Saving the Amazon

A forest that makes its own weather, and the arguments about how to keep it

A The forest is usually discussed as a store of carbon, which is accurate and incomplete. It is also a machine for moving water. Trees draw moisture from the soil and release it from their leaves; the released vapour forms clouds that travel west, fall as rain, and are released again by the next stretch of forest, so that a single molecule of water may cross the basin several times before reaching the Andes. The flows involved are enormous — comparable in volume to the river itself — and they supply rainfall to agricultural regions thousands of kilometres away in southern Brazil and northern Argentina. The forest is not merely a beneficiary of the climate. It is a substantial part of the mechanism that produces it.

B This is what makes the concept of a tipping point more than rhetorical. If enough forest is removed, the remainder generates less vapour, the dry season lengthens, and the surviving trees become less able to withstand it; beyond some threshold the process sustains itself and much of the basin converts to degraded savannah regardless of what is done afterwards. Estimates of where that threshold lies have ranged between a fifth and a quarter of the original area, and something like a fifth has already gone. The estimates are model-dependent and are disputed in detail. The mechanism itself is not. Nor is it purely theoretical: measurements over the southern and eastern parts of the basin, where clearance has been heaviest, already show a dry season lengthening by several weeks compared with fifty years ago, and parts of that region have shifted from absorbing carbon to releasing it.

C The causes of clearance are better understood than they are usually presented. The overwhelming majority of cleared land becomes pasture, and cattle ranching in the region is often barely profitable, which puzzles anyone who assumes that forest is destroyed for gain. The explanation is that clearing land establishes a claim to it, and a claim can be sold. Cattle are, in this sense, a low-cost method of demonstrating occupation rather than an agricultural enterprise, and the profit is realised when the land is transferred, frequently after a change in its legal status. This explains a pattern that otherwise looks irrational: clearance accelerates in advance of legislation expected to legalise existing occupation, because the announcement of an amnesty is an invitation to occupy something before it arrives.

D Policy has demonstrated, twice, that the process is reversible. Between the mid-2000s and 2012 annual deforestation in Brazil fell by roughly eighty per cent, through a combination of satellite monitoring that identified clearance within weeks, enforcement directed at the largest offenders, the withholding of agricultural credit from municipalities with high clearance rates, and agreements under which traders refused to buy soy grown on recently cleared land. Rates rose again when enforcement was relaxed and fell again when it was resumed. The lesson is not that any single measure works but that the outcome depends on decisions that are made and unmade within a political cycle.

E The most consistent protection has come from a source that receives little of the funding. Land under the legal control of Indigenous communities shows deforestation rates dramatically lower than comparable land outside, including land inside conventional protected areas, and the difference persists when the comparison controls for remoteness and terrain. The mechanism is not mysterious: people who live in a forest, whose claim to it is recognised, and who are present continuously, detect and resist incursion in a way that a distant agency cannot. Securing and enforcing such tenure is among the cheapest interventions available and is a fraction of what is spent on schemes to pay for carbon.

F Those schemes have had a difficult decade. Projects generating credits for avoided deforestation depend on estimating what would have happened without them, and investigations have repeatedly found that the counterfactual was set so pessimistically that credits were issued for protecting forest under no particular threat. Several large certifiers have revised their methods. The underlying idea — that the world should pay the countries holding the forest to keep it standing — remains sound and has not yet been implemented at a scale that would matter. The sums involved are not large by the standards of the industries whose emissions they would offset, and the obstacle has never been the arithmetic. It has been that a payment for something not happening is difficult to verify and politically easy to attack.

G A final correction concerns what the forest is. The image of an untouched wilderness is a product of a demographic catastrophe: the basin supported large populations before European contact, who managed the land extensively, and whose numbers collapsed after 1500 through disease. They left evidence in the soil itself — patches of dark, fertile earth created by centuries of settlement — and in the distribution of useful trees, which is measurably non-random around ancient sites. Much of what looks primeval is regrowth over abandoned management. This matters practically as well as historically: a forest that has been inhabited and shaped for millennia is a different object from a wilderness, and the policies that follow from each description are not the same.