Isotope fingerprinting for traceability in the amazon bioeconomy: A Bayesian assignment approach with açaí

• First isotopic model mapping Amazon açaí provenance using δ18O and δD. • Model localizes 80–90% of true sites within the top 20% highest-probability area. • High exclusion strength: 95% of credible regions restrict origins to ∼2% of Amazonia. • δ¹⁸O and δD isotopes demonstrate feasibility for traceability in the bioeconomy. Açaí ( Euterpe spp. ) is a flagship product of the Amazon bioeconomy, recognized for its high nutritional value and economic importance. The commercialization of açaí could benefit from a straightforward certification or traceability system that verifies geographic origin and product authenticity. This study introduces the first isotopic assignment model for açaí berries from the Brazilian Amazon, utilizing stable isotopes of oxygen (δ¹⁸O) and hydrogen (δD). We developed Random Forest-based isoscapes and employed Bayesian assignment techniques to create spatial posterior probability surfaces for 59 samples with known origins. The model’s evaluation included spatial performance metrics at sample level: posterior quantile rank, distance to the highest posterior cell, and the size of the 95% credible area. These metrics were combined into a composite score to assess individual sample performance and overall model accuracy. The model demonstrated moderate to good discriminatory ability. Although actual origins often fell outside the peak posterior zones, the median 95% credible region covered only 2.0% of the Brazilian Amazon, indicating a strong capacity for exclusion. Sixty-eight percent of samples had quantile ranks below 0.10, and 65% were within their respective 95% credible areas. The overall composite score was 0.29 (95% CI: 0.25–0.34), classifying the model as “good”. We conclude that this model is best suited for exclusion-based applications like isotopic provenance certification rather than for exploratory geographic assignments lacking prior location information. This research demonstrates the practical feasibility of using stable isotopes for traceability in tropical forest products and introduces transferable spatial metrics for future isotopic provenance modeling.