrestrial species of snails. The aquatic gastropod Cleopatra bulimoides was abundant up to 30 km west of the present Blue Nile channel and decreased in frequency from the base of the pits upwards. Across the Gezira Melanoides tuberculata was common on both sides of the White Nile. The two semi-aquatic species Pila wernei and Lanistes carinatus were ubiquitous. Of particular interest was the widespread presence of the large land snail Limicolaria flammata, which lives today in the acacia-tall grass savanna region of central Sudan where the rainfall amounts to 400–800

146

Gezira Alluvial Fan and Blue Nile Palaeochannels

mm per year (Tothill, 1946a, 1948). Tothill (1946a) concluded that the presence of both adult and juvenile Cleopatra pointed to flooding by the Blue Nile for 3–4 months every year, followed by rapid retreat of the flood waters and death of the snails once the waters withdrew. He interpreted the occurrence of Melanoides along the White Nile as indicative of permanent flooding and the presence of a White Nile lake or greatly expanded river. In Chapter 8 we noted Tothill’s discovery of Melanoides tuberculata shells at 382 m near the summit of a wave-trimmed sand ridge at Hashaba. The presence of Limicolaria in Gezira clay soils meant that as the Blue Nile cut down and the zone of flooding diminished, the climate was still much wetter than today, with more than twice the present rainfall and a longer summer wet season. The age of the clays was still unresolved, although Tothill (1946a) surmised that they were probably Holocene in age. Williams (1966) obtained the first radiocarbon ages for subfossil shells in the western Gezira and showed that the alluvial clays east of the lower White Nile were indeed Holocene in age. One shell sample (I-1485) came from a 1.9–1.6 m depth in a pit near Tagra village and had a 14C age of 8,700 ± 350 BP (calibrated age 9,750 ± 440 BP). The other shell sample (I-1486) came from 1.7 to 1.45 m depth in a pit 4.5 km east of Esh Shawal and had a 14C age of 11,300 ± 400 BP (calibrated age 13,230 ± 400 BP). Three additional 14C ages from the Tagra pit site and five from the Esh Shawal shell bed subsequently confirmed the accuracy of these two initial ages (Williams, 2009a). During the 1970s and early 1980s Williams and Adamson (1982) embarked on a systematic program of radiocarbon dating subfossil shells collected from soil trenches dug across the Gezira as well as from the White and Blue Nile to test for any reservoir effects. The results showed that during the Holocene high flood levels in the Blue Nile had calibrated 14C ages of 13.9–13.2 ka, 8.6 ka, 7.7 ka and 6.3 ka (Williams, 2009a). The corresponding high White Nile flood ages are 14.7–13.1 ka, 9.7–9.0 ka, 7.9–7.6 ka, 6.3 ka and 3.2–2.8 ka (see Chapter 8). The Blue Nile flood record is less complete than that of the White Nile because the Blue Nile has a much steeper gradient, is highly seasonal and is a far more erosive river during flood so that part of the depositional record is missing. When the shell data of Tothill (1946a) are collated and plotted in graphical form against depth they reveal an informative vertical distribution (Adamson et al., 1982, Fig. 9.8A) (Fig. 11.2). Cleopatra shows a steady decrease in frequency from 160 to 125 cm depth, increases briefly to a secondary peak at 105 cm depth and then dwindles to zero at 50 cm depth. The aquatic gastropod Cleopatra requires permanent water, so that its disappearance implies that widespread flooding by the Blue Nile had ceased, most likely because the Blue Nile was cutting down vertically from about 8000 years ago onwards (Arkell, 1949b; Adamson et al., 1982). The land snail Limicolaria flammata is most abundant in the upper few cm of the Gezira clay and declines to zero by 150 cm depth. This snail is intolerant of flooding and is today a denizen of the acacia-tall grass savanna region to the south, with a minimum annual precipitation of 400 mm. Precipitation in the Gezira during 1921–1950 amounted to 150 mm in the far north and increased progressively to 400–450 mm in the far south (Randell, 1961; Williams et al., 1982, Fig. 7.15).

11.2 Age and Origin of the Gezira

147

Figure 11.2 The changing proportions of aquatic, amphibious and terrestrial mollusca in the upper two metres of Gezira clay during the past 15,000 years. Data from Tothill (1946a) collated and plotted against depth. The numbers represent the total number of individual shells of each species. Cleopatra bulimoides (Oliv.) is an aquatic gastropod dependent on permanent water. Limicolaria flammata (Caill.) is a land snail intolerant of flooding; it now lives in areas with a minimum rainfall of 400 mm. Lanistes carinatus (Oliv.) is an amphibious species with gills and lungs. (Modified from Adamson et al., 1980, Fig. 9.8A.)

Tothill (1948, p. 138) also drew attention to the presence of gypsum in clay soils laid down by the Blue Nile and its absence from White Nile clay soils south of Renk. He noted that gypsum is deposited in the presence of sodium carbonate, which is abundant in the Ethiopian headwaters of the Blue Nile, so that clay soils with gypsum are likely to have been originally deposited by that river. Andrew (1948, p. 110) cited the presence of augite and zeolites in support of an Ethiopian volcanic provenance of the clays and underlying alluvial sands and gravels, and Shukri (1949) confirmed that the heavy minerals found in the Blue Nile differed substantially from those of the White Nile and its tributaries, work that has since been greatly amplified by Padoan et al. (2011) and by Garzanti et al. (2006, 2015). Tothill (1946a, p. 162; 1948, p. 139) also noted that there was a salty layer at depths of 1–2 m in many Gezira clay soils, which he attributed to a drier phase post-dating the time when the large land snail Limicolaria flammata was abundant across the Gezira. At the site of Jebel Tomat, Adamson et al. (1982, Table 9.1) obtained a 14C age of 4,500 ± 130 BP (calibrated age 5,160 ± 206 BP) for Limicolaria at a depth of 60–80 cm on the contact between surface colluvium with occupational debris and the underlying dark alluvial clay. This may denote the final phase of Limicolaria

148

Gezira Alluvial Fan and Blue Nile Palaeochannels

occupation in the southern Gezira, after which desiccation set in and the snails migrated south. One further comment by Tothill (1948, p. 139) is worth citing: ‘The adult semi-fossil shells of Limicolaria obtained by A. J. Arkell at his Hospital site excavation [see Arkell, 1949b] are notably smaller than is normal to-day for Suki and Gedaref specimens, which suggests that the climate was changing to the present one in which this conspicuous coneshaped land mollusc is uncommon north of Singa and unknown except as a semi-fossil north of Wad Medani.’ In fact, as we shall see in Section 11.3.2, the Gezira clay is not a single depositional unit, but was deposited in stages and at different times, so that the surface of the Gezira alluvial fan is time-transgressive.

11.3 Blue Nile Palaeochannels 11.3.1 Mapping the Blue Nile Palaeochannels The Gezira Irrigation Scheme is one of the largest and most enduring major irrigation projects on Earth. Blue Nile water from the Sennar Dam reservoir flows under gravity all the way across the Gezira into a network of major and minor canals. Because the irrigation is gravity fed, detailed contour maps were constructed to facilitate optimum location of the canals. As a result, the entire Gezira has been mapped at 50-cm contour intervals. Steel beacons with the coordinates shown in English and Arabic marked the precise location of every minute of latitude and longitude. Villagers have long since purloined many of the steel posts for other purposes but the stumps remain and still allow accurate navigation when plotted onto 1:25,000 scale air photos and 1:50,000 scale air photo mosaics, and checked against Global Positioning Satellite (GPS) hand-held devices. The 1:50,000 contours maps and survey beacons allowed a series of former Blue Nile channels to be identified and mapped with some accuracy (Williams and Adamson, 1980, Fig. 12.6; Adamson et al., 1982, Fig. 9.3). These channels radiated out northwest across the Gezira from a variety of former Blue Nile outlets between Sennar and Wad Medani. Earlier soil surveys in the northern Gezira had revealed a series of discontinuous linear sandy ridges whose summits were a few metres above the surrounding clay plains (Fig. 11.1). These low sandy rises, locally known as qoz, are associated with the former channels and are often the site of present-day villages. They represent local dunes blown out from the channel sands. Pits dug into the channel sands show that they consist of alternating coarse and fine sand, sometimes displaying clear current-bedding (Hunting Technical Services, 1963; Williams and Adamson, 1973). The channels are former Blue Nile distributary channels and were active until the Blue Nile incised its channels and beheaded the distributary channels, which then ceased to flow.

11.3 Blue Nile Palaeochannels

149

11.3.2 Age of the Blue Nile Palaeochannels Initial attempts to date the palaeochannels relied on chance finds of shell or charcoal for C dating. On this basis, it was possible to show that the channels had been intermittently active between about 50 ka and 8–5 ka (Adamson et al., 1982). A new chapter opened with the advent of luminescence dating. This method can show the time when grains of quartz or feldspar were last exposed to sunlight (Williams, 2014, Table 6.1) and does not rely on finding fossil organic remains. This is especially important because in arid environments occurrences of organic matter in former alluvial or eolian deposits can be very rare. Another advantage of luminescence dating is that it can provide reliable ages back to at least 500 ka and under ideal conditions back to about a million years (see Williams, 2014, pp. 92–94), whereas 14C dating has an effective range of 0–50 ka. The oldest Blue Nile palaeochannel sediments logged and dated so far are those exposed in a series of gravel quarries near the small town of Masoudia located 50 km southeast of Khartoum and 0.5 km west of the Blue Nile (Fig. 11.1). The sediments consist of planar and cross-bedded fluvial sandy gravel beds capped by up to 5 m of dark Gezira clay. Etheria elliptica (Nile oyster) shells in the base of the clay unit had calibrated AMS 14 C age of 40,949 ± 477 years BP and 43,411 ± 545 BP. The Optically Stimulated Luminescence (OSL) age near the base of the grey clay was 52 ± 7 ka (Williams et al., 2015a, Tables 1 and 2). The presence of the Nile oyster shells in the lower 10–15 cm of the clay unit is of interest because it seems to reflect sudden death caused by mass burial of the oysters resulting from a sudden influx of abundant fine sediment – the overbank flood deposits from the suspension load of a rejuvenated Blue Nile. The preferred habitat of Etheria elliptica is a rocky substrate and they can withstand fast flowing water but die if smothered in mud. The age of this event is also of note. Across the Gezira just east of the White Nile and 2 km north of Naima village alluvial gravels with oyster shells were exposed in three gravel pits in 1973. The small quarries were located at the distal end of a former Blue Nile distributary channel. The shells had a radiocarbon age of 38,900 ± 350 14C years BP and a calibrated age of 43,766 ± 355 years BP, essentially the same age as the Masoudia oysters. The interval 52 ± 7 ka was a time of active flow in the Blue Nile and deposition of clay across the northern Gezira flood plain and coincides in time with the formation of Sapropel 2 (52 ka) in the eastern Mediterranean, during a phase of much enhanced Nile discharge to the sea (see Chapter 21). Sand lenses in two of the lower gravel units have OSL ages of 270 ± 30 ka and 190 ± 20 ka (Williams et al., 2015a, Table 1). A prolonged phase of soil formation followed the older of the two gravel units and was followed by a further interval of gravel deposition and a second phase of soil development. Both phases of soil formation were characterised by precipitation of large irregular calcium carbonate nodules and rhizocretions, most probably under a semi-arid climate comparable to that of today. In the banks of the modern Blue Nile, carbonate can be seen precipitated around the roots of trees growing along the edge of the channel to form the cylindrical deposits of calcium carbonate known as rhizocretions. A minor phase of erosion followed the second period of soil formation and carbonate 14

150

Gezira Alluvial Fan and Blue Nile Palaeochannels

precipitation, leading to deposition of water-transported and well-rounded carbonate gravels. The alluvial gravels thus have a polygenic origin. The surface clay mantle was then laid down during the flood events of 50–40 ka. There is one very intriguing feature about a number of the palaeochannels visible on the surface of the Gezira, which has never been adequately explained. In their proximal and middle reaches the channels are relatively straight and linear, as one might expect from a bed-load channel transporting sand and gravel. Bed-load channels are generally braided, with relatively steep gradients and non-cohesive banks. Braided channels are often associated with seasonal fluctuations in discharge. A dramatic change takes place in their distal reaches, when the once linear channels become highly sinuous, with large meander wavelengths (Fig. 11.3a). Auger holes drilled into one of these meanders revealed a finingupward depositional sequence from gravel and gravelly sands to fine and medium sands to clayey sands and finally to clay (Fig. 11.3b). A series of trenches dug into the meanders and adjacent former back swamps showed that the channels were transporting and depositing sand in this area between 100 ka and 70 ka, with clay accumulating in the back swamps between 70 ka and 50 ka. Some of the uppermost clay units capping the channel sands had OSL ages of 12–5 ka at depths of 0.8–1.1 m. These are surprisingly young ages. They may relate to a final pulse of Blue Nile overbank flooding, as suggested by Williams et al. (2015a) or they could reflect deposition during times of very high White Nile floods. The sudden change in channel pattern from straight to meandering could be because the channels in this sector were flowing across the floor of a former White Nile seasonal lake, such as the 386-m Last Interglacial Mega-lake, as detailed in Chapter 8. Several channels pass beneath the dunes east of the White Nile, which have OSL ages of 115–105 ka, 60 ka and 12–7 ka (see Section 11.4.3). The dunes also show signs of scalloping along their margins, which was effected during times of high White Nile flooding (Williams, 2009a). Additional ages were obtained from three other localities. On either side of the village of Branko (Fig. 11.1) on the Blue Nile right bank, carbonate cemented fine gravel and sand contain abundant fossil Etheria elliptica (Nile oyster) shells as well as calcified tree trunk fragments up to 4 m long and even fossil rhinoceros bones. The gravelly sand in which the fossil rhinoceros bones were found is located at 14°58 0 N and 33°15 0 E and lies 2 km south of the village. It has an OSL age of 105 ± 6 ka (Williams et al., 2015a). This age is synchronous with that of Sapropel S4 in the eastern Mediterranean, indicative of a phase of sustained and very high Nile discharge into the sea at that time (see Chapter 21). The Dinder is a major right bank tributary of the Blue Nile and rises in the highlands of eastern Ethiopia (Fig. 11.1). Horizontally bedded fine sandy clays and silts exposed near the base of a 15 m high bank of the Dinder at 12°57 0 N and 34°38 0 E near the village of Azaza Damos had OSL ages of 60 ± 4 ka and 58 ± 4 ka. Fossil gastropod and mussel shells in the same stratigraphic unit but 150 m to the south had calibrated AMS 14C ages of 41,303 ± 906 and 39,113 ± 203 cal years BP. One sample gave an AMS 14C age of 47,814 ± 412 14C years BP, or beyond the present calibration age (Williams et al., 2015a). These ages are similar to the calibrated 14C ages of the oyster shells in the Naima and Masoudia gravel pits discussed earlier, and are coeval with Sapropel S2 in the eastern Mediterranean. Williams et al.

11.3 Blue Nile Palaeochannels

151

Figure 11.3 (a) Blue Nile palaeochannels adjacent to the White Nile near El Geteina (see Fig. 11.1 for location). (After Williams, 2009a, Fig. 5a and Williams et al., 2015a, Fig. 4b.)

152

Gezira Alluvial Fan and Blue Nile Palaeochannels

Figure 11.3 (cont.) (b) Stratigraphic sections through the main palaeochannel, indicated by the vertical arrow in (a), show upward fining from the late Pleistocene to the Holocene. (After Williams, 2009a, Fig. 5b and Williams et al., Fig. 4a.)

(2015a) suggested very tentatively that these ages ‘mark the closing stages of a regime characterised by the transport of coarser sediments by the late Pleistocene Dinder and Blue Nile rivers, and its replacement with a suspension load regime of silt and clay’ (Williams et al., 2015a, p. 96). The third site is a sand quarry located on a well-defined former Blue Nile channel north of Singa at 13°16.50 N and 33°430 E. This channel is much younger than those in the northern Gezira and was active during the terminal Pleistocene and Early to Middle Holocene. Aquatic gastropod shells at depths of 1.7 m, 4.7 m and 6.0 m beneath the present ground surface had calibrated AMS 14C ages of 12,952 ± 42, 13,438 ± 42 and 12,943 ± 85 cal years BP, respectively. Three sand samples from depths of 8.8, 7.75 and 7.25 had OSL ages of 8.2 ± 0.7 ka, 14 ± 1 ka and 10 ± 1 ka, respectively. Two of the OSL ages are slightly younger than the shell ages, which are essentially of similar age throughout the deposit. The return of the summer monsoon in the Blue Nile headwaters began about 14.5 ka ago, so that this palaeochannel was active during an interval of much greater Nile flow. Adamson et al. (1982) have described similar deposits of cross-bedded sands exposed in a sand and gravel quarry at Atra located opposite the Rahad-Blue Nile confluence at 14° 28 0 N and 33°29.5 0 E, the shells in which have a 14C age of 11,800 ± 195 years BP and a calibrated 14C age of 13,710 ± 210 cal years BP. Further south, the 14C age of Nile oyster shells from a depth of 0.5 m on the north bank of a palaeochannel 30 km south of Singa located at 13°18 0 N and 33°40 0 E is 11, 330 ± 150 years BP with a calibrated age of 13,260 ±

11.3 Blue Nile Palaeochannels

153

170 cal years BP. About midway between these two localities a pit was dug through typical Gezira clay soil on the Gezira Research Station at Wad Medani at 14°24 0 N and 33°30 0 E. Cleopatra and Lanistes shells from 1.6 to 1.7 m depth had a 14C age of 11,975 ± 260 BP and a calibrated age of 13,900 ± 300 cal years BP (Adamson et al., 1982; Williams, 2009a). What is of particular interest here is that the age of the Gezira clay is coeval with that of the Blue Nile palaeochannel sands. Overbank flooding was therefore active at the same time as the channels were ferrying sand and gravel along the eastern Gezira. In other words, the main Blue Nile channel had not yet begun to incise.

11.3.3 Blue Nile Incision and Demise of the Late Quaternary Distributary Channels A series of high-level abandoned meanders or oxbows are clearly visible on both sides of the Blue Nile north of Sennar and on the west side of the Blue Nile north of Abu Hugar (Fig. 11.1). These oxbows are remnants of a meandering Blue Nile that showed few signs of being incised. Adamson et al. (1982, Table 9.2) collated the elevation of the presently exposed floor of three of these defunct oxbows and found that at five localities they were 6–7 m above the present Blue Nile maximum flood level near those sites. The oxbows consist of fine fluvial sands beneath ≥3 m of fine silt and clay. The sands have a Blue Nile mineral suite. These oxbow depressions would have been a source of water on the surface of the Gezira fan away from the main rivers for several months during and after the rains. When and why did these meandering channels and the Blue Nile distributary channels in the northern Gezira cease to operate? One possible cause could have been a reduction in Blue Nile discharge linked to a decline in regional precipitation. The progressive decrease in the number of permanent water Cleopatra shells is certainly consistent with this hypothesis, but the presence of abundant Limicolaria land snails and the fact that they become progressively more abundant towards the top of the Gezira clay soils shows that the climate at that time was still significantly wetter that it is today. We therefore need to look for another explanation. Adamson et al. (1982) suggested that ‘the explanation lies in a switch of fluvial regime from one characterised by multiple interconnected aggrading channels to one with a single major incising channel’ and concluded that the Gezira palaeochannels became progressively beheaded between Sennar and Wad Medani (Adamson et al., 1982, p. 187). However, these authors were not able to explain why the incision took place in the first place, beyond a vague appeal to changes in discharge and load, and mention of the great increase in Blue Nile flood discharge during the Early Holocene. We need to consider another possible explanation linked to changes in stream power and channel bank erodibility. The ability of a stream to transport its load and erode its channel bed depends upon stream power and resistance to erosion. Stream power (W) is the rate of energy loss per unit length of stream (Bagnold, 1966), which is expressed as the product of tractive force (r) and velocity (v) per unit width of channel:

154

Gezira Alluvial Fan and Blue Nile Palaeochannels

W ¼ rv

ð11:1Þ

Tractive force is the product of the hydraulic radius (R), the slope (S) and the specific weight of the fluid (y). (The hydraulic radius is the submerged channel cross section divided by the submerged channel perimeter.) r ¼ yRS

ð11:2Þ

Schumm (1977) found that both stream power and sediment transport rate were roughly proportional to stream velocity cubed. A reduction in stream power would therefore lead to a reduction in bank erosion and sediment transport and cause local sedimentation within the channel. The nature of the channel bank sediments is important. Unconsolidated sands and gravels are much easier to erode than compact and cohesive clays, which have far greater shear strength than loose sands. Once the Blue Nile became a suspension load channel its channel bed and banks would become lined with clay and its cross section would alter from broad and shallow to deep and narrow. The smooth clay channel sides would mean less frictional loss of kinetic energy and would in turn allow the river more energy to cut down. The Early Holocene increase in discharge would also increase stream velocity and hence stream power, all of which could promote channel incision. This somewhat theoretical model could account for why the Blue Nile entrenched its bed but should be considered a working hypothesis only and not a definitive explanation. When did the Blue Nile incision begin? The abandoned meanders with ages between 14 ka and 13 ka were not deeply incised into the surface of the Gezira fan. The Early Khartoum site excavated by Arkell (1949b) revealed that the flood level at the time of occupation was 10 m higher than today. The age of this site we now know to be about 9 ka (see Chapter 8). Palaeochannels were still active at the sites of Wad es Said (14°40 0 N; 33° 13 0 E.) and Qoz Kabaro (Fig. 11.1) at 7.8 ka (14C age 6,890 ± 115 years BP; calibrated age 7,730 ± 110 cal years BP). The age came from Pila shells near the surface of cross-bedded sands at Wad es Said. The sands show no signs of later disturbance by running water and no signs of burial beneath later fluvial deposit (Adamson et al., 1982). The age of 7.8 ka may therefore denote when this major palaeochannel ceased to flow. Adamson et al. (1982) considered that the Blue Nile was firmly established in its present channel by about 7 ka and that incision had already deprived some of the palaeochannels of their water, although the deepest distributary channels still received some water during peak Blue Nile floods. These conclusions still appear valid.

11.3.4 Blue Nile Palaeochannels Filled with Volcanic Ash In January 1973 my colleague Dr Don Adamson and Dr Hassan Hag Abdulla, a senior soil scientist from the Gezira Research Station at Wad Medani, discovered deposits of very pure volcanic ash exposed in gullies (kerrib) along the left bank of the Blue Nile 25 km NNW of Wad Medani. (Figure 11.4 shows a small segment of an ash-filled former Blue Nile channel

155

399

.2

11.3 Blue Nile Palaeochannels

4.2

2

.2

NILE

398.

2

40

1.

N

40

40

40

3.

2

2.2

14°30'N

400

K173

Volcanic ash

40 3.

BLUE

Gully K174

2 K174

0

Contours in metres a.s.l. Kilometre post on road: distance from Khartoum

500 m

K175

14°29'N 33°27'E

33°28'E

Figure 11.4 Volcanic ash deposits within a former Blue Nile channel west of the present channel of the Blue Nile. (Adapted from Adamson et al., 1982, Fig. 9.2.)

in some detail). They found that these deposits extend in a linear fashion from just north of Wad Medani for at least 83 km. The ash occupies what appear to be one or more former Blue Nile channels. There is a very clear contact between the sloping surface of the clay banks and the ash, indicative of a stream channel that was not entrenched into its flood plain. The ash is finely laminated, with fine cross-bedding and ripple marks. At one site the ash was mixed with clay and with phytoliths, sponge spicules and algal remains (Adamson et al., 1982). The stream channels were only 3–4 m deep and 20–30 m wide, and the biological remains point to an abundance of aquatic vegetation, notably grasses, to supply the abundant phytoliths. The fine sedimentary structures in the ash indicate low energy flow, and the lack of any cut-and-fill structures is consistent with one continuous interval of ash

156

Gezira Alluvial Fan and Blue Nile Palaeochannels

deposition in the channels. Some parts of the channel contain rolled pumice fragments up to 4 cm in diameter. It is not possible to say whether the channels were characteristic of the entire stream system in this area at that time, or whether they were overflow channels connected to a major parent channel. A plausible interpretation is that the ash-filled channels represent avulsion of possible anabranch channels of the main Blue Nile. It seems likely that a massive eruption of pumice and ash in the Ethiopian headwaters covered part of the catchment and choked the river with ash. There is a possible candidate for this volcano on the interfluve between the upper Blue Nile and the upper Omo River in Ethiopia. Dr Frank Brown, who has analysed the geochemistry of tephra (volcanic ash) deposits throughout Ethiopia, very kindly carried out detailed geochemical analysis of ash samples collected by the writer in February 2006 north of Wad Medani. He found that the ash geochemistry in the former Blue Nile channel is similar to that in the lower part of the Kibish Formation (see Chapter 9) in the lower Omo Valley of southern Ethiopia. This would suggest a maximum age of 190 ka and a minimum age of 100 ka for the Blue Nile volcanic ash (McDougall et al., 2005, 2008). The Ethiopian volcano that may have provided the ash is located on the interfluve between the upper Omo and the upper Blue Nile. Although the Blue Nile has changed course in the last 250,000 years, the evidence we have now reviewed from the ancient Blue Nile channels at Masoudia, Branko and Wad Medani shows that it has flowed very close to its present channel at least three times in that period.

11.3.5 Subsurface Stratigraphy of the Gezira Andrews (1948) coined the term Gezira Formation to refer to all unconsolidated clays, silts, sands and gravels overlying the Nubian Sandstone and older Basement Complex rocks in the Gezira area. The Gezira Formation is very variable lithologically, both laterally and vertically, is more than 180 m thick near the centre of the Gezira, and is not always easy to distinguish from weathered Nubian Sandstone in borehole data. In his MSc thesis on the ground water geology of the Gezira, Abdel Salam (1966) compiled data relating to the sand content in boreholes drilled by the Ground Water Section of the Ministry of Natural Resources and Rural Water Development in Khartoum and produced a lithofacies map of the Gezira Formation. What is most striking on this map are three ‘hypothetical Proto-Blue Nile channels’ (El Boushi and Abdel Salaam, 1982, Fig. 5.2). These buried sandy channels radiate northwest across the Gezira from about the present junctions of the Blue Nile with the Rahad and Dinder Rivers (Fig. 11.5), just as do their modern surface counterparts (Fig. 11.1). In 1973 the author received access to recently drilled borehole samples from across the northern Gezira and found in every case that there was an alternation of sands and clays indicative of a series of fining-upward sequences. The simplest explanation is that the Blue Nile alternated between two regimes. During intervals when the climate in the

11.3 Blue Nile Palaeochannels

157

KHARTOUM

Nile

15°30'N

Blu

e

Jebel Aulia

N

W hit e

15°00'N

Ni

El Geteina

le

14°30'N

Wad Medani

Ra

Wad ez Zaki

ha d

0

50

D in de

km

r

14°00'N Ed Dueim

32°30'E

33°00'E

33°30'E

Figure 11.5 Lithofacies map showing buried sandy palaeochannels in the Gezira Formation. (Adapted and simplified from Ismail M. El Boushi and Yassin Abdel Salaam, 1982, Fig. 5.2.)

Ethiopian headwaters was colder and drier than today and the plant cover less extensive, accelerated hill slope erosion would have provided the upper catchment rivers with an abundant load of sand and gravel. The Blue Nile was a bed-load river, aggrading its Gezira fan, with multiple distributary channels transporting and depositing coarse sediment across the fan. Once the climate became warmer and wetter and the plant cover re-established across the headwaters, hill slope erosion would be curtailed. Weathering rates would increase, soils would develop, and slope processes would furnish the drainage channels with a suspension load of clay and silt. Seasonal flooding across the Gezira fan would deposit a progressively thicker layer of clay. In due course the parent Blue Nile channel would incise its bed, the distributary channels would be beheaded and deprived of flow, and the zone of annual deposition of flood silts and clays would become restricted to within a few kilometres of the river. With a return to drier conditions the river would once again carry sand and gravel and resume aggrading its alluvial fan.

158

Gezira Alluvial Fan and Blue Nile Palaeochannels

11.4 Source-Bordering Dunes 11.4.1 Mineralogy of the Source-Bordering Dunes We saw in Chapter 8 that sand dunes occupy about 160,000 ha east of the lower White Nile Valley between Hashaba in the south and Jebel Aulia in the north. Six sites were sampled at depths of 1.0, 2.0 and 2.7 m near the western margin of the sand dunes (Fig. 11.6) and all were found to have a heavy mineral assemblage similar to that found in the bed load sands of the Blue Nile in Ethiopia (Williams and Adamson, 1973, Table 1; Williams, 2013, Appendix 1). The sample collected by the author in December 1971 came from a point-bar in the Abbai/Blue Nile channel at the bottom of the Blue Nile gorge just upstream of the old Italian Bridge in Ethiopia. It contained abundant hornblende and magnetite; common epidote, pyroxene, rutile and quartz; rare biotite, apatite and plagioclase; and traces of allanite, topaz, sillimanite, garnet and chlorite. In the upper Blue Nile catchments granite, syenite, sandstone, gypsite, shale, limestone and basalt crop out, so that we should not expect an exclusively volcanic mineral suite.

33°E

N

Bl ue

Jebel Aulia

1

15°N

Nile

2

3 4 5 6 Hashaba 14°N

ite

Wh

Ni

50 km

le

Figure 11.6 Location of sand dunes in the western Gezira showing sites sampled for heavy mineral analysis. (Adapted from Williams and Adamson, 1973, Fig.1b.)

11.4 Source-Bordering Dunes

159

Tothill (1946a, pp. 163–164) initially considered that the dunes east of the lower White Nile were once part of the main qoz formation of Kordofan formed during a long dry interval and later cut off by the White Nile during a time of renewed flow. He subsequently amended his initial interpretation and considered that the sand dunes were ‘not in origin similar to the continental “qoz” or sand of Kordofan and Darfur but are the remains of lakeside dunes’ (Tothill, 1948, p. 134). However, neither hypothesis helps to explain why all of the dunes are located at the distal end of former Blue Nile distributary channels (Fig. 11.1), and why their heavy mineral assemblage is similar to that of the Ethiopian Blue Nile. The most likely explanation is that these dunes originated as source-bordering dunes derived from Blue Nile distributary channels that once flowed across the Gezira between the Blue Nile and the White Nile.

11.4.2 Formation of Source-Bordering Dunes Three conditions are necessary for source-bordering dunes to form (Williams, 2014, pp. 120–121). As the name implies, source-bordering dunes are dunes formed in close proximity to a well-defined sediment source, which can be a sandy beach next to a lake or a sandy river channel. First, for the dune to develop over time, a regular replenishment of the supply of sand is required, whether from point-bars or mid-channel bars exposed during the dry season. Second, the dry season winds need to have sufficient velocity and regularity to shift the channel sands downwind. Weak or irregular winds will not produce dunes. Third, to allow free downwind movement of sand from the exposed river channel, there needs to be a lack of riparian vegetation. Source-bordering dunes do not therefore denote extreme aridity but do imply a strongly seasonal river flow regime in a semi-arid environment with minimal tree cover in the vicinity of the river channel. An active series of source-bordering dunes can be seen today opposite Hasaheisa (Fig. 11.1) in the lower Blue Nile Valley along the right bank of the river (Williams et al., 1982, p. 127).

11.4.3 Age of the Palaeochannel Source-Bordering Dunes Williams et al. (2015a) obtained a series of OSL ages for the source-bordering dunes at the distal end of former Blue Nile channels east of the lower White Nile. The ages fall into three clusters, indicating that these dunes were active at 115–105 ka, 60 ka and 12–7 ka, which were all times of extreme Blue Nile floods. They may have been active at other times too, but we so far lack such evidence. Many of the dunes in the northwest Gezira have now become so severely depleted of their former stabilising cover of Panicum turgidum tussock grass and of Salvadora persica, Capparis decidua and Acacia seyal trees that they have been reactivated within the upper metre or more of dune sand. All this has taken place within the past few decades.

160

Gezira Alluvial Fan and Blue Nile Palaeochannels

11.5 Prehistoric Occupation Sites 11.5.1 The Singa and Abu Hugar Middle Stone Age Sites In 1924 Mr W. R. G. Bond, governor of what was then Fung province, found a fossilised human skull in the bank of the Blue Nile at Singa below the house of the District Commissioner. The skull came from a layer of fossil-bearing limestone conglomerate which also cropped out in the river bank at Abu Hugar village 20 miles (32 km) upstream of Singa, where the overall bank sequence was better exposed than at Singa (Arkell, 1949a). The base of the section consisted of 2 m of greenish clay overlain by 1. 5–2.3 m of carbonate nodule gravel with fossils, artefacts and lumps of red ochre (Arkell, 1949a, p. 47). Lacaille (1951) considered that the artefacts were Middle Stone Age tools characterised by an advanced Levalloisian facetted platform technique. The skull of an extinct buffalo was recovered in 1932 from a bed of rolled carbonate gravels at the base of the riverbank below Arbatashar village, slightly downstream from Singa. Bate (1951) examined the almost complete skull of this extinct long-horned African buffalo, which she found to be quite distinct from the recent African Syncerus, and classified it as Homoiceros singae. The bank sediments were unconformably overlain by several metres of Gezira clay at both Singa and Abu Hugar. Both she and Arkell considered that the fossiliferous limestone conglomerate at Singa and Abu Hugar was laid down before the humid interval during which the Gezira clay accumulated. There are no available direct ages for the Singa skull or the extinct buffalo skull. Since they occur within a fluvially transported carbonate gravel, the fossils and the MSA stone tools may not be the same age. An age within the range 50–300 ka seems plausible, but very little useful information can be gleaned about the former environment(s) and prevalent climate(s) in this area until future workers have dated and analysed these sediments. On 13 January 1976, the author studied several bank sections at and south of Singa. A 9.5 m high bank section at Singa revealed five distinct clay units with a total thickness of 7.5 m overlying a 120-cm thick bed (or beds) of very fine silty sand partly obscured by fallen clods of clay. The basal 40 cm of the section consisted of rolled carbonate gravel forming a low indurated bench. Between the upper two clay units was a 1.3 m bed of pale brownish yellow fluvial sand with aquatic shells. The lowest clay unit had an irregular upper and lower boundary indicative of pre- and post-depositional erosion. The fossilbearing carbonate bench at Branko on the Blue Nile right bank (Williams et al., 2015a) may be equivalent to the Singa carbonate bench. If this supposition proves to be correct, it would indicate an age of about 100 ka for the Singa carbonate bench at the base of the bank section.

11.5.2 Holocene Sand Ridges and Prehistoric Occupation Sites in the Gezira The Gezira plain is dotted with low sandy knolls and dunes, many of which were occupied from Mesolithic times onwards. Very few have been systematically

11.5 Prehistoric Occupation Sites

161

excavated and many show considerable signs of disturbance from later occupation, including up to the present day. What does seem clear from the excavations at Shabona just east of the lower White Nile (see Chapter 8) is that during the time of Mesolithic occupation 9–8 ka ago, the wet season was longer, the river flood levels were higher and the fauna and flora were consistent with the presence of an acaciatall grass savanna where there is now a semi-desert biome of xerophytic grasses and sporadic acacias in areas not cultivated for rain-fed sorghum and millet by the local peasant farmers.

11.5.3 The Early Khartoum Archaeological Site and Its Significance A. J. Arkell (1949b) directed a major excavation of a Mesolithic (‘Early Khartoum’) site located on a low sand mound adjoining the Blue Nile northeast of Khartoum central railway station. He later excavated a Neolithic site on a 200 m long gravel ridge near Shaheinab on the west bank of the Nile 30 miles (48 km) north of Omdurman and an eroded ‘Early Khartoum’ Mesolithic site 1000 m due west of the Shaheinab Neolithic site (Arkell, 1953). Both volumes need to be read together, because in the later volume he reinterprets some of the evidence from the Early Khartoum site and adds important new data on Nile flood levels. The fauna at the Early Khartoum site included a number of species that live today in swampy environments and thick vegetation, such as the Nile Lechwe (Limnotragus), a water mongoose and extant reed rats (Thryonomys) as well as the extinct reed rat Thryonomys arkelli. In addition, the land snail Limicolaria flammata was common at the site together with the seeds of Celtis integrifolia, both of which are only found south of Sennar today ‘indicating an average annual rainfall at Khartoum . . . at least equal to that of Sennar at the present day (461 mm), or three times as heavy as the present Khartoum average’ (Arkell, 1949b, p. 3). Fish remains were abundant in the occupation layer, including catfish (Clares sp.), Synodontis sp. and Nile perch (Lates cf. niloticus), as were antelope bones, indicating plentiful grazing nearby. There were no domestic animals. These Mesolithic people lived by fishing, hunting and gathering wild plant foods. Arkell (1949b, p. 109) concluded that the Blue Nile flood level was at least 4 m above that of today. He later revised this estimate upwards on the grounds that ‘it was not only possible but likely that it was a low river hunting and fishing camp, which meant that high river must have been at least 5.4 m, and probably 6 or 7 metres, higher still’ (Arkell, 1953, p. 8). The eroded Mesolithic site west of the Shaheinab Neolithic site was located 10 m above present high flood level of the Nile. Arkell (1949b, p. 29) initially considered that the abundant Ampullaria wernei (i.e., Pila wernei) shells at the Early Khartoum site were used as bait for fishing, but later considered it more likely that they were boiled in pots and eaten (Arkell, 1953, pp. 97–98), a conclusion similar to that arrived at by Clark (1989) from his 1973 work at the Shabona Mesolithic site in the lower White Nile Valley (see Chapter 8).

162

Gezira Alluvial Fan and Blue Nile Palaeochannels

11.5.4 The Jebel Moya Late Holocene Burial Sites During four field seasons from 1911 to 1914, Sir Henry Wellcome directed excavations in the valley on the northeast flank of the Jebel Moya massif. His primary goal was philanthropic rather than scientific in that he aimed to employ as many diggers as possible. By April 1914 some 4,000 men were employed and 1,700 graves were exhumed in that fourth and final season (Addison, 1949). In all, 2,883 supposed graves were recorded, 2,792 were cleared and 3,137 individuals were exhumed, together with pottery, stone tools, beads, amulets and figurines (Mukkerjee et al., 1955). Brass and Schwenniger (2013) indicate 3,135 human burials from 2,791 graves, making Jebel Moya ‘the largest cemetery yet excavated in Northeast Africa’ (Brass and Schwenniger, 2013, p. 455). Wellcome died in 1936 and the results were never published until Addison agreed, somewhat reluctantly, to take on the task, having never taken part in the excavations (Addison, 1949). He identified four stratigraphic units (A, B, C and D), with A (the dark brown topsoil) the youngest. He noted that the soil texture was the same in all four layers so that it was hard to separate C from D (Addison, 1949, p. 15). Based on his analysis of the pottery, Addison (1949) initially proposed that Jebel Moya was occupied between roughly 750 BC and 400 BC (Addison, 1949, p. 214) but later he had ‘second thoughts’ (Addison, 1956) and changed his chronology to be roughly contemporary with the main Meroitic (c. 300 BC to c. AD 350), a conclusion already in part implicit in his earlier work (Addison, 1949, p. 215, Fig. 113). His apparent volte-face proved disconcerting for later workers engaged on a reanalysis of the pottery, and prompted Brass and Schwenniger (2013) to obtain six OSL ages on fragments from two of the pottery assemblages they identified. The OSL ages were 3.4–3.2 ka for pottery assemblage 2 and 1.7–1.5 ka for pottery assemblage 3 (Brass and Schwenniger, 2013, Table 2). These are the first reliable ages obtained so far for the Jebel Moya occupation, which is now reinterpreted as a burial complex on the southern periphery of the late Meroitic state (Brass and Schwenniger, 2013; Brass, 2014, 2015), although this does not preclude earlier occupation of the massif. During a very short visit to Jebel Moya in February 1973, graduate students Andy Smith and Ken Williamson, under the direction of J. Desmond Clark, assisted by the author, excavated and sampled two shallow trenches in the valley on the northeast flank of the main massif. Clark (1973a, 1973b), who had spent the previous month excavating at Jebel Tomat (Clark and Stemler, 1975), 15 km east of the White Nile, concluded that there were strong cultural similarities between the pottery and artefacts recovered from Jebel Moya, Jebel Tomat and the Butana Industry described by Shiner and his colleagues (Shiner, 1971). The available evidence from Jebel Tomat is consistent with a higher White Nile flood level, swampy conditions close to the Jebel, a longer wet season and a more humid climate than at present towards 3.5–1.5 ka (see Chapter 8). Jebel Moya is only 60 km ESE from Jebel Tomat. It is therefore probable that the cattle herders around Jebel Moya (Brass, 2015) also benefited from a somewhat wetter climate at that time.

11.6 Conclusion

163

11.6 Conclusion The Blue Nile Basin has an area of roughly 330,000 km2. After having carved a gorge up to 30 km wide and 1,500 m deep in its upper reaches downstream of Lake Tana in the Ethiopian Highlands, the Blue Nile emerges from its confined upland reaches and debouches onto the lowlands of the Sudan to form a vast (>25,000 km2) low-angle alluvial fan known as the Gezira. This was built up during the course of the Cenozoic by Blue Nile distributary channels and episodic overbank flooding. We do not yet know when the Gezira sediments first accumulated but at least 180 m of fluvial gravel, sands, silts and clays were deposited during the past few million years and more. Borehole data indicate a succession of fining-upwards alluvial sequences indicative of an alternation between a fluvial regime characterised by bed-load channels carrying and depositing sands and gravels across the Gezira and a regime characterised by widespread overbank flooding from suspension-load channels and extensive deposition of alluvial clays. Former distributary channels carrying a traction load of sand and gravel radiate across the Gezira fan but are inactive today. They were deprived of flow once the Blue Nile began to incise into its former floodplain early in the Holocene, leading to the abandonment of once active distributary channels. The Gezira has undergone a repeated alternation between two main modes of deposition. During cold drier glacial intervals when the plant cover was sparse in the Ethiopian headwaters the Blue Nile became a bed-load stream, highly seasonal, with braided or linear distributary channels that carry sand and gravel. During warmer wetter interglacial intervals, the Blue Nile became a suspension load channel, ferrying silt and clay across its flood plain. Optical and radiocarbon ages obtained so far show that the channels on the surface and near surface of the Gezira were active at 270 ± 30 ka, 190 ± 20 ka, 105 ± 6 ka, 100–70 ka and 50 ka to 8–5 ka. Source-bordering dunes derived from the Blue Nile distributary channel sands were actively accumulating at 115–105 ka, 60 ka and 12–7 ka during a time of seasonal river flow and semi-arid climate. Thick overbank clays were deposited across the Gezira between 50 ka and 40 ka and between 14.5 ka and 5 ka, during intervals when the prevailing climate was wetter than present. The Mesolithic ‘Early Khartoum’ site on the bank of the Blue Nile near the present Blue and White Nile confluence was occupied during a time (c. 9 ka) when the rainfall was perhaps three times greater than now, the high Blue Nile flood level possibly 10 m above present, and the adjacent land wet and marshy and in terms of its fauna reminiscent of South Sudan today.

12 The Atbara

On the night of June 23 the flood came roaring down, and the travellers, waking in the morning, beheld a stream 500 yards [ca. 500 m] wide and fifteen to twenty feet [5–7 m] deep. Floating islands of broken bamboos and other debris were being carried along in the muddy current. Alan Moorehead, The Blue Nile (1972, p. 231) (Onset of the Atbara flood season witnessed by Sir Samuel Baker in 1861)

12.1 Introduction The Atbara Basin covers an area of 180,000 km2, more than half of which is located in the Ethiopian uplands. The upper Atbara rises 30 km west of the Ethiopian town of Gondar and 50 km north of Lake Tana in the Northwest Highlands of Ethiopia (Fig. 12.1). Its source is close to the headwaters of the Blue Nile and its two main tributaries the Rahad and the Dinder. The close proximity of the Atbara and Blue Nile headwaters implies that their response to any late Cenozoic changes in temperature, precipitation, plant cover, runoff and erosion would have been very similar. The Tekezze is the major tributary of the Atbara and provides much of the water and sediment to that river (Garzanti et al., 2015). The headwaters of the Tekezze flow east from the mountainous divide that separates them from the westward-flowing Atbara (Fig. 12.1). The Tekezze then flows north and finally west in a large anticlockwise bend around the great volcanic massif of the Semien Mountains, with its summit peak Ras Dashan attaining an elevation of 4,543 m (Werdecker, 1967), although some maps suggest an elevation of 4,620 m. In this sector, it flows through a deep and sometimes narrow gorge, the sides of which are often highly unstable and prone to landslides and rock-falls (F. M. Williams, 2016, pers. commun.). A number of deeply entrenched rivers, including the Mai Shaha, flow south and east from the Semien Highlands to join the upper Tekezze (Fig. 12.2). After emerging from its gorge onto the clay plains of eastern Sudan the Tekezze becomes known as the Setit and finally joins the Atbara 50 km northwest of the Sudanese town of Gedaref and just south of the Khashm el Girba dam on the Atbara River near the town of that name. From there the river flows northwest to join the main

164

12.1 Introduction

165

Figure 12.1 The Atbara Basin. Shaded area above 2000 m.

Nile at the town of Atbara, after a total journey of just over 800 km. The dam was built to provide water for perennial irrigation to assist the Halfawis and other Sudanese communities dispossessed by the flooding of their land in northern Sudan brought about by completion of the Aswan High Dam in 1970 and the creation of Lake Nubia/ Lake Nasser. Until then the Atbara had always been a highly seasonal river that dried out into a series of deep pools for about half the year. In this chapter, we describe the cold climate landforms in the Semien Highlands and use the combined evidence from glacial and periglacial deposits to reconstruct late Quaternary environments in the upper Tekezze catchment. We then review the various estimates of long-term and short-term denudation rates in the Atbara–Tekezze drainage basin and the light they may shed upon Cenozoic tectonic uplift in this region as well as upon possible recent human impact on erosion. We conclude with a survey of Quaternary alluvial formations in the lower Atbara and their associated fossils and prehistoric artefact assemblages.

166

The Atbara

12.2 Cold Climate Landforms and Glaciation in the Semien Highlands A number of major tributaries of the Tekezze–Atbara rise in the Semien Mountains of Ethiopia. It is therefore appropriate to consider the geologically recent climatic history of that spectacular massif in some detail. The Semien Mountains consist of an eroded Oligocene shield volcano, for which Kieffer et al. (2004, Fig. 3, p. 797) have obtained ages between 29.9 and 18.7 Ma, superimposed on horizontal Cenozoic Trap Series lavas and underlying Phanerozoic sedimentary rocks and Precambrian Basement rocks that form the Ethiopian Plateau. That the Semien Mountains were glaciated during the late Quaternary is not in dispute. What is in question is the areal extent of past cold climate processes and landforms. Geomorphic evidence for former glaciation may be considered under two headings: erosional and depositional (see Mahaney, 1990 and Williams, 2014, chapter 13 for comprehensive reviews). The erosional landforms include U-shaped valleys, truncated spurs, cirques, roches moutonnées (ice-scoured and ice-plucked bedrock), glacial striations and rock-basin lakes. The depositional landforms include glacial erratics, lateral and terminal moraines, boulder clay or till, and glacial outwash deposits. Using such evidence, Osmaston et al. (2005) concluded that about 180 km2 of the Bale Mountains in the Southeast Highlands of Ethiopia had been glaciated during the late Quaternary, most probably during the Last Glacial Maximum (LGM) culminating about 20,000 years ago. Hurni (1982, chapter 5) used similar evidence to construct a detailed map of glacial cirques and associated glacial moraines in the Semien Highlands, and concluded that 20 catchments had been glaciated covering a total area of only13 km2. Although this estimate seems very small when compared to the 180 km2 considered to have been glaciated in the Bale Mountains, we need to bear in mind that the Northwest Highlands of Ethiopia receive somewhat less precipitation today than the Southeast Highlands (Griffiths, 1972c, Fig. 3; Schaller and Kuls, 1972, map 2; Hurni, 1982; Conway, 2000) and that this trend may have been even more pronounced during the LGM when lakes were low or dry across Ethiopia and the summer monsoon was relatively ineffective, its influence probably not then extending as far to the northwest as it does today. Accordingly, Hurni’s (1982) estimate seems well founded and is supported by the author’s own observations during visits in December 1971, April 1975 and November–December 2009. Certain earlier workers seem to have overestimated the extent of glaciation in this area. For example, Nilsson (1940) considered that the Pleistocene snowline lay between 3,500 and 4,100 m while Hövermann (1954) suggested that it fell as low as 3,000 m. More reasonable estimates are those of Minucci (1938) who proposed a snowline of 4,100–4,300 m and Hastenrath (1974), who argued for a late Pleistocene snowline of 4,200–4,300 m, based on mapped cirque floor levels rather than questionable deposits of ‘moraine’ at low elevations. Seven 36Cl exposure ages have now been obtained for glacial moraines in two cirques in the Semien Highlands (Williams et al., 2015a, Table 3). Moraines associated with the former cirque glacier on the north face of Mount Bwahit (13°15.6 0 N, 38°11.5 0 E; elevation 4,430 m) (Fig. 12.2) have 36Cl ages of 68.3 ± 3.5 ka, 37.7 ± 1.4 ka and 15.3 ± 0.7 ka.

12.2 Cold Climate Landforms in Semien Highlands

Cheru 13°15´N

Tefew Leser a

4449m

R.

Gabriko

Gabriko R. A

Ras Dashan 4543m

Ma

i Sh aha

Bwahit 4430m

Leb

167

Mesarerya 4353m

Abbat Dashan

N 0

5

13°10´N

km 38°15´E

Land > 4000m

38°20´E

Land > 3000m

Cirque

38°25´E

A

Angular rubble sectioin. See Fig.12.4

Figure 12.2 Geomorphic map of the Ras Dashan summit region, Semien Highlands, Ethiopia. (After Werdecker, 1966; Hastenrath, 1974; Williams et al., 1978; Hurni, 1982; field observations in 1971, 1974, 2009; and air photo interpretation.) The location of the two glaciated summits where the moraines were sampled for 36Cl exposure dating is shown as B (Mount Bwahit) and M (Mount Mesarerya).

The cirque moraines on the northwest face of Mount Mesarerya (13°12.8 0 N, 38°12.8 0 E; elevation 4,353 m) (Fig. 12.2) have 36Cl ages of 46.2 ± 2.1 ka, 36.4 ± 1.6 ka, 27.4 ± 1.0 ka and 18.1 ± 0.9 ka. These ages can be interpreted in several ways. They may indicate one or more periods of glaciation between about 70 ka and 15 ka. Alternatively, as Williams et al. (2015a) surmise, some of the older ages may represent inheritance from older erosional surfaces owing to shallow glacial erosion within the small cirques. Until more ages become available it is hard to choose between these possibilities. However, we can safely conclude that ice was present on Mount Mesarerya 18.1 ka ago and on Mount Bwahit 15.3 ka ago. These ages are maximum ages for the onset of ice retreat in the Semien Highlands.An important issue is whether the glacial climate in the Ethiopian Highlands was wet or dry. Nilsson (1931, 1935, 1940, 1949), Büdel (1954) and Scott (1958) all considered that the times of maximum glaciation were times of wetter than present climate – the ‘glacialpluvial’ hypothesis (see Williams, 2014, chapter 12 for a review). By this account, episodes of high lake levels in tropical Africa were coeval with glacial intervals in Europe and North America. However, with the advent of radiocarbon dating it soon became apparent that the

168

The Atbara

lakes in the Ethiopian, East African and Afar Rifts were either dry or very low during the Last Glacial Maximum (LGM), with lake levels rising rapidly during the warm wet Holocene postglacial (Butzer et al., 1972; Gasse, 1975; Williams et al., 1977; Street, 1979a, 1979b). It is now apparent that precipitation was lower in the Semien Highlands during the LGM (Williams et al., 1978; Hurni, 1982). At present precipitation in the Semien Highlands comes from moist summer monsoon air masses for elevations up to 3,000–3,200 m. Above 3,200 m the dominant winds are the NE Trade winds (Hurni, 1982, chapter 3). Water vapour from the underlying SW moist air masses ascends into the trade winds as high as 6,000 m in elevation, so that the NE winds also bring rain to the mountains. For glaciers to form, snow needs to persist throughout the year. If the summer temperatures are too hot, the snow will melt. We can therefore expect that temperatures were lower when the glaciers were present, but from glacial evidence alone it is hard to say by how much the temperatures were depressed. This is because for snow to accumulate two things are necessary: enough precipitation in the form of snow and lower temperatures so that precipitation occurs as snow. What is needed is an independent means of estimating temperature during times of peak glaciation. Periglacial landforms are cold climate landforms that reflect the influence of temperature rather than that of precipitation because they owe their origin to freeze–thaw processes (Davies, 1969; Washburn, 1973, 1979). A number of freeze–thaw features occur above 3,700 m elevation in the Semien Mountains, including stone-banked terraces, stone stripes and polygons, fine-earth polygons, recently frost-riven boulders and fields of unstable, angular basalt blocks (Hastenrath, 1974; Williams et al., 1978). There is no evidence of any present-day bedrock frost shattering (gelifraction) or of movement of the resulting angular debris below an elevation of about 4,250–4,300 m, which is the present-day upper limit of tussock grass (Büdel, 1954; Hastenrath, 1974; Williams et al., 1978; Hurni, 1982). Periglacial solifluction mantles are widespread deposits that are characteristic of former seasonal freeze–thaw processes beyond the limits of recently glaciated areas (Galloway, 1965b, 1970; Davies, 1969; Washburn, 1973, 1979; West, 1977; Williams, 2014, chapter 13). The lower limit of periglacial solifluction coincides very roughly with the upper limit of the tree-line, which in turn coincides approximately with the 10°C isotherm for the warmest month (Galloway, 1965b). Williams et al. (1978) examined a number of exposures of now stable angular rubble in two tributary valleys of the Mai Shaha (Fig. 12.2). To be reasonably certain that these slope mantles were indeed of periglacial origin rather than simply colluvial mantles, they noted that the angular rubble deposits had the following characteristics: (a) The rubble consisted of angular to subangular, very poorly sorted and unweathered basalt blocks up to 0.4 m long, set in a matrix of 1.5 mm platy fragments and granules. (In contrast to fluvial sediments, which are clast-supported, these mass-flow deposits are matrix-supported (Fig. 12.3). (b) Matrix texture ranged from gritty loam to clayey sand, with hues lighter than the overlying dark surface clays and loams.

12.2 Cold Climate Landforms in Semien Highlands (a)

169

(b)

Figure 12.3 (a) A clast-supported fluvial deposit and (b) a matrix-supported periglacial rubble deposit.

(c) The rubble deposits became thicker and more widespread with increasing elevation. (d) The long axes of the larger fragments were aligned roughly downslope in some sections. (e) Some sections revealed an upward-coarsening sequence, with fine particles near the base and coarser particles towards the top. (f) The angular rubble deposits were often several metres thick on bedrock slopes of 10°– 15°, in contrast to presently active scree and angular colluvium which was rare at these elevations except on slopes steeper than 20°–30°. (g) The rubble often formed part of a fill beneath gently sloping colluvial-alluvial benches that were now being dissected by small tributaries of the Mai Shaha. Figure 12.4 illustrates one of the angular rubble sections studied in the field. In this section, the rubble consists of two beds, each about 3 m thick, separated by up to 3 m of brown clayey coarse sand with very few angular rock fragments. The lower rubble unit (a) has a strongly indurated and iron stained matrix and may be much older than the upper rubble unit (d), which has a porous earthy fabric. In each section studied, a surface layer of dark grey-brown clay or loam with a moderate to strong polyhedral structure overlies the angular rubble. Stones are rare in this surface layer. The observed lower limit of angular basalt rubble was about 3,000–3,100 m in both the Cheru Leba and Gabriko River Valleys, or roughly 1,200 m below the late Pleistocene snowline at 4,200–4,300 m inferred by Hastenrath (1974). The upper limit was less well defined. In the col north of the upper Cheru Leba it was seen at 3,400–3,600 m. Temperatures of 0°C or less would have been necessary in order to shatter bedrock prior to the movement of these angular boulders downslope by periglacial solifluction to rest at elevations between 3,100 m and 3,600 m. Frost shattering is now active only above ca. 4,250–4,300 m. If we accept a mean temperature lapse rate of 0.6°C/100 m, which is a reasonable rate for high

170

The Atbara

grassy surface e

d

1m c b

a gully floor

Figure 12.4 Angular rubble in a gully section near Gabriko village in the Mai Shaha valley, Semien Highlands, Ethiopia. a, d, angular basalt blocks in a matrix of clayey sand; b, a gritty brown sand with occasional basalt blocks; c, e, a dark loam or stone-free clayey sand. (Modified and simplified from Williams et al., 1978.)

mountains in East Africa and Ethiopia (Fantoli, 1966; Brown and Cochemé, 1973; Hurni, 1982), we can infer a possible temperature lowering of between 4°C and 8°C (Williams et al., 1978). Hurni (1982, chapter 5) arrived at a value of 7°C cooler during the late Pleistocene cold period in the Semien Mountains using independent evidence of mapped periglacial deposits. In November 2009, the author and colleagues carried out a reconnaissance survey near Mount Bwahit in the Semien Mountains and examined a section (Fig. 12.5) at an elevation of 3,590 m at 13°15.6070 N, 38°11.5340 E. The sequence of events identified was as follows:Hurni (1982, chapter 6) identified a broadly similar sequence and considered on the basis of a few radiocarbon ages that the dark silty clay unit was Holocene in age and that the recent incision was anthropogenic, most likely triggered by cultivation on steep, recently cleared slopes. (a) Bedrock erosion and creation of at least one rock-cut terrace. (b) Glaciation of the upper slopes above about 3,800 m. (c) Reworking of possible glacial moraine by a debris flow at the distal end of an alluvial fan. The basalt blocks in the poorly sorted fanglomerate were up to one metre in size and matrix-supported.

12.3 Denudation Rates in the Tekezze Basin

Debris flow Organic loam

171

Bedrock LT Lower terrace UT Upper terrace

UT UT 2m

LT 20m

Figure 12.5 Episodic erosion and deposition revealed in a section at 3,600 m elevation on the lower slopes of Mount Bwahit, Ethiopia.

(d) Incision by local streams by at least 3 m. (e) Aggradation and widespread deposition of silt and clay from suspension load streams. (f) Renewed incision to a depth of at least 1.5 m to form an inset terrace of black (10YR2/ 1) silty clay loam. This incision was probably caused by deforestation during the last hundred years. Hurni (1982, chapter 6) identified a broadly similar sequence and considered on the basis of a few radiocarbon ages that the dark silty clay unit was Holocene in age and that the recent incision was anthropogenic, most likely triggered by cultivation on steep, recently cleared slopes.

12.3 Denudation Rates in the Tekezze Basin Several different methods have been used to estimate denudation rates in the Atbara and Tekezze catchments. By far the simplest technique is to use contour maps of the dissected upland plateaux in northwest Ethiopia to estimate the volume of rock eroded above a given elevation from a catchment of given area. Provided the age of the uppermost lava flows is known, it is a simple matter to calculate the mean denudation rate per unit area. This was the approach used by McDougall et al. (1975), who estimated that roughly 31,000 km3 of rock had been eroded from the upper Tekezze Basin since its inception about 30 Ma ago, amounting to a mean rate of denudation of 12 m3 per km/yr, which is equivalent to a mean annual rate of surface lowering of 0.012 mm/yr. They obtained a similar value for the upper Blue Nile headwaters region. More recently Puchol et al. (2016) have used the cosmogenic 3He content measured in detrital pyroxene found in river sands to obtain basin-averaged denudation rates of 0.057 ± 0.011 mm/yr for the Tekezze valley. They obtained comparable rates of 0.07 ± 0.02 mm/yr for the Mille River, which flows east from the Ethiopian escarpment into the western Afar

172

The Atbara

Rift. However, it should be noted that these results are based on a single 2 kg sand sample from each of the two rivers. Garzanti et al. (2006, 2015) and Padoan et al. (2011) analysed the petrography and the strontium and neodymium isotopic composition of sediment samples collected along the entire length of the Nile and its tributaries and were able to discriminate between samples transported as bed load as opposed to samples entrained as suspension load. For the Main Nile prior to construction of the High Dam the total sediment load amounted to 230 ± 20 Mt/ yr. The Atbara had a suspension load of 74 ± 9 Mt/yr and a bed load of 7 ± 2 Mt/yr. Corresponding figures for the Blue Nile were a suspended load of 127 ± 15 Mt/yr and a bed load of 12 ± 2 Mt/yr. This means that the Atbara contributes 36 ± 4% of the annual sediment load of the Nile and the Blue Nile a further 60 ± 4%, with the balance of 3 ± 2% provided by the White Nile and local inputs from wind-blown sand in northern Sudan and Egypt. Allowing for slight differences in bedrock density, Garzanti et al. (2015) calculated that the current denudation rates in both the Atbara mountain catchments and the upper Blue Nile basin amounted to 0.29 ± 0.04 mm/yr. This rate is an order of magnitude (20–40 times) faster than the long-term denudation rates obtained by McDougall et al. (1975) and about five times faster than the rate obtained by Puchol et al. (2016). It is tempting to use these data to argue that the rates of tectonic uplift and associated fluvial incision have been accelerating, but a more plausible explanation for this increase is accelerated erosion in the mountainous catchments caused by clearing of the natural vegetation cover and cultivation on steep bare slopes, as documented by Hurni (1999) and Nyssen et al. (2004).

12.4 Quaternary Alluvial Formations in the Atbara Valley Abbate et al. (2010) have described in exemplary detail a sequence of alluvial deposits up to 50 m thick along a 20-km stretch of the middle Atbara between Khashm el Girba and Halfa al Jadida. They identified two major alluvial formations or synthems (a synthem is a major stratigraphic unit that is unconformity bounded). The lower formation they termed the Butana Bridge Synthem (BBS). The BBS is up to 10 m thick and consists of braided stream gravels at the base and high-sinuosity stream sands at the top. A band of massive calcrete separates these two fluvial units. Abundant vertebrate fossils indicative of an arid savanna environment are found within the BBS, together with Acheulian stone tools. An approximate age of late Early Pleistocene to early Middle Pleistocene was assigned to the BBS. The upper formation or Khashm el Girba Synthem (KGS) is separated from the underlying BBS by a major erosional unconformity that may represent a time gap of several hundred thousand years. The KGS is up to 40 m thick and consists of three stratigraphic units: KGS1, KGS2 and KGS3. Evolved Acheulian and Middle Stone Age artefacts are abundant in the KGS. The fossil vertebrate fauna is more diverse than that in the BBS and indicates a grassland savanna with scattered pools of water or small lakes. The KGS fluvial deposits range from sinuous sandy channels at the base (KGS1) through braided river

12.5 Holocene Environments

173

deposits grading up into sandy meandering streams (KGS2) to pebbly sands of braided streams and sheet flow silts (KGS3). Deposits of Nile oysters (Etheria elliptica) overlain by stromatolite coatings at the base of KGS1 and KGS2 have uranium/thorium ages of 126.1 ± 1 ka and 99.2 ± 0.7 ka, respectively. These ages indicate a last interglacial age, when the global climate was generally wetter and warmer than it is today. Both of these intervals were also times of very high Nile water and sediment discharge into the Mediterranean resulting in the formation of sapropel deposits under anoxic conditions on the floor of the sea (Williams et al., 2015a), which we discuss later in Chapter 21. Abbate et al. (2010) attribute the fluctuations between alluvial sedimentation and channel incision to fluctuations in the level of a hypothetical Atbara lake. Apart from one brief mention of a possible delta feature, no evidence is provided in support of this hypothesis, no lake strandlines are identified, and no dam for the putative lake is suggested. Accordingly, it seems wiser to seek other explanations for changes in river behaviour. One obvious possibility relates to changes in the load to discharge ratio and in the type of sediment carried associated with climatic changes in the headwaters and concomitant changes in plant cover, weathering, soil formation, erosion and hill slope stability (see Chapter 6). Another possible explanation is the type of complex response demonstrated experimentally by Schumm and Parker (1973) in which the incision of a small channel under flume conditions led to the formation of a multiple set of river terraces. As these authors noted, ‘initial channel incision and terrace formation were followed by deposition of an alluvial fill, braiding and lateral erosion, and then, as the drainage system achieved stability, renewed incision followed by a low alluvial terrace’ (Schumm and Parker, 1973, p. 99). Chialvo (1975) carried out a brief survey of the geology in the vicinity of the Atbara–Setit confluence. He considered that the dark alluvial clays overlying the youngest Atbara terrace were a result of Pleistocene flooding by the river, followed by Holocene incision. He then went on to argue that most of the dark clays away from the river were locally derived sheet flood sediments: ‘Mais il est évident que la majeure partie dérive du lessivage et de l’érosion des massifs proches, surtout des pointements basaltiques nombreux dans la région’ (Chialvo, 1975, p. 67) (‘But it is clear that most [of this material] comes from sediment winnowing and erosion of nearby hills and in particular the basalt plugs common in this area.’) However, he also noted that the Atbara has shifted course many times and that buried channels and high level abandoned meander channels are common along the upper Atbara in eastern Sudan after its emergence from the Ethiopian uplands. It is therefore quite possible that some at least of the high-level clays in the middle Atbara valley are of fluvial origin, and that they were laid down before the Atbara had fully cut down into its gorge in Ethiopia.

12.5 Holocene Environments The most widespread sedimentary unit in the middle and lower Atbara valley is a dark cracking clay or vertisol (Arkell, 1949a; Chialvo, 1975; Abbate et al., 2010). It is tempting

174

The Atbara

to equate it to the Gezira clay (see Chapter 11) but we need to remember that the Gezira clay is diachronous, and has yielded ages of 55–50 ka as well as more recent ages of 14–5 ka. The presence of Mesolithic, Neolithic and younger prehistoric sites lying on and within the uppermost dark clays in the Atbara valley suggests that some of these clays may indeed be Holocene in age. By analogy with the Holocene clays in the Gezira, it is likely that these clays were laid down during one or more phases when the climate was somewhat wetter than it is today. Nevertheless, this assumption, however plausible, still needs to be verified (or refuted) by future fieldwork involving direct dating of the clays and analysis of any fossils within them, especially snails of known ecological association such as the acacia-tall grass savanna land snail Limicolaria flammata.

12.6 Quaternary Fossils and Prehistoric Artefacts Arkell (1949a) carried out a rapid reconnaissance of prehistoric sites in the Atbara valley and identified a number of Early Stone Ages sites located close to the present river and its confluence with the Nile. The artefacts from these localities range from Oldowan and Acheulian to Evolved Acheulian, consistent with the later discoveries by Abbate et al. (2010) in the Middle Atbara Valley, except that they also found Middle Stone Age artefacts in alluvial sediments dating back to last interglacial times (Marine Isotope Stage 5). Arkell (1949a) also identified a few Late Stone Age sites, which he termed Khartoum Mesolithic. Later workers have excavated and in some cases dated both Mesolithic and Neolithic sites in the Atbara valley as well as in the Atbai region to the north and the Butana plains to the south (Marks, 1989, 1993; Marks and Fattovich, 1989). The faunal remains recovered from these Holocene sites point to hunting of small and large savanna grassland and local swamp or riverine animals as well as fishing for catfish, tilapia and Nile perch. The stone tool assemblages and pottery styles do not show much resemblance to those from the Nile valley but appear to reflect more easterly influences. There is some suggestion of Neolithic domestication of millet and sorghum as well as possibly earlier herding of domestic sheep, goats and cattle. The chronology so far available for many of these sites is still too meagre to allow any insights into possible responses to Holocene environmental changes to be adequately tested (Hassan, 1986a; Marks, 1989, 1993). Brandt (1986) makes a similar point in his thoughtful review of the Late Pleistocene and early Holocene prehistory of Ethiopia, Somalia and Djibouti.

12.7 Conclusion The Atbara Basin has an area of about 180,000 km2 and rises in the Ethiopian Highlands about 50 km north of Lake Tana and flows for some 800 km to reach the main Nile at the town of Atbara. Its main tributary is the Tekezze River, which rises in the Semien Highlands of Ethiopia and has cut a deep gorge from which about 31,000 km3 of rock have been eroded. Before construction of the Khashm el Girba

12.7 Conclusion

175

dam near Kassala to help resettle people displaced by the Aswan High Dam, the Atbara was a highly seasonal river that dried out for over half the year. When in flood it contributed a discharge of 12 ± 5 km3 (14%) to Nile flow and a sediment load of 82 ± 10 km3 (36%) to the main Nile. During the LGM small cirque glaciers were present on the highest mountains, frost shattering was widespread, temperatures in the headwaters region were 4–8°C lower than today, periglacial solifluction was active along the upper catchment and hill slopes were highly unstable. The flow regime of the river was even more seasonal than in historic times, with an abundant supply of coarse debris ferried by the Atbara towards the Nile. Warmer wetter conditions during the Holocene saw widespread deposition of alluvial clays during the summer floods. Mesolithic and Neolithic communities settled in the lower valley of the Atbara, with fishing providing an important part of their diet.

13 Jebel Marra Volcano

The northwest and north are less populated, and the east is almost uninhabited. This is a consequence of the differences in fertility between the individual regions, and this again depends on the adequacy of the water supply, which comes from the numerous rivers and streams of the Marra range [Jebel Marra]. These do not indeed carry water throughout the whole year, but there is a water-holding stratum a few feet below their sandy beds. Gustav Nachtigal (1834–1885), Sahara and Sudan. IV. Wadai and Darfur (1889)

13.1 Introduction The Saharan uplands (see Chapter 3, Fig. 3.6) contain a unique record of past environmental and climatic changes. The evidence for such environmental changes can be quite varied. In the case of the Hoggar Mountains (2,908 m), there is good evidence from alluvial terraces, lake deposits and glacial landforms to show that these uplands have been subject to multiple glaciations during the late Quaternary and have been significantly wetter during the early Holocene (Rognon, 1967, 1976a, 1976b; Conrad, 1969; Messerli et al., 1980). River terraces in the valleys of Tibesti (3,445 m) and deltaic deposits linked to former drainage from Tibesti into the adjacent Chad Basin also indicate an alternation of wetter and drier climatic episodes during the late Quaternary (Ergenzinger, 1968; Hagedorn and Jäkel, 1969; Servant et al., 1969; Maley et al., 1970, Maley, 2004), as do lake sediments within certain of the volcanic craters (Maley, 2000). The Aïr Mountains (2,022 m) have a relict fauna of baboons (Papio anubis) and patas monkeys (Cercopithecus patas) which came originally from the rain forests of West Africa and must have made the journey overland when permanent rivers flowed south from the Aïr in densely forested valleys during one or more wetter phases in the Quaternary (Monod, 1963). The now isolated and arid Adrar Bous (1,123 m) ring-complex located 65 km ENE of the northern Aïr was occupied intermittently from Early Stone Age times onwards, and during the Early Holocene Mesolithic people hunted crocodiles and hippos and fished for turtles and Nile perch in the lakes that once occupied one of the deep southern valleys between the more resistant 176

13.1 Introduction

177

concentric ridges of this mountainous ring complex. Their Neolithic successors were pastoralists with herds of short-horned cattle (Bos brachyceros) who could find adequate grazing until about 5,000 years ago, when desiccation set in (Williams et al., 1987; Williams, 2008). In common with these other Saharan uplands, Jebel Marra (3,042 m) volcanic massif, located in Darfur province, northern Sudan, also contains a varied repertoire of evidence both inside and outside the main caldera that allows us to reconstruct a remarkable series of Quaternary fluctuations in lakes, rivers and dunes in this region located on the watershed between the Nile and Chad Basins. The German explorer Dr Gustav Nachtigal may have been the first European to see Jebel Marra during his visit to Darfur in 1876 and he has left us a valuable account of the ethnography and land use in this region at that time, with some details about the local rivers (Nachtigal, 1889). Most of the world’s big and recently active volcanoes are clustered along the edge of tectonic plates or above subduction zones (Kearey and Vine, 1996). Jebel Marra volcano is unusual in that it lies close to the centre of the African lithospheric plate and is certainly not located over any subduction zone. A circle of radius 1,500 km centred on Jebel Marra would only just graze the Atlantic coast of West Africa and the western coast of the Red Sea (Fig. 13.1, inset). The Mediterranean coast is 2,000 km due north of Jebel Marra. The present-day volcanic massif covers an area of 13,000 km2 and is up to 80 km wide from east to west and 140 km long from north to south. It attains an elevation of 3,042 m and is composed of some 2,000 m of lavas and pyroclastic rocks. Francis et al. (1973) estimated that the original volume of material erupted from Jebel Marra was about 8,000 km3, compared to roughly 3,000 km3 for Tibesti. The volcano is dormant today, with sporadic fumaroles and hot springs (Hammerton, 1966, 1968; Williams et al., 1980). It last erupted nearly 4,000 years ago, when ash from the volcano buried wood from an Olea tree growing near the northern outer rim of the caldera, most likely Olea laperrinei according to Wickens (1976a, p. 8). The wood has a radiocarbon age of 3,520 ± 100 14C years BP (Frances et al., 1973) and a calibrated age of 3,813 ± 128 years BP. Another unusual feature of Jebel Marra is its flora. Dr Gerald Wickens (1976a, 1976b) carried out an exhaustive survey of the present-day vegetation and found that it consists of a mixture of trees, shrubs and grasses from quite different initial sources, including the tropical African savanna region to the south, west and east, the Ethiopian highlands and Red Sea Hills to the east, and the Mediterranean uplands of northwest Africa. The relative lack of savanna species from the vast region to the west of the Chad basin prompted Wickens (1976a, p. 63) to speculate that when Lake Chad was at its maximum extent (‘Mega-Chad Lake’) it acted as a barrier to the movement of savanna plants from west to east. The highest parts of the massif receive on average over 900 mm of rain a year (Wickens, 1976a, Fig. 5), and springs are common along the slopes of the mountain. The rivers flowing to the west from the massif are perennial, while those to the north such as Wadi Howar are ephemeral today, but show evidence of intermittent or even seasonal flow during the Quaternary (see Chapter 14).

178

Jebel Marra Volcano

Finally, fossil oil palm (Elaeis guineensis) and Combretum leaves and other remains of tropical rain forest species have been recovered from alluvial deposits south of the massif. These were at first attributed to a wet climatic phase in the early to mid-Holocene (Wickens, 1975a, 1975b, 1976a), as was the diatomite brought to Khartoum from a then unknown location west of Jebel Marra and analysed by Dr Gisela Prasad (Prasad, 1971). Later work, discussed in Section 13.4, indicates an age closer to a million years for both the fossil leaves and the diatomites. There is very little information about the Miocene and Pliocene environments in the Jebel Marra region, but recent work in the Chad Basin offers some valuable insights. Otero et al. (2011) analysed the oxygen isotopic composition of phosphate in the fossil tooth enamel of the large open water tiger fish (Hydrocynus) from four fossil-bearing Neogene sites between present-day Lake Chad and Tibesti. The sites had 10Be/9Be ages between 7 Ma and 3.6 Ma. The ∂18O values showed that at these four former lake sites there had been four successive wet phases, each one less humid than its predecessor, with an accentuation of aridity during the Messinian (7.2–5.3 Ma) at the end of the Miocene. Later work on a 200-m long sediment core recovered from the southern Chad Basin revealed the presence of a permanent lake (or recurrent lakes) in this area between 6.7 Ma and 2.4 Ma, during which the climate was significantly wetter than today and the vegetation was indicative of a longer wet season (Moussa et al., 2016). Similar environments were most likely present around Jebel Marra during the very late Miocene and Pliocene.

13.2 Geological History of Jebel Marra Jebel Marra is located where two major Cenozoic volcano-tectonic lineaments intersect (Fig. 13.1). One runs SSE from northwest Libya through Jebel Marra towards Mt Elgon via the Aswa wrench fault in South Sudan and Uganda. The other runs ENE from Mt Cameroun through southern Chad, Jebel Marra, the Tagabo and Meidob Hills and the Bayuda volcanic field to the northern Red Sea Hills (Vail, 1972a, 1978). Vail (1978, p. 42) noted that a possible third band of volcanic rocks may run NE from Jebel Marra across the Libyan Desert via Laquiya Arba ‘in to the Wadi Natash volcanic area of the Eastern desert of Egypt. Jebel Marra lies on a foundation of Precambrian Basement Complex rocks that was uplifted during the late Cenozoic, and now sits across the divide between the Nile and Chad drainage basins, at roughly equal distances from the Atlantic, Mediterranean, Red Sea and Indian Ocean (Fig. 13.1). The Precambrian rocks have undergone regional metamorphism in the amphibolite facies and have been folded and faulted to provide great structural complexity. The rocks are highly varied and include foliated granitic gneisses, mica schists, quartzites, amphibolites, and areas of massive or weakly foliated granites (Vail, 1972a, 1972b). A survey of this region by the author and his colleagues during January–February 1976 revealed that amphibolites were more widespread in certain areas than previously recorded (Williams et al., 1980). Certain calcareous clay soils were found to have developed in situ on

13.2 Geological History of Jebel Marra

20°E

179

30°E JM

MEDITERRANEAN SEA

30°N 30°E

D RE

le Ni

Hoggar

Tibesti

20°N

Bayuda Volcanics

Aïr

A SE

Nile

JEBEL MARRA

ite Wh

10°N

Mt Cameroon

Ethiopian Highlands

Aswa wrench fault

0° 0

e Nil

Lake Chad

ue Bl

Meidob Hills

km

1000

Lake Victoria

N

Cenozoic volcanic rocks

International boundary

Fracture zone

Circle of radius 1500 km centred on Jebel Marra (shown on inset map)

Figure 13.1 Location of Jebel Marra in the western Nile Basin in relation to major tectonic lineaments, as inferred by Vail (1972a). (Adapted from Williams et al., 1980, Fig.13.1.)

weathered amphibolites and so were neither alluvial nor colluvial in origin, as earlier workers had thought (FAO, 1969). The Precambrian basement rocks are overlain unconformably by a series of gritstones, sandstones, siltstones and shales. These sedimentary rocks have not been metamorphosed, with the gritstones and sandstones showing fluvial cross bedding. They are generally mapped as Nubian Sandstones of Cretaceous age (Vail, 1972b, 1974, 1978), but some of these sedimentary rocks could be older – in eastern Chad they are mapped as Ordovician

180

Jebel Marra Volcano

and Devonian (Vail, 1972a), and, in some cases, younger. Many of the sedimentary rocks have been eroded so that the Jebel Marra volcanic rocks often directly overlie the Precambrian basement rocks. The basement rocks beneath the volcanic massif have a mean elevation of about 1,100 m, which is 500 m higher than the Basement Complex surface on the plains 100 km away from the massif. Basalt flows occupy deep valleys up to 80 km from the mountain, and may date back to Miocene times some 15 Ma ago, when the volcano is thought to have begun to form (Andrew, 1948; Lebon and Robertson, 1961), although this is guesswork, since none of the flows appear to have yet been dated. The volcanic rocks consist of trachytes, basalts and pyroclastic rocks. Vents, plugs and scoria cones are ubiquitous across the massif. The Deriba caldera is 5 km in diameter and contains two lakes, one deep and brackish, and the other shallow and very saline, which we discuss in Section 13.5. There were at least two main intervals of volcanic activity with mixed lava and ash eruptions, separated by a period of erosion long enough for deep valleys to be carved by rivers flowing away from the mountain. The main explosive phase then ensued, followed by caldera collapse. Episodic pyroclastic eruptions have continued ever since, the most recent one shown by the presence of carbonised wood in a pumice deposit on the side of the caldera (Francis et al., 1973). The wood is exposed in the side of a recent gully to the north of the caldera wall (Wickens, 1976, p. 8) and, as noted earlier, has a calibrated 14C age of only 3,813 ± 128 years BP. Adamson and F. Williams (1980) noted that the headwaters of Wadi Howar north of Jebel Marra and of Wadi Azum southwest of the massif are curved backwards and appear to have once flowed east. They speculated that the uplift and eruption of the Jebel Marra volcanic complex caused the westward diversion to the Chad Basin of drainage that previously flowed east to the Nile and concluded that a catchment of more than 60,000 km2 had been diverted from the Nile Basin with a shift of the Nile–Chad divide 300 km to the west (Adamson and F. Williams, 1980, p. 239, Fig. 10.8). They further concluded that the thick mantle of colluvial and alluvial material filling the valleys and covering the gentle slopes west of the massif was also a result of this uplift. These sediments had previously been designated as ‘drift’ soils (FAO, 1969, chapter 3), but their origin had never been satisfactorily explained. Since the term ‘drift’ is widely used for glacial till, it is in any case not a suitable term for these sediments, which we consider in Section 13.4.

13.3 Flora of Jebel Marra and Its Significance Jebel Marra is located between the desert steppe to the north and the tropical savanna woodlands to the south. The eastern slopes are covered with many metres of volcanic ash. The soils developed on these pyroclastic deposits are high in potassium and have given rise to relatively fertile soils (FAO, 1969). For several thousand years the local Fur farmers have terraced the slopes of Jebel Marra to benefit from its high rainfall and fruitful soils. It would thus seem probable that Jebel Marra has long operated as a staging post for plants, animals and human groups, linking the northern deserts with the seasonally wet tropics to the south,

13.4 Piedmont Sediments

181

and the rolling dune country of Kordofan and the White Nile Valley in the east with the savanna and sahel regions to the west of the Nile Basin. This portrait is not entirely correct and the reality is both more complex and more interesting. As an adjunct to the soil surveys carried out by Hunting Technical Services on and around Jebel Marra and in Southern Darfur (Hunting Technical Services Limited, 1958, 1974, 1976), plant ecologist Dr Gerald Wickens carried out detailed collections of the plants growing in this region, the results of which he published as a monograph (Wickens, 1976a) and related papers (Wickens, 1975a, 1975b, 1976b, 1982). Wickens (1976a, p. 71) noted that the Jebel Marra massif is the result of ‘a relatively recent volcanic upheaval, hence taxa can be considered as migrating to and not from the massif’. He found that some elements of the flora on Jebel Marra pointed to an origin in the Ethiopian Highlands and the Red Sea Hills (Wickens, 1976a, p. 69). Possible links to East Africa were more tenuous. A number of Mediterranean elements were also present in the Jebel Marra flora, raising the question of possible migration routes. Two likely routes are from NW Africa along the river valleys draining the Atlas Mountains via the Hoggar and Tibesti to Jebel Marra. This is certainly plausible, and there is ample evidence of rivers flowing south from the Atlas (Alimen and Chavaillon, 1963; Alimen et al., 1966; Conrad, 1969) and north across the Sahara from Miocene times onwards (Griffin, 1999, 2002, 2006, 2011; Drake et al., 2011; Coulthard et al., 2013). Another possible route is via the Red Sea Hills, and Wickens pointed out that Suakin on the east coast of the Red Sea Hills receives much of its precipitation between September and January, which is similar to the Mediterranean winter rainfall regime. ‘It would require very little increase in rainfall during the winter months to facilitate the unimpeded migration of Mediterranean species along the Red Sea Hills’ (Wickens, 1976a, p. 78). Both options are of course possible. Maley (1980, p. 82) has raised the question of whether the fragment of Mediterranean olive tree identified as Olea lapperinei on Jebel Marra might not in fact be Olea chrysophylla – a topic for future plant physiologists to resolve. One surprise arose from Wicken’s analysis of the savanna flora (Wickens, 1976a, pp. 61–64). Contrary to expectation, fewer than 60% of the species now growing in Sudan are also found in the savanna of Niger and Nigeria, suggesting that there was restricted exchange between these two great regions south of the Sahara. He proposed that the barrier was provided by Lake Chad during times when it occupied much of the Chad Basin, which it certainly did during the late Quaternary, if not earlier (Grove and Pullan, 1963; Servant, 1973; Servant and Servant-Vildary, 1980; Drake and Bristow, 2006).

13.4 Piedmont Sediments 13.4.1 Fossil Oil Palm Leaves and Early Stone Age Artefacts Following the advice of Dr Gerald Wickens, the author and his colleagues (see Williams et al., 1980) investigated the piedmont deposits near the village of Umm Mari, 20 km west of Kalokitting and 10 km south of Nyama village (Fig. 13.2). At least 20 m of alluvium overlies Precambrian Basement Complex granitic gneiss and weakly foliated granite in this

100 km

Deriba crater

MARRA

Nyala

Tertiary volcanic rocks and minor centres

Quaternary fluviatile, lacustrine and aeolian sediments

50

Barbis

JEBEL

D

El Fasher

B

m

Fa

Fb

Precambrian basement with thin or sporadic sedimentary cover

F

D

basalt flow

clay

quartz gravel

quartz sand

tuffaceous silt

m

0

2

4

wadi floor

bedrock

ash

stone tools sand, gravel and stone tools finely laminated diatomite diatomaceous ash

D

Fossil oil palm locality

Diatomite locality

pumice gravel

volcanic gravel

terrace: sand/silt

palaeosoll

0

10

20

m

leaf and stem fossils and invertebrate tracks

oil palm fossils basalt choppers

fossil oil palm leaf impressions

Nubian (?) sandstone

0

5

10

15

19

Figure 13.2 Fossil soils associated with prehistoric artefacts and oil palm leaf fossils in the piedmont zone of Jebel Marra volcano, northwest Sudan. (After Williams, 2014, Fig.15.5.)

A

0

Kas

F

Baya

Umm Mari

Zalingei

Kebkabiya

JM

Middle Member

Lower Member

13.4 Piedmont Sediments

183

locality. The basement rocks crop out in wadi channels and also as steep hills or jebels rising 30–50 m above the alluvial plain. One kilometre south of Umm Mari the wadi floor consists of aphanitic basalt and lies 50 m below the summits of the jebels. The basalt appears to occupy a former valley cut into the basement rocks and probably dates from early in the formation of the Jebel Marra massif. Detailed study of eleven well-exposed stratigraphic sections located within a four km radius of Umm Mari village revealed the following sequence of events: 1. Weathering of the Basement Complex rocks and formation of a soil in situ on the granitic gneiss 2. Extrusion of basalt and basalt flows down pre-existing valleys eroded in the Basement Complex rocks 3. Deposition of Red Beds (Bed 1): red gravelly sands and sandy gravels derived from Basement Complex rocks, with minor lenses of fine basalt gravel, but no pumice 4. Erosion and incision 5. Deposition of pumiceous sands and fine gravels, and volcanic tuffs containing fossil oil palm (Elaeis guineensis) and Combretum leaf and stem impressions (Bed 2) 6. Erosion and channelling of Beds 1 and 2 (Red Beds and pumiceous sediments) 7. Deposition of volcanic ash as channel-fill deposits (Bed 3) within Beds 1 and 2 8. Deposition of Basement Complex–derived gravels (Bed 4) over the surface of Beds 1, 2 and 3, which together constitute the Lower, Middle and Upper Members of the Umm Mari Formation, an informal stratigraphic name given by Williams et al. (1980) for this very widespread set of deposits 9. Erosion and incision 10. Deposition of Younger Ash (Bed 4) as shower-fall deposits masking the talus slopes of Bed 2, and occupying the heads of small valleys eroded into the Umm Mari Formation 11. Erosion and incision 12. Deposition of alluvial silts (Bed 5) 13. Incision and formation of Upper Brown Loam Terrace (+8 m) 14. Deposition of Recent alluvial silts (Bed 6) 15. Incision and formation of Recent Grey-Brown Loam Terrace (+2 m) 16. Deposition of sub-recent coarse alluvial sands (Bed 7) 17. Incision and deposition of modern wadi sands (Bed 8) The three major alluvial terraces identified in this area are the +20 m alluvial terrace, corresponding to the top of the Umm Mari Formation (Beds 1, 2 and 3), the +8 m terrace corresponding to the top of the Upper Brown Loam (Bed 5), and the +2 m terrace corresponding to the top of the Recent Grey-Brown Loam (Bed 6). We discuss the two younger terraces in Section 13.4.3. The age and full extent of the Umm Mari Formation is not known but prehistoric stone artefacts discovered within the Formation offer some useful clues. In the base of a deposit of mixed granite and basalt cobbles resting unconformably on Bed 1 there were rolled basalt

184

Jebel Marra Volcano

flakes and polyhedra together with a slightly abraded chopper of Oldowan/Acheulian aspect. The gravels in which the tools were found are angular to subangular in shape, so it seems unlikely that the stone tools have been moved very far from their original source. An early Pleistocene age for the Red Beds is therefore plausible. A fresh in situ basalt chopper was found just beneath the thin fossil-bearing indurated tuff of Bed 2. Elsewhere in the Umm Mari Formation there were a number of similar crude stone choppers and flaked basalt pebbles, including bifacial and unifacial choppers, pushplanes, discoids, hammer-stones, and flake scrapers of Developed Oldowan or Early Acheulian aspect. Archaeologist Dr John Gowlett, who very kindly examined the author’s collection in February 1976 in Khartoum, considered that an age range between 1.5 Ma and 0.8 Ma was the most probable. This implies that the Umm Mari Formation, and the fossil oil palm leaves in Bed 2, would also fall within this time range. They are certainly not Holocene in age.

13.4.2 Pleistocene Lake Diatomite and Early Stone Age Artefacts Colchester (1927) reported the presence of diatomaceous earth in the vicinity of Kebkabiya, a small town west of Jebel Marra, but offered no exact location. The precise provenance of the diatomite samples brought to Kebkabiya by donkey and camel and from there by merchants to the markets in Khartoum was not known, nor was their age. Dr Prasad in the Geology Department at the University of Khartoum had earlier analysed some bulk samples and considered them to be of possible Holocene age (Prasad, 1971). Dr Mohamed Shaddad (Geology Department, University of Khartoum) suggested to the author in January 1976 that he might first enquire at Kebkabiya, and then enquire further at Barbis village and at Bara (or Baya) hamlet, on the western flank of the mountain, where the local people had referred to Hofrat el Jirr or lime pits. Figure 13.2 shows the location of these places. In fact, the diatomite site lies 30 km southeast of Kebkabiya and 8 km east of Baya (or Bara) in a basin (13°27 0 N, 24°18 0 E) drained through a small gorge incised into deeply weathered basalts to the southwest. The village of Baya is close to the flat-topped basalt hill Jebel Baya, the steep volcanic pinnacle of Jebel Bara Simbila and the basalt mesa of Jebel Ergi with its magnificent columnar jointing. Broad ridges of volcanic rock up to 100 m high border the basin to the north and west, and directly overlie Precambrian granite. The basin is separated to the east from a deep SSE trending valley by lobes of a rhyolite flow that seems to have come from the north. This flow is less weathered than the basalt and was probably responsible for damming the local drainage and creating a lake. The diatomite crops out on a gently sloping surface with a visible vertical thickness of 5.3 m. Augering revealed that the diatomite continued for a further metre below the surface and was underlain by weathered basalt. The diatomites are finely laminated to massive apart from the basal 40 cm, which are best described as tuffaceous diatomites (Williams et al., 1980). The diatom assemblage, analysed at 5-cm intervals, and the fine laminations are consistent with a deep freshwater lake (Philibert et al., 2010). Published rates of diatom accumulation

13.4 Piedmont Sediments

185

in East African lakes and in Lake Malawi range from 0.1 mm/yr to 2–3 mm/yr (WashbournKamau, 1971; Owen et al., 2008). If we apply these rates to the former lake near Barbis, we obtain a maximum duration of about 60 ka and a minimum duration of about 2–3 ka for the lake. A thin layer of colluvial-alluvial sand and scattered basalt and trachyte gravel lies on top of the diatomite. Within the gravels Williams et al. (1980) found sporadic large trimmed flakes, flake scrapers, bifacially worked choppers and scrapers and the broken butt of a hand axe. One artefact was very fresh with sharp edges, some were slightly abraded and three were heavily abraded. Dr John Gowlett from the Department of Archaeology at the University of Khartoum examined all the artefacts in February 1976 and considered them typologically similar to the artefacts excavated at Olorgesailie in the Kenya Rift and from Bed IV at Olduvai in Tanzania. These sites straddle the Brunhes–Matuyama palaeomagnetic boundary (0.78 Ma), so he suggested an age range of 0.8 ± 0.3 Ma for the stone tool assemblage (Gowlett, in litt., February 1976). Similar Developed Oldowan/Early Acheulian stone tool assemblages elsewhere in East Africa and Ethiopia have a maximum age range from 1.5 Ma to 0.3 Ma and a probable age range from 1.2 Ma to 0.8 Ma (Clark and Kurashina, 1979; Williams et al., 1979; Isaac, 1982; Gowlett, 1984; Clark, 1987; Clark and Schick, 2000; Clark et al., 2008b; Owen et al., 2008).

13.4.3 River Terraces and Middle and Late Stone Age Artefacts South of Jebel Marra in the area around Umm Mari there are well-preserved river terraces, some of them erosional and some depositional (Fig. 13.3). The surface of the highest terrace lies 20 m above the modern wadi floor and comprises the Lower, Middle and Upper Members of the Umm Mari Formation described in Section 13.4.1. The Early Stone Age artefacts within this Formation suggest a probable age range between 1.5 Ma and 0.8 Ma. A second terrace is cut across the surface of the Lower Member and lies 8 m above the wadi floor. Strewn across the surface of this erosional terrace is a discontinuous lag gravel of mixed volcanic and granitic stones and pebbles. Lower still there is another terrace, which rises 2–4 m above the main wadi floor. This alluvial terrace is banked against the base of the Umm Mari Formation. It consists of moderately indurated grey-brown clayey sand, ranging in texture from loamy medium sand near the base to medium sandy loam near the top. It is distinctly more organic and less sandy than the modern wadi channel sands. At one site near the base there was a band of coarse Basement Complex river gravels and a mica schist upper grinder of possible Neolithic age. In the small tributary channels the third terrace consists of pale brown horizontally bedded loamy sands and fine gravelly loamy sands and usually lies only 1.5–2.0 m above the channel floor. A sub-active alluvial bench lies above the channel bed and is banked against the lowest terrace. It consists of medium to coarse sand. The grey-brown loams of the 2–4 m terrace are highly valued by local farmers for the readily tilled structure, excellent water-holding capacity, and high innate fertility owing to their volcanic provenance. The terrace silts extend at least 0.5 m beneath the level of the

186

Jebel Marra Volcano m

15 gb

(a)

m 20

Upper terrace

(b)

Upper Member

gb

14

Upper gb

13

(Tuffaceous and pumiceous)

16

10

member

Holocene terrace

12 gb

Lower Member

14 Leaf fossils

(Quartzose)

11 gb 10 pb

m

9 gb

12

0

Basalt

8 g pb

7 6 5

rb rb

Upper terrace (20m)

(c) 10

Modern terrace

8

Lower terrace (2m)

Middle terrace (8m)

Rock

6 4

rb

Sandy clay loam 4

3 rb

2 rb 1

gb

Lower member

Silty clay

Coarse sandy gravel Quartz sand

2

Reworked ash Pumice gravel and tuffaceous sand/silt

0

Silt

Figure 13.3 Alluvial terraces south and west of Jebel Marra. (Based on author’s field notes and Williams et al., 1980, Fig. 13.7.)

modern coarse bed load sands of the wadi channels. They were probably laid down during an interval characterised by perennial stream flow, a moderately dense plant cover in the catchment area, and gentle well distributed rains (Williams et al., 1980). Very similar dark loams have been described in the Hoggar, Niger, Chad and Ethiopia, where they are all of Early to Middle Holocene age (Rognon, 1976b; Messerli et al., 1980; Williams et al., 1981; Williams et al., 1987; Maley, 1981,1982; Williams, 2008).

13.5 Deriba Crater Lakes and Late Pleistocene High Lake Levels

187

Further investigations by Williams et al. (1980) west of Jebel Marra shed fresh light on the possible age and significance of the alluvial terraces. In the area between Nyertete and Zalingei many of the soils were mapped as ‘Drift Soils’ (FAO, 1969, p. 44). The soils that fall under this umbrella term are so varied that the term has little diagnostic value. However, in this area the seasonal wadis are flanked by at least four sets of alluvial terraces, at heights of 1.5 m, 3 m, 4 m and 8 m above the wadi bed. The dark brown loam comprising the 1.5 m terrace is in all respects the same as that in the 2 m terrace in the Umm Mari area south of the massif, and is also prized by local farmers. There are three older alluvial terraces above the 1.5 m terrace, at elevations of 3 m, 4 m and 8 m. At one site a large worked Middle Stone Age flake was found eroding from the surface of the 8 m terrace, on the surface of which there was a mixed concentration of both Middle and Late Stone Age quartz flakes. Late Stone Age flakes were also strewn across the surface of the 4-m terrace, but were not seen on or within the red-brown loam 3-m terrace or the dark brown loam 1.5-m terrace. Elsewhere in Africa the Middle Stone Age has an age range between about 300 ka and about 50 ka, and the Late Stone Age from about 50 ka to about 10 ka (Williams, 2014, p. 310). The 8-m terrace is therefore probably of very late Middle to Upper Pleistocene age, the 4-m terrace is probably late Upper Pleistocene and the 1.5-m terrace Holocene. We can speculate that during times of proximal stream erosion, widespread sediment transport and distal stream aggradation, the prevailing climate was probably less arid than it is today, when the wadis are carrying coarse sand and are engaged in incising their channels. Further evidence of previously wetter conditions is provided by the alkaline springs south of Kalla, where a bench of silicified limestone 10 m above the modern wadi floor contains subfossil shells of Bulinus and Biomphalaria – both freshwater gastropods and vectors of Schistosomiasis – together with a fossil tortoise shell. The mollusca would have required permanent fresh water and an abundance of aquatic vegetation, both now lacking from this area.

13.5 Deriba Crater Lakes and Late Pleistocene High Lake Levels In their engaging and informative book Volcanoes, Francis and Oppenheimer (2004) wrote about the great caldera of Jebel Marra with a touch of forgivable hyperbole: ‘Its 5-km diameter caldera was formed by a colossal eruption about 3500 years ago, an event that may have showered ash on Pharaonic Egypt’ (Francis and Oppenheimer, 2004, p. 27). Perhaps so, but it would also have depended on which way the winds were blowing when the volcano last erupted. For over half the year today the winds blow from the north and east. In actual fact, the main caldera is a great deal older than late Holocene, and has a complex and very interesting history. The following comments are largely based on a short visit to the caldera by the author and his colleagues in January 1976, partly summarised in Williams et al. (1980). No subsequent geological work has been possible since then owing to the insecurity in Darfur.

188

Jebel Marra Volcano

SHALLOW SALINE LAKE F 1

ASH DAM

3

DEEP LAKE 2

N

0

2 km

Trachyte forming caldera walls

Caldera rim Ridge

Pumiceous ash (not including explosion crater deposits). Fine ash and tuff with fragments of trachyte and basement material.

Outer rim Summit ridge Inner rim

}

Explosion crater rim deposits. Mostly tuff with fragments of trachyte and very rare basement material. Inner rim deposits numbered in probable order of deposition

Hotspring F

Fumarole

Quaternary alluvial and lacustrine deposits

Lakes

Figure 13.4 Geological map of the Deriba caldera, Jebel Marra. (Adapted from Williams et al., 1980, Fig.13.1.)

The uneven rim of the caldera rises 300 to 1,000 m above the caldera floor. The caldera walls consist of very steep to vertical cliffs of trachyte. Pyroclastic deposits associated with the caldera collapse are abundant inside and outside the caldera. They consist of air-fall and water-borne pumiceous tuff and ash, with fragments of trachyte, Basement Complex rocks, and rare pitchstone. Lynes and Campbell Smith (1921, p. 211) reported the presence of

13.5 Deriba Crater Lakes and Late Pleistocene High Lake Levels

189

‘large blocks of . . . metamorphosed, foliated igneous rocks’ on one of the ridge crests of the caldera, including ‘a biotite-bearing hornblende gneiss . . . a fine-grained white gneiss poor in biotite . . . and a hornblende-schist’. Ignimbrites are less common but are visible near the crater rim and apparently flowed down both the outer and inner caldera walls. Inside the caldera, horizontally bedded deposits of pumiceous ash and tuff are plastered against the caldera walls up to 250 m above the caldera floor, and form a 100 m high barrier across the eastern gap in the caldera wall (Fig. 13.4). Inside the main caldera there are two lakes, known as the Deriba lakes. The deeper of the two is about 1 km in diameter, is moderately saline, has a maximum depth of 108.8 m, and occupies an explosion crater within the main caldera (Hammerton, 1968). This crater consists of five nested craters, each formed during successive phases of explosive activity. The highest point of the crater rim is over 200 m above the lake surface, which our aneroid measurements showed to be 2,150 ± 20 m, or about the same level as the main caldera floor (Williams et al., 1980). Hot saline springs appear to be episodically active at the bottom of this lake, as shown by the intermittent presence of a saline layer at the bottom of the lake (Hammerton, 1966, 1968). Hammerton (1968) also found evidence of two former beach levels at depths of 3 m and 5–7 m in the deeper lake, shown by drowned trees and other plant remains. Both he and we found evidence for a recent rise in water level, with a number of dead trees standing in 2–3 m of water. Hammerton (1968) proposed that landslips from the crater rim caused the lake to rise; we considered that a recent increase in rainfall was a more likely cause of the rise in lake level (Williams et al., 1980). The second shallow lake proved to be of particular interest during our visit to the caldera in late January 1976. This lake is about 2.5 km long with a maximum depth of 11.6 m. It is at present highly saline and alkaline. Hammerton (1968) observed that in 1966 the lake was twice the length reported by Hobbs (1918) and by Gillan (1918) during their visit in March 1918, and was less saline than when monitored by Lynes in 1920 during the first of his two plant-collecting visits to Jebel Marra (Lynes, 1921). Lake Chad was also very low during the first two decades of the twentieth century, as was the summer flood level of the Nile, and was high again during the 1950s and early 1960s, prompting Williams et al. (1980) to conclude that the historic fluctuations in the two Deriba lakes were more likely to have been caused by climatic fluctuations, reflecting changes in the balance between precipitation and evaporation, than by purely local changes in spring activity. The historic fluctuations in precipitation are far from trivial. For example, at the end of 1972, following 4 years of drought, the volume of Lake Chad had diminished from 60 million km3 to 17 million km3. After two wet years the lake had risen 2 m above its very low 1972 level. Sadly, by 2016 Lake Chad had shrunk to a fraction of its former self, prompting international concern. There are five well-preserved shorelines surrounding the shallow saline lake, at elevations of +1 m, +2.5 m, +8 m, +16 m and +25 m. Each of the benches is dissected by narrow gullies. Figure 13.5 shows the probable extent of the +8 m lake and was based on foot traverses and 1:40,000 scale air photos (Williams et al., 1980). Calibrated radiocarbon ages for algal carbonates associated with the higher shorelines revealed that the lake was stable

190

Jebel Marra Volcano

SHALLOW SALINE LAKE

DEEP LAKE N

0

2 km

Summit of caldera rim

Residuals of former higher crater floor

Explosion crater rim summit Ridge Active and incised fans

Present lakes

Former lake (+5 to +8m) Estimated highest shoreline

Figure 13.5 Geomorphic map of the Deriba caldera, Jebel Marra. (Adapted from Williams et al., 1980, Fig.13.12.)

at +5 to +8 m between 22.6 ka and 19.4 ka, then fell, but rose again to +9 m at 16.8 ka. This material is not ideal for radiocarbon dating, so that these ages need to be regarded as probable rather than definite, but it is worth noting that similar ages have been obtained for crater lakes in Tibesti situated 1,100 km to the NW (Maley, 2004). On the northern side of the lake the vertical trachyte cliffs are coated with a crust of algal carbonate up to 30 cm thick, which vanishes abruptly at +25 m. On the southern and eastern margins of the lake deltaic deposits with top-set beds at +25 m are cut into by the +5 m and +8 m strandlines, which are in turn older than the two lower shorelines (Fig. 13.6). Two auger holes were

13.5 Deriba Crater Lakes and Late Pleistocene High Lake Levels

191

Distal end of talus slope

30 m 25

+ 25 m strandline Topset beds

20

Current-bedded finely laminated sands and gravels with occasional trachyte blocks. Deltaic foresets dipping 10-20° graded to + 25 m lake level

15

10

+ 8 m bench 5 + 2 m bench + 1 m bench

Sub-horizontally bedded pumice sands and gravels

0 NE

Lake level: January 1976 c 2140 ± 20 m

SW

Figure 13.6 Section at southwest edge of the shallow Deriba caldera lake, Jebel Marra, showing former lake shorelines and eroded deltaic sediments. (Adapted from Williams et al., 1980, Fig. 13.14.)

drilled, one 60 m from the southwest edge of the lake and the other on the +5 m bench above the lake. The main constituent in both was pumice with minor fragments of trachyte embedded in a matrix of volcanic ash. The combined evidence reviewed here now allows us to reconstruct the main stages in the geomorphic history of the caldera: 1. Formation of the main caldera (? Early Pleistocene) 2. Accumulation of pyroclastic deposits as high-level talus deposits along the inner wall of the caldera 3. Erosion and reworking of the talus deposits to form a high-level caldera floor and a probable deep lake 4. Overflow through a high level (+40 m) gap in the southwest caldera wall 5. Ash dam blocks the +40 m gap 6. Deposition of finely laminated deltaic tuffaceous sediments within a caldera lake at least 25 m deep, most probably somewhat before 23 ka 7. Formation of lacustrine algal carbonates up to 25 m above the caldera floor level 8. Fall in lake level 9. Lake stable at +5 m to +8 m between about 23 ka and 19 ka 10. Fall in lake level, followed by a possible rise to +9 m towards 17 ka

192

Jebel Marra Volcano

11. Formation of the inner crater in five or more stages, the latest of which was most likely about 3.8 ka, after which the deep lake formed 12. Shallow lake stable at +2.5 m in main caldera 13. Fall in shallow lake level 14. Lake stable at + 1 m 15. Fall in level of deep lake to –2 m 16. Modern lake levels, with minor fluctuations of ±1 m during the first 70 years of the twentieth century. Deep lake level lower in 1918, high in the 1960s and probably lower again in the 1970s

13.6 Pleistocene and Holocene Erosion and Sedimentation Desert regions often experience complex interactions between fluvial and eolian geomorphic systems and processes (Williams, 2015). The Jebel Marra region is no exception. Linear sand dunes blown in from the north and east are deflected around Jebel Marra. Vegetated linear dunes are the dominant landform east and south of Jebel Marra, and extend up to 200 km east of the massif and 150 km to the south. They are breached in many places by the east and south flowing channels of the Wadi el Ku drainage basin and further south still by the headwater tributaries of Wadi Ibra (Fig. 13.7). During wetter climatic phases Wadi Ibra flowed into the Bahr el Arab which itself joins the White Nile (Hunting Technical Services, 1976; Williams et al., 1980). Wadi Ibra flows along the northeast margin of the Qoz Dango, which is bounded to the south by the Bahr el Arab. Qoz Dango is very different from the linear dunes that were blown in from the north and represent the former southern margin of the Sahara. It is highly localised and has never migrated any distance from its source (Parry and Wickens, 1981). During drier phases the Qoz Dango dunes were devoid of plant cover and moved a short distance to block any nearby river channels, such as Wadi Ibra. Once wetter conditions resumed these channels breached the dunes, which soon became stabilised by vegetation. Boreholes and channel-bank sections in this area show two main phases of late Quaternary alternating fluvial and dune activity, followed by midHolocene clay deposition, renewed dune activity, fluvial silt deposition, river incision and deposition of the modern wadi sands (Williams et al., 1980). Fossil river channels of possible early Holocene age north of the Bahr el Arab were once wide and sinuous but are now segmented, discontinuous and silting up. Desiccation had probably set in by about 5,000 years ago, although there were brief wetter interludes after that. Neolithic and younger sites are common in this area. Both Servant (1973) and Maley (1981) have documented (and dated) a similar sequence of late Quaternary climatic fluctuations in the Chad Basin and the Bahr el Ghazal catchment. During the first half of the Holocene, Lake Chad rose 40 m above its historic 280 m elevation (Servant, 1973). Even more striking, Lake Assal in the Afar Rift is now a highly saline lake at 150 m below sea level. During the Early Holocene it was a freshwater lake stable at 150 m above sea level (Gasse, 1975).

13.6 Pleistocene and Holocene Erosion and Sedimentation

193

Wadi Howar

14°N Kebkabiya El Geneina Zalingei

W ad

JEBEL MARRA

El Fasher i e l

Ku

Wa d

i Azum

Deriba crater

Nyala 12°N

Wa d

N iI

br

0

a

100 QOZ DANGO

km

24°E

Cenozoic volcanics

Hofrat en Has

Bah

r el A r

ab

26°E

linear dunes

Figure 13.7 Sand dunes and drainage systems in the Jebel Marra area. (Adapted from Williams et al., 1980, Fig. 13.5.)

One striking and somewhat enigmatic feature of Jebel Marra is the system of terraces used and maintained by the Fur cultivators living along the slopes of the massif. The terraces have a dual purpose – to conserve soil from moving downslope and to direct

194

Jebel Marra Volcano

water from permanent springs onto the cultivated fields. Paterson (1948, pp. 857) offers a vivid description: ‘Terraces, massively built of enormous boulders, rise tier upon tier to the very crowns of some of the highest peaks.’ The age of these terraces is not known, with suggestions ranging from prehistoric to late mediaeval. The Chagga farmers on the southern slopes of Mount Kilimanjaro make good use of very similar terraces; when I asked their age, they became vague.

13.7 Conclusion Jebel Marra is a dormant volcano close to the western edge of the Nile Basin in Darfur Province of Sudan. It probably began to form during the Miocene, was intermittently active during the Pleistocene, and last erupted late in the Holocene about 3 ka ago. There are two lakes inside the main caldera, one deep and brackish, the other shallow and saline. The shallow lake was deep and fresh at intervals before, during and after the Last Glacial Maximum. Deep alluvial deposits occupy the piedmont zone south and west of the main volcanic massif. The oldest of these sediments were eroded from the Basement Complex rocks through which the volcanic lavas were erupted. Overlying these red alluvial sands and gravels there are layers of reworked volcanic ash. Oil palm (Elaeis guineensis) and Combretum leaf fossils within the ash beds occur in association with Lower Palaeolithic stone tools with a probable age range of 1.5–0.8 Ma. At about this time a deep freshwater lake was in existence on the western piedmont and 6.3 m of finely laminated diatomite accumulated within the lake. Stone tools on the surface of the former lake bed have a probable age range of 1.2–0.8 Ma, once again indicating that the regional climate during the Lower Pleistocene was on occasion far wetter than it has been since. The Combretum and oil palm (Elaeis guineensis) leaf fossils found in reworked volcanic tuffs near Umm Mari village 90 km southwest of the diatomite site are strong evidence of the former presence of tropical rainforest in this region. The wide age range suggested by the Developed Oldowan/Early Acheulian stone tool assemblages present at both localities does not allow us to conclude that the fossil sites and diatomite site are the same age. We can conclude that the climate indicated by the oil palm leaf fossils was very much wetter than today to allow tropical rain forest to grow in this now semi-arid locality. The deep freshwater lake is also consistent with a wetter than present climate. Most likely these wet phases were coeval with one or more Lower to Middle Pleistocene wet interglacial episodes within the 1.5–0.3 Ma maximum time range. If the long duration for the former lake near Barbis is accepted, then it may date to the MIS 11 interglacial, which lasted from 420 ka to 395 ka, or about 25,000 years. Conversely, if the short duration for the lake is accepted, then any one of a number of interglacial phases becomes possible. The important point is that the piedmont deposits south and west of Jebel Marra contain a very long record of environmental change extending well back into the Pleistocene – a record that no longer needs to be telescoped into the few thousand years of the Holocene.

13.7 Conclusion

195

The presence of now vegetated and stable linear dunes as far south as latitude 12°S is evidence of a 500 km southward shift of the effective southern limit of the Sahara at the time these dunes were active. Equally, the Umm Mari oil-palm fossil leaves and the lacustrine diatomites near Barbis show that the northern limits of the savanna once extended well north of the massif.

14 The Desert Nile

My own observation bears out the statement made to me by the priests that the greater part of the country I have described has been built up by silt from the Nile. Herodotus (ca. 485–425 BC), The Histories (1954, p. 105)

14.1 Introduction After receiving its last increment of water and sediment from the River Atbara (Chapter 12), the main Nile or ‘Desert Nile’ pursues a course of 2,689 km across the hyper-arid eastern Sahara and today receives no further tributaries before finally reaching its delta and flowing into the Mediterranean Sea (Fig. 14.1). The total catchment area of the Desert Nile is just over 1 million km2, which is roughly a third of the total Nile Basin area. Except for the Nile itself, this vast area is presently devoid of perennial drainage. We have already reviewed the diverse opinions relating to the age of the Nile in Chapters 2 and 4. For the purposes of this chapter we will concentrate on the Quaternary record of erosion and sedimentation in the valley of the Desert Nile. The local landscape offers opportunities as well as constraints for present-day human land use and settlement, just as it did in the past. To understand the links between the everchanging physical landscape and the pattern and tempo of prehistoric occupation, we need to have a clear understanding of the origin and distribution of the major elements in the landscape bordering the Desert Nile. A series of alluvial and erosional terraces run roughly parallel to the Nile and contain a variety of stone tools that range back in age to the Lower Palaeolithic or Early Stone Age. The terraces bear witness to repeated episodes of Nile aggradation and incision, but because they are not particularly well dated they can usually only provide a relative chronology of former river behaviour. Numerous previously active wadis (desert rivers) flowed west from the Red Sea Hills to join the Nile; their sediments inter-finger with Nile alluvium and appear to reflect times of more humid local climate possibly associated with winter rains. Throughout the Holocene (and, very probably, much of the Pleistocene) there was a seasonal, albeit fluctuating, supply of water and sediment from the Ethiopian headwaters 196

14.1 Introduction

197

30°E

40°E

MEDITERRANEAN SEA

N CAIRO

30o N

0

250

500

LIBYA

km

EGYPT

e Nil

RE

Kom Ombo

I

Aswan

D

Aswan Dam

II

Laqiya

a

CHAD

E

R

Me

E

IT

KHARTOUM

A

Qoz Abu Dulu

e

Wa d el

VI

lik

Wh

Nil e

ite Nile

Jebel Marra

Atb ar

Blu

SUDAN

V

IV

R

W

ar Ho w

Seleim Basin

a

di

Dongola Old Dongola

Wad el Arab Kerma

A SE

Qaab III Basin

20°N

ET HI OP Semien IA Mts Lalibela

Figure 14.1 The Desert Nile (shown as heavy black line) and the location of places mentioned in the text. Roman numerals indicate cataracts. Nile Basin shown in grey shading.

of the Blue Nile and Atbara, while environments near the Desert Nile oscillated between more and less arid. As the Holocene climate became drier, most notably during the Late Holocene from about 4.2 ka onwards, the local wadis provided less and less sediment to the Nile and the influx of wind-blown sands and dust increased. In northern Sudan, the Holocene alluvial sediments adjoining the Nile are particularly well dated by several independent methods and reveal the successive stages in prehistoric cultural evolution from Mesolithic hunters, fishers and gatherers to Neolithic pastoralists and cultivators to town dwellers and monument builders.

198

The Desert Nile

14.2 Deciphering Nile Alluvial history The Quaternary alluvial record along the Desert Nile is fragmentary and therefore hard to interpret (Williams, 2009a, 2012). Three quite independent factors can influence erosion and deposition by rivers. The first concerns changes in base level caused by fluctuations in sea level. For example, during the Last Glacial Maximum (LGM), around 21 ± 2 ka, at a time when ca. 50 million km3 of water was abstracted from the oceans and locked up in the ice sheets of both hemispheres, global sea level fell 120–130 m, and the distributary channels flowing across the Nile Delta cut down accordingly. A more extreme example occurred during the Messinian Salinity Crisis (5.96–5.33 Ma), at which time the Mediterranean dried out to become a salt desert and the base level of the Nile was lowered several thousand metres (Chapter 2). The result was deep incision by the ancestral Nile and the carving out of a canyon as far as the 1st Cataract at Aswan (now the bedrock foundation of the Aswan High Dam). Rushdi Said referred to this late Miocene Nile canyon as the ‘Eonile’ (Said, 1993). The impact of base level changes caused by fluctuations in sea level can have repercussions far upstream, but only as far as the first major nick point in the river, which in the case of the Nile is the most northerly or 1st Cataract. The second factor can be described very simply as tectonic, and refers to earth movements that are sometimes relatively slow, such as uplift or subsidence caused by movement of tectonic plates, or uplift beneath a mantle ‘hot spot’ (Chapter 4), and sometimes quite fast, as in the case of earthquakes and volcanic eruptions. Tectonic uplift has certainly caused river incision in the Ethiopian Highlands during the last 30 Ma (Chapter 6), as has uplift of the Nubian Swell in northern Sudan and southern Egypt during the latter half of the Cenozoic (Thurmond et al., 2004). As we come closer to the present, the third and most important factor concerns changes in river discharge and sediment type and amount; for a detailed analysis see Williams, 2014: chapter 10. We saw in Chapter 11 that the ability of a river to transport its sediment load and erode its channel is determined by what is termed its ‘stream power’. Schumm (1977) noted that sediment transport and stream power are proportional to stream velocity cubed. If stream power falls below a certain threshold value, bed and bank erosion and sediment transport will decline, leading to sediment deposition within the channel. As a first approximation, when the ratio of stream sediment load to stream discharge rises to the level at which the stream can no longer transport its sediment load, deposition will occur. Also important is the proportion of sediment carried in suspension as opposed to being transported as traction load or bed load. Streams that have become mudflows or debris flows can carry much larger particles than streams of clear water. Stream velocity is of paramount importance. A doubling of stream velocity allows the river to carry particles up to sixty times larger in diameter (the Sixth Power Rule of G. K. Gilbert, 1914). Velocity increases with discharge. In the case of the Blue Nile, which is the most important feeder of water and sediment to the main Nile, the flood discharge in the wettest month (August) is on average forty times greater than in the month of lowest discharge (May). The Blue Nile sediment concentration is about forty times higher during August than during May, so that

14.2 Deciphering Nile Alluvial History

199

the Blue Nile ferries more than a thousand times more sediment to the main Nile during the month of peak flood than during the month of lowest flow. Before dams were built, much of this sediment used to reach the Mediterranean Sea, but a significant amount was also deposited across the flood plains of the Nile, as Herodotus astutely observed. The Blue Nile and the Atbara together provide 68% of the flood discharge and 97% of the total sediment load delivered to the main Nile (Chapter 3). We therefore need to remind ourselves that any changes in climate and vegetation cover in the headwaters of these rivers will be reflected in changes in the seasonal flow regime as well as changes in the amount and type of sediment carried downstream to the Nile (Chapters 6 and 12). Nor should the role of the White Nile be forgotten (Chapter 8). This river provides 83% of the discharge into the main Nile during the month of lowest flow. During years of extreme and sustained drought in the Ethiopian headwaters of the Blue Nile and Atbara, it was water from the White Nile that maintained perennial flow in the main Nile, at least until the Aswan High Dam was built. If the contribution of the White Nile is cut off and flood flow from the Blue Nile and Atbara greatly reduced, as happened most recently during and after the LGM until about 15,000 years ago, then perennial flow in the main Nile would cease (Chapter 8). With these key facts in mind, we can now appreciate why correctly interpreting the climatic significance of alluvial deposits along the Nile is far from straightforward. A simple example will suffice to illustrate this complexity. During the very late Pleistocene, coarse gravels accumulated along the Desert Nile in Nubia (southern Egypt and northern Sudan). Butzer considered that a higher than present discharge was required to transport such gravels and so concluded that the climate in Ethiopia during LGM times was wetter than today – in effect, a glacial pluvial climate (Butzer and Hansen, 1968; Butzer, 1971b). Fairbridge (1962, 1963) reached the opposite conclusion, albeit after far less detailed fieldwork, and argued for glacial aridity in the Nile headwaters at that time, partly because that was the global model he favoured (Fairbridge, 1965, 1970), and partly because he considered that the Nile had deposited the gravels in Nubia because it lacked the hydraulic competence to transport them to the sea. How do we resolve this impasse in which essentially the same evidence provokes opposing climatic interpretations by two highly experienced observers? The simple hydro-climatic model put forward to resolve this dilemma (Adamson et al., 1980; Williams and Adamson, 1980) was developed after fieldwork by the author amidst the periglacial and glacial deposits of the Semien Highlands in Ethiopia (Williams et al., 1978) and in the semi-arid lower Blue and White Nile valleys of central Sudan (Williams and Adamson, 1973, 1974; Williams, 1975; Williams et al., 1975). During the long arid interval between ca. 33 ka and 15 ka, Lake Tana in Ethiopia and Lake Victoria in Uganda were mostly closed basins, if not dry. The White Nile became a mere trickle and virtually ceased to flow. The Blue Nile became a highly seasonal river, somewhat akin to the modern Atbara, and the Atbara flowed for only a few short months each year. In the upper catchments of the Atbara and the Blue Nile, the lower limit of periglacial processes extended about 1,000 m below its present lower limit and the timberline was depressed

200

The Desert Nile

in elevation by a similar amount (Chapter 12). Slopes were unstable, landslides and debris flows were active, and an abundant supply of coarse debris was washed down into the upper catchment rivers. The Blue Nile, Atbara and main Nile became braided aggrading streams, carrying a coarse bed load of sand and gravel as far as Nubia and, on occasion, further downstream. With the return to warmer, wetter climates towards 15–14 ka, the previously unstable slopes became forested, and soils began to develop on the weathered volcanic rocks. The Nile now became an incised and sinuous river with a suspension load of mostly siltand clay-sized particles. Lake Victoria and Lake Albert overflowed, flow resumed in the White Nile, and severe floods in Egypt heralded the new hydrologic regime. Butzer (1980, p. 267 and p. 272, Table 11.2) called this the time of the ‘wild’ Nile. We might therefore expect to find an alternation between coarse traction load deposits laid down by the Nile during cold dry intervals in NE Africa and suspension load sediments laid down along the Nile flood plain during warm wet intervals in this region. Times of weakened tropical monsoon would be reflected in enhanced dune activity and desert dust flux; times of stronger summer monsoon would see an advance to the north of the Intertropical Convergence Zone (ITCZ), resulting in a proliferation of lakes, expansion of semi-desert and tropical savanna vegetation to the north, often along active water courses on both sides of the Nile, and movement of animals and prehistoric human groups into areas previously too dry to support much life. These two contrasting scenarios represent either end of an environmental spectrum, with, needless to say, multiple variants in between.

14.3 Pleistocene Erosion and Sedimentation in Southern Egypt 14.3.1 Pleistocene Nile Terraces in Egypt Hull (1896) was among the first to study the Nile terraces in Egypt and to propose that the former discharge of the Nile was once far greater than it is today. Later studies of the alluvial terraces along the Desert Nile by K. S. Sandford and W. J. Arkell during the 1920s and 1930s demonstrated that prehistoric humans had been present in this region at least intermittently from Lower Palaeolithic times onwards (Sandford and Arkell, 1929; Sandford and Arkell, 1933; Sandford, 1934; Sandford and Arkell, 1929a, 1929b). Later work by Wendorf and his team added detail but still suffered from a lack of coherent and long-term chronology (Wendorf, 1968; Wendorf and Schild, 1976). In addition, problems arose in interpreting the Nile flood history, with De Heinzelin (1968, p. 49, Fig. 5) equating maximum Nile flood height with maximum Nile discharge, which is not appropriate in the context of an aggrading river. Salvage archaeology in the late 1960s prompted by the UNESCO efforts to move and conserve the monuments in southern Egypt that were in danger of being submerged beneath the reservoir upstream of the Aswan High Dam did result in some impressively detailed studies. For instance, Butzer and Hansen (1968) produced large-scale maps of the Kom Ombo area in Egyptian Nubia (277 × 3 km) and a comprehensive effort at interpreting the Quaternary alluvial history of the Nile and its now

14.3 Pleistocene Erosion and Sedimentation in Southern Egypt

201

defunct tributary wadis. Their interpretation was based on the lithology, sedimentary structures, fossils, palaeosols, thickness and elevation of the various deposits as well as their lateral extent and surface gradient. The main problem facing these authors was obtaining reliable ages for the different sedimentary units they had mapped. Radiocarbon ages of shells within the alluvium were almost always minimum ages because the actual ages of the shells were well beyond the range of this dating technique. Butzer (1980) summarised this work and concluded that ‘the Egyptian Nilotic record has major import for an understanding of the sub-Saharan Nile Basin. It raises more questions than it provides answers, pointing to obvious areas in Ethiopia and the Sudan that require equally searching investigation’ (Butzer, 1980, p. 277). The Pleistocene sediments are in general poorly dated but do provide a relative sequence of Nile aggradation and incision. Because the local wadi sediments inter-finger with the Nile alluvium, pointing to deposition at different seasons, Butzer and Hansen (1968) and Butzer (1980) concluded that the wadis were active in winter, and that the Red Sea Hills were under the influence of moist winter air masses. An alternative explanation might be that there was a time lag of several months between wadi flow in southern Egypt and peak Nile floods, and that both systems were dependent on the seasonal movement of the ITCZ, which would have reached southern Egypt several months after it brought rain to the upper Nile catchments. The presence of particular types of fossil soil was another line of evidence used by these authors to infer times at which the local climate in southern Egypt was arid or less arid. However, soils are often unreliable indicators of past climate (Chapter 4). There are several reasons why this is so. In the case of soils developed on alluvial sediments, the physical and chemical properties of these sediments are likely to have a stronger influence on soil attributes than local climate. Furthermore, in arid regions it is hard to distinguish between soils developed under a long interval of semi-arid climate and soils formed during a shorter interval of somewhat wetter climate. Independent evidence in the shape of pollen grains, opal phytoliths and other fossil remains is needed to confirm or rebut inferences about past climate drawn from fossil soils.

14.3.2 Late Pleistocene Ponded Drainage and Late Palaeolithic Occupation in Southern Egypt One puzzling feature of the Upper Palaeolithic record in southern Egypt is the abundance of Late Palaeolithic sites concentrated along the Nile during a time of minimal Nile flow and extreme local and regional aridity. In order to account for this seeming anomaly, Vermeersch and Van Neer (2015) proposed that desert dunes blown in from the north and west blocked certain reaches of the Desert Nile during the cold dry LGM, and, more broadly, Marine Isotope Stage 2 (MIS 2: Chapter 4) and created ephemeral lakes. Mud in suspension in the river blocked the pores between sand grains and reduced water losses from lateral seepage. The abundance of Tilapia among the

202

The Desert Nile

fish bones is consistent with the existence of relatively deep lakes with a life span of decades or centuries. The Late Palaeolithic occupation sites occur at varying elevations above the Nile, as one might expect from dune-dammed lakes. With the return of a strong summer monsoon ca. 14,500 years ago bringing renewed high summer floods in the White Nile, Blue Nile and Atbara, the dune dams were breached and the lakes ceased to exist. This model seems eminently logical and now requires testing by obtaining optical ages for the dune sands.

14.3.3 The ‘Wild’ Nile A phenomenon noted by Butzer (1980, p. 272, Table 11.2) was the ‘wild’ Nile – an interval of extreme Nile floods between ca. 12,000 and 11,500 14C years BP, which is equivalent to a calibrated age of 13.5–14 ka. The Blue Nile and White Nile Valleys in central Sudan contain a well-dated record of extensive flooding at and slightly before this time, consistent with the return of the summer monsoon and renewed overflow from Lake Tana in Ethiopia and Lakes Victoria and Albert in Uganda (Chapters 6, 7, 8 and 11). The ‘wild’ Nile episode thus marks the onset of a wetter climate in the Nile headwaters after the long dry intertropical Late Pleistocene climatic interval culminating in the cold and arid LGM.

14.4 Late Quaternary Depositional Environments in Northern Sudan The advent of luminescence dating and its widespread use in the last two decades has revolutionised our approach to reconstructing past environmental changes in the Nile Basin. Thermo-luminescence (TL) dating was initially used to determine the time when archaeological materials such as burnt clay, hearthstones and pottery had been subject to high temperatures, but was also used more widely to date grains of quartz within alluvial and eolian sediments. Optically Stimulated Luminescence (OSL) dating (or optical dating: Huntley et al., 1985) has become the preferred method of obtaining accurate and reasonably precise ages for particles of quartz and feldspar within a wide range of sediments and soils. Serious workers now try to use both radiocarbon and optical dating methods to determine the ages of individual sedimentary units, supplemented where appropriate by cosmogenic exposure dating (Williams, 2014, chapter 6). At the present time, the late Quaternary depositional record in northern Sudan is probably better dated and therefore better understood than that in southern and central Egypt. The main reason for this is that in Egypt much of the archaeological work has very understandably been concentrated on the last 5,000 or so years with their rich archival and monumental records (Trigger, 1982; Toonen et al., 2017). To discuss these latter records in any detail is beyond the scope of this book and would only duplicate comprehensive accounts that are already widely available (Shinnie, 1967; Edwards, 2004; Welsby and Phillipson, 2008).

14.4 Late Quaternary Depositional Environments in Northern Sudan

203

14.4.1 Late Quaternary Environments West of the Desert Nile in Northern Sudan Although the late Quaternary history of the Nile Valley area immediately west of the Desert Nile in northern Sudan remains poorly understood, a slowly increasing body of evidence is adding to our knowledge of past environmental changes in this region. Optical ages obtained on late Pleistocene Nile alluvium have revealed that the Nile was shifting laterally and actively aggrading at 145 ± 20 ka, 83 ± 24 ka, 32 ± 8 ka and 20.7 ± 0.2 ka (Williams et al., 2010). The 32 ± 8 ka phase of active aggradation coincides with the recently recognised interval of very high energy flow in the White Nile at 27.8 ± 3.2 ka and with alluvial fan aggradation near Jebelein on the White Nile right bank (Chapter 8). Holocene terraces and former Nile channels on either side of the present river near Dongola have 14 C and optical ages of 11 ka, 6.5–5.0 ka and 4.8–4.0 ka, after which flood discharge diminished (Williams et al., 2010). A lake in the Qaab Basin west of the Nile was fed from an overflow channel from the Nile between 9.5 ka and 7.5 ka and occupied an area up to 450 km2 (Williams et al., 2010). This entire area is now very arid and the Qaab Basin lake marls and diatomites that overlie Nubian Sandstone at shallow depth are being actively eroded by the wind. We are often inclined to think that during wetter intervals dunes were vegetated and stable, while during times of drier local climate the plant cover died and the dunes became mobile once more. However, the reality is more complex. For example, the Qoz Abu Dulu west of the Nile was briefly active during 9.9 ± 2.0 ka and 9.0 ± 2.8 ka (see Chapter 8), which appears to have been a time of wetter climate and high Nile flow. There may have been brief intervals of intense drought causing sand to move, or, perhaps more likely, the proximal edge of the dune received seasonal supplies of sand blown up from the channel in winter, and so operated as a source-bordering dune (see Chapter 11). The sand sheets west of the Nile near the 4th Cataract were active at 6.6 ± 0.9 ka, 4.8 ± 0.9 ka and 2.9 ± 0.5 ka (Williams et al., 2015a), which, within the dating error limits, coincide with times when Lake Challa was low (Verschuren et al., 2009; and Chapter 7), and with phases of channel and flood plain contraction identified by Macklin et al. (2015) and discussed in Section 14.4.

14.4.2 Late Quaternary Environments East of the Desert Nile in Northern Sudan (a) Prehistoric Environments in the Kerma Area The Kerma area in northern Sudan is one of the hottest and driest regions in the Sudan (Sudan Meteorological Service Climatological Normals, 1931–1950; Griffiths and Soliman, 1972). Annual precipitation is only a few millimetres, with a high degree of variability from year to year, and is associated with the northern limit of the advancing ITCZ during summer (July, August, September). Summer temperatures often exceed 40°C during the day and can fall to near freezing during winter nights. Winds are almost invariably from the north and dunes are active on either side of the Nile.

204

The Desert Nile

The Kerma alluvial plain lies south of the 2nd Cataract and is bounded by the Desert Nile to the west and the horizontally bedded Mesozoic Nubian Sandstone plateau to the east (Williams, 2012; Honegger and Williams, 2015). The plain is up to 15 km wide and is traversed by three former Nile channels that are clearly visible on satellite imagery but less easily discerned on the ground owing to disturbance from irrigation canals and cultivation. The Nubian Sandstone outcrops are in places capped by a bed of highly resistant iron- and silica-cemented sandstone. This hard rock was often used to make stone tools. Sporadic plugs of Cenozoic basalt cut through the sandstone and form steep isolated hills up to 150 m high that were used as lookouts by Mesolithic hunters. The sandstone next to one such basalt plug was tilted 15–20° from its initial horizontal state as a result of the drag exerted as the plug pushed through the sandstone. The plateau rises 50–100 m above the alluvial plain and has a crenulated and dissected western margin from which emerge a series of ephemeral stream channels. On leaving the plateau the channels radiate out westwards as shallow distributary channels and vanish on reaching the eastern edge of the Nile alluvial plain. Two gently sloping erosional surfaces or pediments are cut across bedrock at the base of the plateau and are covered by a nearly continuous desert pavement cover. Desert pavements are ‘armoured surfaces composed of angular or rounded fragments, usually one or two stones thick, set on or in matrices of finer material comprising varying mixtures of sand, silt or clay’ (Cooke et al. 1993, p. 68). Winnowing of finer particles from the surface by wind or running water can lead to the formation of a protective layer of surface stones. In this area, deflation by wind is the dominant agent of desert pavement formation. The lower pediment surface is overlain by a vesicular fossil soil that is in turn capped by one or more hardpan units with a protective cover of desert pavement at the surface. The lower pediment shows evidence of an episodic influx of calcareous eolian dust, followed by leaching and re-precipitation of carbonate at the base of the soil within cracks and cavities in the weathered Nubian Sandstone. During intense downpours following dust storms today dust is scavenged from the atmosphere and deposited on the present surface. In January 2012, silt drapes 6–8 mm thick were plastered across the vertical east, south and west faces of one dry well site and had probably been washed in from the surface during a severe rainstorm following a haboob or major dust storm the previous year. Dust storms occur in this area at least once a year. The vesicular fossil soil mantling the lower pediment shows abundant signs of former termite activity in the form of galleries (Lee and Wood, 1971; Paton et al., 1995), denoting a previously wetter period when the grass cover was at least seasonally abundant. In places this fossil soil has 10–20% brown ferruginous mottles and up to 3% manganese nodules 2–8 mm in size, indicative of seasonal waterlogging and of slightly acidic conditions during soil formation. Soils subjected to termite activity are hard to date (Pillans et al., 1997; Johnson et al., 2014). Two samples collected at the same depth (35 cm) from two separate fossil termite sites on the lower pediment had OSL ages between ca. 55 and 52 ka. In addition, there is no evidence of any bleaching of the quartz silt grains in these two samples after about 6.5 ka (Frances Williams, unpublished report, 2012). It is possible that

14.4 Late Quaternary Depositional Environments in Northern Sudan

205

Figure 14.2 View of a part of the Wadi El-Arab archaeological section, January 2012, Kerma area, northern Sudan. The protective hardpan layer at the surface has protected the Neolithic artefacts within the sand unit from subsequent erosion. (Based on the author’s field notes.)

soil formation was active during a moister climatic interval at 55–50 ka and resumed in the wetter early Holocene, ceasing when aridity was increasing in this area after 6.5 ka. The Blue Nile was depositing alluvial clays in the northern Gezira at ca. 55–50 ka (Williams et al., 2015a and Chapter 11). In the eastern Mediterranean Sapropel 2 has an age of ca. 55 ka and reflects a time of high Nile discharge into the sea (Williams et al., 2015a). Thin layers of silty clay and fine sandy clay form relatively hard platy layers up to 20–30 cm thick on the surface of the underlying fossil soil and sandy colluvial-alluvial pediment mantle, indicating entrainment and deposition of wind-blown dust by surface runoff. Once deposited these platy layers impede infiltration, become hard and protect the underlying coarser sediments from wind erosion. The archaeological site of Wadi El-Arab east of Kerma owes its preservation to a hardpan layer (Fig. 14.2), and was occupied at least intermittently for 3,000 years between 10.3–9.96 ka and 7.56–7.38 ka (Honegger and Williams, 2015). Following the formation of the termite soil and its overlying hardpan, the previously vegetated and stable dunes and sand sheets in this area became active once the present arid climate set in. Strong northerly winds transported and deposited a surface layer of coarse sand and fine granules. A protective layer of platy gravel formed a desert pavement over the upper slopes of both pediment surfaces. When disturbed this can reform and cover 80% of the surface within 6 years in this area. The oldest Nile alluvial sediments seen in this area are planar- and cross-bedded gravels and coarse sandy gravels. Such gravels are of late Pleistocene age or older and are often visible only because they are being quarried. In a gravel quarry located at 19°38.8 0 N, 30° 29.7 0 E, the lowest unit exposed in January 2012 yielded two OSL ages of ca. 50 ka at depths of –5 m and –4 m, while the overlying unit yielded an OSL age of ca. 75 ka at depth –2 m (Frances Williams, unpublished report, 2012). These are approximate ages only and the reversed stratigraphic order of the ages prompts caution. It is highly likely that the sandy matrix of the gravels was not fully bleached, so that the three ages are likely to be minimum

206

The Desert Nile

ages only. The gravels consist of 90–95% sub-rounded quartz pebbles 1–5 cm in diameter in a matrix of coarse sand and fine granules. Occasional rolled pebbles of chert, agate and chalcedony occur within the gravels, together with rolled Middle Palaeolithic flakes. The gravels represent the bed load of the Nile during a late Pleistocene phase of very highenergy channel flow and likely highly seasonal discharge. A weakly developed cracking clay soil overlies the gravels and is probably much younger. In addition to the gravels, two other broad categories of alluvial sediment were identified based on their grain size, mineral composition, sedimentary structures, and associated fossils, all of which reflect the processes and environment of deposition. The most widespread Late Pleistocene and Holocene sediments on the Kerma alluvial plain are horizontal beds of silty clay or very fine sandy clay that are usually separated by thin beds of fluvial sand but sometimes occur as vertically stacked beds of clay each up to a metre thick, with occasional fine laminations visible within the beds. Occasional freshwater gastropod shells and Nile mussel shells occur within the clays. Cracking clay soils (‘vertisols’) have formed on the clay beds, with carbonate concretions or diffuse carbonate in the lower part of the profile – a characteristic feature of vertisols in the Gezira plains between the lower Blue and White Nile (Chapters 4 and 11). The horizontal clay units originated as flood plain deposits laid down by the Nile during the summer flood season, a process clearly recognised by Herodotus more than 2,500 years ago in Egypt. The clays thin out eastwards. The eastern margin of the late Quaternary Nile flood plain consists of a veneer of alluvial clays overlying sheet flood sands and gravels from the Nubian Sandstone plateau with sandstone bedrock at shallow depth. The next most common Late Pleistocene and Holocene Nile alluvial sediments in this area are horizontal beds of fine, medium and coarse sand, showing planar and cross-bedding sedimentary structures. Bed thickness varies from a few centimetres to several metres. In a few cases, bands of sand a few centimetres thick alternate with clay-rich bands of similar thickness. The fluvial sands are bed load channel sediments or crevasse splay deposits laid down during exceptional floods. The alternating fine sand and clayey units appear to be levee deposits that were laid down adjacent to former channels of the Nile. Using a combination of 14C and OSL analyses, the sediments on the alluvial plain proved easier to date than the Pleistocene gravels and the fossil termite soil. Intervals of fluvial sand deposition from the Nile and its anabranches range from late Pleistocene (86 ± 11, 66 ± 5, 64 ± 8 and 63 ± 7 ka) (Frances Williams, unpublished report, 2012) to early-mid Holocene (10.2 ± 0.4, 7.9 ± 3, 7.6 ± 0.3). Clay deposits were laid down across the Nile flood plain at 11.5, 10.2, 10.0, 9.5 and 8.3 ka. These clays later underwent moderate pedogenesis to form vertisols. The sand deposition phase centred on 66–63 ka may have been quite widespread. Intervals of high Holocene Nile flow in this area have ages of 11.5, 10.2, 10.0, 9.5, 8.3, 7.9 and 7.6 ka (Honegger and Williams, 2015). We are now in a position to compare these environmental changes to the changing pattern of human occupation in this area. After a long dry and cold interval in the very late Pleistocene the climate became warmer and wetter and Mesolithic hunter-gatherer groups began to move into the Sahara from about 10.5 ka onwards (see Chapters 15 and 16). Conditions remained sufficiently attractive in the

14.4 Late Quaternary Depositional Environments in Northern Sudan

207

Sahara and desert areas east of the Nile to enable Neolithic pastoralists to tend their flocks of cattle, later supplemented by flocks of sheep and goats, until about 7.3 ka, after which progressive desiccation, especially after ca. 5.5–4.5 ka, forced them to move east to the Nile valley or south into what are now the savanna grasslands bordering the southern Sahara. Solidly based upon 90 14C ages, the Holocene human occupation of the area around Kerma is now known in some detail (Honegger and Williams, 2015). The earliest occupation sites near Kerma date back to 10.3 ka and were used by Epipalaeolithic/Mesolithic and early Neolithic groups who appear to have avoided the alluvial plain and settled along the plateau margin and on small erosional outliers of Nubian Sandstone. Pottery is present by 8.3 ka, making it the oldest so far known in northern Sudan. The Neolithic in this area dates back to 8 ka. After ca. 7.3 ka, people began to settle on the alluvial plain, although their settlements were sometimes subject to extreme Nile floods, shown by successive Neolithic occupation layers during 6.9–6.3 ka separated by Nile flood silts. An early Neolithic cemetery at one site was abandoned at 7.5 ka, and no sites are evident during the ensuing 400 years between 7.5 ka and 7.1 ka, after which there is a proliferation of Neolithic sites until another sharp drop in occupation between 6 ka and 5.4 ka. The two gaps in Neolithic occupation in the Kerma area are broadly coeval with the two lake regressions L4 (8.0–6.7 ka) and L3 (5.9–4.7 ka) in Lake Challa near the White Nile headwaters (Verschuren et al., 2009; and Chapter 7). The post-Neolithic Kerma Period in this area lasted from 4.4 ka to 3.45 ka (Table 14.1). The area fell under the control of New Kingdom Egypt between 3.5 ka and 3.07 ka. The Nile flood history during this time is discussed in the next section. (b) Holocene Floods East of Kawa, Northern Sudan The area immediately east of Kawa close to the North Dongola reach of the Desert Nile has been the focus of intensive geo-archaeological enquiry and has provided well-dated evidence of changing patterns of occupation linked to local channel changes during the Holocene (Macklin et al., 2013; Woodward et al., 2015). Satellite imagery shows very clearly that a major branch of the Nile bifurcated from the main Nile at El-Ugal township at about latitude 18°40 0 N, flowed NNE for about 7 km and then split into two channels or anabranches that flowed north on the Nile alluvial plain bounded to the east by the western escarpment of the Nubian Sandstone plateau. The most easterly channel is termed the Alfreda Nile; its more westerly partner is the Hawawiya Nile. Both former channels joined roughly 40 km to the north to form the Seleim Nile, named after the Seleim Basin located between the Dongola Nile (i.e., the present-day Desert Nile) and the Seleim Nile (Fig. 14.3). Table 14.1 shows the ages of the main archaeological periods identified in northern Sudan. Neolithic occupation sites, with ages between >7 ka and 5.5 ka, were clustered along the Hawawiya, Alfreda and Seleim Nile channels, with some sites concentrated locally along either side of the Dongola Nile. OSL ages of fluvial sands show that all of the former channels and the Dongola Nile were active at this time. A substantial amount of land was therefore available for cultivation between the high and low flow

208

The Desert Nile

Table 14.1 Main archaeological periods in northern Sudan Cultural period

Time period (ka) [BC]

Neolithic Pre-Kerma Kerma New Kingdom control Kushite Medieval

>7.0 to 5.5 [>5,500 to 3,500] 5.5 to 4.4 [3,500 to 2,400] 4.4 to 3.45 [2,400 to 1,450] 3.5 to 3.07 [1,500 to 1,070] Ninth century BC to fourth century AD Fourth century AD to fifteenth century AD

Adapted from Woodward et al. (2015a).

(a)

20°N

Second Cataract

30°N

40°N

(b) CAIRO

30°N

20°N

Seleim Nile

Seleim

N 21ºN

KHARTOUM

Basin

10°N

N

Measured section

Dongola

0°N 0 oN 0

250

500

Nile

km

Kawa

20ºN

Third Cataract

Cataract.

Kerma

0 Kawa

Northern Dongola Reach (Fig. 14.8b)

19ºN

km

50

Dong

Fourth Cataract

ola N

El-Ugal

Ni

le

ile

Mulwad

Nile

Alfreda Nile

Dongola

Hawawiya Nile

Kerma sites between the second and fourth cataracts of the Nile.

0

18ºN

5 km

31ºE

32ºE

El-Ugal

Figure 14.3 (a) Holocene occupation sites between the 2nd and 4th Cataracts. (b) Former active Nile channels east of Kawa, northern Sudan. (Adapted from Macklin et al., 2013, Fig. 1; Woodward et al., 2015a, Fig. 12; and Williams et al., 2015a, Fig. 6.)

14.4 Late Quaternary Depositional Environments in Northern Sudan

209

stages without need for irrigation (Welsby et al., 2002; Macklin et al., 2013). Charles Bonnet has for many years investigated sites belonging to the Kerma Period (4.4–3.45 ka) and has defined three Kerma cultural phases (Bonnet, 1992). The three Kerma phases are the Kerma Ancien (4.4–4.05 ka), the Kerma Moyen (4.05–3.75 ka) and the Kerma Classique (3.75–3.45 ka). During the Kerma Ancien, sites are distributed along all three channel belts in roughly similar numbers, but during the Kerma Moyen and Kerma Classique times, people moved from the Hawawiya channel belt to the Seleim and Alfreda Nile channel sectors, resulting in a doubling of the number of sites from forty to ninety. No sites younger than 3.45 ka occur along the Hawawiya Nile channel, which had dried out by ca. 3.99 ± 0.15 ka. It seems very probable that this desiccation event equates to a period of extreme drought centred on 4.2 ka and evident in the Nile Delta (Stanley et al., 2003) as well as in Lake Albert (Williams et al., 2006). The 4.2 ka interval of very low Nile flow may have precipitated the demise of the Old Kingdom in Egypt (Bell, 1971; Hassan, 1997; Stanley et al., 2003). Notwithstanding the drought and sudden change in Nile flow, the resilient Kerma farmers moved to the eastern edge of the alluvial plain and occupied levees along the Alfreda and Seleim channels. Invasion from Egypt during the 18th Dynasty (ca. 3.5 ka) was followed by renewed regional drought that led to an influx of wind-blown sand and the filling of two brick-lined wells that had been dug into the ancient Hawawiya channel. The eolian sands are well bleached and have optical ages of 3.17 ± 0.12 ka and 3.44 ± 0.14 ka at the two well sites (Macklin et al., 2013). The drying out of the Alfreda Nile occurred shortly before 3.29 ka and may have been accomplished in less than a century. The combined effects of invasion and drought proved catastrophic and farmers abandoned much of this area. This exodus marked the end of the once flourishing Kerma culture. There were later floods in the Alfreda Nile between 780 BC and 730 BC, with an interval of very high floods during early Kushite times (Table 14.1). The Alfreda Nile last flowed in AD 280. As Macklin et al. (2013, p. 695) have pointed out, ‘the dynamics of the local alluvial environment were critical in determining whether climatic fluctuations and changes to river flow represented an opportunity for floodwater farmers (5000–3500 B.C.), a hazard that could be managed (2400–1300 B.C.), or an environmental catastrophe that resulted in settlement abandonment (after 1300 B.C.)’. Woodward et al. (2015a) have mapped and dated alluvial and eolian sediments along the North Dongola Reach of the Desert Nile. They found that the isotopic composition of the Nile flood plain sediments in this sector has varied throughout the Holocene in response to major changes in sediment sources and regional climate. During the Early Holocene, local wadis connected to the main Nile provided up to and sometimes in excess of half the sediment load transported by the Nile north of Khartoum. Detailed analysis of the strontium and neodymium isotopic composition of alluvial and eolian sediments east of the Desert Nile in northern Sudan has revealed that local wadis were active during the Early and Middle Holocene, while reworked eolian sand and dust became an increasingly important component of the Nile alluvial load during the Late Holocene.

210

The Desert Nile

14.5 Meta-analysis of the Desert Nile Holocene Fluvial Archive In an effort to understand how the Holocene Nile has responded to major climatic changes in its headwaters, Macklin et al. (2015) completed a meta-analysis of 115 fluvial units in the Nile Basin, 66 of which have 14C ages and 49 have OSL ages. Along the valley of the Desert Nile in both Sudan and Egypt, 37 fluvial units have 14C ages and 46 have OSL ages. The fluvial units comprised flood plain clays and silts and sands from former Nile channels. The ages obtained for the channel sands were several centuries younger than the closest ages obtained for the flood plain clays. Comparison with well dated lacustrine sediment records from the upper Blue and White Nile catchments supported the hypothesis that the channel ages denoted times of flood plain contraction, reduced Nile flow and drier conditions upstream, while the flood plain ages coincided with wetter intervals, increased Nile flow, and widespread overbank flooding. Times of inferred Holocene channel and flood plain contraction along the Desert Nile have calibrated ages of 8.15–7.75, 6.4–6.15, 5.7–5.45, 4.7–4.55, 3.35–2.9, 2.8–2.55 ka and AD 1600. The pattern of events reconstructed using this approach should not be regarded as definitive, but rather as a working hypothesis in need of testing from detailed local studies.

14.6 Conclusion The Quaternary alluvial record of the Desert Nile is one of repeated aggradation and incision, with sporadic occupation by humans from Lower Palaeolithic/Early Stone Age times onwards. Much of this record is fragmentary and poorly dated. The most recent welldated phase of relatively high Nile flows and wetter climatic conditions in the desert regions adjoining the Nile was between ca. 15–14 ka and 5–4 ka and is sandwiched between two much drier intervals marked by a southward shift of the ITCZ, a reduction in summer monsoonal precipitation, and reactivation of desert dunes. It was also interspersed with intervals of severe drought and reduced Nile flow, which led to people abandoning areas previously occupied in Nubia, to which they returned when conditions improved. A catastrophic drought at the start of the Late Holocene 4,200 years ago may have contributed to the demise of the Old Kingdom in Egypt, although further south in Sudanese Nubia the farmers adapted to the changed conditions and flourished at a time of famine and social break-down in Egypt. Disentangling cause and effect in relation to changing prehistoric land use in the lower Nile Valley is often hard and can lead to uncritical appeals to environmental determinism. The geomorphic and human occupation history of each local area need to be thoroughly investigated before credible lessons can be drawn in regard to human responses to past environmental, social, economic, political and military events.

15 West of the Nile: The Western Desert of Egypt and the Eastern Sahara – Part 1

Level plains of smooth sand are interrupted only by occasional peaks of rock – black, stark, and shapeless. Rainless storms dance tirelessly over the hot, crisp surface of the ground. The fine sand, driven by the wind, gathers into deep drifts, and silts among the dark rocks of the hills, exactly as snow hangs about an Alpine summit; only it is a fiery snow, such as might fall in hell. Winston S. Churchill (1874–1965), The River War (1899)

15.1 Introduction The eastern Sahara is one of the driest and hottest regions on our planet. The Western Desert of Egypt is located west of the Nile within the hyperarid zone of the eastern Sahara (Fig. 15.1). This is an area with a notional mean annual precipitation of less than 5 mm and intervals of half a century or more with no recorded rainfall. Against this present-day background of extreme heat in summer and unremitting aridity, we are confronted with a paradox. Scattered across the eastern Sahara Desert there are abundant signs that this hyperarid region was once vegetated and capable of supporting a tropical savanna flora and fauna as well as occupation by successive prehistoric human groups from Lower Palaeolithic/Early Stone Age times onwards. How was it possible that the wilderness we see today was once a green and pleasant land capable of sustaining herds of domestic cattle as recently as 5,000 years ago? In addition to the present desert being clothed in trees, shrubs and grasslands at various times during the Quaternary, major rivers once flowed across the Sahara to the Mediterranean or to the Nile. The early to mid-Holocene lakes and once active but now ephemeral streams west of the Nile in northern Sudan and western Egypt became the locus of Mesolithic and Neolithic occupation and saw the progressive transition from hunter-fisher-gatherer economies to ones founded on animal and plant domestication. The word ‘Sahara’ in Arabic denotes a wilderness to be traversed as rapidly as possible, just as the word ‘Arab’ in colloquial Sudanese Arabic today denotes a wanderer, vagrant or nomad. Despite these pejorative

211

212

Western Desert of Egypt and Eastern Sahara – Part 1

Figure 15.1 The Western Desert of Egypt.

modern connotations, we can learn a great deal from the rich environmental history and archaeological legacy bequeathed to us by the presently hyperarid eastern Sahara.

15.2 Early Exploration The great pioneering journeys to the west of the Desert Nile in the eastern Sahara and across the Libyan Desert in the 1920s and 1930s by Hassanein Bey (1924, 1925), Prince Kemal el Din Hussein (Kemal el Din Hussein, 1928) and Ralph Bagnold and his travel companions (Bagnold, 1933,1935, 1990) were successful in revealing the potential environmental and archaeological significance of this vast and now hyperarid region (see also Chapter 16). For Bagnold and his fellow explorers, what they called the ‘Libyan Desert’ also included the desert areas of northern Sudan (Sandford, 1933a, 1933b, 1933c, 1935, 1936; Shaw, 1933, 1936; Peel and Bagnold, 1939). Douglas Newbold, who joined Bagnold on one of his expeditions (1930), had earlier completed a 1,600-km (1,000 miles) odyssey by camel across northern Sudan in late 1923, a feat he replicated in late 1927 accompanied by the botanist and amateur archaeologist W. B. K. Shaw (Newbold, 1928; Newbold and Shaw, 1928). Newbold spoke excellent Arabic, had spent many years working in and travelling across northern Sudan, knew the camel-herding tribes in those regions very well, and had a keen interest in archaeology. He visited and put on record a number of hitherto littleknown localities, including Wadi Howar and the Selima sand sea (Figs. 15.2 and

15.3 Wadi Howar and Adjacent Areas

213

15.3), meticulously noting the occurrences of prehistoric stone artefacts observed in the course of his journeys. The exploratory surveys of Newbold, Bagnold and A. J. Arkell (1961) in northern Sudan also helped to focus attention on the now defunct drainage systems west of the Nile, of which Wadi Howar, Wad el Melik and Wadi Muqadam (Fig. 15.2) are perhaps the best known (Arkell, 1948; Barbour, 1961; Whiteman, 1971). For Arkell (1948) Wadi Howar and Wad el Melik were a long established albeit intermittent connecting link between the Sahara and the Nile Valley, facilitating exchange of ideas and a two-way trade. Bagnold (1935, p. 254) was less impressed with what he saw of Wadi Howar, as the following quotation plainly shows: ‘In these latitudes a vague hollow called the Wadi Howar, whose precise nature is not yet understood, emerging from the western border hills north of Kutum (Fig. 15.2), runs northeast for 300 miles [480 km] across the open desert till it is finally lost in the sands immediately south of Bir Natrun Oasis. Were it not for a broad belt of low trees which marks its bed (in which no water has ever been known to run), the wadi would be passed unnoticed, for as a valley it is so shallow as to be almost imperceptible.’ These comments could equally apply to certain sections of Wadi el Melik and Wadi Muqadam today. However, Bagnold did observe that: ‘All along the banks [of Wadi Howar], where the rolling sand dunes began on each side, the ground was strewn with stone implements – grinding stones and an occasional beautifully polished diorite axe, all of unknown date, together with bits of pottery which may possibly be identified with that of the old Ethiopian kingdom of Meroe of 2,000 years ago’ (Bagnold, 1935, pp. 254–255). The Soil Conservation Committee of Sudan, in its major post–World War II review, drew attention once more to the widespread but scattered archaeological and fossil evidence, especially the subfossil shells of aquatic mollusca, indicating wetter Holocene climates in northern Sudan and the southern Libyan Desert (Whyte, 1951). Sandford (1936) had also reported on the sporadic occurrences of fossil freshwater and land snail shells in what he termed the South Libyan Desert, and had drawn similar conclusions.

15.3 Wadi Howar and Adjacent Areas Seventy years after the pioneering work of Bagnold and his colleagues, Wadi Howar and its adjoining region have been intensively investigated by well-equipped teams of geomorphologists, geochemists and archaeologists (Pachur and Kröpelin, 1987; Kuper, 1989; Richter, 1989; Pachur et al., 1990; Kröpelin, 1993a, 1993b; Abell et al., 1996; Abell and Hoelzmann, 2000; Hoelzmann et al., 2000, 2001, 2004; Pachur and Hoelzmann, 2000; Rodrigues et al., 2000; Kuper and Kröpelin, 2006).

15.3.1 Wadi Howar Wadi Howar is more than 1,200 km long and was once a major tributary of the Nile (Fig. 15.2). Its former headwaters flowed from the eastern flanks of Jebel Marra volcano

Western Desert of Egypt and Eastern Sahara – Part 1 Jebel Uweinat

SUDAN

Selima

ile

LIBYA

N

N

214

Laqiya

20°N Oyo

Dongola

Debba

Kutum Jebel Marra

Gabolab

i Muqadam

i W ad

Jebel Tageru

Merowe

el ad W

0

km

200

Ni

le

KHARTOUM 15°N

Nile

H

Ennedi Plateau

El Atrun Jebel Rahib owar

Wa d

Nubian Palaeolake

Jebel Abiyad

lik

West

Me

CHAD

El Fasher 25°E

30°E

Figure 15.2 Wadi Howar, Wad el Melik, Wadi Muqadam and the West Nubian Palaeolake, northern Sudan.

(3,042 m) in NW Sudan (see Chapter 13) and the SE margins of the Ennedi sandstone plateau (1,311 m), with its spectacular sandstone natural arches, in NW Chad. During wetter times in the past, Wadi Howar used to flow eastwards below its emergence from the uplands for roughly 640 km along the southern margin of the Sahara to reach the Nile opposite Old Dongola (Fig. 15.3), the capital of the ca. AD 550 Early Christian kingdom of Makouria (Arkell, 1961) about 35 km north of the former confluence of Wad el Melik and the Nile near the present town of Debba (Fig. 15.2). The lower reaches of Wadi Howar between Jebel Rahib (Fig. 15.2) and the Nile are today mantled by wind-blown sand and are quite hard to identify on the ground apart from the acacia trees aligned along the former channel with their roots reaching down to the shallow groundwater table located about 5 m beneath the present ground surface. A roughly rectangular dry-stone fort or compound about 150 m wide, with walls more than 5 m thick near the base is located roughly 150 km west of the Nile on the south bank of Wadi Howar in what is today a desert (Pachur and Kröpelin, 1987, footnote 5). The former inhabitants of this fort presumably obtained their water either from wells fed by shallow groundwater and/or from Wadi Howar itself at a time when it was flowing at least seasonally some 1,500 years ago. Pachur and Kröpelin (1987) reviewed evidence from alluvial sediments, vertebrate and invertebrate fossils (freshwater mollusca and ostracods) and archaeological remains in the Wadi Howar drainage basin and concluded that between 9.5 and 4.5 ka the lower reaches of

15.3 Wadi Howar and Adjacent Areas

215

Figure 15.3 The ruins of Old Dongola (AD fifth century) on the Nile right bank opposite its former confluence with Wadi Howar.

the river flowed through a region with numerous groundwater outlets and freshwater lakes. The savanna fauna present at this time included buffalo (Bubalus bubalis), elephant (Loxodonta sp.), giraffe (Giraffa camelopardalis), wild ass (Equus africanus), black rhinoceros (?Diceros bicornis), warthog (Phacochoerus aethiopicus), roan antelope (Hippotragus equinus), antelope (Tragelaphus sp.), dorcas gazelle (Gazella dorcas), zebra (?Hippotigris quagga), hartebeest (Alcelaphus bruselaphus) and ostrich (Struthio camelus). The remains of Addax (Addax nasomaculatus), which is well adapted to very dry environments, suggest that the climate may not have remained uniformly wet throughout this time but may have witnessed episodes of severe drought. Also present were fossil remains of crocodile (Crocodylus sp.) and hippo (Hippopotamus amphibius), consistent with a former drainage link to the Nile. The subfossil freshwater gastropods are typical of those found in the Nile during the early Holocene (see Chapters 8, 11 and 14) as well as today, and include Biomphalaria pfeifferi, Bulinus truncatus, Cleopatra bulimoides, Lymnaea natalensis and Melanoides tuberculata. The presence of these gastropods again points to a link with the main Nile. The semi-aquatic snails Pila wernei and Lanistes carinatus were also present, although whether they showed an increase over time (indicating the onset of desiccation) is not indicated. The Neolithic sites contained remains of

216

Western Desert of Egypt and Eastern Sahara – Part 1

domestic cattle (Bos primigenius), goats (Capra aegragus) and sheep (Ovis ammon), which we discuss further in Section 15.3.2. This region presently receives about 25 mm of annual rainfall, although the amount is highly variable from year to year. Pachur and Kröpelin (1987) argued that to replenish the groundwater in the lower Wadi Howar would have required a substantial increase in local precipitation (rather than simply being fed by rainfall in the headwaters), equivalent to a northward shift of the isohyets by about 500 km. They noted that such a shift was supported by evidence from now dry early Holocene lakes in Chad and Mali. Certainly, the abundant evidence from former Holocene lakes, active drainage networks and associated Mesolithic and Neolithic occupation sites across the southern and central Sahara is strongly indicative of a northward shift of the Intertropical Convergence Zone (ITCZ: see Chapter 3) by at least 500 km further to the north in summer, coupled to a stronger summer monsoon (H. J. Hugot, 1962; Faure et al., 1963; Faure, 1966, 1969; Servant et al., 1969; Clark et al., 1973; Servant, 1973; G. Hugot, 1977; Rognon and Williams, 1977; Servant and Servant-Vildary, 1980; Petit-Maire and Riser, 1983; Fontes et al., 1985; Haynes, 1987; Petit-Maire, 1991; Grove, 1993; Hassan, 1997; Guo et al., 2000; Gasse, 2002b; Drake and Bristow, 2006; Bubenzer and Riemer, 2007; Armitage et al., 2007; Williams, 2008). It is also possible that positive feedbacks between the local precipitation, lake evaporation and plant cover could have enhanced the effects of the increase in local rainfall (Hoelzmann et al., 1998; Claussen et al., 1998, 1999; Bonfils et al., 2001; Braconnot et al., 2001; Carrington et al., 2001). This was the last time that there was any significant recharge of shallow ground water aquifers in this vast region. The use of this water for irrigation today is, quite simply, mining a non-renewable resource. Additional evidence relating to the pattern and tempo of early Holocene precipitation in Wadi Howar comes from stable isotope analysis of lake carbonates and sub-fossil gastropod and Nile oyster shells (Abel and Hoelzmann, 2000; Rodrigues et al., 2000). The Nile oyster (Etheria elliptica) shells are of interest for several reasons. The oysters need clear, well oxygenated and preferably fast-flowing water and a rocky substrate. At a site in the lower Wadi Howar located 50 km upstream of its junction with the Nile, the oyster shells have a radiocarbon age of 6,835 ± 110 and 6,635 ± 105 14C years BP and a calibrated age of 7.6 ka (Rodrigues et al., 2000). The oxygen isotopes show two intervals of increased negative oxygen isotope ratios during about one year of growth, which is consistent with two wet seasons during the life of the oyster, one about 4–5 months long and the second just a few months in duration. If the overall interpretation proves to be correct (and only additional work can confirm it), it would indicate that Wadi Howar was ‘a major freshwater stream with continuous flow for some periods of time during the early to mid-Holocene, with water derived largely from two rainy seasons, but with very minor post-precipitational alteration of the isotope ratios’ (Rodrigues et al., 2000, p. 186). It is not immediately clear from this work how one might distinguish between two wet seasons in successive years and two wet seasons within the same year. A very different picture emerges when we examine the oxygen isotopic composition of subfossil gastropod shells and lake marls from sites to the north of Wadi Howar and from

15.3 Wadi Howar and Adjacent Areas

217

two lake sites located along its lower reaches (Abell and Hoelzmann, 2000). The most northerly site is at Dry Selima, located a few kilometres south of Selima Oasis, which we discuss further in Section 15.4.3. This site lies at what was most probably the northern fringe of the ITCZ during the wettest period of the Holocene. The ∂18O values of groundwater north of this latitude (21°22 0 N) indicate recharge from moist air masses transported from west to east across the continent from the Atlantic (Sultan et al., 1997; and Chapter 16). South of this latitude the groundwater was replenished from precipitation of tropical convective origin associated with the seasonal migrations of the ITCZ (Joseph et al., 1992; and Chapter 16). The lake sediments at Dry Selima are bracketed between 8,820 ± 80 and 5,570 ± 60 14C years BP. Both the bulk carbonate and the gastropod shells have generally positive oxygen isotope ratios indicative of more evaporative conditions and/or reduced rainfall relative to the more southerly sites. The gastropod shells from the two lower Wadi Howar sites come from ephemeral lakes and show substantial changes in their oxygen isotope ratios during the time when the mollusca were living (Abell and Hoelzmann, 2000). This observation finds support from the presence of the gastropod Cleopatra bulimoides, which thrives in stagnant waters and is tolerant of moderate levels of salinity. These ephemeral lakes formed when Wadi Howar was in flood and then began to evaporate during the drier times of the year. Wadi Howar was flowing at least seasonally from about 9,500 to 5,500 14C years BP (Kröpelin, 1993). Archaeological surveys in the Wadi Howar region during 1996–1998 (Keding, 1998–2002) revealed that the earliest Holocene occupation in this area (the Dotted Wavy Line Mesolithic horizon) was confined primarily to the dunes adjoining Ennedi, Jebel Tageru and the lower reaches of Wadi Howar along the 300 km of the valley between Jebel Rahib and the Nile (Fig. 15.2). In the final phase of occupation at about 3.5 ka the lower Wadi Howar had been abandoned and occupation was confined to the middle reaches of the valley and the area around Jebel Tageru. This occupation pattern is consistent with a once active and perennial river becoming progressively more seasonal and ultimately becoming a chain of freshwater ponds fed by local rainfall until increasing desiccation in the last 1–2,000 years forced most of the inhabitants to seek refuge further south or in the Nile Valley.

15.3.2 The West Nubian Palaeolake and Adjacent Areas The West Nubian Palaeolake (Pachur and Hoelzmann, 1991; Hoelzmann, 1993; Pachur, 1997; Pachur and Rottinger, 1997; Hoelzmann et al., 2000, 2001) is located approximately 150 km north of Wadi Howar, 150 km NE of Ennedi and 150 km NNW of Jebel Rahib (Fig. 15.3). Initial attempts to map the area of the palaeoloake yielded equivocal results, which made it hard to estimate the balance between water inputs and losses. Hoelzmann et al. (2000) used a simple water balance model to estimate precipitation when the lake was full. They concluded that if the lake area amounted to 1,100 km2 an annual rainfall of about 500 mm would have been needed to balance estimated losses from evaporation. If,

218

Western Desert of Egypt and Eastern Sahara – Part 1

however, the lake area amounted to 7,000 km2 then a mean annual rainfall of about 900 mm would have been necessary to keep the lake full. In the first case, a northward shift of the summer rainfall belt of roughly 500 km is implied. In the second case, the rainfall regime would have been similar to that in South Sudan today, implying a northward shift of the isohyets by at least 800 km. Subsequent more stringent mapping of the lake deposits and partly obliterated strandlines revealed that the former lake had an area of 5,330 km2, a catchment area of ca. 78,000 km2, and a maximum elevation of ca. 540–545 m (Hoelzmann et al., 2001). Radiocarbon ages obtained on lake marls and subfossil gastropod shells indicated that the lake was at its maximum between ca. 9,400 and at least 7,500 14 C years BP, during which time a revised water balance model and associated fossil evidence suggested that the mean annual precipitation may have amounted to between 600 and 460 mm. These ages are expressed as radiocarbon ages and not as calibrated ages for two reasons. The first is that the ages obtained on organic and inorganic carbon differed by 600–1000 years (Hoelzmann et al., 2001, p. 2,000); the second is that the reservoir effect appears to have varied over time, by an unknown amount. In the archaeological sites bordering the former lake six main types of pottery have been identified (Hoelzmann et al., 2001). The oldest group is the Dotted Wavy Line pottery dating back to between about 6,300 and 5,300 14C years BP and was found only above an elevation of 545 m, when the lake was at its highest level. Arkell (1949) also recorded abundant Dotted Wavy Line pottery at the ‘Early Khartoum’ Mesolithic site he excavated. The makers of this pottery lived by hunting, fishing and gathering wild plant foods. A curious and as yet unexplained phenomenon concerns the time lag of roughly 3,300 calendar years between the inception of the former lake at ca. 9,400 14C years BP (if indeed that age is correct) and the first recorded Holocene prehistoric human occupation around the lake towards 6,300 14C years BP. Of course, there may have been earlier human occupation but if so it remains ‘archaeologically invisible’. As the lake level fell, Laqiya pottery replaced the Dotted Wavy Line pottery and sites with this style of pottery occur below 545 m elevation. The makers of Laqiya pottery were once again hunter-fisher-gatherers. Below a level of 540 m elevation the lake broke up into a number of smaller lakes and the total lake area dwindled to less than 400 km2. The Laqiya pottery was now replaced by Leiterband-decorated pottery and then by Halbmond–Leiterband pottery. Leiterband pottery is found also in Wadi Howar (Keding, 1993). The makers of both these pottery styles raised domestic cattle, and first appeared in this area about 500–1000 years after their initial appearance at the site of Shaheinab on the Nile (Arkell, 1953; Caneva, 1988), once again assuming that both sets of ages are indeed correct. It is hard to know just when the lakes dried out because the uppermost lake sediments have been eroded by the wind, but desiccation was certainly under way by about 4,000 14C years BP. Two later styles of pottery succeeded the Halbmond–Leiterband pottery. The first (fine geometric pottery) is associated with the presence of both cattle and goats; the second (coarse geometric pottery) is associated with the remains of cattle, goats and sheep (Hoelzmann et al., 2001, Fig. 9). The increasing dependence upon goats and sheep is consistent with a progressive decrease in annual rainfall in this region.

15.3 Wadi Howar and Adjacent Areas

219

Abell and Hoelzmann (2000) analysed the stable isotopic composition of shells from four species of aquatic mollusca collected from a site located along the southern margin of the West Nubian Palaeolake. The species were Melanoides (Melania) tuberculata, Biomphalaria pfeifferi, Lanistes carinatus and Cleopatra bulimoides. As we saw in Chapter 11 (Fig. 11.2), L. carinatus is amphibious, being endowed with gills and lungs, and so can tolerate seasonal desiccation. B. pfeifferi is sensitive to desiccation and soon dies if left stranded in seasonal pools. M. tuberculata can tolerate quite shallow water and moderate salinity. Between 9,400 and 8,800 14C years BP, the oxygen isotope ratios in the gastropod shells become progressively more negative, but after that time they become more positive in both shells and bulk carbonates, indicating the onset of more evaporative conditions and a decline in the amount of seasonal rainfall. These data are consistent with a northward penetration of the ITCZ during the early Holocene by 500–800 km north of its present front in this region (Abell and Hoelzmann, 2000). Other workers have arrived at essentially similar conclusions for sites in the southern Sahara located several thousand kilometres further west (Petit-Maire and Riser, 1983; Petit-Maire, 1991; Guo et al., 2000).

15.3.3 Wadi Muqadam and Wad el Melik Wadi Muqadam is an ephemeral stream channel and except during sporadic summer downpours is presently inactive. It rises among dissected and horizontally bedded Nubian Sandstone hills west of the lower White Nile and is fed by a number of small tributaries of which the most southerly is at approximately 15°10 0 N, 31°40 0 E (Sudan Geology, 1988: Khartoum 1: 000,000 sheet). Wadi Muqadam is separated from the valley of the White Nile by the Qoz Abu Dulu, a N–S aligned massive linear dune complex that extends northwards for at least 300 km west of the lower White Nile and the southernmost Desert Nile (see Chapter 14, Fig. 14.1). Almost nothing was known about the prehistory of Wadi Muqadam until the construction of a sealed road (Shariyat Shemal or North Road) north across the Bayuda Desert linking Omdurman and Khartoum to the main Nile at Gabolab. This road follows the Wadi Muqadam Valley in its lower reaches. In 1997, Dorian Fuller and his colleagues carried out a survey of archaeological sites exposed at the surface along the road and in the northern sector of Wadi Muqadam (Fuller, 1998; Smith, 1998–2002, Fuller and Smith, 1998 but published in 2004). They found a number of Acheulean hand axes (bifaces), many of them similar to those described by Arkell (1949b) from Khor Abu Anga near Omdurman (see Chapter 8), but the majority of finds were much younger, ranging mostly from Mesolithic to Meroitic. Two Mesolithic sites revealed a modest abundance of pottery with Wavy Line and Dotted Wavy Line decoration similar to that from the Early Khartoum site excavated by Arkell (1949) (see Chapters 8 and 11). The remains of the large land snail Limicolaria flammata point to an acacia-tall grass savanna environment with an annual rainfall of at least 400–500 mm (see Chapters 8 and 11), in contrast to the present-day annual precipitation

220

Western Desert of Egypt and Eastern Sahara – Part 1

of about 100 mm or less (Fuller, 1998; Fuller and Smith, 2004). Although it is true that occasional finds of extant Limicolaria flammata have been reported in very local habitats in the northern deserts of Sudan (Arkell, 1945; Haynes and Mead, 1987), the numbers of subfossil shells reported from Wadi Muqadam and their association with early Holocene prehistoric sites do seem to suggest that the climate and vegetation were quite different at that time. Fish bones of Tilapia and catfish indicate that the wadi was flowing, at least seasonally. It is not yet known if the catfish are Clarias or Synodontis. Clarias lives on floodplains and can estivate during the dry season whereas Synodontis lives today in the main channel of the Nile and is not adapted to living in wet burrows. Shell concentrations of Pila wernei, an edible semi-aquatic snail, offer further evidence for the importance of aquatic resources in the diet of the Mesolithic inhabitants of the lower Wadi Muqadam during the early Holocene, very much as in the lower White Nile Valley at this time (Williams et al., 2015b and Chapter 8). The presence of grindstones most likely implies that the seeds of wild grasses were being collected and eaten. Given the probable rainfall in this area at this time, there would have been plenty of edible species available such as Brachiaria, Digitaria, Cenchrus and Panicum, and possibly also the wild progenitors of millet and sorghum. To this day, the Zaghawa people of Sudan and Chad spend a month or two each year in places where the wild cereals grow (National Research Council, 1996, pp. 258–259). The women in particular collect certain annual grasses for food and beer, including Panicum, Dactyloctenium aegyptium, Echinochloa colona (shama millet), Eragrostis pilosa (wild tef) and Oryza breviligulata (wild rice). [Domesticated Eragrostis or tef is widely cultivated across the Ethiopian uplands today and is a highly nutritious and staple cereal crop.] The Zaghawa women visit different sites at intervals of 15–30 days and cover the piles of drying grass with thorny branches to deter goats and wild animals. A symbolic stone denotes each woman’s clan and serves to deter thieves. The collection can yield four or five camel loads of grain. The seeds of Cenchrus biflorus (cram-cram) and Tribulus terrestris are only used during times of famine. Their seeds are small and hard to collect. Further afield in the central and south-central Sahara collecting wild cereal grasses and many other plants is still an important activity among the Tibu of Tibesti, the Tuareg of the Hoggar, Tassili and Aïr Mountains, and the Fulani/Peul pastoral nomads of northern Nigeria and southern Niger (Rodd, 1926; Bernus, 1974; Benchelah et al., 2000; Gast, 2000; Baroin, 2003). In addition to wild cereal grasses, many other plants would have been a potential source of food. For example, the fruits of Balanites aegyptiaca, Adansonia digitata, Capparis decidua and Boscia salicifolia are collected and eaten across the savanna regions of central Sudan today (Obeid et al., 1982, Table 8.2). During the early to midHolocene, these trees would have been able to grow up to 800 km north of their present limits. Subsequent very brief visits to Wadi Muqadam in 1997 by other workers have revealed the presence of an Upper Acheulian biface and a probable Levallois core and flake (Mallinson, 1997, 1998). The climate in the eastern Sahara was significantly wetter on a number of occasions during the Middle and Late Pleistocene (Wendorf and Schild, 1980;

15.3 Wadi Howar and Adjacent Areas

221

Schild and Wendorf, 1981; Joseph et al., 1992; Wendorf et al., 1993a; Sultan et al., 1997; Osmond and Dabous, 2004; Geyh and Thiedig, 2008) so that the presence of Early and Middle Stone Age artefacts in this area is no surprise. Fuller (1998) and Fuller and Smith (2004) speculated that the White Nile could once have flowed into Wadi Muqadam, perhaps during the times when it had become a lake. This hypothesis seems improbable for several reasons. First, as we have shown in Chapter 8, the lower White Nile formed a seasonal lake with a maximum elevation of 386 m during the last interglacial because of the much greater Blue Nile discharge at that time, which obstructed flow in the White Nile and caused its waters to become ponded and form a seasonal lake. The reconstructed shorelines of this lake are consistent with the Blue Nile joining the White Nile at roughly its present point of confluence. The same argument is true of the terminal Pleistocene–early Holocene White Nile seasonal lake with its shoreline at a maximum elevation of 382 m. Second, the new road was aligned along the northern valley of Wadi Muqadam in part because alluvial gravels close to the surface were abundantly available as road ballast. The White Nile carries a suspension load of silt and clay and has no bed load gravel. Had the White Nile overflowed into Wadi Muqadam it would have flooded the valley with mud, not gravel. Wad el Melik is another major ephemeral stream channel that may once have experienced perennial or seasonal flow and attracted savanna plants and animals. It would repay investigation. I am not aware of any reports of prehistoric artefacts from this now defunct drainage system, although alluvial gravels have been reported there (Bussert et al., 2017). The age of these gravels is not known. Bussert et al. (2017) report the presence of alluvial gravels and conglomerates in northern Sudan, located up to 100 m above present Nile level in presently dry tributary valleys in the area around Shendi as well as detached outliers to the west of the present Nile. They mapped a series of fluvial conglomerates, of which the type locality is the Wadi Awatib Conglomerate, and observe that the conglomerates overlie Eocene marine sediments (at least in certain localities). The upper age limit of the conglomerates is harder to establish. In one locality, basalts with a claimed age of ca. 4 Ma overlie the conglomerates. A Neogene age for the conglomerates bracketed between Miocene and Pliocene therefore seems reasonable, although it is likely that the conglomerates in this region may cover a wide range of ages, as the authors acknowledge. They infer deposition under very high seasonal flow linked to a stronger than present monsoonal climate in this region but with a long dry season and minimal chemical weathering under a relatively sparse vegetation cover. Very limited palaeocurrent measurements (N = 5) and the generally meridional alignment of the remnant gravel-filled channels indicate a general flow direction from south to north, consistent with current surface declivities. Many of the ancient river channels in the Western Desert of Egypt are now the highest elements in the landscape (Butzer and Hansen, 1968; Zaki and Giegengack, 2016). This inversion of relief is common in arid regions, and results from post-depositional cementation of the river gravels by silica, carbonate or iron, leaving them resistant to later denudation. Similar features have also been observed on Mars (Zaki et al., 2018).

222

Western Desert of Egypt and Eastern Sahara – Part 1

15.4 The Darb el Arba’in Desert: Oyo, El Atrun and Selima Oasis Although it is notoriously difficult (Lézine and Casanova, 1989; Williams, 2014, chapter 16) to carry out worthwhile pollen analysis in arid areas – the pollen grains are all too prone to oxidation and destruction in dry environments – a few favoured localities such as Oyo, El Atrun and Selima in hyperarid northern Sudan are the exceptions.

15.4.1 Oyo Holocene lake muds buried beneath wind-blown sands occupy part of the Oyo Depression in NW Sudan at 19°16 0 N, 26°11 0 E and 510 ± 20 m elevation (Ritchie et al., 1985). Oyo lies 50 km NE of the West Nubian Palaeolake described in Section 15.3.2 and roughly 150 km NW of El Atrun/Bir Atrun (Fig. 15.4). The chronology of the Oyo Holocene Lake is based on 8 14C ages obtained on five algal sapropel samples and three charcoal samples collected from a 5 m pit and from hand auger boreholes. The 55 pollen taxa identified include (a) Sudano-Sahelian taxa similar to those now living in the tropical savannas of north-central Sudan, (b) Sahelo-Saharan taxa similar to those in the thorn scrub desert of the southern Sahara, (c) Saharan taxa, (d) montane elements similar to those now living in the Tibesti Mountains, and (e) rare Mediterranean taxa. Ritchie et al. (1985) distinguish three main sedimentary and associated pollen zones. They conclude that between 8,500 and 6,100 14 C years BP there was a deep stratified lake in the Oyo depression surrounded by continuous deciduous savanna vegetation akin to that now growing some 500 km further south, indicating a seasonally wet tropical climate with at least 400 mm of annual summer rain. Towards 6,000 14C years BP a decline in precipitation to about 300 mm and/or an increase in evaporation led to a fall in lake level. Between 6,000 and 4,500 14C years BP the lake became even shallower, rainfall decreased to less than 100 mm, and acacia thorn savanna and scrub-grassland replaced the tropical Sudano-Sahelian savannas. The lake eventually dried out by about 4,500 14C years BP, wind-blown sand covered the lake basin, and the desert scrub vegetation died out except in occasional oases and wadis.

15.4.2 El Atrun The oasis of El Atrun is located at 18°10 0 N, 26°39 0 E within a depression 4 km wide and 8 km long surrounded by bedrock hills. Pollen-bearing sediments overlie bedrock at a depth between 4.7 m and 2.5 m. The surface consists of wind-blown sand and crystalline trona or natron, an evaporite mineral of sodium carbonate that is mined by hand locally and sold in distant markets across Egypt and Sudan. Trona has long been used for cleaning and for making an early form of soap. The pollen-bearing sediments are composed of finely laminated organic and carbonate muds. These were laid down between 9,900 and 6,200 14C years BP, at which time the surrounding area was covered in savanna woodland similar to that now found some 400 km to the WSW and SSW (Ritchie and Haynes, 1987).

15.4 The Darb el Arba’in Desert

25°E

223

35°E

30°E

30°N N

CAIRO

LIBYA

EGYPT Asyut

Nil

Kharga

Darb el Arba’in Kufra

25°N

e

Dakhla

Aswan

Desert

Jebel ‘Uweinat

Selima

Wadi Halfa

20°N Nile

CHAD

Oyo El Atrun

di Wa

Ho

wa

r KHARTOUM

SUDAN El Fasher

15°N

100 km

Figure 15.4 The Darb el Arba’in Desert and Selima Oasis.

15.4.3 Selima Oasis Ritchie and Haynes (1987) also recovered pollen from a 3.7 m pit dug at Selima Oasis located at 21°22 0 N, 29°19 0 E in the hyperarid heart of the eastern Sahara. Selima lies 450 km NE of El Atrun. The 14C chronology was based on twenty-nine samples (charcoal, algal sapropel, or carbonate) and indicated that between about 10,000 and 6,000 14C years BP the area close to Selima was sparsely covered in wooded desert steppe, dominated by Acacia, Commiphora and Maerua, similar to the Sahelian vegetation that now grows 400 km to the south. The combined pollen evidence from Oyo, El Atrun and Selima is consistent with a 400–450 km northward displacement of the Saharo-Sahelian vegetation belt (and of the associated 440–100 mm summer rainfall zones) between about 10,000 and 5,000 14C years BP (Haynes, 1980; Ritchie and Haynes, 1987).

224

Western Desert of Egypt and Eastern Sahara – Part 1

Lézine and Casanova (1989) reviewed the dated evidence from Holocene lake sediments and fossil pollen grains investigated across tropical West Africa. They concluded that two abrupt changes in the vegetation could be discerned during wetter climatic episodes in the Sahelian and Saharan zones bracketed between about 9,500 to 7,000 14C years BP and 4,000 to 2,500 14C years BP. They also noted possible evidence for an increase in rainfall seasonality at 7,500 14C years BP, as had Maley (1982) somewhat earlier on the basis of changes in dominant sediment types in this region. He considered that the change from fine-grained alluvial deposits to coarse sandy deposits reflected a change in precipitation from gentle early Holocene rains scavenging dust from the atmosphere at the start of the summer wet season to more intense convectional downpours after about 7,500 14C years BP. The suggestion is interesting albeit very speculative, but does seem to have received some support from pollen analysis.

15.4.4 The Selima Sand Sea The Western Desert of Egypt consists of a broad monocline surface dipping gently to the north. The rocks that make up the monocline consist of Middle to Late Mesozoic continental and shallow marine sediments (mostly sandstones and siltstones) to the south and early Cenozoic marine sediments (mainly limestones and minor siltstones) to the north (Issawi, 1981). Fluvial dissection of the early Cenozoic and older rocks gave rise to a landscape characterised by high and often deeply dissected escarpments such as the crenulated eastern margins of the Gilf Kebir (Peel, 1939, 1966) and the south facing scarps of Kharga, Dakhla and Bahariya oases (Maxwell and Haynes, 2001). The age of these ancient river systems is poorly known, with suggestions that some may date back to Oligocene times some 30 Ma ago (McCauley et al., 1982, 1986; Breed et al., 1987). Further west the vast Sahabi Rivers flowed from northern Chad across the present Libyan Desert to the Mediterranean during the Miocene (Griffin, 1999, 2002, 2006, 2011). It therefore seems entirely probable that big rivers once flowed across the eastern Sahara during mid to late Cenozoic times. Their relations to the ancestral Nile remain a subject for considerable conjecture (Goudie, 1985; McHugh et al., 1988, 1989; Bussert et al., 2017) and need detain us no further. The Selima Sand Sea (Bagnold, 1933, 1935) is a flat, waterless sand sheet that is today quite devoid of any plant cover. It occupies an area of roughly 60,000 km2 between latitude 21°N in the south, the Gilf Kebir sandstone plateau and Jebel ‘Uweinat ring complex to the west, Selima Oasis to the east and Bir Tarfawi to the north. It is bounded by the somewhat notional 0 mm isohyet and lies squarely within the hyperarid core of the eastern Sahara and the central part of the Darb el Arba’in Desert (Haynes, 1987, 1989; Stokes et al., 1998; Maxwell and Haynes, 2001). This desert derives its name from the 40-day camel caravan track (Darb el Arba’in) used regularly by Arab merchants up to the close of the nineteenth century and also notorious as the route by which slaves were led from El Fasher in Darfur

15.4 The Darb el Arba’in Desert

225

via Kharga Oasis to Asyut on the Nile in Egypt (Bagnold, 1935, map on p. 184). To this day it is lined with sporadic camel skeletons, silent witnesses to the rigours of the journey. Maxwell and Haynes (2001) found that the sediments in the upper 2 m of the Selima sand sheet showed varying degrees of soil development and consequent obliteration of primary sedimentary structures. These sediment units and incipient fossil soils are very similar to those mapped by the author in the desert east of the Nile in northern Sudan (Chapter 14). Maxwell and Haynes (2001) identified five main stages (0–4) in sediment modification by soil forming processes (Maxwell and Haynes, 2001, Table 1). The stage 0 unit consisted of bimodal fine sand and granules with a planar horizontal structure and was the surface unit; it ranged from one lamination a few millimetres thick to several tens of centimetres in thickness. This unit was mobile, devoid of cohesion and locally accumulating in shallow depressions. The stage 1 unit was similar (planar horizontal bimodal fine sand and granules) but more cohesive, keeping a vertical face when excavated, and up to 20 cm but usually only 2–4 cm thick. The stage 2 unit again comprised bimodal fine sand and granules, was up to 10 cm or more in thickness, but its planar horizontal stratification was interrupted by a medium prismatic soil structure (United States Department of Agriculture, 1951, p. 227), indicative of incipient soil formation. The stage 3 unit was up to 20 cm thick, with a weak coarse prismatic structure, little sign of any horizontal stratification, and common surface cracks. It was made up of fine sand and granules with some clay and minor carbonate, both derived from wind-blown dust (see also Williams, 2014, chapter 9). The stage 4 unit was up to 30 cm thick or more, had no obvious stratification and had a well-developed coarse prismatic structure and surface cracks. It consisted of fine sand and granules, some clay and in places a fine pebble alluvium. Two other depositional units were also evident: a fine pebble alluvium with subrounded to subangular pebbles in a clay matrix, and a coarsegrained alluvial-colluvial deposit cemented by calcium carbonate which sometimes graded laterally into calcrete. In places, Neolithic stone artefacts occurred at the contact between the surface stage 0 unit and the immediately underlying stage 4 unit. Stokes et al. (1998) obtained two optical ages for the stage 2 unit and three for the stage 3 unit. The stage 2 sedimentary unit had OSL ages of 3.37 ± 0.25 and 3.65 ± 0.31 ka, and the stage 3 unit had OSL ages of 15.94 ± 1.54 17.41 ± 2.97 and 21.32 ± 6.93 ka (Stokes et al., 1998, Table 1). On the basis of these five optical ages they concluded that the stage 3 unit had accumulated in the very late Pleistocene during and somewhat after the Last Glacial Maximum (21 ± 2 ka). From 4 ka onwards, sand sheet deposition was widespread apart from a short humid interval at ca. 2.5–2.0 ka. How to interpret the varying degrees of soil development seen in the near surface sediments of the Selima Sand Sea is no easy task. Six main factors control soil development, namely, parent material, topography, biological activity, soil climate and time. We saw in Chapter 4 that the two factors that may be considered as passive factors, notably topography and parent material, are very important in the early stages of soil development. Over time the more active factors of soil climate and biological activity exert an increasing influence over soil development. In this area of level to very gently undulating topography, we can consider topography as relatively uniform in space and time. Parent material will vary very little also,

226

Western Desert of Egypt and Eastern Sahara – Part 1

apart from incremental additions of calcareous dust bringing carbonate and fine particles (clay to silt size) to the sandy parent material. The two key transforming agents are therefore likely to be soil climate and biological activity. Soil climate will reflect the regional climate. During wetter climatic intervals, moist soils or incipient soils will be able to foster plant growth and associated microbial and faunal activity. Other things being equal, soils will develop more differentiated profiles during wetter climatic interludes. However, it is still hard to distinguish between the effects upon soil formation of a modest increase in soil moisture sustained over a long interval of time and those of a substantial increase in soil moisture over a shorter interval of time (see Williams, 2014, chapter 15 for details). A further problem is how to separate the effects of sporadic but intense local downpours from those of more widespread but less intense precipitation. The problem is compounded when the different units discussed above do not occur in an expected stratigraphic sequence. For instance, stage 0 or stage 1 units may (and do) lie beneath stage 2 or stage 3 units rather than above them, as one might intuitively expect. Whatever conclusions are drawn from sediments showing varying degrees of pedogenesis in a region such as this will need to be evaluated very rigorously against independent evidence based on microfossils such as pollen and macro fossils such as snail shells or vertebrate remains, all of which present their own interpretative pitfalls.

15.5 Conclusion Although there are tantalising indications of a human presence extending back to Early Stone Age/Lower Palaeolithic times in this presently very dry region, it is not until the Holocene that we find abundant evidence of human occupation in the southern portion of the eastern Sahara. The Middle Stone Age sites associated with a series of Last Interglacial lakes will be discussed in the next chapter. The interval from ca. 13.7 to 8.9 ka and locally up to 5 ka coincides with a time when freshwater lakes were widespread west of the Desert Nile and in the SE Sahara and supported a varied savanna flora and fauna. In its lower reaches, the now dry Wadi Howar flowed through a region with abundant groundwater springs and local lakes between 9,500 and 4,500 14C years BP and was a major tributary of the Nile during that time. Its flow may have been perennial for some of that time and seasonal at other times. It is probable that Wadi Muqadam and Wad el Melik, like Wadi Howar, ceased to flow to the Nile by about 4.5 ka. As an aside, it is also highly likely that the many hundreds of now defunct or very ephemeral wadis that originate in the Red Sea Hills had a similar history and once brought water and sediment to the Nile from the east, a topic we review in Chapter 18.

16 West of the Nile: The Western Desert of Egypt and the Eastern Sahara – Part 2

It seemed we had strayed into a secret Stone Age world . . . A well-worn path led inland from the brink [of the plateau] to successive factories of stone implements. Debris from the ancients’ works lay strewn everywhere. It looked so fresh we half expected to come upon a group of uncouth artisans round the next corner. Yet we found no scrap of evidence that anyone had been here since prehistoric times. Ralph A. Bagnold (1896–1990), Sand, Wind, and War: Memoirs of a Desert Explorer (1990, p. 117)

16.1 Introduction In the present-day oases of Kharga and Dakhla in the Western Desert of Egypt, thick tufa deposits bear witness to past spring activity and higher groundwater levels during the past half million years and more. Similar tufa deposits are found far to the west in the Acacus Massif of central Libya. Pleistocene lakes, some of them very large, occupied topographic depressions in what is now hyper-arid western Egypt, northern Sudan and south-central Libya. A succession of Middle and Late Pleistocene lakes at Bir Sahara and Bir Tarfawi in the southern sector of the Western Desert of Egypt attracted prolonged but intermittent Middle Palaeolithic/Middle Stone Age (MSA) occupation. Scattered across the eastern and southern Sahara there are concentrations of stone tools ranging from Lower Palaeolithic/Early Stone Age (ESA) through Middle Palaeolithic and Upper Palaeolithic to Neolithic and younger, mute witnesses to times past when the climate was less harsh than it is today. The great sandstone plateau of the Gilf Kebir (Fig. 16.1) was able to support Neolithic herders as recently as 5,000 years ago. Even more dramatic than the remains of Neolithic pottery and stone tools are the galleries of prehistoric rock engravings and rock paintings, including the spectacular scenes of Neolithic cattle at Jebel ‘Uweinat, an isolated mountainous ring complex rising nearly 1,300 m above the surrounding sandy desert to an elevation of 1,934 m and sitting astride the present-day frontiers of Egypt, Libya and Sudan. In spite of difficulties associated with dating many of these sites and interpreting their associated deposits, a much clearer picture is now

227

Western Desert of Egypt and Eastern Sahara – Part 2

228

32°N

SEA

AN AN E Benghazi ERR T I D ME

80 km

GR

EA T

SEA

erir

SAND

oS

LIBYA

ND

sci

lan

SA

CALANSCIO

Ca

28°N

EGYPT

N

SEA REBIANA SAND SEA 24°N

Gilf Hamada el Fayoud

Wadi Bakht

Hamada el Akdamin TIBESTI MTS

CHAD

Sand sea

Jebel Arkenu Jebel Uweinat

Hamada Ibn Battutah 20°E Sandstone plateau

Kebir

SUDAN 24°E Volcanic mountains

Ring complex

Figure 16.1 Sites discussed in the Western Desert of Egypt and the SE Libyan Desert.

emerging of how the environments in this huge region responded to past climatic changes and of how prehistoric communities made use of the opportunities offered by these favourable habitats.

16.2 Dakhla and Kharga Oases Kharga and Dakhla are two of a series of oases (Fig. 15.1) that are cut into the underlying Nubian Sandstone and bordered by high tablelands of Eocene limestone (Wendorf and Schild, 1980; Embabi, 2004, pp. 190–202). The oases are fed primarily from groundwater, some of which comes from great depth and so can be quite warm and full of dissolved chemicals. Springs are common, and are often encircled by mounds of spring tufa, generally formed of calcium carbonate, although bands of gypsum and halite are not

16.3 The Gilf Kebir, Jebel ‘Uweinat, Jebel Arkenu

229

uncommon. Many of the springs are no longer active and there is frequent evidence that the eye of the spring has migrated over time. (The Arabic word ‘ain’ denotes both the eye and a spring or well). Associated with the spring deposits, and with tufa deposits precipitated along the waterfall reaches of formerly active stream channels, there is abundant evidence of a prehistoric human presence in the form of Lower Palaeolithic/ ESA and Middle Palaeolithic/MSA stone tools as well as Neolithic pottery, tools, shelters and burials. The classic work at Kharga is that of Gertrude Caton-Thompson (1952), which followed an earlier survey by Caton-Thompson and Gardner (1932). Wendorf and Schild (1980, pp. 216–222) also carried out excavations at Kharga. Their excavations simply confirmed the original findings of Caton-Thompson (1952) as to the presence of ESA (Acheulian) and MSA (Mousterian and Aterian) artefacts, followed by various stages of the Neolithic. However, because the artefacts were found in spring deposits, or on heavily deflated surface sites, there had been considerable mixing and reworking, so that little could be gleaned about past prehistoric behaviour except at the most general level. Of far greater significance are the tufas and the light they shed on past episodes of groundwater recharge, a topic reviewed in Section 16.6. 16.3 The Gilf Kebir, Jebel ‘Uweinat, Jebel Arkenu and Environs 16.3.1 The Gilf Kebir Linstädter and Kröpelin (2004) provide useful overviews of the Holocene environments in this region. The Gilf Kebir (or big cliff, barrier or wall in Arabic) is a vast sandstone plateau aligned roughly SE–NW composed primarily of Nubian Sandstone with some younger volcanic outcrops. It is located ca. 600 km west of the Nile (Figs. 16.1 and 16.3) and consists of two plateaux separated by a wide sandy valley that forms a natural col between the two tablelands. The northern plateau is known as Abu Ras plateau and the southern plateau as Kemal el Din plateau in honour of Prince Kemal el Din Hussein who explored this area in 1925 and 1926 (Kemal el Din Hussein, 1928). The overall plateau covers an area of 5,800 km2 and attains a maximum elevation of ca. 1,050 m in the south and ca. 900 m in the north. The southern cliffs are over 300 m high and hard to climb but are scalloped and deeply dissected by now dry valleys up to 20 km long and 4 km wide. As the quotation at the head of this chapter indicates, Bagnold (1990) has given us a vivid account of his visit to the Gilf Kebir in February 1938 with Ronald Peel (Peel and Bagnold, 1939; Peel, 1939, 1941, 1966) and other colleagues. This visit led to discoveries of stone tools, pottery, prehistoric tracks and rock art on the plateau (Bagnold et al., 1939; Winkler, 1939a) and of a series of Neolithic sites at Wadi Bakht (Fig. 16.1), where Bagnold’s archaeologist colleague Myers collected artefacts as did Wendorf et al. (1980, pp. 217–222) on a later visit in 1975. Pre–World War II expeditions had skirted the plateau since the cliffs seemed impossible for vehicles. Subsequent more detailed

230

Western Desert of Egypt and Eastern Sahara – Part 2

exploration was to wait nearly half a century (El Baz et al., 1978; McHugh, 1978; Wendorf et al., 1980). Four main phases of Holocene prehistoric occupation have been identified on the basis of pottery styles and associated stone tool assemblages (see Linstädter and Kröpelin, 2004, Fig. 5, p. 765, and Linstädter, 2007, Fig. 2, p. 36). The oldest known phase is an Epi-Palaeolithic/Mesolithic hunter-gatherer phase (Gilf A), which dates back to ca. 10.3 ka; whether pottery was part of this culture is uncertain. The Gilf B phase (8.8–6.3 ka) is one of hunter-gatherers who did make pottery. Neolithic herders of sheep, goats and cattle (Gilf C: 6.3–5.3 ka) were present in and around the Gilf Kebir at a time when Linstädter and Kröpelin (2004) have suggested that there was a change from a summer to a mainly winter rainfall regime. The Gilf D phase (5.3–4.7 ka) ended with increased desiccation and lack of reliable water, in contrast to Jebel ‘Uweinat to the west, which has four permanent water points to this day (Menardi Noguera and Zboray, 2011, p. 101). The inference by Linstädter and Kröpelin (2004) that winter rains were prevalent during 6.3–5.3 ka is based on some subtle – even tenuous – evidence and reasoning. However, it does receive some support from the palaeobotanical work of Schulz (1994). The movement of Neolithic pastoralists onto the plateau surface and away from the deeply embayed wadis dissecting the plateau margin took place at this time, so that adequate supplies of grass must have likewise been available then. An 8-m alluvial-lacustrine (playa) section exposed in Wadi Bakht first noted by Bagnold et al. (1939) and also studied by Wendorf and Schild (1980) covers a span of about 5,500 years between ca. 10.3 ka and ca. 4.7 ka. In the lower 7 m of the section playa muds form discrete layers of mud and intercalated sands with a mean thickness of 14 mm for the mud units. Intense rainstorms linked to infrequent northward incursions of the ITCZ and associated summer monsoonal rains are considered responsible for creating the ephemeral dune-blocked pools in which the thin mud layers accumulated (Wendorf and Schild, 1980, p. 757 and pp. 774–775). The uppermost metre of sediment is massive and is hypothesised to reflect slow and sustained sedimentation under a regime of gentle nocturnal winter rains (Wendorf and Schild, 1980, p. 774). Neither the fossil vertebrate remains found in association with archaeological sites overlying the Wadi Bakht playa nor the plant macrofossils preserved within the playa sediments are strictly diagnostic of a winter rainfall regime. Ziziphus, Tamarix and Maerua crassifolia all survive in favourable sites scattered across the desert regions of the eastern Sahara and northern Sudan today. The arguments for winter rain are based on a combination of inductive, deductive and actualistic reasoning. Wendorf and Schild (1980) assert that ‘the dynamics of present-day weather patterns are apparently similar to those of 6000 years ago’ (Wendorf and Schild, 1980, p. 763), adducing in evidence a present-day advance of winter rains as far south as latitude 20° N, and sporadic northward extension of monsoonal rains. They further assert that winter rains are non-erosive, fall mainly at night when temperatures are lower and

16.3 The Gilf Kebir, Jebel ‘Uweinat, Jebel Arkenu

231

evaporation less, so that infiltration into the surface soil is enhanced and runoff reduced, allowing more grass to grow. In contrast, summer monsoonal rains are alleged to occur mainly during the day and to be high intensity rains leading to reduced infiltration and so less soil moisture storage and hence reduced pasture growth. Each of these assumptions is plausible but each is no more than an unproven assumption. The issue is best considered unresolved. 16.3.2 Jebel ‘Uweinat Jebel ‘Uweinat (or ‘mountain of small springs’ in Arabic) is a ring-complex ca. 25 km in diameter located in the heart of the Libyan Desert across the frontiers of Libya, Egypt and Sudan (Figs. 16.1 and 16.3). Like many similar ring-complexes in the Sahara and in the Bayuda Desert of northern Sudan, ‘Uweinat has a central core of granite flanked by concentric aureoles of other igneous rocks such as granodiorite, syenite, gabbro and dolerite. In common with the smaller ring complex of Jebel Arkenu 35 km to the north (Figs. 16.1 and 16.4), the ‘Uweinat ring complex was intruded into older rocks in Eocene times during the mid-Cenozoic (Atherton et al., 1978; Derek et al., 1978, both cited by Menardi Zoguera and Zboray, 2011, p. 112). The older rocks are more or less horizontal sandstones ranging in age from Palaeozoic to Mesozoic and have been eroded to expose the western sector of the massif, which is now deeply dissected by relatively wide, flatfloored dry valleys flanked by one or more alluvial terraces. The remote ancestors of these presently defunct drainage systems may have been superimposed from the overlying sedimentary rocks that would probably have been warped upwards during the successive episodes of ring dyke intrusion. The Eocene was a time of prolonged deep weathering under a hot wet tropical climate and a dense forest cover in the central Sahara (Faure, 1962b; Greigert and Pougnet, 1967; Williams, 2014, chapter 18). The more highly weathered sandstones would have probably offered little resistance to erosion unless protected by more resistant beds that had been strongly silicified and/or iron-cemented during or after deposition. ‘Uweinat is well known to rock art specialists for its many scores of rock shelters adorned with paintings and, less commonly, engravings of animals and humans. Early descriptions date back to the 1920s and 1930s (Hassanein Bey, 1924; Breuil, 1926, 1928; Caporiacco and Graziosi, 1934; Almásy, 1936, 1939; Winkler, 1939b) and in more recent times every issue of the former journal Sahara (1989–2013: 24 volumes) was devoted to often highly detailed descriptions of the rock art across the Sahara including in and around ‘Uweinat (Le Quellec, 2009; Zboray, 2009; Menardi Noguera and Zboray, 2011). At the time of writing, more than a thousand individual rock paintings have now been documented at ‘Uweinat. Le Quellec (2009, Fig.1) provides a clear satellite image on which are shown the main rock-art sites scattered across the massif. He also cautions against over facile interpretations of the art, citing Breuil’s comparison with Bushman (San) art as an example (Breuil, 1928).

232

Western Desert of Egypt and Eastern Sahara – Part 2

Alfred Muzzolini (1995) has summarised the history of Saharan rock art discovery and in his 1995 monograph provides a useful introduction to many of the better-known localities. He was critical of the lack of any attempts to go beyond mere description and cataloguing of finds based on a bewildering array of locally identified ‘styles’, ‘styles’ sometimes spawned by different workers describing the same paintings. He contrasted what he regarded as the embryonic status of Saharan rock art studies with the more rigorous interpretation emanating from southern Africa and Australia. In fairness, one could argue that in contrast to Australia and southern Africa, the Neolithic and earlier artists have long since vanished from the Saharan rock art sites, so that no living descendants and informants are available to help interpret the enigmatic symbols and depictions. For example, Patricia Vinnicombe (1976) and David Lewis-Williams (1981, 1990, 1991), among others, have drawn attention to the spiritual and ritual significance of certain animals, dancing, and behaviour during trance among the San people of southern Africa as a guide to the, to us, somewhat esoteric ways in which the prehistoric and historic rock artists in southern Africa have sought to portray possible behaviours and beliefs. Coulson and Campbell (2001) make the same point in their beautifully illustrated book on African rock art, noting, for example, the symbolic and ritual significance of paintings of giraffes for the San artists of SW Africa. Mark Borda (2011, p. 127) has sounded an additional note of caution: ‘It perhaps cannot be assumed that wherever paintings are found, the scenes of life they document actually occurred in the immediate vicinity of the rock art.’ Borda was describing a painted rock shelter located in an elevated sandstone basin surrounded by rugged sandstone cliffs at Jebel Arkenu. He went on to propose that ‘one therefore has to allow for the possibility that these paintings represent activities that were only taking place down on the plains’ (Borda, 2011, p. 127), observing that the giraffe paintings were consistent with this conclusion, since the actual presence of giraffes would be hard to explain in such rugged and inaccessible terrain. Zboray (2009) and Menardi Noguera and Zboray (2011) provide a comprehensive inventory of the rock art at ‘Uweinat and have attempted to place the art into context with the landscape and the presence of animal and possible prehistoric trackways (see also Le Quellec, 2011). Of the various styles of painting specified by Zboray and his colleagues, the Miniature style appears to be pre-pastoral in that cattle are absent from the art and the figures in paintings of the ‘Uweinat Cattle Pastoralists are often superimposed upon the Miniature style figures. The Round Head style, sometimes portrayed as elongated figures, is also very evident at Arkenu. Pottery fragments are often found in and near the mostly granite rock shelters where the art galleries occur, together with sporadic scatters of stone artefacts, usually fashioned from basalt. There have been no systematic modern archaeological excavations at ‘Uweinat so that the possible occupation history is inferred by analogy with dated occupation sites elsewhere in the wider region. By comparison with the Gilf Kebir ceramics, as well as those further afield for which a reliable radiocarbon chronology exists, four occupation phases have been identified at ‘Uweinat (Sukova, 2011). Phase A is an early Holocene Epi-Palaeolithic phase characterised by no pottery

16.3 The Gilf Kebir, Jebel ‘Uweinat, Jebel Arkenu

233

and probably older than ca. 8.6 ka. It probably coincides with the humid phase dated to 9.4–8.1 ka during which tufas accumulated in pools within some of the wadis (Marinova et al., 2014). The main Neolithic occupation associated with the paintings of cattle, goats and sheep, dogs, archers, and domestic scenes involving adult males and females as well as juveniles, seems to be bracketed between ca. 8.6 and ca. 5 ka and is subdivided into the two phases B and C. The latest phase identified (D) is younger than ca. 5 ka (Kuper and Riemer, 2010). Lenka Sukova (2011) has described a unique piece of carved and polished veined Nubian Sandstone 6.5 cm high which she discovered in the western sector of ‘Uweinat. She referred to this small sculpture as the ‘Venus’ of Jebel ‘Uweinat. The figurine was found lying face down in front of a group of granite boulders located at the foot of the mountain between Ain Zueia and Ain Doua, both of which have permanent water. Two similar highly stylised small (?) female human figures carved from veined Nubian Sandstone came from two graves in the Neolithic cemetery of Kadruka next to the Nile in northern Sudan. Four calibrated 14C ages indicate an age of 6.8–6.7 ka for this cemetery. We are left with the tantalising possibility of long distance exchange between the Desert Nile and Jebel ‘Uweinat in the far SE of Libya, perhaps taking place over several generations. Andrew Smith (2004) has argued from ethnographic data, Saharan rock art and funerary monuments that we may have underestimated the former links between prehistoric and proto-historic pastoralists in the Sahara, Nile Valley and Red Sea Hills. Contemporary pastoral societies such as the Fulani and Tuareg, and, to a lesser extent, the Beja and Tibu, certainly require a vast area of land in which to seek pasture and water for their herds. During wetter intervals in the early to mid-Holocene Neolithic herders would have sought refuge for their animals away from the former tsetse fly breeding grounds in the southern Sahel by moving their cattle further north into drier areas.

16.3.3 Jebel Arkenu Jebel Arkenu is a ring complex ca. 20 km in diameter similar to but smaller than that of Jebel ‘Uweinat, which lies 35–40 km to the SE. It stands about 450 m above the surrounding desert plains to attain a maximum elevation of 1,435 m. The various rings consist of andesites, rhyolites, dolerites, microgranite, granite, felsite and porphyry (Williams and Hall, 1965, Fig. 2). The NW sector of the massif is a highly dissected plateau made up of roughly horizontal Cambro-Ordovician sandstones (Williams and Hall, 1965, Fig. 2). The following brief sketch of the geological history of Arkenu provides some insight into how the mountain attained its present relief and morphology (Burollet, 1963; Vittimberga and Cardello, 1963; Williams and Hall, 1965). Precambrian folding and regional metamorphism was followed by prolonged erosion and the creation of a gently undulating erosion surface. Deposition of coarse subangular quartzose gravel under continental conditions during the Cambrian was followed by renewed erosion during

Western Desert of Egypt and Eastern Sahara – Part 2

234

Stone circle on upper terrace Neolithic tools in situ

Wadi channel

? 4.5m Lower terrace

? 2m

Sand

Gravelly sand

Gravel

Bedrock

Figure 16.2 Late Quaternary alluvial terraces at Jebel Arkenu. (After the author’s field notes.)

the Lower Ordovician. Fine and coarse sand and pebbles were laid down in the Upper Ordovician, followed by Upper and Middle Devonian deposition of fine sands, silts and muds some 100 km to the N and NW of Jebel Arkenu. Late Devonian and Lower Carboniferous subsidence was accompanied by further accumulation of silts and muds. In post-Carboniferous times, episodic volcanic activity resulted in outflows of andesite and rhyolite onto the Carboniferous series. This was followed by deposition of the Nubian series (i.e., Nubian Sandstone Formation) sands and silts during the Jurassic and Cretaceous. Intrusion of the granitic ring dykes during the mid-Cenozoic resulted in metamorphism of the older sedimentary and extrusive rocks, and was accompanied and followed by tectonic uplift, volcanism and erosion. As at Jebel ‘Uweinat, ancestral rivers flowing across the Mesozoic and older sedimentary formations would have cut down vertically into the ring complex rocks during times of tectonic uplift. The rocks most resistant to erosion now form concentric ridges; the less resistant rocks form valleys or broad depressions. Williams and Hall (1965, p. 482) reported that ‘Gebel Archenu was named after the solitary acacia which stands opposite the outlet of the main wadi, over a kilometre to the south-west of the mountain’ (Hassanein Bey, 1924, p. 354). However, as Théodore Monod kindly pointed out to the author following his own visit to Arkenu a few years later, this tree is a Maerua crassifolia and not an acacia. This is all the more interesting in that Menardi Zoguera and Zboray (2011, p. 101 and plate A5, p. 111) concluded that one of the paintings they described from Jebel ‘Uweinat depicted a goat browsing a tree. They commented (Menardi Zoguera and Zboray, 2011, p. 101) that ‘the geometry of branches and leaves suggest this plant to be a Maerua crassifolia’. Andrews (1948, p. 37) observed that Maerua crassifolia grows along the larger sandy beds of wadis in rocky terrain west of the Nile in northern Sudan. Snow (1948, p. 684), in his review of animal foodstuffs in the Sudan, noted in regard to Maerua crassifolia Forsk that the ‘twigs are greedily stripped of their small

16.3 The Gilf Kebir, Jebel ‘Uweinat, Jebel Arkenu

235

leaves by camels’. They were doubtless just as attractive to Neolithic goats as they are to present-day camels. Two major dry valleys traverse Arkenu, each of which reaches the desert plain along the southern margin of the mountain. The longest and most westerly of these two wadis reaches about 20 km into the massif. Hassanein Bey (1924) reported a spring of fresh water in the upper reaches of this wadi during his visit in 1923, but by the time of our visit there in August 1962 (Williams and Hall, 1965) the spring had almost ceased to flow and the water had become salty and polluted with gazelle dung. Two discontinuous gravel terraces border the wadi channel for several kilometres upstream of its outlet (Fig. 16.2). The surface of the higher terrace is between 5 m and 7 m above the wadi floor; it consists of large rounded boulders, many in excess of 30 cm diameter. The lower terrace is seldom more than 2 m high and consists mostly of boulders 15–30 cm in diameter. Sporadic stone waste flakes, fragments of pottery and two stone circles were recorded by Williams and Hall (1965) on the surface of the upper terrace. Noting that Peel (1939) and Peel and Bagnold (1939) had described two similar sets of boulder terraces in the Gilf Kebir, Williams and Hall (1965) suggested that the higher terrace might have been in existence before the arrival of Neolithic pastoralists into the massif. They considered that the lower terrace was etched into the higher terrace during a renewed phase of incision in postNeolithic times. Until these alluvial deposits have been dated using optical luminescence and/or cosmogenic nuclide dating methods (see Chapter 4) further speculation is not warranted.

16.3.4 The Hamada el Akdamin and Adjacent Region Three Nubian Sandstone sandstone plateaux are located between latitudes 22°N and 24°N and roughly 200 km west of Jebel Arkenu and 100 km east of the NE tip of Tibesti (Fig. 16.1). Until 1963 the existence of these three plateaux was not widely known. From north to south their informal names are the Hamada el Fayoud, the Hamada el Akdamin (Williams and Hall, 1965) and the Hamada Ibn Battutah, the latter name in honour of the great fourteenth-century Arab explorer of the Sahara (Pesce, 1969). Although not as impressive in terms of relief and steepness as the Gilf Kebir, the two northerly sandstone plateaux also contain evidence of prehistoric occupation. To the best of my knowledge, the Hamada Ibn Battutah has yet to be explored for any prehistoric remains. During the Miocene very large rivers flowed between Tibesti volcanic massif to the west and the sandstone tablelands to the east (Griffin, 1999, 2002, 2006, 2011). Traces of these great rivers are very obvious on satellite imagery, and speak of a time when active and well integrated drainage systems flowed across the Sahara from the northern Chad Basin, western Sudan and western Egypt in the south to the southern Mediterranean coast in the north (McCauley et al., 1982, 1986; Drake et al., 2011). The two northern plateaux extend for 200 km between latitudes 23°50 0 N and 22°10 0 N. The plateau surface has an elevation of more than 615 m in the SW and 520 m in the NW,

236

Western Desert of Egypt and Eastern Sahara – Part 2

Figure 16.3 Late Quaternary alluvial terraces in the Hamada el Akdamin. (After the author’s field notes.)

with a mean gradient from S to N of 1:1,800. The dissected margins of the plateaux seldom rise more than 100–120 m above the gently sloping base. The upper 7–10 m consist of a bed of strongly silicified sandstone overlain by sporadic blocks of highly weathered ferruginous sandstone and clinker-like fragments of laterite. Sandford (1935) described similar lateritic remains in NW Sudan and the adjacent SE Libyan Desert, and suggested that they were formed during one or more phases of early Tertiary deep weathering under a wet tropical climate, when this region was covered in tropical forest. Beneath the resistant caprock, the current-bedded multicoloured sandstones and mudstones are highly weathered except for occasional intra-formational pebble beds and a sandstone conglomerate that crops out at the base of several of the wadis surveyed by Williams and Hall (1965). Whether this conglomerate is at the base of the Nubian Sandstone formation in this area or is simply intraformational is not known. Williams and Hall (1965) reported archaeological remains from five main sites located close to the eroded margins of the Hamada el Akdamin (Williams and Hall, 1965, Fig. 7). The artefacts included ESA/Lower Palaeolithic Acheulian hand axes and cleavers, MSA/Middle Palaeolithic blades and tanged points, LSA/Mesolithic crescents and borers, and Neolithic (and possibly younger) pottery, upper and lower grindstones, stone circles and stone hearths. A Neolithic (or younger) occupation site lies on the surface of a gravel terrace that rises 3–5 m above the floor of a wadi incised into the NW scarp of the Hamada el Akdamin (Figs. 16.1 and 16.3). Several U-shaped windbreaks or walls of unroofed dwellings over a metre high are built of large, sub-rounded boulders and are aligned along the inner edge of the 5 m terrace, while a series of stone hearths are scattered along the outer edge of the terrace (Fig. 16.3). Stone scrapers, bone harpoons, charcoal, fish (one bone suggesting catfish: Dr A. J. Arkell, pers. comm., 1964) and animal bones and an ostrich eggshell bead were found within the upper metre of one hearth, suggesting prolonged occupation. On the southern edge of this plateau a level

16.3 The Gilf Kebir, Jebel ‘Uweinat, Jebel Arkenu 15°E

20°E

25°E

237

30°E

35°E

Mediterranean

Sea

BENGHAZI 30°N

CAIRO

LIBYA ile

Re

N

N

dS

EGYPT

ea

Rebiana Sand Sea 25°N

Zumma

Aswan

Eghei massif

Kufra

Gilf Kebir

Nil

e

Bisciara

J. Uweinat

0 15°N

km

Wanyanga Kebir 500

Ennedi massif

SUDAN Wadi Zirmei

Ho

di Wa

wa

Dongola

le

CHAD

Wanyanga Serir

r

Ni

20°N

Tibesti

J. Medob KHARTOUM

Figure 16.4 The area between Kufra and Ennedi visited by Arkell in 1957 during his archaeological reconnaissance. (Modified from Arkell, 1964, Plate 58.)

area of sand was impounded between two dunes that blocked the head of a wadi, reminiscent of a similar situation at Wadi Bakht in the Gilf Kebir. A pit dug to a depth of 2.7 m revealed scrapers and borers stratified within the loamy sand deposit to the base of the pit. A small lake or pond may have been present here at least seasonally during the early Holocene. Perhaps the most intriguing find was near the head of a wadi on the plateau surface adjoining the SE edge of the Hamada el Akdamin. Here a series of small channels up to 30 cm deep, from which the coarser gravel has been removed by people to form low banks on the downslope side of the channel, run obliquely across the slope and converge on the head of the wadi. Higher up they intersect one another at right angles. The upper 10 cm of deposit within the channels is stony very fine sand; beneath that depth there is a slight increase in silt and clay content. To quote Williams and Hall (1965, p. 495): ‘Coarse aggregates of almost pure rock-salt were found localized beneath these channels, reinforcing the picture one has of a scanty rainfall, painstakingly preserved and canalized, and

238

Western Desert of Egypt and Eastern Sahara – Part 2

a high prevailing rate of evaporation.’ A further possibility besides irrigation or water conservation is that the channels were in fact deliberately contrived in order to harvest a scanty supply of salt. The Fur farmers in western Darfur obtained salt from the slopes of Jebel Marra in western Sudan before and during the 1940s by collecting and washing salty earth using a very simple form of salt works (Paterson (1948, Fig. 359: photograph by J. D. Tothill). Finally, several water holes with an estimated yield of 2,000 gallons or about 10,000 litres were located on the far NW of the Hamada el Akdamin during the height of the northern summer in August–September 1963 (Williams and Hall, 1965, Fig. 7). During times of wetter climate such sheltered rock pools or gueltas would have been more common. Another observation is worth mentioning: ‘[Just] as at Archenu, ancient trackways are ubiquitous and clearly visible along the plateau surface. In one locality, many of these paths converge upon what must once have been a sizeable edifice, now collapsed, built of slabs of hard, fine-grained sandstone similar in size to modern paving stones’ (Williams and Hall, 1965, p. 495). Beyond the confines of the plateau there is further evidence of occasional prehistoric occupation. Arkell (1964) found an Upper Palaeolithic occupation site ca. 1.5 km south of Bisciara well and both Mousterian and possible Neolithic artefacts in the Rebiana Sand Sea north of the Hamada el Fayoud. Williams and Hall (1965, p. 494) noted that ‘Palaeolithic scrapers and handaxes were found on the summits of several sandstone mesas flanking the main plateau, whereas the more recent implements were concentrated around the plateau margin and on gravel terraces within the wadis’. The reasons for this distribution are not known. During his journey in 1957 from Kufra to Ennedi via the Wanyanga/Ounianga lakes and later west from Kufra across the Rebiana Sand Sea to Zumma in the Eghei Massif, Arkell (1964) made a number of important discoveries (Fig. 16.4). He found evidence of ESA/Lower Palaeolithic sites with pebble tools as well as Acheulian hand axes and cleavers; MSA/Middle Palaeolithic sites with Levallois cores, Mousterian points and Aterian tanged points; Epi-Palaeolithic borers and backed blades; and Neolithic grindstones, pottery, ostrich eggshell beads, arrow heads and cattle paintings. In the vicinity of the shrunken lake at Wanyanga Kebir, he noted the presence of now defunct river channels, in the sediments of which there were Acheulian bifaces. Further south, at Wadi Zirmei, he recorded petroglyphs of the extinct elk-like Megaceroides, elephant, rhinoceros, giraffe, antelope, ostrich, a possible lioness, hunting dog (Lycaon pictus) and cattle. In the gorge at Archei he observed that ‘there is now in the gorge a series of five pools, with potholes in the upper stretches. There are fish in several of the pools, and crocodiles at least 6 ft. (1.8 m) long’ (Arkell, 1964, p. 14). This relic fauna had once been connected to Lake Chad during wetter climatic intervals. On the return journey he was able to visit the amazon-stone quarry at Eghei Zumma discovered by Théodore Monod during World War II. This very hard blue-green microline feldspar occurs in veins running through the weathered granite and shows evidence of having been quarried for thousands of years, possibly as far back as the Neolithic. The beads made from this stone have been recovered from the Fayum and from Khartoum, and were

16.4 Bir Sahara, Bir Tarfawi and the Tushka Lakes

239

traded over a vast area. From the scattered observations mentioned above, it is clear that once peace returns to this now arid region, future archaeological enquiry will reveal many more surprises.

16.4 Bir Sahara, Bir Tarfawi and the Tushka Lakes 16.4.1 Bir Sahara and Bir Tarfawi The sustained geo-archaeological work west of the Nile by Wendorf, Schild and their multidisciplinary teams is remarkable for several reasons. They published the results of their field and laboratory observations swiftly and in great detail in a plethora of impressively illustrated volumes complete with specialist appendices. At the same time, each successive study took critical stock of what had gone before, pointing out previous errors of interpretation and the reasons behind new insights. As a result, their publications have the allure of an ever-changing work in progress, which is of course the hallmark of all good science. I will therefore adopt something of an historical approach in attempting to review this vast body of work. We saw in Chapter 14 that the initial work by Wendorf and colleagues in the Nile Valley yielded somewhat equivocal results (Wendorf, 1968; Wendorf and Schild, 1976). The team then turned their attention to the eastern Sahara in the presently hyperarid region west of the Desert Nile at about the latitude of Aswan (Wendorf and Schild, 1980; Schild and Wendorf, 1981; Wendorf et al., 1984, 1993; Schild and Wendorf, 2001). It is to this presently hyperarid region that we now turn. The nominal precipitation amounts to no more than 5 mm, which simply means decades with no measurable rain interspersed with rare cloudbursts, which provide temporary growth from long dormant seeds and ephemeral grazing for small herbivores. The two localities generally known as Bir Tarfawi (BT) and Bir Sahara (BS) are 10 km apart and are situated 350 km SW of Kharga Oasis and 400 km west of the Nile at latitude 23°N (Fig. 15.1). They both occupy broad shallow depressions caused by wind erosion and deflation. Such deflation hollows are common in the eastern Sahara (e.g., Glennie, 1970, Map 4). Bahay Issawi and his colleagues from the Geological Survey of Egypt mapped the geology of the vast region extending from the Gilf Kebir and Jebel ‘Uweinat in the west to Bir Sahara and Bir Tarfawi in the east. He provides a graphic description of the latter region: ‘The area is one of the most utterly desolate parts of the earth. No vegetation, no traces of any kind of life, absolutely nothing but vast endless sheets of sands and the sun. Only at Bir Tarfawi and Bir Sahara does seepage from underground water sustain clusters of trees. At Tarfawi, a corridor of date palms, dom palm, acacia and tamarisk trees extends 15 km . . . in . . . discontinuous strips. At Bir Sahara, clumps of tamarisks are often growing on a sand dune surface for a distance of eight kilometres’ (Issawi, 1981, p. 20). As we shall see, it was not always thus, particularly during Last Interglacial times, when Middle Palaeolithic groups lived around the lakes that once flourished in this area (Wendorf et al., 1993). Issawi (1981, 1993) has described the geology and geomorphology of this

240

Western Desert of Egypt and Eastern Sahara – Part 2

region. The Basement Complex granites and granite gneisses (see Chapter 4) are overlain unconformably by two sedimentary formations of similar lithology (sandstone) but of different age. West of Bir Sahara the sandstone is of Palaeozoic age. To the east continental Upper Cretaceous Nubian Sandstone rests directly on the Basement rocks. To the east and north Lower Eocene shallow marine clays, sands and limestones cover parts of the sandstone plateau surface and cap the scarp summits. The oldest Quaternary sediments are the ‘Basal Sands and Gravels’. The age of these sediments is not known. They are older than the Bir Sahara (BS) and Bir Tarfawi (BT) depressions, which are eroded into them, and appear to mantle at least some of the large and now defunct ‘radar channels’ identified by McCauley et al. (1982) using ground-penetrating radar. Again, the age of these channels is not yet well established. A series of younger braided channels is in places inset within the far wider ‘radar channels’ and locally contain Middle Acheulian cleavers and handaxes; these artefacts are likely to be ca. 0.5–1.0 Ma old. The oldest lake and spring sediments at BS and BT contain Late and Final Acheulian stone tools, suggesting an age of >300 ka, but are otherwise not dated. A small playa lake at BT occupied a deflation hollow eroded into a former lacustrine carbonate layer; it was fed by local runoff and has a maximum age of 175 ka. The fossil fauna (gazelles) and ephemeral nature of this seasonal playa lake imply a local precipitation of ca. 50–200 mm (Gautier, 1993). Five successive lake stages are so far evident at both BT and BS, each occupying shallow deflation hollows eroded into the preceding set of lake deposits during ensuing arid intervals, but each located within the same depressions. The lakes are considered as being a single lake that dried out intermittently, during which some of the pre-existing lake sediment was blown away, especially the former lake margins and beaches, so that often only the deepwater sediments have been preserved. Each lake has some MSA/Middle Palaeolithic material preserved within or on its sediments as well as remains of birds, rodents, vertebrates, reptiles and fish, in varying amounts. Dating the lakes has been hard. Five independent dating methods (luminescence, both thermoluminescence and Optically Stimulated Luminescence [OSL]), uranium-series, amino acid racemisation, and electron spin resonance), gave a scatter of ages between ca. 500 ka and ca. 30 ka, although the 30 ka ages are 14C ages, and are far too young. Only the OSL ages seem to have yielded stratigraphically consistent results (Wendorf et al., 1993b, pp. 552–573). About all we can safely conclude is that the inception of the Saharan Middle Palaeolithic is probably no older than 230 ka, so that the Saharan Acheulian industries would be older than ca. 230 ka, a conclusion consistent with that of Clark et al. (2008b). Philibert et al. (2010) have also discussed attempts to date Saharan Pleistocene lakes (see Chapter 13). Wendorf et al. (1993b) tentatively considered that the Middle Palaeolithic wet phase represented by the ‘White Lake’ at BT could either belong within MIS 7 (starting at ca. 190 ka) or, possibly, late in MIS 6 (see Chapter 4, Table 4.2). The trace element studies of ostracods from this lake show that the lake was permanent, with fresh to slightly saline

16.4 Bir Sahara, Bir Tarfawi and the Tushka Lakes

241

water, and at the site analysed relatively shallow with fluctuating temperatures (De Deckker and Williams, 1993). The ‘West Lake’ at BS was also a ‘fairly well developed lake, fresh to slightly saline, with permanent water, deep at times’ (De Deckker and Williams, 1993, p. 117). A long arid interval separated the two lakes or sets of lakes, with the West Lake being tentatively assigned to MIS 5 (Wendorf et al., 1993b, p. 566), which is plausible. Gautier (1993) described the BT fauna from one lake and noted its resemblance to a modern savanna fauna. He concluded that the regional environment included dense reed beds around the lake set amidst a savanna grassland with scattered acacias frequented by rhino, giraffe, large bovids and antelopes large and small. He inferred a tropical summer monsoonal rainfall regime amounting to ca. 400 mm. The presence of crocodiles and Nile perch (Lates niloticus) in certain of the deeper lake sediments raises the question of provenance – and whether it was from the Nile, Chad Basin or Niger Basin. The recently mapped Tushka Lakes west of the Nile throw some light on this question and indicate a Nile connection.

16.4.2 The Tushka Lakes Using topographic data obtained from the most recent Space Shuttle Radar Topography Mission (STRM), Maxwell et al. (2010) identified a series of linear topographic depression, which they called the Tushka Lakes (Fig. 15.1), that appear to have formed part of a once integrated drainage system flowing northeast from Selima Oasis in northern Sudan (see Chapter 15) towards the Nile valley in southern Egypt. They note that these channels fade out before reaching the Nile, near the margin of a broad depression that they interpret as a former lake with a surface elevation of ca. 247 m. We described earlier (Section 16.4.1) that at Bir Tarfawi some 400 km west of the Nile a series of Middle to Late Pleistocene palaeolakes associated with Acheulean and Middle Palaeolithic artefacts and associated fauna have in places yielded Nilotic fish remains. The fish fossils have been discussed in detail by Van Neer (1993), who suggested a possible Nile connection to account for the presence of Nile perch (Lates niloticus) and other nilotic species (Van Neer, 1993, pp. 152–154). The ca. 247 m surveyed elevation of one of these lakes (Schild and Wendorf, 1993, pp. 20–21: Figs. 3.4 and 3.5) is the same as that revealed by the SRTM data. Maxwell et al. (2010) concluded that a former lake occupied much of the valley to the west of the Nile up to the 247 m contour. What is less clear is the age of this lake, although a Last Interglacial age seems possible. We have seen that little can be definitively concluded concerning the ages of the suite of lakes at Bir Sahara and Bir Tarfawi associated with Acheulian and Middle Palaeolithic stone artefacts, mapped by Wendorf et al. (1993a). Based on the elevation of the Wadi Tushka spillway towards the Nile, Maxwell et al. (2010) identified a second and lower lake at an elevation of ca. 190 m. At Bir Kiseiba, Wendorf and Schild (1984) had earlier mapped the distribution of ‘Terminal Palaeolithic’ and Neolithic sites (Wendorf and Schild, 1984, Fig. 1.2), with most sites

242

Western Desert of Egypt and Eastern Sahara – Part 2

bracketed between 200 m and 180 m, the Neolithic Holocene sites occurring lower in the landscape. Maxwell et al. (2010) speculate that overflow from the Nile during times of high floods was one of the sources of water to feed the 247-m and the 190-m lakes, which seems quite possible. Recent evidence from northern Sudan has also revealed a former lake of early Holocene age in receipt of Nile floodwaters (see Chapter 14 and Williams et al., 2010).

16.5 Saharan Groundwater Recharge during the Quaternary During wetter intervals in the Quaternary the vast groundwater aquifers in the eastern Sahara were replenished. Two critical questions relating to these intervals of groundwater recharge are when did the recharge occur and what was the source of precipitation? The answers to these questions are not straightforward and are based on a growing body of evidence from stable isotope geochemistry. The focus on Holocene prehistoric sites in the eastern Sahara and along the Nile has tended to colour our perspective on past sources of precipitation, leading to an overemphasis on former movements of the ITCZ and on summer monsoonal rainfall. For example, Kuper and Kröpelin (2006) mapped the distribution of some 150 dated archaeological excavations in the presently hyperarid eastern Sahara. They concluded that initial settlement in this region followed the relatively sudden inception of a wetter climate towards about 10.5 ka. Desiccation set in gradually from about 7.3 ka onwards, leading to early abandonment of the most northerly sites and later abandonment of the more southerly sites. The age and distribution of the archaeological sites is consistent with a progressive retreat southwards of the northern front of the tropical summer monsoon. Thirty years earlier Haynes (1987) had reached a somewhat similar conclusion in regard to time-transgressive changes in the Holocene wetting front, except that his concern was with the onset of wet conditions in the eastern Sahara and not their demise. His argument is based upon the time when early Holocene lakes began to form in the region he termed the Darb el Arba’in Desert (see Chapter 15), which presently receives less than 10 mm of mean annual rainfall – a notional figure, given the high degree of annual variability. He concluded that the ‘model of the [early Holocene] migration of the monsoonal circulation is supported by the south-to-north trend of decreasing ages among pluvial lakes of the Arba’in Desert’ (Haynes, 2006, p. 79). Haynes estimated that the rate of northward advance of the early Holocene wetting front amounted to between 0.15 and 0.83 km/yr, with a mean rate of about 0.63 km/yr the most plausible (Haynes, 2006, p. 79). A different picture emerges when we consider a much longer time frame than that of the Holocene. Osmond and Dabous (2004) obtained 230Th/234 U ages for times of enhanced groundwater movement in the Egyptian Sahara. They analysed secondary uranium (U) in iron ores and carbonates (tufa and speleothems) from Kharga Oasis, Bahariya Oasis and Wadi Sannur in the Western Desert of Egypt as well as at four sites in the Eastern Desert and

16.5 Saharan Groundwater Recharge during Quaternary

243

along the Red Sea coast. Their overall conclusion was that times of wetter climate were evident during marine oxygen isotope stages 4, 5, 6 and 7 (MIS 4, 5, 6 and 7: see Chapter 4). From this they inferred that the wet phases were the result of migration of the tropical monsoonal belt driven mainly by the 23 ka precessional cycle (see Chapter 4). In fact, as they acknowledge, their chronology is imprecise – too imprecise, indeed, to demand this interpretation. A different approach is needed to resolve this conundrum and is provided by two earlier studies. As expected, the reality is both more complex and more interesting. Sultan et al. (1997) investigated the stable isotopic composition of Pleistocene groundwater and carbonate deposits in the Western Desert of Egypt. They obtained ages for the tufa deposits of >450 ka, 270–290 ka (MIS 8), 160–190 ka (MIS 6) and ca. 45 ka. An unexpected outcome of their work is that westerly air masses from the Atlantic were the source of the precipitation responsible for replenishing the groundwater and for providing water to enable the spring deposits to form. This could imply that replenishment occurred during glacial episodes when the dominant westerlies extended further to the south, bringing winter rain to North Africa, and tallies with the MIS 6 and MIS 8 ages for two of the tufa formations. Possible independent confirmation for this conclusion may come from the work of Geyh and Thiedig (2008) who obtained 230Th/U ages of >420 ka, 380–290 ka, 260–205 ka and at least 140–125 ka for northern Sahara lacustrine carbonate deposits in the Murzuq Basin of Libya. Although these ages coincide with humid Middle Pleistocene interglacial phases, the authors noted that the intervening arid phases ‘might sometimes have been interrupted by very short humid episodes in North Africa during Pleistocene glacial periods (MIS 10 and MIS 6)’ (Geyh and Thiedig, 2008, p. 19). However, they did caution that ‘the reliability of the corresponding 230Th/U ages does not, however, exclude the possibility that these episodes are artefacts of an erroneous chronological framework’ (Geyh and Thiedig, 2008, p. 19). They also found that the four humid episodes identified decreased progressively in intensity over time, a conclusion in accord with the work of Szabo et al. (1995) in the eastern Sahara and Western Desert of Egypt. An earlier study by Joseph et al. (1992) of the oxygen and hydrogen stable isotopic composition of groundwater and meteoric precipitation in the Sahelo-Sudanian zone reached some interesting conclusions. Water precipitated over the Ethiopian uplands comes from the Indian monsoon. The East African jet then transports these air masses westwards across the Sahelo-Sudanian zone. These air masses are recycled vertically as a consequence of evapo-transpiration and convection. The squall lines at the start of the wet season mobilise water vapour that comes from the Atlantic in the lower layers but from the east in the upper layers. Subsequent squall lines make use of vapour previously mixed in the lower layers of the atmosphere. Overall, the isotopic decrease in 18O in rainwater and groundwater shows a gradient from east to west. The water vapour comes from evaporation over the Indian Ocean and is transported from east to west across the southern Sahara and Sahel zones by the East African Jet, the Tropical Easterly Jet and Easterly Waves.

244

Western Desert of Egypt and Eastern Sahara – Part 2

The work of Sultan et al. (1997) indicates that groundwater in the northern Sahara was replenished by moist air masses from the Atlantic that moved eastwards across North Africa most likely during glacial intervals when ice caps over NW Europe had diverted the rain-bearing winter westerly winds further south across North Africa. In contrast, Joseph et al. (1992) have shown that in the southern Sahara and Sahelo-Sudanian zone, groundwater was replenished by moist air masses of Indian Ocean provenance moving westwards from the Ethiopian uplands across the tropics, most likely during times of stronger summer monsoon. The respective northern limits of the tropical summer rainfall belt and the southern limits of the temperate winter rainfall belt would have oscillated back and forth over time, with westerly air masses dominant in the northern Sahara during glacial phases and easterly air masses dominant along the southern Sahara during interglacial phases. Times of groundwater replenishment would thus have been out of phase in northern and southern latitudes. In central latitudes, some replenishment may have taken place during alternating phases of strong summer monsoon and strong westerly airflow.

16.6 Late Quaternary Environments in the Sahara: Implications and Cautions We have seen in the preceding sections that there is strong evidence in the form of dated tufa deposits, now defunct river channels, and lake sediments with their associated microfossils (diatoms, ostracods, cladocera) that the eastern Sahara was more humid on a number of occasions during the Quaternary. The most recent of the late Pleistocene wet intervals was during the Last Interglacial, which reached its peak 125,000 years ago (125 ka: MIS 5e). Kieniewicz and Smith (2007) conducted stable isotope and minor element analyses of calcite silts and associated Melanoides tuberculata shells from a wadi in Kharga Oasis that they attribute to MIS 5e. They found that the spring water that fed the ponds, in which the silts accumulated and the gastropods lived, remained fresh throughout the year. The groundwater feeding the springs was derived in the main from the Atlantic, with a minor component of possible Indian Ocean provenance. McKenzie (1993) had earlier studied the isotopic composition of calcite precipitated in some of the Bir Sahara lakes attributed by Wendorf et al. (1993) to MIS 5e. She concluded from the virtual absence of detrital particles within the carbonate that the plant cover around the lake was sufficiently dense to preclude erosion. When the lakes dried out, they did so very rapidly, as a result of the southward displacement of the ITCZ. The oxygen isotopic ratios (16O/18O) of the calcite and hence of the lake waters were consistently negative, suggesting that the lakes received rainfall from the southwest monsoon. Many authors have commented on the remains of relict drainage systems adjoining the Desert Nile, often evident as sinuous ridges capped with alluvial gravels, which are variously cemented with iron, carbonate or silica (Butzer and Hansen, 1968, pp. 229–230; Embabi, 2004; Zaki and Giegengack, 2016; Zaki et al., 2018). These ridges result from an inversion of the former relief, in which the alluvial gravels proved more

16.6 Late Quaternary Environments in the Sahara

245

resistant to erosion than the softer sediments adjoining the former channel. Their age is uncertain; some of the gravel ridges contain rolled Acheulian bifaces, which could indicate an age almost anywhere between ca. 1.5 Ma and ca. 0.3 Ma. Some of these channels such as the Wadi Kurkur tributary of the Nile were fed by springs at a time when the groundwater level was higher and the climate wetter than at present. They point to a time when the drainage system in the vicinity of the Nile was more integrated than it is today (Nicoll and Sallam, 2016). Remote sensing techniques, including the use of Shuttle Imaging Radar, have revealed a whole series of buried or partially buried ancient river channels. Drake et al. (2011) compiled maps of these former drainage basins and associated lakes, and concluded that dispersal of people across the ‘green Sahara’ would have been possible during the Last Interglacial at ca. 125 ka, allowing hunter-gathering groups to move out of Africa at that time. Castañeda et al. (2009) analysed the carbon isotopic composition of leaf waxes in a marine sediment core collected from latitude 9°10 0 N off the coast of West Africa and within the path of the Harmattan dust plume. The leaf waxes can be used to distinguish between plants that follow the C3 and C4 photosynthetic pathways (see Williams et al., 1998, pp. 238–239; Williams, 2014, pp. 106–110). During the early to mid-Holocene, and again towards 45–50 ka and 110–120 ka, the leaf waxes indicated that C3 plants were common inland, suggesting that trees may have been present in parts of the southern Sahara at those times. They went on to conclude that movement across the Sahara was possible at these times, which is plausible but not actually demanded by the evidence they present. Coulthard and his colleagues adopted a more critical approach to the question of whether rivers did indeed flow across the Sahara during the Last Interglacial (Coulthard et al., 2013). Using a combination of hydrologic and hydraulic modelling, they found that the river system offering the most likely migration route for early humans was not the Sahabi and Kufra palaeo-drainage basins identified by earlier workers but the Irharhar River that flows from north to south today in the west-central Sahara and is ephemeral. Once again, this theoretical approach needs to be tested against well-dated geomorphic and archaeological data from the localities concerned. There is considerable interest in just how prehistoric humans might have responded to environmental change, especially if such change was detrimental to sources of food and water (Brookfield, 2010; Cremaschi et al., 2010; Gatto and Zerboni, 2015). Did the human groups affected migrate, for instance from the desert to the river – from Sahara to Nile? Did they adapt? Or did they become extinct? One aspect often raised in such discussion concerns the apparent rapidity with which such changes may have occurred (Fontes et al., 1985). The record from rivers and lakes in many parts of the Sahara has been viewed, somewhat myopically, as indicating desiccation setting in quite rapidly from about 5.5 ka onwards. A key culprit behind this notion derives from a single marine sediment core taken off the coast of Mauritania, which showed a decline in dust input at 14.8 ka and an increase in dust flux at 5.5 ka (De Menocal et al., 2000). The interval 14.8 ka to 5.5 ka was dubbed ‘the African Humid Period’ (note the definite article), which is both a misnomer and very misleading – a misnomer because many parts of Africa were not more humid at that time,

246

Western Desert of Egypt and Eastern Sahara – Part 2

and misleading because it ignores the influence of a multitude of local factors upon the hydrologic and ecological responses to regional fluctuations in climate (Rognon and Gasse, 1973; Street, 1980). Stefan Kröpelin and his team offer a salutary rejoinder to this erroneous concept in a paper deemed of such interest (Kröpelin et al., 2008) that it was reviewed at length in Le Figaro five days after its publication in Science on 9 May 2008 (Miserey, 2008). Kröpelin et al. (2008) obtained two sediment cores from the upper 7.47 m of deposit in Lake Yoa, a small and hyper-saline permanent lake at Ounianga visited briefly by Arkell in 1957. This lake is located about halfway between Tibesti volcano and the Ennedi massif, and lies in the path of the NE trade winds, which are blowing sand dunes into the northern sector of the lake. The finely laminated lake silts contain pollen, ostracods, diatoms and cladocera and span the last 6,000 years. Aquatic productivity increased in Lake Yoa at about 5.6 ka and remained high until ca. 3.3 ka. Between 4.2 and 3.9 ka the lake changed from fresh to saline. Until 4.3 ka the surrounding landscape was savanna grassland with scattered Acacia trees, and rivers flowing down from Tibesti brought in pollen from montane shrubs such as Erica arborea. This influx ceased at ca. 4.3 ka, denoting curtailed runoff from Tibesti. A change from savanna to semi-desert vegetation had already begun at 5.6 ka, and accelerated after 4.8 ka when the tropical trees vanished. Desert plants appear at ca. 2.7 ka, together with an influx of sand from the north and establishment of the present NE trade wind regime, with winds channelled southwards between the two mountain masses of Tibesti and Ennedi for much of the year. The overall conclusion from this detailed study is that desiccation in the second half of the Holocene was gradual, not abrupt. Furthermore, different ecosystems, whether aquatic or terrestrial, responded at different times and in different ways. The main lesson to be drawn is that we cannot ignore the influence of local factors, such as topography, soils and runoff, in modulating the response of ecosystems to regional changes in climate.

16.7 Conclusion The Western Desert of Egypt is the driest region adjoining the Nile and is part of the hyperarid eastern Sahara. In spite of its present-day aridity and absence of life, the Western Desert contains tantalising evidence of at least episodic human occupation from Lower Palaeolithic (notably, Acheulian) times onwards. In addition, a number of large former lakes were present in the vicinity of Bir Tarfawi and Bir Sahara. At Bir Tarfawi and Bir Sahara located some 400 km west of the Nile a series of Middle to Late Pleistocene palaeolakes associated with Acheulian and Middle Palaeolithic artefacts and associated fauna have in places yielded Nilotic fish remains. Attempts to date these lakes have proven inconclusive, yielding a scatter of ages from ca. 500 ka to 1000m

RE

N 0

D

Nil e

Quseir

10

N

km

(c) (b)

ARABIA

N

RE

A

SE

Aswan

0

SAUDI

Quseir (c)

100 200 km

(b)

EGYPT

25°N

EA

JO RD A

N

Land > 2000m

(b)

DS

I SR

AEL

40°E

RE

30°E MEDITERRANEAN SEA

D

Lake Nasser

Nil

e

Kom

alluvial KOM OMBO

A

LS

SE

HIL

20°N

di Wa

Khari

t

plain

Nile (a)

0

5 km

ra

ba

At

SUDAN

Wad Shait i

Ombo

Figure 18.1 The Red Sea Hills showing places mentioned in the text.

ring dyke intrusions (Whiteman, 1971; Vail, 1978; Said, 1993; Geological Research Authority of the Sudan and Robertson Research International Limited, 1995). These Basement Complex rocks form part of the Arabian-Nubian Shield and underwent uplift and erosion at intervals from early Cenozoic times onwards as the northern extension of the Afro-Arabian Swell (Almond, 1986). The final phase of uplift was associated with the opening of the Red Sea that began about 25 Ma ago and was associated with subsidence and rifting along the axis of the Red Sea and concomitant uplift of the rift margins (Said, 1993; Embabi, 2004; F. M. Williams, 2016). The process of rifting in the Red Sea was initiated during the Late Oligocene, which was also a time of uplift of the Ethiopian Highlands and associated widespread eruptions of basaltic lavas, and was largely complete by the Middle Miocene, although widening of the Red Sea is presently occurring at a rate of about 1.5 cm/ yr (F. M. Williams, 2016). The Cenozoic uplift of the Red Sea Hills was accompanied by widespread erosion and removal of the uppermost Eocene limestone followed by removal of the Mesozoic and Palaeozoic sedimentary cover rocks, leading to exhumation of the underlying Precambrian basement rocks (Macgregor, 2012; Hamdan and Brook, 2015). This process of cover rock removal was assisted by deep weathering during the Miocene and Pliocene, when the tropical climate was wetter than today (Hamdan and Brook, 2015). The present topography is one of deep valleys and intensely dissected plateaux.

18.3 Pleistocene Rivers Flowing from Red Sea Hills

259

18.3 Pleistocene Rivers Flowing from the Red Sea Hills During wetter intervals in the Quaternary, major rivers flowed both east and west from the Red Sea Hills, bringing sediment to the Nile in the west and the Red Sea to the east. Today, flow from the channels within these desert wadis is strictly ephemeral, and may only occur about once a decade or less often (Butzer and Hansen, 1968, pp. 68–69). In the case of the wadis that drain east towards the Red Sea, the most recent phase of wadi aggradation (sandy gravels and cobbles) coincides with a time of low sea level (Butzer and Hansen, 1968, p. 429), indicating a wetter than present climate in this region during MIS 2 (see Chapter 4, Table 4.2). In no case do the Eastern Desert wadi deposits merge into the series of coral reefs adjoining the Red Sea at elevations of +10 m, +5 m, +6 m, +8.5 m and +3.5 m above present sea level (Butzer and Hansen, 1968, p. 429). The reefs formed during warmer and wetter interglacial periods when sea level was as high as or higher than it is today. In fact, the wadis draining from the Red Sea Hills to the Red Sea have cut deep and narrow valleys through the fringing reefs that run parallel to the present shoreline. The very coarse nature of the alluvial gravels points to greater runoff and a rainfall amounting to about 100 mm a year on the uplands (Butzer and Hansen, 1968, p. 426). A more detailed account is available for certain of the wadis that once flowed west from the Red Sea Hills into the Nile, notably Wadi Shait and Wadi Kharit that merge together on the Kom Ombo alluvial plain on the eastern side of the Nile to the north of Aswan (Butzer and Hansen, 1968, Fig. 4–4). These authors identified a number of episodes during the Lower and Middle Pleistocene when wadis flowing west from the Red Sea Hills brought down thick deposits of coarse gravels and cobbles derived from Basement Complex rocks, often drowning out the finer sediments deposited by the Nile. However, no chronometric ages are available for any of these deposits. The situation is somewhat clearer in the case of Late Pleistocene and Holocene wadi gravels, with ages between ca. 17–14.5 ka, 14–13 ka and 11.5–8.5 ka indicating coeval Nile aggradation and accelerated wadi activity in the Red Sea Hills (Butzer and Hansen, 1968, Table 3-7, p. 149). Lake Tana in the Ethiopian headwaters of the Blue Nile was dry during the Last Glacial Maximum (23–19 ka) but began to refill from 17 ka onwards, reaching its highest level at ca. 14.5 ka (Lamb et al., 2007; Marshall et al., 2011 and Chapters 6 and 11). Lake Victoria in the Ugandan headwaters of the White Nile has a similar history (Williams et al., 2006 and Chapters 7 and 8). High flow in the Desert Nile at these times reflects the impact of a stronger summer monsoon. It is hard to say whether the accelerated wadi activity in the Red Sea Hills during the times identified by Butzer and Hansen (1968) is a result of somewhat higher rainfall in summer, in winter, or in both seasons. The work by Hamdan and Brook (2015 and Section 18.4) on tufas in the Egyptian Red Sea Hills lends support to the inference of a more northerly influx of moist air masses from the Indian Ocean, although this does not preclude the possibility of an influx of moist air from the north during the winter. The question is best left open.

260

The Red Sea Hills

18.4 Pleistocene and Holocene Spring Tufas and Their Climatic Significance When fossil tufa deposits are found in presently hyperarid areas, they constitute evidence of a previously wetter climate in which groundwater levels were higher and fed springs and streams flowing from the springs. Tufas form when calcium bicarbonate dissolved in groundwater is precipitated as calcium carbonate as a result of carbon dioxide degassing (Butzer and Hansen, 1968; Pentecost, 2005; Goudie, 2013; Williams, 2014, chapter 14). Such degassing is common where groundwater emerges above ground in springs to form spring tufas. Another common location for tufas to accumulate is at waterfalls, where carbon dioxide degassing is fostered by the turbulence. In a major contribution to our knowledge of the climatic history of the Egyptian Red Sea Hills and Sinai Desert, Hamdan and Brook (2015) analysed 58 samples from spring and waterfall tufa deposits collected from 11 sites in the Sinai Desert and the Eastern Desert. They used a combination of 14C dating, petrography, and evaluation of the stable carbon and oxygen isotopic composition of selected tufa samples to reconstruct past changes in temperature and humidity. They found that the tufas had developed during two distinctly wetter phases, the most recent dating to the early to mid-Holocene (12–6.7 ka) and an earlier phase of late Pleistocene age (ca. 31.2–22.5 ka). A single tufa sample also yielded a possible age of ca. 62–56 ka, but additional samples will be needed to confirm whether this was indeed a significantly wetter time in this region. The late Pleistocene tufas have δ18O values that are more depleted than the corresponding Holocene values. The authors inferred from this and other previously published data that during the late Pleistocene the tufas were fed by shallow ground waters replenished from precipitation derived from the Mediterranean at a time when colder temperatures over Europe had driven the westerly wind belt further south. These late Pleistocene tufas were inferred to have been deposited at temperatures within the range 14.0°–20.8°C, compared to the Holocene temperature range of 18.4°–23.4°C. They also concluded that the Holocene precipitation was associated with the Indian Ocean summer monsoon and with somewhat warmer temperatures. In this context, it is worth noting that Arz et al. (2003) concluded from analysis of the stable isotopic composition of microfossils from two marine sediment cores retrieved from the Red Sea that Late Pleistocene Red Sea surface temperatures were 4°C lower than during the Holocene. The absence of Holocene tufas between latitudes 28°N and 25°N seems to mark the most northerly limit in the Eastern Desert to which summer monsoonal rains were able to penetrate. These latitudes are also the driest in the Red Sea Hills today. In contrast, late Pleistocene tufas occur at intervals throughout the Eastern Desert down to 24°N, indicating that moist air masses were able to penetrate considerably further south than they do today. Because tufa non-organic carbonate 14C ages can differ, often significantly, from 14 C ages obtained from organic carbon materials within the tufa, such as plant remains, Hamdan and Brook (2015) were at pains to obtain ages for both types of carbonate from some of the tufa samples. One conclusion they drew from the difference in age between the two types of material is that during brief wet interludes

18.6 Wetter Climate in Red Sea Hills 2,000 Years Ago

261

within otherwise dry periods, conditions were locally sufficiently wet for plant growth at tufa sites but not wet enough for any widespread tufa carbonate precipitation.

18.5 Mesolithic and Neolithic Occupation in the Red Sea Hills Surprisingly few Holocene archaeological sites have yet been fully investigated in the Red Sea Hills. One exception is the Tree Shelter site (26°16.332 0 N, 33°57.379 0 E) located in a tributary valley of the Wadi Sodmein 25 km inland from Quseir on the Red Sea coast of Egypt. The present climate at the site is hyperarid, with a single solitary acacia at the entrance to the valley (Marinova et al., 2008). Moeyersons et al. (1999) report occasional winter rains. Marinova et al. (2008) recovered roughly a thousand faunal remains from the site, together with 436 fragments of charcoal. The animal remains included sheep (Ovis ammon f. aries), goat (Capra aegagrus f. hircus), Barbary sheep (Ammotragus lervia), a small bovid, Dorcas gazelle (Gazella dorcas), small lizards, a small snake, a small rodent, ostrich eggshell and Red Sea fish. The charcoal indicated the former presence of Tamarix, Salvadora persica, Maerua crassifolia, Balanites aegyptiaca, Boscia salicifolia, Capparis decidua and Ziziphus. Ten radiocarbon ages (Marinova et al., 2008) show that the sediments at the Tree Shelter site were laid down between 8,000 and 4,900 14C years BP (equivalent to ca. 9.0–5.8 ka). The early Holocene rains were torrential and highly erosive, but the vegetation was sufficiently abundant to offer food to the animals that were hunted during the Epipalaeolithic/Mesolithic. After 8.0 ka the rainfall regime seems to have become more regular and less intense. Both here and in the nearby Sodmein cave, ovicaprine bones are fairly sparse, but this apparent lack is contradicted by the abundance of dung from the domestic Neolithic herds inside the cave (Linseele et al., 2014, p. 9). Although incursions of rain during both summer and winter supported a semi-desert flora, it seems likely that the Neolithic herders visited these sites on an episodic or seasonal basis, as they moved with their herds in search of good grazing for their animals, much as is the practice today among the Hadendowa nomads in the Red Sea Hills to the south, near the border between Sudan and Eritrea.

18.6 A Wetter Climate in the Red Sea Hills 2,000 Years Ago Shortly after about 60 BC, the Greek historian Diodorus Siculus described cave-dwelling cattle herders in the Red Sea Hills who carried out cattle raids and fought each other over pasture before retreating into the safety of their upland swamps (Jackson, 1957, p. 58). According to Diodorus, these people lived on milk and blood from their cattle, reminiscent of the present-day Maasai cattle herders in southern Kenya and northern Tanzania (Western, 1997). They also made an intoxicating drink from the fruit of Ziziphus, which is still found in sheltered valleys in the Red Sea Hills. Ziziphus berries are presently collected and eaten across the central Sudan and at least as far west as central Niger, where they are a prized food in times of famine, as witnessed by the author in December 1973. What is especially

262

The Red Sea Hills 34°E

36°E

EGYPT

38°E 0

200

km

Re

22°N

100

d RED

N a Se

NUBIAN DESERT 20°N

IV

Port Sudan Suakin Erkowit

Ni

le

Erkowit

SUDAN

SEA

u

Wadi Amur Berber

Hi lls

18°N

ea

at

Pl

V BAIYUDA DESERT

Nile

Atbara

ra

ba

At

X Meroe Shendi X X VI X 16°N Naga

Kassala

Khartoum

ERITREA

Land > 1000m VI

Cataract

Figure 18.2 The Red Sea Hills, NE Sudan, showing location of Erkowit. Crosses are Meroitic sites (ca. 2 ka); dotted lines are major wadis; land above 1,000 m is dotted; Roman numerals are Nile cataracts. (Adapted from Mawson and Williams, 1984, Fig.1.)

interesting in regard to the account by Diodorus is that there are no swamps in the Red Sea Hills today. The Erkowit plateau rises to over 1,200 m (Fig. 18.2) and receives frequent winter fogs from the Red Sea that provide enough moisture to support scattered Euphorbia candelabra and other evergreen Ethiopian plants (Andrews, 1948; Barbour, 1961). Mean annual rainfall at Erkowit amounts to about 300 mm. Tothill (1946b) visited the mist oasis of Erkowit in June 1941 and collected aquatic snail shells from one of the ephemeral stream channels near the former WW2 golf course as well as shells of land snails from the slopes adjacent to the channel. By analogy with the shells he had collected from the Sudan Gezira (Tothill, 1946a) he concluded that the Erkowit shells were of early Holocene age and indicative of a wetter climate. Mollusc shells identified in the June 1941 Tothill collection

18.6 Wetter Climate in Red Sea Hills 2,000 Years Ago

263

Melanoides tuberculata (Müller) cf. dautzenbergi Pilsbry

gb

Perennial flow valley-bottom silts and clays b

Biomphalaria pfeifferi

(Krauss)

Lentorbis junodi (Connolly) Gyraulus cf. connollyi Brown and Van Eeden (lateral view)

g Ca

Lymnaea natalensis (Krauss)

dg

Perennial flow valley-bottom clays and sandy clays

db

Il

g

Da Ca Il Cy

b b

Ephemeral or seasonal flow : upward fining sequence Ostracods and Chara

Sub-fossil mollusca

Bedding and other features

g vcs cs ms ts silt clay

b

Figure 18.3 Stratigraphy and local geomorphic context of the Erkowit shell-bearing clays. The ostracods are Ilyocypris sp. (Il), Cyprinotus sp. (Cy), Canodopsis sp. (Ca) and Darwinula sp. (Da). Broken arrows denote scattered distribution. The star denotes Chara sp. A and B. Bedding is massive (open dashes) or flat (closed dashes). Colour symbols denote brown (b), grey (g) and dark (d). The lithological symbols are clay as black, silt as dashes, sand as dots, and clay-sand mixtures as dots and wavy lines. (Adapted from Mawson and Williams, 1984, Fig. 2.)

were Xerophila sp., Cerastus abyssinica Pfv., Planorbis herbini Byt, Bulinus truncatus Aud., Limnaea caillaudi Boury and Melanoides tuberculata Mull. In January 1973, I visited Erkowit and sought directions from the local Beja (Hadendowa) pastoralists as to the whereabouts of any shell-bearing sites within ephemeral stream channels. They suggested that I should investigate Khor Harasab (Figs. 18.3 and 18.4), where up to 6 m of shell-bearing clays overlie coarse angular gravels, and advised me to examine a well dug recently into the alluvial sediments. Shells were scattered throughout the upper 4 m of alluvium in the well (Fig. 18.3). The mollusc species identified in the January 1973 Williams collection were Melanoides tuberculata (Muller) cf. dautzenbergi Pilsbry, Lentorbis junodi (Connolly), Biomphalaria pfeifferi Krauss, Lymnea natalensis Krauss and Gyraulus cf connollyi Brown & Van Eeden. The ecological requirements of each of these species are well established (Brown, 1965, 1980, 1994). Melanoides is an aquatic river or lake bottom species. Lentorbis junodi is found in slow-moving streams and marshes. Both Biomphalaria pfeifferi and Gyraulus connollyi live in stony and vegetated

264

The Red Sea Hills

Figure 18.4 Valley-fill sediments, Khor Harasab, Erkowit, Sudan.

streams. Lymnaea can move outside water but its eggs dry up if removed from water, so that reproduction is only possible in water. It can therefore be regarded as aquatic. The shells of four types of ostracod (Cyprinotus, Darwinula, Canodopsis and Ilyocypris) were also found in the fine-grained alluvial sediments. Most species of Darwinula and Cyprinotus inhabit fresh water, although some species can live in saline (oligohaline to mesohaline) water. Canodopsis is found in fresh water, while most species of Ilyocypris live in wet mud (Mawson and Williams, 1984). Fossil Chara were found at a depth of 4 m. Non-marine charophytes grow submerged in fresh standing water on a base of sand or mud and can form extensive submerged meadows. Four samples of Melanoides tuberculata from near the base and near the top of the alluvial silts and clays were selected for radiocarbon dating (two conventional and two AMS) and each sample was tested for recrystallisation using X-ray diffraction. The calibrated 14C ages were 1,840 ± 130, 1,620 ± 180, 1,680 ± 50 and 1810 ± 50 years BP. The combined fossil and sedimentary evidence indicates permanent stream flow across swampy meadows, with streams depositing sand and silty clays along their channels. The climate on the Erkowit plateau was therefore wetter than today between ca. 1.8 and 1.6 ka, and most probably for several centuries before then. To quote Mawson and Williams (1984, p. 49): ‘More enduring rains during both winter and summer may have raised groundwater levels to the point where perennial base flow became possible.’

18.7 Conclusion

265

Today there are sporadic ephemeral and, possibly, perennial pools near Erkowit (18°5 0 N, 37°10 0 E) that allow dragonflies (Odonata) to breed (Dumont and Martens, 1984), as well as Cladocera (Dumont et al., 1984), freshwater ostracods (Martens, 1984) and algae, especially diatoms (Compère, 1984), but there are no extensive and permanent swamps or marshes. The Red Sea Hills dragonflies are quite distinct from those in the main Nile Valley of Sudan, being Eremian species of Palaearctic origin, indicating migration from the north rather than from the tropics (Dumont and Martens, 1984). Some of the present-day ostracods are also of Palaearctic provenance lending support to the hypothesis of a migration from the north during times of Quaternary glaciation in Eurasia (Martens, 1984). Martens (1984, p. 160) also noted that ‘ . . . the main part of the fauna has an African origin. This is not surprising, because the Red Sea Hills are connected with the River Nile through temporary river systems such as the Wadi Amur and the Nile constantly introduces Ethiopian faunal elements’. There is further evidence in support of the inference that the climate was wetter in the Sudan about 2,000 years ago. In about AD 60 the Roman Emperor Nero despatched a military party under the command of two centurions to discover the sources of the Nile. They travelled up the White Nile but eventually were forced to desist from their task, later reporting that their way was blocked by impenetrable swamps from which emerged two rocky hills. If, as seems likely, these twin hills are Jebel Ahmed Agha, they are now at least 200 km downstream of the northern edge of the present-day Sudd swamps (Jackson, 1957). We saw in Chapter 8 that at some time between ca. 3.5 and 1.5 ka, swamps reached close to the foot of Jebel et Tomat, which is now 12 km away from the nearest riparian White Nile swamps. Further north, in the Butana Desert and adjacent Nile Valley, the Meroitic Kingdom (ca. 300 BC to ca. AD 350) was the major regional iron-smelting centre of its time (Shinnie, 1967). Considerable supplies of charcoal would have been needed to sustain this activity. Although the charcoal could have been brought in by boat from further south, it is much more likely that the vegetation at that time consisted of dense stands of acacia woodland. Furthermore, the earth dams or hafirs out in the Butana Desert no longer fill under current conditions of rainfall and runoff, suggesting a somewhat wetter climate when they were in active use (Jackson, 1957).

18.7 Conclusion The Red Sea Hills are a range of highly dissected Precambrian Basement Complex igneous and metamorphic rocks that parallel the western shores of the Red Sea for roughly 1,200 km between northern Egypt and the border between Sudan and Eritrea. The present-day climate ranges from arid to hyper-arid. At intervals during the Quaternary wadis originating in the Red Sea Hills flowed down to the Red Sea during times of low glacial sea level. Other much longer wadis ferried a bed load of coarse gravel and cobbles as far as the Nile, most recently during the terminal Pleistocene and early

266

The Red Sea Hills

Holocene. Fossil tufa deposits occur sporadically along the Red Sea Hills in the Eastern Desert of Egypt. The two most recent phases of tufa formation have calibrated 14C ages of 31–22 ka and 11–7 ka. Temperatures were lower than today when the 31–22 ka tufas were accumulating and it is probable that they owe their existence to the southward displacement of humid westerly air masses during the time when much of Europe was glaciated. The Holocene tufas precipitated under warmer climatic conditions and appear to reflect the influence upon local groundwater levels of rainfall associated with the more northerly penetration of moist tropical air masses originating over the Indian Ocean. Between 9 ka and 6 ka Mesolithic hunter-gathers and Neolithic herders of sheep and goats made occasional forays into the Red Sea Hills, which supported a more luxuriant vegetation cover than is present today. The most recent occasion when the climate in the southern Red Sea Hills was wetter than today was some two thousand years ago, at which time streams were flowing throughout the year and now arid wadis were home to extensive swampy meadows that provided a refuge for bands of cattle herders who fought among themselves for pasture and cattle.

19 The Sinai Peninsula

At the top of the hill I could appreciate as never before the awful proposition of crossing endless sand dunes. These great hills and valleys stretched, it seemed, to the uttermost parts of the earth . . . there was nothing here but sand, sculptured by the wind into overwhelming cliffs and sensuous crescent curves. Geoffrey Moorhouse, The Fearful Void (1974, p. 70)

19.1 Introduction The Sinai Peninsula is a triangular shaped mass of land, roughly 60,000 km2 in area, bounded by the Mediterranean Sea to the north, the Gulf of Suez and northern Red Sea to the west and the Gulf of Aqaba and southwest Israel to the east (Fig. 19.1). The southern sector consists of rugged mountains, the highest (St Catherine’s peak, Mount Sinai) rising to 2,285 m. Desert dunes occupy much of the northern sector of the Sinai (Embabi, 2008). One noteworthy feature of the narrow valleys running through the mountains and dissected central plateaux is the fine-grained valley-fills made up of reworked wind-blown desert dust. These fine-grained sediments accumulated in low-energy stream channels and wetlands during times when the local climate was wetter than today. Throughout prehistory, bands of early human hunter-gatherers moved from Africa into Eurasia after first crossing the Sinai Peninsula. During the Neolithic, domesticated plants and animals were brought in to the Nile Valley from the Near East via the Sinai Peninsula at least 7,000 years ago. The Sinai has thus been a migration corridor for more than a million years. In historic times the copper and turquoise mines of the Sinai were highly valued by the Predynastic and Dynastic rulers of Egypt (Trigger, 1982).

19.2 Origin and Evolution of the Sinai Peninsula The Sinai Peninsula consists of four major physiographic regions (Fig. 19.1). The southern third of the peninsula is made up of highly dissected uplands consisting mostly of very resistant Precambrian igneous and metamorphic rocks. These Basement Complex rocks

267

268

The Sinai Peninsula

34°E

N

Tel Aviv Jerusalem

Mediterranean Sea

ISRAEL

Suez Canal

Port Said

Dead Sea

4 NEGEV DESERT

SINAI

ba f Aqa

6 35

Gulf o

ez

f Su

lf o

Gu

2 1

JORDAN 30°N

DESERT

Suez

EGYPT

32°N

SAUDI ARABIA 28°N

Land > 1000m 0

100

Red Sea

km

Figure 19.1 The Sinai Peninsula. 1: Wadi Feiran. 2: Wadi Watier. 3: Wadi Kid. 4: Wad el Arish. 5: Wadi Madsus. 6: Mt Sinai.

form part of the Arabian-Nubian Shield and were intensively metamorphosed during the East African Orogeny of ca. 650–630 Ma (Collins and Pisarevsky, 2005; see also Chapter 4). Prolonged erosion of the Basement Complex rocks was followed by deposition of Palaeozoic, Mesozoic and Cenozoic sedimentary rocks. Slow regional uplift at the end of the Eocene was followed by up to 10 million years (6–10 Ma) of gradual denudation, resulting in the development of the Oligocene ‘Regional Truncation Surface’ (RTS) (Avni et al., 2012; Avni, 2017). The Late Eocene to Late Oligocene uplift of the Northeastern Afro-Arabian tectonic plate led to the formation of a crustal dome 3,000 km by 1,500 km in extent located across the present Red Sea (Avni, 2017). Increased tectonic activity during

19.3 Periglacial Landforms in the Sinai Mountains

269

the Early Miocene resulted in rifting along the Red Sea accompanied by uplift along the Red Sea Rift margins. The central portion of the Sinai Peninsula consists of a series of dissected plateaux developed primarily in Mesozoic and Cenozoic sedimentary rocks, their accordant summits coinciding with the Oligocene Regional Truncation Surface. The northern sector consists of Quaternary desert dunes resting on an erosional surface of gently tilted Cenozoic limestones and other marine sedimentary rocks. The coastal plain adjoining the Mediterranean Sea makes up the smallest physiographic region in the Sinai. During times of lower Quaternary glacial sea level the coastal plains were wider and were often invaded by windblown sands travelling eastwards from the Nile Delta across the northern Sinai to the Negev Desert in southern Israel.

19.3 Periglacial Landforms in the Sinai Mountains There is no evidence that the highest uplands in the Sinai Peninsula were ever glaciated during the Quaternary but there is good evidence that periglacial processes are active today and were more active during the late Pleistocene. Butzer and Hansen (1968, p. 427) summarise the observations of earlier workers in regard to modern cold-climate landforms on Mt Sinai (also known as Gebel Musa and Jebel Catharina), including stone rings, patterned ground and stone stripes. These features denote a temperature at or below freezing in the month of minimum temperature (see Chapter 12). Messerli et al. (1980) provide a comprehensive review of modern and late Pleistocene glacial and periglacial landforms and processes in the mountains of North Africa, including Sinai. They describe ‘a well-preserved set of periglacial landforms . . . on Precambrian quartz porphyry near the summit of Jebel Catharina in the Sinai’ ( Messerli et al., 1980, p. 111), the lower limits of which have not been mapped. They also commented that ‘periglacial solifluction mantles near the road below Jebel Catharina are a legacy of cold climate processes’, noting that these deposits are overlain by brown Holocene soils, much as in the Hoggar and Tibesti Mts of the Sahara (Messerli et al., 1980 p. 111). Sheet floods are now eroding the Holocene soils. The lower limit of periglacial solifluction mantles coincides very roughly with the 10°C isotherm in the warmest month, and with seasonal freeze–thaw cycles during the colder months (see Chapter 12). Similar periglacial deposits have been described further west in the northern Sahara. For example, in the Jebel Akhdar uplands of Cyrenaica in northeast Libya, McBurney and Hey (1955, p. 82) attributed the angular blocks forming the late Pleistocene screes and alluvial fans along the coastal margin of the Jebel to ‘a time when frost was a frequent occurrence during a large part of the year’, a conclusion endorsed by Vita-Finzi (1969, p. 99) during his investigation of late Pleistocene valley-fill deposits in this region. The two main conclusions to be drawn from the Sinai periglacial landforms are, first, that late Pleistocene winter temperatures were colder than they are today, and second, that in presently cold and arid mountain summits, there must have been sufficient moisture to

270

The Sinai Peninsula

promote frost shattering and periglacial solifluction. Existing data based solely on periglacial landforms in the Sinai Peninsula do not allow us to propose quantitative estimates of late Pleistocene changes in temperature and precipitation, a topic we discuss in the following section.

19.4 Tufa Deposits in the Sinai Peninsula and Their Climatic Significance Hamdan and Brook (2015) recently completed a comprehensive survey and analysis of relict tufa deposits in the Sinai Peninsula and in the northern Red Sea Hills (see Chapter 18). They sampled one locality (site 7: Wadi Watier) in the central uplands at 29.0°N, 34.01°E and two localities in the far southeast of the Sinai Peninsula at 28.2°N, 34.3°E (site 1: Wadi Kid) and 28.3°N, 34.4°E (site 2: Wadi Madsus). The tufas at site 7 have calibrated 14C ages of 28.4–27.6 ka and 21.2–20.4 ka. The site 1 tufas have two almost identical calibrated 14 C ages of 6.88–6.67 ka. The site 2 calibrated 14C ages are 7.30–7.17 ka and 7.6–7.5 ka. The apparent absence of tufa formation between ca. 20 ka and ca. 8 ka is consistent with aridity in the Sinai at this time. A similar arid interval in the northern Red Sea Hills between ca. 19 ka and ca. 9 ka is consistent with this inference (Hamdan and Brook, 2015). The stable oxygen isotopic composition of the late Pleistocene Sinai tufa samples points to a temperature range between 14.0°C and 20.8°C, with a mean temperature value of 17.73°C during tufa precipitation. This is in contrast to the inferred Holocene temperatures for tufa precipitation at sites 1 and 2 of 21.22°C to 23.7°C, with a mean temperature value of 22.46°C, or 4.73°C higher than the mean late Pleistocene temperature. Hamdan and Brook (2015) further concluded from their oxygen isotope composition that the late Pleistocene tufas indicated moist air mass circulation from the west from the Mediterranean, ‘when the Westerly circulation was pushed southwards during the coldest periods of the Late Pleistocene’ (Hamdan and Brook, 2015, p. 185). They considered that the Holocene tufas ‘are almost certainly a response to increased rainfall from the north’ (Hamdan and Brook, 2015 p. 185), when precipitation from the Mediterranean reached the northern Red Sea.

19.5 Late Pleistocene Valley-Fills of the Sinai Peninsula 19.5.1 Late Pleistocene Valley-Fills of the Sinai Peninsula and Cognate Regions Desert rivers flowing through uplands composed of igneous and metamorphic rocks, as in the southern Sinai, or coarse-grained sedimentary rocks, as in central Sinai, will tend to have very coarse bed loads and transport much of their load by rolling and saltation during rare torrential floods. It is therefore something of a surprise to discover that the present-day channels in wadis, such as Wadi Feiran and Wadi Arish, are incised into very fine-grained late Pleistocene alluvial silts and clays 10–20 m thick (Butzer, 1958, p. 39 and 74; Rögner et al., 2004), the origin of which has long been an enigma. The Sinai valley-fills remain poorly dated, although Issar and Eckstein (1969) report ages of 24–20 ka for wetland sediments in Wadi Feiran.

19.5 Late Pleistocene Valley-Fills of Sinai Peninsula

271

Figure 19.2 Riverbank sediments derived from reworked eolian dust, Matmata Hills, Tunisia.

Similar late Pleistocene fine-grained valley-fill deposits have been found in other mountainous desert regions, including the Negev (Avni, 2005; Avni et al., 2006), the Matmata uplands of Tunisia (Fig. 19.2) (Coudé-Gaussen and Rognon, 1983; CoudéGaussen et al., 1987), the Namib (Eitel et al., 2001, 2005; Heine and Heine, 2002), the semiarid Chifeng region of Inner Mongolia in northern China (Avni et al., 2010) and the arid Flinders Ranges of South Australia (Williams et al., 2001; Williams and Adamson, 2008; Glasby et al., 2010; Haberlah et al., 2010a, 2010b). It now seems clear that they are best interpreted as reworked desert dust or loess.

19.5.2 The Late Pleistocene Wetlands of the Flinders Ranges in South Australia: A Possible Analogue for the Late Pleistocene Valley-Fills of the Sinai Peninsula The fine-grained valley-fills in the Flinders Ranges of South Australia (Fig. 19.3) have been investigated in greater detail than similar deposits elsewhere and are very precisely dated. The study of these deposits provides insights into how the Sinai deposits might have accumulated – insights not yet available from the comparatively limited fieldwork and analytical work that has been carried out to date on the Sinai Desert valley-fills. The work in the Flinders Ranges showed that in order for these fine-grained valley-fills to accumulate, four things are necessary. First, there needs to be a regular supply of windblown dust. In the case of the Flinders Ranges, the source areas were the surfaces of dried

272

The Sinai Peninsula

Figure 19.3 Late Pleistocene alluvium derived from reworked eolian dust, Flinders Ranges, South Australia.

playa lakes and the exposed late Pleistocene continental shelf, all of which were located upwind of the Ranges and roughly perpendicular to them. In the case of the Sinai, eolian dust blown eastwards from the northern Sahara is the most likely source, possibly supplemented by calcareous dust blown inland from the exposed continental shelf during times of lowered glacial sea level. Second, once the dust has fallen on the surface of the hills and valleys, it needs to be trapped by the vegetation. If not, it can be remobilised by the wind, much as occurs today in northern Nigeria during the Harmattan season (McTainsh and Walker, 1982; McTainsh, 1987). If dust is to remain on the surface and form a dust mantle, appropriate natural dust traps are essential (Coudé-Gaussen and Rognon, 1983; CoudéGaussen et al., 1987; Pye, 1987; Tsoar and Pye, 1987; Williamson et al., 2004). CoudéGaussen and Rognon (1983) concluded that a more humid late Pleistocene climate that promoted a dense cover of grasses and shrubs was responsible for desert dust accumulation in the Matmata limestone uplands of southern Tunisia. In fact, there may be no need to invoke a very much wetter climate. During times of maximum glaciation, most recently during the very late Pleistocene, the atmospheric carbon dioxide concentration was reduced to 180–200 ppmv (Jouzel et al., 2007). Such low concentrations would have favoured the expansion of grasses and herbs at the expense of trees, and the resulting denser grass cover would have provided an effective dust trap (Williams and Adamson, 2008).

19.5 Late Pleistocene Valley-Fills of Sinai Peninsula

273

Third, as the dust deposits accumulated on the hill slopes and valley sides, some of the dust mantle needs to be washed downslope under the influence of slope wash under a regime of gentle rain and limited runoff. Torrential rains and intense runoff would not result in the progressive accumulation of several tens of metres of homogeneous and well sorted silts and clays along the valley bottoms but rather in the accumulation of a heterogeneous and poorly sorted mixture of silts and gravels. It is worth observing that fine-grained valley-fills are not accumulating in any of these regions today (Williams et al., 2001). The hydrological conditions when the valley-fills were accumulating to depths of 20 m must therefore have been unlike those of today. At present, rains in these arid mountains tend to be sporadic and often very intense, erosion is locally severe, and the flash floods in the valleys are capable of moving large cobbles and very coarse gravels as traction load. The fourth prerequisite is a stream regime characterised by limited erosion and very low stream power, allowing progressive accumulation of the loess washed off the valley sides to form a fine-grained valley fill. An important factor promoting the persistence of the late Pleistocene wetlands in this presently arid upland were the greatly reduced rates of evaporation, especially in summer, which were linked to lower glacial temperatures. Fewer or no trees and reduced losses of soil moisture from evapotranspiration would have led to a rise in local and regional groundwater levels, contributing water to springs and wetlands. The winter cloud base would have been much lower during the late Pleistocene, leading to a regime of prolonged and gentle winter rains during which loess could be washed down the hill slopes by runoff to accumulate in the valley bottoms and be reworked by low-energy stream flow. The reduction in dust flux and the return of high intensity rainfall caused abrupt incision of the Flinders Ranges fine-grained valley fills and the demise of the wetlands (Williams et al., 2001). There were a series of rapid changes in global climate from about 17,000 years ago (17 ka) onwards. Global warming is associated with synchronous retreat of mid-latitude mountain glaciers in both hemispheres commencing at 17 ka (Schaefer et al., 2006). [In North America, the Laurentide Ice Sheet had retreated from a significant part of the Agassiz Basin by 14.5 ka (Teller et al., 2018), which is also when the return of the summer monsoon triggered major floods in the Blue and White Nile Rivers.] From about 17 ka onwards, the concentration of atmospheric carbon dioxide began to increase, the influx of eolian dust diminished, and hitherto treeless desert environments were colonised by trees, shrubs and tussock grasses. The warm Bølling–Allerød interstadial (14.6–12.8 ka) was succeeded by the cold Younger Dryas stadial (12.8–11.5 ka). During this time of millennial scale rapid climatic fluctuations, the Sinai climate oscillated between warm and relatively wet to cold and dry. The late Pleistocene climatic regime of gentle winter rains gave way to one of more erratic rainfall characterised by highly localised extreme rainfall events, flash floods, and intermittently very high stream energy flows. The result was a hydrologic change from one of fine-grained valley-fill accumulation to one of channel incision and widespread gully erosion.

274

The Sinai Peninsula

19.6 Desert Dunes of the Sinai Peninsula and Adjacent Northern Negev Desert In the northern third of the Sinai Peninsula a complex series of linear and crescent-shaped sand dunes are clearly visible on satellite imagery and probably reflect more than one generation of dunes. The northern Sinai dune field extends eastwards into Israel where an area of 1,300 km2 in the northwest Negev Desert is covered by presently vegetated and stable linear dunes. The precise age of the Sinai dunes is not well established. However, 97 Optically Stimulated Luminescence (OSL) ages are available for the Negev dunes, which were starting to accumulate in this area from about 100 ka onwards, but especially after 23 ka, at a time of low sea level and stronger winds (Roskin et al., 2011a). Thirty-five dunes and inter-dune swales dated by OSL show that these dunes were active during 18–11.5 ka (post-LGM), 2.0–0.8 ka (very late Holocene), and in the last 150 years. The post-LGM interval was when dune movement was most widespread. During this time, the Negev dunes formed dams across small valleys, resulting in the formation of lakes and ponds between the dunes. The Sinai and Negev linear dunes are aligned roughly from west to east, consistent with transport by sand moving winds blowing from the west. Roskin et al. (2011b) studied these dunes in detail and concluded that they were actively moving during the late Pleistocene under the influence of stormy winter cyclones from the eastern Mediterranean. The climatic regime at that time would have been one of very strong winds and higher winter rainfall, also evident from the occurrence of lakes, fossil soils and prehistoric occupation sites in the swales between the linear dunes. This finding runs counter to the popular view that active dunes indicate a more arid climate. In this case, sand supply and wind velocity proved to be more critical agents of dune activity than any reduction in precipitation and plant cover. The Holocene climate in the northern Negev and adjoining northern Sinai Peninsula was more arid than that of the late Pleistocene, but the reduction in seasonal storms and in wind velocity gave rise to dune stabilisation. Once again, the wind regime exerted a greater influence than aridity in controlling dune movement and stabilisation. In this region, maximum dune activity seems to have been synchronous with a phase of increased rather than decreased rainfall. Sea level fluctuations also had a major influence on sand supply in the northern Sinai and northern Negev. During the Last Glacial Maximum (23–19 ka), sea level was up to 120 m lower than it is today, so that previously submerged sands in the Nile Delta became subject to deflation. Along the coast of that time the sands that were still submerged were carried eastwards by longshore drift and were subsequently blown southwards and eastwards to feed the sand dunes of the northern Sinai Peninsula and northern Negev. According to Roskin et al. (2011b) and Roskin and Tsoar (2017), movement of linear dunes from the Sinai into the northern Negev was already in progress during the LGM, accelerating during 16.0–13.7 ka, with some movement at 12.4–11.6 ka. The first phase coincides reasonably closely with Heinrich Event 1 (16.8–14.6 ka) in the North Atlantic (Heinrich, 1988) and the second with the Younger Dryas (12.8–11.6 ka)) cold interval in Greenland.

19.6 Desert Dunes of Sinai Peninsula

275

These ages for peak dune activity are of interest, because there is a long-held view that the dunes across North Africa were particularly mobile during the LGM (Sarnthein, 1978; Sarnthein et al., 1981; Lancaster et al., 2002), which was a time of extreme aridity and enhanced transport of eolian dust from the Sahara to the Atlantic (Parmenter and Folger, 1974; deMenocal et al., 2000). This idea dates back well over half a century, and is based on the inference that the Trade Winds were stronger at this time as a result of steeper temperature and pressure gradients between tropical and equatorial latitudes during times of maximum glaciation in Europe and North America. For a comprehensive review, see Rognon (1989). Grain size analysis of marine sediment cores retrieved off the west coast of the Sahara reveal that grains blown out to sea by the Trade Winds were coarser during the last glacial, indicating higher wind velocities associated with a stronger anticyclonic circulation over the Sahara (Parkin and Shackleton, 1973; Parkin, 1974). The question of when desert dunes were active in the past is not easy to answer because it depends to a considerable degree on preservation of the dune sands, and on whether or not they have been repeatedly mobilised. Swezey’s research on the desert dunes of Tunisia prompted him to propose that dune sands are best preserved if sand deposition is closely followed by a humid phase during which soil development or deposition of fluvial and lacustrine sediments protects the underlying sands from deflation (Swezey, 2001, 2003). Swezey (2003) has argued that preservation is less likely if sand accretion is followed by an arid phase with concomitant dune deflation and redeposition. In his comprehensive review of dune activity in the Sahara (Swezey, 2001), he found that most of the ages for dune activity were younger than the LGM, most probably as a result of reworking during the arid Younger Dryas episode (12.8–11.6 ka), which, he noted, was followed by a wetter Saharan climate between 11.5 ka and 7 ka. The Sinai dunes may also have influenced the lands downwind in a more indirect manner. The loess deposits in the Negev Desert consist of a relatively coarse quartz fraction (50–60 µm in diameter), with OSL ages extending back to about 180 ka, in a matrix of much finer particles (3–8 µm in diameter) (Crouvi et al., 2008, 2010). The finer particles are believed to have travelled in suspension in the atmosphere for several thousand kilometres across the northern Sahara before being removed from the atmosphere by falling rain or settling out as dry dust fall. The most common process of dust deposition is in rain, and rainstorms often occur soon after major dust storms. The source of the coarser fraction remained a puzzle until the idea was put forward that it originated from sand particle abrasion during the eastward movement of the Sinai desert dunes (Crouvi et al., 2008, 2010; Enzel et al., 2010). The Nile Delta was the ultimate source of these dunes, with the sands blown inland from deltaic sediments exposed during times of lower glacial sea level (Amit et al., 2009). The Nile has been transporting sediments to the delta, at least intermittently, for well over 2 million years (Williams and Talbot, 2009; Davis et al., 2012), so that this process of coarse loess formation probably has a long history.

276

The Sinai Peninsula

19.7 Prehistoric Occupation in the Sinai Peninsula The Sinai Peninsula has played an important role in world prehistory, because it was one of the major routes by which prehistoric humans were able to move out of Africa into Eurasia and eventually to Australia and the Americas, a topic we consider in Chapter 23. It was also the route that enabled Neolithic groups from the Near East to bring domestic plants and animals to the Nile Basin and to other parts of Africa (Chapter 22). Given its very significant role as a prehistoric migration corridor, it may seem surprising that so few prehistoric sites have been identified and rigorously dated within the 60,000 km2 of the Sinai Peninsula. However, the area is remote, rugged, often insecure, and much of the archaeological evidence is probably buried beneath younger sediments. This dearth of prehistoric sites contrasts with the wealth of sites that have been carefully investigated further east in the Levant and further west in the Nile Valley and eastern Sahara. For example, Lower Palaeolithic Acheulian bifaces are found in alluvial terrace sediments in the southern Negev Desert (Ginat et al., 2003), at the prehistoric site of ‘Ubeidiya in Israel, where they are dated to ca. 1.4 Ma (Goren-Inbar et al., 2004) and at many other sites in Israel and other regions in the Levant (Sharon, 2017). Humans were therefore present more than a million years ago in these presently arid areas. Sporadic surface finds of Acheulian bifaces have been made in the Sinai, but it seems that no Lower Palaeolithic sites in primary context have yet been discovered there (Phillips, 1987). Middle Palaeolithic artefact assemblages and associated human skeletal remains with ages of 150–100 ka occur in the Omo Valley of Ethiopia as well as at Es Skhul and Qafzeh in Israel (Stringer et al., 1989; Bräuer et al., 1997), but only a few scattered Middle Palaeolithic artefacts have been recorded from the Sinai. The picture becomes much clearer as we enter the late Pleistocene and Holocene. Upper Palaeolithic and Epi-Palaeolithic hunter-gatherer sites associated with former wetlands and springs show a range of activities as well as of artefact types (Phillips, 1987; Belfer-Cohen and Goring Morris, 2017; Goring-Morris and Belfer-Cohen, 2017). Microwear patterns on the edges of stone tools indicate use of meat, scraping horn and bone, and scraping of hides (Phillips, 1987). Resource use ranged from intensive and possibly seasonal use of wetland plants and hunting of small game to more extensive foraging and hunting. Obtaining suitable raw materials for stone tool manufacture was another factor governing site location, with certain types of flint being widely used. These Upper Palaeolithic and Mesolithic sites range in age from at least 40 ka to less than 15 ka (Phillips, 1987; Belfer-Cohen and Goring-Morris, 2017; Goring-Morris and Belfer-Cohen, 2017), which is coeval with the late Pleistocene interval of fine-grained valley-fill accumulation described in Section 19.5. The sites are well preserved because they were formed during a prolonged interval of widespread valley floor aggradation under conditions of low stream energy. Flash floods associated with torrential and highly localised desert rainstorms in southern and central Sinai would have destroyed much of the earlier evidence of human occupation along the valley bottoms and sand dune migration has also buried many former sites in northern Sinai.

19.8 Conclusion

277

In the Near East and especially in the ‘Levantine Corridor’, the Epi-Palaeolithic Natufian Culture (15–11.5 ka) marks the transition from hunter-gatherer to agricultural village societies and has been described as ‘the harbinger of food-producing cultures in the southern Levant’ (Grosman and Munro, 2017, p. 699). In Sinai, certain of these Epi-Palaeolithic sites have been given local names on account of their distinctive stone tool assemblages. These local names need not concern us here; the important point to retain is that there were widespread cultural links between the southern Levant and the Sinai Peninsula during the terminal Pleistocene, and that from this culture there emerged the various Neolithic cultures that ultimately gave rise to urban civilisation in the Nile Valley. The complex and fascinating question of the causes and consequences of plant and animal domestication in the Nile Basin is the subject of Chapter 22.

19.8 Conclusion The Sinai Peninsula covers an area of 60,000 km2 and is bounded by the Mediterranean Sea to the north, the Gulf of Suez and northern Red Sea to the west and the Gulf of Aqaba and southwest Israel to the east. The southern desert consists of rugged mountains eroded into Precambrian igneous and metamorphic rocks. The central sector is a dissected plateau composed of Mesozoic and Cenozoic sedimentary rocks. The latter formations extend to the north and are covered by active desert dunes. The present climate in the Sinai ranges from arid to hyperarid, but at intervals during the late Pleistocene the climate was less arid, springs were active, streams were flowing, and wetlands provided food for animals and Upper Palaeolithic hunter-gatherers. The ubiquitous late Pleistocene fine-grained valley fills in the Sinai upland wadis represent reworked eolian dust carried and deposited by low energy stream channels during times of prolonged but gentle winter rains. A moist episode during the early to mid-Holocene saw the movement of Neolithic groups into the Nile Valley, bringing their herds of domestic animals and their cereal grains from the Near East into the Nile Basin and eventually into the Sahara and East Africa.

20 The Nile Delta

A much bigger gulf than this could have been turned into dry land by the silt brought down by the Nile – for the Nile is a great river and does, in fact, work great changes . . . I have observed for myself that Egypt at the Nile Delta projects into the sea on either side; I have seen shells on the hills and noticed how salt exudes from the soil to such an extent that it affects even the Pyramids. Herodotus (ca. 485–425 BC),The Histories (1954 p. 106)

20.1 Introduction The Nile Delta (Fig. 20.1) covers an area of 22,000 km2 and is one of the best-studied regions in the entire Nile Basin as well as one of the richest agricultural regions in the world. It provides 63% of Egypt’s agricultural land and a livelihood for more than 50 million of its people (Marriner et al., 2012b). The Delta is an important industrial area, leading to unresolved issues of soil and water pollution and ecosystem damage (Stanley and Warne, 1993a; Stanley and Warne, 1998; Hamza, 2009). In this chapter, we focus on the late Pleistocene and Holocene history of the Nile Delta and the light it sheds on hydro-climatic changes in the upper reaches of the Nile Basin.

20.2 Origin and Evolution of the Nile Delta Deltas are gently sloping alluvial plains formed where rivers enter a body of standing water such as a lake or the sea. At the proximal edge of the delta the parent river channel splits into a series of distributary channels, which radiate out across the delta alluvial plain. The distal margin of the delta may consist of lobes, cusps or a combination of the two, as in the case of the Nile Delta (Fig. 20.1). Many deltas, including the Nile Delta, are roughly triangular in shape and resemble the Greek capital letter Δ (‘Delta’). It was this feature that prompted Herodotus to apply the term ‘Delta’ to the mouth of the Nile some 2,500 years ago. He noted that the two main branches of the Nile entered the sea at Pelusium and Canopus (Fig. 20.1), and commented that he was ‘convinced – and the Egyptians themselves admit the fact – that

278

20.2 Origin and Evolution of the Nile Delta

279

Figure 20.1 Nile Delta. (Modified from Butzer, 1976, Fig. 4; Stanley and Warne, 1993b, Fig. 1; Bird, 2000, Fig. 10.1; and Pennington et al., 2017, Fig. 1).

the Delta is alluvial land and has only recently (if I may so put it) appeared above water’ (Herodotus, trans. 1954, p. 107). Herodotus also made the prescient observation that ‘the Egyptians who live below Lake Moeris in the Delta and thereabouts will, if the Nile fails to flood, suffer permanently’ (Herodotus, trans. 1954, p. 106). Increasing human impact on the Delta, exacerbated by more and more river regulation in the last 150 years, has confirmed the accuracy of this prediction. Indeed, Stanley and Warne (1998, p. 820) concluded that ‘The Nile delta has been transformed within the past 150 years from a constructive wavedominated delta that took 7,000 years to develop, to an eroding coastal plain that is now well into its destruction phase.’ Deltas have a very characteristic structure when viewed in cross-section parallel to the direction of channel flow (Fig. 20.2). The outermost or ‘bottom-set beds’ consist of relatively thin horizontal layers of fine sediment. The uppermost or ‘top-set beds’ consist of thin and gently sloping layers of sediment, which grade into the thicker and more steeply sloping ‘fore-set beds’. Where former deltas are now exposed above lake or sea level and incised by gullies, these deltaic structures are often very evident. Examples include the early Holocene Angamma Delta in the northern Chad Basin formed during a high stand of Lake Chad (Ergenzinger, 1968; Servant et al., 1969; Servant, 1973; Servant and ServantVildary, 1980) and the much smaller late Pleistocene delta inside Jebel Marra caldera formed when the now saline and shallow Deriba lake was much deeper (see Chapter 13, Fig. 13.6; and Williams et al., 1980).

280

The Nile Delta Topset beds

Foreset beds Bottomset beds

Figure 20.2 Deltaic bedding. (Modified from Sparks, 1972, Fig. 8.35.)

The presently visible portions of the Delta originated during the Holocene marine transgression, which came to a halt about 6,500–6,000 years ago. The Holocene deltaic sediments overlie a late Pleistocene alluvial plain, which had developed prior to the fall in sea level centred on the Last Glacial Maximum (LGM) 21,000 ± 2,000 years ago, during which global sea levels fell to ca. 125 m below present sea level. Global sea level has fluctuated continuously during the last 2.6 Ma of the Quaternary in tandem with the waxing and waning of the great Northern Hemisphere ice sheets. During warm interglacial intervals of high sea levels the associated marine transgressions were accompanied by delta progradation and the formation of sandy barrier ridges parallel to the prevailing coastline. Times of falling sea level (marine regressions), culminating in low glacial sea levels and a fall in base level, were accompanied by river channel incision and reworking of the Nile sediments deposited offshore (Stanley and Warne, 1993b; Stanley and Warne, 1998). As a consequence of these repeated fluctuations in base level, together with constant shifts in the location of Nile Delta distributary channels, the alluvial architecture of the Delta and its offshore extension (the Nile Cone; Chapter 21) is quite complex. This complexity is revealed in frequent lateral and vertical changes in sediment type and associated fossil assemblages (Stanley and Warne, 1998). Adding to this complexity are variations in sediment influx and differential rates of sediment compaction, topics we discuss in Sections 20.5 and 20.6. Like any delta, the Nile Delta consists of a distinct suite of physiographic elements and ecosystems. At the proximal or upstream apex of the Delta the Nile consists of a single welldefined channel flanked on either side by a narrow flood plain consisting of alluvial clays and silts. Further downstream the channel bifurcates into a series of distributary channels that radiate out across an alluvial plain. At the distal end of the fan-shaped Delta the alluvial plain merges into marine sands and clays. Barrier ridges consisting of shelly sand run parallel to the shore. Impounded behind them are lagoons and swampy depressions. Depending on the influx of Nile water into the lagoon, they may range from fresh to brackish to saline. Where the lagoons have been cut off from any contact with the Nile for several centuries or more, they become highly saline sebkhas characterised by a predominance of evaporitic minerals and sediments, including dolomite [CaMg(CO3)2], halite [NaCl] and gypsum [CaSO4].

20.3 Holocene History of Maryut Lagoon

281

Historic maps of the Delta sometimes show traces (real or imagined) of former Nile distributary channels. For example, clearly marked west of the Delta on John Low’s 1811 map of Egypt is an ‘Ancient bed of the Nile filled up by a King of Egypt the traces of which are still found in the Desert’ (Low, 1811, and Fig. 20.3). This putative channel does not appear on Bonne’s 1762 map of Egypt (Bonne, 1762, and Fig. 20.4), although he portrays the distributary channels in the Delta in much greater detail than Low (1811).

20.3 Holocene History of Maryut Lagoon, Western Nile Delta Cores collected from the four main lagoons have shed some light on the recent evolution of the northern sector of the Delta. From west to east, these lagoons are Maryut, Idku, Burullus and Manzala (Fig. 20.1). We discuss one in particular – Maryut lagoon – because it has been studied in some detail and has a fairly reliable chronology for part of the Holocene. Maryut lagoon is located near Alexandria on the NW margin of the Nile Delta. It is the oldest Holocene lagoon in the Delta and originated about 8,000 years ago during the closing stages of the Holocene marine transgression. Flaux et al. (2011) have obtained a thorough record of the history of this lagoon between 8 ka and 3.2 ka based on detailed analysis of the fossil assemblages within seven sediment cores collected from the bed of the lagoon and four stratigraphic sections, buttressed by 27 calibrated 14C ages. The present-day lagoon is slightly brackish and has no connection to the Mediterranean Sea. At intervals during the Holocene the micro- and macrofossils (foraminifera, mollusca, ostracods), calibrated against samples from nine present-day surface habitats, indicate a lagoon regime with a strong marine influence, a lagoon regime with enhanced Nile sediment and water inputs, and a regime with neither marine nor perceptible Nile influences. Soon after the 7.7 ka birth of the lagoon it became isolated from the sea at ca. 7.5 ka by a sand barrier of probable late Pleistocene age and remained isolated as the northern edge of the Delta prograded and advanced seawards. Between ca. 7.5 ka and ca. 4.8 ka, the lagoon was a freshwater swamp in which accumulated highly organic muds mostly devoid of shells. Conditions changed quite suddenly from 4.8 ka onwards, after which there was an alternation between shell-rich deposits and fine-grained organic sediments devoid of shells. Each sedimentary couplet (shell/mud) deposited between 4.8 ka and 3.2 ka represented about 45 ± 6 years (N = 39). Flaux et al. (2011) attribute the change in depositional regime after 4.8 ka to a general decline in Nile sediment transport and Nile flood discharge brought about by climatic desiccation in the Nile headwaters. They interpret the alternation between shell accumulation and mud deposition as an alternation between intervals of very low Nile flow, enabling marine shells to invade the lagoon, and intervals of higher Nile flow, during which Nile silts accumulated on the floor of the lagoon, causing anoxic conditions and death of the marine mollusca. We saw in Chapter 3 that historic Nile floods between AD 622 and 1470 AD revealed cycles of high and low flows with a periodicity of 35–45 years (Fraedrich et al., 1997; De Putter et al., 1998).

282

The Nile Delta

Figure 20.3 Map of the Nile Delta, as depicted in the author’s copy of the map by John Low: Egypt from the best Authorities. New and Complete American Encyclopedia or Universal Dictionary of Arts and Sciences, by E. Low, successor to John Low, New York, 1811.

20.3 Holocene History of Maryut Lagoon

283

Figure 20.4 Map of the Nile Delta, as depicted in the author’s copy of the Carte de L’Egypte Ancienne et Moderne by M. Bonne. Paris, Lattré, 1762.

284

The Nile Delta

20.4 Variations in Nile Delta Sediment Provenance The pioneering work by Shukri (1949) on the mineral composition of Nile sediments demonstrated very clearly the significant differences between sediments transported by the White Nile, the Blue Nile and the Atbara, and has been elaborated in impressive detail by Garzanti et al. (2006, 2015). In particular, the White Nile sediments are rich in amphibole and lacking in pyroxene, whereas the Blue Nile and Atbara sediments are especially rich in pyroxene, as expected from their headwaters in the volcanic uplands of Ethiopia, and are correspondingly poor in amphibole. The Atbara carries a higher proportion of pyroxene than the Blue Nile. Hassan (1976) built on Shukri’s work in an effort to identify when the Ethiopian headwaters of the Nile began to contribute sediment to the Desert Nile in Egypt and northern Sudan, and proposed a minimum age for this conjunction towards the end of the Middle Pleistocene. Given the many gaps in the Nile alluvial record, and in particular the ubiquitous erosional gaps between the older Nile conglomerates and the late Pleistocene alluvial silts (see, e.g., Giegengack and Zaki, 2017, Fig. 2), such an approach is unlikely to yield unequivocal results as to when the Ethiopian tributaries first flowed into the Desert Nile. Foucault and Stanley (1989) used changes in the heavy mineral content of Holocene sediments in the Nile Delta to determine times when the Blue Nile and Atbara were major contributors of sediment to the Delta, evident in high proportions of pyroxene within the sediments. Conversely, times when the White Nile was a significant contributor of sediment to the Delta are evident in high proportions of amphibole relative to pyroxene. They found that times of relatively low Blue Nile and Atbara sediment input coincided with times of high lake levels in the Ethiopian Rift and Chad basin. In contrast, dry and cold climatic intervals in the Ethiopian headwaters, such as during the LGM, were associated with relatively high Blue Nile and Atbara sediment contribution to the Delta. They concluded that during warm wet climatic phases, vegetation cover in the Ethiopian headwaters of the Nile protected the hill slopes from erosion so that although flood discharge was high, sediment loads were relatively low. The opposite was true of times of cold dry climate in the Ethiopian Highlands. At such times, the vegetation cover was greatly reduced, the tree line was up to 1,000 m lower, periglacial processes were active over a large area, and the unstable slopes provided abundant sediment to the rivers in the upper catchments of the Blue Nile and Atbara (Adamson et al., 1980; see also Chapters 11 and 12). Foucault and Stanley (1989) further noted that during wetter climatic intervals when the Intertropical Convergence Zone (ITCZ) reached further north across the eastern Sahara and southern Egypt, many of the presently inactive wadis would have been flowing to the Nile. The heavy minerals in many of these wadi sediments would have been relatively high in amphibole and low in pyroxene, reflecting the influence of the underlying Basement Complex rocks (see Chapter 4). Hamroush and Stanley (1990) used instrumental neutron activation analysis of samples from three well-dated Nile Delta sediment cores to detect subtle changes in trace elements and rare earth elements during the past 30,000 years. They focused on changes in the ratios

20.4 Variations in Nile Delta Sediment Provenance

285

of chromium to scandium (Cr/Sc) and lanthanum to lutetium (La/Lu) to detect subtle changes in sediment inputs from the White Nile on the one hand and the Ethiopian tributaries of the Nile (Blue Nile, Atbara) on the other. Peaks in the Cr/Sc ratio denote a high input of Ethiopian sediment and peaks in the La/Lu ratio denote an influx of sediment from the Ugandan headwaters of the White Nile. During the wet interval between about 7 and 4 ka in Ethiopia, the Cr/Sc ratio was initially high, consistent with a substantial input of sediment from Ethiopia to the Delta. As time progressed and a dense vegetation cover became re-established in the Blue Nile and Atbara headwaters, flood discharge remained high but erosion rates diminished, leading to a reduction in the sediment loads, reflected in a lower Cr/Sc ratio. During the arid interval between 17.5 and 13 ka, both the La/Lu and the Cr/Sc ratios remained low, consistent with much reduced water and sediment discharge in all three rivers. In addition to using heavy minerals to identify Nile sediment sources in the Delta, another powerful tool is to use the stable isotopic composition of the deltaic sediments. Krom et al. (2002) investigated fluctuations in the strontium isotopic composition of Nile Delta clays during the last 7,000 years to see how events in the upper catchments of the Nile had influenced events downstream as well as offshore. They concluded that times of very high discharge from the Ethiopian headwaters of the Nile are reflected in very high Nile floods downstream and in the formation of sapropel deposits offshore. Sapropels are highly organic deposits that form on the floor of the sea under anoxic conditions during episodes of above average freshwater influx. The age of the youngest sapropel in the Eastern Mediterranean is still in dispute. It may have begun to form shortly after 14 ka and may have ceased forming by about 6.5 ka or even as recently as 5 ka (see overview in Williams et al., 2015a, pp. 104–105). In Chapter 21 we discuss in some detail the sapropel record preserved in the Nile Cone. Fluctuations in the 87Sr/86Sr values preserved in Nile Delta clays have been used to good effect to reconstruct a Holocene history for the main Nile (Stanley et al., 2003). One dramatic event recorded by the Nile Delta strontium isotopic data is the hydrologic background to the sudden collapse of the Old Kingdom in Egypt, which was precipitated, or at least aggravated, by the catastrophic drought centred at 4.2 ka (Stanley et al., 2003). We saw in Chapter 14 that in strong contrast to the collapse of the Egyptian Old Kingdom, the Kerma civilisation in the Nile Valley of northern Sudan survived this event (evident in the strontium ratios in local Nile alluvium) and persisted for a further thousand years, before succumbing to invasion from Egypt and a further decline in Nile flow (Macklin et al., 2013; Woodward et al., 2015a). Using the Palermo Stone record of Nile high flood levels, Barbara Bell has had success in reconstructing a progressive decline in flood level between 3,100 BC and 2,800 BC (Dynasty I and Dynasty II), after which the mean flood level remained low until soon after 2,500 BC (Bell, 1970, Fig. 1). The Palermo Stone is a former Nilometer used to record Nile flood levels and is now housed in the Palermo Museum. Bell (1970) also studied the records from other fragmentary remains of the original Nilometer.

286

The Nile Delta

Drawing on the archival accounts of what she termed the First Dark Age in Egypt (2,200–2,000 BC), she provided a graphic account of the social unrest, breakdown of law and order, famine and occasional references to cannibalism caused by several decades of extreme drought and very low Nile floods (Bell, 1971). It was also a time of dust storms, wind-blown sand encroachment into the Nile and onto the Nile flood plain, and desiccation across the Delta. The Egyptian word tzw refers to sand banks exposed in the Nile channel during low water and was widely used by contemporary writers as a synonym (or euphemism – it was not done to criticise the Nile) for very low Nile floods. Bell (1971) identified two short intervals of very low Nile flow (2,180–2,130 BC and 2,000–1,990 BC) during which she concluded that the White Nile had ceased to flow. Strontium isotopic evidence near the White Nile headwaters shows that at about this time there was indeed an erosional gap and no overflow from Lake Albert in Uganda into the White Nile (Williams et al, 2006, Fig. 9). This must have been a crucial factor in accentuating the severity of this drought in Egypt. Before dams were constructed throughout the Nile Basin, it was the White Nile that maintained the Desert Nile as a perennial river during times of extreme drought in the Ethiopian headwaters of the Blue Nile and Atbara (Williams and Adamson, 1973; Williams and Adamson, 1982). They emphasised the easily overlooked importance of the White Nile: ‘In very dry years the White Nile is the guarantor of perennial flow in the Nile; the swamps of the southern Sudan therefore play a quite crucial role today in buffering the seasonal fluctuations in Nile discharge and in helping to maintain perennial flow in the main Nile’ (Williams et al., 2006, p. 2653).

20.5 Holocene Fluctuations in Nile Delta Sedimentation When analysed appropriately, changes in sedimentation rates in the Nile Delta can provide a high-resolution archive of climatic fluctuations in the Nile headwaters. For example, Marriner et al. (2012a) examined 105 sediment cores and sections across the Nile deltaic plain and produced a detailed time series of changes in sedimentation during the past 8,000 years supported by 318 calibrated 14C ages. The mean rates ranged up to 220 mm/century. They reconstructed separate records for different components of the Delta, consisting of pro-delta sediments, marine sands and muds, lagoon or marsh muds, fluvial sands and silts and peat deposits. They then compared dated peaks and troughs in sedimentation rates with various proxy palaeoclimatic records, such as cave records from Oman and China, which show a strong Early Holocene monsoon signal, also evident in the Nile fluvial record, and, indirectly, in high rates of sedimentation in the Nile Delta. More directly pertinent as proxy records of past climatic changes in the Nile Basin is a diatom record from Lake Victoria (Stager et al., 2003; and Chapter 7) and a diatom record from Lake Abiyata in the Ethiopian Rift (Chalié and Gasse, 2002; and Chapter 6). When the Nile Delta sedimentation record was compared to the Holocene record of moderate to strong El Niño Southern Oscillation (ENSO) events obtained by Moy et al. (2002) from Laguna Pallcacocha in southern Ecuador, Marriner et al. (2012a) found a reasonably good

20.6 Holocene Variations in Nile Delta Subsidence

287

correlation between times of frequent ENSO events in Ecuador and times of fluctuating sedimentation in the Delta. The influence of ENSO events on historic Nile flow is well established, with El Niño events coinciding with times of low Nile flow and La Niña events with times of high Nile flow (Whetton et al., 1990; Whetton and Rutherfurd, 1994; Ortlieb, 2004; see Chapter 3). Marriner et al. (2012a) have successfully shown that this relationship extends back at least 8,000 years. Marriner et al. (2012a) proposed that as the summer monsoon weakened and the northern limit of the ITCZ shifted progressively further to the south, the influence of ENSO events on Nile discharge and sediment transport seems to have dwindled. It is not clear what the mechanism contributing to this may be, particularly given that Moy et al. (2002) identified far fewer ENSO events during the first half of the Holocene than in the last 5,000 years. Human modification of the Delta through dams, irrigation canals and drainage is increasingly evident from about 4 ka onwards (Marriner et al., 2012a, 2012b), and would have had a significant impact on sedimentation rates. Distinguishing between human impact and that of natural climatic fluctuations on Nile sediment transport and deposition remains a difficult task. Stanley and Warne (1993b, 1998) have provided a solid stratigraphic framework for the late Pleistocene and Holocene sectors of the Nile Delta with special attention to the northern sector of the Delta where the interplay between marine and fluvial sedimentation was most evident. Pennington et al. (2017) have built upon this earlier work with a geographical focus on the 15,000 km2 fluvial plain in the southern sector of the Delta. Using information from 1,640 boreholes and a somewhat eclectic set of 71 ages (OSL, 14C and archaeological), they were able to reconstruct changes in deltaic sedimentation between 8 and 4.5 ka. During the initial phase of high relative sea level rise, the channels steepened their gradient, flood plain aggradation was rapid, the braided or anastomosing channels were highly unstable and prone to avulsion and sudden shifts in location, and the adjacent land was a swampy wetland. As the rate of relative sea level rise diminished, lateral channel migration replaced vertical channel aggradation, the channels began to meander, and soils began to develop on the now welldrained flood plain. This change began sooner in the south than in the north, so that fluvial aggradation of the Delta was time-transgressive. The southern locus of landscape stability initiated soon after 7.4–7.2 ka expanded progressively northwards until by 4.5 ka almost all the Delta was covered by laterally meandering river channels flanked by well-drained flood plains. These conditions proved favourable to a change from a hunter-gatherer economy to one based on plant and animal domestication. This in turn provided the foundation for the emergence some 5,000 years ago of the Predynastic human settlements, which ultimately culminated in the urban Egyptian civilisation of Dynastic times (Stanley and Warne, 1993b).

20.6 Holocene Variations in Nile Delta Subsidence All of the major deltas in the world are prone to subsidence under the weight of the overlying mass of sediment. Such subsidence can lead to a subsurface influx of salt water

288

The Nile Delta

from the sea, and has been causing changes in groundwater chemistry of the Delta for a number of decades already (Appelo, 1990). The question of which portions of the Nile Delta are likely to be subsiding most rapidly in the future is one of great potential importance for the future well-being of the inhabitants of the Delta. In order to resolve this question what is required is a detailed record of geologically recent subsidence rates across the Delta, particularly along the northern margins of the Delta adjacent to the sea. Marriner et al. (2112b) analysed the subsidence rates revealed in organic-rich peat and lagoon deposits in the northern sector of the Delta, for which 194 calibrated 14C ages were available. Subsidence rates ranged from 0.03 to 4.5 mm/yr. The fastest rates were in four large lagoons, including the Maryut Lagoon discussed in Section 20.3. Each lagoon was located above a late Pleistocene to Holocene set of alluvial deposits forming valley-fills within deeply entrenched LGM Nile distributary channels. Subsidence was brought about by a combination of compaction from the weight of the alluvial overburden, oxidation of organic material, and dewatering. A decline in Nile sediment input from about 4 ka onwards meant that the sediments laid down across the Delta after that time were more prone to interstitial water loss and showed subsidence rates that were significantly faster than during the previous millennia (Marriner et al., 2012b). Human impact on the Delta during the past 4,000 years involved digging canals, draining swamps and erecting dams across distributary channels. All of these activities acted to accelerate subsidence rates during the late Holocene. These trends are likely to continue, especially because much of the sediment that used to accumulate on the Delta is now trapped upstream of the Aswan High Dam, which was completed in 1964. Garzanti et al. (2015) calculated the most recent Nile sediment budget from the volume of sediment trapped in reservoirs between 1964 and 1998. They noted that coastal erosion in northern Egypt is a direct result of the inability of trapped sediment to reach the Mediterranean and contribute to coastal aggradation. The long-term dynamic equilibrium between coastal erosion and coastal deposition has now been irreversibly upset, with coastal erosion likely to continue in the future.

20.7 Human Occupation of the Nile Delta The Nile Delta formed during the last 6,000–8,000 years, so that we cannot expect to find human occupation sites older than the oldest Delta sediments. Butzer (1976, p. 25) commented that ‘since 10 m of alluvium were deposited during the last 6,000 years, it is not surprising that there is no Predynastic record from the delta proper’. Under the general term Early Predynastic, he included Fayum A Neolithic (6.7–6 ka) and Merimde (6.9–6.5 ka) (Butzer, 1976, Table 1). The Neolithic site of Merimde is located on the left bank of the most westerly major Nile Delta distributary channel roughly 40 km downstream from the apex of the Delta (Fig. 20.1). Hassan (1980, p. 438–439) considered that ‘the Fayum A and Merimde Neolithic are the oldest farming cultures in

20.7 Human Occupation of the Nile Delta

289

the Nile Valley dating back to ca. 7,000–5,500 years B.P. (tree-ring corrected radiocarbon ages’. Given the strategic location of both of these sites in relation to the likely Neolithic migration routes across the Sinai, it is no surprise that they are comparatively old. Butzer (1976) may not have been entirely correct in dismissing any Predynastic record from the Delta proper. A recent study by Véron et al. (2013) of lead isotopes in sediment cores from the marine bay of Alexandria and in the nearby Maryut Lagoon indicated relatively high levels of lead contamination during the Egyptian Early Dynastic (2,897 ± 187 BC) and Predynastic (3,520 ± 145 BC). The absence of such lead pollution in the nearby Canopic branch of the Nile (Fig. 20.1) means that we can rule out the Nile as the source of the lead. The lead isotopes are consistent with possible sources in Cyprus, Crete, Turkey and the Oman Gulf, where both lead and copper were being mined at these times. Alexander the Great established Alexandria in 331 BC, on the sites of the ancient town of Rhakotis, mentioned by Herodotus during his visit to Egypt some 2,500 years ago. The inhabitants of the western sector of the Delta thus appear to have had metal trading links within the Eastern Mediterranean during the Chalcolithic–Early Bronze Age transition 6,000 years ago, while those in the eastern sector of the Delta had more direct overland links with Sinai and the Levant (Véron et al., 2013).

Figure 20.5 Depth profile of 87Sr/86Sr from coastal core S-21 in the Nile Delta east of the Suez Canal. (After Krom et al., 2002 and Stanley et al., 2003.)

290

The Nile Delta

20.8 Conclusion The Nile Delta occupies an area of 22,000 km2, which amounts to 63% of Egypt’s agricultural land, and provides a living for more than 50 million people. The presently visible portions of the Delta began to form during the Holocene marine transgression, which came to a close about 6,500–6,000 years ago. Depending on the input of Nile sediment ferried in from upstream, the Delta lagoons have fluctuated from being brackish to saline, reflecting a strong marine influence, to being full and fresh, reflecting a strong Nile influence. Fluctuations in the 87Sr/86Sr values preserved in Nile Delta clays have been used to reconstruct a Holocene history for the main Nile over the past 8,000 years. Strontium isotope analyses of Delta sediments together with archival evidence also indicate that the collapse of the Old Kingdom some 4,200 years ago was associated with several phases of intense drought in the Nile headwaters and greatly reduced Nile floods in Egypt. During the past 5,000 years the variations in sedimentation rates along different parts of the Delta appear to display a strong El Niño Southern Oscillation (ENSO) signal. Fluctuations in sediment subsidence rates along the northern margins of the Delta reflect fluctuations in inputs of Nile sediment and have increased during more recent years. Following completion of the Aswan High Dam in particular, which trapped a very large proportion of the Nile sediment, coastal erosion has led to a loss of deltaic land to the sea.

21 The Nile Cone

All the rivers run into the sea; yet the sea is not full; unto the place from whence the rivers come, thither they return again. Ecclesiastes 1.7 If you take a cast of the lead a day’s sail off-shore, you will get eleven fathoms [ca. 20 m], muddy bottom – which shows how far out the silt of the river extends. Herodotus (ca. 485–425 BC), The Histories (1954, p. 104)

21.1 Introduction The Nile Cone is the seaward extension of the Nile Delta and, on present evidence, appears to have been accumulating for at least 30 million years (Macgregor, 2012; Fielding et al., 2016, 2018). The older portions of the Nile Cone (Oligocene and Miocene in age, or 30–5.3 Ma) extend north across the floor of the Eastern Mediterranean for roughly 400 km and span an E–W distance of ca. 800 km (Fig. 21.1a). The younger portions of the Cone (Pliocene and Pleistocene in age, or 5.3 Ma to 11 ka) are almost as extensive, with an E–W extent of ca. 700 km and a northward extent of ca. 300 km (Fig. 21.1b). Apart from its actual and potential economic value as a source of oil and gas (Makled and Mandur, 2016), the Nile Cone contains a relatively complete record of major Nile floods. The purpose of this chapter is to review this record and to offer a synthesis of the Quaternary depositional record preserved within the younger and better-dated portion of the Cone.

21.2 Age and Volume of the Nile Cone The Nile is one of the most intensively studied rivers in the world. Despite more than a century of detailed investigation throughout the basin, until very recently there was still very little agreement over when the Nile drainage network arose. Estimates varied from Eocene to Holocene. One group of workers argued that the Nile is no older than Pleistocene, when pyroxenes and other minerals diagnostic of an Ethiopian volcanic source first become evident (Shukri, 1949). Another group advocates a much earlier connection between the 291

292

The Nile Cone

Figure 21.1 The Nile Cone at different stages in its formation. (a) Oligocene and Miocene (30–5.3 ma). (b) Pliocene and Pleistocene (5.3 Ma to 11 ka). LB denotes Levantine Basin, R the Rosetta Depocentre, and D the Damietta Depocentre. (Simplified from Macgregor, 2012, Fig. 5.)

Ethiopian headwaters of the Nile and the desert Nile in Sudan, with evidence of Oligocene incision by the Blue Nile and Tekezze Rivers (McDougall et al., 1975). Fielding et al. (2016, 2018) have now demonstrated quite conclusively that the modern Nile flow regime was already underway by 31 Ma, confirming an Oligocene connection between the Ethiopian headwaters and the Nile Cone. They analysed sediment samples retrieved from the Nile Cone with biostratigraphic ages of 31, 27.5, 17, 15.5 and 15.2 Ma, 3.25–2.65 Ma and 1.295 Ma. Their analysis of Sr-Nd bulk data and of detrital zircon U-Pb and Hf-isotope data showed a persistent signal in offshore sediments extending back to at least 31 Ma, consistent with a sustained and relatively steady input of sediments derived from the Ethiopian Continental Flood Basalts via the Blue Nile and Atbara tributaries of the desert Nile. Fielding et al. (2018) also noted that Eocene sediments onshore in northern Egypt may yet reveal a still earlier Ethiopian connection, and await future study. There has also been considerable debate about the volume of the Nile Cone and its sources of sediment. Macgregor (2012, Table 2) estimated that the volume of compacted sediment deposited onto the Nile Cone between 30 and 10 Ma amounted to 188,097 km3. This was followed by deposition of a further 393,265 km3 during the last 10 Ma. The total compacted volume for the period 30–0 Ma is therefore 581,362 km3, which is substantially higher than earlier estimates. For example, Harrison (1955) used gravity anomaly data to estimate the areal extent and thickness of the Nile Cone, and came up with a figure of 95,000 km3 assuming no crustal sag and with a figure of 220,000 km3 allowing for the late Cenozoic crustal sag that has indeed taken place. Emery et al. (1966) used Harrison’s submarine contour map and considered that the volume was closer to 140,000 km3 and most likely less. Wong and Zarudski (1969) considered this an underestimate and McDougall

21.3 Analysis of Marine Sediment Cores

293

et al. (1975) concluded that a volume between 100,000 and 200,000 km3 was plausible, on the grounds that this volume matched that of the gorges eroded in the headwaters of the Blue Nile and Tekezze–Atbara Rivers in Ethiopia. However, as Macgregor has shown, the Ethiopian headwaters of the Nile were not the only contributors of sediment to the Nile Cone. Talbot and Williams (2009, pp. 39–40) have also pointed out that incision by the ancestral Nile during the late Miocene created a canyon between the coast and as far upstream as Aswan, from which roughly 80,000 km3 of rock had been eroded before the Pliocene transgression converted the canyon into a vast estuary that was filled with a similar volume of sediment from the aggrading Nile. Very rapid rates of Pliocene sedimentation are also evident in two marine sediment cores (Baltim-1 and NDOB-1) drilled into the Nile Cone (Makled and Mandur, 2016, Figures 3 to 5). Macgregor (2012) estimated likely rates of denudation in different sectors of the Nile Basin over the past 30 Ma using a combination of geomorphic evidence (incision into surfaces of known age) and apatite fission track analysis for certain portions of the Red Sea Hills, where he inferred 1.3 km of mean surface lowering in the past 30 Ma. He concluded that the Red Sea Hills region alone had contributed 217,249 km3 or 37% of the total amount of sediment in the Nile Cone (Macgregor, 2012, Table 1). Earlier workers had neglected the contribution from the Red Sea Hills and had greatly underestimated the quantity of rock eroded from northern Sudan and the now arid Western Desert of Egypt, so that if we exclude the post-depositional carbonate present in the sediments, a Nile Cone volume slightly in excess of 500,000 km3 seems reasonable.

21.3 Analysis of Marine Sediment Cores from the Nile Cone A variety of analyses (terrestrial and marine microfossils, geochemical and isotopic composition) of marine sediment cores obtained from various depths within the Nile Cone have been used to reconstruct past changes in sediment input and freshwater discharge from the Nile Basin (Rossignol-Strick et al., 1982; Rossignol-Strick, 1985, 1999; Wehausen and Brumsack, 1998; Ducassou et al., 2008, 2009; Rohling et al., 2009; Revel et al., 2010, 2015; Zhao et al., 2011, 2012; Blanchet et al., 2013, 2015; Hennekam et al., 2014, 2015; Makled and Mandur, 2016). Some of these studies have focused on the presence or absence of sapropel layers within the Nile Cone and further afield, which we discuss in greater detail in Section 21.4. In this section, we focus more on geochemical and isotopic studies. Zhao et al. (2011) investigated the oxygen isotope variations in marine core MD90-964 on the distal Nile fan in the Levantine Basin (Fig. 21.1) of the eastern Mediterranean. The core extended back in time to 1.75 Ma. They identified 21 sapropel layers, which they defined as sediment layers with at least 1% organic carbon. These layers they found to be enriched in barium. They identified another 21 dark layers enriched in barium but from which the carbon had been oxidised and removed. These they termed ‘ghost sapropels’ and ‘hidden sapropels, depending on the strength of the geochemical evidence that they had once been sapropels. Fluctuations in the ratio of titanium to vanadium (Ti/V) at different

294

The Nile Cone

depths in the core were used as a proxy for fluctuations in the amount of suspended sediment delivered to the Nile from the volcanic highlands of Ethiopia. These fluctuations showed a 23-ka periodicity, indicative of strong precessional influence (see Chapter 4). They also found a 78-ka signal, the cause of which remains enigmatic, although it could represent two obliquity cycles. Each obliquity cycle lasts about 41 ka and reflects changes in the tilt of Earth’s axis from a maximum of 24°36 0 to a minimum of 21°59 0 . Times of maximum inclination are times of hotter summers and colder winters in high latitudes; times of minimum tilt are associated with mild winters and mild summers. The precessional cycle over the past million years has varied between 16.3 ka and 25.8 ka and over the past 10 Ma has varied between 23 ka and 19 ka (Chapter 4). This cycle reflects the changing season of the year during which Earth is closest to the sun and is controlled by the direction in which the spin axis of Earth points in space. At times when the sun is closest to Earth during the northern summer, the tropical monsoons will tend to be stronger and summer precipitation will also be both more intense and greater in amount. Zhao et al. (2011) considered that during episodes of higher monsoonal precipitation in the Ethiopian Highlands, the vegetation cover would be denser and more widespread, leading to a reduction in erosion and hence in the sediment load of the Blue Nile and Atbara. They concluded, somewhat cryptically, that fluctuations in Nile sediment discharge ‘are more the result of river transport capability than of erosion potential in source areas’ (Zhao et al., 2011, p. 239). In a subsequent study of the ratio of iron to aluminium (Fe/Al) within the same core (MD90-964), Zhao et al. (2012) used variations in this ratio as an index of fluctuations in the contribution of iron-bearing heavy minerals derived from the Ethiopian headwaters of the Nile. They also considered this index to be an indirect proxy for precipitation changes in that region. Their key finding was that times when the Fe/Al ratio was highest often coincided with times of sapropel formation, with a spectral peak indicating a 23 ka precessional signal. They reviewed earlier studies from the Nile Delta and proximal Nile Cone based on Sr and Nd isotopic ratios (Foucault and Stanley, 1989) and the clay mineral composition of sapropels (Revel et al., 2010). These studies indicated reduced inputs of fine-grained Nile sediments during times of sapropel formation in the eastern Mediterranean but higher Nile freshwater discharge during times of stronger summer monsoon. They did not attempt to explain this apparent paradox. The factors determining Nile discharge and Nile sediment load are complex and are reviewed in Section 21.5. We now return to the question of Nile floods and sapropel formation.

21.4 Nile Floods and Sapropel Formation Marine sediment cores from the floor of the Mediterranean have revealed the presence of multiple sapropel layers in Neogene (Miocene and Pliocene) and Quaternary sediments (Lourens et al., 1996; Cramp and O’Sullivan, 1999; Larrasoaňa et al., 2003). The word sapropel comes from the Greek words saprós, meaning putrid or decayed, and pēlós,

21.4 Nile floods and Sapropel Formation

295

meaning clay or earth. The Glossary of Geology defines sapropel as ‘an unconsolidated, jellylike ooze or sludge composed of plant remains, most often algae, macerating and putrefying in an anaerobic environment on the shallow bottoms of lakes and seas. It may be a source material for petroleum and natural gas’ (Bates and Jackson, 1987, p. 588). The Encyclopaedic Dictionary of Physical Geography defines sapropel as ‘a mud or ooze composed predominantly of anaerobically decomposing organic material, usually in aquatic environments’ (Goudie et al., 1985, p. 376). The anaerobic conditions that allow sapropels to form on the bed of the eastern Mediterranean are thought to reflect a change in seawater stratification caused by a sustained influx of freshwater from major rivers such as the Nile (Rossignol-Strick et al., 1982; Rossignol-Strick, 1985,1999), as well as possible outflows from the Black Sea. Another possibility is increased primary productivity as a result of an influx of nutrients followed by preservation under anoxic conditions (Cramp and O’Sullivan, 1999). These two sets of processes are not mutually exclusive. Lourens et al. (1996) provided a chronology of sapropel formation in the Mediterranean calibrated against the astronomical time scale. Their ages are midpoint ages for sapropel formation and there appears to be a 3,000-year time lag between sapropel formation and the correlative precessional minimum (see Chapter 4) shown in their insolation index. The sapropels are numbered consecutively from the youngest (S1) back in time. Cramp and O’Sullivan (1999) noted the strong lack of any spatial and temporal continuity of the sapropels, which they considered was a result of redeposition and/or post-depositional geochemical alteration of the sapropels, the former process shown by the fact that some sapropel layers are more than 4 m thick, which would be unusually thick for an undisturbed sapropel layer. They further noted that although most sapropels coincided with warmer climatic phases on land (i.e., interglacial and interstadial phases), some also formed during times of colder climate, such as sapropel S8 and S6. The conspicuous sapropel S5 was synchronous with marine isotope stage 5e (125 ka), or the peak of the last interglacial. In the case of the Nile, the floodwater hypothesis is, in principle, eminently testable by comparing well-dated phases of very high Nile discharge with dated phases of sapropel formation. While such hypothesis testing is certainly possible for about the last 125,000 years (Williams et al., 2015a), it becomes more difficult with earlier Nile flood episodes simply because the ages obtained for these earlier events have much larger error terms, making it hard to establish whether the flood phases were indeed synchronous with the sapropels. The most recent sapropel S1 in the eastern Mediterranean is a composite unit, with cited ages showing a high degree of variability (Thomson et al., 1999; Mercone et al., 2001). For example, Cramp and O’Sullivan (1999) suggest an age range of 12–6 ka. Other age estimates suggest 13.7–12.4 ka near the base and 9.9–8.9 ka near the top (Ducassou et al., 2009), 9.5–6.6 ka (Revel et al., 2010) and 10.1–6.5 ka with a gap at 8.2–7.9 ka (Hennekam et al., 2014). In fact, formation of sapropel S1 may have ended as recently as 5 ka (Higgs et al., 1994), which is also when the Nile deep-sea turbidite system became inactive as a result of reduced sediment discharge from that river associated with reduced rainfall in the Ethiopian headwaters of the Nile (Ducassou et al., 2008, 2009). There are

296

The Nile Cone

Table 21.1 East Mediterranean sapropel ages for the past 200,000 years Sapropel unit

Sapropel age (after Lourens et al., 1996)

Sapropel age range (after Ducassou et al., 2009)

Marine isotope stage

Sea surface temperature

S7 S6 S5 S4 S3 S2 S1

195 172 124 102 81 55 8

200–194 180–170 125–118 100–96 82–78 56–54 10–6

MIS 7 MIS 6 MIS 5e MIS 5 c MIS 5a MIS 3 MIS 1

W C W W W W W

The ages in column 2 are from Lourens et al. (1996); the ages in column 3 are estimated from Ducassou et al. (2009), Figures 4 and 5; column 4 shows Marine Isotope Stages; in column 5, W is a warm climatic interval and C is a cold climatic interval.

several reasons for these age discrepancies, of which the most obvious is differential rates of post-depositional oxidation of the sapropel. Another is that the sediment cores come from different sectors of the Nile Cone, some of which might have experienced low rates of sedimentation whereas others might have experienced more rapid sedimentation at the same time, which would have been more conducive to sapropel formation. Since the Nile Delta distributary channels were active at different times, we would expect to find varying rates of sediment deposition across the proximal sector of the Nile Cone (Hennekam et al., 2015). Three deep-sea sediment cores were retrieved from the western sector of the Nile Cone at depths of 1389 m (MS27PT), 2823 m (FKS05) and 2221 m (FKS04), at up to 200 km NW of the present coast (Ducassou et al., 2008). The clastic mud beds rich in terrestrial organic matter have been reworked by turbidity currents and are characterised by an upward coarsening particle size composition followed by an upward fining sequence, indicative of a waxing flow succeeded by a waning flow regime. Ducassou and her co-authors concluded that the mud beds rich in organic matter indicated higher Nile discharge and wetter intervals in the Nile headwaters, with the most recent wet phase lasting from 12–10 ka to 6–4 ka (Ducassou et al., 2008). A comprehensive analysis of more than forty deep-sea sediment cores collected across the Nile Cone and spanning the last 200 ka enabled Ducassou et al. (2009) to derive a detailed chronology for changes in sediment input and for times of sapropel formation associated with episodes of very high Nile discharge (Ducassou et al., 2009, Figs. 4 and 5). Table 21.1 summarises this work. Turbidity flows in deep-sea fans on the Nile Cone were most active during times of rising and high sea level associated with wetter climates, and least active during times of low sea level or high sea level stands coupled to arid periods, as in the last 5,000 years. Capozzi and Negri (2009) independently confirmed the important role played by sea level fluctuations upon deep water circulation in the Mediterranean and indirectly on the timing of sapropel formation.

21.4 Nile floods and Sapropel Formation

297

Revel et al. (2010) carried out a detailed study of fluctuations in the strontium and neodymium isotopic composition of the terrigenous component of sediment in marine core MS27PT located 100 km WNW of the Rosetta distributary outlet on the Nile Delta (see Chapter 20, Fig. 20.1). They also analysed changes in major elements, with a 10-year resolution for iron fluctuations during wetter periods across North Africa in the past 100,000 years (98–69 ka, 60–50 ka, 38–30 ka and 14–5 ka). They found high amounts of desert dust influx during the arid phases coinciding with MIS 4 and MIS 2 (Last Glacial Maximum [LGM]), which were also times of eolian dust deposition in East Antarctica. The 98–69 ka wet phase coincides with the formation of sapropels S4 and S3, the 60–50 ka wet phase with sapropel S2, and the 14–5 ka wet phase with sapropel S1. Their final conclusion was that the Ethiopian headwaters of the Blue Nile and Tekezze–Atbara Rivers were significantly wetter during 8–5 ka, with a longer rainy season and/or highly variable rainfall intensity. However, a 12-m sediment core obtained from the Dendi eastern crater lake near the Blue Nile headwaters in the central Ethiopian uplands indicates peak wetness (and, presumably, high Blue Nile flow) between 10.0 and 8.7 ka and peak aridity between 4.0 and 2.6 ka (Wagner et al., 2018). The response of the Nile to millennial-scale climatic change since the LGM was the focus of a study by Box et al. (2011) in which they analysed the 87Sr/86Sr ratios and major element geochemistry from two sediment cores. Core 9501 was retrieved on the distal edge of the Nile Cone, south of Cyprus; core 9509 came from the eastern margin of the Nile Cone off the coast of southern Israel. The authors noted that the sediment in these cores represented suspended sediment discharged from the Nile into the Levantine Basin, and so provided a record of erosion and sediment transport in the Nile headwaters. They found that during the dry interval of the last 5,000 years as well as before 11 ka, in core 9509, the influx of sediment deduced to be from the Blue Nile/Atbara (10–12 g/cm2 per yr and ca. 6 g/cm2 per yr, respectively) was greater than during the 11–5 ka period of peak humidity, when it averaged only 2 g/cm2 per yr. The sediment flux from the White Nile showed a very different response, increasing from 5 g/cm2 per yr at ca. 13 ka to >15 g/cm2 per yr by 5 ka. The influx of Saharan dust decreased during this interval, but was high both before and after the 13–5 ka humid episode. The authors postulated that during wetter intervals in the seasonally wet Ethiopian Highlands, expansion of the plant cover caused a reduction in erosion and sediment yield in the Blue Nile/Atbara headwaters, but in the perennially wet Ugandan headwaters of the White Nile, an increase in precipitation would generate an increase in erosion and in the transport of sediment through the Sudd marshes. This interpretation raises several questions. First, until about 5 ka, there was overflow from Lake Turkana via the Pibor into the Sobat and so into the White Nile (see Chapter 9). White Nile sediment would therefore show an enhanced volcanic signature during wetter climatic intervals. How is this sediment input into the lower White Nile to be distinguished from the contributions emanating from the Blue Nile and Atbara? Second, wadis were actively contributing sediment to the Desert Nile during wetter climatic periods (Woodward et al., 2015 and Chapter 14). Some of this sediment came from the Red Sea Hills, where the bedrock geology is not too dissimilar to that in the White Nile headwaters. Some came from

298

The Nile Cone

erosion of the Mesozoic Nubian Sandstone Formation, and some represented recycled desert dust. It would be easy to mistake the sediment input from these wadi sediments for a higher White Nile sediment input. Third, and equally important, it is by no means certain that an increase in precipitation in the equatorial headwaters of the White Nile would necessarily lead to an increase in sediment yield. The opposite is more likely. Peak sediment yields in present-day rivers occur in the monsoonal tropics and in semiarid regions, but not in rivers flowing through densely vegetated catchments in equatorial climatic regions (Douglas, 1967, 1969; Milliman and Meade, 1983; Milliman, 1997; Williams, 2012a). These caveats are not intended as a criticism of the quality of the analytical work conducted on the marine sediment cores. Rather, they illustrate the complexity of the Nile’s response to climatic change and the difficulties involved in construing subtle environmental changes in the Nile headwaters from the geochemical, isotopic and microfossil evidence preserved in marine sediment cores retrieved from the Nile Cone. A further example of complex response to climatic change in the Nile Basin is shown by the work of Cécile Blanchet and her co-workers (Blanchet et al., 2013) on Holocene marine sediment core P362/2–33 collected from the SW margin of the Nile Cone offshore from the Rosetta distributary outlet in the Nile Delta. Using a combination of grain size analysis, bulk element composition and variations in the strontium and neodymium isotope values, they inferred high Blue Nile sediment input during 9.5–7.3 ka, which they attributed to high spring insolation, and low Blue Nile sediment influx during 7.3–4 ka, which they related to high autumn insolation in the Ethiopian Highlands. Two arid phases were identified between ca. 8.5 and 7.3 ka and between 4.5 and 3.7 ka. The White Nile was thought to have had a high sediment input to the Nile during 8–4 ka, when the inferred Blue Nile sediment influx was relatively low. Once again, disentangling the White Nile signal is far from simple, so that inferences about the past role of the White Nile should be viewed with considerable caution. Hennekam et al. (2014) analysed the stable oxygen isotopic composition of the planktonic foraminifer Globigerinoides ruber (∂18Oruber) and the bulk sediment inorganic geochemistry in a Holocene marine sediment core (PS009PC) from the southeast Levantine basin of the Nile Cone (Fig. 21.1). The high rates of sediment deposition at this site enabled them to compile a high-resolution record of changes in Nile discharge, as well as to identify times when the main source of precipitation in the Nile headwaters was via the Indian Ocean or the Atlantic. They found that early Holocene Nile discharge came primarily from the Indian Ocean, with a peak at 9.5 ka. Oscillations in the summer monsoon at this time appeared to reflect fluctuations in solar activity. They identified five periods of enhanced Nile flow, with a recurrence interval of 500 to 1,000 years, associated with evidence of increased anoxia in the marine sediment core. The calibrated ages they obtained for these periods of very high Nile flow were ca. 9.7 ka, ca. 9.1 ka, ca. 8.6 ka, 7.7 ka and 6.6 ka. These ages are in remarkably good agreement with independent evidence of high White Nile flow at 9.7–9.0 ka, 7.9–7.6 ka and 6.3 ka, and for high Blue Nile flow at 8.6 ka, 7.7 ka and 6.3 ka (Williams, 2009a, and Chapters 8 and 11). In this core, sapropel S1 is a composite unit, with ages of 10.1–8.2 ka for the lower portion S1a, and 7.9–6.5 ka for the upper portion S1b

21.4 Nile floods and Sapropel Formation

299

Figure 21.2 Phases of sapropel formation in the Nile Cone in relation to episodes of high Nile flow during the last 250 ka. (After Williams et al., 2015a, Fig. 8.)

(Hennekam et al., 2014). Whether the 8.2–7.9 ka hiatus relates to the 8.2-ka cold event in the North Atlantic and Greenland remains uncertain but it is coeval with a phase of much reduced Nile flow. To summarise thus far: during phases of very high Nile flow, clastic muds rich in continental organic matter and highly organic sapropels accumulated on the floor of the eastern Mediterranean (Fig. 21.2) (Krom et al., 2002; Ducassou et al., 2008, 2009; Rohling et al., 2009; Revel et al., 2010, 2015; Zhao et al., 2011, 2012; Blanchet et al., 2013, 2015; Hennekam et al., 2014, 2015; Makled and Mandur, 2016). The sapropel record reveals that episodes of middle to late Pleistocene high flow in the Blue and

300

The Nile Cone

White Nile coincide very broadly with sapropel units S8 (217 ka), S7 (195 ka), S6 (172 ka), S5 (124 ka), S3 (81 ka), S2 (50 ka) and S1 (Williams et al., 2003, 2010, 2015a; McDougall et al., 2008). Sapropel 5 (124 ka) was synchronous with major flooding and the formation of the 386-m lake in the lower White Nile Valley (Barrows et al., 2014) and with a prolonged wet phase at ca. 125 ka at Kharga Oasis in the Western Desert of Egypt (Kieniewicz and Smith, 2007). Although the sapropel record in the eastern Mediterranean is incomplete, with some evidence of complete removal of sapropels by post-depositional oxidation (Higgs et al., 1994), it is nonetheless a longer and more complete record than that presently available on land, and so can serve as a useful surrogate record for Nile floods and phases of enhanced summer monsoon precipitation.

21.5 Conclusion The Nile Cone extends north from the present coastline for 300–400 km and spans an E–W distance of 700–800 km (Fig. 21.1). Its total sediment volume amounts to about 580,000 km3, of which 180,000 km3 were deposited between ca. 30 Ma and 10 Ma (Oligocene and Miocene), and the remaining 400,000 km3 during the last 10 million years (Pliocene and Quaternary). About 37% of the total sediment in the Nile Cone (220,000 km3) came from erosion of the Red Sea Hills and roughly 100,000 ± 50,000 km3 of rock came from erosion in the headwaters of the Blue Nile and Tekezze/ Atbara Rivers. Deep-sea sediment cores contain a useful, if sometimes enigmatic, record of past changes in Nile sediment influx and an indirect record of possible changes in precipitation, erosion and sediment yield in the Ethiopian and Ugandan headwaters of the Nile. Clastic muds rich in continental organic matter and highly organic sapropels accumulated on the floor of the eastern Mediterranean during phases of very high Nile flow. Episodes of middle to late Pleistocene high flow in the Blue and White Nile coincide very broadly with sapropel units S8 (217 ka), S7 (195 ka), S6 (172 ka), S5 (124 ka), S3 (81 ka), S2 (50 ka) and S1. Sapropel S1 is a composite unit, with ages of 10.1–8.2 ka for the lower portion and 7.9–6.5 ka for the upper portion in cores studied by Hennekam et al. (2014). The 8.2–7.9 ka hiatus coincides with a phase of much reduced Nile flow.

22 Origins of Plant and Animal Domestication in the Nile Basin

Or ever the silver cord be loosed, or the golden bowl be broken, or the pitcher be broken at the fountain, or the wheel broken at the cistern. Then shall the dust return to the earth. . . Ecclesiastes 12/6–7

22.1 Introduction The quotation at the head of this chapter is an elegant expression of the ephemeral nature of archaeological remains. Our focus in this book has been concerned primarily with the reconstruction of prehistoric environments in the Nile Basin, with occasional forays further afield. The delicate forensic task of reconstructing past environmental changes requires us to draw on evidence from a wide range of disciplines, including the earth and biological sciences as well as archaeology, anthropology and linguistics (see Table 4.1 in Chapter 4). Throughout the last 2.6 million years of Quaternary time, as well as long before then, regional climates within the Nile Basin and adjacent areas have fluctuated from more to less humid, with cooler temperatures on land and sea during global glacial phases and warmer temperatures during interglacial phases (see Chapters 1 and 4). Seasonal movements of the Intertropical Convergence Zone (ITCZ) have an important influence on precipitation in the African tropics (see Chapter 3). During the Northern Hemisphere summer, the northward progression of the ITCZ is associated with summer rain across the southern half of the Nile Basin. Monsoonal influences accentuate the influence of the ITCZ and bring moist air from the Atlantic and Indian Oceans to the uplands of Ethiopia and Uganda and the vast lowland plains of South Sudan and southcentral Sudan. In central Sudan, this influence is felt as far north as the confluence of the Blue and White Nile Rivers at Khartoum, with that city receiving on average about 175 mm of rain a year during the past 80 years, but far less during years of prolonged drought. North of Khartoum rainfall decreases rapidly so that there is virtually no rain today in the Nile valley in the vicinity of Aswan in southern Egypt, as well as in the Western Desert of Egypt and in the Eastern Sahara more generally. Modest winter rains bring some relief to the farmers on the Nile Delta north of Cairo, but even then, their life would be perilous indeed

301

302

Origins of Plant and Animal Domestication

without resort to irrigation waters from the Nile. It is against this background of fluctuating climates, which also controlled the flood regime of the Nile, that we need to consider some of the questions relating to the adoption of plant and animal domestication in different parts of the Nile Basin and adjacent areas. In the Preface to this book I stated that ‘the most recent northward excursion of the ITCZ was during the Early to Middle Holocene, when groups of Mesolithic people made a living from fishing, hunting and gathering wild plant foods. By about 8,000–7,000 years ago we see the inception of plant and animal domestication in the Nile valley, several thousand years after its adoption by Neolithic communities in the Fertile Crescent of the Levant and Anatolia. Why was the onset of the Neolithic so late in the Nile Basin compared to further north?’ The aim of this chapter is to provide some possible answers to this and related questions. Throughout this work, the term Mesolithic represents the intermediate cultural stage between Upper Palaeolithic and Neolithic. Mesolithic people relied on hunting, gathering and fishing to obtain food. Their tool kit included grindstones, pottery, bone points and the use of small, carefully flaked stone tools, some of which were attached to bone or wooden handles and used as sickles to harvest wild grass seeds. Some workers use the term Epipalaeolithic or Terminal Palaeolithic as synonyms for Mesolithic, and we have used these terms in reviewing their work in earlier chapters, but for simplicity and clarity we will retain the term Mesolithic in this chapter. The term Neolithic refers to those Stone Age cultures in which food production was based partly or wholly on the use of domesticated plants and/or animals. The tool kit included pottery and grindstones as well as edge-ground and polished stone implements that were hafted and used as hatchets, chisels and adzes. Perforated stones attached as weights near the hardened and pointed end of digging sticks were used in cultivation, much as they are today in parts of south-eastern Ethiopia. The transition from Mesolithic to Neolithic was often very gradual, so that it is not particularly useful to become too preoccupied with precise identifications of when the one ended and the other began. Indeed, Holdaway et al. (2017a, b, c) suggest that in the context of their geo-archaeological work in the northern Fayum, the terms Neolithic and Epipalaeolithic should perhaps be abandoned. An example from soil science may serve to illustrate this dilemma. In some instances, the boundary between different soil units or horizons can be very clear and sharp, but in other cases the vertical transition can be very diffuse and gradual, so that it is not easy to specify where one horizon ends and the other begins. Lateral transitions from one soil type to another can be equally indistinct, prompting Milne (1935, 1936) to refer to spatially linked hill slope soils that exhibit progressively different physical and chemical properties as a soil ‘catena’ (Latin for chain). Likewise, boundaries between different vegetation zones can also be sharp or gradual. In considering prehistoric lifestyle changes in the Nile Basin during the very late Pleistocene and early to middle Holocene, we need to remain aware that the transition from one state to another can be erratic, slow and not always in a single direction. The entire question of the origins of food production in the Nile Basin is not an easy one to resolve but there has been considerable progress in the last few years (Brass, 2017;

22.2 Some General Considerations

303

Holdaway and Wendrich, 2017; Phillipps et al., 2017; Wendrich et al., 2017a, b; Winchell et al., 2017; Wright, 2018). It is also a question best considered in the context of a much wider region, including the ‘Fertile Crescent’ lands of the Levant and Anatolia in which wheat and barley were first domesticated towards 11–10 ka, along with sheep, goats and cattle (Gopher et al., 2017; Grosman and Munro, 2017; Kislev and Simchoni, 2017; Özdoğan, 2017; Vigne et al., 2017).

22.2 Some General Considerations Some scholars have argued that simply to ask ‘when’ and ‘where’ questions of archaeological and other environmental evidence will ultimately yield quite trivial answers unless we also ask ‘how’ and ‘why’ (Stahl, 1984; Wenke, 1991). So, for example, merely to determine when the inhabitants of the Nile Basin first began to use domesticated plants and animals may be a necessary start, but of greater import is to understand what processes were involved in such domestication and how people were interacting with their ever-changing environment. In seeking to comprehend why the adoption of domesticated wheat and barley in the Nile Basin was so slow relative to their adoption in the Near East we need to steer a careful passage between the Scylla of environmental determinism and the Charybdis of the equally vacuous extremes of geographical possibilism. Spate (1952) provides an incisive and witty critique of these matters in his discussion of the work of the great historian Arnold Toynbee and the geographer Ellsworth Huntington. To assert that the Nile is simply another Indus or Jordan is to ignore its hydrologic complexity, as Butzer (1976) has shown so well, just as the claim that past (and present) societies can cheerfully ignore the very real constraints imposed by soil and groundwater salinity, sparse vegetation and mobile sands on desert dwellers is to ignore reality (Williams, 2014, chapter 4). In his account of growing up in semiarid Western Australia, Tim Winton (2015, p. 17) expresses this reality with admirable brevity: ‘Geography trumps all’. We therefore need to consider very carefully what might have been the role of environmental changes, including fluctuations in Nile floods and sediment type and amount, in impeding (or facilitating) plant and animal domestication. The concept of ‘geological opportunism’ (Vita-Finzi, 1969, 1978) implies that prehistoric peoples could not use parts of the landscape until they were available to be used. Put simply, if a Neolithic farmer wanted to grow sorghum or wheat and the land was perennially flooded, that land was not available for that purpose, although it could of course be used for fishing. Once the land drained, perhaps as a result of river channel incision, the wetlands would then become available for cultivation (Williams, 2009a). Were social, political and economic factors not also important influences? Wenke (1991) argues very cogently that they were. The timing and possible causes of plant and animal domestication have long been a matter of considerable interest to prehistorians as well as plant and animal ecologists (Clark, 1971; Heiser, 1973; Harlan, 1975; Cohen, 1977; Clark and Brandt, 1984; Clark, 1984a; A. B. Smith, 1992; B. C. Smith, 1995). Was such domestication synchronous or was

304

Origins of Plant and Animal Domestication

one a precursor to or even a prerequisite for the other? Within the Nile Basin as a whole, the adoption of plant and animal domestication was patently far from synchronous. Does this reflect diffusion of ideas, local experiments, movement of peoples, or even, in the case of cattle, sporadic raiding and rustling? In regard to the question ‘why’, W. H. Auden (1907–74) noted very aptly in the conclusion to the last poem he wrote, entitled Archaeology, that ‘Knowledge may have its purposes, but guessing is always more fun than knowing.’ We cannot enter the minds of early Neolithic herders and farmers, let alone those of Palaeolithic hunters and gathers, although ethnographic studies conducted over many years do offer helpful insights. Can linguistics assist? Perhaps, but only with great discretion. In an early survey, Ehret (1984) concluded that linguistic evidence from Africa pointed to an earlier onset of plant and animal domestication in East and West Africa than much of the extant archaeological evidence indicated. He may have been broadly correct, although that is debateable, but the problem here is one of accurate timing. Glottochronology, or the use of linguistic evidence to establish cultural histories and human migrations, often appears more an art than a science. Using insights from linguistics as a chronometer for prehistoric cultural change will always be fraught because we cannot assume constant rates of change nor can we assume that humans always act rationally. There is no good reason to assume uniform rates of language development in prehistoric times: the lessons from studies of historic language evolution suggest otherwise (Bragg, 2003; Crystal, 2004). A possible analogue for language development comes from the earth sciences, in which long intervals of uniform activity were punctuated by short sharp catastrophic events, triggering rapid and often unpredictable change. Of course, any searching discussion of these topics demands a thorough appreciation of the quality of the evidence used to infer that we are indeed dealing with domesticated plants and animals rather than with the wild ancestors of the domesticates. Such evidence may range from bones and teeth in a primary context in rigorously dated and stratigraphically controlled excavations to genetic evidence of both plant and animal domestication. Flotation devices (admittedly cumbersome and thirsty devices) have been in use for half a century but have been employed far too infrequently to obtain seeds, fruits and other plant remains from archaeological sites in the Nile Basin and virtually never in the surrounding deserts. Pottery is a valuable source of information about possible plant domestication because pots and pot fragments may retain impressions of plants and seeds growing in the vicinity of the site when and where the pots were made. They may also contain grains of the domesticated plants, which can then be directly dated by AMS 14C analysis to obtain very precise ages. Similar rigour is needed in determining the age of a stone or bone artefact, potsherd or the fossil remains of a possibly domesticated plant or animal. Excavation in rigid sediment ‘spits’ of fixed horizontal depth and vertical thickness may be convenient and pleasing for the statistician but it can all too often ignore the subtle lessons to be learned from using undulations or topographic irregularities in the former ground surface to control the method of excavation. Whatever the dating technique employed, we still need to be clear about the

22.2 Some General Considerations

305

scope and limitations of that method. How we control for sample contamination should be a matter for explicit evaluation, and not simply relegated to the outer darkness of an appendix buried away in the supplementary information section of a major article. Assessing site disturbance caused by ants, worms and termites is all too often evaded, ignored or dismissed with the cavalier comment that no signs of ‘major’ disturbance were seen. The cumulative effects of low-magnitude but high-frequency geomorphic and bioturbation processes is only now beginning to be more widely acknowledged, however reluctantly, and this despite clarion calls from earlier workers about the reality of such disturbance in archaeological sites across Africa (Cahen and Moeyersons, 1977; Moeyersons, 1978; McBrearty, 1990). Dating the sediments within an excavation is one thing, but can we be certain that the remains of putative domesticated wheat or barley, which have not been directly dated, are indeed coeval with the sediments in which they occur? Past experience suggests otherwise. A salutary lesson is provided by Wendorf et al. (1984b) and especially by the Addendum at the end of that chapter. The domesticated barley grains earlier claimed to be late Pleistocene and Terminal Palaeolithic in age (Wendorf et al., 1979) later proved to be intrusive and well over 10,000 years younger than the sediments in which they were found. No attempt was made to account for this anomaly. There is no persuasive alternative to obtaining direct ages for the archaeological remains, if at all possible, using whatever dating techniques are most suitable. A further issue is the thorny one of determining whether or not the excavations at a particular site are indeed representative of the entire site. Short of excavating the entire deposit, which runs counter to excavation ethics (new techniques in the future will provide insights not now possible, but not if all the material has been dug up), we can never be quite sure. The best strategy is to devise a system of sampling that allows for varying site usage in the past – hence the need for careful excavation and analysis over a number of field seasons, allowing the evidence obtained during one season to guide future work. The efforts of Donatella Usai and Sandro Salvatori at El Khiday, of Lech Krzyzaniak at Kadero, of Derek Welsby, Charles Bonnet and Matthieu Honegger in Nubia and of Simon Holdaway and Willeke Wendrich in the Fayum are all fine examples of this counsel of perfection being put into practice in the Nile Valley (Krzyzaniak, 1991; Bonnet, 1992; Welsby et al., 2002; Salvatori et al., 2011, 2014; Honegger and Williams, 2015; Holdaway and Wendrich, 2017). Use of the term site also begs the question of what is a site. Is it the entire landscape traversed and used by former people? Given the growing evidence that prehistoric peoples ranged across vast areas, can we ever be sure that our excavations and mapping of material deposited across the landscape, be it stone or bone, can ever tell us all we would like to know about past events and processes? Probably not. Nevertheless, we can still learn a great deal from sustained, imaginative and careful interdisciplinary work. Before attempting any form of regional syntheses of plant and animal domestication in the Nile Basin and adjacent areas, it will be useful to consider a few specific case studies that illustrate some of the pitfalls confronting geo-archaeological enquiry into the transition

306

Origins of Plant and Animal Domestication

from an economy based on hunting, gathering and fishing to one based on food production through herding and cultivation of domesticated plants. The terms Mesolithic and Neolithic are convenient and concise labels to describe such cultures or lifestyles and will be used in this sense here. My intention in the following sections is not to offer gratuitous criticism of earlier studies but rather to demonstrate that any work of science is a work in progress, to be confirmed, refuted or otherwise modified as new techniques and insights become available. In the earth sciences, a prime example is the widespread adoption of plate tectonics in the late 1960s as a model with superior predictive power to that of all earlier geological models concerned with the origin of the continents and oceans (Paton, 1986).

22.3 The Transition from Mesolithic to Neolithic in the Fayum and Main Nile Valley We begin with the Fayum (Fig. 22.1) because it has been studied by archaeologists for more than a century and appears to have some of the earliest evidence of domesticated plants and animals in the Nile Basin. In their sustained, innovative and pioneering archaeological and geomorphological investigations in the northern Fayum, Gertrude Caton-Thompson and Elinor Gardner laid the foundations for all subsequent geo-archaeological work in this region (Caton-Thompson and Gardner, 1929, 1932, 1934). Their interpretations and conclusions have not always stood the test of time, but their groundwork encouraged, and in some cases even made possible, the plethora of more recent studies (see Chapter 17). Many of these are still continuing, in spite of widespread, irreversible and accelerating damage to the archaeological remains from burgeoning agricultural development projects in the Fayum (Holdaway and Wendrich, 2017). Despite the claims in the wider literature, there is in fact relatively little difference between the occupation sites adjacent to the north shore of Holocene Lake Moeris that have been claimed to be Epi-Palaeolithic or Mesolithic and those considered to be Neolithic (Holdaway et al., 2017a, b, c; Wendrich et al., 2017a, b). Both communities relied primarily on fishing, most likely on a seasonal basis. Both used grindstones to mill the seeds from wild grasses, although the Neolithic folk also supplemented their use of wild grains with domesticated emmer wheat (Triticum turgidum spp. dicoccon) and hulled six-row barley (Hordeum vulgare spp. vulgare). Both hunted wild animals to be found in the vicinity of the northern Fayum, although here again the Neolithic people supplemented their diet of meat from wild animals with the food and other resources provided by small numbers of domesticated goats, sheep, cattle and pigs, all of which came originally from outside the Nile Valley, and most likely from the Levant or Near East. Almost nothing is known about how the Neolithic people cultivated their wheat and barley, both of which are winter crops that do best under seasonally cool conditions, although they can be grown under irrigation in areas of summer rainfall. The many abandoned small wheat fields in northern Sudan today are a stark reminder that growing

22.3 Mesolithic to Neolithic Transition in Fayum and Main Nile Valley

30°E

307

40°E

MEDITERRANEAN SEA N

Saïs Merimde

30o N

CAIRO El Omari

Fayum

0

250

500

km

Ni le

LIBYA

EGYPT Farafra Oasis

Wadi Kubbaniya

D RE

Bir Kiseiba Bir Tarfawi

Nabta Playa

A SE

CHAD

Nile

20°N

Shaqadud

Sabaloka Gorge

SUDAN

KHARTOUM

Butana

ER

IT

Kassala

R

EA

Bl ue

White

Nile

Nile

ET HI OP IA

Figure 22.1 Mesolithic and Neolithic sites in Egypt and northern Sudan cited in the text.

wheat outside its optimum habitat can be precarious. There is some reasonably persuasive circumstantial evidence in support of the proposition that a zone stretching from the northern Red Sea Hills across the northern Sahara (including the Fayum and Farafra Oasis: Fig. 22.1), came under the influence of Mediterranean winter rains during roughly 6.7–5.8 ka (see Chapters 17 and 18). We do not know whether wheat and barley were grown along gently shelving sections of the mid-Holocene Lake Moeris shoreline as the summer flood levels receded, using the system French scholars describe as décrue (i.e., post-flood), or whether they depended solely on sporadic winter rains, or whether they received floodwaters from the ephemeral stream channels draining the northern uplands. Of course, any

308

Origins of Plant and Animal Domestication

combination of these different water harvesting techniques may have been used, but we simply do not know. The Neolithic groups used storage pits lined with fibre mats or clay, in which they stored grain, pots, baskets, sickles and in one instance a throwing stick or type of boomerang. (Such weapons are still in occasional use for sporadic hunting of small animals among the farmers of the Gezira in central Sudan.) The Fayum pits were sealed with locally manufactured circular covers of clay, sand, salt and shell grit that set hard when dry. Radiocarbon ages obtained on charcoal recovered from hearths indicated a general age range back to at least 9.2 ka for the Mesolithic sites and of roughly 7.0–6.0 for the Neolithic sites (Holdaway et al., 2017c; Wendrich et al., 2017a, b). Whether the hearths dating between 8.5 and 7.5 ka belong in the Mesolithic or Neolithic is probably a moot point. The oldest bone from a domestic animal has a 14C age of ca. 7.0 ka and the oldest domesticated cereal grains have 14C ages of 6.5–6.3 ka (Holdaway et al, 2017c). The alert reader will note that there are slight discrepancies in some of the calibrated 14 C ages cited in the chapters of The Desert Fayum Reinvestigated (Holdaway and Wendrich, 2017) and those in some of the prior publications, reviewed in Chapter 17. These minor discrepancies do not really matter, for the reasons that now follow. A word of caution is required here in regard to radiocarbon ages and their calibration. (For details of the scope and limitations of this versatile dating technique see Williams et al., 1993, pp. 256–261; Williams et al., 1998, pp. 274–277; Walker, 2005, pp. 17–55; Williams, 2014, pp. 86–91.) Converting radiocarbon ages into calendar ages involves using one of the regularly updated calibration software packages. These take into account fluctuations in the production of atmospheric radiocarbon or carbon-14 (14C) as well as the analytical error terms involved in laboratory measurements of the sample. The rate of production of 14C in the atmosphere has not been constant during the last 50,000 years for which we now have potential 14C chronologies. This is because the cosmic ray flux has not been constant within this time interval, and has fluctuated in tandem with fluctuations in the strength of Earth’s magnetic field, which shields us from cosmic radiation. So, for example, a sample with a mid-Holocene 14C age of 4,500 ± 50 years is represented by a number of separate points on the calibration curve, giving a calibrated range of 5,050–5,275 years BP (Walker, 2005, p. 33). Once the confidence limits at two standard deviations are also taken into account, the calibrated age range now becomes 4,950–5,350 years BP. For these reasons, I have chosen in most cases to round off the calibrated ages cited by different authors and to express them as thousands of calibrated years before present, denoted by the abbreviation ka. Painstakingly detailed analysis of the relative proportions of cores and flakes across the landscape of the northern Fayum indicates that cores were both taken out of the area and brought in, as were large stone flakes, indicating persistent if intermittent movement of people into and out of the region (Holdaway et al., 2017c; Wendrich et al., 2017a). Quite clearly, the people who made occasional use of the northern Fayum region were attracted by the fishing potential in the lake and associated basins, but also roamed across a much wider region in what is now the surrounding desert, which on wetter occasions during the early to mid-Holocene would have been desert scrub or even savanna grassland.

22.4 Mesolithic to Neolithic Transition in Eastern Sahara

309

The archaeological record left behind in the northern Fayum is but a subset of a much wider pattern of subsistence for which we at present have no record. No permanent dwelling structures in the form of post-holes, stone slabs or mud-bricks have been found in association with the Mesolithic and Neolithic occupation of the northern Fayum. What we appear to be dealing with in this region is a gentle progression from fishing, hunting and gathering wild foods during the early Holocene to a relatively slow adoption of domesticated plants and animals as a supplement to the traditional fishing and hunting lifestyle. Such a gradual transition is more an evolution than a revolution, and this seems equally the case elsewhere in the lower Nile Valley. In the Fayum, the use of domesticated wheat and barley dates back to 6.5–6.3 ka, while domesticated bones of sheep and/or goats date back to at least 7 ka, and possibly to 7.4–6.8 ka. The arrival of domesticated cattle was somewhat later (Holdaway et al., 2017a, b). The Fayum was abandoned by 6 ka, possibly because the cessation of winter rains at that time made it very difficult to grow wheat and barley. It was not occupied again until several thousand years later, possibly during the Old Kingdom period and certainly during Greco-Roman times (Holdaway et al., 2017a). Sites in the adjacent lower Nile Valley and Delta (Saïs, Merimde Beni Salama and El Omari: Fig. 22.1) have evidence of domesticated pigs, goats and sheep, cattle, wheat and barley and occasional stone and mud-brick house structures, with ages extending back to 6.5–6.3 ka, 6.8–6.4 ka and 6.4–5.7 ka for these three localities, respectively. There is little evidence of progressive vertical accumulation of midden material through time, as in the early to mid-Holocene ‘tells’ of Mesopotamia. Rather, people appear to have spread laterally during repeated visits to particular localities, making it hard for the modern archaeologist to define site boundaries (Holdaway et al., 2017b).

22.4 The Transition from Mesolithic to Neolithic in the Eastern Sahara In Section 22.2 we noted that early claims by Wendorf et al. (1979) and Wendorf and Schild (1980, pp. 273–280) that barley was being cultivated at Wadi Kubbaniya in the Nile Valley (Fig. 22.1) during Upper Palaeolithic times towards 18 ka have since been shown to be wrong (Wendorf et al., 1984b). The barley grains in question were later dated directly by these authors and found to be mid- to late Holocene in age; they had been incorporated into late Pleistocene sediments and so were intrusive (Wendorf et al., 1984b). These claims of an entirely African and independent centre for barley domestication west of the Nile need detain us no longer. We saw in Section 23.3 that all the evidence from Merimde in the upper Nile Delta region and from the Fayum west of the lower Nile indicates a relatively late arrival of wheat and barley derived from original centres of domestication in the ‘Fertile Crescent’ of the Levant. We turn now to an even more contentious issue – that of whether cattle were originally domesticated in North Africa before making their first appearance in the Nile Valley. Once again, Wendorf and Schild and their colleagues (1980; 1984a) were the protagonists for the claim that domestic cattle descended from wild African cattle were present at Bir Kiseiba

310

Origins of Plant and Animal Domestication

and Nabta Playa in the eastern Sahara (Fig. 22.1) several thousand years before they appear to have reached the adjacent Nile Valley. Smith (1984, 1992) has vigorously contested this claim and Wendorf et al. (1987) have equally vigorously defended it. Wendorf and Schild (1980, p. 278) had explained earlier why early cattle domestication in the Sahara is such a thorny issue: ‘It must be recognised that the problem of domestic cattle in the Sahara is one of the most difficult and controversial questions pertaining to the origin of domestication. In large part this is due to the scarcity of Bos remains in archaeological sites and the fragmentary nature of the few that are found, so proper anatomical observations are generally not feasible.’ The central issue here is whether the fossil remains of putative domestic Bos taurus recovered from this area do indeed represent direct descent from the wild African aurochs (Bos primigenius). A related issue is whether some (or all) of the cattle bones from Nabta Playa (Schild and Wendorf, 2001) and Bir Kiseiba are in fact bones of domesticated animals rather than of wild aurochs hunted at that time. Finally, is it not also possible that the Neolithic Bos taurus herds represent an admixture of genes, some derived from cattle initially domesticated in the Near East from Eurasian wild aurochs, which later interbred sporadically with African wild aurochs after being introduced to the northern Nile Valley and adjoining desert regions? To answer these questions, we need ideally to consider evidence from archaeology, palaeoecology, palaeohydrology, geomorphology, genetics, isotope geochemistry and geochronology – evidence which is not always available where and when we most need it. Nabta Playa and Bir Kiseiba are located several hundred kilometres west of the Desert Nile in the now hyperarid eastern Sahara. One of the (far from convincing) arguments put forward in support of cattle domestication at this site is the claim that the climate at that time was too dry to support much wildlife apart from hares and gazelles, so that hunting was not a reliable source of food. Michael Brass has refuted this assertion in his rigorous and thorough reappraisal of the subfossil plant remains recovered from excavated sites in both localities (Brass, 2017). He concluded that much of the vegetation is tolerant of dry conditions and, in the case of deep-rooted plants and trees, would in any case have drawn its moisture from the local aquifers at a time of high early to mid-Holocene regional water tables (see Chapters 15 and 16). The semi-desert scrub and acacia savanna that Brass envisaged would have been more than adequate to provide food and shelter for a wide range of African herbivores and browsers. The scanty fossil remains could reflect bone disintegration, selective choice of animals to be brought on-site and consumption of the larger animals off-site – a pattern common in many hunter-gatherer societies today. Brass (2017) also reviewed some of the more recent evidence from genetics, and concluded that domesticated cattle from an initial centre of domestication in the Middle Euphrates Valley began spreading into the Nile Valley and North Africa by ca. 8.3 ka, reaching Bir Kiseiba and Nabta Playa well after that date at some time in the late Middle Holocene. He also suggested that domesticated caprines (sheep and/or goats) were present in NE Africa before domestic cattle arrived. Finally, he concluded that the early Holocene Bos remains at Nabta Playa and Bir Kiseiba were those of hunted wild aurochs, so that ‘the

22.5 Mesolithic to Neolithic Transition in Central and Eastern Sudan

311

time has come to abandon the long-standing hypothesis of an early Holocene independent centre of cattle domestication in Northeast Africa’. These conclusions notwithstanding, it remains possible, but not fully proven, that there was indeed some interbreeding between Bos taurus and Bos primigenius in North Africa, as Clark et al. (2008c) very tentatively speculated, with two authors of that chapter in favour of this hypothesis and one opposed. Future genetic studies are needed to clarify this matter.

22.5 The Transition from Mesolithic to Neolithic in Central and Eastern Sudan The pioneering excavations by A. J. Arkell (1949b, 1953) at the Mesolithic ‘Early Khartoum’ site and at the Neolithic Esh Shaheinab site (Fig. 22.2) have provided solid foundations for all subsequent work in this region. It is important to read both volumes together, because in his 1953 book Arkell offers new and different interpretations of some of the conclusions he reached in his 1949 monograph. We have reviewed his work in Chapters 8 and 11, but it will be useful to reiterate the main conclusions here. The ‘Early Khartoum’ site is located on a low sand mound adjacent to the Blue Nile north-east of the Khartoum central railway station, and is slightly south of the confluence of the Blue and White Nile. The Neolithic site at Esh Shaheinab is located 48 km north of Omdurman (Fig. 22.2) on a 200-m-long gravel ridge near the present-day left bank of the Nile. One kilometre due west of the Shaheinab Neolithic site there is an eroded ‘Early Khartoum’ Mesolithic site. Arkell (1949b) speculated that the ‘Early Khartoum’ tradition would prove to be about 8,000 years old, a prediction that was later confirmed by 14C ages of 9–8 ka from the Mesolithic sites of Tagra (Fig. 22.2) (Adamson et al., 1974) and Shabona (Fig. 22.2) (Adamson et al., 1982; Clark, 1984b, 1989). The presence of an abundant swamp fauna including the Nile Lechwe (Limnotragus), a water mongoose and extant reed rats (Thryonomys) as well as the extinct reed rat Thryonomys arkelli indicated to Arkell that the vegetation at this time was similar to that growing today in the swamps and dense gallery forest of South Sudan. Other evidence in support of this inference included abundant seeds from the tree Celtis integrifolia, which requires a mean annual rainfall of roughly 500 mm, as well as subfossil shells of the large land snail Limicolaria flammata, which is found today only south of Sennar (Fig. 22.2), which now receives on average 460 mm of rain each year. Tothill (1948) described Limicolaria flammata as a denizen of the acacia-tall grass savanna ecosystem in central Sudan, concluding that it is seldom found north of the 500-mm isohyet. Needless to say, isolated individuals can occur well north of this limit, provided the micro-habitat is suitable (Haynes and Mead, 1987). Arkell (1949b) concluded from the combined fossil evidence that the early Holocene climate at Khartoum during Mesolithic times was characterised by a longer wet season and a mean annual rainfall of at least 500 mm, or about three times more than Khartoum receives today during its short summer rainy season. Blue Nile flood silts found above present maximum flood levels prompted Arkell (1949b, p. 109) to infer that the Blue Nile flood level was at least 4 m higher at that time. Based on his subsequent experience at Esh

312

Origins of Plant and Animal Domestication Esh Shaheinab Zakiab Kadero KHARTOUM

N

El Khiday 50 km

15°N

Shabona

14°N

Tagra

Sennar Jebel Tomat Jebel Moya Rabak 33°E

34°E

Figure 22.2 Mesolithic and Neolithic sites in central Sudan cited in the text.

Shaheinab, where an eroded Mesolithic site west of the main Neolithic site was situated 10 m above modern Nile flood level, he revised his initial +4 m estimate for flood levels at the ‘Early Khartoum’ site. It is worth repeating his reasoning here. He observed that: ‘It was not only possible but likely that it was a low river hunting and fishing camp, which meant that high river must have been at least 5.4 metres, and probably 6 or 7 metres, higher still’ (Arkell, 1953, p. 8). From this it follows that the Blue Nile channel has become incised into its flood plain since that time, which in turn indicates incision by the main Nile, to which the Blue Nile is tributary. The effect of such incision would be to reduce the extent of mid- to late Holocene flooding on the alluvial plains bordering the Nile except during intervals of very high Nile flow. Fish remains were abundant at the ‘Early Khartoum’ site and included catfish (Clares sp.), Synodontis sp. and Nile perch (Lates cf. niloticus). Bones of various antelope species were also common, together with large herbivores and browsers, such as Cape buffalo, giraffe and elephant. The site must therefore have been flanked by trees and grasslands, indicating a savanna woodland habitat able to provide enough food for these animals. The shells of the semi-aquatic snail Ampullaria wernei (i.e., Pila wernei) were very

22.5 Mesolithic to Neolithic Transition in Central and Eastern Sudan

313

common at the site, prompting Arkell to conclude that they were used as bait in fishing. He later considered it more likely that they were boiled in pots and eaten (Arkell, 1953, p. 97–98). Clark (1984, 1989) arrived at a similar conclusion following his 1973 excavation of the Shabona Mesolithic site on the right bank of the lower White Nile (see Chapter 8). These snails can grow up to 5 cm in diameter and are quite easy to collect from shallow ponds and seasonal wetlands. The Early Khartoum Mesolithic folk lived by fishing, hunting and gathering wild plant foods, snails and shell fish (Nile oysters and mussels). They did not have any domesticated animals. Arkell (1953) recovered bones of a small domestic goat and possible sheep from the Neolithic site at Esh Shaheinab, but it was not until the 1972–89 excavations at Kadero (Fig. 22.2) directed by Krzyzaniak (1991) that an abundance of domesticated cattle remains and associated human burials were recovered. Kadero lies 18 km north of Khartoum and 6 km east of the Nile. It is primarily a Neolithic site, with some Mesolithic and intrusive Meroitic remains. This pattern of sporadic sequential occupation at the same locality is also evident at El Khiday (Fig. 22.2) near the left bank of the lower White Nile as well as at other sites in central Sudan (see Chapters 8 and 11). The Early Khartoum tool kit included broad shallow pots displaying the very characteristic dotted wavy line pattern of decoration. Bone spear points and barbed bone harpoon points were common, the latter likely used for securing large Nile perch and, perhaps, hippos. Shell fish hooks were used to obtain smaller fish and nets may have been used to catch catfish in shallow ponds left isolated at the end of the flood season, just as at the Mesolithic site of El Khiday on the left bank of the lower White Nile (Salvatori et al., 2011, 2014; Williams et al., 2015a). Both Arkell (1953) and Clark (1984) considered that the Shaheinab Neolithic site was one occupied seasonally by people who still obtained most of their food from hunting, fishing and gathering. Hunting Technical Services Limited (1964) and Obeid et al. (1982) provide lists of plants used today for food, medicine, skin tanning and building in the northern Gezira area between the lower Blue and White Nile and surrounding region. Even today, under a semi-arid climate, there is still an abundance of economically valuable plants. Both Krzyzaniak (1991) and Haaland (1981) concluded that the site at Kadero was one devoted primarily to cattle rearing. The abundance of scrapers found at the site is entirely consistent with their present-day use in parts of south-central Ethiopia to scrape fat and other tissue from skins and hides (Gallagher, 1976; Haaland, 1981). Leather skins have multiple uses. Goat skins are still used today across Sudan and the Sahara to make porous water bags or girbas, and are not too hard to make, provided the initial skinning is done with care. The inflated bags can also be used as floats to prevent wounded hippos or crocodiles from sinking. Leather mats are still used by stone tool makers in Ethiopia to protect them from cuts and to collect waste flakes. Tents or temporary shelters made from skins and/or mats woven from reeds or other suitable plants are still in use today across Africa. The Aïr Tuareg routinely use leather bags to carry dried food and cereal grains on their camels, as do the Somali women in the Jubba valley. Suitably treated, leather is a durable and versatile material.

314

Origins of Plant and Animal Domestication

Haaland (1981, 1984) conducted some limited excavations at the Neolithic sites of Kadero (6.5–6.4 ka), (Fig. 22.2) (6.8–6.5 ka) and Shaheinab (6.5–6.4 ka). The ages in brackets are calibrated 14C ages. She formulated a series of working hypotheses to guide her excavation strategy, and concluded that the main cattle herding site was under the control of the men, with the women responsible for collecting plant foods and pot-making. Hunting remained important and was no doubt the responsibility of the men, at least in the case of large and dangerous animals. A possible modern parallel comes from the Anbarra people living along the Arnhem Land coastal wetlands in tropical northern Australia, where Betty Meehan (1991) has described in some detail a traditional division of labour between the adult male hunters and the female and juvenile gatherers. Her observations were based on living with the Anbarra over a period of twenty years. Meehan observed that in one particular month (April 1973) the Anbarra obtained their food from sixty different species of plants and animals and that all members of the community, young or old, had a role in obtaining food. Some words of caution are needed here in regard to pottery making. Haaland (1981) was influenced by her observations of Fur women making pots in Darfur Province of NW Sudan. Shriainen (1984) has described two distinct pottery traditions in South Sudan. Dinka pottery is made from coils of clay and decorated with a twisted palm roulette, and the potters were men. On the other hand, the pots made by the Central Sudanic speaking Jur people of South Sudan – the Jur Molo pottery – are fashioned from a single lump of clay and decorated with impressions from a rough fibre mat. He does not specify the sex of the Jur Molo potters, but did consider this to be a much older tradition than the Dinka pottery. If this surmise is correct, it means that older and newer cultural traditions can occur within the same general region and are likely to have done so in the past. Using pottery styles as a dating tool might therefore prove unwise. Krzyzaniak (1991) found that certain of the Neolithic human burial sites at Kadero contained grave goods indicative of wealth and prestige. The cowrie shell necklaces pointed to a trade in cowrie shells from the Red Sea to the east, and the volcanic and metamorphic rocks (porphyry and rhyolite) used to make the polished mace heads and axe heads came from the rocks cropping out along the cliffs of Sabaloka Gorge (Fig. 22.1) at the 6th Cataract. He suggested that the leaders of these pastoral groups were responsible for maintaining trade and other social and economic links across a wider area, and, probably, determined the timing and direction of seasonal migration of the herds, in response to seasonal access to suitable grazing. Both the Mesolithic or ‘Early Khartoum’ cultural tradition and the Shaheinab Neolithic have their counterparts across the Early to Middle Holocene Sahara. French archaeologists refer to the Early Holocene Sahara at this time as ‘le Sahara des Tchads’ or ‘le Sahara des lacs’ in recognition of the abundance of lakes and integrated drainage systems prevalent at that time (Faure, 1969; Rognon and Williams, 1977; Servant and Servant-Vildary, 1980; Vernet, 1995). The evidence for wetter climates across the Sahara is widespread and compelling (Rognon,1989; Gabriel, 1986; Williams 1984; 2008; see also Chapter 4). At the isolated mountain of Adrar Bous, a rugged ring complex in the geographical heart

22.5 Mesolithic to Neolithic Transition in Central and Eastern Sudan

315

of the Sahara, Mesolithic people fished for Nile perch and caught turtles, hippos and crocodiles in a lake along the southern edge of the central granitic massif (Clark et al., 1973; Williams et al., 1987; Williams, 2008). This lake dried out for several thousand years and refilled to a lower level (Williams et al., 1987), attracting herders of short-horned domestic cattle (Clark et al., 2008c) who could graze their cattle on the surrounding grassy plains and water their animals at the lake and its tributary streams. The complete skeleton of one such animal (Bos brachyceros) was found embedded in the sediments of a former swamp, the clay from which had later set like concrete. It had an AMS calibrated 14C age on tooth enamel of 4.8 ka (Clark and Carter, 2008). Whether or not the progenitors of these cattle originated in Africa, in the Near East or were a hybrid mixture of both African and Asian species, as some of the genetic evidence suggests, remains inconclusive (Clark et al., 2008c). The abundance and widespread nature of the Mesolithic lake side dwellers across the southern half of the Sahara and further south prompted John Sutton to describe their lifestyle as aquatic (Sutton, 1974). He referred to their tool kit as aqualithic (Sutton, 1977). Both terms are of course an exaggeration, but Sutton’s point about intensive use of aquatic resources at this time remains valid. It also seems probable that the riparian Mesolithic occupants of the Nile Valley in Sudan and southern Egypt were in contact with their counterparts across the southern Sahara. Desiccation of the Sahara proceeded in stages, with sporadic drying out of lakes and rivers. The final phase of desertification began about 4,500 years ago and would have prompted an exodus of Neolithic pastoralists to the west along the Niger, to the south via the Tilemsi and other valleys in Mali (Smith, 1980), and to the east across the Libyan Desert and Western Desert of Egypt into the Nile valley, which still offered reliable water and aquatic food resources. The strictly aquatic resources would have been supplemented by wild and probably also domesticated cereal grasses such as sorghum and other annual plants growing to maturity after the retreat of the summer floods from all but the back swamps situated along the outermost margins of the flood plains. The back swamps would have likewise been a source of fish, Pila snails and edible reeds such as Typha. The apparent absence of any occupation between the youngest of the Mesolithic sites and the earliest of the Neolithic sites in the Khartoum area prompted Haaland (1984) to question the reason for this putative gap of 2,000 years. Rejecting a climatic cause, she proposed that a change to a more nomadic lifestyle would have been harder to detect given the few durable material possessions used by pastoral nomads. It is also possible that the locus of visible settlement may have shifted away from the Khartoum area to the savanna grasslands of the western Butana (Fig. 22.1), as the evidence from a number of sites including the 7–4 ka site of Shaqadud 75 km NE of Khartoum (Fig. 22.1) and 20 km SW of the Nile suggests (Mohammed-Ali and Marks, 1984). Haaland (1984) and colleagues excavated a site near Rabak (Fig. 22.2) which yielded calibrated 14C ages of 9.5–9.0 ka near the base, with wavyline pottery akin to that at the Mesolithic Early Khartoum site, and 6.9–5.3 ka in the Neolithic horizon.

316

Origins of Plant and Animal Domestication

Clark (1984) reviewed his 1973 reconnaissance excavations at Jebel Tomat (Fig. 22.2) (4.5–2 ka) and Jebel Moya (Fig. 22.2) (see Chapter 8). He concluded that there was progressive cultural evolution in this region from late Neolithic to metal using and Meroitic, with no obvious signs of any significant discontinuity or influx of ethnically different people from outside the area. New excavations at Jebel Moya by Dr Michael Brass and colleagues from Sudan and UCL, which began in October 2017, will no doubt help shed light on these and related issues. A further important question concerns the onset of sorghum and millet domestication in central and eastern Sudan. Both sorghum and millet are staple cereal grains throughout this region, and are used to make porridge and local beer (Merissa). Sorghum is a tropical panicoid grass with a very tough stem. It is grown today on heavy clays soils across much of central Sudan using a system of earth bunds to conserve rainwater within the fields under cultivation. The seeds are sown in early summer (May–June) at the onset of the wet season and the ripe heads of sorghum are harvested about 3 months later. Millet is grown on lighter soils and thrives better on sandy soils than sorghum does. Ann Stemler (1980, 1984) has reviewed the conditions necessary for domesticating wild cereal grasses such as wheat, barley, millet and sorghum. She noted that in wild grasses, the seeds are dispersed throughout the growing season to maximise the survival chances of the wild grass. In some mutant plants the genetic message contained in the abscission zone at the base of the inflorescence or grain-bearing portion of the plant is defective, and the seeds remain on the plant until the end of the growing season. Such plants, if harvested, stored and sown the next growing season, will produce similar plants, with the grains remaining on the inflorescence until all the grains are ripe, which is what a farmer requires, but runs contrary to the survival strategies of wild grasses. Stemler (1980) pointed out that the most efficient way to harvest these mutant plants is with a sickle. Sickles were used to harvest wild grasses during the Mesolithic, when small tools such as backed blades were in common use. Attached to a handle of wood, bone or antler with resin or gum, they make efficient sickles. The winnowed grains could be stored in rodent-proof pots or, less effectively, in pits lined with fibre mats, as at Jebel Tomat (Fig. 22.2), where Clark and Stemler (1975) reported the presence of early domesticated sorghum (Sorghum bicolor), the carbonised grains of which had 14C ages of ca. 2 ka (80 BC to AD 350). Recent analysis of plant impressions on pots excavated near Kassala (Fig. 22.2) in the Butana region of eastern Sudan have demonstrated the presence of Sorghum bicolor (L.) Moench associated with Neolithic remains dated to 5.6–5.1 ka (Winchell et al., 2017). These authors endorsed the now widely held view that Sorghum bicolor originated from wild stands of sorghum (Sorghum arundaceum [Desv.] Stapf) found across Africa. Winchell et al. (2017, Fig. 1) provide a map showing twenty sites in central, eastern and northern Sudan in which they distinguish between wild, cultivated and domesticated sorghum. Sites with wild sorghum date between ca. 8 ka and 2 ka, those with cultivated sorghum are bracketed between ca. 5.5 ka and 2 ka and those with fully domesticated sorghum have ages between ca. 3 ka and 1.65 ka. It is not clear to me why the sorghum impressions from the site near Kassala are shown as cultivated rather than domesticated.

22.6 Mesolithic to Neolithic Transition in Ethiopia and East Africa

317

In any event, it now appears proven that fully domesticated sorghum was being cultivated in eastern Sudan by 5.6–5.1 ka, or nearly a thousand years after domesticated wheat and barley were grown in the Fayum (6.65–6.35 ka) and cattle, sheep and goats were present at Kadero (6.5–6.4 ka). One possible explanation for the relatively late appearance of domesticated sorghum at Jebel Tomat is suggested by the reticulate network of shallow drainage channels apparent on satellite imagery and air photos in the area between the two small granite hills at Jebel Tomat situated 11 km east of the White Nile (see Chapter 8, Fig. 8.4). The drainage pattern is consistent with recent draining of extensive swamps and wetlands that were present on the White Nile flood plain until about 2,000 years ago (Williams, 2009). Such drainage was triggered by incision of the White Nile.

22.6 The Transition from Mesolithic to Neolithic in Ethiopia and East Africa The earliest evidence of domestic animals so far recovered from Lake Besaka in the Ethiopian Rift (Fig. 22.3) and Laga Oda (Fig. 22.3) (Clark and Williams, 1977; Brandt, 1984) at the foot of the escarpment flanking the southern Afar Rift in SE Ethiopia date back to about 3,500 years ago (Brandt, 1984). Somewhat earlier evidence comes from northern Lake Turkana (Fig. 22.3) (Barthelme, 1984; Robbins, 1984), and a few sites in NE Ethiopia have possible ages of 7,000–6,000 years BP, although the chronology from all these sites relies on too few reliable 14C ages and sometimes indirect inferences as to age. Many factors were doubtless responsible for the delayed appearance of domestic sheep, goats and cattle in Ethiopia and East Africa, not least was the prior use in the Ethiopian uplands of locally domesticated plants such as ensete (Ensete ventricosum) and teff (Eragrostis tef). These reliable and highly nutritious sources of food could supply the needs of local farmers even in times of erratic seasonal precipitation, just as they do today (Brandt, 1984). Another important factor relevant to the slow southward movement of cattle into Ethiopia and East Africa was the obstacle posed by the presence of tsetse flies in the wetter regions of the southern Nile Basin during the early Holocene. The many species of tsetse fly belong to the genus Glossina, and feed on the blood of vertebrate animals, in the process transmitting sleeping sickness (African trypanosomiasis) to animals and humans. Their preferred habitats are dense equatorial rain forest and dense gallery forest between about latitudes 15°N and 20°S. As the Holocene climates in this region became less humid, the forest and woodland habitats of these flies shrank and sometimes moved further south into more equatorial latitudes. (The tsetse fly is not to be confused with the Tabanidae flies or horse flies, whose bites madden humans and their animals during the wet season in northern Sudan.) We saw earlier that drying out of the Saharan lakes and rivers during the second half of the Holocene and especially after about 4.5 ka, caused the cattle herders of the Sahara to move south or into the Nile valley in search of water and suitable grazing. In his still very useful compendium on The Grass Cover of Africa, Rattray (1960, p. 9) used the term

318

Origins of Plant and Animal Domestication 36°E

40°E

44°E

16°N

N ER

ITR

EA 250 km

I

AFAR DJ IB O UT

12°N

Blu

eN

ile R

.

ADDIS ABABA

ETHIOPIA

EY LL A V

Lake Besaka

ET HIO P

IAN

RI FT

8°N

Laga Oda

4°N

Lake Turkana

Figure 22.3 Neolithic sites in Ethiopia cited in the text. (Pale grey shading areas above 1,000 m; grey, above 2,000 m; dark grey, above 3,000 m)

‘grassland’ to refer to ‘open treeless country which has developed either as a direct expression of the climate or as a result of a certain set of soil conditions, both of which are unfavourable to the growth of trees’. He also described a third category of grassland as ‘secondary’ or ‘derived’ grassland resulting from biotic influences, including fire. However, it would be misleading to assume that all grasslands were equally palatable: what might be attractive grazing for cattle could prove problematic for sheep, even disregarding the oft forgotten fact that different breeds of sheep and cattle have different dietary preferences. And so, a further factor leading to a delay in the expansion of domestic grazing animals into the southern Nile Basin and beyond is the time required for the different breeds to adapt to their new habitats. Consider, for example, the Baggara cattle in western Sudan. As Rattray (1960, p. 59) commented: ‘Baggara cattle are poor milkers (but their owners generally have enough cows to discount this). They mature in about 5

22.6 Mesolithic to Neolithic Transition in Ethiopia and East Africa

319

years and are able to withstand long daily marches between water and grazing and are used as pack animals for their owners’ camping equipment.’ Neolithic rock paintings in the Tassili show cattle carrying women, small children and some form of shelter (Museen der Stadt Köln, 1978), so that using oxen as transport animals is not new. The tribal name Baggara comes from the Arabic word for cow (ba’qara, pl. ba’qar). A further factor is the innate conservatism of pastoral nomads, and, indeed, peasant farmers worldwide. To this day, the inhabitants of the vast plains between the Blue and White Nile Rivers bring their herds of cattle down to the rivers during the winter dry season, in much the same seasonal routine as their Neolithic forebears. The abundance of water now available from irrigation canals in this land between two Niles means that there is no vital need to practise this type of transhumance today. When I asked them why they continued this practice they replied simply that this was what they had always done. It is an over-simplification to assume that once the central and southern Sahara had become arid then life was no longer possible there. Quite apart from the refuges offered by the high mountains, which to this day provide water and grazing in sheltered valleys for the Tuareg of the Hoggar and Aïr Mountains, the Tibu or Goran of Tibesti and the Fur of Jebel Marra, there is the phenomenon of sporadic cloudbursts and the resultant sudden appearance of grazing know in southern Libya and northern Sudan as gizzu grazing. Rattray (1960, p. 50) comments as follows: ‘In the extreme west of the Sudan there is a rather peculiar area of “gizzu” grazing. This makes its appearance if rain has fallen, with the onset of intensely cold nights. Its value to camels and sheep is great, as herds and flocks will journey 300 to 400 miles (500 to 650 kilometres) to take advantage of it. In a year without “gizzu” the animals are in much poorer condition than in a normal year when rain showers are sufficient. There is no water in the “gizzu” area and animals live entirely on the moisture derived from the grazing.’ Rattray lists eight of the more important ‘gizzu’ plants, including Indigofera bracteolata, Neurada procumbens, Indigofera arenaria, Aristida papposa and Crotelaria thebaica, noting that Neurada procumbens is the main provider of water for the animals. There are many hundreds of grass species in Africa, many more still to be fully described and often up to 150 species of grass growing in an area of only a few square kilometres (Rattray, 1960, p.10). It is also worth bearing in mind that ‘the savanna country provides the most extensive grazing in the continent’ (Rattray, 1960, p. 19). The savanna regions are the habitats of many large and small wild herbivores and browsers, and are home today to small groups of hunter-gatherers. The diet of these people is often far more varied than that of peasant farmers and pastoralists, and such a broadly based economic lifestyle would have proven a strong incentive to prehistoric hunter-gatherers to modify their way of life with considerable circumspection. In this regard, Neolithic pastoral groups embraced a wide spectrum of adaptive strategies (Smith, 1992; Wright, 2017) ranging from continued foraging with only limited use of domesticated stock, to mainly relying upon cattle, sheep and/or goats supplemented by fishing in lake shore and riparian sites and/or hunting in places where wild animals were still abundant.

320

Origins of Plant and Animal Domestication

Given the many lack of incentives to change from a successful hunter-fisher-gatherer Mesolithic lifestyle to one based on sedentary agriculture or more or less nomadic pastoralism, it comes as no surprise to find that animal domestication came late to Ethiopia and Kenya.

22.8 Conclusion Four broad themes pervade models put forward that attempt to explain how and why Mesolithic hunter-fisher-gatherer societies morphed into Neolithic herders of sheep, goats and cattle and cultivators of cereal crops and vegetables within the Nile Basin. These themes are (a) movement of farmers and herders from the Levant or North Africa into and up the Nile Valley; (b) diffusion of ideas about plant and animal domestication through trade links and other well-established social and economic networks; (c) local experimentation and domestication of indigenous plants and animals; and (d) the possible reasons for time lags in the transition from Mesolithic to Neolithic in the Nile Basin. Many of the existing models draw their support from linguistics and from appeals to cultural history. However, as Brandt (1984) pointed out more than 30 years ago in his thoughtful review of the origins of food production in Ethiopia, the paucity of supporting archaeological evidence means that these models remain, at best, working hypotheses to be tested by future archaeological research and, at worst, essentially untestable. The situation has

Table 22.1 Some key dates relating to plant and animal domestication in the Nile Basin Location

Evidence

Fayum

Sheep/goats Cattle wheat, barley Lower Nile Sheep/goats, cattle, wheat, barley Red Sea Hills Sheep/goats Western Desert of Sheep/goats Egypt Cattle Barley Northern Sudan Sheep/goats Cattle Central and East Sudan Cattle, sheep/goats Sorghum Ethiopia Cattle, sheep/goats Lake Turkana, North Cattle Kenya Based on sources cited in the text.

Age (ka)

Chapter

>7.0–6.8 6.8–5.7

17

7.1–7.0 ~7.9 ~7.6