At the time, it must have sounded like a sensible case of sustainable development. During the 1980s, about 2,500 square kilometres of Amazonian rainforest was flooded to create the Balbina dam to feed the energy demands of the Brazilian city of Manaus. A sizeable chunk of rainforest was lost, but Brazil gained access to a non-polluting energy source. It's a compromise Brazil has made many times; more than 80% of the country's domestic electricity is generated by hydropower plants.

Yet the clean, green image of dams may have been seriously overstated. Researchers are gathering in Paris next week to discuss greenhouse-gas emissions from tropical reservoirs. Some of the latest findings point to a disturbing conclusion: that the global-warming impact of hydropower plants can often outweigh that of comparable fossil-fuel power stations. If that's correct, current energy strategies, particularly in developing nations, will need to be rethought.

The greenhouse-gas emissions from regions flooded by dams may have been grossly underestimated. Credit: R. ANTONIO/GAMMA

The problem lies with the organic matter in the reservoir. Large amounts are trapped when land is flooded to create the dam, and more is flushed in after that. In the warm water of tropical dams, this matter decays to form methane and carbon dioxide. Although both are greenhouse gases, the main worry is methane, which has more than 20 times the warming impact of carbon dioxide over a 100-year period. In the specific case of Balbina, there is now a rough consensus: in terms of avoiding greenhouse-gas emissions, a fossil-fuel plant would have been better.

But that is where the agreement ends. On one side of the debate is Philip Fearnside, a conservation biologist at the National Institute for Research in the Amazon in Manaus. His work, based mainly on theoretical calculations, looks at water leaving dams. Many dams release water from several metres below the surface, so the flow goes through an abrupt pressure change. Fearnside calculates that this causes methane release, much as carbon dioxide fizzes out when carbonated drinks are opened. His latest results suggest that a typical tropical hydropower plant will, during the first ten years of its life, emit four times as much carbon as a comparable fossil-fuel station.

Lining up against him in a decade-long dispute are Luiz Pinguelli Rosa and his colleagues at the Federal University of Rio de Janeiro, who accuse Fearnside of exaggerating reservoir emissions. They complain in particular that Fearnside has extrapolated from measurements taken on the Petit Saut dam in French Guiana; the data were taken in the years immediately after the reservoir was created, when the store of organic matter would have been greatest.

If these estimates are correct, figures for annual global methane emissions need to be increased by a fifth.

With few data sets available on tropical dams, the debate has increased in acrimony without approaching a conclusion. Environmental groups question the impartiality of Rosa's work, which is funded in part by the hydropower industry. Rosa strongly denies any bias, and in turn accuses Fearnside of seeking to show that “something is wrong with dams”.

The Paris meeting, which runs on 5–6 December and is organized by the United Nations Educational, Scientific and Cultural Organization (UNESCO), is unlikely to settle their dispute, but researchers will discuss new methane data. On 14 November, for example, Frédéric Guérin of the Laboratory of Meteorology in Toulouse, France, and his colleagues published results on methane release from sites downstream of three tropical dams1. They found that so much methane builds up in the dam that downstream emissions, which are rarely factored into estimates of a reservoir's impact, should account for between a tenth and a third of total emissions. Another new paper estimates that, for Balbina, downstream emissions alone have the same greenhouse warming potential as 6% of all the fossil fuels consumed by São Paulo, a city of more than 11 million people2.

Even without these downstream emissions, the global impact of dams may be significant. Danny Cullenward, an energy-policy expert at Stanford University, has made preliminary calculations of the impact of Fearnside's findings. Cullenward stresses that more data are needed, but his estimates suggest that dams release between 95 million and 122 million tonnes of methane per year. If correct, estimates of annual global methane emissions (which do not generally include dam emissions) need to be increased by a fifth. Even extrapolating Rosa's figures gives Cullenward a total of 23 million tonnes.

Many think enough is known to start acting now. Some worry about the huge dam projects tentatively planned for tropical areas, such as a $5-billion project on the Congo river. Another concern is the Clean Development Mechanism (CDM), a system that allows developed nations to fund clean-energy projects in developing nations in return for credits that can be used to meet Kyoto Protocol targets. Current rules allow certain hydropower projects to be funded under the CDM, a situation some scientists and environmental groups would like to see revised.

But matters are unlikely to change without more data, so researchers at the UNESCO meeting will discuss which questions to prioritize and how best to work together. More substantial progress could begin in 2008, when the Intergovernmental Panel on Climate Change (IPCC) will decide whether or not to start work on a special report on renewable energy. Previous IPCC special reports have had significant political impact, and the dams question is likely to fit very well into the scope of the proposed energy study, says Bert Metz, a climate-policy expert at the Netherlands Environmental Assessment Agency and co-chair of one of the IPCC's three working groups.