Food Chemistry 129 (2011) 312318

Contents lists available at ScienceDirect

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

Comparing physicochemical properties of banana pulp and peel ours

prepared from green and ripe fruits
Abbas F.M. Alkarkhi a, Saifullah bin Ramli b, Yeoh Shin Yong b, Azhar Mat Easa b,
a
School of Industrial Technology, Environmental Technology Division, Universiti Sains Malaysia, 11800 Penang, Malaysia
b
School of Industrial Technology, Food Technology Division, Universiti Sains Malaysia, 11800 Penang, Malaysia

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

Article history: Banana pulp and peel our prepared from green and ripe Cavendish banana were assessed for physico-
Received 13 October 2008 chemical properties such as pH, total soluble solids (TSS), water holding capacity (WHC) and oil holding
Received in revised form 22 February 2011 capacity (OHC) at 40, 60 and 80 C, colour values L, a and b, back extrusion force (BEF) and viscosity.
Accepted 22 April 2011
Data obtained were analysed by MANOVA, discriminant analysis and cluster analysis. All statistical anal-
Available online 4 May 2011
yses showed that physicochemical properties of our prepared from pulp and peel, and green and ripe
banana were different from each other. TSS, WHC40, WHC60 and BEF can be used to discriminate
Keywords:
between peel and pulp our, whilst TSS and viscosity can be used to discriminate between our prepared
Cavendish banana
Physicochemical properties
from green and ripe banana.
Banana pulp our 2011 Elsevier Ltd. All rights reserved.
Banana peel our
MANOVA
Cluster analysis
Discriminant analysis

1. Introduction
Banana is one of the most consumed fruits in tropical and sub-
tropical regions. New economical strategy to increase utilisation of
banana includes the production of banana our when the fruit is
unripe, and to incorporate the our into various innovative prod-
ucts such as slowly digestible cookies (Aparicio-Saguilan et al.,
2007), high-bre bread (Juarez-Garcia, Agama-Acevedo, Sayago-
Ayerdi, Rodriguez-Ambriz, & Bello-Perez, 2006) and edible lms
(Rungsinee & Natcharee, 2007). The clear advantage presented by
green banana our includes a high total starch (73.4%), resistant
starch (17.5%) and dietary bre content (14.5%) (Juarez-Garcia
et al., 2006). Due to the high content of these functional ingredi-
ents, regular consumption of green banana our can be expected
to confer benecial health benets for humans (Rodriguez-Ambriz,
Islas-Hernndez, Agama-Acevedo, Tovar, & Bello-Prez, 2008). On
the other hand, ripe banana our that is less known banana prod-
uct, may offer high sugar content to dishes requiring sweetness.
Other than health and sensory reasons, the stages of ripeness are
also important for technical aspect of processing. Banana pulp our
prepared using fruits at different stages of ripening has been
shown to behave differently during manufacture of food products
Corresponding author. Tel.: +60 4 6533888x2222; fax: +60 4 6573678.
E-mail address: azhar@usm.my (A.M. Easa).
such as cakes (Yomeni, Njoukam, & Tchango Tchango, 2004) and
extruded products (Gamlath, 2008). Being able to differentiate be-
tween banana our prepared from different stages of ripeness
could help food processors to control the quality of food incorpo-
rated with banana our.
As the fruits of the banana trees are consumed at green, average
ripe and ripe stages (Emaga, Andrianaivo, Wathelet, Tchango, &
Paquot, 2007), the amount of fruit waste from the peels is expected
to increase with the development of processing industries that
utilise the green and ripe banana. Like its pulp our counterpart,
banana peel our can potentially offer new products with stand-
ardised compositions for various industrial and domestic uses
(Emaga et al., 2007). The peel of banana that represents about
40% of the total weight of fresh banana (Tchobanoglous, Theisen,
& Vigil, 1993) has been underutilised. Various studies have been
conducted to investigate banana peel, and this include the produc-
tion of banana peel our (Ranzani, Sturion, & Bicudo, 1996), the
effects of ripeness stage on the dietary bre components and pec-
tin of banana peels (Emaga, Robert, Ronkart, Wathelet, & Paquot,
2008) and the chemical composition of banana peel, as inuenced
by the maturation stage and varieties of banana (Emaga et al.,
2007).
The potential applications of banana pulp and peel our depend
on their chemical composition (Emaga et al., 2007; Rodriguez-
Ambriz et al., 2008), as well as physicochemical and functional
properties. However, once the fruit is processed into our, identi-

0308-8146/$ - see front matter 2011 Elsevier Ltd. All rights reserved.
doi:10.1016/j.foodchem.2011.04.060
 A.F.M. Alkarkhi et al. / Food Chemistry 129 (2011) 312318
cation of the parts and stages of ripeness of banana used for our
preparation becomes a challenge. As far as we know, no study
has been conducted to compare physicochemical properties of ba-
nana our prepared from pulp or peel of green and ripe banana.
One study that compared the antioxidant compounds in banana
peel and pulp extracts found that the content of antioxidant com-
pounds was higher in the peel than in the pulp (Someya, Yoshiki, &
Okubo, 2002) implying a potentially higher value of the peel in
terms of antioxidant content. Perhaps in the future it may be pos-
sible for technologist to mix and match the pulp and peel our in
order to achieve technofunctional properties without sacricing
the aesthetic values of their products. This development can spur
the utilisation of banana peel as innovative ingredients in various
food and nutraceutical products thus increases the possibility of
banana peel in addition to banana pulp to be processed into our.
The physicochemical properties of the our are expected to vary
with stage of ripeness as it is known that the composition of
banana changes dramatically during ripening. It seems worthwhile
to study the physicochemical data of banana our and devise
methods to discriminate banana pulp and peel our based on its
physicochemical data. Statistical techniques that can be applied
to perform this task include MANOVA, cluster analysis and
discriminant analysis (Markus, Elena, Jacinto, & Carlos, 2002;
Ricardo, Pablo, Elena, Jacinto, & Carlos, 2003; Surez, Rodrlguez,
& Romero, 2007). Therefore the objective of this study was (i) to
compare selected physicochemical properties of pulp and peel
our prepared from green and ripe Cavendish banana, (ii) to use
the data to discriminate between the ours, and (iii) to single out
most appropriate physicochemical methods that differentiate the
ours.
2. Material and methods
2.1. Preparation of banana peel our
Cavendish (Musa acuminate L., cv cavendshii) banana, was
purchased from 12 markets around Penang, Malaysia. A total of
222302 green (stage 1 of ripening: all green) and ripe (stage 6
of ripening: yellow with green tip) banana of each stage of ripeness
were obtained from each market. The fruits were washed and
separated into pulp and peel. The pulps were cut into transverse
slices of about 2 mm thickness. To reduce enzymic browning, pulp
slices and peels were then dipped in 0.5% (w/v) citric acid solution
for 10 min, drained and dried in oven (AFOS Mini Kiln, at 60 C
overnight). The dried pulp slices and peels were ground in a Retsch
Mill Laboratory (Retsch AS200) to pass through 60 and 40 mesh
screen to obtain banana pulp and peel our, respectively. The yield
of our was calculated by dividing the amount of our produced by
the amount of fresh banana used, and the results were converted to
g/kg (g of our/kg of banana). Four types of our produced were
ripe Cavendish pulp (R-pulp), ripe Cavendish peel (R-peel), green
Cavendish pulp (G-pulp) and green Cavendish peel (G-peel). All
ours were stored in airtight plastic packs in cold storage
(15 2 C) for further analyses.
2.2. pH, TSS and viscosity
The pH of the our was measured using a Coming pH metre,
model 10. Flour suspension (8% (w/v)) was stirred for 5 min,
allowed to stand for 30 min, ltered and the pH of ltrate mea-
sured (Suntharalingam & Ravindran, 1993). Total soluble solids
(TSS) in the same our slurries were measured using an Atago
refractometer (Atago PAL-1, Co. Ltd., Tokyo, Japan) (Salvador, Sanz,
& Fiszman, 2007).
313
Viscosity was determined as described by Fagbemi (1999). Flour
was dispersed in water at 8% (w/v) concentration using a magnetic
stirrer (1000 rpm) and heated from 30 to 95 C in a shaking water-
bath (memmert, Gmbh-Germany) and kept at this temperature for
20 min. The slurry obtained was stirred constantly and cooled at
room temperature. The viscosity was measured using a Vibro
Viscometer (SV-10, A & D Japan).
2.3. Water- (WHC) and oil-holding capacity (OHC)
Twenty-ve millilitres of distilled water or commercial olive oil
were added to 1 g of dry sample, stirred and incubated at 40, 60 or
80 C for 1 h. Tubes were centrifuged at 3000g for 20 min, the
supernatant was decanted, and the tubes were allowed to drain
for 10 min at a 45 angle. The residue was weighed and WHC
and OHC calculated as g water or oil per g dry sample, respectively
(Rodriguez-Ambriz et al., 2008).
2.4. Colour
The instrumental measurement of our colour was carried out
with a Colorimeter Minolta CM-3500d (Minolta, Spectrophotome-
ter, USA) and the results were expressed in accordance with the
CIELAB system with reference to illuminant D65 and a visual angle
of 10. The measurements were performed through a 6.4-mm-
diameter diaphragm with an optical glass, placing the our directly
on the glass. The parameters determined were L (L = 0 [black] and
L = 100 [white]), a (a = greenness and +a = redness) and b
(b = blueness and +b = yellowness).
2.5. Back extrusion force of slurry
A TA-XTplus Texture Analyser (Stable Micro Systems, Godal-
ming, UK) was used to evaluate the texture of the our slurry
(8% w/v). Back ward extrusion tests were conducted with the disc
diameter 45 mm, setting the probe travel distance at 30 mm. Both
tests were performed with a test speed of 2 mm/s, a trigger force of
5 g, and force in compression mode. Forcetime curves were
recorded at a crosshead speed of 5 mm/s and recording speed
was 5 mm/s to enable evaluation of back extrusion force (BEF) of
the slurry.
2.6. Statistical analysis
2.6.1. Multivariate analysis of variance (MANOVA)
Multivariate analysis of variance is used where several
dependent variables (p) are measured on each sampling unit
instead of one variable. The objective of MANOVA is to com-
pare the mean vectors of k groups for signicant difference.
Equality of the mean vectors implies that the k means are
equal for each variable, and if two means differ for just one
variable then we conclude that the mean vectors of the k
groups are different.
2.6.2. Discriminant analysis
Discriminant analysis is a multivariate technique used for two
purposes, the rst purpose is description of group separation in
which linear functions of the several variables (discriminant
functions (DFs)) are used to describe or elucidate the differences
between two or more groups and identifying the relative contribu-
tion of all variable to separation of the groups. Second aspect is
prediction or allocation of observations to group in which linear
or quadratic functions of the variable (classication functions
(CFs)) are used to assign an observation to one of the groups (Alvin,
2002; Richard & Dean, 2002).
314 A.F.M. Alkarkhi et al. / Food Chemistry 129 (2011) 312318
2.6.3. Cluster analysis
Cluster analysis (CA) is a multivariate technique, whose primary
purpose is to classify the objects of the system into categories or
clusters based on their similarities, and the objective is to nd an
optimal grouping for which the observations or objects within each
cluster are similar, but the clusters are dissimilar to each other.
Hierarchical clustering is the most common approach in which
clusters are formed sequentially. The most similar objects are rst
grouped, and these initial groups are merged according to their
similarities. Eventually as the similarity decreases all subgroups
are fused into a single cluster. CA was applied to heavy metals in
sediment data using a single linkage method. In the single linkage
method, the distances or similarities between two clusters A and B
is dened as the minimum distance between a point in A and a
point in B:
DA; B minfdyi ; yj for yi in A and yj in Bg
where dyi ; yj is the Euclidean distance in (1).
At each step the distance is found for every pair of clusters and
the two clusters with smallest distance (largest similarity) are
merged. After two clusters are merged the procedure is repeated
for the next step: the distances between all pairs of clusters are cal-
culated again, and the pair with minimum distance is merged into
a single cluster. The result of a hierarchical clustering procedure
can be displayed graphically using a tree diagram, also known as
a dendrogram, which shows all the steps in the hierarchical proce-
dure (Alvin, 2002; Richard & Dean, 2002).
3. Result and discussion
3.1. General descriptive statistics
The average length and diameter of Cavendish banana used for
the study were 18.0 and 6.0 cm, respectively, and the average
weight per fruit was 174 g. Banana peel our produced was brown-
ish in colour and green pulp our was creamy pale-yellow in
colour, whilst ripe pulp our was light brown in colour. All our
had visible dark spots scattering about the our samples, and
presented banana avour. The average yield of G-peel, R-peel,
G-pulp and R-pulp ours were 38.5, 48.6, 159.6 and 118.5 g/kg of
fresh banana fruit, respectively. It is evident that pulp yielded more
our than the peel, and green pulp yielded the highest amount of
our compared to others.
Table 1 summarises descriptive statistics including the mean,
standard deviation, maximum and minimum values for all physi-
cochemical properties of our. The spread around the mean value
(Standard deviation (Std)) was small and random in all samples,
indicating consistency of samples. The mean pH of our ranged
between 4.80 to 5.47 and R-peel our showed the highest pH
whilst G-peel our showed the lowest.
The mean TSS ranged between 1.22 to 4.26 Brix with the order;
R-pulp > R-peel > G-peel > G-pulp. TSS indicates soluble solid con-
tent of our, and high TSS has been associated with high sucrose
content in banana pulp (Bugaud, Chillet, Beaute, & Dubois, 2006).
It has been reported that the average starch content drops from
70% to 80% in the pre-climacteric period to less than 1% at the
end of the climacteric period, while sugars, mainly sucrose,
accumulate to more than 10% of the fresh weight of the fruit
(Zhang, Whistler, BeMiller, & Hamaker, 2005). The lower TSS of
green banana our is acceptable since it is known that amylase,
glycosidase, phosphorylase, sucrose synthase and invertase can
act in the degradation of starch and the formation and accumula-
tion of soluble sugars (Emaga et al., 2007; Terra, Garcia, & Lajolo,
1983). Since TSS of R-pulp was higher than R-peel, it can be
concluded that R-pulp had higher sugar content than R-peel our.
In principle, this could cause less sweetness to be perceived in
R-peel our which in turn could inuence consumer acceptance.
This suggestion however, needs further testing. Three types of
soluble sugars, i.e. sucrose, glucose and fructose that have been
detected in banana peel (Emaga et al., 2007) may represent the
TSS of peel our.
The mean L value ranged between 37.6 and 74.2. This indicates
a substantial colour difference existed between the pulp and peel
our. From visual observation this difference was obvious, i.e. the
peel our was a lot darker than the pulp. Major change in colour
took place during drying of the peel that yielded dark brown
powder, particularly in the ripe samples. As banana peel contains
glucose, fructose and protein (Emaga et al., 2007), an extend of
the Maillard reaction could had occurred, while certain enzymes
such as polyphenol oxidase may be present in banana peel that
could contribute a certain degree of enzymatic browning (Thipay-
1 arat, 2007). This later explanation seems acceptable since the
enzymatic browning of banana is well known problem.
During ripening of banana, the esh colour changes from the
typical opaque white of a product with a high starch content to
a very soft yellow as the yellowing of the skin intensies (Salva-
dor et al., 2007). In fresh banana, the colour changes of peel during
storage as a result of ripening has been observed as a loss of green-
ness, and increase in reddish and yellowness tones (Salvador et al.,
2007) which correspond to increase in the a and b values (Chen &
Ramaswamy, 2002) that took place as a result of the breakdown of
the chlorophyll in the peel. a value of the ripe our was marginally
higher than the green, meanwhile b value of the ripe our was
lower than the green that does not reect the actual colour charac-
teristics of green and ripe banana pulp and peel. Inconsistency in b
value might have been attributed to the excessive browning occur-
ring in the peel during drying, and to some extend the presence of
dark spots scattering about the our and interfere with colour
analyses. No study could be found on the effect of drying on
banana peel, but quality of banana paste as inuenced by vacuum
dehydration has been studied by Thipayarat (2007). It was found
that banana paste dehydrated with vacuum dehydration had
darker colour (lower L value) and more intense yellow colour
(higher b value) that was hypothesised to be due to condensation
and to moisture loss, enzymatic and non-enzymatic browning
(Thipayarat, 2007).
Mean WHC of all our samples increased with temperature, and
ranged between 1.4 and 8.2 g/g dry sample. These values are lower
than those reported in mango dietary bre (12 and 15 g water/g
dry sample) and mango peel dietary bre (11 g/g) (Larrauri, Ruper-
ez, Borroto, & Saura-Calixto, 1996), but were comparable with
bre-rich unripe banana our (2.5 g/g) (Rodriguez-Ambriz et al.,
2008). The mean WHC at all temperatures tested were the highest
in R-peel our (6.18.2 g/g dry sample), and was the lowest in
R-pulp our (1.44.7 g/g dry sample). WHC could be related to
the physical state of starch (Waliszewski, Aparicio, Bello, & Mon-
roy, 2003), dietary bre and protein in the our. According to
Rodriguez-Ambriz et al. (2008) amylose has the capacity to effec-
tively bind water molecules, yielding a higher WHC. However since
starch was low in ripe banana peel (Emaga et al., 2007), the high
WHC noted in R-peel could be attributed to the dietary bres
and protein. The increase in WHC at 80 C in all our samples
was due partly to protein denaturation, solution properties of
dietary bre such as hemicelluloses and pectin polysaccharides
(Zhang et al., 2005) and to a smaller extend to the gelatinisation
of starch in the our that absorbs water into starch granules with
concomitant swelling (Rodriguez-Ambriz et al., 2008).
Another functional property of banana our is oil holding
capacity (OHC). In general the mean OHC of almost all samples
increased with temperature, and ranged between 0.50 and
1.30 g/g dry sample. These values are lower than that reported in
 A.F.M. Alkarkhi et al. / Food Chemistry 129 (2011) 312318 315

Table 1
Descriptive statistics for selected physicochemical properties of banana our.

Parameter Green (G-peel) Ripe (R-peel)

c d
Min Max Mean Std Min Max Meanc Std
(a) Cavendish peel our
pH 4.30 5.33 4.80 0.42 4.86 5.69 5.47 0.24
TSS (Brix) 1.53 1.90 1.73 0.12 3.20 3.63 3.46 0.14
L value 34.83 48.73 40.88 4.46 32.43 41.08 37.62 3.07
a value 3.79 6.42 5.20 0.78 4.77 6.34 5.55 0.39
b value 21.01 27.07 23.27 1.94 11.02 14.01 12.47 0.88
WHC40a 4.14 5.20 4.91 0.36 5.39 6.55 6.10 0.33
WHC60a 4.81 5.85 5.23 0.33 5.59 6.72 6.34 0.33
WHC80a 5.15 6.50 5.88 0.34 6.65 9.26 8.19 0.68
OHC40b 0.69 0.85 0.76 0.04 0.78 1.06 0.93 0.08
OHC60b 0.68 0.80 0.76 0.03 0.92 1.05 0.98 0.04
OHC80b 0.95 1.17 1.03 0.06 1.07 1.39 1.28 0.08
Viscosity (mPa s) 46.73 60.07 54.24 4.38 66.80 83.90 76.47 5.56
BEF (N) 32.70 40.91 37.29 2.53 35.94 63.24 50.68 8.73
(b) Cavendish pulp our
Green (G-pulp) Ripe (R-pulp)
pH 4.37 5.65 5.06 0.52 4.76 5.60 5.13 0.29
TSS (Brix) 1.03 1.43 1.22 0.12 3.77 4.57 4.26 0.24
L value 64.37 79.25 74.18 4.62 67.12 74.86 70.85 2.53
a value 1.57 3.67 2.53 0.78 2.42 5.07 3.22 0.80
b value 14.69 21.69 17.36 2.32 11.51 20.23 14.15 2.59
WHC40a 3.72 4.08 3.94 0.12 1.08 1.69 1.37 0.18
WHC60a 5.37 5.99 5.66 0.17 1.56 1.88 1.71 0.10
WHC80a 6.03 6.53 6.31 0.17 3.99 5.03 4.67 0.41
OHC40b 0.64 0.91 0.80 0.09 0.69 0.87 0.79 0.05
OHC60b 0.42 0.64 0.50 0.07 0.73 0.87 0.82 0.04
OHC80b 0.71 0.97 0.85 0.06 0.94 1.15 1.05 0.07
Viscosity (mPa s) 35.07 47.47 40.94 3.65 84.13 91.67 87.88 2.18
BEF (N)e 0.54 0.81 0.67 0.07 1.97 2.72 2.32 0.20
a
Water holding capacity (g water/g dry sample).
b
Oil holding capacity (g oil/g dry sample).
c
n = 12.
d
Standard deviation.
e
Back extrusion force.

Table 2 bre-rich banana powder that could hold 2.2 g oil/g dry sample
Multivariate test (MANOVA) for four different groups (G-peel, R-peel, G-pulp, and R-
pulp).
(Rodriguez-Ambriz et al., 2008), but are similar to that of mango
dietary bre with OHC in the range 1.01.5 g oil/g ((Larrauri
Effect Test Value F Sig. et al., 1996). Other products tested for OHC by other researchers in-
Group Pillais trace 2.94 123.48 <0.0001 clude mango peel dietary bre (4 g oil/g), (Larrauri et al., 1996)
Wilks Lambda 0.00 225.58 <0.0001 and citrus peel bre (2.355.09 g oil/g) (Chau & Huang, 2003).
Hotellings Trace 393.79 309.65 <0.0001
OHC relates to the hydrophilic character of starches present in
Roys largest root 215.94 564.75 <0.0001
the our (Rodriguez-Ambriz et al., 2008) that is present in high
quantity in green our (Rodriguez-Ambriz et al., 2008; Zhang
et al., 2005), and in lesser quantity in ripe our.
Mean viscosity ranged between 41.0 and 87.9 mPa s, and the
Table 3
Wilks Lambda for testing DFs. mean BEF ranged between 0.67 and 50.7 N. The order of viscosity
and BEF was similar; R-pulp > R-peel > G-peel > G-pulp. The func-
Test of function(s) Wilks Lambda P-value
tionality of starch is largely related to its gelatinisation and pasting
1 through 3 0.000 <0.0001 characteristics. When our sample was heated in water, starch
2 through 3 0.000 <0.0001 granules swell at their gelatinisation temperature, and when amy-
3 0.051 <0.0001
lose leaches out of the granules and swell, viscosity and textural

Table 4
The results of classication for discriminant analysis of the four groups.a

Group % Correct Predicted Group Membership

Cavendish Green Peel Cavendish Green pulp Cavendish Ripe Peel Cavendish Ripe pulp
Cavendish Green Peel 100 12 0 0 0
Cavendish Green pulp 100 0 12 0 0
Cavendish Ripe Peel 100 0 0 12 0
Cavendish Ripe pulp 100 0 0 0 12
a
100.0% of original grouped cases correctly classied.
316 A.F.M. Alkarkhi et al. / Food Chemistry 129 (2011) 312318

changes result. In green our, starch gelatinisation may contribute low in ripe banana peel (Emaga et al., 2007), the viscosity and
to a certain extend to viscosity and texture. Since starch content is texture of ripe our could have been attributed mostly by

(a) 25.00

20.00

15.00
Scores of the second discriminant

10.00

5.00 Green Peel

Ripe Peel
Green Pulp
0.00 Ripe Pulp
-20.00 -15.00 -10.00 -5.00 0.00 5.00 10.00 15.00 20.00 25.00 30.00

-5.00

-10.00

-15.00

-20.00
Scores of the first discriminant function

(b) 6.00
Scores of the third discriminant function

4.00

2.00

Green Peel
0.00 Ripe Peel
-20.00 -15.00 -10.00 -5.00 0.00 5.00 10.00 15.00 20.00 25.00 30.00 Green Pulp
Ripe Pulp
-2.00

-4.00

-6.00

-8.00

-10.00
Scores of the first discriminant function

(c)
Scores of the scores of the third discriminant

6.00

4.00

2.00

Green Peel
0.00 Ripe Peel
-20.00 -15.00 -10.00 -5.00 0.00 5.00 10.00 15.00 20.00 25.00 Green Pulp
Ripe Pulp
-2.00

-4.00

-6.00

-8.00

-10.00
Scores of the second discriminant

Fig. 1. The scores of discriminant functions for (a) rst and second (b) rst and third, and (c) second and third, discriminant functions.
 A.F.M. Alkarkhi et al. / Food Chemistry 129 (2011) 312318 317

hemicelluloses and pectin polysaccharides. This explanation is
acceptable since banana peel contain a vast quantity of dietary -
bre, mainly hemicelluloses and pectin polysaccharides (Emaga
et al., 2007; Zhang et al., 2005). The total dietary bre content of
banana peel can be as high as 50% (based on dry basis) (Emaga
et al., 2007), while the hemicelluloses of banana peel constitute
20% of peel, and have solution properties (Zhang et al., 2005).
Hemicelluloses of banana peel may even be further developed into
gums or hydrocolloids (Zhang et al., 2005). The exceptionally high
viscosity of ripe our was also attributed to its high sugar content
as indicated by TSS, and protein contents since protein has been
shown to increase with increasing ripeness of banana cultivars
(Yomeni et al., 2004). The differences in these functionalities may
have implications in products incorporated with banana ours.
Choice of our to be used in product development will be based
on these properties. For instance, our prepared from ripe banana
peel or pulp will be suitable for products requiring sugars, viscosity
and texture. These results also indicated the potentials of Caven-
dish banana peel our, together with banana pulp our to be
exploited as food ingredients.
3.2. Multivariate analysis
The results of MANOVA for physicochemical parameters are
shown in Table 2. According to these results, the physicochemical
properties in all groups (G-peel, R-peel, G-pulp and R-pulp) exhib-
ited a strong signicant difference in terms of selected parameters
(P < 0.0001). These results indicate that ours prepared from peel
and pulp, using green and ripe banana are different based on
physicochemical properties. Variation in physicochemical proper-
ties was evaluated through DA. Three DFs were found to discrimi-
nate the four groups of our (Eqs. (2)(4)). Wilks Lambda test for
validity of each discriminant function showed that DFs are statisti-
cally signicant (Table 3). Furthermore 100% of the total variance
between the four groups was explained by three DFs. The relative
contribution for each parameter in each DF is given below:

Z 1 0:15pH 0:82TSS  0:64L  0:06a 0:30b
0:54WHC40 0:70WHC60  0:04WHC80
0:47OHC40 0:14OHC60 0:40OHC80
0:16Viscosity 0:64Texture 2

Z 2 0:06pH 0:61TSS  0:15L 0:15a  0:04b
 0:31WHC40  0:43WHC60  0:22WHC80
0:12OHC40 0:31OHC60 0:24OHC80
0:64Viscosity  0:12Texture 3

Z 3 0:47pH 0:27TSS 0:96L 0:87a  0:86b
 0:25WHC40 0:67WHC60 0:18WHC80
0:14OHC40 0:12OHC60  0:31OHC80
 0:10Viscosity  0:03Texture 4
TSS, L values, WHC40, WHC60 and texture (BEF) exhibited
strong contribution in discriminating the four our groups and ac-
count for most of the expected variations in physicochemical prop-
erties (Eq. (2)). Eq. (2) explains 54.8% of the variation between
groups, while Eq. (3) explains only 40.5% of the variation, and this
amount belongs mainly to the contribution of TSS and viscosity,
whilst other parameters showed less contribution in explaining
the variation between the groups. L, a, b values and WHC60
exhibited strong contribution in discriminating the groups and ac-
count for 4.7% of the variation. It can also be seen that OHC40,
OHC60, OHC80 and WHC80 did not show a strong contribution
in all equations (Eqs. (2)(4)).
The classication matrix for variety of banana (Table 4) showed
that 100% of the cases were correctly classied to their respective
groups. The results of classication also showed that signicant
differences existed between groups of banana our, which are ex-
pressed by in term of one DF (Eq. (2)).

Fig. 2. Dendrogram showing clustering of sampling sites based on physicochemical properties of banana peel and pulp ours produced using two different stage of ripeness.
318 A.F.M. Alkarkhi et al. / Food Chemistry 129 (2011) 312318
Relationship between the scores of DFs and the samples studied
is depicted in Fig. 1a. Horizontal axis corresponded to the scores of
the rst DF, whilst the vertical axis corresponded to the second DF.
It could be seen from Fig. 1 that all the rst DF was responsible for
discriminating pulp and peel our, whilst the second DF was
responsible in discriminating ripe and green our. It can also be
seen that R-peel and G-peel ours are closer to each other com-
pared to R-pulp and G-pulp our (Fig. 1b and c). This observation
has been explained in Table 1, i.e. the viscosity and TSS of G-peel
and R-peel ours were closer to each other compared to those of
G-pulp and R-pulp.
In general, it can be concluded that TSS, L value, WHC40,
WHC60 and BEF were useful in discriminating the peel and pulp
our of Cavendish banana. TSS and viscosity were useful in dis-
criminating our prepared from different stage of ripeness. How-
ever since major colour changes occurred during drying of the
peel, the use of L value may be inappropriate for discriminating
between pulp and peel ours.
3.2.1. Cluster analysis
Cluster analysis (CA) was used to identify the similarity groups
between the samples. CA rendered a dendrogram as shown in
Fig. 2, grouping all 48 samples into four statistically signicant
clusters. Cluster 1 (samples 112 for green peel (G-peel)), cluster
2 (samples 1324 for ripe peel (R-peel)), cluster 3 (samples 25
36 green pulp (G-pulp)), and cluster 4 (3748 ripe pulp (R-pulp)).
This result reveals that the four groups have different characteris-
tics in terms of physicochemical properties. This supports the re-
sults of MANOVA and DA, since the same result was obtained by
these two techniques. This grouping gives evidence that samples
in each group share each other the sources of physicochemical
properties. It implies that for rapid assessment and quality control
of banana pulp and peel our using physicochemical properties,
only one sample in each group will be sufcient to represent the
our prepared from parts of banana (pulp or peel) or banana from
different stage of ripeness (green or ripe).
4. Conclusion
Based on the physicochemical properties data and analysis,
multivariate statistical techniques can help in discriminating bana-
na our prepared from pulp and peel, and green and ripe fruits of
Cavendish banana. TSS, WHC40, WHC60 and BEF were useful in
discriminating the peel and pulp our, whilst TSS and viscosity
were useful in discriminating our prepared from different stage
of ripeness. Future work should aims at studying similar tech-
niques of differentiation of banana peel our produced from all
stages of ripeness. Improved drying methods to yield better quality
our should also be researched.
Acknowledgements
The authors gratefully acknowledge the nancial assistance
from Universiti Sains Malaysia and the research facilities by Dean
of the School of Industrial Technology, USM, Penang. USM short
term grant (304/PTEKIND/638098) is acknowledged.
References
Alvin, C. R. (2002). Methods of multivariate analysis. USA: John Wiley & Sons. INC..
Aparicio-Saguilan, A., Sayago-Ayerdi, S. G., Vargas-Torres, A., Juscelino, T., Ascencio-
Otero, T. E., & Bello-Perez, L. A. (2007). Slowly digestible cookies prepared from
resistant starch-rich lintnerized banana starch. Journal of Food Composition and
Analysis, 20, 175181.
Bugaud, C., Chillet, M., Beaute, M. P., & Dubois, C. (2006). Physicochemical analysis
of mountain bananas from the French West Indies. Scientia Horticulturae, 108,
167172.
Chau, C. F., & Huang, Y. L. (2003). Comparison of chemical composition and
physicochemical properties of different ber prepared from the peel of Citrus
sinensis L. cv. Liucheng. Journal of Agricultural and Food Chemistry, 51, 26152618.
Chen, C. R., & Ramaswamy, H. S. (2002). Color and texture change kinetics in
ripening bananas. Lebensmittel-Wissenschaft und Technologie, 35, 415419.
Emaga, T. H., Andrianaivo, R. H., Wathelet, B., Tchango, J. T., & Paquot, M. (2007).
Effects of the stage of maturation and varieties on the chemical composition of
banana and plantain peels. Food Chemistry, 103, 590600.
Emaga, T. H., Robert, C., Ronkart, S. N., Wathelet, B., & Paquot, M. (2008). Dietary
ber components and pectin chemical features of peels during ripening in
banana and plaintain varieties. Bioresource Technology, 99, 434643454.
Fagbemi, T. N. (1999). Effect of blanching and ripening on functional properties of
plantain (Musa aab) our. Plant Foods for Human Nutrition, 54, 261269.
Gamlath, S. (2008). Impact of ripening stages of banana our on the quality of
extruded products. International Journal of Food Science and Technology, 43(9),
15411548.
Juarez-Garcia, E., Agama-Acevedo, E., Sayago-Ayerdi, S. G., Rodriguez-Ambriz, S. L., &
Bello-Perez, L. A. (2006). Composition, digestibility and application in
breadmaking of banana our. Plant Foods for Human Nutrition, 61, 131137.
Larrauri, J. A., Ruperez, P., Borroto, B., & Saura-Calixto, F. (1996). Mango peels as a
new tropical ber: Preparation and characterization. Lebensmittel-Wissenschaft
und Technologie, 29, 729733.
Markus, P. F., Elena, R. R., Jacinto, D. M., & Carlos, D. R. (2002). Statistical
differentiation of bananas according to their mineral composition. Journal of
Agricultural and Food Chemistry, 50, 61306135.
Ranzani, T. D. C. M. R., Sturion, L. G., & Bicudo, H. M. (1996). Chemical and biological
evaluation of ripe banana peel. Archivos Latinoamericanos de Nutricion, 46(4),
320324.
Ricardo, C. R., Pablo, S. H., Elena, M. R. R., Jacinto, D. M., & Carlos, D. R. (2003).
Mineral concentration in cultivars of potatoes. Food Chemistry, 83, 247253.
Richard, A. J., & Dean, W. W. (2002). Applied multivariate statistical analysis. London:
Prentice-Hall.
Rodriguez-Ambriz, S. L., Islas-Hernndez, J. J., Agama-Acevedo, E., Tovar, J., & Bello-
Prez, L. A. (2008). Characterization of a ber-rich powder prepared by
liquefaction of unripe banana our. Food Chemistry, 107, 15151521.
Rungsinee, S., & Natcharee, P. (2007). Oxygen permeability and mechanical
properties of banana lms. Food Research International, 40, 365370.
Salvador, A., Sanz, T., & Fiszman, S. M. (2007). Changes in colour and texture and
their relationship with eating quality during storage of two different dessert
bananas. Postharvest Biology and Technology, 43, 319325.
Someya, S., Yoshiki, Y., & Okubo, K. (2002). Antioxidant compounds from banana
(Musa Cavendish). Food Chemistry, 79, 351354.
Surez, M. H., Rodrlguez, E. M. R., & Romero, C. D. (2007). Mineral and trace element
concentrations in cultivars of tomatoes. Food Chemistry, 104, 489499.
Suntharalingam, S., & Ravindran, G. (1993). Physical and biochemical properties of
green banana our. Plant Foods for Human Nutrition, 43, 1927.
Tchobanoglous, G., Theisen, H., & Vigil, S. (1993). Integrated solid waste management:
Engineering principles and management issues. New York: McGraw-Hill.
Terra, N. N., Garcia, E., & Lajolo, F. M. (1983). Starch-sugar transformation during
banana ripening: The behavior of UDP glucose pyrophosphorylase, sucrose
synthetase and invertase. Journal of Food Science, 48, 10971100.
Thipayarat, A. (2007). Quality and physiochemical properties of banana paste under
vacuum dehydration. International Journal of Food Engineering, 3(4), 1 (article 6).
Waliszewski, K. N., Aparicio, M. A., Bello, L. A., & Monroy, J. A. (2003). Changes of
banana starch by chemical and physical modication. Carbohydrate Polymers,
52, 237242.
Yomeni, M. O., Njoukam, J., & Tchango Tchango, J. (2004). Inuence of the stage of
ripeness of plaintains and some cooking bananas on the sensory and
physicochemical characteristics of processed products. Journal of the Science of
Food and Agriculture, 84(9), 10691077.
Zhang, P., Whistler, R. L., BeMiller, J. N., & Hamaker, B. R. (2005). Banana starch:
production, physicochemical properties and digestibility a review.
Carbohydrate Polymers, 59, 443458.