The great desert

A landscape that was green within human memory, and what it still sends abroad

A The largest hot desert on earth is younger than civilisation in the sense that matters, which is that it has not always been a desert. Rock paintings and engravings across its central massifs show cattle, giraffe, elephant and hippopotamus, and in one celebrated shelter a group of human figures in postures that have been interpreted as swimming. The art is not fanciful. It was produced by people living in a landscape of lakes, rivers and grassland, and the sediments beneath the sand confirm what they drew.

B The scale of the change is difficult to overstate. Lakes covered areas the size of small countries; the largest of them, in what is now Chad, was among the biggest bodies of fresh water on earth and has left shorelines visible from orbit. Fish bones, crocodile remains and the tools of people who lived on both are recovered from the middle of what is now absolute desert. The mechanism behind the change is understood in outline and argued about in detail. The earth's axis wobbles on a cycle of roughly twenty-three thousand years, altering the distribution of sunlight through the year; between about eleven and five thousand years ago the northern summer received more solar radiation, which strengthened the monsoon and drove rain-bearing systems far further north than they reach today. The vegetation that resulted reinforced the effect, since plants return moisture to the air and darken the surface so that it absorbs rather than reflects heat. Green ground makes rain, and rain makes green ground.

C How the period ended is the point of dispute. Sediment cores from the sea floor off the coast, which record dust blown from the continent, were originally read as showing an abrupt transition within a few centuries, and this was taken to demonstrate that the system had crossed a threshold and collapsed. Later work on lake beds and on other cores has suggested a slower and regionally uneven drying spread over a millennium or more. The distinction matters beyond the local history, because a landscape that can flip is a different kind of thing from one that dries gradually, and the same question is now asked of other vegetated regions, including the Amazon. Neither reading is comfortable: a slow drying implies that a system can be lost without any obvious moment of failure, and an abrupt one implies that the moment of failure arrives without warning.

D What the desert produces today travels a very long way. A single depression in the central Sahara, the bed of a lake that dried out, supplies a large share of the dust carried west across the Atlantic each year by the trade winds. That dust contains phosphorus, and the quantity deposited on the Amazon basin is of the same order as the quantity the forest loses each year to its rivers. The connection is not a curiosity: the fertility of the largest rainforest on earth depends partly on the erosion of a dry lake bed on another continent, which is among the better illustrations of how poorly the world divides into separate systems.

E Beneath the sand lies water that the rain of the humid period left behind. Vast sandstone aquifers underlie several countries, holding water tens of thousands of years old, and they have been tapped on a large scale — most dramatically by a project pumping it northwards through pipelines to the coastal cities of Libya. The water is not being replaced at any meaningful rate, so its extraction is mining rather than irrigation, and the arithmetic of how long it will last is a matter of public policy in several states that prefer not to discuss it.

F The desert's edge has been the subject of the most ambitious ecological project attempted in Africa, and of a useful correction. The original proposal was a continuous belt of trees planted across the width of the continent to halt an advancing desert. Two things undermined it. The first was that the desert is not advancing in the manner assumed: satellite records show the boundary moving north and south with rainfall over decades rather than marching steadily southward. The second was that planted trees in dry country largely died, while land managed so that existing roots and seeds could regenerate recovered at a fraction of the cost.

G The project was consequently redefined, and its present form has little to do with a wall. It now supports a mosaic of practices — restoring degraded land, digging crescent-shaped pits that catch what rain falls, protecting the natural regrowth of trees on farmland — across a band of the Sahel, with results that are patchy, real and unevenly documented. The revision is the interesting part. A scheme conceived on a misdiagnosis was not abandoned but reformulated around what the evidence showed, which happens rarely enough in development policy to be worth recording.