Multivariate methods for characterization and classification of espresso coffees from different botanical varieties and types of roast by foam, taste, and mouthfeel

Three espresso coffee (EC) samples of different botanical varieties and types of roast were prepared in standard conditions using an experimental EC prototype: Arabica coffee, Robusta Natural blend, and Robusta Torrefacto blend (a special roast by adding sugar). The ECs were characterized with regard to the physical parameters, amount of total solids, total solids on filtrate, lipids, caffeine, trigonelline, and chlorogenic acids by HPLC, and sensory descriptive analysis related to foam appearance, taste, and mouthfeel. Principal component analysis (PCA) was applied to differentiate the EC samples. Arabica and Robusta samples were separated successfully by principal component 1 (55.3% of variance) including physicochemical and sensory parameters related to foam and taste of ECs. Torrefacto and Robusta Natural EC samples were separated by principal component 2 (20.7% of total variance) including mouthfeel and other attributes of color foam. Some interesting correlations among sensory and physicochemical variables were found. A very simple discriminate function was obtained by discriminate analysis allowing the classification of each EC sample into its respective group with a success rate of 100%.