Dams are among the largest structures humans build, and they shape the landscapes they sit in for generations. Large dams provide hydroelectric power, regulate water supply for cities and agriculture, control floods, and enable inland navigation. They also flood upstream valleys, change river ecosystems, displace populations, and sometimes trap enormous volumes of sediment. The balance among these effects has become one of the most contested issues in water management.

The twentieth century saw a dramatic expansion in dam building. Major projects in the United States during the 1930s, including the Hoover and Grand Coulee dams, were followed by similar projects in Europe, the Soviet Union, China, Brazil and many developing countries. By the late twentieth century, most of the world's large rivers had been dammed at least once, and the total reservoir storage behind dams was estimated at more than 7,000 cubic kilometres. The largest individual projects, such as the Three Gorges Dam on the Yangtze, now provide substantial shares of national electricity generation.

The engineering challenges are considerable. A large concrete gravity dam must resist the enormous pressure of the water behind it, at the same time handling earthquakes, sediment accumulation, and the thermal expansion and contraction that curing concrete produces. Earth- or rockfill dams, which rely on mass rather than on tensile strength, face similar pressures but different failure modes. Overtopping - water flowing over the crest - can rapidly erode an earth-fill structure and is considered one of the most dangerous possible failures. Spillways and bottom outlets allow operators to release water when conditions require it.

The decision to build a particular dam has always involved trade-offs. Upstream, the valley to be flooded may contain farmland, settlements, archaeological sites and distinctive habitats. The Three Gorges reservoir displaced over a million people and submerged many archaeological sites; compensation programmes were on a large scale but, according to subsequent studies, often fell short of restoring displaced communities' livelihoods. In the American Southwest, the Glen Canyon Dam, completed in 1964, flooded a canyon that many later regarded as one of the country's most extraordinary landscapes; the loss has been one of the most bitterly argued environmental decisions of the post-war period.

Ecological effects extend far beyond the reservoir itself. Dams block fish migration, and even well-designed fish ladders are imperfect substitutes. Atlantic and Pacific salmon populations have been severely reduced by dams on their spawning rivers, and expensive programmes have been required to maintain remaining runs. Downstream of a dam, the reduced flow and altered timing of water release change the behaviour of the river itself. Sediment that a river would ordinarily carry is trapped in the reservoir, and downstream ecosystems that depend on periodic sediment deposition - from riparian forests to delta wetlands - deteriorate. Hydrologist Dr. Maria Vlasova has shown that the Aral Sea and the Mekong Delta are both textbook cases of the downstream consequences of large dam programmes.

Climate change has complicated dam operation. Some regions now experience more extreme rainfall events, requiring greater spillway capacity than original designs provided. Other regions face longer droughts, reducing the reservoir levels that hydroelectric plants need to operate. In California, drought years have repeatedly forced major hydroelectric plants to reduce output. Rising temperatures also affect the snowpack that feeds many reservoirs, often resulting in earlier spring runoff that is harder to capture and harder to release.

The dam-removal movement, which began seriously in the 1990s, has taken on growing importance. Older dams whose economic value has declined and whose environmental costs are high have been removed in several parts of the United States and Europe. The most notable example is the Elwha Dam in Washington State, removed in 2011 to restore a major salmon river, with dramatic early results: salmon returned to upstream habitats within a year. Dr. Vlasova has argued that dam removal should be considered seriously whenever dams approach the end of their design lives and the original economic case has weakened.

Not every dam will be removed, and new dams continue to be built. What has changed is the framework in which decisions are made. International guidelines, including those of the World Commission on Dams in the early 2000s, have pushed projects towards more rigorous evaluation of social and environmental consequences. Implementation has been uneven, and political and commercial pressures often override formal procedures, but the idea that a dam's costs need to be weighed against its benefits on something approaching a balanced basis is now widely accepted.

The dam, in short, is an object whose promises and costs interact with its surroundings in ways that are difficult to predict in full. Engineering has become better at building safe structures, but the deeper question - when a dam is worth building at all - is one that societies face afresh for each project.