Introduction

1The Nile is the longest river on Earth (6,853 km) and has the third largest catchment area (3,310,000 km²) after the Amazon and the Orinoco. It spans 35 degrees of latitude and flows from the equatorial uplands of Rwanda, Burundi and Uganda (the headwaters of the White Nile) and the volcanic highlands of Ethiopia (the headwaters of the Blue Nile and Atbara rivers) across the seasonally-wet tropical lowlands of South Sudan and Sudan. It then flows north through the arid eastern Sahara and Western Desert of Egypt to reach the Mediterranean Sea after a journey through waterless desert of 2,689 km downstream of the Nile-Atbara confluence. The Nile basin (figure 1.1) consists of four subsidiary basins (figure 1.2), namely, the White Nile basin (1,730,000 km²), the Blue Nile basin (330,000 km²), the Atbara basin (180,000 km²), and the Desert Nile basin (1,070,000 km²).

2Before their flow was regulated by large dams and water and sediment was impounded in reservoirs upstream of the dams, the White Nile, Blue Nile and Atbara rivers provided 10%, 68% and 22% of Nile peak monthly flow and 83%, 17% and 0% of Nile low monthly flow, respectively (Williams et al. 1982; Woodward et al. 2007). Their respective contributions to total Nile sediment load were 3%, 61% and 36%. The unregulated Blue Nile and Atbara thus provided most of the sediment and flood discharge, but the White Nile provided most of the water during the time of minimum annual flow (figure 1.3). When flow in the White Nile was cut off, as it was during the Last Glacial Maximum (LGM: 24‑18 ka) (Mix et al. 2001) the main Nile would likely have ceased flowing throughout the year (Williams et al. 2006). In terms of their modern contribution to total Nile discharge (88 ± 5 km³) and total sediment load (230 ± 20 million tonnes [Mt]), the White Nile furnishes a mere 3 ± 2% or 7 Mt, despite its huge catchment area, amounting to 52% of the total Nile basin. The Blue Nile basin occupies just 10% of the total Nile basin but provides 48 ± 10 km³ or 54% of the annual Nile discharge and 140 ± 20 Mt or 61% of the annual Nile sediment load. The Atbara River, which joins the main Nile or Desert Nile 320 km downstream of the Blue and White Nile confluence at Khartoum, covers a mere 5.4% of the Nile basin but provides 12 ± 5 km³ (14%) of mean annual Nile discharge and 82 ± 10 Mt (36%) of the Nile’s mean annual sediment load (Garzanti et al. 2006; Garzanti et al. 2015; Woodward et al. 2015).

figure 1.1

figure 1.2

figure 1.3

3The aim of this contribution is to show how the flood discharge and sediment load of the Nile and its major tributaries have varied during the course of the Late Pleistocene in response to fluctuations in climate, plant cover, erosion and sediment deposition within the Nile Basin. A further aim is to show how river response to climate change has been complex, with environmental fluctuations within the catchment headwaters having a profound influence upon the alluvial history of the Desert or Main Nile and upon the changing pattern of sedimentation across the Nile delta and Nile cone. The account that follows is highly condensed. A detailed account is given elsewhere (Williams 2019).

Global climatic cycles

4The tempo and periodicity of late Quaternary environmental fluctuations in the Nile basin between 75,000 and 15,000 years ago (75‑15 ka) are best understood by considering some of the key causes of global climatic oscillations. The Quaternary Period (2.6 Ma to present) is characterised by a series of glacial-interglacial climatic cycles (Williams et al. 1998; Williams 2014). During the last 0.7 Ma, the length of each glacial-interglacial cycle was governed by the changes in the elliptical path of the earth around the sun over a time span of 96,600 years reflected in a 3.5% variation in solar radiation received in the outer atmosphere. This cycle is termed the orbital eccentricity cycle. Two additional cycles also had a strong influence upon global climatic fluctuations at time scales of 10⁴ years. The tilt of the earth’s axis (presently 23° 27´) has varied between a maximum tilt of 24° 36´ and a minimum tilt of 21° 59´, with times of mildest winters and summers in high latitudes during times of minimum tilt, and times of hottest summers and coldest winters during times of maximum tilt. This cycle (the obliquity cycle) lasts 41,000 years and controls climatic seasonality. The third cycle (the precessional cycle) has varied between 16.3 ka and 25.8 ka over the last million years and reflects the changing season of the year when the Earth is closest to the sun and is controlled by the direction in which the Earth’s spin axis points in space.

5Around the world, the interval from 75 ka to 15 ka was one of increasing but fluctuating global cooling. The cooling was both a cause and a consequence of the gradual accumulation of vast ice caps in North America and Eurasia as well as the expansion of the Antarctic ice cap and of glaciers in South America and Africa. Global sea levels fell as the ice caps expanded, culminating in the time of lowest sea level and maximum global ice volume known as the Last Glacial Maximum (LGM: 24‑18 ka) (Mix et al. 2001). The LGM sea level was about 120 m lower than today (Chappell 1974; Chappell et al. 1996; Lambeck & Chappell 2001), resulting in coastal rivers (the lower Nile included) cutting down into their former floodplains. In addition, vast areas of the continental shelf became exposed, leading to a general expansion of land mass area. The cooling and sea level lowering was not a continuous process but fluctuated in tandem with both the 41 ka obliquity cycles and the c. 20 ka precessional cycles during this interval of time. As a result, the pattern of late Quaternary sea level fluctuations was saw-toothed (Chappell & Shackleton 1986; Omta et al. 2015), with gradual lowering as the ice caps expanded and relatively rapid rise of sea level as the ice melted. Times of relatively high sea level during the last interglacial-glacial cycle are c. 125, 85, 60, 45, 28 and post-15 ka, with relatively low sea levels at c. 75, 55, 35 and 20 ka. As a first approximation, we might expect the high sea level phases to be relatively warm and wet and the low sea level phases to be relatively cold and dry. The timing of these events indicates some form of precessional climatic influence.

6A series of shorter climatic cycles with a duration of 10³ years were superimposed upon the longer astronomical cycles we have just described. Analysis of the oxygen isotopic composition of Greenland ice cores has revealed a series of relatively brief but intense climatic cycles with a periodicity of very roughly 2,500 years known as Dansgaard-Oeschger (D-O) cycles (Williams et al. 1998; Williams 2014). Each D-O cycle consists of a cold stadial event and a warm interstadial event. During the last 40,000 years the duration of individual interstadial events has ranged between c. 500 years to c. 2,000 years. Each interstadial began quite abruptly, with temperature increases of up to 7°C within a few decades. Cooling thereafter was slower, amounting to a gradual drop of 12‑13°C below present before the next rapid warming. A note of caution is needed here. Climatic signals tend to be amplified in high latitudes and attenuated in low latitudes, so that a more modest response is to be expected in the Nile headwaters to any global climate forcing revealed in the Greenland ice core records. Nevertheless, the now well-dated evidence collected from Africa over the past several decades shows very clearly that during times of peak glacial cooling, the climate in the intertropical zone was distinctly more arid, with lake levels falling, forests retreating, rivers becoming more seasonal, and previously vegetated and stable dunes becoming reactivated. Conversely, during the warmer wetter climatic phases, forests expanded, dunes became vegetated and stable once more, rivers became less seasonal and lake levels rose once more (Gasse 2000; Hoelzmann et al. 2004; Gasse et al. 2008; Williams 2014).

7One additional factor with a possible influence on global and regional climate deserves mention here, namely, the 74 ka super-eruption of Toba volcano in Sumatra, now precisely dated to 73.88 ± 0.32 ka (Storey et al. 2012). This eruption was one to two orders of magnitude greater than the historic eruptions of Agung, Tambora, Krakatoa and Pinatubo volcanoes, all of which had a discernible influence on global climate (Williams et al. 2009). Stable carbon and oxygen isotopic analyses supported by fine-resolution uranium series age obtained on a speleothem from a cave in New Mexico in SW USA revealed that a prolonged drought lasting 1559 ± 431 years was synchronous with Greenland ice core stadial 20 (GS-20), the coldest Greenland stadial on record, the inception of which coincided with the Toba eruption (Polyak et al. 2017). It is therefore possible, perhaps even probable, that this cooling and desiccation episode also affected the Nile basin at this time (c. 74‑72.5 ka).

8In the light of this general environmental background, we now turn to review the specific evidence from the Nile basin for the 75‑15 ka time period, beginning in the headwaters because they are the primary sources of water and sediment. In doing so, we need always to bear in mind that except for certain lake cores and deep-sea sediment cores, most of the proxy evidence for environmental change in the Nile basin suffers from only having a very low-resolution chronology. As a consequence, short-term fluctuations in Nile flow that are clearly evident in the historic and Holocene records become hard to discern when we examine the late Pleistocene records. For example, Holocene high flood levels in the lower White and Blue Nile valleys are consistent with a periodicity of about 1.5 ka (Williams 2009). Annual fluctuations in Nile flow during the past thousand years have revealed cycles of 75 and 35‑45 years (Fraedrich et al. 1997; Williams & Nottage 2006). The latter periodicity is also evident in the Holocene Maryut lagoon on the Nile delta (Flaux et al. 2011). Few of these cycles are evident in the Pleistocene record simply because the age control is at present inadequate to reveal fluctuations in Nile flow at time-scales of 10¹ and 10² years, although such fluctuations must have occurred. Finally, fluctuations in Indian Ocean dipole phases as well as El Niño-Southern Oscillation (ENSO) events, which have a roughly 5 to 8-year periodicity, are apparent in the Holocene Nile Delta sediment records (Marriner et al. 2012) as well as in historic Nile flow records (Whetton et al. 1994; Whetton et al. 1996; Ortlieb 2004) but would be imperceptible in most Pleistocene records, despite exerting a major influence on the occurrence of floods and droughts in the Nile basin (Adamson et al. 1987; Whetton et al. 1990; Williams 2014; Macklin et al. 2015; Fan et al. 2017).

The Blue Nile and Atbara headwaters

9The Ethiopian headwaters of the Nile (figure 1.4) have been contributing sediment to the Nile Cone in the eastern Mediterranean for at least 30 million years (Fielding et al. 2016; Fielding et al. 2018). Episodic uplift during that time triggered intervals of intense fluvial incision by the progenitors of the modern Atbara and Blue Nile, resulting in a highly dissected landscape consisting of a rolling plateau surface capped by high peaks and flanked by near-vertical escarpments, some over a thousand metres high (Williams 2016). The highest peak in the Semien Highlands is Ras Dashan (4,543 m) and shows clear evidence of recent glaciation (glacial cirques, lateral and terminal moraines, glacially-striated rock pavements), as do a dozen of the other mountain peaks (Hurni 1982). We have recently obtained seven ³⁶Cl exposure ages for glacial moraines flanking two of the peaks (Mount Mesarerya and Mount Bwahit) (figure 1.5) in the Semien Highlands adjacent to the Tekezze river valley, a major tributary of the Atbara river (Williams et al. 2015a). Moraines associated with the former cirque glacier on the north face of Mount Bwahit (13°15.6´N, 38°11.5´E; elevation 4,430 m) have ³⁶Cl ages of 68.3 ± 3.5 ka, 37.7 ± 1.4 ka and 15.3 ± 0.7 ka. The cirque moraines on the northwest face of Mount Mesarerya (13°12.8´N, 38°12.8´E; elevation 4,353 m) (figure 1.5) 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. The moraine ages therefore range from 70 to 15 ka. The ages may indicate one or more periods of glaciation between 70 ka and 15 ka, but another interpretation is possible. Some of the older ages may reflect inheritance from older erosional surfaces owing to shallow glacial erosion within the small cirques. Ice was certainly present on Mount Mesarerya 18.1 ka ago and on Mount Bwahit 15.3 ka ago. These latter ages should therefore be considered as maximum ages for the onset of ice retreat from the Semien Highlands.

10Periglacial landforms are more extensive in the Semien Highlands than the strictly glacial landforms. Widespread periglacial solifluction mantles developed in the uplands in response to frost shattering and downslope movement of hillslope mantles under the influence of freeze-thaw processes and gravity. Periglacial processes were active down to 3,000 m, slopes were unstable (figure 1.6), and the upper tree line was roughly 800‑1200 m lower during the LGM (24‑18 ka) than it is today (Williams et al. 1977). Temperatures in the headwaters of the Atbara and Blue Nile were 4‑8°C lower than today (Williams et al. 1977). At that time the Blue Nile was a highly seasonal river with a bed-load of sand and gravel (table 1.1A), much of which was deposited by the main Nile in northern Sudan and southern Egypt. The ecotone between savanna and desert extended 300‑500 km further south, causing reactivation of formerly fixed dunes, and the Atbara probably ceased to flow for much of the year (Adamson et al. 1980; Williams & Adamson 1980).

figure 1.4

figure 1.5

figure 1.6

11The LGM in both Ethiopia and Uganda was not only colder than today but was also considerably drier. Lake Tana on the Blue Nile dried out (Lamb et al. 2007; Marshall et al. 2011; Costa et al. 2014) as did Lakes Albert and Victoria in Uganda (Livingstone 1980; Johnson et al. 1996), curtailing flow into the White Nile (Williams et al. 2006), although the Blue Nile was still sustained by its major tributaries. As global climate became warmer towards 17 ka, mountain glaciers in both hemispheres began to retreat (Schaefer et al. 2006) and Lake Tana began to refill. By 15.0‑14.5 ka the summer monsoon had intensified in the upper Nile area (Williams et al. 2006). Lake Tana overflowed once more (Lamb et al. 2007; Marshall et al. 2011; Costa et al. 2014), and the hitherto closed lake basins in the Ugandan headwaters of the White Nile also filled and overflowed northwards, creating a shallow seasonal lake up to 25 km wide and over 400 km long south of the Blue and White Nile confluence of that time (Williams 2009). The Blue Nile now became a less seasonal mixed load river transporting a suspension load of silt and clay (table 1.1B) and a bed load of sand (Williams et al. 2015a).

The White Nile headwaters and lower White Nile valley

12Sediment cores collected close to the centre of Lake Albert show that it was dry during the late Pleistocene around 21.15‑18.25 ka and 16.7‑15 ka, with soils developing on the dry bed of the lake (Williams et al. 2006). There was no overflow of water into the White Nile during these times. There was a brief interval at 27 ka (27.2 ± 3.8 ka) when the White Nile experienced very high floods and was able to transport a considerable volume of medium sand at least as far as Ed Dueim (Williams et al. 2010), where very large cross-beds over 40 m wide are exposed in a sand quarry (14°00´ 24.3´´N, 32°16´13.1´´E, elevation 379 m). Alluvial fans along the flanks of a granite inselberg (12°35´N, 32°50´E, elevation 394 m) near Jebelein just east of the White Nile were also active at this time (27.5 ± 2.7 ka).

13The transport of a sandy bedload in the lower White Nile valley implies that the Sudd swamps cannot have been in existence at that time. Today these swamps operate as a gigantic filter, and only allow the passage of very fine sediment at their distal northern points of outflow. A likely explanation is that an interval of reduced precipitation over Uganda and South Sudan led to desiccation of the swamps. Once rainfall increased again, towards 27 ka, there was a brief return to active flow in the White Nile before the Sudd swamps had time to become re-established (Williams et al. 2010, 2015a). The late Pleistocene White Nile has therefore experienced three distinct hydrologic modes (table 1.2). At one extreme, when the Ugandan lakes were dry, there was no overflow into the White Nile, which was reduced to a trickle and blocked by wind-blown dunes. At the other extreme, during warmer wetter climatic intervals, the Ugandan lakes overflowed into the swamps of South Sudan and the White Nile carried a meagre suspension load of fine silt and clay. On occasion between these two extremes the White Nile experienced brief intervals of high energy flow and sand transport unimpeded by the swamps that had dried out but fed from overflow from the Ugandan lakes.

14When they were in existence during wetter climatic intervals the upland swamps of Uganda and the lowland swamps of South Sudan helped to damp down the flood peaks of the White Nile and sustain its perennial flow. The ratio of discharge in the wettest month to that in the driest month was 40:1 in the unregulated Blue Nile but only 5:2 in the White Nile, so that the ratio was reduced to 16:1 in the main Nile (Williams et al. 1982; Woodward et al. 2007). The White Nile provides 83% of the low season flow in the Nile. Without this moderating influence the Nile would probably cease to flow during the low water season, and without the swamps the White Nile would become a highly seasonal river, as occurred at 27 ka, when the Sudd seems to have dried out (Williams et al. 2010, 2015a). The Sudd swamps are important for other reasons also. When the White Nile first enters these swamps, its discharge is roughly 23 km³. On emerging its discharge has been reduced by seepage and evapotranspiration to c. 10 km³. The Bahr el Ghazal river joins the White Nile north of the Sudd and brings 3 km³ to that river. Further north the Sobat river flows from Ethiopia and brings 10 km³ of water to the White Nile (Hurst & Phillips 1931; Hurst 1952; Williams et al. 1982; Shahin 1985), bringing its total discharge back to c. 23 km³, although by then the chemical composition of the river has greatly altered (Bishai 1962; Talling 1957; Talling 1976; Williams & Adamson 1973). The Sudd swamps therefore act as a gigantic biogeochemical filter. The water chemistry of the Nile will have varied over time depending upon whether or not the swamps were in existence.

15Lake Challa (3.3°S; 37.7°E) is a small crater lake situated but outside the Nile catchment on the lower eastern slope of Mt. Kilimanjaro and provides a more complete record of hydro-climatic events close to the upper White Nile catchment (Verschuren et al. 2009; Moernaut et al. 2010). The chronology of late Quaternary lake fluctuations is based on 164 AMS ¹⁴C ages for bulk organic carbon samples, corrected for lake-carbon reservoir ages that vary with depth from c. 200 to c. 450 years. The dated sediment cores extend back to 20.5 ka and indicate that the lake was low during 20.5‑14.5 ka although the climate may have been somewhat wetter for a long interval before then. Whether Lake Challa’s inferred hydro-climatic record can be used as a proxy for the late Pleistocene White Nile remains unclear. Certainly, the Holocene flood record in the White Nile accords with the inferred Holocene high levels in Lake Challa (Williams et al. 2015b). However, a contrary view (Tryon et al. 2016) but one based on a very incomplete chronology, proposes that the climate around Lake Victoria was drier than present between about 100 ka and 35 ka, when faunal evidence and stable carbon isotopic analyses point to extensive C₄ grasslands growing in and around the lake basin during that time. The question as to the nature of the late Pleistocene climate in the upper White Nile basin therefore remains unresolved.

16A further factor adding complexity to the late Pleistocene White Nile depositional record is the input of volcanic sediments during times of overflow from Lake Turkana, which is fed by water from the Omo river in Ethiopia. The Omo headwaters lie close to those of the Blue Nile. The addition of the Omo and Lake Turkana basins would have increased the catchment area of the Nile Basin by 146,000 km². There was a well-dated episode of overflow from Lake Turkana into the White Nile via the Pibor and Sobat rivers at 102 ka (McDougall et al. 2008) and again in the early Holocene (Butzer 1971; Harvey & Grove 1982; Johnson 1996; Johnson & Malala 2009) but there may have been other as yet undiscerned overflow events between 75 and 15 ka, which would have led to enhanced Nile floods.

17The two wettest intervals in the late Pleistocene were the last interglacial and the terminal Pleistocene. During the last interglacial the lower White Nile valley was occupied until 110 ka by a seasonal lake 650 km long and up to 80 km wide (Barrows et al. 2014), which attained an elevation of 386 m. During the terminal Pleistocene between 14.5 and 12.5 ka when integrated drainage was re-established in the Nile Basin (Talbot et al. 2000) a seasonal lake 400 km long and up to 25 km wide flooded the lower White Nile valley up to an elevation of 382 m (Williams 2009). Both intervals were times of wetter climate across North Africa and Arabia, with generally drier climates between those times (Williams 2019).

The Gezira alluvial fan and Blue Nile distributary channels

18The distal portion of the Blue Nile valley is a large, low-angle alluvial fan known as the Gezira (figure 1.7). It is traversed by a series of distributary channels, all of which are now inactive although they flowed at intervals during the late Pleistocene, when sand-bearing channels radiated across the surface of the Gezira to join the White Nile up to 120 km south of its present confluence with the Blue Nile (Williams 2009, Williams et al. 2015a). Later Blue Nile incision beheaded these channels, reduced seasonal flooding, and caused desiccation of the wetlands that were present along the lower Blue and White Nile valleys. Borehole samples from the northern Gezira show a vertical alternation of sands and clays (Williams et al. 1982; Woodward et al. 2007). A likely explanation is that the Blue Nile oscillated between two hydrologic regimes. During cold, dry climatic phases the plant cover in the uplands was less abundant and accelerated hill slope erosion provided the headwater channels with abundant sand and gravel (Adamson et al. 1980; Williams & Adamson 1980). The Blue Nile was then a bed-load river with multiple distributary channels transporting and depositing coarse sediment (sand and gravel) and aggrading its alluvial fan. With a return to warmer, wetter climatic conditions the plant cover spread across the headwaters and hill slope erosion was reduced. The surface rocks were exposed to chemical weathering and fine-grained soils developed across the landscape. The slopes now provided the stream channels with a suspension load of silt and clay. Seasonal floods deposited a layer of clay across the Gezira fan until the main Blue Nile channel began to incise its bed, ultimately depriving its distributary channels of flow, confining the zone of annual deposition of flood silts and clays to within a few km of the river. Renewed climatic desiccation would once again be reflected in deposition of sand and gravel and renewed fan aggradation. The channels now visible on the surface of the Gezira were active at least intermittently during 100‑70 ka and 50‑8 ka and source-bordering dunes fed by sediments blown from the seasonal channels were active towards 60 ka. There was a major humid interval with widespread clay deposition across the Gezira at a time of very high Blue Nile flow during 50‑40 ka (Williams et al. 2015a).

figure 1.7

The Red Sea Hills and the Desert Nile

19Erosion of the Red Sea Hills during the past 30 million years (Macgregor 2012) has contributed 220,000 km³ to the Nile Cone or 37% of its total volume of 580,000 km³. (In Egypt the Red Sea Hills and Sinai Desert are sometimes called the Eastern Desert). Rivers flowing west from the Red Sea Hills contributed sediment to the Desert Nile at intervals throughout the Quaternary and seem to have been most active during times of lower glacial sea level in the Red Sea. These rivers are now ephemeral wadis and seldom flow. In their distal reaches the late Pleistocene sediments derived from the Red Sea Hills interdigitate with strictly Nilotic sediments, suggesting that the wadis were active in winter and were fed by winter rains (Butzer & Hansen 1968). Late Pleistocene tufa deposits in the Red Sea Hills lend further support to these conclusions. Tufas form when calcium bicarbonate dissolved in groundwater is precipitated as calcium carbonate as a consequence of carbon dioxide degassing. Such degassing takes place where groundwater emerges in springs or where there is turbulent flow at waterfalls. When tufas are found in now arid areas they are evidence of a previously wetter climate in which higher groundwater levels fed springs and the streams flowing from those springs (Hamdan & Brook 2015). Late Pleistocene tufas in the Red Sea Hills have ¹⁴C ages between 31.2 ka and 22.5 ka (Hamdan & Brook 2015). A single tufa age of c. 62‑56 ka may indicate a wetter phase at this time, but it is unwise to rely on one age, and one also beyond the present range of radiocarbon calibration. The late Pleistocene tufas have δ¹⁸O values that are more depleted than the corresponding values in Holocene tufas, suggesting that the tufas were formed by shallow groundwater springs fed by precipitation from the Mediterranean during an interval when colder glacial temperatures over Europe had forced the westerly wind belt further south (Hamdan & Brook 2015). The late Pleistocene tufas were deposited at temperatures of 14.0°-20.8°C, compared to the equivalent Holocene temperature range of 18.4°-23.4°C (Hamdan & Brook 2015). The stable oxygen isotopic composition of microfossils from two marine sediment cores from the Red Sea also shows that the late Pleistocene sea surface temperatures were 4°C lower than during the Holocene (Arz et al. 2003). The late Pleistocene tufas in the Red Sea Hills extend down to latitude 24°N, showing that moist air masses derived from the Mediterranean were able to penetrate much further south than is generally the case today. If this was true of the Eastern Desert of Egypt, it may also have been true of the Western Desert of Egypt, but independent evidence is needed to confirm this.

20The late Pleistocene alluvial record from the Desert Nile is characterised by repeated aggradation and incision but is fragmentary and not well dated. Optical ages obtained on late Pleistocene Nile alluvium (Williams et al. 2010) indicate that the Nile was migrating laterally and actively aggrading at 83 ± 24 ka, 32 ± 8 ka and 20.7 ± 0.2 ka. The 32 ± 8 ka phase of aggradation may coincide with the 27.8 ± 3.2 ka interval of very high energy flow in the White Nile and with the phase of alluvial fan aggradation near Jebelein on the White Nile right bank discussed earlier (Williams et al. 2010). In the Kerma region of northern Sudan phases of fluvial sand deposition by the Nile and its anabranches (Williams 2019) have optical ages of 86 ± 11, 66 ± 5, 64 ± 8 and 63 ± 7 ka. Very high floods resumed in the Nile at c. 15‑14 ka after a long interval of reduced summer monsoonal precipitation and desert dune reactivation associated with the southward displacement of the Intertropical Convergence Zone (Williams 2019).

21The recurrent phases of late Pleistocene channel incision along the main Nile would have converted previously swampy flood plains into well-drained alluvial terraces more suitable for sustained or seasonal prehistoric occupation.

The Nile Delta and the Nile Cone

22By far the most complete records of past changes in Nile sediment type and amount are preserved in the submerged portions of the Nile Delta and its much larger submarine extension known as the Nile Cone (figure 1.8). However, even these sedimentary archives are hard to interpret without a clear understanding of environmental changes in the Nile headwaters. Within the Nile Cone marine sedimentary sequence there are bands of highly organic material known as sapropels. Sapropels form under anoxic conditions that allow the preservation of organic matter. A sustained influx of freshwater from major rivers like the Nile is thought to be responsible for a change in seawater stratification, enabling organic matter to be preserved at depth under anaerobic conditions (Rossignol-Strick et al. 1982; Rossignol-Strick 1985, 1999). Testing this hypothesis is possible by comparing independently well-dated phases of very high Nile discharge with dated phases of sapropel formation and confirms that at least over the past 125 ka phases of very high Nile discharge do indeed coincide with sapropel events S1 to S5 in the eastern Mediterranean, notably at 125 ka (S5), 104 ka (S4), 81 ka (S3), 55 ka (S2) and 13.5‑6.5 ka (S1) (Williams et al. 2015a). Sapropel ages are numbered consecutively from the youngest (S1) back in time. Except for sapropel S1, the sapropel ages are mid-point ages calibrated against the astronomical cycle and seem to show a 3,000-year time lag between sapropel formation and the correlative precessional minimum shown in the insolation index (Lourens et al. 1996). A more comprehensive depositional chronology based on over forty deep-sea sediment cores collected across the Nile Cone covers the last 200 ka and reveals changes in sediment input and times of sapropel formation associated with phases of very high Nile discharge (figure 1.9 and table 1.3) (Ducassou et al. 2008, 2009). Detailed isotopic and geochemical analysis of core MS27PT located 100 km WNW of the Rosetta distributary outlet on the Nile Delta indicate wetter periods at 98‑69 ka, 60‑50 ka, 38‑30 ka and 14‑5 ka (Revel et al. 2010). The 98‑69 ka composite wet phase is coeval 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. If we include the 38‑30 ka wet phase, the times of high Nile discharge evident from the sapropel record all appear to reflect a precessional signal. Desert dust deposition in this sediment core took place during Marine Isotope Stages 4 (71‑57 ka) and 2 (29‑14 ka), the latter broadly synchronous with the Last Glacial Maximum (Revel et al. 2010).

figure 1.8

figure 1.9

Conclusion

23The combined evidence from terrestrial archives (rivers, lakes, dunes, glaciers) and from marine sediment cores retrieved from the Nile Delta and Nile Cone provides a patchy and often inadequately dated record of prehistoric environments in the Nile basin. The late Pleistocene climates in the upper Nile basin inferred from such evidence have fluctuated between two extremes: cold and arid or warm and wet. Between 75 and 15 ka the regional climate in the upper Nile catchments was mostly dry (apart from moist phases at c. 55‑50 ka and c. 38‑30 ka) with active dunes encroaching on the Nile during times of low Nile flow between c. 71‑57 ka and c. 29‑14 ka. The Holocene and historic records of Nile flow clearly indicate an intricate series of hydrologic fluctuations at time scales ranging from years to decades to centuries. Such evidence is lacking from the late Pleistocene record because the time-resolution of the various available environmental chronologies is too coarse to reveal such fluctuations, although they must have occurred. It is also important to note that late Pleistocene hydro-climatic fluctuations were not always in phase across the entire Nile Basin. For example, in the Red Sea Hills moist air masses from the Mediterranean extended as far south as latitude 24°N between 31.2 ka and 22.5 ka, which is when lakes in the headwaters of the Blue and White Nile rivers were drying out or already very low.

Acknowledgements

24I thank Dr Alice Leplongeon, Dr Mae Goder-Goldberg and Dr David Pleurdeau for their superb organisation of the workshop “Not just a corridor: Human occupation of the Nile Valley and neighbouring regions between 75,000 and 15,000 years ago” held in Paris during May 31-June 2, 2018, from which this paper arose. An unknown reviewer offered much appreciated constructive suggestions. I thank Frances Williams who drew the figures with her customary skill and clarity.