Food Chemistry 135 (2012) 2575–2583

Contents lists available at SciVerse ScienceDirect

Food Chemistry
journal homepage: www.elsevier.com/locate/foodchem

Climatic factors directly impact the volatile organic compound fingerprint

in green Arabica coffee bean as well as coffee beverage quality
B. Bertrand a,⇑, R. Boulanger b, S. Dussert c, F. Ribeyre d, L. Berthiot b, F. Descroix b, T. Joët c
a
CIRAD, UMR RPB, BP 64501, 34394 Montpellier, France
b
CIRAD, UMR QUALISUD TA B95/16, 73 rue Jean-François Breton, 34398 Montpellier Cedex 5, France
c
IRD, UMR DIADE, BP 64501, 34394 Montpellier, France
d
CIRAD, UMR UR Bioagresseurs de Pérennes, F-34398 Montpellier Cedex 5, France

a r t i c l e i n f o a b s t r a c t

Article history: Coffee grown at high elevations fetches a better price than that grown in lowland regions. This study was
Received 31 January 2012 aimed at determining whether climatic conditions during bean development affected sensory perception
Received in revised form 15 May 2012 of the coffee beverage and combinations of volatile compounds in green coffee. Green coffee samples
Accepted 15 June 2012
from 16 plots representative of the broad range of climatic variations in Réunion Island were compared
Available online 11 July 2012
by sensory analysis. Volatiles were extracted by solid phase micro-extraction and the volatile compounds
were analysed by GC–MS. The results revealed that, among the climatic factors, the mean air temperature
Keywords:
during seed development greatly influenced the sensory profile. Positive quality attributes such as acid-
Coffee
Volatile compounds
ity, fruity character and flavour quality were correlated and typical of coffees produced at cool climates.
Climatic factors Two volatile compounds (ethanal and acetone) were identified as indicators of these cool temperatures.
Beverage quality Among detected volatiles, most of the alcohols, aldehydes, hydrocarbons and ketones appeared to be pos-
itively linked to elevated temperatures and high solar radiation, while the sensory profiles displayed
major defects (i.e. green, earthy flavour). Two alcohols (butan-1,3-diol and butan-2,3-diol) were closely
correlated with a reduction in aromatic quality, acidity and an increase in earthy and green flavours.
We assumed that high temperatures induce accumulation of these compounds in green coffee, and would
be detected as off-flavours, even after roasting. Climate change, which generally involves a substantial
increase in average temperatures in mountainous tropical regions, could be expected to have a negative
impact on coffee quality.
Ó 2012 Elsevier Ltd. All rights reserved.

1. Introduction fee are also subsequently found in roasted coffee. According to

Volatile aroma compounds are secondary metabolites gener-
ated mostly from isoprenoids, phenylpropanoids, amino acids,
and fatty acids (Goff & Klee, 2006). Combinations of volatile organ-
ic compounds determine the aroma of fruits and vegetables
(Schwab, Davidovich-Rikanati, & Lewinsohn, 2008). Although a sin-
gle fruit or vegetable synthesizes several hundred volatiles, only a
small subset generates the flavour fingerprint. For example, in to-
mato, although 440 different volatile compounds have been found,
only a few have a marked impact on the organoleptic properties of
this vegetable (Carli, Fogliano, & Frusciante, 2009). The volatile
composition of roasted coffee is likely one of the most important
factors for determining coffee character and quality (Flament,
2001). Volatile compounds generated by roasting has been the fo-
cus of several studies (Buffo & Cardelli-Freire, 2004; Mayer, Czerny,
& Grosch, 1999). Some compounds that are present in clean green
c-
⇑ Corresponding author. Tel.: +33 467 4162 73; fax: +33 04 67 41 61 83. o-
E-mail address: benoit.bertrand@cirad.fr (B. Bertrand). f-
Flament (2001) and Grosch (2001), out of the 300 volatile com-
pounds detected in clean green coffee, around 200 are still present
after roasting. Pre-existing volatiles in green coffee have been stud-
ied (Agresti, Franca, Oliveira, & Augusti, 2008; Cantergiani et al.,
2001; Spadone, Takeoka, & Liardon, 1990; Toci & Farah, 2008),
especially as they enable the detection of defective elements before
roasting (spoilage, fat oxidation, stinking beans, etc.).
In recent years, with the double digit growth in the specialty
coffee market, there is an increasing need to differentiate coffee
origin (Giovannucci, Liu, & Byers, 2008). Elevation is the environ-
mental factor most frequently mentioned with respect to origin ef-
fects. Coffee grown at high elevations fetches a higher price than
that grown in lowland regions. Although elevation is empirically
known to have beneficial effects on coffee quality, only a few scien-
tific reports have actually documented these effects. Guyot et al.
(1996) found that coffee from higher elevations in Guatemala
exhibited better beverage quality, and that was later confirmed
in several Central American countries (Avelino et al., 2005;
Bertrand et al., 2006; Decazy et al., 2003). However, it was not

0308-8146/$ - see front matter Ó 2012 Elsevier Ltd. All rights reserved.
http://dx.doi.org/10.1016/j.foodchem.2012.06.060
2576 B. Bertrand et al. / Food Chemistry 135 (2012) 2575–2583

possible to determine whether the positive effects of elevation
were due to a temperature gradient, to other climatic variables
or even to specific soil conditions. The purpose of our work was
to assess, for the first time, whether climatic conditions during
bean development affect volatile compounds in green coffee, and
whether such modifications are reflected in the sensory perception
of the coffee beverage. Reunion Island that associates strong altitu-
dinal and rainfall gradients over short distances and fine meteoro-
logical survey has recently been shown as an adequate location to
address the question of climate influence on chemical composition
of the coffee seed (Joët, Descroix, Bertrand, & Dussert, 2010). In the
following study, we analysed volatile organic compounds from
green coffee samples derived from 16 Arabica coffee plots located
throughout Reunion Island and encompassing a wide range of
tropical climatic conditions.
2. Materials and methods
2.1. Biological material
Experiments were carried out in Reunion Island on seeds of Cof-
fea arabica cv. Laurina (a natural dwarf mutant of cv. Bourbon) dis-
playing two specific traits that have been identified so far: a rather
low caffeine content and a typical oblong shape. The experimental
plots were the same as those previously described by Joët et al.
(2010). They were planted in 2003 without shade and were in their
second year (2006) of production (about 2 m high). Plant spacing
was 2 m between rows and 1 m along rows. Fertilisers were ap-
plied and pests and diseases were controlled according to locally
recommended practises. Out of 107 experimental plots available
throughout the island, 16 that maximise variations in elevation
(150–1032 m a.s.l.) and climatic conditions were selected (Table 1).
The survey unit was a compact plot with a stand of about 240 cof-
fee trees.
2.2. Meteorological observations and harvest dates
The island of Réunion hosts a dense meteorological network of
more than 50 automated stations dedicated to sugarcane produc-
tion (this service is under Météo France and CIRAD supervision).
The experimental design included 16 coffee plots that were care-
fully selected to present a very broad range of climatic variations
(Table 1). Temperatures were also recorded locally (under the cof-
fee canopy) using portable temperature recorders. The average
temperature (Taver) gradient ranged from 13.4 to 25.3 °C, thus
Table 1
Geographical and climatic data for the last 5 months preceding harvest. Plots, date of the harvest; Alt, elevation above sea level (m), longitude and latitude (degrees), mean of
minimum, average, and maximum daily temperatures (°C) (Tmin, Taver and Tmax, respectively); R, rainfall (mm d 1); SR, solar radiation (J cm 2 d 1); PET, potential
evapotranspiration (mm d 1); Tferm, mean temperature during wet processing (°C). Cv (%) = coefficient of variation.
covering what is considered (Mitchell, 1988) to be the optimum
growing range for this species (15–24 °C). Tropical climates are
characterised by a reduced seasonal temperature variation, with
elevation being the main factor modulating this physical variable.
As expected, Taver, Tmin and Tmax were highly correlated with ele-
vation (R > 0.92, P < 0.0001) and highly intercorrelated (R > 0.95,
P < 0.000). Only the average temperature (Taver) was thus taken
into account in the rest of this study. In most Arabica coffee grow-
ing areas, annual rainfall is in the 1500–2500 mm range. In our net-
work plots, rainfall mostly ranged from 807 to 1918 mm/year, with
two outlier spots located on the eastern rainy coast, where annual
rainfall was around 5000 mm. Solar radiation, which ranged from
973 to 1783 J cm 2 d 1 was negatively correlated with elevation
(r = 0.72) due to the frequent cloudy weather in the highlands,
and positively correlated with temperature (R = 0.83, P < 0.01). It
should be noted that the network was set up without shade,
whereas C. arabica evolved in a shady environment in the Ethio-
pian highland forest region. Potential evapotranspiration ranged
from 1.9 to 4.6 mm d 1 and did not show any correlation with
other climatic parameters. Finally, the water temperature observed
during postharvest processing (Tferm) ranged from 19 to 25.3 °C
(Table 1) and was disconnected from climatic variables measured
in fields. Harvest dates differed depending on the location
(Table 1).
2.3. Berry harvest and processing
Coffee berries were collected at the harvest peak for each plot
(April to December 2006). Only just-ripe cherries (when the peri-
carp turns purple) were harvested. Fruits were harvested in the
morning (2000 g of coffee cherries), transported to the CIRAD pro-
cessing factory (Ligne Paradis – Saint Pierre), and directly pro-
cessed in the afternoon by the wet method (depulping,
fermentation and drying). Temperature probes inserted inside
the batch (Tferm) were used to record temperatures during the fer-
mentation process (Table 1), which was divided into three succes-
sive steps: dry fermentation with only mucilage (24 h), maceration
by adding water to cover the seeds (15 h) and finally washing,
while replacing the water three times during this step (11 h). The
green coffee was screened through a size 15 sieve and defective
beans were discarded, so as to obtain a batch of approximately
500 g of green coffee beans (11% moisture) that was used for both
chemical and sensorial analyses. The samples were frozen at
80 °C in plastic flasks pending their use.

Plots Harvest day Alt Long Lat Tmin Taver Tmax R SR PET Tferm
1 04–2006 150 55 23 38 21 16 00 21.8 25.3 28.6 3.8 1728 4.6 24.0
2 04–2006 270 55 40 39 21 03 16 22.5 23.9 26.4 9.0 1783 2.1 25.3
3 07–2006 315 55 38 39 21 20 08 16.3 19.3 24.3 4.7 1284 2.5 19.6
4 08–2006 463 55 21 08 20 57 36 18.4 21.2 25.2 1.5 1488 2.5 20.3
5 05–2006 500 55 21 08 20 57 36 18.0 22.0 26.6 2.7 1150 2.9 21.6
6 08–2006 549 55 31 32 21 17 38 15.8 19.5 24.1 3.8 1374 3.4 19.7
7 07–2006 585 55 16 48 21 05 18 14.6 18.9 23.8 3.0 1043 2.2 18.7
8 09–2006 661 55 36 23 21 19 36 12.6 16.2 20.9 5.1 1199 2.4 21.3
9 09–2006 757 55 32 52 21 17 29 13.7 17.0 21.1 4.2 1148 2.6 20.3
10 10–2006 790 55 24 02 21 13 17 10.8 14.4 19.2 2.7 1164 2.7 23.0
11 08–2006 805 55 18 12 21 07 20 14.0 17.5 22.0 3.0 1031 2.2 19.5
12 09–2006 880 55 30 24 21 03 16 12.3 16.4 21.6 1.4 1270 1.9 18.9
13 09–2006 980 55 21 03 21 00 31 9.6 13.4 17.3 2.6 1023 1.9 20.8
14 10–2006 985 55 18 41 21 07 09 10.3 14.3 19.1 2.9 973 2.2 22.1
15 09–2006 1014 55 33 39 21 16 59 10.7 14.9 20.0 4.2 1133 2.6 19.7
16 12–2006 1032 55 25 02 21 12 35 9.9 14.0 17.7 2.0 1349 3.3 26.4
Cv (%) 28.2 20.2 15.0 51.1 19.0 26.0 10.8
 B. Bertrand et al. / Food Chemistry 135 (2012) 2575–2583 2577

2.4. Extraction of volatile compounds from ground coffee by ables. Principal components analyses (PCA) were applied to the
headspace-SPME means of the sensory, climatic and volatile compound data.

Green coffee bean samples were ground and brought to room

3. Results
temperature for 90 min prior to sampling for headspace analysis.
A CAR/PDMS (Carboxen/polydimethylsiloxane, 75 lm) SPME fibre
3.1. Sensory attributes of coffee beverage quality linked with climatic
(Supelco Co., Bellefonte, PA, USA) was used to extract volatile con-
factors
stituents from the coffee headspace as its affinities for all classes of
aroma compounds found in coffee have been previously docu-
The sensory quality of the samples displayed significant differ-
mented (Akiyama et al., 2003; Bicchi, Iori, Rubiolo, & Sandra,
ences (P < 0.01) between sites for aroma quality, body, acidity, bit-
2002; Roberts, Pollien, & Milo, 2000), and notably for trace com-
terness, fruitiness, green taste, earthy taste and overall appreciation
pounds or low molecular weight compounds. Two grammes of
of the beverage (as tested by one-way ANOVA). Relationships be-
ground coffee were placed in a 2 ml hermetically sealed glass flask,
tween significant quality attributes and climatic factors were ana-
which corresponded to a headspace of 1/3 of the sampling flask.
lysed by correlations (Table 2). Aroma quality, acidity, fruitiness
The flasks were placed for 15 min in a thermostatically regulated
and overall quality were favoured by cool climates, whilst the unde-
oven at temperatures of 50 °C until the sample headspace equilib-
sirable earthy and green tastes were increasingly present as the
rium was reached. The equilibration time and temperature, which
temperature increased. It therefore appears that the quality was
are important factors that influence the quality of volatile com-
weakest under warm climates. A negative correlation between
pound extraction, were fixed in accordance with prior studies
the fruity and earthy attributes was also obtained (r = 0.74). We
undertaken by this laboratory (Gonzalez-Rios et al., 2007). Volatile
showed that, of the climatic factors, temperature variations played
compounds were then extracted by placing the SPME fibre in con-
a paramount role on quality since it was correlated with six (of the
tact with the headspace for 45 min at equilibrium temperature. For
eight) sensory attributes. Solar radiation also played an important
compound desorption, the fibre was placed in the GC injector and
role as three sensory attributes were correlated with this factor.
heated to 250 °C for 4 min. All samples were analysed in duplicate.
However, as temperature and SR were highly correlated, it was dif-
ficult to determine their respective roles. The other climatic vari-
2.5. Combined gas chromatography–mass spectroscopy
ables played a minor role. However, it should be mentioned that
factors such as rainfall and potential evapotranspiration were cor-
The coffee SPME extracts were analysed on a GC–MS apparatus
related with bitterness and green taste, respectively.
(HP-6890A GC connected to an HP-5973N MS) equipped with a
DB-WAX capillary column (60 m  0.32 mm, with a 0.25 lm phase
coating (J&W Scientific)). Injection was performed in splitless 3.2. Volatile compound composition of green coffee beans as a function
mode for 4 min at 250 °C with a Supelco specific SPME insert of of climatic conditions
0.75 mm i.d. The carrier gas (helium) flow rate was 1.5 ml/min.
The temperature ranges in the oven were programmed from 50 Table 3 gives the list of the 44 volatile compounds identified in
to 200 °C at 3 °C/min, followed by a rise from 170 to 250 °C at green coffee beans by HS-SPME/GC–MS. Out of the 44 metabolite
8 °C/min. The electronic impact ionisation method was used with variables analysed, 40 (91%) showed significant differences
an ionisation energy of 70 eV. The mass range scanned was 40 to (P < 0.05) between experimental plots, as tested by one-way ANO-
300 amu at a scanning rate of 5.36 scans/s. The transfer line tem- VA (Table 3). Two compounds, i.e. hexanal and delta valerolactone,
perature was 260 °C. presented P values slightly higher than the significance threshold
(P = 0.058 and 0.072, respectively), while pentanoic and hexanoic
2.6. Identification of volatile compounds acids clearly did not present any significant differences between
environments. Relationships between the volatile compounds
The volatile constituents of the headspace were identified by and environmental factors were further analysed by drawing up
comparing their calculated relative retention indexes with those a correlation matrix (Table 4). Of the 44 volatile compounds tested,
given in the literature, and their mass spectra with those in the 18 were not correlated with the climatic variables.
database (Wiley Mass Spectral Data). The mean air temperature appeared again to be the predomi-
nant factor since it was correlated with 21 volatile compounds (Ta-
2.7. Analysis of sensory characteristics ble 4). The regression slope sometimes differed with respect to
temperature-dependent compounds (Fig. 1). Strikingly, in the alco-
The coffee beverage was prepared by brewing 50 g of roasted hol class, only positive correlations were detected, indicating that
coffee in 1 l of water for 5 min. The cup quality of the coffee sam-
Table 2
ples was assessed twice by 12 expert tasters using 7 sensory crite- Correlation matrix (Pearson’s r linear-correlation coefficients) between significant
ria: aroma (intensity and quality), body, acidity, bitterness, sensory attributes and mean environmental factors as calculated during the last
astringency, sourness, flavour attributes (green, earthy, fruity, 5 months of fruit development. Underlined r values indicate P values < 0.05 and r
medicinal). A hedonic assessment score, i.e. an overall apprecia- values in bold indicate P values < 0.01. Taver, average daily temperatures; Tferm,
mean temperature during wet processing; R, rainfall; SR, solar radiation; PET,
tion, was also assigned to the coffee beverage. Scoring was on a potential evapotranspiration.
scale of 0 to 10, where a score of 0 corresponded to the total ab-
sence of the criterion in the coffee. Sensory attributes Taver Tferm R SR PET
Aroma quality 0.57 0.47 0.10 0.51 0.55
2.8. Statistical analysis Body 0.00 0.36 0.45 0.02 0.20
Acidity 0.51 0.48 0.40 0.45 0.18
Bitterness 0.08 0.48 0.59 0.10 0.18
StatisticaÓ software (Statsoft, France 2004) and the SAS system
Fruity 0.67 0.30 0.22 0.51 0.22
for the Windows SAS System V9.1 were used to perform all statis-
Green 0.61 0.46 0.19 0.66 0.78
tical analyses. Following ANOVA, the sensory and volatile com-
Earthy 0.83 0.45 0.13 0.73 0.47
pound values were compared by the Duncan test (P 6 0.05). Overall quality 0.30 0.31
0.59 0.51 0.54
Pearson’s correlation coefficients (r) were calculated between vari-
2578 B. Bertrand et al. / Food Chemistry 135 (2012) 2575–2583

Table 3
Comparison of sixteen plots maximising variations in elevation (150–1032 m a.s.l.) and climatic conditions for volatile compounds in green coffee by HS-SPME/GC–MS, using one-
way ANOVA. Quantities of volatile compounds measured by means of peak areas. (MS, mass spectrum, LRI linear retention index, SD, standard deviation, F Fisher ratio, P
probability).

Chemical class Identification method Volatile compounds Mean ±SD F P

Alcohol MS, LRI Ethanol 72718591 27883751 195.81 0.00
Alcohol Propan-2-ol 1929713 992162 7.07 0.00
Alcohol MS, LRI Butan-2-ol 1334564 536856 7.59 0.00
Alcohol MS, LRI Pentan-2-ol 1297137 437296 36.73 0.00
Alcohol MS, LRI Butan-1-ol 2011134 722862 5.28 0.00
Alcohol MS, LRI 2 and 3-Methylbutan-1-ol 17543367 8679307 31.39 0.00
Alcohol MS, LRI Pentan-1-ol 3268017 1097557 32.90 0.00
Alcohol MS 2-Butoxyethanol 6536559 1357063 40.17 0.00
Alcohol MS, LRI 2-Ethylhexan-1-ol 3148161 2572345 100.51 0.00
Alcohol MS Butan-2,3-diol 16142168 20020595 110.51 0.00
Alcohol MS Butan-1,3-diol 19472925 20732996 160.50 0.00
Alcohol MS 2-Methylpropan-1-ol 2977529 862771 21.46 0.00
Alcohol MS, LRI Hexan-1-ol 16763938 3585781 26.54 0.00
Alcohol MS Hexan-2-ol 2233515 855749 35.65 0.00
Alcohol MS, LRI Benzyl alcohol 1265110 354225 10.16 0.00
Alcohol MS, LRI 2-Phenylethanol 3411679 1737340 109.48 0.00
Aldehyde MS Ethanal 5308199 929601 8.79 0.00
Aldehyde MS, LRI Hexanal 6962196 2813923 2.25 0.06
Aldehyde MS, LRI Benzaldehyde 4865368 2915233 23.12 0.00
Aliphatic acid MS, LRI Acetic acid 28330099 19048995 61.26 0.00
Aliphatic acid MS, LRI 3-Methylbutanoic acid 59166121 39373387 187.51 0.00
Aliphatic acid MS Pentanoic acid 7765020 1187516 1.18 0.37
Aliphatic acid MS, LRI 3-Methylbut-2-enoic acid 6497552 1278942 3.92 0.00
Aliphatic acid MS Hexanoic acid 10171000 2327168 1.87 0.11
Furan MS 2-Furanmethanol 1709635 678698 13.48 0.00
Furan MS 2-Methylfuran 3611560 1004160 4.03 0.00
Furan MS 3-Methylfuran 3127022 1648102 31.56 0.00
Hydrocarbon MS, LRI Pentane 12888317 5928736 7.90 0.00
Hydrocarbon MS, LRI Toluene 5792483 1098807 5.39 0.00
Hydrocarbon MS, LRI Ethylbenzene 11472895 4196387 16.13 0.00
Hydrocarbon MS Styrene 43698104 9481515 5.17 0.00
Hydrocarbon MS,LRI Methane 1553873 1461154 59.48 0.00
Ketone MS, LRI Butan-2-one 6388182 3233684 56.52 0.00
Ketone MS Butan-2,3-dione 12106753 6350286 10.96 0.00
Ketone MS, LRI 3-Hydroxybutan-2-one 2734257 1260648 14.93 0.00
Ketone MS, LRI Acetophenone 5181692 2201598 19.06 0.00
Ketone MS, LRI Acetone 10411577 9052544 259.10 0.00
Lactone MS Gamma butyrolactone 8251671 3123284 29.14 0.00
Lactone MS Gamma valerolactone 2042472 980511 164.16 0.00
Lactone MS Delta valerolactone 1310595 241594 2.13 0.07
Phenol MS, LRI Phenol 1178913 1376017 19.41 0.00
Pyrazine MS 2-Metoxy-3-(2methylpropyl)pyrazine 3908137 1987826 32.31 0.00
Pyrrole MS Pyrrole 4005671 1563679 2.86 0.02
Sulfide MS Dimethylsulfide 7533249 5011895 47.80 0.00

3.3. Volatile compounds as possible indicators of sensory attributes

these compounds, notably 2-butoxyethanol, and diols such as bu-
tan-2,3-diol and butan-1,3-diol, were positively affected by warm
climatic conditions. Among the aldehydes, benzaldehyde was pos-
itively correlated with temperature while, by contrast, ethanal
accumulated in cool climates. Other compounds were found to
be negatively correlated with temperature, i.e. 2-methylfuran, bu-
tan-2-one and dimethylsulfide. Finally, it should be noted that the
highest correlations (r2 > 0.7) were found for one alcohol (2-phen-
ylethanol), one aliphatic acid (acetic acid) and one lactone (gamma
butyrolactone) that accumulated in warm climates.
Solar radiation, as a temperature co-variable, was found to be less
accurate in explaining the variance in volatile compounds since sig-
nificant correlations were only noted for seven volatile compounds.
Rainfall was linked to four volatile compounds (2-ethylhexan-1-ol,
3-methyl-2-butenoate, methane and gamma valerolactone, respec-
tively), since PET was linked only with butan-2,3-diol and butan-1,
3-diol. Finally, temperature during wet processing was positively
correlated with pyrrole and pentan-2-ol contents.
The volatile compounds were correlated with the sensory data
(Table 5). The majority of compounds that were positively affected
by warm temperatures (namely, 2-butoxyethanol, 2-ethylhexan-1-
ol, butan-2.3-diol, butan-1.3-diol, benzaldehyde, hexan-2-ol, acetic
acid, 3-methylbutanoic acid, benzyl alcohol, 2-phenylethanol bu-
tan-2-one, butan-2,3-dione, and gamma butyrolactone) were also
negatively correlated with aroma quality, acidity, fruity character
or overall quality. Moreover, we found that these compounds (ex-
cept for 3-methyl butanoate) had high positive correlations with
the negative earthy attribute, indicating that these compounds, if
not directly involved in off-flavours were at least valuable indica-
tors of them. From this standpoint, butan-1.3-diol and butan-2.3-
diol showed the highest correlation coefficients with the earthy
attribute, with r2 values of around 0.86 (Fig. 2). Conversely, among
the five compounds that were associated with cool climatic condi-
tions (ethanal, 2-methylfuran, butan-2-one, acetone and dimethyl-
sulfide), butan-2-one was negatively correlated with the earthy
attribute, while ethanal (Fig. 2) was associated with the fruity attri-
bute. Moreover, some volatile compounds that were not clearly
correlated with the climatic parameters recorded, such as ethanol,
2-furanmethanol, hexanal, 2-methylpropan-1-ol, 3-methylfuran
 B. Bertrand et al. / Food Chemistry 135 (2012) 2575–2583 2579

Table 4
Correlation matrix between volatile compounds in green coffee and climatic factors. R = Pearson’s linear-correlation coefficients. Underlined R values indicate P values < 0.05 and
R values in bold indicate P values < 0.01. Taver, average daily temperatures; Tferm, mean temperature during wet processing; R, rainfall; SR, solar radiation; PET, potential
evapotranspiration.

Compounds Taver Tferm R SR PET

Ethanol 0.08 0.32 0.31 0.05 0.31
Propan-2-ol 0.35 0.02 0.23 0.04 0.45
Butan-2-ol 0.48 0.21 0.08 0.28 0.43
Pentan-2-ol 0.01 0.52 0.31 0.17 0.13
Butan-1-ol 0.14 0.09 0.31 0.35 0.17
2 and 3-Methylbutan-1-ol 0.52 0.11 0.47 0.23 0.01
Pentan-1-ol 0.06 0.07 0.07 0.16 0.07
2-Butoxyethanol 0.73 0.06 0.41 0.55 0.02
2-Ethylhexan-1-ol 0.76 0.44 0.80 0.76 0.09
Butan-2,3-diol 0.82 0.50 0.35 0.67 0.63
Butan-1,3-diol 0.84 0.49 0.44 0.68 0.51
2-Methylpropan-1-ol 0.13 0.35 0.11 0.08 0.28
Hexan-1-ol 0.18 0.14 0.33 0.24 0.28
Hexan-2-ol 0.79 0.06 0.43 0.31 0.23
Benzyl alcohol 0.78 0.18 0.22 0.38 0.41
2-Phenylethanol 0.86 0.19 0.35 0.52 0.36
Ethanal 0.62 0.16 0.24 0.52 0.09
Hexanal 0.03 0.27 0.09 0.10 0.34
Benzaldehyde 0.56 0.36 0.15 0.27 0.24
Acetic acid 0.88 0.17 0.44 0.47 0.26
3-Methylbutanoic acid 0.55 0.11 0.50 0.50 0.37
Pentanoic acid 0.35 0.19 0.11 0.04 0.14
3-Methylbut-2-enoic acid 0.35 0.23 0.67 0.26 0.49
Hexanoic acid 0.51 0.18 0.05 0.01 0.02
2-Furanmethanol 0.01 0.48 0.10 0.03 0.41
2-Methylfuran 0.72 0.32 0.12 0.39 0.04
3-Methylfuran 0.20 0.30 0.10 0.13 0.36
Pentane 0.38 0.09 0.16 0.34 0.12
Toluene 0.09 0.34 0.20 0.19 0.17
Ethylbenzene 0.21 0.42 0.34 0.14 0.22
Styrene 0.35 0.03 0.43 0.43 0.04
Methane 0.45 0.44 0.64 0.34 0.24
Butan-2-one 0.78 0.31 0.24 0.77 0.34
Butan-2,3-dione 0.57 0.33 0.50 0.46 0.04
3-Hydroxybutan-2-one 0.36 0.44 0.39 0.35 0.27
Acetophenone 0.66 0.22 0.23 0.50 0.11
Acetone 0.57 0.08 0.19 0.39 0.05
Gamma butyrolactone 0.86 0.01 0.26 0.41 0.42
Gamma valerolactone 0.29 0.32 0.64 0.46 0.26
Delta valerolactone 0.09 0.16 0.11 0.28 0.14
Phenol 0.24 0.08 0.12 0.25 0.27
2-Metoxy-3-(2methylpropyl)pyrazine 0.68 0.28 0.10 0.29 0.16
Pyrrole 0.33 0.58 0.03 0.31 0.48
Dimethylsulfide 0.75 0.14 0.16 0.46 0.33

and toluene, presented close correlations with body, acidity and

bitterness attributes. Finally, pyrrole, which was shown to be cor-
related with temperature during wet processing, was associated
with weak aroma quality, acidity and overall quality.
To draw up a summary representation of our results, the sen-
sory analysis variability structure was analysed by principal
component analysis. In this analysis, the volatile compounds
and climatic variables were considered as supplementary vari-
ables. The graphic representation (Fig. 3) is based on the first
two principal components, which accounted for 84.5% of the
overall variance. As expected, the aromatic quality, fruity charac-
ter, overall quality attributes contrasted with the negative attri-
butes (earthy and green), while acidity contrasted with
bitterness and body. The earthy and green attributes appeared
to be linked to high temperatures, solar radiation and PET. Vol-
atile compounds such as butan-2,3-diol, butan-1,3-diol and 2-
phenylethanol, appeared to be indicators of the earthy attributes,
whereas toluene, 2 methyl-1-propanal and hexanal appeared to
be indicators of body and bitterness. High fermentation temper-
atures were linked with an increase in pyrrole, benzaldehyde
and 2-furanmethanol and contrasted with the fruity character
and overall quality. Finally, acidity, overall quality, fruity charac-
ter and aroma quality were located opposite high temperatures
and solar radiation and close to acetone and ethanal. These latter
volatile compounds could be considered as highland coffee
indicators.
4. Discussion
4.1. Climatic conditions during seed development have a major effect
on beverage quality
Tropical climates are characterised by a reduced seasonal tem-
perature variation, with elevation being the main factor modulat-
ing this physical variable. High quality coffees are produced at
the higher elevations. Our study confirmed that temperature dur-
ing seed development has a major effect on roasted coffee sensory
profiles. Coffees produced in regions with a cool climate (more ele-
2580 B. Bertrand et al. / Food Chemistry 135 (2012) 2575–2583

2E8 8E7

1,8E8
7E7

1,6E8
R = 0,85 6E7
R = 0.82
1,4E8
P < 0,01 P < 0,01

butan-2,3-diol
5E7
1,2E8
Alcohols

1E8 4E7

8E7
3E7

6E7
2E7
4E7

1E7
2E7

0 0
12 14 16 18 20 22 24 26 12 14 16 18 20 22 24 26

Taver Taver

8E6 1E7

7E6 9E6 R = 0.73

R = 0,85 P < 0.01
6E6 P < 0,01 8E6
2-phenylethanol

2-butoxyethanol

5E6 7E6

4E6 6E6

3E6 5E6

2E6 4E6

1E6 3E6
12 14 16 18 20 22 24 26 12 14 16 18 20 22 24 26

Taver Taver
8E7 1,6E7

7E7
R = 0.84 1,4E7
P< 0.01 R = 0,86
6E7
P < 0,01
gama butyrolactone

1,2E7
butan-1,3-diol

5E7

4E7 1E7

3E7
8E6

2E7

6E6
1E7

0 4E6
12 14 16 18 20 22 24 26 12 14 16 18 20 22 24 26

Taver Taver

Fig. 1. Volatils compounds (Retention Index Standard) in function of TAVER, the mean of average daily temperatures (°C). R and P = Pearson’s linear correlation coefficient and
probability of significance. Dotted lines represented the 95% confidence level.

vated) are more acidic, have a better aroma quality and display
fewer flavour defects than those produced in warmer regions (less
elevated). Conversely, coffees grown under the hottest tempera-
ture conditions have lower acidity, lower aromatic quality, as well
as the presence of green and earthy off-flavours.
In our network of coffee plots, we did not find any clear rela-
tionships between rainfall and the quality of the coffee produced.
This was in line with the findings of Da Silva Emerson et al.
(2005), showing that temperature, but not water availability, was
likely the most important factor responsible for differences in the
quality and chemical composition of coffee from three Brazilian re-
gions. Concerning evapotranspiration, we found that high PET val-
 B. Bertrand et al. / Food Chemistry 135 (2012) 2575–2583 2581

Table 5
Correlation matrix (Pearson’s linear-correlation coefficients), between sensory attributes and volatile compounds. R = Pearson’s linear-correlation coefficients. Underlined R
values indicate P values < 0.05 and R values in bold indicate P values < 0.01.

Compounds Aromatic quality Body Acidity Bitterness Frutty Green Earthy Overall quality
Ethanol 0.10 0.59 0.46 0.62 0.34 0.01 0.30 0.33
Propan-2-ol 0.27 0.26 0.00 0.23 0.18 0.41 0.22 0.06
Butan-2-ol 0.25 0.300 0.04 0.18 0.24 0.28 0.31 0.21
Pentan-2-ol 0.07 0.13 0.14 0.08 0.25 0.27 0.25 0.03
Butan-1-ol 0.01 0.19 0.09 0.01 0.06 0.01 0.03 0.07
2 and 3-methylbutan-1-ol 0.31 0.18 0.45 0.24 0.31 0.04 0.46 0.49
Pentan-1-ol 0.05 0.08 0.03 0.06 0.18 0.33 0.04 0.12
2-Butoxyethanol 0.45 0.12 0.43 0.11 0.59 0.30 0.54 0.45
2-Ethylhexan-1-ol 0.47 0.09 0.52 0.29 0.61 0.39 0.72 0.68
Butan-2,3-diol 0.78 0.01 0.60 0.21 0.68 0.69 0.93 0.78
Butan-1,3-diol 0.74 0.07 0.63 0.28 0.66 0.60 0.93 0.77
2-Methylpropan-1-ol 0.33 0.54 0.63 0.67 0.37 0.23 0.35 0.58
Hexan-1-ol 0.09 0.20 0.03 0.06 0.25 0.12 0.06 0.13
Hexan-2-ol 0.34 0.11 0.40 0.09 0.52 0.39 0.53 0.37
Benzyl alcohol 0.52 0.11 0.55 0.25 0.54 0.36 0.68 0.56
2-Phenylethanol 0.66 0.04 0.58 0.23 0.62 0.37 0.80 0.67
Ethanal 0.28 0.19 0.37 0.31 0.68 0.36 0.51 0.51
Hexanal 0.22 0.44 0.57 0.59 0.26 0.27 0.25 0.48
Benzaldehyde 0.61 0.42 0.73 0.48 0.49 0.18 0.58 0.66
Acetic acid 0.51 0.01 0.48 0.17 0.57 0.36 0.76 0.52
3-Methylbutanoic acid 0.25 0.60 0.12 0.39 0.29 0.51 0.49 0.23
Pentanoic acid 0.04 0.14 0.01 0.21 0.29 0.15 0.05 0.02
3-Methylbut-2-enoic acid 0.37 0.31 0.02 0.35 0.00 0.46 0.37 0.11
Hexanoic acid 0.31 0.13 0.18 0.08 0.22 0.06 0.40 0.15
2-Furanmethanol 0.43 0.32 0.56 0.38 0.41 0.35 0.43 0.38
2-Methylfuran 0.30 0.18 0.20 0.19 0.42 0.18 0.39 0.21
3-Methylfuran 0.03 0.55 0.33 0.59 0.12 0.20 0.20 0.14
Pentane 0.22 0.13 0.16 0.20 0.41 0.20 0.28 0.35
Toluene 0.05 0.72 0.46 0.58 0.09 0.05 0.16 0.11
Ethylbenzene 0.18 0.40 0.31 0.45 0.03 0.05 0.01 0.15
Styrene 0.05 0.19 0.16 0.26 0.02 0.17 0.19 0.04
Methane 0.30 0.36 0.56 0.60 0.52 0.00 0.48 0.62
Butan-2-one 0.45 0.04 0.34 0.06 0.48 0.45 0.68 0.38
Butan-2,3-dione 0.26 0.49 0.65 0.51 0.35 0.06 0.58 0.46
3-Hydroxybutan-2-one 0.23 0.01 0.08 0.05 0.08 0.26 0.42 0.25
Acetophenone 0.07 0.01 0.25 0.07 0.39 0.34 0.32 0.09
Acetone 0.34 0.00 0.36 0.21 0.58 0.21 0.45 0.38
Gamma butyrolactone 0.64 0.04 0.45 0.03 0.61 0.49 0.71 0.51
Gamma valerolactone 0.23 0.15 0.10 0.04 0.21 0.03 0.25 0.25
Delta valerolactone 0.18 0.01 0.22 0.05 0.14 0.07 0.01 0.24
Phenol 0.10 0.23 0.16 0.25 0.24 0.02 0.04 0.12
2-Metoxy-3-(((2methylpropyl)pyrazine 0.27 0.01 0.32 0.02 0.34 0.08 0.32 0.28
Pyrrole 0.60 0.37 0.62 0.39 0.37 0.37 0.50 0.60
Dimethylsulfide 0.17 0.20 0.05 0.19 0.31 0.39 0.43 0.13

ues corresponded to the presence of a green taste. This negative

attribute is often attributed to incomplete ripening.
4.2. Temperatures during seed development affect volatile compounds
In this report, we detected 44 volatile compounds that could be
measured accurately, which is within the range generally expected
with the extraction technique we used (Akiyama et al., 2003;
Czerny & Grosch, 2000; Yeretzian, Jordan, & Badoud, 2002). Of
the 44 extracted volatile compounds, all except 2-butoxyethanol
had already been listed by other authors working on green coffee
(Akiyama et al., 2003; Flament, 2001; Gonzalez-Rios et al., 2007).
To our knowledge, this is the first time a study has focused on cor-
relating climatic factors and volatile compound content in green
coffee. The hypothesis is that volatile compounds in green coffee
represent a signature of the biochemical composition of coffee
beans and its development in the last weeks prior to ripening. In
previous studies (Villarreal et al., 2009; Joët et al., 2010), we al-
ready showed that variations in coffee bean aroma precursor com-
position were temperature-dependent.
In this study, we showed that around half of the volatile com-
pounds detected seemed to be linked with climatic factors experi-
enced during berry development. Our study revealed that the
volatile compound composition varies with temperature during
seed development. We showed, in particular, that warm climates
favour the accumulation of volatile compounds associated with
the appearance of off-flavours (earthy, green). We therefore
showed that the temperature conditions prevailing during fruit
development have a major effect on beverage quality. The enzymes
and genes that controlled most of the volatile compounds corre-
lated with climatic conditions have barely been studied in coffee
seeds until now. However, based on the knowledge acquired on
model plants (Goff & Klee., 2006; Schwab et al., 2008), as well as
on the increasing genomic resources dedicated to coffee such as
the coffee genome sequencing initiative (Wincker P. et al., 2011)
and oligonucleotide arrays for transcriptome survey (Privat et al.,
2011), rapid progresses are awaited in the near future.
In our study of 21 green coffee volatile compounds that were
correlated with climatic conditions, 18 would remain after roasting
2582 B. Bertrand et al. / Food Chemistry 135 (2012) 2575–2583

8E7
1
7E7

R = 0,93 body
bitterness
6E7
P< 0,01
toluene
butan-1,3-diol

5E7 2-methylpropan-1-ol
hexanal Aromatic quality

4E7 butan-2,3-dione
methane

PC 2 : 27.07%
Fruity
acetone
Overall quality
benzaldehyde ethanal
3E7 Tferm pyrrole
2-furanmethanol
0
2E7
2-metoxy-3-(2methylpropyl)pyrazine
@ 2-butoxyethanol R
hexan-2-ol
2-phenylethanol acetic acid acidity
1E7
Earthy
butan-1,3-diol
SR ethanol
Taver gamma butyrolactone
0 butan-2,3-diol 3-methylfuran
0,0 0,2 0,4 0,6 0,8 1,0 1,2 1,4 1,6
PET
Earthy Green
3-methyl-butanoate

8E7

-1
7E7

-1 0 1
R = 0.92
6E7
P < 0.01 PC 1 : 57.43%
butan-2,3-diol

5E7 Fig. 3. Scatterplot of a principal components analysis based on sensorial analysis

findings. The PCA resulted from the sensorial analysis of beverages produced from
4E7 green coffee samples harvested on 16 plots representative of the usual full range of
variations in elevation (150–1032 m a.s.l.) and climatic conditions on coffee plots.
The first two components which accounted for 84.5% of total variability are
3E7
represented. Sensory criteria (active variables) are represented in blue. Volatile
compounds and climatic variables are supplementary variables (respectively in
2E7
black and red). Taver, average daily temperatures; Tferm, mean temperature during
wet processing; R, rainfall; SR, solar radiation; PET, potential evapotranspiration.
1E7 The @ symbol represents 2-ethylhexan-1-ol and benzyl alcohol with similar
patterns.
0
0,0 0,2 0,4 0,6 0,8 1,0 1,2 1,4 1,6

Earthy
butan-1,3-diol). In our study, they seemed to be associated with a
7,5E6 decrease in aromatic quality, acidity, fruity character and an in-
crease in green and earthy flavours. Butan-2,3-diol develops a but-
7E6 R=0,68 tery odour that is considered unpleasant and classified as an off-
P<0,01 flavour (Gonzalez-Rios et al., 2007). This compound is known to
6,5E6 play a role in aroma development during fermentation. It has been
detected in stinking beans (Vincent, Barel, & Challot, 1976). We
6E6
hypothesise that high temperatures during fruit development in-
Ethanal

duce accumulation of this volatile compound in green coffee beans

5,5E6
and that high concentrations give rise to off-flavours that are read-
ily detected by an earthy or green taste in the sensory analysis.
5E6
Conversely, we showed that ethanal was positively correlated
4,5E6
with the fruity attribute. Ethanal is the main aliphatic aldehyde
in roasted coffee, where it is formed by a thermal reaction between
4E6 sugars and certain amino acids. It is also present in green coffee,
where its concentration was found to be greater when tempera-
3,5E6 tures were cool and it was detected in coffees scored as fruity. In
0,0 0,2 0,4 0,6 0,8 1,0 1,2 1,4 1,6 1,8 2,0 2,2 2,4 2,6 2,8
this study, ethanal could be considered as an indicator of the sen-
Frutty sory quality of coffee. The summary representation of the data
clearly showed that when sensorial variables were linked with cer-
Fig. 2. Three volatils compounds (Retention Index Standard) in function of sensory
attributes (scoring 0–10 where score = 0 corresponded to the total absence of the tain volatile compounds, the presence of these compounds never
criterion in the coffee). R and P = Pearson’s linear correlation coefficient and completely accounted for these attributes, except with respect to
probability of significance. Dotted lines represented the 95% confidence level. the earthy odour, which seemed to be very closely correlated with
Butan-1,3-diol and butan-2,3-diol were associated with earthy flavours as ethanal the presence of butan-2,3-diol and butan-1,3-diol.
seem positively associated with the fruity attribute.

5. Conclusion
according to Flament (2001) and Gonzalez-Rios et al. (2007).
Amongst the volatile compounds that were closely correlated with Specialty coffees are made from exceptional beans grown in
a quality attribute, there were two alcohols (butan-2,3-diol and ideal coffee-producing climates. Highland coffees from mountain
 B. Bertrand et al. / Food Chemistry 135 (2012) 2575–2583
regions have the best sensory quality reputation. They tend to fea-
ture distinctive flavours, notably acidity and pleasant aromas
(Rinehart, 2009). In order to meet the demands of the huge and
rapidly growing specialty coffee market, over 20 million farmers
from 35 different countries are striving to produce increasing vol-
umes of specialty coffee. However, the supply of quality coffee is
reaching its upper limit as already noted through the high differen-
tials and limited availability for washed Arabicas, including Ken-
yan, Colombian and Guatemalan coffees in 2009 and the high
prices for all mild Arabicas in 2010 and 2011 (Timothy Schilling,
Borlaug Institute, personal communication). Our results indicated
that environmental temperature during seed development nega-
tively influenced the sensory profile. Recent studies have shown
that even minimal increases in mean temperatures due to climate
change will have disastrous consequences on coffee production. In
some regions, climate change will considerably reduce the area
presently suitable for coffee production (Laderach, M. Lundy, R.
Julián, & K. Schepp., 2009). Our study suggests that climate change,
which in mountainous tropical areas leads to a marked increase in
average temperatures, will also very likely have a substantial neg-
ative impact on the quality of highland coffees.
References
Agresti, P. D. C. M., Franca, A. S., Oliveira, L. S., & Augusti, R. (2008). Discrimination
between defective and non-defective Brazilian coffee beans by their volatile
profile. Food Chemistry, 106, 787–796.
Akiyama, M., Murakami, K., Ohtani, N., Iwatsuki, K., Sotoyama, K., Wada, A., et al.
(2003). Analysis of volatile compounds releases during the grinding of roasted
coffee beans using solid-phase microextraction. Journal of Agricultural and Food
Chemistry, 51(7), 1961–1969.
Avelino, J., Barboza, B., Araya, J. C., Fonseca, C., Davrieux, F., Guyot, B., et al. (2005).
Effects of slope exposure, altitude and yield on coffee quality in two altitude
terroirs of Costa Rica, Orosi and Santa Maria de Dota. Journal of the Science of
Food and Agriculture, 85(11), 1869–1876.
Bertrand, B., Vaast, P., Alpizar, E., Etienne, H., Davrieux, F., & Charmetant, P. (2006).
Comparison of bean biochemical composition and beverage quality of arabica
hybrids involving Sudanese-Ethiopian origins with traditional varieties at
various elevations in Central America. Tree physiology, 26, 1239–1248.
Bicchi, C., Iori, C., Rubiolo, P., & Sandra, P. (2002). Headspace sorptive extraction
(HSSE), stir bar sorptive extraction (SBSE) and solid phase microextraction
(SPME) applied to the analysis of roasted Arabica coffee and coffee brew. Journal
of Agricultural and Food Chemistry, 50, 449–459.
Buffo, R., & Cardelli-Freire, C. (2004). Coffee flavour: An overview. Flavour Fragrance
Journal., 2004(19), 99–104.
Cantergiani, E., Brevard, H., Krebs, Y., Feria-Morales, A., Amado, R., & Yeretzian, C.
(2001). Characterisation of the aroma of green Mexican coffee and identification
of mouldy/earthy defect. European Food Research and Technology, 212, 648–657.
Carli, P., Arima, S., Fogliano, V., Tardella, L., Frusciante, L., & Ercolano, M. R. (2009).
Use of network analysis to capture key traits affecting tomato organoleptic
quality. Journal of Experimental Botany, 60(12), 3379–3386.
Czerny, M., & Grosch, W. (2000). Potent odorants of raw Arabica coffee. Their
changes during roasting. Journal of Agricultural and Food Chemistry, 48(3),
868–872.
Da Silva Emerson, A., Mazzafera, P., Brunini, O., Sakai, E., Arruda, F. B., & Mattoso, L.
H. C. (2005). The influence of water management and environmental conditions
on the chemical composition and beverage quality of coffee beans. Brazillion
Journal of Plant Physiology, 17(2), 229–238.
2583
Decazy, F., Avelino, J., Guyot, B., Perriot, J. J., Pineda, C., & Cilas, C. (2003). Quality of
different Honduran coffees in relation to several environments. Journal of Food
Science, 68, 2356–2361.
Flament, I. (2001). Coffee flavour chemistry. Chichester, UK: John Wiley and Sons.
Giovannucci, D., Liu, P., & Byers, A. (2008). Adding value: Certified coffee trade in
North America. In Pascal Liu (Ed.), Value-adding Standards in the North American
Food Market – Trade opportunities in certified products for developing countries.
Rome: FAO.
Goff, S. A., & Klee, H. J. (2006). Plant volatile compounds: Sensory cues for Health
and Nutritional Value? Science, 311, 815–819.
Gonzalez-Rios, O., Suarez-Quiroz, M., Boulanger, R., Barel, M., Guyot, B., Guiraud, J.-
P., et al. (2007). Impact of post-harvest processing on the volatile fraction of
coffee beans: I. Green coffee. Journal of Food Composition and Analysis, 311.
Grosch, W. (2001). Chemistry III: volatile compounds. In R. J. Clarke & O. G.
Vitzthum (Eds.), Coffee recent developments (pp. 68–89). London, UK: Blackwell
Science.
Guyot, B., Gueule, D., Manez, J. C., Perriot, J. J., Giron, J., & Villain, L. (1996). Influence
de l’altitude et de l’ombrage sur la qualité des cafés Arabica. Plantations,
Recherche, Développement, 3, 272–280.
Joët, T., Laffargue, A., Descroix, F., Doulbeau, S., Bertrand, B., De Kochko, A., et al.
(2010). Influence of environmental factors, wet processing and their
interactions on the biochemical composition of green Arabica coffee beans.
Food chemistry, 118(3), 693–701 [20091002]. http://dx.doi.org/10.1016/
j.foodchem.2009.05.048..
Laderach, P., M. Lundy, A. Jarvis, R. Julián, E. Pérez Portilla and K. Schepp.2009.
Predicted impact of climate change on coffee-supply chains. Climate change
CIAT conference: pp. 20 http://www.uci.ac.cr/descargas/conferencias/
Predicted-impact-of-climate-change-on-coffee-supply-chains.pdf.
Mayer, F., Czerny, M., & Grosch, W. (1999). Influence of provenance and roast degree
on the composition of potent odorants in Arabica coffees. European Food
Research and Technology, 209, 242–250.
Mitchell, H. W. (1988). Cultivation and Harvesting of the Arabica coffee tree. In R. J.
Clarke & R. Macrae (Eds.), Coffee, Agronomy (pp. 334). London, UK: Elsevier
Aplied Science.
Privat, I., Bardil, A., Gomez, A. B., Severac, D., Dantec, C., Fuentes, I., et al. (2011). The
‘PUCE CAFE’ project: The first 15K coffee microarray, a new tool for discovering
candidate genes correlated to agronomic and quality traits. BMC Genomics,
12(5). Available at: http://dx.doi.org/10.1186/1471-2164-12-5.
Rinehart, R., 2009. What is Specialty Coffee? Article published by the Specialty
Coffee Association of America (SCAA) – June 2009. Available at: http://
www.scaa.org/?page=RicArtp1.
Roberts, D. D., Pollien, P., & Milo, C. (2000). Solid-phase microextraction method
development for headspace analysis of volatile flavor compounds. Journal of
Agricultural and Food Chemistry, 48, 2430–2437.
Spadone, J.-C., Takeoka, G., & Liardon, R. (1990). Analytical investigation of Rio off-
flavor in green coffee. Journal of Agricultural and Food Chemistry, 38(1), 226–233.
Schwab, W., Davidovich-Rikanati, R., & Lewinsohn, E. (2008). Biosynthesis of plant-
derived flavor compounds. The Plant Journal, 54, 712–732.
Toci, A. T., & Farah, A. (2008). Volatile compounds as potential defective coffee
seeds’ markers. Food Chemistry, 108, 1133–1141.
Villarreal, D., Laffargue, A., Posada, H., Bertrand, B., Lashermes, P., & Dussert, S.
(2009). Genotypic and environmental effects on coffee (Coffea arabica L.) bean
fatty acid profile: Impact on variety and origin chemometric determination.
Journal of agricultural and food chemistry, 57(23), 11321–11327.
Vincent, J. C., Barel, M., & Challot, F. (1976). Contribution à l’étude des fèves
défectueuses du café vert. 7th International Colloquium on the Chemistry of
Coffee (Hambourg, 9-14.6.1975) ASIC 1976, 133–140.
Wincker, P., Albert, A., Andrade, A, Argout, X., Bertrand, B., Kochko, A, Giuliano, G.,
Graziosi, G., Henry, R., Jaramaya, J., Lashermes, P., Ming, R., Nagai, C., Rounsley,
S., & Sankoff, D. (2011). Sequencing the Coffee Genome. Plant & Animal
Genomes XIX Conference January (pp. 15–19), San Diego, CA.
Yeretzian, C., Jordan, A., & Badoud, R. (2002). From the green bean to the cup of
coffee: investigating coffee roasting by on-line monitoring of volatiles. European
Food Research and Technology, 214, 92–104.