With that in mind, a team from Germany and Kenya has projected land-use changes in Brazil to 2020. Its results indicate that the biofuel and cattle ranching sectors should work together to ensure that the increased growth of biofuels does not lead to destruction of rainforest for rangelands. One way to achieve this could be by increasing livestock density.

"Indirect land-use change could considerably compromise the greenhouse-gas savings from growing biofuels, mainly by pushing rangeland frontier into the Amazon forest and Brazilian Cerrado savanna," write the researchers in PNAS.

The team, from the University of Kassel, Max Planck Institute for Meteorology, Potsdam Institute for Climate Impact Research, and Helmholtz Centre for Environmental Research, all in Germany, and the United Nations Environment Programme in Kenya, found that the expansion of biofuel croplands would cause few carbon emissions from direct land-use change as most would replace rangeland areas.

But by 2020 this replacement would cause an expansion of 121,970 sq km of rangeland into forest areas and 46,000 sq km of rangeland into other native habitats, creating carbon emissions from indirect land use change. In fact, the carbon debt created would take about 250 years to repay by using the biofuels in place of fossil fuels. Sugarcane ethanol would be responsible for 41% of this indirect deforestation and soybean biodiesel would account for 59%.

The researchers calculated that increasing average livestock density by 0.13 animals per hectare across the country could be enough to avoid these indirect land-use changes associated with biofuel expansion.

"Brazilian biofuels can cause indirect deforestation and increase carbon emissions if a cooperation towards intensification of cattle ranching is not settled," David Lapola of the University of Kassel and Max Planck Institute for Meteorology told environmentalresearchweb.

Lapola and colleagues used a spatially explicit modelling framework that incorporated a land-use model, a model of the economy of the agricultural sector, and a dynamic vegetation model that included climate-change effects.

"We chose a modelling approach because it allows us to view land use changes in a systematic way," said Lapola. "We were able see the separate contribution of each forcing we fed into the model, such as future biofuel production demands, or future food production demands, or both of them together. Specifically, the indirect land-use changes depicted in the paper are very difficult to detect in reality, and are more easily captured with a geographical model like the one we used."

Now Lapola and colleagues plan to establish ways of detecting these indirect land-use changes in the field.