Scientists unveil breakthrough tool that could help stop the world’s third‑biggest driver of deforestation

Scientists at the Royal Botanic Gardens, Kew, World Forest ID, University of Sheffield and international collaborators have developed a new technique that can identify where soybeans – the third largest driver of tropical deforestation – are grown to within roughly 200 kilometres, a breakthrough that could transform efforts to stop deforestation linked to global food supply chains.

The Defra-funded study, published today in Communications Earth and Environment, combines chemical fingerprinting of soybeans with advanced geospatial machine learning to estimate where crops were harvested across South America. Researchers say the method could help regulators, scientists and companies verify the origins of commodities that are often traded through complex international supply chains.

Agricultural expansion remains the biggest driver of tropical forest loss, with 3.7 million hectares of tropical forest lost in 2023 alone, while 71.6 million hectares were lost between 2001 and 2015. Soy, primarily produced for pig and poultry feed, accounts for around 11.5% of commodity-driven deforestation, particularly in South America where production is rapidly expanding to meet global demand. The crop is the third-largest driver of tropical deforestation, behind cattle and oil palm, yet tracing where soy was grown is difficult because shipments are often mixed and traded across multiple countries. 

The new study shows it is possible to estimate soybean harvest origin far more precisely than previous methods, which could only classify by country or broad region. By analysing stable isotope ratios and trace elements across 267 soybean samples collected throughout South America, and combining them with environmental data, scientists have developed a machine-learning model that predicts crop origin to within 192.52 (±23.51) km from the harvest location. This is significant as deforestation risk varies dramatically over short distances, sometimes even between neighbouring farms. 

Working across complex supply chains, scientists describe this as “a leap forward in commodity traceability” capable of verifying whether a shipment’s declared origin matches where the crop was grown. The model is already being applied to other deforestation associated commodities, including timber, and can also be applied to cacao, coffee, palm oil, and rubber.

Caspar Chater, Senior Research Leader at RBG Kew says, ‘Supply chains for commodities like soy are incredibly complex, but this approach provides transparency regardless of supply-chain complexity. This represents a significant advance in our ability to trace agricultural commodities back to where they were grown.’

There is an urgent need for importing economies to regulate this trade, and this study could help support the implementation of new environmental legislation such as the EU Deforestation Regulation which is due to come into force in December 2026. This legislation requires companies importing certain commodities to prove they were not produced on recently deforested land, and this model provides regulators with a scientifically defensible way to confirm whether origin claims are plausible. Similar UK legislation is expected to follow under the Environment Act’s Forest Risk Commodity regulation, with Defra’s funding for this project reflecting an ambition for legislation that requires the use of traceability tools for all listed Forest Risk Commodities. UK businesses have also committed to eliminating deforestation from soy supply chains through the UK Soy Manifesto (2021), a cross-industry initiative involving retailers, manufacturers and food companies.

While consumers do not buy soybeans directly, choosing supermarkets with strong soy sourcing policies is a good first step. Until now, it has not been possible to verify these claims, but this new technology reveals which supermarkets are actually compliant. Similarly, products with ‘deforestation‑free” or “verified origin” claims can now be checked for the first time.

Jade Saunders, Executive Director at World Forest ID says, ‘By making it possible to verify where soybeans are grown with unprecedented precision, this innovative tool gives companies and regulators a powerful new way to turn deforestation-free commitments into real-world accountability, enabling robust verification, detecting misreported origins in high-risk regions and strengthening compliance.’

While the technology described is not meant as a ‘silver bullet’, it can now be added to the tools available in regulating deforestation, presenting a scenario in which we can conceivably foresee the elimination of deforestation and land conversion from many globally significant supply chains.  

ENDS

For high-res images, please download from the following link and credit as named: https://we.tl/t-JNeaVrmyKtHnqHv3

For interview requests please contact: Sebastian Kettley, Senior Press Officer (s.kettley@kew.org) / Chloe Wells, PR Manager (c.wells@kew.org) / Kew Press Office (pr@kew.org)

Key findings from the study:

• Average origin prediction error of ~192.5 km
• Over 94% of South American soybean production regions are represented in the dataset
• Over 88% accuracy for predictions to country level
• 3.8% of the study region contained the highest probability of origin
• Soy production is concentrated in Brazil, Argentina and Bolivia
• 267 soybean samples analysed across South America