Illegal logging remains one of the greatest threats to the Amazon rainforest, driving deforestation, biodiversity loss, and ecosystem degradation. While environmental agencies use species identification and wood anatomy to determine where timber originated, these methods are not always sufficient to verify whether wood has been legally harvested.

Researchers are now turning to the chemistry locked within the wood itself.

A new paper published in the Open Access journal Molecules presents the first regional map of oxygen isotope patterns in Amazonian wood. The researchers show how natural chemical fingerprints could help determine where timber originated, providing a powerful new tool to combat illegal logging.

Dr. Luiz Martinelli, lead author of the study, explains:

Our goal is to provide the Federal Police with a tamper-proof method. You can’t falsify stable isotopes.

Every tree carries an environmental signature

Stable isotopes are naturally occurring forms of elements which contain the same number of protons but different numbers of neutrons, this means they do not decay over time.

Elements such as carbon, hydrogen, nitrogen, and oxygen naturally exist in different forms of isotopes. The ratio of these isotopes in living beings varies depending on environmental conditions and the biological processes involved as a plant grows, creating a unique chemical fingerprint. In plants, isotopes become incorporated into cellulose within the bark as they grow, which can preserve a long-term record of local hydroclimatic conditions.

Researchers often refer to this as nature’s barcode.

Because rainfall, temperature, humidity, and evaporation vary across large geographical regions, trees growing in different locations develop distinct isotopic signatures. By mapping these variations across landscapes, scientists can create isoscapes, spatial maps of isotope values, that can be used to estimate where unknown samples originated.

Such approaches have already been used to trace the geographic origin of materials ranging from wildlife products to human remains and narcotics. However, despite the urgent need for better methods to verify timber provenance, no regional oxygen-18 (δ18O) isoscape previously existed for Amazonian wood.

Building the first Amazonian wood isoscape

To create a regional map, the researchers analysed oxygen isotope ratios from α-cellulose, a major part of bark structure, extracted from 387 trees collected across 25 locations spanning the Amazon Basin. The sampling included species from 24 botanical families, capturing the remarkable ecological diversity of the region.

The team then combined these isotope measurements with climatic variables, including mean annual temperature, relative humidity, vapour pressure deficit, potential evapotranspiration, and ratio of stable oxygen isotopes in rainfall.

The researchers compared two modelling strategies, which provide higher accuracy and assist in spotting errors:

• Multiple Linear Regression (MLR): Assumes linear relationships between environmental variables and isotope values.
• Random Forest (RF): A machine learning technique capable of modelling more complex, non-linear interactions.

The researchers also assessed whether models trained using individual trees performed differently from those using site-averaged measurements. This distinction is important because forensic investigations typically involve analysing a single timber sample rather than an average from an entire location.

Mapping a climatic fingerprint

Both statistical models revealed remarkably similar large-scale patterns across the Amazon.

The researchers identified a clear southwest-to-northeast gradient in δ18O values, with higher isotope values in the eastern Amazon and progressively lower values towards the western basin. These patterns closely mirror changes in rainfall isotopes as moisture travels inland from the Atlantic Ocean.

As moisture moves westward across the continent, rainfall gradually becomes depleted in the heavier oxygen-18 isotope. Trees absorb this changing water composition, recording the regional climatic signal within their cellulose. As a result, wood from the western Amazon contains proportionally less oxygen-18 than wood from the east, creating a distinct geographical fingerprint that can be mapped across the basin.

Although Random Forest can model more complex relationships, it offered little improvement over the MLR model. The MLR model explained around 70% of the observed variation in isotope values, while the RF model achieved a comparable 67%, indicating that broad climatic gradients account for much of the isotopic variation observed across the Amazon.

The study also found that the most accurate models were not always the most useful in real investigations. Models based on average data from each site gave the best statistical results, while models using individual trees better reflected the challenges investigators face when analysing a single piece of timber.

Why does this matter?

Illegal logging continues to drive deforestation throughout the Amazon, making reliable methods for verifying timber origin increasingly important.

Current systems rely heavily on documentation, such as Brazil’s Forest Origin Document (DOF), alongside species identification and anatomical analyses. While valuable, these approaches cannot always independently confirm whether timber genuinely originated from its claimed source.

Stable isotope analysis provides an additional line of forensic evidence because the environmental conditions recorded within wood cannot easily be altered after harvesting. Rather than relying solely on paperwork, investigators can compare the isotopic composition of seized timber against regional isoscapes to estimate where it likely originated from.

The researchers found that their current oxygen isotope model can already exclude approximately 92% of the Amazon rainforest when assessing the potential origin of a timber sample. While this represents a significant advance for forensic timber tracking, the remaining uncertainty still covers a substantial area, demonstrating that further refinement is needed.

The study also identified regions with greater prediction uncertainty, particularly within the “arc of deforestation”, an area where illegal logging is especially prevalent but sampling density remains relatively low. Expanding sampling in these regions could substantially improve the accuracy of future provenance models while strengthening their value for environmental monitoring and law enforcement.

Looking beyond a single isotope

While the results are promising, the authors emphasise that oxygen isotopes alone are unlikely to provide a sufficient resolution across such a vast and environmentally diverse region.

Dr. Luiz Martinelli explains:

Since the Amazon is so complex, a very large area, it isn’t possible to use just one isotope.

The research team is therefore expanding the framework to include carbon, nitrogen, and strontium isotopes while increasing the number of sampled trees across the Amazon. They are also exploring complementary chemical approaches, such as mapping the concentration of elements within wood. Combining multiple chemical fingerprints could provide a far more precise tool for determining timber provenance.

This study concludes that future improvements will also depend on denser sampling throughout the basin and incorporating additional ecophysiological variables, such as wood density and plant water-use traits, to better capture the biological processes that influence isotope variation.

Ultimately, this research provides the first regional δ18O isoscapes for Amazonian wood while demonstrating that climate-driven statistical models can successfully capture large-scale isotopic patterns across the rainforest. As these approaches continue to evolve, stable isotope analysis could become an increasingly valuable forensic tool for verifying timber origin, supporting law enforcement, and helping protect one of the world’s most important ecosystems from illegal exploitation.

More research exploring stable isotope analysis and its forensic applications can be found across the Open Access journals Metabolites and Molecules. Alternatively, you can access the full MDPI journal list here.