J Food Sci Technol

DOI 10.1007/s13197-015-1958-1

ORIGINAL ARTICLE

Chemical composition and temperature influence on honey

texture properties
Mircea Oroian 1 & Sergiu Paduret 1 & Sonia Amariei 1 & Gheorghe Gutt 1

Revised: 11 June 2015 / Accepted: 14 July 2015

# Association of Food Scientists & Technologists (India) 2015

Abstract The aim of this study is to evaluate the chemical Among the texture characteristics, hardness (firmness) is one
composition and temperatures (20, 30, 40, 50 and 60 °C) in- of the most important parameters of fruit and vegetables
fluence on the honey texture parameters (hardness, viscosity, (Konopacka and Plocharski 2004). Crispness is the key trait
adhesion, cohesiveness, springiness, gumminess and of cellular, brittle and crunchy food (Taniwaki and Kohyama
chewiness). The honeys analyzed respect the European regu- 2012). Given gelled products such as muscle food, springiness,
lation in terms of moisture content and inverted sugar concen- cohesiveness, adhesiveness and gumminess are significant
tration. The texture parameters are influenced negatively by the properties for the texture evaluation (Akwetey and Knipe
moisture content, and positively by the °Brix concentration. 2012; Stejskal et al. 2011, Chen and Opara 2013).
The texture parameters modelling have been made using the Texture and food structure are inextricably linked; the
artificial neural network and the polynomial model. The poly- micro- and macro- structural composition of foods will deter-
nomial model predicted better the texture parameters than the mine the sensory perception, and any change in structure
artificial neural network. carries the risk of changing perceived texture and violating
consumer expectations (McKenna and Kilcast 1999).
Keywords Honey . Texture . Water . Sugar composition . Texture profiles are curves which monitor and record the
HMF . Modelling spatial or temporal characteristic events of samples during
food texture measurements. Analysis of the profiles of me-
chanical and acoustic measurements is an important aspect
Introduction of food texture research. Texture profile analysis (TPA) sets
up a ‘bridge’ from objective measurement to subjective sen-
Texture is primarily the response of the tactile sense to phys- sation and makes food texture characteristics more predictable
ical stimuli that result from contact between some part of the (Chen and Opara 2013). The TPA profile has been used for
body and the food. The tactile sense (touch) is the primary achieving the textural properties of: commercial cooked meat
method for sensing texture but kinesthetics (sense of move- products (de Avila et al. 2014), fresh cut pineapple (Montero-
ment and position) and sometimes sight (degree of slump, rate Calderon et al. 2008), date flesh (Rahman and Al-Farsi 2005),
of flow) and sound (associated with crisp, crunchy and crackly salmon fillet (Wu et al. 2014), mushroom (Jaworska and
texture) are also evaluate texture (Bourne 2002). Bernas 2010) and sausages (Herrero et al. 2007).
Texture is a key quality attribute used in the fresh and proc- The honey rheological properties have been the purpose of
essed food industry to assess product quality and acceptability. many papers. The rheological properties of honey revealed a
Newtonian behaviour in the great part of the articles published
(Oroian 2012; Oroian 2013, Kayacier & Karaman 2008;
* Mircea Oroian Ramzi et al. 2015), but in some case have been reported as a
[email protected] non-Newtonian fluid (Witczak et al. 2011, Yanniotis et al.
2006; Ahmed et al. 2007). In the last decade, the influences
1
Faculty of Food Engineering, Stefan cel Mare University of Suceava, of different factors on honey rheological parameters have been
Suceava, Romania investigated, as: temperature (Yanniotis et al. 2006), chemical
 J Food Sci Technol

composition (Oroian et al. 2014), shear rate and water content volumetric flask. Add 0.5 ml of Carrez solution I and mix.
(Al-Mahasneh et al. 2013). Add 0.5 ml of Carrez solution II, mix and make up to the mark
To the author knowledge no other study related to the hon- with water. Filter through paper; rejecting the first 10 ml of the
ey texture has been reported. filtrate. Pipette 5.0 ml in each of two test tubes. Add 5.0 ml of
The aim of this study is to evaluate the influence of chem- water to one of the test tubes and mix well (the sample solu-
ical composition and temperature on the honey texture param- tion). Add 5.0 ml of sodium bisulphite solution 0.2 % to the
eters (hardness, viscosity, adhesion, cohesiveness, springi- second test tube and mix well (reference solution). Determine
ness, gumminess, chewiness) using the statistical methods the absorbance of the sample solution against the reference
(Pearson correlation, Principal component analysis) and to solution at 284 and 336 nm.
model the honey texture parameters in function of chemical
composition and temperature using the response surface meth- HMF ðmg=kgÞ ¼ ðA284 −A336 Þ  149:7  5  D=W
odology and artificial neural network.
where A284 – absorbance at 284 nm, A336 – absorbance at
336 nm, 149.7 – constant, 5 – theoretical nominal sample
Materials and methods weight, D – dilution factor, W – weight in g of the honey
sample (Bogdanov 2002).
Materials Glucose, fructose and sucrose were determined using an
HPLC 10 AD VP Shimadzu, with RI detector, according to
6 samples of honeys (H1, H2, H3, H4, H5 and H6) have been the Harmonised methods (Bogdanov 2002). The compounds
purchased from the local producers from Romania. The tex- were separated on an amino column 250 × 4.6 mm i.d. and
ture parameters of honeys, like the rheological parameters, can particle size 5 μm. The samples were prepared as: 5 g of honey
be influenced by the presence of crystals and air bubbles were dissolver in water (40 ml) and transferred quantitatively
(Bhandari et al. 1999; Mossel et al. 2000). Before being used into a 100 ml volumetric flask, containing 25 ml methanol and
they were warmed up to 55 °C to dissolve any crystals, and filled up to the volume with water. The solution was filtered
kept in flasks at 30 °C to remove air bubbles that could inter- through a 0.45 μm membrane filter and collect in sample
fere rheological/textural studies (Oroian 2012). vials. Flow rate 1.3 ml/min, mobile phase: acetonitrile/water
Carrez solution I (15 g of potassium hexacyanoferrate in (80:20, v/v), column and detector temperature 30 °C, sample
water and make up to 100 ml), Carrez solution II (30 g of zinc volume 10 μl. A calibration curve was made for each sugar
acetate and make up to 100 ml), sodium bisulphite solution using standard solutions of different concentrations (0.5–
0.20 %, glucose, fructose, sucrose, acetonitrile and methanol 80 mg/ml). The linear regression factor of the calibration
were purchased from Sigma Aldrich (Germany). curves was higher than 0.9982 for all sugars. Sugars were
quantified by comparison of the peak area obtained with those
Moisture content determination of standard sugars. The results for each sugar were expressed
as g/100 g honey.
The moisture contents of honey samples were obtained by
measuring the refractive index at 20 °C using a digital refrac- Texture profile analysis (TPA)
tometer (Leica Mark II Plus). The sample has been
homogenised. The distilled water has been used as reference The TPA was carried out at 20, 30, 40, 50 and 60 °C with Mark
material (nD) at 20 °C is 1.3330. In order to determine the 10 Texture Analyzer (Mark 10 Corporation, USA) equipped
moisture content, the prism of the refractometer should be with a 50 mm disc probe, the flask diameter was 70 mm. The
clean and dry. The honey is placed on the refractometer prism TPA was operated at a constant speed of 150 mm/min, until a
in order to cover the surface. After 2 min the refractive index depth of 12.5 mm (the honey column had 25 mm). The TPA
has been read. The water content was determined based on a can offer a great number of texture parameters, as: hardness
Chataway Table reading the corresponding moisture content (H), viscosity (V), adhesion (A), cohesiveness (Co), springi-
of the refractive index determined (Bogdanov 2002). ness (S), gumminess (G) and chewiness (Ch) (Chen and
Opara 2013).
Chemical composition determination
Artificial neural network (ANN)
The hydroxymethylfurfural (HMF) content was determined
using White method which is based on the determination of The data used for ANN modelling were used and for polyno-
UV absorbance of HMF at 284 nm. 5 g of honey are placed mial model too. The ANN modelling was made up using the
into a 50 ml beaker. Dissolve the sample in approximately Neuro Solutions trial version (NeuroDimension Inc.,
25 ml of water and transfer quantitatively into a 50 ml Gainesville, FL, USA). The software for ANN has an
J Food Sci Technol

advanced neural networks wizard known as the Neural input neurons and hidden neurons represent the weights.
Builder where it is packed with more powerful nonlinear anal- Each hidden neurons passes its weighted sum through a hy-
ysis methods to build the network Barchitecture^. Network perbolic tangent function as the hidden layer transfer function.
Barchitecture^ is a representation by a graph where the input In this study, 8 neurons in the hidden layer were determined
variables are organized in layers, while the weights that mod- from the lowest value of mean squared errors (MSE) of 0.020.
ulate the combination of the non-linear functions are repre- The choice of one hidden layer is usually sufficient for the
sented as lines connecting units in different layers. The non- purpose of approximation of continuous nonlinear function
linear analysis method chosen in this software is known as the as more hidden layers may cause over-fitting (Madadlou
multilayer perceptron (MLP) where it belongs to the super- et al. 2009; Rai et al. 2005; Torrecilla et al. 2004, Cheok
vised learning neural network and is very important in engi- et al. 2012). Besides the number of neurons in hidden layers,
neering applications. It also provides flexibility and complex- the momentum and learning rate are two important settings for
ity to approximate nonlinear functions to any desired accuracy this MLP network. The recommended momentum and learn-
by changing the number of layers and the number of neurons ing rate values of 0.7 and 0.1 were used in this study to cal-
in each layer (Fathi et al. 2009; Madadlou et al. 2009; Rai et al. culate the weights (Samarasinghe 2007).
2005). The performance of the network is explicitly
measured based on the mean squared error (MSE), de- Polynomial modelling of textural properties
fined as the difference of the output of ANN and a pre spec-
ified external desired signal. The data model regarding the prediction of honey hardness,
The exact MLP network architecture of this study has an viscosity, adhesion, cohesiveness, springiness, gumminess or
input layer of three neurons, one hidden layer of five neurons chewiness according to its chemical composition (fructose,
and an output layer of two neurons. The input layer comprised glucose, sucrose, sugars (the difference between °Brix
of three neurons because of the seven input variables of fruc- concentration –reported as dry matter – and the sum
tose, glucose, sucrose, sugars (the difference between °Brix of fructose, glucose and sucrose), non-sugars compo-
concentration –reported as dry matter – and the sum of fruc- nents (the difference, reported to 100 g, between 100 g and
tose, glucose and sucrose), non-sugars components (the dif- the sum of moisture content and °Brix concentration), mois-
ference, reported to 100 g, between 100 g and the sum of ture content and temperature was made using a 3rd
moisture content and °Brix concentration), moisture content grade polynomial equation with seven variables. The
and temperature. The output layer of one neuron represents measured and predicted values have been compared to
the honey hardness, viscosity, adhesion, cohesiveness, spring- see the suitability of the model. The equation of the model is
iness, gumminess and chewiness. The lines connecting the as given (eq. 10):

X
n
X
n
X
n
X X
n
X
n
X
n
Texture ¼ b0 þ bi x i þ bii x2i þ biii x3i þ bijk xi x j xk þ bij xi x j þ bij x2i x j þ bij xi x2j ð1Þ
i¼o i¼1 i¼1 i< j< k i< j i< j i< j

where Texture is the honey hardness, viscosity, adhesion, co- experimental design, was employed to evaluate the individual
hesiveness, springiness, gumminess or chewiness predicted, and interactive effects of four main controllable variables on
b0 is a constant that fixes the response at the central point of the dye removal efficiency (output response). CCD with four
the experiments, bi – regression coefficient for the linear effect input variables consists of 30 experiments with 30 orthogonal
terms, bij – interaction effect terms, bii – quadratic effect terms two levels full factorial design points (coded as ±1).
and biii – cubic effect terms. The response of predicted honey hardness, viscosity, adhe-
The coded values (Table 1) were applied for statistical cal- sion, cohesiveness, springiness, gumminess or chewiness ob-
culations, according to the following equation (Poroch-Seritan tained from polynomial modelling and ANN was then com-
et al. 2011; Myers and Montgonery 2002): pared using the absolute average deviation (ADD) given in
Eq. (1) (Bas and Boyacı 2007):
zi − z0i ̅n  
xi ¼ ; i ¼ 1; ð2Þ
Δzi ∑Pi¼1 jY i;exp  Y i;cal j =Y i;exp
ADD ¼  100 ð3Þ
P
where: xi denotes the coded level of the variable (dimension-
less value, zi is the actual value of the variable); z0i is the center where Yi,exp. and Yi,cal are the experimental and calculated
point of the variable and Δzi is the interval variation. Central responses, and P is the number of the experimental run.
composite design (CCD), used extensively in polynomial Absolute average deviation (AAD) in this study is a measure
 J Food Sci Technol

Table 1 Correspondence between actual and coded values of design 2014). The moisture content met the threshold requirements of
variables
the Codex Alimentarius (20 % maximum) (Codex
Design variables Symbol Actual values of coded levels StandardCodex standard 2001). The moisture content is in the
same range with those reported by Karabagias et al. (2014);
−1 +1 Kucuk et al. (2007); Escuredo et al. (2012); Rybak-
Moisture content (g/100 g) X1 17.12 18.52
Chmielewska et al. 2013; Szczęsna et al. 2011, Waś
et al.Codex standard 20011a and Waś et al. 2011. The °Brix
Fructose (g/100 g) X2 35.20 44.80
concentration ranged between 80.3 and 81.6°.
Glucose (g/100 g) X3 28.20 36.20
The HMF content, which is an indicator of honey fresh-
Sucrose (g/100 g) X4 0.00 19.60
ness, ranged between 2.1–19.8 mg/kg. This parameter in hon-
Sugars (g/100 g) X5 0.80 11.60
ey is related to its quality and extent of heat processing, but it
Non-sugar substances (g/100 g) X6 1.18 1.28
has not been related to the origin of sample (Anklam 1998).
Temperature (°C) X7 20 60
These low values demonstrate that honey was indeed fresh
and had not been subjected to excessive heat treatments. The
HMF was in the same ranged with the content observed in
of how much the predicated data from models deviates from other studies (Terrab et al. 2003; Zappala et al. 2005; Escriche
the experimental data (Cheok et al. 2012). et al. 2009; Oroian 2012; Szczęsna and Rybak-Chmielewska
1999; Rybak-Chmielewska et al. 2013; Szczęsna et al. 2011,
Statistical analysis Waś et al.Codex standard 20011a and Waś et al. 2011).
All the six honeys had a total content of glucose and fruc-
Statistical analysis was performed using the Statgraphics Plus tose greater than 60 g/100 g honey (Table 2). These values
software system – trial version. The data corresponding to respect the requirements of the EU DirectiveCodex standard
each variable were analyzed by one-factor analysis of variance (2001).
(ANOVA). Multiple comparisons were performed using the
least significant difference test (LSD) and Fisher ratio (F), and Texture profile of honeys
statistical significance was set at α = 0.05. Pearson correlation
was made using SPSS trial version (USA). In the Fig. 1 is presented a typical TPA of a honey at different
temperatures (20, 30, 40, 50 and 60 °C). It can be observed
that the honey response to the stress is decreasing with the
Results and discussions increasing of the temperature, so we can conclude that the
texture profile (and the texture parameters too) are influenced
In the Table 2 are presented the moisture content, °Brix con- negatively by the temperature. As the temperature increases,
centration, HMF content, fructose, glucose and sucrose con- the average speed of the molecules in honey increases and the
centration of the six honeys analysed. amount of the time they spend Bin contact^ with their nearest
The moisture content of the six honeys ranged between neighbours decreases; thus, as temperature increases, the av-
17.12–18.52 g/100 g. Product moisture content of honey is erage intermolecular forces decrease and in conclusion the
widely related to the harvest season and the level of its maturity textural properties are decreasing (Recondo et al. 2006). The
released in the hive. The parameter is highly important for the honey texture properties studied were: hardness (H), viscosity
shelf life of honey during storage. Honey is hydroscopic and (V), adhesion (A), cohesiveness (Co), springiness (Sp), gum-
will pick up moisture from the environment (Karabagias et al. miness (Gu) and chewiness (Ch).
The highest hardness, viscosity, cohesiveness, springiness,
Table 2 Physicochemical parameters of the honeys gumminess and chewiness were observed in the case of honey
H6, while the highest adhesion was observed in the case of
Honey H1 H2 H3 H4 H5 H6
honey H1.
Moisture content (g/100 g) 17.16 17.48 18.52 17.92 18.12 17.12 In the Table 3 are presented the Pearson correlation be-
°Brix concentration (%) 81.5 81.3 80.3 80.9 80.7 81.6 tween the textural properties and physicochemical properties.
HMF (mg/kg) 19.8 10.2 9.5 2.1 9.7 7.9 It seems that all textural properties are strongly positively
Fructose (F) (g/100 g) 37.1 41.8 38.1 44.8 41.8 35.2 influenced one by the other (P < 0.001). The moisture content
Glucose (G) (g/100 g) 33.8 32.7 33.5 28.2 36.2 33.1 has a highly negatively influence on the magnitude of the
Sucrose (S) (g/100 g) 0 1.4 1 1.6 1.9 1.7 textural parameters (P < 0.001) while the °Brix concentration
F + G (g/100 g) 70.9 74.5 71.6 73.0 78.0 68.3 has a highly positively influence on the same parameters
F/G 1.09 1.28 1.14 1.58 1.15 1.06
(P < 0.001). With respect to the chemical compounds, the
other substances presented in the honey (non-sugar defined
J Food Sci Technol

Fig. 1 Texture profile analysis Load [N]

(TPA) of H3 sample 0.6 20 °C
0.5
30°C
40°C
0.4 50°C
0.3 60°C

0.2

0.1

-0.1
0 10 20 30 40 50 60 70 Displacement [mm]
-0.2

-0.3

-0.4

as the difference between 100 g of honey and the sum of °Brix Textural properties modelling
concentration and moisture content) have am important posi-
tively influenced on all the textural parameters (P < 0.001). In order to achieve the evolution of the textural properties
Fructose content is influencing negatively the hardness (H), (hardness (H), viscosity (V), adhesion (A), cohesiveness
viscosity (V), adhesion (A), cohesiveness (Co), gumminess (Co), springiness (Sp), gumminess (Gu) and chewiness
(Gu) and chewiness (Ch) and positively the springiness. The (Ch)) in function of the honey composition and the tempera-
influence of glucose and sucrose on textural parameters is a ture have been used two models: polynomial models and ar-
negligible one. A positive influence has the other sugars pre- tificial networks modelling.
sented in the honey (defined as the difference between the
°Brix concentration and the sum of glucose, sucrose and fruc- Artificial neural network (ANN)
tose) on the textural parameters. The sum of fructose and
glucose and the ratio of these two compounds are influencing The texture parameters prediction using the ANN was limited
negatively the textural parameters. to the selection of a suitable number of neurons in the hidden

Table 3 Pearson correlation of textural and physicochemical parameters of honeys

H V A Co Sp Gu Ch B M F G S F + G F/G Su Non

H 1 0.961** 0.952** 0.923** 0.970** 0.999** 0.999** 0.930** −0.943** −0.675 0.138 −0.256 −0.619 −0.487 0.695 0.987**
V 1 0.851* 0.973** 0.938** 0.970** 0.970** 0.887* −0.901* −0.623 −0.07 −0.063 −0.678 −0.370 0.704 0.980**
A 1 0.815* 0.944** 0.943** 0.943** 0.958** −0.962** −0.565 0.109 −0.462 −0.524 −0.428 0.654 0.896*
Co 1 0.866* 0.937** 0.933** 0.841* −0.856* −0.683 −0.053 −0.172 −0.779 −0.377 0.804 0.952**
Sp 1 0.967** 0.973** 0.959** −0.966** 0.497 −0.016 −0.181 −0.550 −0.296 0.636 0.943**
Gu 1 0.998** 0.925** −0.938** −0.682 0.113 −0.247 −0.646 −0.477 0.716 0.992**
Ch 1 0.926** −0.939** −0.672 0.097 −0.241 −0.649 −0.463 0.717 0.991**
B 1 −0.999* −0.393 −0.051 −0.274 −0.464 −0.208 0.575 0.874*
M 1 0.421 0.039 0.270 0.485 0.232 −0.592 −0.891*
F 1 −0.462 0.409 0.716 0.867* −0.732 −0.731
G 1 −0.079 0.289 −0.841* −0.238 0.116
S 1 0.380 0.30 −0.535 −0.191
F+G 1 0.273 −0.977** −0.698
F/g 1 −0.313 −0.509
Su 1 0.743
Non 1

H hardness, V viscosity, A adhesion, Co cohesiveness, Sp springiness, Gu gumminess, Ch chewiness, B Brix concentration, M moisture content, F
fructose content, G glucose content, S sucrose content, F + G fructose + glucose content, F/G fructose glucose ratio, Su - (the difference between °Brix
concentration – reported as dry matter – and the sum of fructose, glucose and sucrose), Non - non-sugars components (the difference, reported to 100 g,
between 100 g and the sum of moisture content and °Brix concentration)
 J Food Sci Technol

a b
1.20
0.50

Hardness - predicted data 1.00

Viscosity - predicted data

0.40
0.80
0.30
0.60
0.20
0.40

0.20 0.10

0.00 0.00
0.00 0.50 1.00 1.50 0.00 0.20 0.40 0.60
Hardness - experimental data Viscosity - experimental data

c d 0.90
8.000

Cohesiveness - predicted data

7.000 0.85
Adhesion - predicted data

6.000
0.80
5.000
4.000 0.75

3.000 0.70
2.000
0.65
1.000
0.000 0.60
0.000 2.000 4.000 6.000 8.000 0.60 0.70 0.80 0.90
Adhesion - experimental data Cohesiveness - experimental data

e 0.95
f 1.00
0.90 0.90
Springiness - predicted data

Guminess - predicted data

0.85 0.80
0.70
0.80
0.60
0.75
0.50
0.70
0.40
0.65
0.30
0.60 0.20
0.55 0.10
0.50 0.00
0.50 0.60 0.70 0.80 0.90 1.00 0.00 0.20 0.40 0.60 0.80 1.00
Springiness - experimental data Guminess - experimental data

g
0.90
Chewiness - predicted data

0.80
0.70
0.60
0.50
0.40
0.30
0.20
0.10
0.00
0.00 0.20 0.40 0.60 0.80 1.00
Chewiness - experimental data
 J Food Sci Technol

layer as the number of neurons for input and output H ¼ 1:57 þ 0:77 X 1 þ 5:56 X 2 þ 4:64 X 3
layers were already defined from the experimental de-
þ 1:11 X 4 þ 6:28 X 5  3:33 X 7 þ 8:55 X 27
sign. The ANN design, for hardness (H), viscosity (V),
adhesion (A), cohesiveness (Co), springiness (Sp), gum- þ 1:46 X 1 X 7 þ 9:80 X 2 X 7 þ 8:17 X 3 X 7
miness (Gu) and chewiness (Ch), was made using 7
input layers (fructose, glucose, sucrose, sugars (the dif- þ 1:96 X 4 X 7 þ 11:05 X 5 X 7  0:20 X 37
ference between °Brix concentration –reported as dry  3:53 X 1 X 27  24:61 X 2 X 27  20:55 X 3 X 27
matter – and the sum of fructose, glucose and sucrose),
non-sugars components (the difference, reported to 100 g, be-  4:92 X 4 X 27  27:79 X 5 X 27 ð4Þ
tween 100 g and the sum of moisture content and °Brix
concentration), moisture content and temperature), the
V ¼ 0:04 þ 0:09 X 1 þ 0:78 X 2 þ 0:65 X 3
number of neurons in the hidden layer was determined
after running several networks iteratively and observing the þ 0:14 X 4 þ 0:88 X 5  0:62 X 7  0:88 X 27
minimum value in the mean squared errors (MSE) and regres-
sion coefficients. þ 0:26 X 1 X 7 þ 1:69 X 2 X 7 þ 1:41 X 3 X 7
In the Figs. 2 are presented the evolution of exper- þ 0:32 X 4 X 7 þ 1:90 X 5 X 7  0:04 X 37
imental data versus predicted values for ANN predic-
tion of honey hardness (H), viscosity (V), adhesion (A), þ 0:34 X 1 X 27 þ 2:65 X 2 X 27 þ 2:23 X 3 X 27
cohesiveness (Co), springiness (Sp), gumminess (Gu)
þ 0:56 X 4 X 27 þ 3:03 X 5 X 27 ð5Þ
and chewiness (Ch). In the Table 4 are presented the
regression coefficients and ADD values for each
parameter. A ¼ 40:37 þ 17:00 X 1 þ 123:31 X 2 þ 103:15 X 3
The ADD values for the ANN prediction of honey hard-
ness (H), viscosity (V), adhesion (A), cohesiveness (Co), þ 24:67 X 4 þ 139:53 X 5  70:47 X 7  81:09 X 27
springiness (Sp), gumminess (Gu) and chewiness (Ch) ranged þ 29:27 X 1 X 7 þ 204:25 X 2 X 7
between 0.95 and 17.86, while the R2 ranged between 0.9240
and 0.9960. þ 170:78 X 3 X 7 þ 41:34 X 4 X 7
The independent variables importance for the ANN predic-
þ 231:18 X 5 X 7  1:08 X 37  33:51 X 1 X 27
tion of honey hardness (H), viscosity (V), adhesion (A), co-
hesiveness (Co), springiness (Sp), gumminess (Gu) and  234:27 X 2 X 27  195:96 X 3 X 27
chewiness (Ch) are presented in Fig. 3. It seems that temper-
ature have the highest influence on the ANN prediction of the  47:37 X 4 X 27  265:16 X 5 X 27 ð6Þ
textural parameters.
Co ¼ 7:19  2:63 X 1  18:87 X 2  15:81 X 3
Polynomial modelling
 3:80 X 4  21:37 X 5 þ 1:41 X 7  4:61 X 27
The textural parameters (hardness (H), viscosity (V),  0:57 X 1 X 7  4:24 X 2 X 7  3:55 X 3 X 7
adhesion (A), cohesiveness (Co), springiness (Sp), gum-
miness (Gu) and chewiness (Ch)) in function of seven  0:83 X 4 X 7  4:79 X 5 X 7  0:03 X 37
parameters (fructose, glucose, sucrose, sugars (the dif-
þ 1:89 X 1 X 27 þ 13:51 X 2 X 27 þ 11:32 X 3 X 27
ference between °Brix concentration –reported as dry
matter – and the sum of fructose, glucose and sucrose), þ 2:75 X 4 X 27 þ 15:33 X 5 X 27 ð7Þ
non-sugars components (the difference, reported to
100 g, between 100 g and the sum of moisture content
and °Brix concentration), moisture content and tempera- Sp ¼ 5:77  2:04 X 1  14:51 X 2  12:18 X 3
ture) were fitted to cubic model using polynomial models. The  2:93 X 4  16:45 X 5  4:20 X 7  3:98 X 27
equations for each parameter are presented in the equations
presented below: þ 1:69 X 1 X 7 þ 12:10 X 2 X 7 þ 10:13 X 3 X 7

þ 2:44 X 4 X 7 þ 13:70 X 5 X 7  0:05 X 37

ƒFig. 2 Textural properties modeling – experimental vs. predicted data: a þ 1:58 X 1 X 27 þ 11:43 X 2 X 27 þ 9:59 X 3 X 27
hardness, b viscosity, c adhesion, d cohesiveness, e springiness, f
gumminess, g chewiness – rhombus - ANN, square – cubic model þ 2:31 X 4 X 27 þ 12:96 X 5 X 27 ð8Þ
 J Food Sci Technol

Table 4 Artificial neural networks (ANN) methodology statistical Table 5 Polynomial model statistical parameters
parameters
R2 P-value ADD
2
R P-value ADD
Hardness 0.9845 0.003 5.22
Hardness 0.9620 0.001 10.37 Viscosity 0.9814 0.001 3.34
Viscosity 0.982 0.001 4.96 Adhesion 0.9854 0.003 10.34
Adhesion 0.985 0.001 8.82 Cohesiveness 0.9940 0.003 0.38
Cohesiveness 0.970 0.001 1.04 Springiness 0.9699 0.001 1.85
Springiness 0.996 0.001 0.95 Gumminess 0.9841 0.004 6.21
Guminess 0.924 0.001 15.08 Chewiness 0.9870 0.002 6.62
Chewiness 0.945 0.001 17.86

Gu ¼ 1:09  0:35 X 1  2:49 X 2  2:10 X 3  0:51 X 4 In the Table 5 are presented the regression coefficients, P-
values, ADD values for each equation. All proposed model
 2:83 X 5  2:31 X 7 þ 4:81 X 27 þ 1:03 X 1 X 7 are significant (P < 0.05). The coefficients of regressions (R2)
obtained for the above cubic equation indicated that the vari-
þ 6:79 X 2 X 7 þ 5:65 X 3 X 7 þ 1:35 X 4 X 7
ation of hardness (H), viscosity (V), adhesion (A), cohesive-
þ 7:64 X 5 X 7  0:18 X 37  2:00 X 1 X 27 ness (Co), springiness (Sp), gumminess (Gu) and chewiness
(Ch) can be explained by independent variables of 6 of the7
 13:72 X 2 X 27  11:43 X 3 X 27  2:72 X 4 X 27 parameters used (fructose, glucose, sucrose, sugars (the differ-
 15:45 X 5 X 27 ð9Þ ence between °Brix concentration – reported as dry matter –
and the sum of fructose, glucose and sucrose)). The ADD
Ch ¼ 2:23  0:84 X 1  5:96 X 2  5:01 X 3 values ranged between 0.38–10.34, while the R2 ranged be-
tween 0.9814–0.9940. It seems that the cubic model has better
 1:21 X 4  6:77 X 5  3:90 X 7 þ 3:44 X 27
ADD values for hardness, viscosity, cohesiveness, gumminess
þ 1:67 X 1 X 7 þ 11:39 X 2 X 7 þ 9:51 X 3 X 7 and chewiness are smaller than in the case of ANN modelling,
in the case of adhesion and springiness the ADD values
þ 2:28 X 4 X 7 þ 12:85 X 5 X 7  0:17 X 37 of ANN modelling are lower. Keeping into account the
 1:45 X 1 X 27  9:80 X 2 X 27  8:15 X 3 X 27 regression coefficients and ADD values it seems that the
cubic model is a better predictor compared to the ANN
 1:93 X 4 X 27  11:02 X 5 X 27 ð10Þ modelling.

Fig. 3 Parameters importance for

Artificial neural network (ANN) Chewiness

Guminess

Springiness Temperature
Non-sugar
Cohesiveness Sugars
Sucrose

Adhesion Glucose
Fructose
Moisture
Viscosity

Hardness

0 0.1 0.2 0.3 0.4 0.5

Parameters importance, %
J Food Sci Technol

The regression analysis allowed us to observe that X7 and de Avila MDR, Cambero MI, Ordonez JA, de la Hoz L, Herrero AM
(2014) Rheological behaviour of commercial cooked meat products
X5 had significant negatively linear effects on texture param-
evaluated by tensile test and texture profile analysis (TPA). Meat Sci
eters modelling. Just in the case of hardness, viscosity, adhe- 98(2):310–315
sion and cohesiveness have been observed a positively linear Escriche I, Visquert M, Juan-Borras M, Fito P (2009) Influence of simu-
effect on the texture parameters modelling (X7 and X5 too). lated industrial thermal treatments on the volatile fractions of differ-
ent varieties of honey. Food Chem 112:329–338
The interaction between one with another one parameter are in
Escuredo O, Fernandez-Gonzalez M, Seijo MC (2012) Differentiation of
general no significant (P > 0.05). In Fig. 2 is presented the blossom honey and honeydew honey from northwest Spain.
evolution of experimental data versus predicted data of textur- Agriculture 2:25–37
al parameters using the cubic model. Fathi M, Mohebbi M, Ali Razavi SM (2009) Application of image anal-
ysis and artificial neural network to predict mass transfer kinetics
and color changes of osmotically dehydrated kiwifruit. Food
Bioproc Technol. doi:10.1007/s11947-009-0222-y
Conclusions Herrero AM, Ordonez JA, de Avila R, Herranz B, de la Hoz L, Cambero
MI (2007) Breaking strength of dry fermented sausages and their
correlation with texture profile analysis (TPA) and physico-chemical
The honeys analyzed displayed a textural profile influenced
characteristics. Meat Sci 77(3):331–338
by the temperature and °Brix concentration. From all the sev- Jaworska G, Bernas E (2010) Effects of pre-treatment, freezing and fro-
en textural parameters analysed, the adhesion had the highest zen storage on the texture of Boletus edulis (Bull: Fr.) mushrooms.
magnitude and the viscosity the smallest magnitude. Keeping Int J Refrig 33(4):877–885
into account the ADD values, the cubic polynomial Karabagias IK, Badeka A, Kontakos S, Karabournioti S, Kontominas
MG (2014) Characterisation and classification of Greek pine honeys
model predicted better the texture parameters than the according to their geographical origin based on volatiles, physico-
ANN methodology. In the case of ANN methodology, the chemical parameters and chemometrics. Food Chem 146:548–557
temperature has been the parameters with the highest impor- Kayacier A, Karaman S (2008) Rheological and some physicochemical
tance in the prediction. characteristics of selected Turkish honeys. J Texture Stud 39(1):17–27
Konopacka D, Plocharski WJ (2004) Effect of storage conditions on the
relationship between apple firmness and texture acceptability.
Postharvest Biol Technol 32(2):205–211
Kucuk M, Kolayh S, Karaoğlu S, Ulusoy E, Baltaci C, Candan F (2007)
Biological activities and chemical composition of three honeys of
References different types from Anatolia. Food Chem 100:526–534
Madadlou A, Emam-Djomeh Z, Mousavi ME, Ehsani M, Javanmard M,
Sheehan D (2009) Response surface optimization of an artificial
Ahmed J, Prabhu ST, Raghavan GSV, Ngadi M (2007) Physico-chemical, neural network for predicting the size of re-assembled casein mi-
rheological, calorimetric and dielectric behavior of selected Indian celles. Comput Electron Agric 68(2):216–221
honey. J Food Eng 79(4):1207–1213 McKenna BM, Kilcast D (1999) In: Texture in food, vol 2: Solid foods,
Akwetey WY, Knipe CL (2012) Sensory attributes and texture profile of Ed. Woodhead Publishing in Food Science and Technology,
beef burgers with gari. Meat Sci 92(4):745–748 England
Al-Mahasneh MA, Rababah TM, Ma'Abreh AS (2013) Evaluating the Montero-Calderon M, Rojas-Grau MA, Martin-Belloso O (2008) Effect of
combined effect of temperature, shear rate and water content on packaging conditions on quality and shelf-life of fresh-cut pineapple
wild-flower honey viscosity using adaptive neural fuzzy inference (Ananas comosus). Postharvest Biol Technol 50(2–3):182–189
system and artificial neural networks. J Food Process Eng 36(4): Mossel B, Bhandari B, D'Arcy B, Caffin N (2000) Use of an Arrhenius
510–520 model to predict rheological behaviour in some Australian honeys.
Anklam E (1998) A review of the analytical methods to determine the LWT Food Sci Technol 33(8):545–552
geographical and botanical origin of honey. Food Chem 63:549–562 Myers RH, Montgonery DC (2002) DC process and product optimization
Bas D, Boyacı IH (2007) Modeling and optimization II: comparison of using designed experiments. Montgomery Wiley, New York
estimation capabilities of response surface methodology with artifi- Oroian M (2012) Physicochemical and rheological properties of
cial neural networks in a biochemical reaction. J Food Eng 78(3): Romanian honeys. Food Biophys 7(4):296–307
846–854 Oroian M (2013) Measurement, prediction and correlation of density,
Bhandari B, D'Arcy B, Kelly C (1999) Rheology and crystallization viscosity, surface tension and ultrasonic velocity of different honey
kinetics of honey: present status. Int J Food Prop 2(3):217–226 types at different temperatures. J Food Eng 119(1):167–172
Bogdanov S (2002) Harmonised methods of the international honey com- Oroian M, Amariei S, Escriche I, Leahu A, Damian C, Gutt G (2014)
mission (Swiss bee research Centre, FAM, liebefeld, CH-3003 Bern, Chemical composition and temperature influence on the rheological
Switzerland) behaviour of honeys. Int J Food Prop 17(10):2228–2240
Bourne M (2002) Food texture and viscosity: concept and measurement, Poroch-Seritan M, Gutt S, Gutt G, Cretescu I, Cojocaru C, Severin T
2nd edition, Ed. Academic Press, Florida (2011) Design of experiments for statistical modelling and multi-
Chen L, Opara UL (2013) Approaches to analysis and modelling texture resposnse optimization of nickel electroplating process. Chem Eng
in fresh and processed foods – a review. J Food Eng 119:497–507 Res Des 89:136–147
Cheok CY, Chin NL, Yusof YA, Talib RA, Law CL (2012) Optimization Rahman MS, Al-Farsi SA (2005) Instrumental texture profile analysis
of total phenolic content extracted from Garciniamangostana Linn. (TPA) of date flesh as a function of moisture content. J Food Eng
Hull Using response surface methodology versus artificial neural 66(4):505–511
network. Ind Crop Prod 40:247–253 Rai P, Majumdar GC, DasGupta S, De S (2005) Prediction of the viscos-
Codex standard (2001) (Codex alimentarius) 12–1981, rev. 2 revised ity of clarified fruit juice using artificial neural network: a combined
codex standard for honey, effect of concentration and temperature. J Food Eng 68(4):527–533
 J Food Sci Technol

Ramzi M, Kashaninejad M, Salehi F, Mahoonak ARS, Razavi SMA Taniwaki M, Kohyama K (2012) Mechanical and acoustic evaluation of
(2015) Modeling of rheological behavior of honey using genetic potato chip crispness using a versatile texture analyzer. J Food Eng
algorithm–artificial neural network and adaptive neuro-fuzzy infer- 112(4):268–273
ence system. Food Biosci 9:60–67 Terrab A, González AG, Díez MJ, Heredia FJ (2003) Characterization of
Recondo MP, Elizalde BE, Buera MP (2006) Modelling temperature Moroccan unifloral honeys using multivariate analysis. Eur Food
dependence of honey viscosity and of related supersaturated model Res Technol 218:88–95
carbohydrate systems. J Food Eng 77(1):126–134 Torrecilla JS, Otero L, Sanz PD (2004) A neural network ap-
Rybak-Chmielewska H, Szczęsna T, Waś E, Jaśkiewicz K, Teper D proach for thermal/pressure food processing. J Food Eng
(2013) Characteristics of Polish unifloral honeys. IV. Honeydew 62(1):89–95
honey, mainly abies alba L. J Apicultural Sci 57(1):51–59 Waś E, Rybak-Chmielewska H, Szczęsna T, Kachaniuk K, Teper D
Samarasinghe S (2007) Neural networks for applied sciences and engi- (2011) Characteristics of Polish unifloral honeys. III. Heather honey
neering –from fundamentals to complex pattern recognition. Taylor (Calluna vulgaris L). J Apic Sci 55(1):129–136
& Francis Group, Boca Raton Witczak M, Juszczak L, Gałkowska D (2011) Non-Newtonian behaviour
Stejskal V, Vejsada P, Cepak M, Špicka J, Vacha F, Kouril J, Policar T of heather honey. J Food Eng 104(4):532–537
(2011) Sensory and textural attributes and fatty acid profiles of fillets Wu D, Sun D-W, He Y (2014) Novel non-invasive distribution measure-
of extensively and intensively farmed Eurasian perch (Pera ment of texture profile analysis (TPA) in salmon fillet by using
fluviatilis L.). Food Chem 129(3):1054–1059 visible and near infrared hyperspectral imaging. Food Chem 145:
417–426
Szczęsna T, Rybak-Chmielewska H (1999) Determination of
Yanniotis S, Skaltsi S, Karaburnioti S (2006) Effect of moisture content
hydroksymethylfurfural (HMF) in honey by HPLC. Pszczelnicze
on the viscosity of honey at different temperatures. J Food Eng
Zesz Nauk 43:219–225
72(4):372–377
Szczęsna T, Rybak-Chmielewska H, Waś E, Kachaniuk K, Teper D Zappala M, Fallico B, Arena E, Verzera A (2005) Methods for the deter-
(2011) Characteristics of Polish unifloral honey. I. Rape honey mination of HMF in honey: a comparison. Food Control 16:273–
(Brassica napus L. var. Oleifera Metzger). J Apic Sci 55(1):111–119 277