Global Forest Watch (GFW) Commodities recently released new data and analysis for soybean production in Brazil’s Cerrado biome, a mix of forest and savannah rich with biodiversity and carbon. These data show users the spatial extent of soy production and how it relates to deforestation in those areas over time. The relationship between soy production and deforestation is complicated in Brazil and so too is the data that represent it. This blog answers questions that users may have about the data and analyses.
How were the data produced?
What is the Recent Loss index?
The “Recent Loss Index,” available through the analysis feature, shows when tree cover loss occurred in areas that overlap with the soy data. Recent loss (within 2 years of the 2013/2014 crop year) indicates a higher likelihood that the deforestation was due to soy, while historical loss (more than 2 years before the 2013/2014 crop year) indicates a lower likelihood that the deforestation was linked to the soy shown in the data.
Because recent loss presents a higher risk for soy buyers seeking to eliminate deforestation from their supply chains, the index weighs recent tree cover loss more heavily than historical loss. The analysis also accounts for total soy production in an area, making the Index comparable across areas of different sizes. The Index ranges from zero to one, with zero indicating the least amount of risk (tree cover loss is most historical) and one indicating highest risk (tree cover loss is most recent).
Does the Recent Loss Index indicate causation?
No. Although a higher Recent Loss Index indicates a higher likelihood that the deforestation in an area was due to recent soy expansion, it does not definitively show causation. Forests may be cleared for several reasons, including cattle ranching or land speculation in addition to soy expansion (recent expansion or otherwise), and true causation can only be determined through a more careful analysis of land use transitions across many years.
What the data do show is where permanent land conversion (i.e. deforestation) has occurred. If an area currently growing soy experienced forest loss, we can safely say that that land was previously a forest that has since been permanently converted for non-forest use, regardless of whether soy was the primary driver.
How do you define forest loss?
GFW uses tree cover loss data developed by University of Maryland’s Global Land Analysis and Discovery (GLAD) lab run by Dr. Matt Hansen. Tree cover is defined as any vegetation that is at least 5m tall, which includes natural forests and other tree cover types. As a result, the loss indicated by these data is not only forest loss, but inclusive of these other tree cover types, such as plantations. However, users can filter the data by tree canopy density to match a forest definition. For example, the Brazilian government uses a threshold of 10 percent to define their forests. By viewing tree cover loss at a certain tree canopy density, users can get a closer estimate of forest loss. GFW data and analysis default to 30 percent density because it more effectively reflects dense forest, but the Recent Loss Index can be adjusted to calculate at any density.
What is the Cerrado and why does it matter?
The Cerrado is one of the six officially designated biomes of Brazil, spanning central Brazil from the Amazon rainforest in the north to the coastal Atlantic Forest across a range of forest and savannah ecosystems.
The region is a threatened biodiversity hotspot, yet contains most of the country’s planted soy and corn. Soy production has expanded throughout this region, and this expansion has taken place on both previously cleared and newly deforested land. We hope that this layer will help users understand where soy has been expanding recently, and the data and analysis can be used to inform decisions that lead to less deforestation.
Did you include soy growing on areas that were not previously forests?
Yes, the data includes all soy produced across the Cerrado, regardless of the previous land use or native vegetation of that area. However, the year-by-year analysis shows only tree cover loss from 2001-2013, with all other loss labeled as “Converted pre-2001.” Whether the conversion of non-forested areas or savannah ecosystems that fall below a certain canopy density is acceptable is a critical area of debate in Brazil, and this analysis should not be interpreted as a comprehensive indicator of legitimate, approved or legal conversion of areas considered non-forest.
Why doesn’t the analysis include 2014 tree cover loss?
The soy data show the 2013/2014 soy crop year. The land on which soy was planted in late 2013 (October or later) would have to have been cleared in early 2013 (before October) or even further back given the time required to prepare soil for soy cultivation. Any tree cover loss detected in 2014 within these soy production areas would therefore likely be an error.
What about soy production on the border of or just outside the Cerrado?
The analysis tool can be applied to any area of interest (municipalities, uploaded polygons, or freehand selection). However, the soy data only cover the Cerrado biome. Thus, if a user analyzes the soy data in an area or municipality that is partially or entirely outside of the Cerrado, the analysis will apply to only the portion of the selected area the lies within the biome boundary, clipping the rest. We anticipate including soy data for production regions beyond the Cerrado in the future.
Is deforestation legal in Brazil?
Yes, under some circumstances. Brazil’s Forest Code allows landowners in the Cerrado to clear up to 65 percent of forest on their land. Even with this law on the books, and pressure from soy buyers on producers to comply, enforcement in Brazil has been known to be inconsistent. In fact, company sourcing polices and certification schemes are often more stringent, with some prohibiting deforestation completely. The soy data and analysis should not be used to definitively verify legality or compliance, but can help soy buyers implement their zero-deforestation commitments.
Learn more by reading our announcement blog.