106.

Ribeiro J. S., Salva T. J. G., Tomaziello R. A., Augusto F., Ferreira M. M. C., "DISCRIMINATION OF BRAZILIAN ARABICA COFFEE ACCORDING TO OVERALL QUALITY USING AROMA COMPOSITION, SOLID PHASE MICROEXTRACTION (SPME) AND PARTIAL LEAST SQUARES - DISCRIMINANT ANALYSIS (PLS-DA)". Campinas, SP, Brazil, 14-19/09/2008: 22nd International Conference of Coffee Science, Programme & Abstracts (2008) 131. Poster PC767.


PC767
DISCRIMINATION OF BRAZILIAN ARABICA COFFEE ACCORDING TO
OVERALL QUALITY USING AROMA COMPOSITION, SOLID PHASE
MICROEXTRACTION (SPME) AND PARTIAL LEAST SQUARES -
DISCRIMINANT ANALYSIS (PLS-DA)
____________________________________________________________________________________________________

RIBEIRO, Juliano S.*, SALVA, Terezinha J. G**, TOMAZIELLO, Roberto A.**, AUGUSTO, F.* and FERREIRA, Márcia M. C*

*Universidade Estadual de Campinas, SP, Brazil, **Instituto Agronômico de Campinas, SP, Brazil.

Flavor plays an important role in sensory analyses and could be considered  a  “fingerprint”  of  products [1].   The flavor of
coffee is composed of  an  extremely  complex  mixture  of  volatile  compounds  that  presents different  concentrations and
intensities.   Each functional class found i n  roasted coffee flavor shows different compounds  and  derivatives  with distinct
sensorial attributes and odorific impact.   In order to get insight into which peaks  of  gas chromatographic analysis could be
responsible for the discrimination of the Arabica coffee samples according to their overall quality,  eleven samples with high
overall quality (class one), and 9 samples with low overall quality (class two) have been investigated.

Partial least squares-discriminant analysis  (PLS-DA)  [2]  was applied to the pre-selected peaks  and,   from the scores plot
shown  in  Figure 1A,  two distinct groups can be visualized.   Coffee samples with low overall quality are located  on the left
side, with negative scores in LV1 (51.28 % of the original information), well separated from samples with high overall quality
on the right side, with positive scores.
 
 


 

Figure 1 – Scores (A) and Loadings (B) plots. Low  (circles)  and high  (triangles)  overall quality samples.   The numbers in
figure A and B are the peaks responsible for the discrimination
 
 

From the  loadings  plot  of  LV1  (Figure 1B)  it  can be  seen  that  the  higher  the  concentration  of  3-methypropanal  (1),
2-methylfuran (2),  furfural (4),  furfuryl  formate (5), 5-methyl-2-furancarboxaldehyde (6) and 4-ethylguaiacol (13) the higher
the  quality  of   the  arabica  coffee.   On  the  other   hand,    3-methylthiophen   (3),    2-furanmethanol   acetate   (7)   and
2-ethyl-3,6-dimethylpyrazine (9) tend to be more concentrated in worst beverages. The not yet identified compounds (8, 10,
11 and 12) are undergoing new mass spectrometry analyses.

[1] Cuevas-Glory, L. F., Pino, J. A., Santiago, L., Sauri-Duch, E., Food Chem., 103,1032 (2007);

[2] Barker, M., Rayens, W., J. Chemom., 17, 166 (2003);
 
 
 

131