Résumé
The objective of this work was to compare the effectiveness of three chemical families - namely elements, chlorogenic acids (CGA) and fatty acids (FA) - for the discrimination of Arabica genotypes (traditional versus modern introgressed lines) and potential terroirs within a given coffee growing area. The experimental design included three Colombian locations (Location 1, Location 2, and Location 3) in full combination with five (one traditional and four introgressed) Arabica genotypes and two field replications. Elements, chlorogenic acids and fatty acids were analyzed in coffee bean samples by ICP-AES, HPLC and GC, respectively. Analysis of variance (ANOVA), principal component analysis (PCA) and discriminant analysis (DA) were carried out to compare the three methods. A significant effect of the location was observed for almost all compounds measured, as inferred by twoway ANOVA, revealing the potential of the three chemical classes studied for discriminating coffee terroirs within a given country. The effect of the genotype was highly significant with most of the chlorogenic and fatty acids measured. By contrast, most of the elements analysed showed no significant differences among genotypes. Though elements provided an excellent classification of the three locations studied, as estimated by combined PCA-DA approach, this chemical class was useless for genotype discrimination. Chlorogenic acids gave satisfactory results, but fatty acids clearly offered the best results for the determination of both genotypes and environments, with very high percentage of correct classification (79 and 90%, respectively). In order to take advantage of both climatic and soil diversity, one major practical recommendation which can be drawn from the present work would thus to undertake the simultaneous analysis of FA and elements for coffee origin authentication. (Résumé d'auteur)