Now a team from McGill University in Canada and the International Center for Tropical Agriculture (CIAT) in Colombia has discovered evidence that soy production is indirectly causing deforestation. Navin Ramankutty of McGill and colleagues found that soy growth on former cattle pasture in the state of Mato Grosso in the southern Amazon could be leading to the creation of replacement pasture in Pará state further north. In other words, it appears that soy cultivation could be shifting cattle ranching into virgin rainforest.

"Soybean cultivation is booming in the southern Amazon, and Brazil has now become one of the largest exporters of soy," Ramankutty told environmentalresearchweb. "The spurt of research in this area is because of the concern that soybean cultivation in the Amazon may be leading to deforestation. The soybean lobby in the Amazon has refuted this, arguing that soybeans are not causing new deforestation, but simply replacing old abandoned pastures."

Cattle ranching has boomed in the Brazilian Amazon since the late 1970s when state subsidies and infrastructure development kicked in. But export crops such as soy, which is mainly used in animal feed and cooking oil, have increased significantly over the last decade, encouraged by expanding world markets and government incentives. Today Brazil is one of the world's largest exporters of agricultural products.

According to Ramankutty, the emerging consensus seems to be that ranching, and not soy cultivation, is to blame for deforestation. But, while Ramankutty and his colleagues' findings support previous research that most of the deforestation in the Amazon can be explained by pasture expansion in situ and not soy expansion, they also found that soybean expansion could have led to deforestation indirectly.

"We show that soy replacing pastures in some places could then lead indirectly to pastures replacing forests elsewhere," he said. "Others have suggested this indirect mechanism as well. We show some empirical support for it."

To carry out the study, the McGill/CIAT team used annual census data on deforestation, crop harvested areas and livestock population for the period 2000–2006 from municipalities within the Brazilian Legal Amazon. Data for the amount of pasture were only available for 1996 and 2006 but the team estimated pasture for 2000–2005 from the livestock figures. One challenge facing the researchers was the introduction of 16 new municipalities in the region; they overcame this by using a geographic information system (GIS) to develop a consistent approach to the data.

"There are now several studies using remote sensing data that have asked the same question," said Ramankutty. "However, because working with remote sensing data is computer-intensive, time-consuming and expensive, the same question has not been examined basin-wide. On the other hand, census/survey data exist at the municipio level covering the entire Amazon basin."

The analysis showed that deforestation shifted 39 km to the northeast between 2000 and 2006, while pasture moved 87 km to the northwest and soybeans shifted 82 km to the northeast.

"These findings point to the need for more field-based research to test the displacement hypothesis and suggest that policymakers face important tradeoffs to satisfy the demands of the livestock and soybean industries versus conservation of the Amazon," write the researchers in a paper in Environmental Research Letters (ERL). "Policies that benefit one group will likely work to the detriment of the others."