Food Sci. Technol. Res.

, 16 (2), 123 – 134, 2010

Determination of Metal Contents in Sugar Beet (Beta vulgaris) and Its Products:

Empirical and Chemometrical Approach

Biljana Škrbić1*, Nataša Ðurišić-Mladenović1 and Nada Mačvanin2

1
Faculty of Technology, University of Novi Sad, Bulevar cara Lazara 1, Novi Sad, Serbia
2
Faculty of Medicine, Institute of Occupational Health, Novi Sad, Serbia

Received August 20, 2009; Accepted November 3, 2009

Multielements K, Na, Ca, Mg, Zn, Fe and Cu were estimated in 25 composite samples of sugar beet,
extracted sugar beet pulp, dried sugar beet pulp, molasses, and white sugar collected during 50 days of
the 2005 campaign in one beet sugar factory, and in 65 soil samples collected from the fields where beet
was grown.
Mean total contents of analyzed elements in sugar beet and different sugar beet based products were
in the range: 18.11-37510 mg/kg dry matter (d.m.) for K, 6.54-8945 mg/kg d.m. for Na, 14.36-5220 mg/kg
d.m. for Ca, 0.09-2550 mg/kg d.m. for Mg, 0.01-10.85 mg/kg d.m. for Zn, 0.42-360.4 mg/kg d.m. for Fe,
and 0.07-7.09 mg/kg d.m. for Cu. Mean extractable amounts of these elements in the soils sampled in the
fields where beet was cultivated were as follows: 207.6 mg/kg d.m., 60.3 mg/kg d.m., 14.4 mg/kg d.m.,
404.7 mg/kg d.m., 1.44 mg/kg d.m., 9.89 mg/kg d.m. and 1.61 mg/ kg d.m., respectively.
In order to get a better insight into the metal patterns of the investigated samples, three statistical
techniques were used. Principal component analysis (PCA) and cluster analysis (CA) approved to be
more powerful than Spearman’s test in revealing the specific correlations among the variables (i.e. metal
contents). PCA and CA pointed out the specific metal pattern of molasses on one hand and of sugar beet,
extracted and dried sugar beet pulps on the other. Moreover, the chemometrical approach pointed out
that main components that classified the metal behavior in the examined samples were the ones correlated
with Na and K on one hand, and on the other hand with the remaining metals.

Keywords: sugar beet, metals content, AAS, principal component analysis, cluster analysis

Introduction
It is well known fact that the chemical composition of
plants reflects in general the elemental composition of the
growth media. Higher plants take up metals from air (or
water) by the shoots, but the absorption from soil by roots
could be regarded as the main pathway of elements to plants
(Kabata-Pendias and Pendias, 2000). The concentration of
metallic elements in soils is associated with biological and
geochemical cycles and is influenced by anthropogenic ac-
tivities, such as agricultural practices, transport, industrial
activities, waste treatment and disposal (He et al., 2005).
Metals exist in soil in immobile (sulphides, phosphates, sili-
cates, etc.) and mobile forms, with later being more impor-
*To whom correspondence should be addressed.
E-mail: biljana@tf.uns.ac.rs
tant for understanding the migration patterns and the uptake
by plants. It has been well documented that the soil contami-
nation is likely to result in a corresponding contamination of
harvested crops and the food consumers (Fergusson, 1990,
Younas et al., 1998, Nan et al., 2002). Furthermore, process-
ing of raw materials and the process conditions (additives,
pH, etc.) could also influence the quality of the final food
products including their metal contents (de Brujin and Bout,
1999, Mohamed, 1999).
The elemental composition of sugar beet has been a sub-
ject of few previous studies investigating sugar beet quality
and its influence on the quality of the final products (Mauch
and Farhoudi, 1979/80, Škrbić et al., 2005a, Škrbić and
Ðurišić-Mladenović, 2005). Additionaly, the heavy element
contents of sugar beet has been investigated to determine
the beet uptake as a consequence of either the growth on
124
contaminated soils (Bojinova et al., 1996) or different agri-
cultural practices (Saleh et al., 1999, Mantovi et al., 2003).
The contents of alkaline and alkaline earth metals in sugar
beet were also the subject of a sugar beet nutrition study by
Wakeel et al. (2009). The main sugar beet product, white
sugar, is referred as a very pure food product, although it
contains a very small but significant amount of impurities.
The control of inorganic impurities in the sugar, i.e. trace
elements, is important because of its fundamental nutritional
role in the diet of humans and especially in the diet of infants
and the elderly. Besides a safety perspective of the sugar
quality, the content of elements also determines usefulness of
sugar for various industrial applications (de Bruijn and Bout
1999, Wojtczak and Krol, 2002), making the sugar element
analysis more important. The main inorganic cations in sugar
are calcium (Ca), magnesium (Mg), potassium (K), while
there are also some trace elements in sugar such as arsenic
(As), mercury (Hg), lead (Pb), cadmium (Cd), iron (Fe), cop-
per (Cu) and zinc (Zn). The occurrence of toxic elements in
sugar such as As, Hg, Cd, Pb should be in compliance with
current legislations on the maximum permissible levels on
toxic elements. For instance, in Serbia the maximum permis-
sible contents for As, Pb and Cu in white sugar are set to be
1 mg/kg of each. In some EU Member States (Hungarian
Regulation, 2004) even more stringent limits has been speci-
fied for As and Pb in sugar, i.e. 0.1 and 0.5 mg/kg, respec-
tively, while for Hg, Cd, Cu and Zn the limits are 0.02, 0.02,
2.0 and 3.0 mg/kg, respectively. Consequently, the content
of toxic elements in sugar is usually very low and generally
does not result in any safety concerns for sugar. The content
of microelements such as Fe, Cu and Zn in sugar is impor-
tant from the technological point of view for the evaluation
of the technological process as well as for the evaluation of
the quality of sugar. The content of these elements may be
treated as an additional criterion of the purification and fil-
tration process of beet juice and of the sugar crystallization
process (Wojtczak and Krol, 2002). Considering the research
studies dealing with elemental analysis of commercial beet
sugar there are several published papers generally referring
the analysis of heavy elements either essential and/or toxic
ones (Leblebici and Volkan, 1998, Ronda et al., 2001, Wojtc-
zak and Krol, 2002, Škrbić et al., 2002, Škrbić and Gyura,
2006, Škrbić and Gyura, 2007). The content of impurities is
important also to be known for the sugar beet by-products
such as molasses and dried sugar beet pulp (fibers) nowa-
days, because they are widely used as food supplements due
to claimed health benefits. Molasses is a valuable ingredient
with high contents of vitamins and minerals, and is ideal in
specialty bread and bakery products, crackers, cookies, ce-
real bars, cakes, pies, syrups, sweet dressings and as natural
B. Škrbić et al.
colorant, buffering agent and as a raw material for baker’s
yeast production (Hickenbottom, 1996). On the other hand,
due to high content of natural calcium in sugar beet fibers,
their addition to foods and beverages may improve the cal-
cium balance without adversely affecting other minerals
(Wolever and Jenkins, 1997). Also, sugar beet fibers display
many unique functional properties in processed foods result-
ing from its combination of soluble and insoluble fibers (Ang
and Crosby, 2003). Thus, due to high contents of pectic and
cellulosic substances, there is a great potential for dried sugar
beet pulp addition into the health promoting foods.
The objective of this study was quantitative determina-
tion of four macroelements (K, Na, Ca, and Mg) and three
microelements (Zn, Fe and Cu) in sugar beet and its products
(extracted sugar beet pulp, dried sugar beet pulp, molasses
and white sugar) obtained in the 2005 campaign at the lead-
ing sugar factory in Serbia, and the mobile forms of the same
metals in the samples of soils where beet was grown in order
to elucidate the links between metal patterns within the chain
field - raw material - products. Furthermore, to obtain more
reliable information about relationships among the metals
and classification of the samples, these analytical data were
evaluated with statistical techniques as Spearman’s test, prin-
cipal component analysis (PCA) and cluster analysis (CA).
This is the first attempt to get a deeper insight into the metal
patterns of materials used and produced in the sugar industry
and also to find links with the pattern existing in soils where
sugar beet was grown. As this study followed up the contents
of 7 metals, including those important for the sugar beet nu-
trition (Na, K, Ca, Mg, Cu, Zn), and also for the evaluation
of the technological process (Ca, Fe), and their distribution
from the sugar beet to the final products, it could be regarded
as a contribution to field of agricultural and cultivation sci-
ence as well as to a better understanding of the impact of
sugar beet processing on the metal content distribution in the
by-products and final product (white sugar).
Materials and Methods
Sampling Sugar beet (slices) and its products, extracted
sugar beet pulp, dried sugar beet pulp, molasses and white
sugar were collected during 2005 campaign from the leading
Serbian sugar factory accounting for about 17% of the total
national white sugar production. The factory is located in Vo-
jvodina, the province on the north of Serbia, also known as a
main Serbian agricultural region.
In the campaign approximately 15 samples were pooled
together during a day to obtain test samples for particular
day. Ten test samples representing a 10-day period of the
campaign were mixed to obtain composite samples for that
period, so-called “decade”. As the 2005 campaign lasted for
Metals in Sugar Beet and Its Products
about 50 days, it means that for five “decades” of the cam-
paign (i.e. five 10-day periods), five representative composite
samples (nos. 1-5) of sugar beet, its by-products and final
product (white sugar) were prepared. In this way, the sam-
pling procedure provided that the qualities of the composite
samples for each particular period (“decade”) were related
among each other, e.g. sugar beet sample no. 1 was related
with the extracted sugar beet pulp no. 1, etc.
Composite soil samples were prepared from 65 sites lo-
cated in the arable fields of the Province of Vojvodina where
sugar beet was collected during campaign. At each sampling
site, four core subsamples (0-50 cm) were collected by a
plastic tool from a rectangular area of approximately 20 ×
50 m. Then, the subsamples were combined and mixed to a
composite sample of ~500 g. The samples were air-dried and
sieved through a 2-mm polyethylene sieve before analysis.
Then, they were ground until fine particles were obtained.
According to a comprehensive survey of the Vojvodina’s ag-
ricultural soils (Ubavić et al., 1993), soils examined here be-
long to a neutral pH reaction soil type with medium to good
provision with humus and total nitrogen, and good provision
with available phosphorus.
All the prepared samples were stored in polyethylene
bags at the temperature about 4℃ to avoid changes in chemi-
cal composition.
Apparatus Samples were analyzed depending on the
type of metals by Perkin-Elmer atomic absorption spectrom-
eter (AAS) model 5000. Software HG Graphic II (Perkin
Elmer, Norwalk, CT, USA) was employed for spectrum ac-
quisition and data processing.
Reagents All the chemicals used were of ultrapure
grade (Suprapur, Merck). The glassware was cleaned prior to
use by soaking overnight in 10% v/v HNO3 and rinsing with
Milli-Q water. Standard stock solutions of elements were
purchased from Merck (Germany).
Sample preparation and determination of elements The
aliquots from the pooled samples of sugar beet slices, ex-
tracted sugar beet pulp, and molasses were first evaporated
down to a homogeneous state.
All the samples from the sugar factory were wet acid –
digested. Method of sample preparation was described pre-
viously (Škrbić et al., 2005a,b). Briefly, one-gram samples
were dissolved with 10 cm3 of concentrated nitric acid and
10 cm3 of perchloric acid and heated under the reflux. In or-
der to complete the destruction of the organic matter several
portions of 1 cm3 of nitric acid were added. After obtaining a
clear, colorless solution, it was evaporated to dryness under
vacuum and transfered with deionised water Milli-QTM type
(18 MΩ) into a 50 cm3 flask.
Exchangeable (mobile fraction of) macroelements (K,
125
Na, Ca, and Mg) were extracted from the air-dried soil by
mixing 10 cm3 of neutral 1 N ammonium acetate with 10 g
of soil and shaken for 5 min. These elements were measured
by analyzing the filtered extract. Extractable (mobile fraction
of) zinc, copper, and iron (Zn, Cu, and Fe) were determined
by treating 10 grams of air-dried soil with 20 cm3 of DTPA
(diethylene triamine pentaacetic acid) extracting solution
(0.005 molar DTPA, 0.1 M TEA (triethanolamine), and 0.01
M CaCl2 , adjusted to pH 7.3). After shaking for 2 h, the soil
sample was filtered and the extract was analyzed.
K, Na, Ca, Mg, Zn, Fe and Cu, concentrations in the
samples were determined by direct aspiration of the aqueous
solution into air-acetylene flame of AAS at 766.5 nm, 589.0
nm, 422.7 nm, 285.2 nm, 213.9 nm, 248.3 nm and 324.7 nm,
respectively. The measurable range for each element was as
follows: 0.2-1.6 mg/L for K, 0.10-1.20 mg/L for Na, 0.2-3.2
mg/L for Ca, 0.02-0.60 mg/L for Mg, 0.10-1.20 mg/L for
Zn, 1.25-10 mg/L for Fe, and 0.10-3.00 mg/L for Cu. The
samples with the overanged contents of elements were di-
luted in order to be in the specified range. The exception was
the determination of the Zn, Fe and Cu contents in the sugar
samples that was performed by AAS with electro-thermal
atomization. The measurable ranges here were: 0.1-0.5 µg/L
for Zn, 4-10 µg/L for Fe, and 0.7-2.7 µg/L for Cu. The oper-
ating conditions were based on those suggested by the manu-
facturer (Beauty, 1988) and were presented elsewhere (Škrbić
and Čupić, 2004).
Analytical quality control Limits of detection (LOD)
and limits of quantification (LOQ) were calculated on the ba-
sis of three and ten times the standard deviation of the mean
blank determinations, respectively, obtained from 5 measure-
ments. The obtained values of the method LOD (LOQ in pa-
rentheses) for metals in sugar beet and the products were as
follows: 0.2 (0.7) mg/kg for K; 0.02 (0.07) mg/kg for Na; 0.1
(0.3) mg/kg for Ca; 0.001 (0.003) mg/kg for Mg; 1.00 (3.00)
mg/kg for Zn, 1.10 (3.60) mg/kg for Fe and 0.9 (3.0) mg/kg
for Cu. For the method of Zn, Fe and Cu determination in
the sugar samples that was performed by AAS with electro-
thermal atomization, LOD (and LOQ) values were 0.05 (0.16)
μg/kg, 1.0 (3.3) μg/kg, and 1.0 (3.3) μg/kg, respectively. The
LOD (and LOQ) values of the method for soil metal analysis
were: 0.02 (0.07) mg/kg for K; 0.002 (0.007) mg/kg for Na;
0.01 (0.03) mg/kg for Ca; 0.0001 (0.0003) mg/kg for Mg; 0.1
(0.3) mg/kg for Zn; 0.11 (0.36) mg/kg for Fe; and 0.09 (0.30)
mg/kg for Cu.
Quality control consisted of analysis of blank, spiked and
duplicate samples. Blank samples were included in every
batch of samples to check for possible contamination. They
were treated and analyzed with the same method as the ac-
tual samples.
126
The accuracy of the procedure was checked by the analy-
sis of the sugar samples fortified with known quantities of
analyzed elements, i.e. 10 mg/kg for K, Na, and Ca; 0.02 mg/
kg for Mg, Zn, and Cu; and 0.5 mg/kg for Fe. The recoveries
for investigated metals were in the range of 79-107% with
relative standard deviations less than 15%. The obtained data
were not adjusted on the basis of these recoveries and were
presented as mean value of the duplicates.
Statistical data analysis In order to investigate the ele-
ments correlation and samples classification, the obtained
analytical results were arranged into the input matrix in the
following way: the concentrations of the metals (variables)
were arranged in the columns, whereas the investigated sam-
ples (cases) of sugar beet, molasses, dried sugar beet pulp,
extracted sugar beet pulp, and sugar were in the matrix rows.
Additionally, the mean values of the elements determined in
the soil samples was also put in the input matrix in order to
examine links amongst the metal patterns within the chain
soil-raw material-products. In this way the 26 × 7 (cases ×
variables) matrix was obtained. Throughout the study the
software packages Excel 2003 (Miscrosoft Corporation) and
STATISTICA 6.0 (StatSoft Inc., Tulsa, OK, USA) were used.
Spearman’s non-parametric correlation test In order
to quantitatively analyze the relationship among micro- and
macroelement contents in sugar beet, sugar beet products
and soil where the beet was grown, the Spearman’s non-
parametric rank correlation coefficient was applied to the
obtained data. The non-parametric correlation coefficient is
a common parameter used to quantify the strength of asso-
ciation between the pairs of variables when the presumption
about normality is violated. Definition of the Spearman’s
non-parametric correlation coefficient could be found in
standard textbook of statistics (Massarat et al., 1997). A limit
value that separates significant from non-significant correla-
tions is chosen for the probabilities predefined (at 1 and 5%)
at a given number of observation (n) , i.e. for a given degrees
of freedom (n-2) (Massarat et al., 1997).
Principal component analysis (PCA) PCA has suc-
cessful application in various field of chemistry, includ-
ing environmental analysis and food chemistry (Heberger
et al., 1999, Csomos et al., 2002, Heberger et al., 2003,
Golobočanin et al., 2004, Škrbić et al., 2005b, Škrbić and
Onjia, 2007, Škrbić and Ðurišić-Mladenović, 2007, Škrbić et
al., 2009). This is the multivariate analytical tool used to re-
duce a set of original variables and to extract a small number
of latent factors (principal components, PCs) for analyzing
relationships among the observed variables (which are metal
contents in this study) and classification of samples. The
newly extracted variables, i.e. PCs, are linear combinations
of the original ones and are sorted into descending order ac-
B. Škrbić et al.
cording to the amount of variance that they account for in
the original set of variables. The loadings express how well
the new abstract principal components correlate with the old
variables. PCA will show which kinds of metals are similar
to each other, that is, carry comparable information, and
which ones are unique. Relationships among the samples are
evaluated on the score plots, whereas the loading plots show
the extent to which each variable contributes to the sample
separation (grouping).
In this work, the data were firstly mean-centered (column
means subtracted from each matrix element). Then each
matrix element was divided by the standard deviation of the
respective column and the established matrix was submitted
to PCA. In this way, we removed scalar differences and the
overwhelming influence of high-concentration metals that
would otherwise numerically “swamp” major differences in
lower concentration metals, placing all variables on equal ba-
sis in PCA. Number of PCs extracted from the variables was
determined by Kaiser’s rule (Kaiser and Rice, 1974). This
criterion retains only PCs with eigenvalues that exceed one.
According to Morrison (1967) principal components should
account for approximately 75% of the total variance. In order
to interpret the significance of retained PCs in terms of the
original variables, only those loadings (coefficients) whose
absolute value was greater than 60% of the maximum coef-
ficient in absolute value for each PC were considered (Jolliffe,
1986). The algorithm of PCA can be found in the standard
textbooks (Massart et al., 1997).
Cluster analysis (CA) CA is used to classify samples
into groups, i.e. clusters. It can be considered to be an alter-
native to PCA. An important step in clustering is to select a
distance measure, which determines the similarity or dissimi-
larities of the samples. The distance between two points is
well-defined; the simplest distance measure is the Euclidean
distance. However, numerous clustering algorithms exist as
to what is considered to be a distance between two groups.
It may be defined as the distance of the two closest points
or the two farthest points, etc. Weighted schemes are also
reliable alternatives. The most popular clustering algorithm
is perhaps Ward’s method. Definitions of distance measures
and clustering algorithms can be found in standard chemo-
metric textbooks (Otto, 1999, Vandeginste et al., 1998). In
this study, Ward’s method as an amalgamation rule and the
Euclidean distance as a measure of the proximity between
samples are used for the cluster analysis of the input matrix
previously described.
Results and Discussion
Empirical approach The mean total contents and stan-
dard deviations of examined metals in samples of sugar
Metals in Sugar Beet and Its Products 127

beet and its products representative for the production of the
studied sugar factory are presented in Table 1. This table also
gives the mean extractable levels (and standard deviations)
for the metals in soil where the beet was grown.
The highest concentrations of the alkali metals (K and
Na) were determined in molasses followed by sugar beet, ex-
tracted sugar beet pulp, dried sugar beet pulp, soil and sugar
(Table 1). In fact, K and Na are two very important monova-
lent cations in sugar beet nutrition. Both are macronutrients
being taken up and utilized in large quantities by sugar beet
crops producing optimum yield. Sugar beet is a halophyte,
i.e. natrophilic (“sodium-loving”) crop, due to its original
shoreline habitat, absorbing and assimilating Na that partly
replaces K (Draycott and Christenson, 2003; Wakeel et al.,
2009). It has been known that the K and Na contents of sugar
beet fluctuate widely and are dependent, in particular, on lo-
cation, weather conditions, soil, fertilizer usage, and time of
harvesting (Van der Poel et al., 1998). Regarding Mauch and
Farhoudi (1979/80), the sodium content of sugar beet is nor-
mally much lower than level of potassium; this was also the
case in this study with the mean content of K in sugar beet
found to be 3 times higher than Na (Table 1). Similar value
of the K/Na ratio was also observed in extracted sugar beet
pulp, dried sugar beet pulp and sugar, while in molasses this
ratio was even higher (about 4; see Table 1). It is interesting
to note that the ratio of mean mobile fraction of K and Na in
soil samples was also about 3. All these findings confirmed
the fact that K and Na in sugar beet products originated from
sugar beet without notable influence of manufacturing pro-
cess on their contents, whereas their content in sugar beet
reflected the extractable content of K and Na in soil. Com-
parison of the obtained data on K and Na with the published
ones was possible only for sugar as for the other investigated
samples there were no published data. The mean content of
Na and K in white sugar obtained here corresponded to the
values previously reported to be in the ranges 6.5-8.5 mg/kg
and 9.33-48 mg/kg, respectively (de Brujin and Bout, 1999,
Škrbić et al., 2002).
Considering the alkaline earth metals (Ca and Mg), the
highest contents were found in extracted and dried sugar
beet pulps, with the Ca/Mg ratio of about 2 (Table 1). In
molasses and sugar Ca was also predominant with 25 to
159-times higher content than Mg, respectively (Table 1).
However, in sugar beet and soil the opposite was seen as the
content of Mg was approximately 2 and 28 times higher than
the respective Ca content (Table 1). It is obvious that Ca in
sugar beet products predominantly originated from the beet
processing, i.e. from the addition of calcium oxide during
juice purification. During the course of the manufacturing
process, the calcium ions (and magnesium ions present in the
sugar beet) are precipitated as carbonate, sulphite, sulphate
and eliminated. However, this does not happen completely
and depends on the composition of the juices, its alkalin-
ity and pH-values. In this study, the mean Ca content of
the white sugar was high (14.36 mg/kg, Table 1) indicating
low effectiveness of the second carbonatation, in which the
surplus of Ca, previously added to the process in main lim-
ing, should be precitated. Similar result was obtained for the
same factory during campaign of 2001 (Škrbić et al., 2002),
while the low Ca content (up to 1.2 mg/kg) confirmed the
effectiveness of that procedure (de Brujin and Bout, 1999).
In the review paper of Mauch and Farhoudi (1979/80), only
Ca of the alkaline earth metals in sugar has been detected in
a concentration that is worth mentioning. As Mg hydroxide
is highly insoluble, most of Mg extracted from beets leaves
the process with the carbonatation slurry from the first carbo-
natation, and therefore, its concentration in the sugar is low.
In this study, mean level of Ma in white sugar (0.09 mg/kg)
was similar to the data of de Bruijn and Bout (1999) (0.1-0.3
mg/kg).
With respect to three investigated microelements, Fe was
the most abundant in all investigated samples (Table 1). The
mean Fe content in the studied sugar beet (360.4 mg/kg) was
in the range from 17.2 to 502.6 mg/kg reported by Mauch

Table 1. The mean total contents and standard deviations of metals in sugar beet and its products (mg/kg d.m.) and mean extractable
levels and standard deviations of metals in soil (mg/kg d.m.).

Sample K Na Ca Mg Zn Fe Cu

Sugar beet 7315±770.8 2370±103.4 970±53.8 2275±80.5 6.76±1.10 360.4 ± 55.9 7.09 ± 0.75

Extracted sugar beet pulp 3600±667.4 1165±285.7 4720±406.7 2550±64.0 5.89±0.31 116.0 ± 11.2 4.69 ± 0.47
Dried sugar beet pulp 2650±369.2 1025±197.2 5220±189.7 2400±122.5 10.10±0.26 156.2 ± 20.2 6.32 ± 0.37
Molasses 37510±1364 8945±1161 3019±432.0 117.6±38.1 10.85±0.93 134.9 ± 28.2 2.30 ± 0.30
Sugar 18.11±2.10 6.54±0.51 14.36±0.71 0.09±0.03 0.01±0.00 0.42 ± 0.02 0.07 ± 0.01

Soil 207.6±70.3 60.3±32.8 14.4±12.3 404.7±164.4 1.44±3.59 9.89 ± 4.7 1.61 ± 1.86
128 B. Škrbić et al.

and Farhoudi (1979/80). The sugar samples also had the
content of Fe in the range of concentrations given in the lit-
erature (Table 2). It is known that Fe is generally removed
during juice purification, while in the case of a low pH-value
it dissolves from the metal surface of the equipment causing
a rise in the iron content in white sugar (de Brujin and Bout,
1999).
Copper mean level in the investigated sugar beet (7.09
mg/kg) was higher than the concentrations given by Mauch
and Farhoudi (1979/80) of 1.9 - 3.5 mg/kg and also by Man-
tovi et al. (2003) of 1.5-6 mg/kg, but was lower than 11.8
mg/kg determined for beet grown at contaminated area in
Bulgaria (Bojinova et al., 1996). As the major part of Cu
contained in the press juice is eliminated during juice puri-
fication, its mean level in the studied white sugar was 0.07
mg/kg (Table 1). It was in the range of concentrations given
in the literature (Table 2) and lower than allowed level of
Cu in sugar defined by the Serbian (1992) and by Hungarian
Regulations (2004).
The mean Zn content in the sugar beet found to be 6.76
mg/kg was similar with the lower value of the range 7-19
mg/kg reported by Mantovi et al. (2003), and much lower
than the data given for the beet grown at contaminated area
in Bulgaria of 37.1 mg/kg (Bojinova et al.,1996). Mean Zn
level in the investigated white sugar was 0.01 mg/kg; it was
within the concentrations given in the literature (Table 2) and
lower than allowed level of Zn in white sugar (3 mg/kg) de-
fined by the Hungarian Regulation (2004).
Chemometrical approach
Spearman’s test The Spearman’s correlation matrices
were computed and the significant correlations obtained for
the criterion values of probability p<0.05 and p<0.01 were
presented in Table 3 with respect to the metal contents in the
analyzed samples. The limit values that separated significant
from non-significant correlations were found in the correla-
tion coefficient table (Massart et al., 1997) and in our case
for n = 26 they were 0.496 and 0.388 at the 1% and 5% sig-
nificance levels, respectively. According to the results (Table
3) strong (significant) correlation of positive sign can be seen
among majority of the investigated elements: K and Na, K
and Zn, K and Fe, Na and Zn, Na and Fe, Ca and Mg, Ca
and Zn, Mg and Fe, Mg and Cu, Zn and Fe, and Fe and Cu.
Positive correlation coefficient between the contents of two
elements indicates that the high concentration of one element
in the analyzed samples is associated with the high content
of the other and vice versa.
The Spearman’s test was also applied on the pairs of sam-
ples in order to elucidate the similarities among them based
on their element patterns. The obtained diagonal matrix of
the correlation coefficients (all were positive i.e. in the range

Table 2. Ranges of Fe, Cu and Zn concentrations (mg/kg) in white sugar samples reported by various studies.

Fe Cu Zn Origin Reference

0.14-0.23 0.02-0.04 0.02-0.06 The Netherlands de Brujin and Bout (1999)

0.10-1.01 0.01-0.12 0.01-0.05 Serbia Škrbić and Gyura (2007)
1.41-1.79 0.02-0.035 — Turkey Leblebici and Volkan (1998)
0.03-1.13 0.01-0.36 0.01-0.30 European sugar beet refineries Wojtczak and Krol (2002)

Table 3. The Spearman’s non-parametric correlation matrix for metals content in sugar beet, its products
and soil where the beet was grown.

K Na Ca Mg Zn Fe Cu

K 1.000
Na 0.979** 1.000
Ca 0.198 0.231 1.000
Mg 0.151 0.111 0.734** 1.000
Zn 0.660** 0.676** 0.597** 0.237 1.000
Fe 0.583** 0.591** 0.328 0.394* 0.560* 1.000
Cu 0.361 0.382 0.491* 0.636** 0.373 0.788** 1.000

* Correlation is significant at p < 0.05.

** Correlation is significant at p < 0.01.
Metals in Sugar Beet and Its Products 129

0-1) was transformed into “color” map that represented the
magnitude of correlation by a “color” scale (Fig. 1). The
“colors” were scaled from white for the correlations with the
coefficients in the range 0 -0.6 to medium grey (referring to
the ranges of coefficients 0.6-0.8) and dark grey (with coef-
ficients higher than 0.8). In this way, the most similar (cor-
related) samples can be seen from the color map through the
occurrence of dark shades. Thus, as the strongest correlations
existed between the same type of the samples despite of the
time (“decade”) of sampling (Fig. 1) it could be concluded
that rather constant element patterns among them existed
during the campaign. Few more correlations emerged (Fig. 1):
among sugar beet and molasses, sugar beet and most of the
sugar samples, extracted and dried sugar beet pulps, extract-
ed and dried sugar beet pulp with some sugar samples, and
molasses and sugar. Based on these observation the follow-
ing could be drawn: a) the metal pattern of molasses reflected
the one of sugar beet, even though the opposite has been seen
after the direct comparison of analytical (empirical) results,
particularly concerning the contents of Ca, Mg, Zn and Cu
(Table 1); b) pressing and drying of extracted sugar beet pulp
did not impair its metal pattern, resulting in similarity be-
tween the patterns of extracted and dried pulps; and c) metal
pattern of sugar reflected the pattern of sugar beet, extracted

Sugar beet Extracted sugar beet pulp Dried sugar beet pulp Molasses Sugar Soil

1 2 3 4 5 1 2 3 4 5 1 2 3 4 5 1 2 3 4 5 1 2 3 4 5
1

2
Sugar beet

1
Extracted sugar beet

3
pulp

1
Dried sugar beet pulp

2
Molasses

2
Sugar

Soil

SPEARMAN’S
COLOR COEFFICIENT
0 - 0.6
0.6 - 0.8
0.8 - 1

Fig. 1. Color map of the Spearman’s correlation coefficients between samples based on their element patterns. Numbers 1-5 represent 5
composite samples of each type of material representing 5 “decades” (10-day-periods) of the 2005 campaign in the selected sugar factory.
130 B. Škrbić et al.

and dried pulp, and also of molasses. It is interesting to note
that the correlation between soil and sugar beet samples was
found to be statistically significant, coinciding with the fact
that the metal patterns of plants is largely dependent of the
soil composition.
Principal component analysis To gain better insight
into the latent structure (hidden regularities) of the obtained
data and to investigate similarities and dissimilarities of
metal contents of the investigated samples PCA was applied
and the obtained results are shown in Table 4 and Fig. 2.
Three PCs were retained explaining 97.7% of the total
variance in the data (Table 4). The first principal component,
PC1, explained 45.7% of the data variance and it was signifi-
cantly loaded with Cu > Zn > Mg > Fe > Ca (Table 4). The
second component, PC2, accounting for 37.3%, correlated
markedly with alkali metals, i.e. K > Na, while the third one,
PC3, explained 14.7% of the variance and correlated posi-
tively with Fe and negatively with Ca (Table 4).

Table 4. Results of principal component analysis for elements in sugar beet and its
analyzed products and soil where sugar beet was grown(a).

Original variables PC 1 PC 2 PC 3

K -0.981

Na -0.963
Ca -0.678 -0.728
Mg -0.791
Zn -0.814
Fe -0.745 0.652
Cu -0.910

Eigenvalue 3.20 2.61 1.03

Proportion of total variance 45.7 37.3 14.7

Cumulative 45.7 83.0 97.7
(a)
only loadings higher than 60% of the maximum absolute value in each PC are presented.

Fig. 2. The PCA biplots for variables and cases: a) PC1 vs. PC2, b) PC1 vs. PC3 (▲ sugar beet, ▼ extracted sugar beet, ■ dried sugar
beet, molasses, sugar, ● soil).
▲

▲
Metals in Sugar Beet and Its Products
Figires 2a and 2b show the biplots of PC1 vs. PC2 and of
PC1 vs. PC3, respectively, displaying graphically the infor-
mation obtained on samples and variables by PCA. During
the biplot construction, each loading and score values were
divided with the respective maximum values obtained for
particular PC; thus, the biplot presented the relative posi-
tions of the element loadings and the sample scores within
the range from –1 to +1. In this way, the interpretation of the
variables correlation and the influence on the samples clas-
sification could be made more obviously. The correlation
between alkali elements (Na/K) as well as between Cu/Mg
and Fe/Ca in relation to the data is unambiguous and signifi-
cant. It could be also said that the correlated elements carried
out the same amount of information and the influence on the
samples classification. It is interesting to note that Fe/Ca cor-
relation seen by PCA and linked to the influence of the tech-
nological process was not seen by the Spearman’s correlation
analysis (Table 3).
Distribution of the samples along PC1 was influenced
primarily by their contents of Cu, Zn and Mg, whereas the
content of Na and K governed the sample positioning along
PC2 (Fig. 2a). The following groups were observed in the
PC1 vs. PC2 biplot: group I consisting of the sugar beet,
dried and extracted sugar beet pulps, group II gathering all 5
samples of molasses, and group III rounded the soil and sug-
ar samples. Group I had the highest levels of Cu and Mg, and
intermediate level of alkali metals. Moreover, slight differen-
tiation in the metal patterns could be seen in all five samples
of each sample type reflecting slight pattern differences from
the first to the fifth “decade” of the campaign. Group II gath-
ered samples of molasses that contained the highest content
of alkali metals and intermediate content of the remaining
elements; again, the difference among the samples composed
in different “decades” could be observed and it was more evi-
dent than within the sample types in group I. In the group III,
the samples with the lowest levels of all of the analyzed met-
als were gathered; sugar samples had lower metal contents
than the soil. Within this group all five points representing
five sugar samples in the biplot (Fig. 2) that were collected
during five 10-day period (“decades”) of the 2005 campaign
in the factory included in this study were overlaid indicating
that they had identical metal patterns. Thus, it could be con-
cluded that the final step of sugar production (ash removal by
washing procedure in the centrifugals) removed differences
observed in the metal patterns of sugar beets from the first to
the fifth “decade” of campaign that could be also seen in the
patterns of intermediate- and by-products.
The biplot of PC1 vs. PC3 showed slightly different
groupings of the samples governed by the Fe and Ca con-
tents. The sugar beet samples separated clearly from the rest
131
of the samples due to the highest content of Fe. Moreover,
they contained the lowest Ca content in comparison to the
by-products. Thus, separation along PC3 reflected obviously
the influence of technological process on Fe and Ca contents:
decreasing of the Fe level from sugar beet to the by-products
and white sugar is a result of its removal during juice puri-
fication, and increase of the Ca content in the by-products
originates in the addition of the technological material (lime)
in the beet processing. The separation of sugar and soil
samples were influenced again with the lowest levels of all
investigated metals.
Cluster analysis CA uses less information (distances
only) than PCA. Dendrograms (the tree plots) of CA using
Ward’s method and the Euclidean distance as a measure of
the proximity among the metals in the samples and among
the samples, are shown in Figs. 3 and 4, respectively.
Considering the CA dendrogram of variables (Fig. 3) two
main clusters could be seen, cluster with alkali metals and
the one with all remaining metals analyzed in the samples,
within which Ca/Zn and Mg/Cu/Fe created two sub-clusters,
respectively. Thus, in general CA indicated similar correla-
tions among variables as PCA. Nevertheless, CA did not find
close correlation of Fe/Ca like PCA was.
Dendrogram in Fig. 4 shows 3 main branches (clusters)
of samples according to their metal patterns. Within cluster
I that corresponded to the PCA group I, two sub-clusters
could be seen: one that gathered all sugar beet samples, and
another that clearly separated extracted from dried sugar beet
samples. Cluster II analogue to the PCA group II rounded
all molasses samples. All five sugar samples and soil sample
formed the distinct cluster III on the CA dendrogram corre-
sponding to the PCA group III.
Conclusion
This paper represents the first attempt to elucidate the
correlations in metal patterns in the chain soil - sugar beet -
sugar beet products and to classify the samples according to
their contents of some macro- and micro-elements. In gen-
eral, it could be seen that the structure of the data set formed
from the analytical results obtained in this study could not be
interpreted easily and satisfactory by the simple Spearman’s
non-parametric test. The interpretation of the sets was more
completely performed by multivariate data analysis tech-
niques like principal component analysis, PCA, and cluster
analysis, CA, developed for dealing with strongly collinear
and numerous variables. However, a color map used for
graphical presentation of the correlation coefficients between
the samples was the only method that implied association
between metal contents in sugar beet and soil. PCA and CA
showed the same sample classification, but they indicated
132 B. Škrbić et al.

Na

Ca

Zn

Mg

Cu

Fe

0 2 4 6 8 10 12 14
Linkage Distance

Fig. 3. Cluster analysis dendrogram showing association among the metals investigated.
sugar
beet
beet pulp
extracted
sugar
sugar beet
dried

pulp
molasses
sugar

soil

0 5 10 15 20 25 30
Linkage Distance

Fig. 4. Cluster analysis dendrogram showing association among the investigated samples in sugar beet, its
products and soils.
Metals in Sugar Beet and Its Products
slightly different correlations among the metals content in
the samples. Only PCA reflected the opposite correlation
between Fe and Ca that could be attributed to the influence
of technological process. PCA also showed that the samples
classification was primarily governed by Cu, Zn and Mg,
on one hand and Na and K on the other. The metal pattern
of molasses revealed by both multivariate techniques was
quite different from that found in sugar beet, extracted and
dried sugar beet pulp that formed one group (cluster) with
similar patterns. Presence of the investigated metals in sugar
was uniform during the whole campaign, regardless of the
metal contents fluctuation in sugar beet and its by-products,
indicating the efficiency of purification processes. The metal
contents of sugar and soil were the lowest and similar be-
tween each other, which was the reason of gathering these
quite different matrices into the same group (cluster) by PCA
and CA. Naturally, all of these statements are valid within
the scope of the investigation. Samples from other regions
might behave differently.
Acknowledgements This work was supported by the Serbian Min-
istry of Science and Technological Development as a part of the
Project 152001B: Sources identification and correlations amongst
the elements and organic compounds in abiotic and biotic matrices:
risk analysis and a contribution to the monitoring and improvement
of the environmental status, coordinated by prof. B. Škrbić.
References
Ang, J.F. and Crosby, G.A. (2003). A new look at sugar beet fiber.
Cereal Food World, 48, 238-243.
Beauty, R.D. (1988). “Concept, instrumentation and techniques in
atomic absorption spectrophotometry.” Perkin Elmer, Norwalth,
USA.
Bojinova, P., Georgiev, B., Kabakchiev, I., Krasteva, V., Stanislav-
ova, L., Tchuldjian, H., Welp, G. and Brummer, G. (1996). Har-
monization of the Methods for the Investigation of Heavy Metal
Pollution of Soils and the Standardization of the Assessment
Criteria for Soil Protection, Report No. UBA-FB 96-071, (Terytze,
K. ed.) N. Poushkarov Institute of Soil Science and Agroecology,
Sofia and Institut fur Bodenkunde, Bonn.
Csomos, E., Heberger, K. and Simon-Sarkadi, L. (2002). Principal
component analysis of biogenic amines and polyphenols in Hun-
garian wines. J. Agric. Food Chem., 50, 3768-3774.
de Brujin, J.M. and Bout, M. (1999). Analytical approach to white
sugar quality - anions, cations and their probable origin. Zucker-
industrie, 124, 532-535.
Draycott, P. and Christenson, R. (2003) Nutrients for sugar beet
production: soil-plant relationships, CABI, Oxon, UK.
Fergusson, J.E. (1990). “The Heavy Elements: Chemistry, Environ-
mental Impact, and Health Effects.” Pergamon Press, New York.
133
Golobočanin, D., Škrbić, B. and Miljević, N. (2004). Principal
component analysis for soil contamination with PAHs. Chemom.
Intell. Lab. Syst., 72, 219-223.
He, Z., Yang, X. and Stoffella, P. (2005). Trace elements in agrosys-
tems and impacts on the environment. J. Trace Elem. Med. Biol.,
19, 125-140.
Heberger, K., Csomos, E. and Simon-Sarkadi, L. (2003). Principal
component and linear discriminant analysis of free amino acids
and biogenic amines in Hungarian wines. J. Agric. Food Chem.,
51, 8055-8060.
Heberger, K., Keszler, A. and Gude, M. (1999). Principal compo-
nent analysis of measured quantities during degradation of hydro-
peroxides in oxidized vegetable oils. Lipids, 34, 83-92.
Hickenbottom, J. (1996): Use of molasses in bakery products. AIB
Tech. Bull. 18, 1-6
Hungarian Regulation (2004). Allowable contents of contaminants
in food 17, IV, 16.
Jolliffe, I.T. (1986). “Principal Component Analysis.” Springer-
Verlag, New York.
Kabata-Pendias, A. and Pendias, H. (2000). “Trace Elements in
Soils and Plants.” CRC Press.
Kaiser, H.F. and Rice, J. (1974). Little jiffy, mark IV. Educ. Psy-
chol. Meas., 34, 111-117.
Leblebici, J. and Volkan, M. (1998). Sample preparation for arse-
nic, copper, iron and lead determination in sugar. J. Agric. Food
Chem., 46, 173-177.
Mantovi, P., Bonazzi, G., Maestri, E. and Marmiroli, N. (2003). Ac-
cumulation of copper and zinc from liquid manure in agricultural
soils and crop plants. Plant Soil, 250, 249-257.
Massart, D.L., Vandeginste, B.G.M., Buydens, L.M.C., de Jong, S.,
Lewi, P.J. and Smeyers-Verbeke, J. (1997). “Handbook of Che-
mometrics and Qualimetrics, Part A”. Elsevier, Amsterdam, The
Netherlands.
Mauch, W. and Farhoudi, E. (1979/80). Quality factors in commer-
cial white granulated sugar. Sugar Technol. Rev., 7, 87-171.
Mohamed, A.E. (1999). Environmental variations of trace element
concentrations in Egyptian cane sugar and soil samples. Food
Chem., 65, 503-507.
Morrison, D. (1967). “Multivariate statistical methods”. McGraw-
Hill, New York.
Nan, Z., Zhao, C., Li, J., Chen, F. and Sun, W. (2002). Relations
between soil properties and elected heavy metal concentrations in
spring wheat (Triticum aestivum L.) grown in contaminated soils.
Water Air Soil Pollut., 133, 205-213.
Otto, M. (1999). “Chemometrics Statistics and Computer Applica-
tion in Analytical Chemistry.” Wiley-VCH, Weinheim, Germany.
Ronda, F., Sancho, D., del Alamo, M. and Gomez, M. (2001).
Direct determination of arsenic, cadmium, cobalt, copper, chro-
mium, lead, tin and zinc in white beet sugar using graphite fur-
nace atomic absorption spectrophotometry. Zuckerindustrie, 126,
134
208-212.
Saleh, A.A.H., El-Meleigy, S.A., Ebad, F.A., Helmy, M.A., Jen-
tschke, G. and Godbold, D.L. (1999). Base cations ameliorate Zn
toxicity but not Cu toxicity in sugar beet (Beta vulgaris). J. Plant
Nutr. Soil Sci., 162, 275-279.
Serbian Regulation (1992). Contents of pesticides, metals, metal-
loids and other toxic substances in foods. Official J., 5, 67-85.
Škrbić, B., Čupić, S., Cvejanov, J. and Miljević, N. (2002). The
content of heavy metals and some inorganic cations in white
sugar. Centr. Eur. J. Occup. Environ. Med., 8, 142-145.
Škrbić, B. and Čupić, S. (2004). Trace metal distribution in surface
soils of Novi Sad and Bank Sediment of the Danube River. J. En-
viron. Sci. Health, Part A, 39, 1547-1558.
Škrbić, B. and Ðurišić-Mladenović, N. (2005). Toxic and essential
elements in Serbian sugar beet, molasses and white sugar. Zuck-
erindustrie, 130, 913-917.
Škrbić, B., Gyura, J., Davidović-Torma, T., Vujošević, Z. and Šereš,
Z. (2005a). Determination of metals in sugar industry products.
Ecologica, 12, 289-295 (in Serbian).
Škrbić, B., Ðurišić-Mladenović, N. and Cvejanov, J. (2005b). Prin-
cipal component analysis of trace elements in Serbian wheat. J.
Agric. Food Chem., 53, 2171-2175.
Škrbić, B. and Gyura, J. (2006). Survey on some contaminants in
white sugar from Serbian sugar beet refineries. Food Addit. Con-
tam., 23, 31-35.
Škrbić, B. and Gyura, J. (2007). Iron, copper and zinc in white sug-
ar from Serbian sugar beet refineries. Food Control, 18, 135-139.
Škrbiš, B., and Ðurišić-Mladenović, N. (2007). Principal compo-
nent analysis for soil contamination with organochlorine com-
pounds. Chemosphere, 68, 2144-2152.
B. Škrbić et al.
Škrbić, B. and Onjia, A. (2007). Multivariate analyses of microele-
ment contents in wheat cultivated in Serbia 2002. Food Control,
18, 338-345.
Škrbić, B., Szyrwińska, K., Ðurišić-Mladenović, N., Nowicki, P.
and Lulek, J. (2009). Principal component analysis of indicator
PCB profiles in breast milk from Poland. Environ International.
DOI:10.1016/j.envint
Ubavić, M., Bogdanović, D., Hadžić, V. (1993). Basic chemical
properties of soil of the Vojvodina Province and possibilities of
their contamination with heavy metals. Contemporary Agricul-
ture, 1, 47-51. (In Serbian)
Van der Poel, P.W., Schiweck, H. and Schwartz, T. (1998). “Sugar
Technology.” Dr. Albert Bartens KG, Berlin, Germany.
Vandeginste, B.G.M., Massart, D.L., Buydens, L.M.C., Jong, S.D.E.,
Lewi, P.J. and Smeyers-Verbeke, J. (1998). “Handbook of Che-
mometrics and Qualimetrics, Part B.” Elsevier, Amsterdam, The
Netherlands.
Wakeel, A., Abd-El-Motagally, F., Steffens, D. and Schubert, S.
(2009). Sodium-induced calcium deficiency in sugar beet during
substitution of potassium by sodium. J. Plant Nutr. Soil Sci., 172,
254-260.
Wojtczak, M. and Krol, B. (2002). Content of iron, copper and zinc
in white sugar samples from Polish and other European sugar
factories. Food Addit. Contam., 19, 984-989.
Wolever, T.M.S. and Jenkins, D.J.A. (1997). What is a high fi-
ber diet? In “Dietary Fiber in Health and Disease”, ed. by D.
Kritchevsky & C. Bonfield. Plenum Press, New York, pp. 35-42.
Younas, M., Shahzad, F., Afzal, S., Khan, M.I. and Ali, K. (1998).
Assessment of Cd, Ni, Cu, and Pb pollution in Lahore, Pakistan.
Environ. Int., 24, 761-766.