Composition and properties of Indonesian palm civet coffee
(Kopi Luwak) and Ethiopian civet coffee
Massimo F. Marcone
*
Department of Food Science, Ontario Agricultural College, Guelph, Ont., Canada N1G 2W1
Received 19 May 2004; accepted 25 May 2004
Abstract
This research paper reports on the findings of the first scientific investigation into the various physicochemical properties of the
palm civet (Kopi Luwak coffee bean) from Indonesia and their comparison to the first African civet coffee beans collected in
Ethiopia in eastern Africa. Examination of the palm civet (Kopi Luwak) and African civet coffee beans indicate that major physical
differences exist between them especially with regards to their overall color. All civet coffee beans appear to possess a higher level of
red color hue and being overall darker in color than their control counterparts. Scanning electron microscopy revealed that all civet
coffee beans possessed surface micro-pitting (as viewed at 10,000ffmagnification) caused by the action of gastric juices and digestive
enzymes during digestion. Large deformation mechanical rheology testing revealed that civet coffee beans were in fact harder and
more brittle in nature than their control counterparts indicating that gestive juices were entering into the beans and modifying the
micro-structural properties of these beans. SDS–PAGE also supported this observation by revealing that proteolytic enzymes were
penetrating into all the civet beans and causing substantial breakdown of storage proteins. Differences were noted in the types of
subunits which were most susceptible to proteolysis between civet types and therefore lead to differences in maillard browning
products and therefore flavor and aroma profiles. This was confirmed by electronic nose analysis which revealed differences between
the palm civet coffee (Kopi Luwak) and African civet coffee aroma profiles. Analytical techniques for the authentification of palm
civet (Kopi Luwak) and African civet coffee are also explored. It would appear that SDS–PAGE may serve as the most reasonable
and reliable test to help confirm the authenticity of civet coffee. Electronic nose data was able to distinguish both civet coffees from
their control counterparts and further indicated that processing through the civets gastro-intestinal track substantially modified
these coffees.
ff2004 Elsevier Ltd. All rights reserved.
Keywords:Kopi Luwak; African civet; Comparison; Composition
1. Introduction
Coffee is grown in over 80 countries around the world
which lie within 1000 miles north and south of the
equator. Of the many varieties grown world-wide a few
varieties have achieved a special reputation and notoriety
based upon their rarity and overall flavor. Of these Ja-
maican Blue Mountain and Tanzanian Peaberry are the
most notable and as such command a premium price.
Although these coffees are in short supply, no coffee is
perhaps in shorter supply and has a more distinct flavor
and history than a coffee called Kopi Luwak from In-
donesia. With an annual production of under 500 pounds
and a price tag of 600 dollars (Canadian) per pound, it
commands the undisputed reputation of being the rarest
and most expensive coffee or beverage in the world.
Although Kopi Luwak (the Indonesian words for
coffee and civet) comes from the Indonesian islands of
Java, Sumatra and Sulawesi, it is not its exotic location
of origins but rather its unusual and quite unexpected
method of production which contribute to its mystique
and price. The desire to consume unique food products
is a characteristic of passionate coffee drinkers. To this
end, an unique coffee emerged from the jungles of In-
donesia and became know as Kopi Luwak in the West.
This is indeed a rare and unique coffee as it is processed
in the digestive system of the indigenous palm civet
*
Tel.: +1-519-824-4120; fax: +1-519-824-6631.
E-mail address:[email protected](M.F. Marcone).
0963-9969/$ - see front matterff2004 Elsevier Ltd. All rights reserved.
doi:10.1016/j.foodres.2004.05.008
Food Research International 37 (2004) 901–912
www.elsevier.com/locate/foodres

(Paradoxurus hermaphroditus). This 3–10 pound animal
is an expert tree climber and lives in the trees. During
the night it uses its eyesight and smell to seek out and eat
only the ripest, reddish coffee cherries. The coffee cherry
fruit is sweet and is completely digested by the Luwak
but the beans are excreted in their feces. This internal
fermentation and action by different digestive enzymes
add a unique flavor to the beans which has been de-
scribed as earthy, musty, syrupy, smooth, and rich with
both jungle and chocolate undertones. It is believed that
the Kopi Luwak coffee has its origins in the Dutch coffee
plantation estates (approx. 200 years ago) before Indo-
nesian achieved its independence in 1945. It is believed
that the local people could bring in these unusual beans
and, in return, receive a small payment from the estate
managers (Schoenholf, 1999).
In the West, the knowledge about the existence of this
coffee was not well known until the March 1981 issue of
National Geographic which mentioned this coffee by
name in one of their feature articles entitled ‘‘The Bo-
nanza Bean – Coffee’’ (Starbird, 1981).
Curiously, Kopi Luwak is not the first nor only hu-
man food produced by its passage through or part of the
digestive tract of an animal and used for human con-
sumption.
The world’s first sweetener, honey, is a product pro-
duced from a combination of flower nectar and pollen
and bee ‘excretions’. The bee transports its nectar to its
hive to make it into honey by first swallowing it. They
then regurgitate it and mix it with other substances, re-
moving a majority of the liquid from it and it becomes
honey which they use to feed their young and each other.
Another is Bird’s nest soup often referred to as the
‘‘Caviar from the east’’, is produced from the saliva of a
swiftlet (a small bird native to Southeast Asia). It is
consumed in many parts of the world with the United
States being one of its biggest importers. This tiny bird
makes its nest from strands of its own gummy saliva
which hardens when exposed to air.
Argan oil is produced by goats in Morocco who are
encouraged by their herders to climb acacia-like trees
and eat the fruit which is similar to olives. The locals
remove the pits from the feces and grind them to extract
their oil. This unusual and expensive oil is used in mas-
sage and cooking, and is repudiated to be an aphrodisiac.
Presently, the main island producer of Kopi Luwak is
Sulawesi (Indonesia) but due to a local war, few people
venture into the forest to collect this rare, unique coffee for
fear of being killed by the warring parties. As a result, the
coffee supply has virtually dried up and it is extremely
difficult to locate authentic sources of Kopi Luwak coffee.
The desire to posses and drink unique coffee has not
ceased so other sources of animal ‘processed’ coffee was
explored. The search began by determining the coffee
growing areas of the world. Researching the cultivation
areas world of coffee, it was determined that coffee only
grows within 1000 miles north and south of the equator.
The next criteria was to find possible animals within this
region who could possibly eat coffee cherries and, again
through some research, the African civet (Civettictis
civetta) was the best possible candidate. This 15–45
pound animal lives in sub-Saharan Africa, i.e., from
Senegal to Somalia and south to Nambia and eastern
South African. Unlike the Indonesian civet, this animal
lives mainly on the ground in forests, savannah, and
areas with long grasses and thickets. It usually lives near
permanent water systems and is repudiated to be a good
swimmer. It is noctural and is an omnivore – it eats fruit,
carrion, rodents, insects, eggs, reptiles, birds and vege-
tation. It is known that this civet deposits its feces in
special piles called civetries which are located near the
edge of its territory. The Ethiopian people also capture
this animal and harvest the musk from the perineal
glands which are located just under the tail. This musk
has traditionally been used in the perfume industry but
its use has been greatly reduced due to the influence of
people who are concerned about the animals’ welfare.
The third hurdle was to determine which of the coun-
tries within the geographical area of the civet was po-
litically stable. The choice was Ethiopia, the birthplace
of coffee which is known world-wide for its long history
of growing quality Arabica coffee. With this determi-
nation, a research expedition into Ethiopia in December
2003 to find and study this civet, was undertaken to
hopefully determine scientifically if this cat-like animal
eats coffee cherries. This was indeed established, without
a doubt, that this animal does indeed eat the coffee
cherries and excretes the ‘processed’ beans in their feces.
Extensive documentation of the findings and samples
from various locations in Ethiopia were taken back to
Canada for authentication and analysis.
The objectives of this study were first to determine the
physico-chemical properties of Kopi Luwak and to de-
termine and compare them to those of the African civet
to determine if the former may one day serve as a sub-
stitute for Kopi Luwak.
2. Materials and methods
2.1. Materials
Kopi Luwak and control beans (not having gone
through the palm civet) were obtained from Holland
Coffee California Inc. (Novato, CA).
Both the Kopi Luwak and control coffee beans were
of theCoffea robusta(canephora variety) and originated
in Aceh Province in Sumatra, Indonesia during the 2002
harvest season.
African civet coffee and its respective control beans
(not having gone through the African civet) were
collected from both the regions around Abdelah and
902 M.F. Marcone / Food Research International 37 (2004) 901–912

Nekemte in western Ethiopia during the period of
December 3–21, 2003.
All the beans were dried to 11% moisture before
experiments were begun.
2.2. Methods
2.2.1. Colorimetric determination
The color (L*,a*, andb* values) of the raw coffee
beans was determined using a Minolta Chroma Meter
CR-200b (Minolta Camera Co, Ltd., Osaka, Japan).
2.2.2. Scanning electron microscopy
Beans were mounted on aluminum stubs and coated
with gold/palladium (60/40) to a thickness of 25–30 nm
using an Anatech Hummer VII sputter coater (Alexan-
dria, VA). Following coating, beans were viewed under
a Hitachi S-570 scanning electron microscope with a
high voltage setting of 10–20 kV.
2.2.3. Proximate analysis of beans and elemental (P, K,
Mg, and Ca) analysis
Proximate analysis was performed as prescribed in
the official standard methods of the American Associa-
tion of Cereal Chemists, Inc. (American Association of
Cereal Chemists, 1983) after beans were ground using a
coffee grinder (Black & Decker, Toronto, Ont.). For
elemental analysis, 0.250 g samples of oven-dried bean
material were wet digested and subjected to atomic ab-
sorption analysis as described by Marcone and Yada
(1997) and Marcone (2000).
2.2.4. SDS–PAGE
SDS–PAGE electrophoresis was performed on raw
coffee beans after normalization for differences in pro-
tein content. Approximately 50 mg of ground coffee
bean material was dissolved in 1 ml of SDS–PAGE
Standard derivitization and procedure run according to
the method outlined by Marcone and Yada (1997).
2.2.5. Microbial tests
0.5 g of ground coffee bean (unroasted and roasted)
were diluted into 4.5 ml of 0.1% sterile peptone water.
10-fold dilutions were made in peptone water and 0.1 ml
spread plated onto Tryptic Soy Agar (TSA) and Violet
Red Bile Agar (VRB). 1.0 ml of each dilution was also
plated on 3 MEscherichia colicoliform petrifilms. Plates
and films were incubated at 37ffiC for 24 h prior to
enumeration.
2.2.6. Roasting of coffee beans
All green coffee beans (control and civet) were roas-
ted in a Probat Twin Roaster (Probat Burns Inc.,
Memphis, TN) under identical roasting conditions to
achieve a medium colored roast. Roasting times to the
onset of the first and second cracks for each roast were
kept constant at 8 and 11 min, respectively followed by
forced air cooling for 1 min. Each type of roasted bean
resulted in beans that were uniform in color as measured
by a Minolta colorimeter.
2.2.7. Cupping of coffee
All roasted beans (controls and civet coffee) were
ground to a standard cupping grind of 800lm essential
as described by Lyman, Benck, Dell, Merle, and
Murray-Wijelath (2003). Coffee was brewed by adding
six ounces of hot distilled water (98ffiC) to exactly 8 g of
ground coffee into a standard tasting cup/bowl and al-
lowed to steep for 4 min. The crust floating on top of
each cup was then broken and the coffee aroma noted as
per official steep cup methodology. The various roasted
coffee beans were then tasted by a (blinded) certified
cupper and evaluated as per official coffee cupping
procedures.
2.2.8. Electronic nose analysis
Tests were conducted using an Alpha MOS Fox 3000
Electronic nose equipped with 12 metal oxide sensors.
Prepared samples are placed into a sample tray for the
auto-sampler and held at room temperature during
sampling process. Samples were transferred sequentially
to the incubator/heating block and gently agitated at
constant rpm/directional cycle to facilitate headspace
sample production. The headspace sample (1000 or
500lL) was drawn into the syringe and transferred to
the injection port of the electronic nose. Sensor response
data was collected for 120 s followed by a 1080 s delay
before injection of the next sample. The carrier gas (flow
rate maintained at 150 ml/min) is oxygen/nitrogen at
20% (i.e., 19.8–20.2% O
2) and with impurities specified
as H
2O < 5 ppm, CnHm< 5 ppm, O2+N2> 99.95%,
O
2¼20ffi1%. For data analysis the numerical values of
the changes in the sensor resistance are recorded as in-
dividual response patterns for each of the twelve metal
oxide sensors by the computer system operated through
the Alpha MOS AlphaSoft V8.0 software. This data is
used for the multivariate statistical analyses contained in
the program. This allows the evaluation of the presence/
detection of differences within and among sample
groups. The principal component analysis (PCA) sub-
routine in the program was used to generate the maps to
compare the data within and between samples.
2.2.9. Large deformation mechanical rheology
Yield force (F
y) and compression modules (k) mea-
surements were conducted using a Stable Mirco Systesm
Materials Tester model MT-LQ (Surrey, England) fitted
with a 50 kg load cell. The geometry attached to the
load cell was a 38 mm diameter stainless steel com-
pression platen and positioned exactly at the top of the
bean and then lowered at a constant rate of deformation
(10 mm/s). Large deformation test parameters,F
yandk,
M.F. Marcone / Food Research International 37 (2004) 901–912 903

were obtained using the software provided by the
manufacturer.
3. Statistical analysis
Statistical analysis was performed using a SAS Sta-
tistical Analysis System package. Significant differences
among treatments (samples) were determined by Dun-
can’s multiple range test (p60:05) (SAS, 1990).
4. Results and discussion
Physical characterization of the palm civet coffee
beans (otherwise known as Kopi Luwak) collected in the
province of Apec in Indonesia and the two African civet
coffee bean types collected in Abdela and Nekemte in
Western Ethiopia indicate the existence of major dis-
tinguishable differences between them with regards to
their overall size, weight and color.
Preliminary examination of the unroasted beans by
analytical color measurement equipment (Minolta
colourimeter) indicated substantial differences in over-
all color between the three types of civet beans and
their corresponding control beans (control bean being
defined as those not having gone through the civet’s
digestive tract (GI)). (Table 1, Fig. 1). In general,
control beans were more uniformly green in color with
little discernable (quantifiable) color difference between
them (Table 1, Fig. 1). The largest of the civet coffee
beans were from the palm civet (Kopi Luwak) fol-
lowed closely by those from Nekemte with the smallest
beans collected being those from Abdela (Table 1).
Although palm civet (Kopi Luwak) coffee beans were
the largest, they were actually the lowest in weight
compared to Nekemte and Abdela civet beans (with
Nekemte and Abdela civet bean weights agreeing with
their overall size). (Fig. 1). Instron (compression
measurements) on the civet beans as compared to their
control beans indicated that they were substantial
harder and more brittle in nature than their corre-
sponding control beans which were softer and more
elastic in nature (with bean fracturing occurring at
applied perpendicular forces of 200–275 and 80–100 N,
respectively). The observed differences would indicate
that same penetration of gastric juices and or digestive
enzymes were occurring entering into the beans and a
change in the micro structural properties of the civet
beans was occurring during transit through the civet’s
GI track.
The palm civet (Kopi Luwak) coffee beans were
significantly higher in their level of red and yellow hues
rendering them with an overall beige color compared
to the African civet beans, which were greener in color.
Both types of African civet beans were darker in
color (i.e., lowerL* value) than their control beans and
both higher in the reddish color than their respective
controls.
These differences in color would indicate that in the
case of all three civet bean types some form and level of
chemical reaction was occurring at the very surface of
the bean which was rendering them discernibly darker in
color than their corresponding control counterparts.
Although this was true the actual color hue was different
between them with the commonality being that all three
civet bean types were slightly but significantly higher in
the reddish hue tone as well as being significantly darker
in overall color. At this point it can be indicated that the
various digestive biochemicals (gastric juices and pro-
teases) are actually penetrating the outer coffee cherry
(pericarp) after ingestion by the civets and reaching the
actual bean surface where the chemical color change was
occurring. It would be difficult to use color analysis as
the sole means to determine the authenticity of civet
beans as it appears that the magnitude of these differ-
ences differ between palm and African civet beans as
well as within the sub-grouping of the two African civets
studied.
Fecal examination by scanning electron microscopy
of both African civet dung collected from their respec-
tive civetries revealed that similar to its close relative the
palm civet, these animals have similar dietary patterns,
including the consumption of insects, meat (animal) and
vegetable matter thus making them both omnivores
(Fig. 2).
In order to study further the observed changes oc-
curring at the surface of civet beans (indicated earlier)
microscopic examination by scanning electron micros-
copy was conducted. Scanning electron microscopy did
in fact reveal substantial differences between all control
beans and those having passed through the two types of
civets. In the case of all controls, examination of the
beans at 1000ff(Fig. 3) and 5000ffmagnification
(Fig. 4) revealed that their surfaces were substantially
rougher in nature that their civet counterpart. The rel-
ative smoothness of those beans passing through either
the palm civet or African civet could be due to the fact
of exfoliation of the outer surface of the bean during
passage and paralstatis action through the digestive
system of the civet. In a similar way all coffee beans
passing through the civet’s digestive system were ob-
served having surface micro-pitting as observed at
10,000ffmagnification, respectively but not observed in
their control counterparts (Fig. 5). This would indicate
that the acidic substances and proteolytic enzymes
found in the gastric juices of each of the civets were
permeating through the endocarp of the coffee cherry
and reaching and reacting with the actual bean surface
in a similar fashion. This observation together with the
observed changes in color may potentially serve as a
904 M.F. Marcone / Food Research International 37 (2004) 901–912

Table 1
Physical, chemical and microbiological characterization of Kopi Luwak (Indonesian Palm Civet) coffee beans and Nekemte Civet and Abdela Civet
(Ethiopian Civet) coffee beans and respective coffee bean controls
Kopi Luwak
(Indonesian
Palm Civet)
Kopi Luwak
(control)
Nekemte Civet
(Ethiopian Civet)
Nekemte
(control)
Abdela Civet
(Ethiopian Civet)
Abdela
(control)
Proximate analysis (%)
Moisture 9.2
a
11.7
c
10.9
b
12.0
c
11.2
b
13.0
c
Protein 13.5
d
14.5
e
12.1
b
13.2
c
11.4
a
12.7
c
Fat 13.0
c
12.0
a
12.5
b
12.1
a
13.1
c
12.5
b
Ash 3.6
c
3.4
b
3.2
a
3.4
b
3.4
b
3.8
d
Carbohydrate
(by difference)
60.7 58.4 61.3 59.3 63.9 61.0
Seed characteristics
Color
L* 37.62
a
40.00
b
44.66
c
60.12
e
44.94
c
54.3
d
a* 6.23
f
2.78
c
4.10
c
2.21
a
3.0
d
2.57
a
b* 20.41
f
11.93
c
12.92
d
6.31
a
14.00
e
9.00
b
Av. bean wt (g) 1.4
a
1.6
b
1.8
c
2.0
d
1.6
b
1.9
cd
Av. bean (mm) 12
c
12
c
9
b
8
b
7
a
7
a
Large deformation mechanical rheological properties (N)
275
d
100
a
200
b
80
a
240
c
92
a
Minerals (ppm)
Potassium (K) 15,000
a
18,200
d
16,500
b
19,000
e
17,000
c
18,100
d
Phosphorus (P) 20,00
a
24,50
d
22,00
c
25,00
d
21,00
b
24,60
d
Calcium (Ca) 14,50
a
18,00
c
15,00
b
18,00
c
15,50
b
18,00
c
Magnesium (Mg) 14,00
a
17,00
b
14,00
a
17,50
b
14,40
a
17,60
b
Iron (Fe) 12,00
a
15,00
c
12,50
a
14,50
b
13,00
a
15,00
c
Microbial counts (green)
A
APC 8.5 ff10
5b
17ff10
5a
8.0ff10
5b
1.4ff10
5a
7.2ff10
5b
1.1ff10
5a
Enterics 3.2 ff10
2c
2.2ff10
4a
4.0ff10
2d
2.5ff10
4b
4.4ff10
2d
2.9ff10
4b
Coliforms 100
a
1.4ff10
4c
110
a
1.8ff10
4c
1.0ff10
3b
1.4ff10
4c
Mold 5.0 ff10
3c
1.2ff10
3a
4.1ff10
3d
1.9ff10
3d
3.2ff10
3c
2.3ff10
5b
Microbial counts (roasted)
A
APC 5.5 ff10
2a
2.5ff10
2b
1.1ff10
2a
2.0ff10
2a
1.8ff10
2a
2.4ff10
2b
Enterics 50
a
<100
b
<100
b
<100
b
<100
b
<100
Coliforms ND ND ND ND ND ND
Mold ND ND ND ND ND ND
APC, aerobic plate count; Enterics, Enterobacteriaceae count.
Values in each category row with the same letter are not significantly different (pP0:05).
ND, not detected.
A
Colony forming units per gram.
Fig. 1. Photograph of (a) Kopi Luwak coffee beans (b) Nekemte-African Civet coffee beans (c) Abdela-African Civet coffee beans.
M.F. Marcone / Food Research International 37 (2004) 901–912 905

preliminary way of determining the authenticity of civet
coffee beans.
Although it would appear that enzymatic action of
proteolytic enzymes (i.e., pepsin and trypsin, etc.) was
reaching and affecting the surface of the bean more in-
formation was sought in order to determine if these
compounds were actually penetrating and entering into
and modifying the actual bean. If in fact these enzymes
were making their way into these beans, breakdown of
susceptible storage proteins would be expected and ob-
servable. SDS–PAGE in fact confirmed this hypothesis
as civet beans finger prints revealed a loss in intensity of
‘‘protein bands’’ which would be indicative of proteolysis
(Fig. 6). Although this was found to occur, it appeared
that both the acidic (30,000 Da) and basic (20,000 Da)
subunits described by (Marcone, Kakuda, & Yada,
1998), in seeds and beans respectively of the palm civet
(Kopi Luwak) beans were equally susceptible to prote-
olysis whereas only the basic subunit of the African civet
showed substantial breakdown. Since the breakdown of
the proteins would lead to the exposure of more amino
acids (and free amino acids) their reaction through
maillard browning on roasting would lead to the pro-
duction of measurable affects on the aromatic and flavor
characteristics of civet coffee compared to their control
beans. Since the degree and specificity of proteolysis was
observed to be different between the palm civet visa-vi the
African civets (as indicated above) measurable differ-
ences in flavor and aromatic characteristics of the palm
visa-vi African civet could be expected.
Fig. 2. Scanning electron micrographs of (A and a) insect parts found in African Civet dung from Nekemte (B and b) (B) Rodent hair (b) Civet hair
and (C and c) bone material found in African Civet dung from Abdela.
906 M.F. Marcone / Food Research International 37 (2004) 901–912

It is interesting to note that all civet beans were
lower in total protein indicating that during digestion
not only were proteins being partially broken down but
also leached out of the bean. It is interesting to note
that the lower levels of proteins would lead to a de-
crease in the levels of coffee bitterness since proteins
serve as the precursor of certain bitter compounds on
roasting (Mc Camey, Thorrpe, & McCarthy, 1990).
This may help to explain some of the perceived flavor
differences noted by the consumers between Kopi Lu-
wak and its controls. Examination of the other com-
ponents in the various beans (including minerals) did
not reveal any other substantial difference between
them. As such it would appear that little other modi-
fication of the coffee beans occurs as they pass through
the civets’ GI track.
Microbiological testing (aerobic plate counts – APC)
showed that all types of green civet coffee beans were
in fact significantly more contaminated that their re-
spective control beans (Table 1) something that would
be expected. In the civet bean coffee fewer colony
forming types were observed than their respective
control beans. Quite surprisingly the Enteric Organism
counts were on average two log orders lower for the
green civet coffee beans than their controls. In fact,
coliform counts were also shown to be significantly
Fig. 3. Scanning electron micrographs of (A) Kopi Luwak control coffee bean (not gone through animal) (b) Kopi Luwak coffee bean (B) Nekemte
control coffee bean (b) Nekemte African Civet coffee bean (C) Abdela control coffee bean (c) Abdela African Civet coffee bean.
M.F. Marcone / Food Research International 37 (2004) 901–912 907

lower in the civet beans as compare to their respective
control. A possible explanation of this observation
could be that civet coffee beans are typically extensively
washed under running water after collection which
would dislodge bacteria and lower their overall counts.
These results would agree with the above observation
of fewer colony types for civet coffee beans since dif-
ferent bacteria would have different levels of physical
attachment to the beans. Gram stain results indicated
predominately gram positive rods on the surface of the
beans. Other types of microorganism such as molds
were also found on both civet and control beans with
significantly more molds being found on the civet
beans (this could be due to the slow rate of drying after
washing). Upon roasting of all coffee beans reduction
of colony counts occurred to near undetectable levels.
Use of the electronic nose to measure the aroma profile
of roasted ground coffee beans of the three civet coffees
and three control coffees indicate good differentiation
between samples at a discrimination index of 95%
(Table 2). Examination of the distance between clusters
on the PCA map indicated that the control sample
could be differentiated from one another but do not
appear to be very different as a group. On the other
Fig. 4. Scanning electron micrographs of (A) Kopi Luwak control coffee bean (not gone through animal) (b) Kopi Luwak coffee bean (B) Nekemte
control coffee bean (b) Nekemte African Civet coffee bean (C) Abdela control coffee bean (c) Abdela African Civet coffee bean.
908 M.F. Marcone / Food Research International 37 (2004) 901–912

hand Kopi Luwak (Indonesian civet coffee) was the
most different from all coffees with the African civets
(although being differentiable from one another) ap-
pear to be very similar in their overall profiles. It can
be concluded at this point that the electronic nose does
indicate that aroma/flavor profiles of the palm and
African civet coffee beans are affected as they are
processed through the civet’s GI tract but are modified
differently. These results would be in direct agreement
with the differences observed in proteolytic protein
finger prints also mentioned earlier which would lead
to the production of different maillard browning by
products partly responsible for the aroma/flavor pro-
files of the palm and African and civet coffees.
Coffee cupping results by an experience, certified
cupper revealed very little difference in the overall flavor
and aromatic attributes of all three control coffee beans
except that the palm civet control was slightly lower in
body and slightly higher in acidity which correlated well
with the electronic nose data obtained. Cupping results
Fig. 5. Scanning electron micrographs of (A) Kopi Luwak control coffee bean (not gone through animal) (b) Kopi Luwak coffee bean (B) Nekemte
control coffee bean (b) Nekemte African Civet coffee bean (C) Abdela control coffee bean (c) Abdela African Civet coffee bean.
M.F. Marcone / Food Research International 37 (2004) 901–912 909

Table 2
PCA map of electronic nose data of ground roasted coffee tested at 35 ffiC
910 M.F. Marcone / Food Research International 37 (2004) 901–912

also showed that all three civet coffee beans to be dif-
ferentiable from the actual controls which again agreed
well with the electronic nose data. Major differences
were noted in the level of acidity and body (being lower)
for all civet coffee as compared to their controls. Al-
though the palm civet (Kopi Luwak) coffee was found to
be differentiable by the electronic nose procedure from
the two African civet coffees, little difference in the
overall aroma and flavor profile was noted between
them by the certified blinded cupper.
According to several researchers, fermented coffees
(i.e., wet processed) have a better overall quality than
those prepared by dry-processing (Avallone, Brillouet,
Guyot, Olguin, & Guiraud, 2002; Puerta-Quintero,
1999).
In the coffee industry wet processed or fermented
coffees are know to be of superior flavor and as such
command a much higher price than their dry-pro-
cessed coffee counterparts. It is interesting to note that
when the coffee cherries are processed through the
digestive tract of both the palm (Indonesian) and
African civet they do indeed undergo a type of wet
processing due to acidification in the stomach and
then fermentation due to the natural intestinal mi-
croflora. In wet processing of coffee, it is important to
remove all mucilage which could otherwise lead to
secondary fermentation during drying and storage and
ultimately lead to the development of flavor defects
(Avallone et al., 2002; Woelore, 1993). It is interesting
to note that several researchers have found that mu-
cilage degradation seems to be correlated to acidifi-
cation (similar to what the coffee cherries would
experience due to gastric juices in the civets’ stomach).
Although researchers have found that microbial
growth is necessary, it does not directly participate in
mucilage degradation by enzyme production (pectate-
lyase – enzymatic pectolysis) but limits off-flavor de-
velopment due to the production of various organic
acids. It is also interesting to note that research shows
that lactic acid bacteria are preferred in wet processing
systems in order to stay as close as possible to a
natural neutral fermentation. It is interesting to note
that lactic acid bacteria are a major colonizing bac-
teria in the digestive tract of civets. It would be in-
teresting to submit that a possible reason for the
unique and characteristic flavor of civet coffee could
be due to the type of wet process it undergoes in the
GI tract of civets.
In closing, it would appear that color analyses cou-
pled with microscopic surface analysis for the presence
of micro-pitting on civet coffee beans could be poten-
tially used as a method to authenticate the origins (or
process) of civet coffee beans. SDS–PAGE electropho-
resis was also shown to be an excellent tool to determine
the authenticity of civet coffee beans but also having the
added advantage of being able to distinguish between
those having gone through the digestive tract of the
palm civet visa-vi those having gone through the African
civet.
Electronic nose authentification analyses was also
found to be an alternate technique and also being able to
distinguish between those from the palm civet visa-vi
those from the African civet.
References
American Association of Cereal Chemists. (1983).Approved methods of
the American Association of Cereal Chemists(8th ed.). St. Paul,
MN: AACC.
Avallone, S., Brillouet, J. M., Guyot, G., Olguin, E., & Guiraud, J. P.
(2002). Involvement of pectolytic micro-organisms in coffee
fermentation.International Journal of Food Science and Technol-
ogy, 37, 191–198.
Lyman, D. J., Benck, R., Dell, S., Merle, S., & Murray-Wijelath, J. M.
(2003). FTIR-ATR analysis of brewed coffee: Effect of roasting
conditions.Journal of Agricultural and Food Chemistry, 51, 3268–
3272.
Mc Camey, D. A., Thorrpe, T. M., & McCarthy, J. P. (1990). Coffee
bitterness. In Russell L. Rouseff (Ed.),Development in Food
Science: 25.Bitterness in Foods and Beverages(pp. 169–182).
Amsterdam, The Netherlands: Elsevier Science Publishers.
Marcone, M. F., Kakuda, Y., & Yada, R. Y. (1998). Salt-
soluble seed globulins of dicotyledonous and monocotyledon-
ous plants II. structural characterization.Food Chemistry,
63(2), 265–274.
Marcone, M. F., & Yada, R. Y. (1997). Evidence for the phosphor-
ylation and glycosylation of Amaranth 11S Globulin (Amaran-
thin).Food Biochemistry, 21, 341–369.
Marcone, Massimo F. (2000). First report of the characterization of
the threatened plant speciesAmaranthus pumilus(Seabeach
Amaranth).Journal of Agriculture and Food Chemistry, 48, 378–
382.
Fig. 6. SDS–PAGE (reducing) (1llof1mg
1
protein solutions)
applied to Homogenous 20 Phast gels (Pharmacia LKB). (Lane S)
standards [a]-lactalbumin, 14,400 Da; Soybean trypsin inhibitor,
20,100 Da; Carbonic anhydrase, 30,000 Da; ovalbumin, 43000 Da;
bovine serum albumin, 67,000 Da; phosphorylase b, 94,000 Da;
(Lane 1) Kopi Luwak (Lane 2) Kopi Luwak Control (Lane 3) Ne-
kempt African Civet Coffee bean (Lane 4) Nekempt Control (Lane 5)
Abdela African Civet Coffee bean (Lane 6) Abdela Control bean.
M.F. Marcone / Food Research International 37 (2004) 901–912 911

Puerta-Quintero, G. I. (1999). Influencia de proceso de beneficio en la
calidad del cafe.Cenicafffe(Colombia), 50, 78–88.
SAS Institute (1990).SAS user’s guide: Statistics. Statistical Analysis
System: Cary, NC.
Schoenholf, Donald N. (1999). Kopi Luwak: The Stercoaceous
Coffee of Indonesia.Tea and Coffee Trade Journal, 142–
146.
Starbird, Ethel A. (1981). The Bonanza Bean – Coffee.National
Geographic Magazine, 388–404.
Woelore, W.M. (1993). Optimum fermentation protocols for Arabica
coffee under Ethiopian conditions. InAssociation Scientifique
Internationale Du Cafffe(Edited by ASIC), 15
th
International
Scientific Colloquium on Coffee, Montpellier, Paris(pp. 727–733).
Association Scientifique Internationale Du Cafffe (ASIC).
912 M.F. Marcone / Food Research International 37 (2004) 901–912