Coffea arabica, L.

Babova, Oxana, Occhipinti, Andrea & Maffei, Massimo E., 2016, Chemical partitioning and antioxidant capacity of green coffee (Coffea arabica and Coffea canephora) of different geographical origin, Phytochemistry 123, pp. 33-39 : 37

2.2. Principal component analysis and cluster analyses of C. arabica and C. canephora accessions

Different classes of compounds have been used for the authentication of different coffee varieties ( Alonso-Salces et al., 2009 and references cited therein) and for distinguishing the species and origins of green coffee bean samples of C. arabica and C. canephora from different geographic regions ( Rodrigues et al., 2009; Sberveglieri et al., 2011; Serra et al., 2005; Wei et al., 2012). We used the data set of Table 1 to calculate a Principal Component Analysis (PCA). Fig. 4 shows the scatter plot of the two main PCA factor loadings, which explained the 96% and 3% of the total variance for Factor 1 and Factor 2, respectively. A discrimination between the C. canephora accessions and all other C. arabica species was observed, mainly because of Factor 2 scores (negative for C. canephora and positive for C. arabica ). Among C. arabica , a separation was found for C. arabica from Kenya and all other accessions, which were separated by positive values of Factor 1. A separation was also observed for the ‘‘caracol” accession from Brasil (C-Brasil). The ‘‘caracol” (or ‘‘snail” in Spanish, also called peaberry) is a natural mutation of the C. canephora coffee bean inside its cherry that affects about 5% of the world coffee. The phenotype of these ‘‘caracol” green coffee produces a single, rather than a double bean that appears smaller, denser and with a more rounded shape with respect to the wild type ( Cilas and Bouharmont, 2005; Giomo et al., 2008). As expected, the chemical pattern of the C-Brasil accession places it closer to the C. canephora group ( Fig. 4 ).

A cluster analysis (CA) calculated on the data matrix of Table 1, by using Euclidean distance with median linkage method, shows a first cluster that isolates the A-Kenya accessions because of the higher content of neochlorogenic acid (5). A second cluster is made by all C. canephora accessions, with a close statistical linkage between R-Vietnam1 and R-Vietnam2. This cluster is generated because of the high amount of caffeine (16). A third cluster isolates the C-Brasil accession because of the presence of 4-O-feruloyl-5- caffeoylquinic acid (15), whereas a fourth cluster gathers the APeru1 accession containing a higher content of caffeic acid (1). All other C. arabica accessions are present in the last cluster because of the high amount of chlorogenic acids. Several subclusters compose this last cluster (see Supplementary Fig. S1 ).

Figures

Fig. 4. Scatter plot of the principal components analysis factor scores showing the separation of all C. canephora from C. arabica accessions. A clear separation is present inside the C. canephora group between Uganda, Vietnam and the ‘‘caracol” mutation from Brasil (C-Brasil). Among C. arabica accessions, accessions from Kenya show an opposite partition from all other varieties; see Table 2 for code names.

Fig. 1. Structure formulae of the identified compounds extracted from Coffea arabica and Coffea canephora. (1) Caffeic acid; (2) p-coumaroylquinic acid; (3) p-coumaroyl-N- tryptophan; (4) chlorogenic acid (3-O-caffeoylquinic acid); (5) neochlorogenic acid (5-O-caffeoylquinic acid); (6) cryptochlorogenic acid (4-O-caffeoylquinic acid); (7) caffeoyl-N-tryptophan; (8) 3-O-feruloylquinic acid; (9) 5-O-feruloylquinic acid; (10) 3,4-O-dicaffeoylquinic acid; (11) 3,5-O-dicaffeoylquinic acid; (12) 4,5-O-dicaffeoylquinic acid; (13) 3-O-feruloyl-4-caffeoylquinic acid; (14) 3-O-feruloyl-5-caffeoylquinic acid; (15) 4-O-feruloyl-5-caffeoylquinic acid; (16) caffeine.