Ancient Lakes of the Sahara

Kevin White & David J Mattingly. American Scientist. Volume 94, Issue 1. Jan/Feb 2006.

Descending the dune onto a hard white surface that was cracked by desiccation and sculpted by the wind into contorted pinnacles, we soon began to kick up clouds of gray dust as our feet sank through the crust into the powdery silt beneath. The sediments disturbed by our footprints were studded with tiny snail shells and the petrified roots of trees and shrubs long since swept away by the intensifying aridity in the Sahara over the past 5,000 years. Our surroundings were also dotted with the detritus of ancient human occupation: fireplaces, discarded tools ranging from crudely broken pebbles to finely detailed arrowheads, mounds that mark the last resting places of Saharans from the distant past. Why were these artifacts so abundant here? Because we were walking across the bed of an ancient lake.

As creatures whose lifespan is numbered in decades, it is hard for individual human beings to comprehend how much weather patterns shift over millennia. Examination of geological history can thus provide some valuable perspective. Most scientific knowledge of climate change has come from the study of ice sheets at high latitudes, and specialists have learned a great deal about the many cycles of glacial advance and retreat over the last two million years. But in another part of the world swings in climate have been equally profound and have exerted an equally decisive influence on human history: the desert. These sandy wildernesses may seem like unpromising places to look for sedimentary records of past environments. The lack of protective vegetation allows the wind to scour away the sediments laid down when the climate was wetter. However, human artifacts, which are often made of materials selected for their durability, are not so easy to erase.

For the past eight years, we and a group of colleagues from King’s College London and the universities of Leicester, Reading and East Anglia have been searching for information about human prehistory in one of the harshest, most inaccessible spots on Earth, a place called the Fazzan. One of three main provinces in Libya, the Fazzan lies in the dead center of the Sahara Desert. These days it receives at most a few centimeters of rain a year, and many years it gets none at all.

Much of our work takes us deep into the sand sea (edeyen) of Ubari, a place no one-not even the nomads who live in the Fazzan-ventures into casually. But this forbidding expanse holds many clues to the past environment of north Africa.

The Edeyen Ubari contains the desiccated remains of many lakes, which are hidden in the depressions between sand dunes. These lakes formed during wetter times, when groundwater lay much closer to the surface than it does today. These bodies were probably shallow, small and swampy, and they were often covered with reeds. Through a variety of daring techniques, geologists and archaeologists can find out when the wet periods came and went. We can also learn when people lived here, and how they adapted to their fickle environment.

Our research, and that of the handful of archaeologists who explored this region before us, have, among other things, led to a completely new view of an elusive and remarkable society-the Garamantes, who were known to the ancient Romans as a race of warriors and nomads. Despite that perception, it is clear from the archaeology that they had agriculture, cities and a phenomenally advanced system of water extraction that kept their civilization going for 1,000 years as the land was gradually drying up around them. The Garamantes eventually lost their battle with Mother Nature. But their story teaches an important lesson-that human history is inextricably tied up with climate change. You can flee it, or, if you are as resourceful as the Garamantes, you can learn to live with it, but you cannot ignore it.

New Evidence from the Skies

The Sahara, the world’s largest desert, has a vital role to play in studying the way humans adapted to long-term shifts in the weather. Around the fringes of the Fazzan, in the Acacus Mountains, rock art and animal bones document the gradual adoption of pastoralism over the course of several millennia. Early engravings of wild game give way to later images of domesticated animals. Savino Di Lernia, an archaeologist at the University of Rome, has argued that growing aridity forced hunter-gatherers to start penning up their prey (particularly Barbary sheep) in anticipation of food shortages, and this practice eventually led to domestication. A further drying episode around 5,000 years ago led to the gradual transformation of these herding societies into oasis cultivators, marked by the rise of the Garamantian civilization.

On a broader scale, the changing environments of the Sahara probably acted as a spigot in controlling our ancestors’ exodus from the East African Rift Valley into Europe and beyond. When the climate was relatively humid, the passage north was open, but the abrupt onset of arid conditions shut off the pipeline from time to time.

Unfortunately for archaeologists, the Sahara is a very large, inaccessible place to be looking for the scant remaining evidence of early human activity, and these hunter-gatherers left few signs of their occupation. But people’s basic need for water provides a clue as to where Saharan archaeologists should be digging. Grinding stones and hand axes are usually found in greatest abundance near ancient rivers, lakes and springs, although most of these sites have subsequently dried up.

How do you find a river or a lake that isn’t there any more? The answer, it turns out, is to look from space. This unexpectedly powerful approach was first demonstrated in 1982, when a team of NASA scientists analyzing space-radar images of the Selima Sand Sheet on the border between Egypt and Sudan discovered a river network buried beneath the sands. This feat was possible because radar can penetrate some distance into very dry ground-a feature of little use over much of our damp planet but of great potential in deserts because radar can image the subsurface topography beneath the sands.

Follow-up investigations in the field revealed a wealth of archaeological evidence of periodic occupation of the Selima region over the last 200,000 years. As ice sheets grew and shrank at high latitudes, and as the East African monsoon intensified and weakened in response to changing amounts of incoming solar energy, the climate over the Sahara alternated between arid, semi-arid and wet periods. The last of these provided runoff for ancient rivers and also recharged the aquifers that still provide the groundwater that feeds desert oases, albeit in diminishing amounts.

The NASA discovery caused great excitement among the scientists who study the prehistory of the Sahara. We already knew of the archaeological heritage of the Fazzan, thanks to two decades of research by British archaeologist Charles Daniels and Sudanese archaeologist Mohammed Ayoub in the 1960s and 1970s. But these investigators were constrained to keep to known navigation routes through the desert, the location of which was usually governed by factors other than archaeological richness, such as the urgent need to get to the next oasis. The NASA work showed that remote sensing could be used to reconnoiter far from the beaten track, to identify where water was once present and thus where evidence of human habitation was likely to be most abundant.

Indeed, space-based measurements provided our team with two different ways to identify ancient lakes in the Fazzan. As these bodies dried out, the resulting mineral precipitates, particularly calcium carbonate and gypsum, cemented the lake sediments together. This hardened layer, called duricrust, prevents the wind from eroding the underlying sediments, forming prominent residual platforms as the loose surrounding sand is blown away. The mineral signatures of calcium carbonate and gypsum can easily be detected by the “multispectral” images produced by the Landsat Thematic Mapper, a collaborative Earth Observing System mission involving NASA and the U.S. Geological Survey.

In some places the duricrust does not contain much calcium carbonate or gypsum, perhaps because the chemistry of the original water was unusual or was changing as the lakes dried up. In these places the duricrusts are cemented by silica, which is the same mineral that makes up the surrounding sand. Multispectral images are of little use in these cases. However, we can still detect the difference between duricrust and sand in ordinary radar images, such as the ones NASA employed to such good effect in the Selima Sand Sheet. We can do so because the microwave pulses used for radar mapping penetrate into the loose sands but reflect strongly from the hard duricrusts.

By a combination of these two methods, we produced a comprehensive map of duricrusts in the Ubari Sand Sea. These images also proved invaluable for planning our journeys into this no-man’s land. A skilled interpreter can recognize areas where chaotic patterns of dunes with steep slopes would bring an expedition to a halt. Using the Global Positioning System for navigation through the trackless sands, we could pinpoint our position on the map and stick to relatively safe routes.

The Perils of Desert Research

After charting out our course through the desert, we mounted several two-vehicle expeditions to the duricrust sites, at a rate of two to three per field season. We traveled in two well-equipped Land Rovers, which were fitted with special sand tires that can be deflated to run at very low pressures. This tactic has the effect of increasing the footprint of the treads, providing extra traction and “flotation” by spreading the weight over a larger area so as to combat the tendency of the tires to sink in. Even so, expedition members knew that they would be spending a lot of time digging and pushing over the next few days.

There is no other surface quite as challenging to traverse as a sea of sand dunes. Both vehicles were equipped with winches and ropes, and the drivers on our team made sure to keep them some distance apart, so if the lead vehicle got mired in a patch of soft sand, the other could stop in time and pull the stalled Land Rover out backwards. When we got bogged down, we could also put sand ladders-narrow aluminum gratings-underneath the tires to provide a firm surface for them to grip. In this way the stranded vehicle could climb out of the soft patch under its own steam, or perhaps with a bit of a push from a team of weary scientists.

Getting over a big dune requires steady nerves and unflinching concentration. You have to judge the acceleration just right to generate enough momentum to carry the vehicle up the steep, soft slope. Too little momentum and you get stuck halfway up the dune, wheels spinning and sending out rooster-tails of sand. Overdo it and you become airborne at the top, which can be very bad news if there is a steep drop on the other side. Some of the dunes in the Edeyen Ubari are more than 30 meters high, and the semi-controlled slide down the far side can be a frightening experience. If the driver doesn’t keep the vehicle straight, it could roll over, with potentially fatal results.

Added to the above hazards is a very disconcerting phenomenon called “beige-out,” which begins each day about mid-morning. Up until this time, the low sun illuminates the small ripples on the surface of the dunes, and the driver can read the slope of the terrain. But after the sun rises high enough, it no longer highlights these subtle undulations, and all the driver can see is what appears to be a flat, featureless beige-colored plain. It is very hard to imagine this sight until you have experienced it, but without any surface features to indicate the lay of the land, the driver has no idea whether there is a steep slope immediately in front. No matter how slowly you creep forward, you will inevitably have many heart-stopping moments when you suddenly find the vehicle pitching down at a dangerous angle or taking to the air as you encounter an unseen dune.

On top of all these challenges, expeditions deep into the sand sea require very large volumes of fuel and water, to allow a suitable safety margin in case disaster strikes. Inevitably the vehicles take to the dunes in a very heavily laden state, compounding their tendency to get stuck. Fuel consumption is many times higher than it would be over less punishing terrain, because the wheels are churning through soft sand in low gear. We frequently had to stop to allow the engines and transmissions to cool.

But desert travel is not all hardship. Gazing at the ever-changing landscape, precision-sculpted by the wind, makes time pass quickly. Camping out in the dunes is also a rewarding experience. With no clouds and no light pollution to compete with the stars, you are treated to spectacular views of the night sky as you drift off to sleep.

Measuring Time

Even the simplest visual observations provided a fascinating insight into the changing natural and human environments of the Fazzan. Several phases of lake formation and desiccation were evident at each interdune basin we visited, with higher terraces of duricrust-capped lake sediments representing higher and older water levels. Fossil shells in the lake sediments beneath the duricrusts attested to the presence once of brackish perennial lakes between the dunes and provided opportunities for dating those wetter times.

To ascertain the age of the mollusk shells we found around the Saharan palaeolakes, we initially employed the well-known method of radioactive dating, but with some interesting wrinkles that were specific to the lake setting. Uranium-238, a radioactive element, decays eventually to a stable isotope of lead (lead-206), but it does so in a number of steps, each forming a different unstable isotope. The ones that are important for dating shells are uranium-234, which is soluble in water, and its daughter product thorium-230, which is not. Hence the water in the lake contained lots of soluble uranium-234, but no thorium-230 (which was all locked up in the sediments at the bottom). Mollusks living in the lake absorbed trace amounts of the uranium-234 and incorporated it into their calcium carbonate shells. This uranium-234 immediately started to decay into thorium-230 at a predictable rate. A very young shell would have little thorium-230, because the uranium would not have had much time to decay. An old shell would have more thorium-230. Geochronologists can thus use the ratios of thorium-230 and uranium-234 found now to work out how old ancient shells are, back until about 500,000 years ago.

We dated 10 shells from one of the lake deposits using this technique, and the results exhibited a wide range of ages, from 56,000 to 84,500 years ago. Although we tried to pick the shells that looked most pristine, we found evidence of later disturbance: Much of the original calcium carbonate that made up the shell (a mineral known as aragonite) had subsequently changed to calcite. The latter mineral in these particular mollusks is a sure sign of recrystallization. That means the uranium in the shell had a chance to dissolve away and alter the isotope ratio, thus changing the apparent age of the fossil. The errors can go in either direction. Resetting the isotopic clock would make the fossils seem younger than they are; on the other hand, the dissolution of soluble uranium tends to depress the ratio of uranium to thorium, making the shell appear older. Reluctantly we had to conclude that these shells cannot provide reliable dates for when the lakes were full.

A second strategy we used for dating the wet phases also relies on the presence of radioactive material in the sediments, but in a very different way. Optically stimulated luminescence (OSL) exploits the damage caused to surrounding atoms by the emission of energetic particles during radioactive decay. These particles rip through nearby atoms and jostle electrons from their normal orbitals around the nucleus, sending them into to higher-energy orbitals. Sunlight enables displaced electrons to jump back to their normal locations. However, after the sediment has been buried and no longer absorbs sunlight, it starts to accumulate radiation damage. The longer the period of burial, the more damage accumulates. On exposure to light in the laboratory, the damaged atoms repair themselves, absorbing some of the light and giving off a tiny flash as all of the displaced electrons fall back to the ground state. A more damaged crystal will, naturally, emit a larger burst of luminescence. It is essential, of course, to prevent any light from hitting the crystal before it gets to the laboratory, and also to expose it there to a very pure source of light (a laser). The only other information we needed to calculate the period of time since burial was the annual dose of radiation the crystals had received. This quantity was easy enough to determine in the field with a gamma-ray spectrometer.

OSL is particularly useful in deserts because it works best with crystals of quartz, the most abundant material in the Sahara. We used it to date the sands beneath lake sediments, which represent a dry phase before the lake began to fill up. Our sand samples yielded ages ranging from 226,000 to 255,000 years ago. Although these dates don’t tell us when the lakes were full, they provide an upper limit on age; that is, we know that the overlying lake sediments must be younger.

Knowing the shortcomings of OSL, we also employed a third dating technique. Many of the ancient lake basins contain dark organic layers at their base, remnants of the peat left behind as these swampy pools dried up for the last time. Because this material contains carbon, it can be dated using well-established radiocarbon techniques. Carbon also has different isotopes, including the common stable form, carbon-12, and unstable carbon-14. While a plant or animal is alive, it incorporates carbon that comes, directly or indirectly, from the atmosphere, which contains both of these isotopes in a ratio that is roughly constant. (The exact ratio of carbon-14 to carbon-12 has varied over time, a complication that must be taken into account when computing carbon-14 ages.) Following the death of the plant or animal, the carbon-14 in its tissues will decay to carbon-12 and will not be replaced by any new carbon-14; thus, the more carbon-14 one finds in a sample, the younger it is.

Because carbon-14 decays more quickly than uranium-234, carbon dating works on a much shorter time scale. We took three samples of black organic sediments from ancient lake basins, which yielded dates ranging from 6,690 to 9,120 years before present (in the jargon of carbon-14 chronologists, who define the present as 1950). These dates probably represent the most recent humid phases, shortly before the onset of the current aridity.

How People Coped

What is the human story that lies behind these scientific observations of a changing landscape and climate? We found large quantities of stone tools scattered on and around the duricrusts, indicating at least two separate phases of human occupation. The earliest humans in the area in the central Sahara were Paleolithic hunter-gatherers, who lived between about 200,000 and 70,000 years ago. They survived by hunting large and small game in a landscape that was considerably wetter and greener than it is now. A great many studies of lake deposits and other climatic indicators, such as pollen, support the existence of major rivers and lakes in northern Africa at this time, and we can imagine a lush savannah landscape.

A prolonged arid phase from about 70,000 to 12,000 years ago apparently drove humans out of the region. But then the rains returned, and people settled here once again. Paleolithic hand axes can be found in close proximity to finer Mesolithic and Neolithic tools, showing that the settlers tended to colonize the same areas as before-around the lakes and on rocky outcrops. To judge by the differences in lake levels, there was not as much rain in Neolithic times as there was during the Pleistocene humid phase. Nevertheless, archaeologists have found rock-art depictions of water-dependent creatures, such as elephants, rhinos and the massive bubalus bovine, which give a clear indication that this region again was a savannah with ample sources of water. Some of the duricrust deposits in the sand sea certainly belong to this phase. The most recent wet phase in the Sahara lasted to only about 5,000 years ago, after which time the lakes evaporated, and the desert took hold.

There was a remarkable difference between this relatively recent climatic downturn and the earlier Pleistocene arid period. This time, humans found ways to overcome the difficulties and remain living in the increasingly hostile desert environment. Had they still relied on hunting and gathering, survival would have been impossible. But by the mid-Holocene, judging from the changing nature of the rock art, they had made the transition to a pastoral lifestyle.

One of our most remarkable finds at one of the lake sites was a so-called “antenna tomb,” a distinctive burial site having a central cairn with its heaped stones arranged in two 20-meter projecting arms. This was the first such structure to be discovered on a lake bed, and the first to use duricrust (which is almost as tough as stone) instead of rocks. Antenna tombs are believed to be the final resting places of the pastoral people who lived in the Fazzan between 5,000 and 3,000 years ago, after the climate had begun to dry out. It must have taken huge ingenuity to maintain their herds and their lifestyle in a landscape where the surface water sources were progressively shrinking and failing.

However, the long-term answer to living in the desert came through another profound change-the transition to an agricultural civilization with towns and villages. This process culminated with the emergence of the Garamantian society in the first millennium B.C. The Garamantes were sedentary farmers who controlled the Fazzan oases between about 500 B.C. and 500 A.D., and thus found themselves at the hub of important trade routes from sub-Saharan Africa to the Mediterranean. The Greeks and Romans knew about the Garamantes and portrayed them as notorious and unruly nomads. Yet the archaeological evidence of their settlements and irrigation systems is unequivocal. They lived in a centralized society, with impressive dwellings, temples, public buildings and funerary monuments. Analysis of botanical remains from our excavations shows that they cultivated a variety of high-grade cereals (wheat, barley, millet and sorghum) and other crops (date palms, vines, olives, cotton, vegetables and pulses). They lived in a long, linear, valley-like depression to the south of the sand sea where water lay close to the surface, a region now called the Wadi al-Ajal. This area is still relatively sand-free, and the bedrock can be seen in satellite images.

The Garamantes drew their water from a large subterranean aquifer that had built up over earlier, wetter millennia, and they transported it through a network of tunnels called foggaras by modern-day Berbers. (We do not know what the Garamantes called them.) This irrigation scheme required an impressive amount of engineering skill, social organization and manpower-much of which came from slave labor. Presumably the aggressive nature that impressed the Greeks and Romans helped the Garamantes to acquire the necessary slaves.

But even this remarkably advanced society, the first urban civilization ever built in a desert, could not cope forever with the progressive draw-down of the water table. Sometime around 500 A.D., the Garamantian civilization collapsed, and the foggaras fell into disuse. Oasis agriculture on a smaller scale and using simpler technology continued through the medieval period and into modern times, but the pre-Islamic culture of the Garamantes was largely forgotten, even by their descendants.

The environment found today contains many further clues to the Fazzan’s past. There are still a few interdune lakes left in the Ubari Sand Sea, fed by aquifers that reach the surface. These small bodies may provide analogues for the ancient lakes that now exist only as desiccated remnants. Some of these modern lakes are so shallow that they dry to a hard white crust of salt during the summer, but others are deeper and perennial. They form islands of plant and animal life amid the sterile sea of sand.

Continued Struggle

The battle to wrest water from the desert is not over yet. In the past 40 years, improved technology has once again made it possible for people to tap the ancient aquifer on a huge scale. In 1984, Libya began work on one of the largest civil-engineering projects in the world, the Great Man-Made River. The second stage of this project, completed in 1996, pumps about a million cubic meters of water per day from wells in the Fazzan, through underground pipes, to Tripoli and the surrounding regions on the coast. (For comparison, the volume of water moved is about the same as the capacity of the Los Angeles Aqueduct.) The project, a source of great national pride in Libya, has stimulated population growth and the “greening” of the desert. But for how long?

Already the groundwater level in the Fazzan is dropping with ominous rapidity. The most likely prognosis is that agriculture will be forced to contract around a decreasing number of settlements sustained by deep-bore artesian wells. The reduction in the number and area of oases could have a profound impact on ecological diversity bird migration patterns and human commerce. Clearly, the last chapter has not yet been written in this long story of human adaptation to climate change.