Brazilian Canephora Coffee Evaluation Using Nir Spectroscopy and Discriminant Chemometric Techniques

Abstract

High-quality Brazilian Canephora coffees are rising to the level of specialty coffees in the face of a new industry perception. In this framework, spectra from 527 coffees were analyzed in the near-infrared (NIR) region. Principal component analysis distinguished the main Canephora-producing states of Brazil, its botanical varieties, low-quality Canephora from specialty Canephora, specialty Canephora from specialty Arabica, and Canephora with geographical indication (GI) from those without this indication. Also, Canephora coffee cultivars from Western Brazilian Amazon were distinguished in the exploratory analysis. Three multi-class PLS-DA (traditional, hard, and soft versions) were compared for the discrimination of the 5 main classes under study: Robustas Amazônicos from traditional (1) and indigenous (2) producers of Rondônia, Conilon from Espírito Santo (3), Conilon from Bahia (4), and specialty Arabica coffees (5). Binary PLS-DA discriminated GI Canephora and non-GI Canephora with 100% sensitivity and specificity. Carbohydrates, chlorogenic acids, lipids, caffeine, and proteins were dominant absorption bands in coffee classifications. The proposed method is objective, simple, fast, and could be easily and reproducibly used in the routine analysis of coffee to verify claims of its identity, variety, and origin.

Keywords: Conilon, Geographical origin, Multivariate classification, NIR spectroscopy, PLS-DA, Robusta