PIII
ELSEVlER
Food Qvalig and Pwfennce Vol. 9, No. 4, pp. 237-242, 1998 0 1998 Elsevicr Science Ltd. All rights reserved Printed in Great Britain 0950-3293/98 S 19.00 + 0.00
SOSSO-3293(98)00002-O
QUALlTYAllRlBUTESOFPARlZASALAMl
AS
INFLUENCED
BYTHEADDITION OFMECHANICALLY DEBONEDCHICKEN MEAT S. N. Raphaelides,* S. Grigoropoulou & Il. Petridis Department
of Food Technology,
T.E.I.
of Thessaloniki,
(Accepted
17 December
of
ABSTRACT
54101
Thessaloniki,
Greece
products
containing
MD
1997)
the
meat
meat
since
mechanical deboning results in cellular disruption, protein denaturation and increased lipid and heme oxidation. Thus, the colour of MD meat is a dull brownish red
The e$ect of substituting beef with mechanically deboned chicken meat (MDC) on the quality characteristics of pariza salami was investigated. Analysis of variance showed that up to 39% substitution of beef by MDC caused no adverse e$ects to the sensory variables of pariza samples. Redundancy analysis indicated that sensory and physical attributes of par&a samples made of various ratios of beef to MDC were greatb injuenced by a number of compositional variables related to texture and colour. The use of a limited number of instrumental measurements proved to be suj%ient to characterize the quality projle of pariza salami. 0 1998 Elsevier Science Ltd. All rights reserved
if pigment oxidation has occurred, which may affect the colour of MD meat-containing meat products. Another sensory attribute that has to be controlled in products containing MD meat is that of a grainy or gritty texture. The extent to which this occurs depends upon the size and amount of bone particles contained in the final meat product (Field, 1988). Regarding the functional properties of MD meat in comparison to those of hand-deboned meat the information available in the literature is conflicting. Thus, as far as the emulsion ability is concerned, it has been reported (Schnell et al., 1973) that the stability of emulsions containing MD meat is poor and can be improved by adding caseinates, whereas Pisula and Rejt (1979) and Field (1988) reported superior emulsion properties of MD meat relative to hand-
INTRODUCTION
deboned meat. The majority of the researchers, such as Pisula and Rejt (1979) and McMillin et al. (1980)) generally agree that the water-holding capacity of MD meat is better than that of hand-deboned meat, possibly due to the higher pH value of the MD meat. So far, pariza containing MDC meat is not commercially available. The present study was initiated to investigate the potential use of MDC meat in the manufacture of pariza-type salamis. Various proportions of
Pariza is a kind of pasteurized salami that is prepared mainly from beef and/or pork meat and contains small shredded meat fragments embedded in the main meat paste. It is considered to be one of the most popular cooked meat products in Greece. Commercially, either mechanically deboned beef (MDB) or pork (MDP) meat is added to the meat paste for the production of pariza,
beef and MDC meats were experimental pariza salamis.
in amounts up to 30% for the MDB and up to 10% for the MDP based on the total meat content. The use of lower percentage of MDP than of MDB is due to the fact that higher amounts of MDP cause more serious adverse effects to the sensory characteristics of pariza than do
MATERIALSAND
respective amounts of MDB. Although the idea of using mechanically deboned meat in meat product formulations is very attractive due to its much lower cost compared to that of hand-boned meat, its use is still fairly limited. This is attributed to considerable deterioration of the sensory characteristics *To whom correspondence
PO Box 14561,
mixed
together
in
the
METHODS
Materials Six batches of pariza salami were prepared, which differed in the proportion of beef to MDC, based on the total quahtity of meat present in the final product. That is, the percentage composition of meat ranged from 0% MDC to 100% beef in the first batch to 90% MDC to
should be addressed. 237
238
S. JV. Rafihaelides et al.
10% beef in the sixth one (Table 1). In all batches, the frozen beef and the frozen MDC used were from the same lots. The beef was from the forequarter of the carcass and the MDC was from various parts of the chicken except the breast. Both types of meat were obtained from regional meat and poultry processing plants.
Sample preparation Every batch contained as main ingredients meat, lard and water in proportions shown in Table 1 and was prepared as follows. The frozen beef was cut into small pieces and added to the cutter. There it was comminuted to even smaller pieces for a short time. Then the appropriate amounts of phosphates and salt were added. The grinding was continued until the mixture was reduced to a paste. While mixing, part of the required quantity of water was added in portions. When the temperature of the paste had risen over 5°C a mixture of comminuted lard and MDC trimmings was added followed by the addition of the remaining amount of water and the appropriate spices. When the paste was ready, it was filled into synthetic casings to the required volume by means of a WEMAG (Germany) vacuum filler. The formed salamis were left for at least 2 h at ambient temperature before cooking. They were cooked using steam in a specially designed chamber under controlled conditions. The cooking programme applied was: initially at 50°C (chamber temperature) for 20min, then at 60°C (chamber temperature) for 30min and finally the chamber temperature was set to 78°C until the internal temperature of the samples at the centre reached 72°C within 15 min. Then they were held for an extra 10 min at that temperature before they were cooled down and refrigerated at 4°C.
Chemical and physical analyses Moisture content was determined using the direct water distillation method (AOAC, 1990), the fat content was determined with the Soxhlet method (AOAC, 1990), and the protein content with the Kjeldahl method (AOAC, 1990). The total ash content was determined by igniting the charred sample in a muffle furnace at 525°C until a TABLE 1. Pariza salami formulations
Basic ingredients: Samples Beef (kg) MDC trimmings (kg)
1 5.0
-
Other ingredients: expressed as of the basic ingredients Lard 50.0 Sodium chloride 4.8 Polyphosphates 0.8 Sodium caseinate 4.0 Potato starch 4.0
2 4.5 0.5
3 3.5 1.5
4 2.5 2.5
5 1.5 3.5
6 0.5 4.5
percentage of the total weight Water Sodium nitrite Sodium ascorbate Sugar Spices
50.0 0.04 0.2 0.6 1.0
constant weight was reached (LFRA, 1978). The collagen content was determined with the hydroxyproline method (British Standards Institution, 1979). Determinations were replicated three times. Colour was determined on both surfaces of a salami cut in half perpendicular to the long axis. A colourimeter made by the company Dr Lange, model Microcolor, Germany was used. L, a and b values were measured. Determinations were replicated ten times, i.e. 5 salamis twice, Mechanical parameters were determined using an INSTRON UTM, model 1140 (Instron Ltd, U.K.). Three different series of determinations were made: Compression at 10% deformation of the initial sample height: The measurements were obtained on cylindrical samples; 21 mm diameter by 21 mm height. Samples were taken from the centre of both halves of a salami cut in half perpendicular to the long axis, using a metal borer of the same inner diameter as that of the samples. Parameters measured were: force at peak height of 10% compression (FF); recovery force (FR) measured after the sample was relaxed under constant compression for a time period equal to that elapsed for the completion of the 10% compression; percentage recovery (Recovery) to the initial height of the compressed sample at the point of the FR measurement. Compression at 80% deformation. Samples used had the same dimensions as described above. Force recorded at the first significant break of the sample indicated by the compression-deformation curve obtained and it was designated as FB. Figure 1 presents a typical compression curve of pariza salamis obtained at 80% deformation of the sample. Cutting shear, designated as FC, using a WarnerBratzler attachment of cylindrical samples, 2 1 mm in diameter and 50 mm in length. In both series of compression measurements, a 36 mm flat plate probe was used as the compression attachment to the Instron measuring cell. In all series of determinations the cross head and the chart speeds of the instrument were lOmm/min and 50mm/min, respectively. The determinations were replicated four times for each sample treatment.
Sensory evaluation Sensory testing was carried out using an experienced 15member panel selected from the Department’s staff. The panel’s experience on sensory assessment was acquired through their regular participation, for at least 3-4 years, in similar kinds of projects concerning meat products, cheeses, processed fruits, etc. A balanced incomplete
Quality Attributes of Pariza Salami
239
variable were performed according to Student-NewmanKeuls test procedure (Zar, 1984). The structural complexity of chemical, physical and sensory variables on MDC samples was tested using redundancy analysis as described by Ter Braak (1987) and performed by the CANOCO statistical software (Ter Braak, 1988). In simple terms, the set of chemical variables was regressed against the major axes 1 and 2 composed by the principal component analysis between samples and both sensory and physical variables.
RESULTS
L DistanceF’IG. 1. Typical 80% compression curve of experimental pariza salamis. FB is explained in Table 2.
block design was used and each panelist
assessed four
samples, which were in the form of slices. The technique used for the sensory assessment was that of the unstructured scaling: the panelists were asked to taste the samples in two runs. In each run a separate attribute was assessed, i.e. texture cohesiveness and colour. They assessed cohesiveness by biting a slice of the sample using their front teeth. They judged cohesiveness from the magnitude of the effort they made to completely penetrate the sample, i.e. the greater the effort the more cohesive the sample was. The colour was assessed by comparing the degree of redness between slices of the various samples. They recorded their evaluation by drawing a vertical line for each sample across a horizontal line 15cm long at the point that best reflected their perception of the magnitude of that attribute. The left end (Ocm) of the line was marked for the texture as absolutely non-cohesive and for the colour as faintly red. The right end (15 cm) was marked for the texture as very cohesive and for the colour as intensively red.
Statistical
AND
DISCUSSION
The mean values of the compositional and physical parameters of the pariza samples are shown in Figs 2 and 3. It can be seen that there is a reduction in the values of the composition variables as the percentage of added MDC meat was increased, with the exception of the fat content which increases due to the presence of bone marrow in the MDC. The values of the physical variables decreased with the increase of the MDC addition. Despite the fact that MDC contains more hemoglobin (due to the presence of marrow) than the hand-deboned poultry meat, the redness of the pariza samples was reduced with the increase of the percentage of MDC added. This occurred because beef contains more myoglobin, the main colouring substance of meat, than the chicken meat. All mechanical variables of pariza samples were affected by the percentage addition of MDC, which shows that both the presence of higher amounts of connective tissue and protein induced a more rigid texture whereas the presence of higher amounts of fat in the mixtures of beef and MDC pariza samples created a softer texture. Moreover, the presence of a higher amount of water in the first pariza sample made exclusively from beef (control) facilitated the diffusion of the
I-1
21
analysis
Potential differences between treatments for each sensory variable were tested using a balanced incomplete block ANOVA, which included six treatments of MDC addition, four treatments per panelists, 15 panelists and ten replicates per treatment (t=6, k=4, b= 15, r= 10; Cochran and Cox, 1957). The.Minitab statistical software was used for this two-way analysis of variance based on the Generalized linear model with one fixed factortreatments and one random factor-panelists. Whenever the significance of ANOVA results was met, multiple comparisons between pairs of adjusted means of each
-2'.'.""'."'."""'. 10 20 0
30
40
% MDC
50
60
70
80
00
content
FIG. 2. Influence of MDC addition on the chemical composition of the pariza samples. Variables have been equally scaled after data standardization (subtracting their mean and dividing by the standard deviation).
240
S. JV. Raphaelides et al.
0
10
20
30 %
40
50
60
70
80
sample into their mouth, and before they completely crush it with their teeth, it appears that their assessment scores are more closely related to the data obtained with the 10% compression than with the 80% compression tests. Strong correlations (r > 0.900) were obtained between sensory cohesiveness and the Instron 10% compression variables (FF, FR and % recovery; Table 2). These variables are related to the linear (reversible) elastic behaviour shown by the samples and are an indication of the degree of textural rigidity of the samples. Objective colour values correlate strongly with both sensory cohesiveness and objective textural variables, and this is obvious since both depend on the meat type content of pariza, i.e. the more beef contained the redder and tougher it becomes. As would be expected for the chemical attributes, protein, moisture, ash and collagen strongly correlate with the 10% compression Instron variables and less with the 80% Instron compression variables, since all these attributes are related to the protein matrix formation. The matrix is a composite system where particle-forming components such as fat and fibres are dispersed in the network formed by the protein macromolecules and the water. The rigidity of the matrix depends on the presence of permanent and temporary cross-links as well as on the presence of salts, indicated by the ash content, since electrolytes affect the conformation of the macromolecules. It is well established that pH and ionic strength greatly influence the structural characteristics of biopolymer systems by affecting the creation of secondary bonding, i.e. electrostatic interactions in their networks. Besides, in the case of meat systems, the majority of cross-links tend to be constructed from cooperative regions of hydrogenlcationit bonding or hydrophobic patches (Ross-Murphy, 1983). By contrast the aggregation of colloidal particles such as fat are not involved in such co-operative contribution. Hence, the higher the fat content of the system, the less the number of cross-links formed and the weaker the structure becomes, which easily breaks under quite low strains.
90
MDCcontent
F’IG. 3. Influence of MDC addition on the physical profile of the pariza samples. Variables have been equally scaled after data standardization (subtracting their mean and dividing the standard deviation). Abbreviations are as explained Table 2.
by in
protein out of the fibres and the formation of a more cohesive matrix in comparison to the other samples which contained less water and more fat. Besides, when the emulsification process is not sufficient it results in segregation of the lipid from the protein matrix and the formation of a less tenacious structure (Field, 1988). The characterization of the texture of pariza samples, in the present study, was made using mechanical variables other than those normally employed by the General Foods Texture Profile Analysis (TPA) technique (Bourne, 1978). This practice was established after a series of preliminary experiments using the TPA method was carried out and it was realized that the data obtained were poorly correlated with those obtained by the sensory techniques. The 10% compression employed in the present study, instead of the 80% compression employed by the TPA method, guarantees that the linear elastic limit of the sample structure was not exceeded, hence the texture is not irreversibly affected as with the TPA technique. Since the panelists assess most of the textural characteristics as soon as they introduce the TABLE
2. Correlation matrix between sensory, physical and chemical variables of parka than the critical value so,os,4 = 0.811 are statistically significant at 0.05 probability level
Colour Colour a/b FF FR FC FB Reco Moisture Fat Ash Protein Collagen
Cohe
Colour
Cola/b
0.947 0.940 0.932 0.928 0.812 0.799 0.924 0.839 -0.876 0.884 0.800 0.939
0.867 0.824 0.816 0.681 0.696 0.819 0.742 -0.757 0.785 0.701 0.860
0.987 0.985 0.930 0.901 0.937 0.953 -0.978 0.943 0.911 0.900
FF
FR
FC
FB
1.ooo 0.965 0.942 0.950 0.973 -0.977 0.933
0.968 0.943 0.951 0.973 -0.978 0.931
0.985 0.888 0.993 -0.953 0.878
0.843 0.985 -0.915 0.857
0.935
0.935
0.962
0.974
0.937
0.936
0.888
0.901
samples.
Correlation
Reco
Moisture
0.888 -0.889 0.802 0.785 0.933
-0.961 0.898 0.971 0.894
Fat
-0.968 -0.945 -0.845
coefficients
greater
Ash
Protein
0.931 0.809
0.846
Abbreviations: Cohe (Cohesiveness), Colour, sensory variables assessed by the unstructured scaling technique, Co1 a/b (Colour a/b), objective colour parameter, FF (Compression force), FR (Recovery force), FC (Cutting shear), FB (Break force), Reco (Recovery).
Quality Attributes of Pariza Salami When
10%
temporary majority 80%
compression
cross-links
is employed,
ruptured
the number
is fairly small,
of them can be easily reformed.
compression,
irreversibly
are ruptured
destroyed.
Since
the
10%
compression differences
compression
measurements between
texture
salamis
is
of pariza
of cross-links,
are more
the
measurements,
both perma-
and the structure
affected by the nature and the number trace
In the case of
most of the cross-links,
nent and temporary,
of
and the
sensitive
than
the
is the
3. F values and probabilities
TABLE
which are coarser and hence
Variable
F vahep
The fat content
correlates
fat content
is a characteristic
renders its texture The relation chemical
strongly but negatively
This was expected of the MDC
and to the
samples is more clearly
shown in Fig. 4.
Cohesiveness
54.88
1 >3 >4 >5 >6 2 11.2 10.4 9.5 7.9 5.2 4.1 . . .. . . . .
Colour a/b
40.08
3 =2 =l >4 >5 >6 9.9 9.6 8.9 6.9 5.8 3.2
.. ...
as they
were
MDC
the control, taining
the effect of the panelists’ 3 shows
the
results
evaluation of the
of the
Moisture analysis
ANOVA
analysis of the means for the sensory attributes
variance
and colour. of equality
of means
1, 2 and 3 are significantly than
exists in cohesiveness.
1 and 2 do not significantly
fer as also is the case between
1 and 3. However,
dif-
samples
more cohesive than sample 4,
sample
5, which
in turn
is more
cohesive than sample 6. On the other hand, samples and 3 have the same red colour,
MDC
and ash were rejected
inflation
factor
which
1, 2
is significantly
criteria
of variable
selection
regression coefficients
cohesiveness
and
ultimately
reduces
For the same sensory attributes the percentage
colour,
the
the sensory intensity.
Fig. 5 shows the effect of
Moisture plot between the chemical
Fat and sensory/physical
as values of their greatly
(Ter Braak,
the critical 1988). Three
for the redundancy
of variables,
coefficients
chemical
constituents
by protein
1988).
Braak,
Figure ables,
99.2%
6 gives a global
both
objective
The
and causative
overall (chemical)
of the total variation
Ash pattern
of the parka
dependent variable regarding
on the six pariza
with lines with the same direction
Protein
and fat followed
view of the effect of all vari-
and subjective,
samples based on the results of the redundancy Variables
ana-
t values of the
were the most influential (Ter
by
beef
that collagen
increasing
of MDC meat added to the pariza samples
FIG. 4. Draftsman Table 2.
con-
and interset correlation
the first two major axes.
content
from
that pariza
with axes 1 and 2, which indicated
(sensory and physical)
30%
from the redundancy
exceeded
importance
lysis, are forward
ple 5, which in turn is more red than sample 6. It follows MDC
up to
difference
could not be distinguished
due to high autocorrelation
relation explained
in both
For
and this is an indication
more red than sample 4, and this is more red than samthat
panelists.
value of 20 which we adopted
of samples
and this more
the
the panelists from the pariza made of 100% two-way
Overlap
by
there was no significant
up to 30%
Table
Cohesiveness
assessed
addition
samples,
cohesiveness
conclut3ion
with
meat,
of the sensory and physical variables
Regarding
of
since higher
to be softer and more pale in colour.
ones of the six pariza
Overall
value
less sensitive due to the excessive strains employed. all the physical variables.
of significance
ANOVA results on the sensory variables. The overall conclusion on differences between treatment means is based on the SNK paired comparisons test
to
80%
241
analysis.
show positive
C&SW
samples. Abbreviations
are as explained in
242
S. N. Raphaelideset al.
10 : e9 g 6 .f a 7 5 06
f.,.,““,‘,“,“““,
0
10
20
30
40
% MDC
50
60
70
60
90
content
correlation, and those with inverse direction show negative correlation. The intensity of this correlation increases as the angle between the variables diminishes. Two variables with an angle of 90” are totally not correlated. Samples positioned close to a line of a variable show strong relationship. Thus, fat is indicative of samples with high MDC content and inversely related with the bulk of the rest of the variables. Strong positive correlation exists between collagen, cohesiveness, colour-a/b and % recovery, and protein strongly and positively correlates with cutting shear and break force. These variables are also indicative of samples with low MDC content. As a conclusion, up to 30% MDC meat can be added to beef pariza salami. It does not cause adverse effects regarding the texture and the colour of the product since they were not significantly different than those of the 100% beef pariza.
REFERENCES AOAC (1997) OJ’icial Methods of Analysis. Association of Official Analytical Chemists, Washington D.C., USA. Boume, M. C. (1978) Texture profile analysis. Food Technol. 32, 2
0
10
20
30 %
40
60
60
70
60
60
MDC content
Standards
Institution
(1979) Determination of hydroxy-
proline. B.S. 4401, Part II, UK
FIG. 5. Influence of MDC addition on the sensory profile of the pariza samples. Vertical bars represent the standard errors based on the ANOVA’s error mean square.
4
;F l50%
t
FAT
PROT
62-66.
British
.
90%
FIG. 6. Biplot based on redundancy analysis of pariza MDC content and sensory/physical profile with respect to three chemical variables. The lines for sensory/physical and chemical variables display the approximate correlation coefficients between these two sets of variables. Abbreviations are as explained in Table 2.
Cochran, W. G. and Cox, G. M. (1957) Experimental Designs, 2nd edn. John Wiley and Sons, Chichester. Field, R. A. (1988) Mechanically separated meat, poultry and fish. In Edible Meat By-Products. Advances in Meat Research, ed. A. M. Pearson and T. R. Dutson, Vol. 5, pp. 83-l 19. Elsevier Applied Science, London. LFRA, Leatherhead Food Research Association (1978) Determination of Ash. Analytical Methods Manual, 2nd edn. LFRA, UK. McMillin, K. W., Sebranek, J. C., Rust, R. E. and Topel, D. G. (1980) Chemical and physical characteristics of frankfurters prepared with mechanically processed pork product. 3. Food Sci., 45(6), 14551459, 1462. Pisula, A and Rejt, J. (1979) The influence of the addition of mechanically deboned meat (MDM) on the physicochemical properties of meat model blends. In Proceedings of the European Meeting of Meat Research Workers, No 25, 11.5:863-l 1.5:868. Ross-Murphy, S. B. (1983) Rheological Methods. In Biophysical Methodr in Food Research, ed. H. W.-S. Chan, pp. 139-199. Critical Reports in Applied Chemistry, SCI, London. Schnell, P. G., Nath, K. R., Darller, J. M., Vadehra, D. H. and Baker, R. C. (1973) Physical and functional properties of mechanically deboned poultry meat as used in the manufacture of frankfurters. Poultry Sci. 52, 1363-1369. Ter Braak, C. J. F. (1987) Unimodal models to relate species to environment. Agricultural Mathematics Group, Wageningen. Ter Braak, C. J. F. (1988) CANOCO: A FORTRAN program for canonical community ordination by partial detrended and canonical correspondence analysis, principal component analysis and redundancy analysis. Agricultural Mathematics Group, Wageningen. Zar, J. H. (1984) Biostatistical An&is, 2nd edn. Prentice Hall, London.