Michel Rocha Baqueta¹, Noah Larissa Rebuá Thomaz Moreno¹, Enrique Anastácio Alves², Alessandra Sussulini¹

Coffee cultivation in the states of the Legal Amazon has grown significantly. Among these, the state of Rondônia stands out, where indigenous communities cultivate coffee within forested areas. In Amazonas, coffee production follows agroforestry practices in the Apuí region, while in Acre state, cultivation is still emerging. Although all three states produce robusta coffee variety (Coffea canephora), the aromatic profiles of these Amazonian coffees remain undefined. The present study compared the volatile metabolomic profiles of coffees from the three states using headspace solid-phase microextraction coupled with gas chromatography–mass spectrometry (HS-SPME-GC-MS) in an untargeted metabolomics approach, followed by multivariate analysis, favoring the use of entire chromatograms in chemometric analysis to maximize information extraction related to sample origin and quality. Additionally, volatiles extracted from both whole and ground beans were analyzed to assess the influence of sample preparation on the volatile profile. Amazonas sample had a predetermined overall quality score of 81 points on a 100-point sensory scale. Acre sample scored 77 points, while the Rondônia sample, originating from the Sete de Setembro Indigenous Land, achieved an overall quality score of 89 points. Whole coffee beans from Amazonas (18 volatiles), Acre (19 volatiles), and Rondônia (24 volatiles) yielded fewer volatile compounds compared to their respective ground counterparts, which presented 51, 47, and 48 volatiles, respectively. Although direct analysis of whole beans is feasible, ground samples provide a more comprehensive volatile profile. Principal component analysis (PCA) of whole beans provided better separation among the three coffee samples compared to PCA of ground beans. When PCA was performed on both whole and ground beans together, a clear distinction was observed between the two sample preparation methods. The identified volatile compounds mainly belonged to the classes of esters, carboxylic acids, alcohols, phenols, aldehydes, ketones, and caffeine (a xanthine alkaloid). In the ground coffee fraction, PCA evidenced that Amazonas coffee was mainly characterized by 2,3-butanediol (fruity, creamy, buttery), ethyl 2-hydroxybenzoate (sweet, balsamic, medicinal, floral), acetic acid (vinegar-like, sour, pungent), ethyl acetate (fruity, sweet, pineapple-like), caffeine (bitter), and ethanol (alcoholic, slightly sweet). Acre coffee showed a simpler volatile profile, dominated by acetic acid and caffeine, contributing sour, pungent, and bitter notes. In contrast, indigenous Rondônia coffee exhibited a more complex aroma profile, marked by 2,3-butanediol, 2-heptanol (green, herbal, slightly fruity), ethyl 2-hydroxybenzoate, ethanol, and acetic acid, suggesting a broader combination of sweet, floral, herbal, and acidic sensory characteristics. Although dozens of volatile compounds were identified, multivariate analysis revealed only a subset as key contributors to the differentiation of the samples. However, given the complexity of aroma and flavor chemistry, even compounds with lower weights in PCA may play a meaningful role in the sensory perception of coffee during cupping. Therefore, further studies are necessary to more thoroughly characterize the distinctive aromatic attributes of Coffea canephora coffees from the Western Amazon.

Agradecimentos: This study was financed, in part, by the São Paulo Research Foundation (FAPESP), Brasil. Process Number #2024/23102-8.