- Email: [email protected]
Dried sausages fermented with Staphylococcus xylosus at different temperatures and with different ingredient levels — Part III. Sensory evaluation
Meat Science, Vol. 41, No. 2, pp. 211-223, 1995 Copyright 0 1995 Elsevier Science Limited Printed in Great Britain. AU rights reserved 0309-1740/95 $9.50+0.00
ELSEVIER
0309-1740(94)00068-9
Dried Sausages Fermented with Staphylococcus xylosus at Different Temperatures and with Different Ingredient Levels - Part III. Sensory Evaluation
L. H. Stahnke Department
of Biotechnology, Technical University of Denmark, Building 221, Lyngby 2800, Denmark
(Received 15 January 1994; revised version received 3 October 1994; accepted 6 October 1994)
ABSTRACT Sausages with added Staphylococcus xylosus were fermented at dtjerent temperatures and with dtyerent added levels of salt, glucose, nitrite, nitrate and Pediococcus pentosaceus in accordance with a six factor fractional design. The odour of the sausages was evaluated by a quantitative descriptive method with ten descriptors and by gas chromatography olfactometry. The sensory profile was correlated to the experimental design and the volatile compounds by partial least squares analysis. Also, the effects of temperature and different ingredients on the strength of the individual descriptors were tested using multiple linear regression and analysis of variance. The study showed that salami odour was more pronounced in sausages fermented at low temperature than in sausages fermented at high temperature and added nitrite, glucose and P. pentosaceus. High temperature sausages had a more sour and cheesy aroma, but were less fatty and sourdough-like. Salami odour was correlated with the presence of ethyl esters and 2-alkanones as well as with high numbers of Staphylococcus xylosus. 2- and 3-methylbutanal seemed to be influential as well. The sourish note to salami odour was especially caused by acetic andperhaps butanoic acid, the cheesy note by 2-methylpropionic, butanoic and 3-methylbutanoic acid.
INTRODUCTION The aroma of fermented sausage is of great importance to the sausage quality, but little is known about the components that comprise the aroma profile and the ones of more significance to flavour. Neither is much known about the influence of sausage ingredients and fermentation parameters on the final sausage aroma. 211
212
L. H. Stahnke
Several studies have shown that the volatile profile of fermented sausage is very complex, including components from many classes of compounds (Berger et al., 1990; BerdaguC et al., 1993; Johansson et al., 1994; Stahnke, 1994). The previous paper has already given a short introduction to the subject. The unique aroma of most foods is not due to one compound, but usually to many compounds of diverse structures, blended in a specific ratio. There are two approaches to evaluate the components of most importance to final flavour. One approach is to pinpoint the more potent aroma components in the volatile fraction by gas chromatography olfactometry (also called effluent sniffing) and more elaborate, by aroma extract dilution (Grosch, 1993; Acree et al., 1984). Over the years, many workers have employed effluent sniffing and have in this way detected some of those components giving certain foods their characteristic aroma, e.g. 1-octen-3-01 and I-octen-3-one in mushrooms (Fischer & Grosch, 1987). A second approach is to produce a descriptive sensory profile of the food in question and employ different multivariate statistical methods to study the correlation between the volatile profile and the sensory data. In this way, it may be possible to get an idea of which component or class of components that are responsible for specific aroma notes (Chien & Peppard, 1993; Aishima & Nakai, 1991; Martens, 1982). The purpose of the present study was to investigate the effect of fermentation temperature and some basic sausage ingredients on the sensory profile of a fermented sausage added Staphylococcus xylosus. The influence of various parameters, fermentation temperature, salt, nitrite, nitrate and glucose concentration, and the addition of another starter culture, Pediococcuspentosaceus, was tested in a fractional factorial design. Also, sensory data and volatile profile were correlated using multivariate statistics. MATERIALS
AND METHODS
Detailed information on experimental design, sausage production and chemical and bacterial analyses, and on the analysis of volatiles, are given in the two preceeding papers. Table 1 sums up the experimental design and the factorial levels. Sensory analyses
The odour of the sausages was evaluated twice by a trained panel of 11 members using a quantitative descriptive method with 10 descriptors and an unstructured line scale of 150 mm. The line had two anchor points at low and high level, respectively (15 mm from the end). The list of descriptors was developed during training sessions by the panel members. The panel was trained once a week for a total of 3 months prior to the testing sessions. The test samples consisted of approximately 6 g of ground sausage equilibrated in closed 50 ml containers for 1 h at 37°C. The panel was presented with a reference sample at the beginning of each session. The effluent from the gas chromatograph was evaluated once for all of the sausages except the ‘centrepoints’. The effluent evaluation was done by one
Dried sausages -
III. Sensory evaluation
213
TABLE 1 Experimental Design” Factor
Batch no. Fermentation temperature
NaNOz
KNOs
0
A B C D E F G H I J K L M N 0 P
0 0
0 0 0 0 0
Salt
1 0 0 1 0 0 1
1 0 1 1 0 1 0 1
0 1 1 0 1 0 1
6 1 1 1 1 0 0 0 0
1
1
0
1 1
1 1 1 1 1 1 1
15°C 25°C
Glucose
4 1
0
: Factor levels: Low High
Pediococcus pentosaceus
Ob
4.2 x 107cm/g
50ppm
300 ppm
0
0
1.5%
0.20%
1.0%
3.5%
“O= low level, 1 = high level, f = centre level. bPediococcuspentosaceus was not added to the minces but a natural flora of 2.5 x lo3 cfu/g lactic acid bacteria was present from the very beginning.
person sniffing at the outlet of the GC-column via an ‘olfactory detector outlet’ (ODO-1, SGE, Inc., USA). Split ratio 1: 1. The sniffing response was timed and recorded. Statistical analyses
The relationship between the gas chromatographic analysis and the sensory data was evaluated by ‘partial least squares analysis’ (PLS) using a multivariate analysis program (Sirius, version 2.3, Pattern Recognition Systems A/S, Bergen, Norway). The dependent variables (Y-matrix) were the sensory scores in mm divided by 150 mm times 10. The independent variables (X-matrix) were pH, amount of free fatty acids (FFA), and relative GC peak areas and numbers of S. xylosus in the loglo scale. The effects of the different factors on the sensory scores were evaluated and tested using multiple linear regression and analysis of variance (MODDE version 1.2, UMETRI AB, UmeP, Sweden). PLS analysis was also employed, the X-matrix being the experimental design.
52 63 57 61 61 53 56 59 65 53 54 68 60 51 66 64 49 61
f f f f f f f f f f f f f f f f f f
2.6 2.3 4.5 1.0 15 7.4 2.0 7.9 3.1 3.4 8.1 2.2 5.4 9.2 0.1 13 8.6 5.6
Intensity
76 79 58 64 85 61 70 66 83 66 63 49 63 55 79 54 52 60
f f f * f f f f f f f f f f f f f f
1.8 2.6 9.5 6.6 2.5 3.9 11 18 8.5 8.1 8.8 13 13 0.3 1.5 7.9 25 1.7
Salami
0.3 10 15 5 6 9 9 10 8 21 7 37 29 22 31 30 19 16
f + f f f f f f f + f f f f f f f f
0.3 0.6 7.8 2.2 4.4 3.3 5.0 1.4 3.2 7.5 1.1 11 16 1.5 16 19 13 12
Cheese
29 33 26 33 39 21 34 27 38 37 39 18 33 26 45 19 21 30
f 0.4 f 5.0 f 6.3 f 2.5 f 16 It 0.5 If: 6.5 f 3.2 f 2.2 f 5.8 f 3.4 f 3.2 f 6.9 f 8.9 * 16 f 0.9 f 5.3 l l-7
Sourdough 17 22 18 20 27 29 17 19 17 23 12 22 17 17 30 14 28 15
f 0.4 f 1.4 f 1.9 f 1.4 zk 4,2 zk 11 f 0.7 f 13 f 0.7 f 1.2 f 4.3 f 1.2 f 3.0 l 12 & 1.0 f 3.0 f 6.7 f 1.8
Sourish 22 23 23 16 25 17 19 20 20 21 20 17 19 26 18 26 17 20
f 2.5 f 0.6 f 5.5 f 2.5 zk 0.6 f 0.7 f 0.4 f 0.8 f 3.2 f 5.0 f 3.5 f 0.6 f 5.4 f 4.1 + 1.3 f 14 f 1.8 f 4.0
Fatty
DescriptoP
YScores in mm on a line scale of 150 mm, mean and standard deviation of duplicates. %x content of sausage types in Table 1.
A B C D E F G H I J K L M N 0
fYP@
Sausage
TABLE 2 Sensory Scores
2 2 5 2 5 4 4 1 1 3 5 10 6 16 5 17 5 5
f f f f f f f f f f f f f f f f f f
0.8 2.6 0.5 0.1 1.4 4.0 3-3 O-8 0.6 0.5 1.2 2.5 0.8 11 1.0 17 6.9 7.5
Rancid 4 9 12 11 7 9 9 15 7 13 12 16 21 36 8 31 13 11
* f f f f f f f f f f f f f f f f f
2.2 1.2 6.5 1.1 3.6 6,4 2.8 4.2 1.2 4.0 2.4 9.3 5.7 8.4 1.2 21 10 0.6
Nauseous 0.1 5 9 30 3 1 0.5 7 2 21 10 6 6 4 1 2 0.2 3
f f f f f f f f f f f f f f zt f f f
0.1 3.0 7.7 6.5 1.2 1.6 0.3 9.2 1.0 0.9 3.3 2.7 2.5 0.3 0.7 0.4 0.2 1.7
Burned
* f f f f f f f f f f f f f f f f f
13 17 33 13 16 45 26 26 13 15 16 48 38 40 20 35 25 27
6.3 4.1 2.9 6.7 15 16 11 20 7.7 2.9
5.1 0.9 14 6.1 0.1 30 0.3 10
Solvent
? s: k 6 g 6
Dried sausages -
RESULTS
IZZ. Sensory evaluation
AND
215
DISCUSSION
Relationship between factors and sensory data Table 2 shows the sensory scores obtained for the 18 different sausage types and Table 3 the factors that have a significant influence on the strength of the different aroma notes. The size of the individual regression coefficients is proportional to the importance of the factor in question. Figure 1 shows a loading plot from the PLS analysis on the factors and the sensory scores. Factors placed close to the centre of the plot affect the sensory properties to a lesser degree than factors far from the centre. The first principal component (PC) describes 36% of the variation in the data set, the second PC 14%. This means, that factors pulling descriptors apart horizontally are of more significance to the descriptors than factors pulling them apart vertically. Due to the nature of an experimental design it is not possible to test the significance of the principal components by crossvalidation. The Sirius program does not include any other testing abilities. There are small discrepancies between the results in Table 3 and Fig. 1 as the mathematics behind the results are different and the plot only explains 50% of the total variance of the experiment. Table 3 displays the results from a multivariable linear regression on each descriptor separately, the PLS plot arises from an algorithm working on all descriptors and factors simultaneously. The loading
FATTY
NAUSEOUS RANCID
SALAMI SOURDOUGH
tem*nat
nat*glu
BURNED
principal component 1 (36%)
Fig. 1. Loading plot from PLS analysis on sensory scores and experimental design. Percentage on axes indicates the variation explained by the principal component. tern = temperature, ped = P. pentosaceus, nit = nitrite (NaNO& nat = nitrate (KNO& glu = glucose, sal = salt. Two-factor interactions are confounded, cf. Table 2 in Part I.
PED
+ 4*
_1** + 3*** + 0.2* + 5***
+ 8***
__fj++* q*** 4*
TEA4
-3*
-
GLU
SAL
Two-factor interaction.@
-3*** -PED.GLU***, +‘NAT.GLU***, +PEDSAL* + NAT.GLU* + TEM.GLU**, + NIT.NAT* + TEM.GLU* _2*** -l** +TEM*NAT**, +TEM*PED*, -PEDSAL*, -TEM.GLU* +2*** _1*** -l*** +TEM.NAT***,-TEMSAL***,-TEM.GLU***, +PED.NAT**, +4*** _2** -1* + TEM.NAT**, -TEMSAL*, -TEM.GLU*, -NITSAL* -2* +4** -NITSAL* + 9*** + GLU.TEM**
NAT
_1* _5*** -3. -4* -2* +4*
NIT
Regression coeficients of main effects
Significant effects”
+TEM.NIT**
a*** = p 5 O$lOl, ** = P < 0.01, * = P < 0.05, l = P 5 O-10. TEM = fermentation temperature, PED = Pediococcuspentosuceus, NAT = KNOJ, NIT = NaNO*,%LU = glucose, SAL = salt. + or - in front of a factor indicates whether the factor increases or decreases the sensory score. The unit of the coefficients is mm, see Table 2. bTwo-factor interactions are confounded cf. Table 2 in Part I.
Intensity Salami Sourdough Sourish Cheese Fatty Rancid Nauseous Burned Solvent
Sensory descriptor
TABLE 3 Main Effects and Two-Factor Interactions
Dried sausages -
ill. Sensory evaluation
217
plot is included since it gives an overview that describes the coherence between the descriptors. It appears that none of the factors significantly affect the ‘intensity’ of the sausage. At the sensory sessions the ‘intensity’ was described as the overall odour sensation experienced at first on opening the glass jar with the sausage sample. The descriptor is probably too coarse to be of value in profiling the aroma of this type of sausage. The salami and sourdough odours seem to express the same attribute of the sausages since they are placed side by side in the plot (Fig. 1). However, salami aroma is decreased by more factors than sourdough smell (Table 3), this effect appears if a third dimension is included in the loading plot. In fact, salami aroma is decreased by high amounts of all of the ingredients except glucose, although the interaction effect, -PED*GLU, shows that P.pentosaceus has a more negative influence on salami aroma if glucose is added (and if salt concentration is low: +PED*SAL). One should keep in mind that those effects are valid only in the ingredient span shown in Table 1. For example salt and nitrite may very well be essential to salami aroma although their effects are negative according to the results. The results simply state that they will influence negatively if added in excessive amounts above 1.5% and 50 ppm, respectively (cf. Table 1). Concerning the direct influence of glucose on salami aroma, this is one of the cases where the results from the two analyzing methods differ. The loading plot in Fig. 1 suggests that glucose addition reduces the salami aroma, while that effect is not significant according to the linear regression model (Table 3). Rancidity and nauseating smell coincide (Fig. 1). Except for some of the twofactor interactions their ingredient dependence are the same (Table 3). This means that, in future sensory sessions, it is not necessary to include both descriptors. High temperature and addition of nitrate in particular, increase the rancidity of the sausages. It seems reasonable that ‘rancid’/‘nauseous’ and ‘salami’/‘sourdough’ are placed opposite in the plot. A proper salami aroma should not include rancid notes. ‘Salami’ is also opposed to ‘sourish’, ‘cheese’ and ‘solvent’. Burned odour is affected by salt, especially, but also by nitrite and nitrate (Table 3). This is not evident from the loading plot. However, if the principal components 3 and 4 are included, those effects clearly show. Cheesy smell is especially favoured by high temperatures. The previous paper showed that the amount of cheese-smelling components, like 2-methylpropionic, butanoic and 3-methylbutanoic acid, were strongly increased by high temperature. Relationship between volatile components and sensory data
Figures 2 and 3 display two loading plots from the PLS analysis of the volatile components and the sensory data. The quantified volatiles were shown in Table 3 of the preceeding paper. Table 4 shows the variance explained by the principal components (PC). To facilitate a better explanation of the results four PC’s have been included, though the degree of explanation of PC 3 especially is small. The first PC explains 3 1% of the variation in the sensory data set. This means that about one third of the sensory profile in the direction of PC 1 is accounted for by the volatile components. It is seen, in Figure 2, that ‘nauseous’, ‘rancid’ and ‘cheese’, in particular, and their correlation with the aldehydes decanal, pentanal, hexanal and 3-methylhexanal, and the acids methylbutanoic, 2-methylpropionic
218
L. H. Stahnke
BURNED
-
heptanonitrile
*-m*thylfura”
acetic acid + I-penten-3-01 P-propanol
‘OLVENT
6J
- NAUSEoUSde%~kyl-butanoic
F
-
s
_ RANCID
P-
3 .a
-
2 ‘C a
2-methyl-propionic acid P-methylpropanol l-propanol pentanal hexanal
aid
3-methylbutylacetata
Isopropyl PH
xetate
s.xylosus
P-pentanone bmeth lbutanal Bmethylgutanal
diawl
acetoin
+ n ethyl-butanol
FFAFAm *-Penten
INTENSITY
2-heptanone
s”~Y@fouGH
2-methyl-3-buten-2-01
I-butanol
butanal -
4-methyl-2-pentanone acetaldehyde
2-methylprzpanal
3methylhexanal
CHEESsuta”oic
8
2.3butandiol acid
2.nonanone ethyl acetate
3-methyl-2-buten-l-01
ropaoate ethyl bmethylbutan
SALAMI
ethanol
e$~~~~~“~~
te
ethyl P-methyl butanoate
principal component Fig. 2.
1
Loading plot from PLS analysis on sensory scores and content of volatile compounds. Table 4 shows the variation explained by the principal components.
and butanoic acid are responsible for this. ‘Salami’, ‘sourdough’ and ‘sourish’ are opposed to the above descriptors, pulled towards the right by the 2-alkanones and the numbers of S. xylosus and towards the bottom of the plot by the ethyl esters. Also an earlier study showed that certain ethyl esters are essential to salami aroma (Stahnke, 1994). It seems that, if S. xylosus is important in producing salami aroma, this is not caused merely by producing ethyl esters, since S. xylosus is placed on the first principal component. The ethyl esters are placed close to zero of the first PC (Fig. 2). ‘Rancid’, however is placed opposite to the numbers of S. xylosus. It may be that this is caused by the catalase-producing capability of S. xylosus. Catalase catalyzes the decomposition of peroxides involved in autoxidative reactions leading to rancidity. Lactic acid bacteria, other than P. pentosacew, are present in the sausages. Their presence is not included in the statistical analysis, but they may also be responsible for the formation of ethyl esters. There is evidence that several lactic acid bacteria are capable of performing esterification of acids (Hosono et al., 1974; Bills et al., 1965). According to Fig. 3 the sourish note to salami odour is especially correlated with acetic acid, but perhaps also with butanoic acid. 2-Heptanone does not have a sour note (cf. below and Table 5) but may be of importance in the total perception of ‘sourish’. As to be expected, ‘sourish’ and acetic acid are strongly opposed to pH. The same is the case for ‘sourdough’. Figure 3 also indicates that the short-chain aldehydes 2- and 3-methylbutanal may be important to ‘salami’
Dried sausages -
III. Sensory evaluation
219
PH
3.methylbutanal 2,:methy’b”%%thylpropanal s. .ylosus z-pen&I 101
2,3-butandiol
FATTY
BURNED SOURDOUGH
ethyl octawate 1-butanol
SALAMI ethyl P-methylpropanoate
isopropyl acetate p~~~ialhe,~~l~g~~
2-methy’propano’
NAUSEOUS
methylfuran
2-i-manol i-propanol
RANCID
ethyl 3.methylbut; ‘le 2.methyl-3.buten-2.olethy
2-pentanone acetaldehyde
3.methyl-2.butenol
ethyl proplonate
noate 3-methylbytyl 2_methylbutanoate
dwetyl 4.methyl-2.pentanone eth
2-nona%GvENT ethanol
P-methylproplonic acid _~______. methylbutanolc acid 3.methylhexanal acetate methylbutanol
I acetatk-Penten-got
+ acetoin
+ acetic acid butanoic
CHEl!SE
acid
FFA 2.heptanone
SOURISH
I
I
I
I
I
I
principal component 3 Fig. 3.
Loading plot from PLS analysis on sensory scores and content of volatile compounds. Table 4 shows the variation explained by the principal components.
and ‘sourdough’. Both aldehydes have high volatilities with characteristic smells. Sourdough aroma seems to be influenced by the cheesy smelling acids as well (2-methylpropionic and methylbutanoic acid). Solvent odour may be caused by a combination of all of the alcohols and the acetaldehyde and isopropyl acetate, pulling ‘solvent’ towards the left (Fig. 3). It is interesting that ‘fatty’ and diacetyl coincide in Fig. 2. Diacetyl is a highly volatile compound with a buttery and perhaps fatty smell in connection with other compounds. Most likely the aldehydes pentanal, hexanal and 3-methylhexanal. The amount of free fatty acids (FFA, 16 : 0 + 18 : 3) also coincides with ‘fatty’ odour. It is doubtful, though, if FFA have any direct influence on odour since their volatility are almost zero. The burned odour does not seem to correlate with the presence of any single component, but is opposed to the ethyl esters (Fig. 2). Ethyl esters may mask TABLE 4
PLS Analysis of Relative GC Peak Areas and Sensory Scores % Variance explained by principal components
X-matrix (GC peak areas) Y-matrix (sensory scores)
PC1
PC2
PC3
PC4
19 31
12 8
14 5
9 10
Total 54 54
220
L. H. Stahnke
TABLE 5
Results from GC-Effluent Sniffing RT-index” Description of odour 486 520 603 647 681 711 731 742 773 774 777 790 788 811 831 845 857 863 866 874 882 886 901 902 908 931 932 943 951 958 961 966 976 979 985 988 992 1000 1013 1017 1024 1043 1044 1049 1058 1067 1079
rotten green green, weak green, weak butter, very strong dry, green, pelargonium, weak sour, nailpolish, cheese, bit fruity breath of garlic burned rubber, coffee, cooked meat rotten toast, bread vinegar butter, weak fruit bubblegum, strong fruit, weak sour, weak fruit, strong dry, exhaust coffee, toasted onion, bush green leaves sharp, metallic fruit old cooked potatoes, nauseous chutney, fruity cheesy, weak boiled milk/meat, vitamins solvent green leaves fruit, weak cheesy, very intense pelargonium, nettle, fruity green leaves coffee, toasted bread, toasted rancid unpleasant bread breath of garlic sweaty socks, very intense fungi, mouse, popcorn, strong chutney?, cucumber?, weak green cooked, toasted potatoes, pleasant metallic, sharp, soil chutney, fruit, weak mushroom nettle, pelargonium, plant
Sausage type@
Component’
all all
acetaldehyde
A, E, L J, K M 0, Q
F, K L all 2, 3-butadione = diacetyl A, C, F, J, K all 2- and 3-methylbutanal all - 0, Q A C, JA L K, N, 0, Q all L, M, P, Q all acetic acid 2, 3pentadione B, D, E, J, M all ethyl-2-methylpropanoate all 2-methylpropylacetate all propanoic acid all ethylbutanoate all - N, Q A, F, L L, M, Q
D, E, G, M 0 all all - P, Q all all all
J, K M, P all A, C, D, L all all - F, L, Q all A L J, K M all A, & C, E, F, G K, M, 0, Q A, B, F, Q N, P, Q all all -J, K, L, M, 0 all all A B, K, N E, J, M, N all all B, J% L, Q all all
hexanal ethyl-2-methylbutanoate (cis-x-hexenal) ethyl-3-methylbutanoate 2-methylpropanoic acid N-camp. (amine) 3-methylhexanal ethylpentanoate butanoic acid 2-heptanone heptanal dimethylpyrazine dimethylpyrazine c&t-heptenal 3-methylbutanoic acid 2-methylbutanoic acid
(ethylhexanoate) (1octen-3-one)
Dried sausages -
221
III. Sensory evaluation
TABLE 5-continued 1084 1091 1106 1139 1145 1154 1165 1171 1177 1186
nettle, dry, pelargonium orange nettle, plastic, weak hot soil, mushroom, sharp, metallic popcorn, fungi, mouse, weak plastic, nettle, plant fruity, weak mushroom deepfried, weak honey
all octanal all - P, Q all-C,D,F,I A, C, D, E, J, L, M, 0 3-octen-2-one all - A, D, E, F all - D, E, F, I K N, 0, P (x-nonen-y-one) all - I, 0 trans-2-octenal G I, K Q B, E, G I, K, L, M, 0, Q
1187
nettle, dry, pelargonium
B, C F, G I, K M N I-’
1190 1193 1196 1199 1202 1219 1233 1250 1255 1266 1282 1296 1308 1324
coffee, burned, cress orange fresh, nauseous Jerusalem artichoke, burned plastic orange, weak cooked meat bandage, cooked frankfurter, strong deepfried, weak popcorn vitamins, cooked meat cucumber cucumber staple, horse dried flowers
1330 1339 1357
bandage, cooked frankfurter pelargonium deepfried
all - A, B, M, P, 0 all - J, L, Q all - 0 all - A, C, J, L, N, P K L, M, N, 0, P C, F, G N all E, I, K M, Q all - D, F all - C all D, E, I, K, M, N, P all - F B, C, E, C. K, M, N, 0, Q all - A, D, I, L C, K L, M all
nonanal
methoxyphenol
= guiaicol
trans-2-nonenal methylphenol
= cresol
‘Kovats indices of the used GC-method. bOnly odours detected in more than two sausages are included; cf. Table 1 for explanation of sausage types. ‘Identified components that may be responsible for odour. Compounds in parenthesis are suggestions based on odour and RT-index, cf. also Table 2 in Part II.
burned smell when present in larger amounts. ‘Burned’ could arise from a mixture of components or from components not identified because of their very low concentrations, but having low sensory threshold values (e.g. the dimethylpyrazines, cf. Table 5). The fact that the four principal components only explain 54% of the variance in the sensory data set (Y-matrix, see Table 4) also shows that many important aroma components may not have been detected in this experiment. Table 4 also shows that only 54% of the variation of the X-matrix is explained by the first four principal components. This could indicate that many of the volatiles are of no importance to aroma. Even if this is so, leaving out compounds like the saturated alcohols (Table 5 shows that they have little aroma value) only results in a slight increase in degree of explanation.
222
L. H. Stahnke
Effluent snifling
Table 5 shows the outcome of the effluent sniffing. The identified components which are proposed as responsible for the odours are compiled in Table 2 of the previous paper. The sniffing procedure was quite simple there being no attempt to quantify the smells, but simply detecting their presence. In order to find any correlation between the detected odours and the sensory data, attempts were made to do a PLS analysis on the results; but these were not successful, partly because the sniffed odours were not quantified. Binomial data ( = present/non-present) were used instead. However Table 5 clearly shows that there are many components with characteristic odours that have not been detected/identified by the gas chromatographic and mass spectrometric analyses. An earlier study showed this as well (Stahnke, 1994). In order to obtain a fuller picture of the aroma development in fermented sausages it would be necessary to identify all of the components with sensory value. The identified components of special importance to aroma seem to be the volatile acids, many aldehydes and several esters and ketones. Also two different dimethylpyrazines may be of importance. It appears that saturated alcohols have too high sensory threshold values to be detected individually, but perhaps they are of some importance in the complete aroma of the sausages. Many of the unidentified odours are of character which suggests that they probably arise from breakdown of sulfuric amino acids, e.g. the odours at RTindex 486, 742, 774, 1000. Other odours have notes reminiscent of long, polyunsaturated aldehydes, e.g. those at RT-index 1250, 1296, 1357; and some smell like amides, e.g. the odours at 1017, 1145, 1255. It is notable that more than half of the sausage types possessed the same odours. This indicates that differences in the overall aroma of fermented sausages are caused by differences in the ratio of aroma components and not so much by development of specific compounds in certain types of sausages. TABLE 6 Aroma Profile of Two Types of Sausages Sausage type
Sensory descriptoP Old-fashioned
Modern
Low temperature, added nitrate
High temperature, nitrite, glucose, P. pentosaceus
Salami Sourdough Sourish Cheese Fatty Rancid Nauseous Burned Solvent “Aroma strength is increased in direction of arrow.
+ t + -+ + ct t+ u +
Dried sausages -
Ill. Sensory evaluation
223
CONCLUSIONS Table 6 summarizes the aroma profile of two types of sausages. The data are based on the linear regression model, assuming independency between the sensory descriptors. Table 6 shows that salami odour is more pronounced in ‘oldfashioned’ sausages. Also, ‘old-fashioned’ sausages have more sourdough, fatty and solvent-like smells while ‘modern’ types of fermented sausages have more sourish and cheesy aroma notes. ‘Salami’ seems to correlate with the presence of ethyl esters and 2-alkanones as well as with high numbers of S. xylosus. Also 2- and 3-methylbutanal may be of importance. Sourdough odour is influenced by the same components as salami odour, mixed with cheesy-smelling acids like 2-methylpropionic and 3-methylbutanoic acid. Fatty aroma is probably related to the content of diacetyl as well as the aldehydes hexanal and 3-methylhexanal. Sourish notes to salami odour seem to correlate especially with acetic and perhaps butanoic acid. Cheesy aroma is correlated with high amounts of volatile acids like 2-methylpropionic, butanoic and 3-methylbutanoic acid. ACKNOWLEDGEMENTS This study was sponsored by the Center for Food Research at Technical University of Denmark, DK-2800 Lyngby, Denmark. REFERENCES Acree, T. E., Barnard, J. & Cunningham, D. G. (1984). Food Chern., 14, 273. Aishima, T. & Nakai, S. (1991). Food Rev. Znt., 7(l), 33. Berdague, J.-L., Monteil, P., Montel, M. C. & Talon, R. (1993). Meat Sci., 35, 275. Berger, R. G., Macku, C., German, J. B. & Shibamoto, T. (1990). J. Food Sci., 55, 1239. Bills, D. D., Morgan, M. E., Libbey, L. M. & Day, E. A. (1965). J. Dairy Sci., 48, 1168. Chien, M. 8z Peppard, T. (1993). In Flavor measurement, eds. C.-T. Ho & C. H. Manley. Marcel Dekker, Inc., New York. p. 1. Fischer, K.-H. & Grosch, W. (1987). Lebensm. Wiss. Techn., 20, 233. Grosch, W. (1993). Trends Food Sci. Techn., 4(3), 68. Hosono, A., Elliott, J. A. & McGugan, W. A. (1974). J. Dairy Sci., 57(5), 535. Johansson, G., Berdague, J.-L., Larsson, M., Tran, N. & Borch, E. (1994). Meat Sci., 38, 203. Martens, H. (1982). In Food research and data analysis, eds. H. Martens & H. Russwurm, Jr. Appl. Science Publishers, London, p. 5. Stahnke, L. H. (1994). Meat Sci., 38, 39.
ELSEVIER
0309-1740(94)00068-9
Dried Sausages Fermented with Staphylococcus xylosus at Different Temperatures and with Different Ingredient Levels - Part III. Sensory Evaluation
L. H. Stahnke Department
of Biotechnology, Technical University of Denmark, Building 221, Lyngby 2800, Denmark
(Received 15 January 1994; revised version received 3 October 1994; accepted 6 October 1994)
ABSTRACT Sausages with added Staphylococcus xylosus were fermented at dtjerent temperatures and with dtyerent added levels of salt, glucose, nitrite, nitrate and Pediococcus pentosaceus in accordance with a six factor fractional design. The odour of the sausages was evaluated by a quantitative descriptive method with ten descriptors and by gas chromatography olfactometry. The sensory profile was correlated to the experimental design and the volatile compounds by partial least squares analysis. Also, the effects of temperature and different ingredients on the strength of the individual descriptors were tested using multiple linear regression and analysis of variance. The study showed that salami odour was more pronounced in sausages fermented at low temperature than in sausages fermented at high temperature and added nitrite, glucose and P. pentosaceus. High temperature sausages had a more sour and cheesy aroma, but were less fatty and sourdough-like. Salami odour was correlated with the presence of ethyl esters and 2-alkanones as well as with high numbers of Staphylococcus xylosus. 2- and 3-methylbutanal seemed to be influential as well. The sourish note to salami odour was especially caused by acetic andperhaps butanoic acid, the cheesy note by 2-methylpropionic, butanoic and 3-methylbutanoic acid.
INTRODUCTION The aroma of fermented sausage is of great importance to the sausage quality, but little is known about the components that comprise the aroma profile and the ones of more significance to flavour. Neither is much known about the influence of sausage ingredients and fermentation parameters on the final sausage aroma. 211
212
L. H. Stahnke
Several studies have shown that the volatile profile of fermented sausage is very complex, including components from many classes of compounds (Berger et al., 1990; BerdaguC et al., 1993; Johansson et al., 1994; Stahnke, 1994). The previous paper has already given a short introduction to the subject. The unique aroma of most foods is not due to one compound, but usually to many compounds of diverse structures, blended in a specific ratio. There are two approaches to evaluate the components of most importance to final flavour. One approach is to pinpoint the more potent aroma components in the volatile fraction by gas chromatography olfactometry (also called effluent sniffing) and more elaborate, by aroma extract dilution (Grosch, 1993; Acree et al., 1984). Over the years, many workers have employed effluent sniffing and have in this way detected some of those components giving certain foods their characteristic aroma, e.g. 1-octen-3-01 and I-octen-3-one in mushrooms (Fischer & Grosch, 1987). A second approach is to produce a descriptive sensory profile of the food in question and employ different multivariate statistical methods to study the correlation between the volatile profile and the sensory data. In this way, it may be possible to get an idea of which component or class of components that are responsible for specific aroma notes (Chien & Peppard, 1993; Aishima & Nakai, 1991; Martens, 1982). The purpose of the present study was to investigate the effect of fermentation temperature and some basic sausage ingredients on the sensory profile of a fermented sausage added Staphylococcus xylosus. The influence of various parameters, fermentation temperature, salt, nitrite, nitrate and glucose concentration, and the addition of another starter culture, Pediococcuspentosaceus, was tested in a fractional factorial design. Also, sensory data and volatile profile were correlated using multivariate statistics. MATERIALS
AND METHODS
Detailed information on experimental design, sausage production and chemical and bacterial analyses, and on the analysis of volatiles, are given in the two preceeding papers. Table 1 sums up the experimental design and the factorial levels. Sensory analyses
The odour of the sausages was evaluated twice by a trained panel of 11 members using a quantitative descriptive method with 10 descriptors and an unstructured line scale of 150 mm. The line had two anchor points at low and high level, respectively (15 mm from the end). The list of descriptors was developed during training sessions by the panel members. The panel was trained once a week for a total of 3 months prior to the testing sessions. The test samples consisted of approximately 6 g of ground sausage equilibrated in closed 50 ml containers for 1 h at 37°C. The panel was presented with a reference sample at the beginning of each session. The effluent from the gas chromatograph was evaluated once for all of the sausages except the ‘centrepoints’. The effluent evaluation was done by one
Dried sausages -
III. Sensory evaluation
213
TABLE 1 Experimental Design” Factor
Batch no. Fermentation temperature
NaNOz
KNOs
0
A B C D E F G H I J K L M N 0 P
0 0
0 0 0 0 0
Salt
1 0 0 1 0 0 1
1 0 1 1 0 1 0 1
0 1 1 0 1 0 1
6 1 1 1 1 0 0 0 0
1
1
0
1 1
1 1 1 1 1 1 1
15°C 25°C
Glucose
4 1
0
: Factor levels: Low High
Pediococcus pentosaceus
Ob
4.2 x 107cm/g
50ppm
300 ppm
0
0
1.5%
0.20%
1.0%
3.5%
“O= low level, 1 = high level, f = centre level. bPediococcuspentosaceus was not added to the minces but a natural flora of 2.5 x lo3 cfu/g lactic acid bacteria was present from the very beginning.
person sniffing at the outlet of the GC-column via an ‘olfactory detector outlet’ (ODO-1, SGE, Inc., USA). Split ratio 1: 1. The sniffing response was timed and recorded. Statistical analyses
The relationship between the gas chromatographic analysis and the sensory data was evaluated by ‘partial least squares analysis’ (PLS) using a multivariate analysis program (Sirius, version 2.3, Pattern Recognition Systems A/S, Bergen, Norway). The dependent variables (Y-matrix) were the sensory scores in mm divided by 150 mm times 10. The independent variables (X-matrix) were pH, amount of free fatty acids (FFA), and relative GC peak areas and numbers of S. xylosus in the loglo scale. The effects of the different factors on the sensory scores were evaluated and tested using multiple linear regression and analysis of variance (MODDE version 1.2, UMETRI AB, UmeP, Sweden). PLS analysis was also employed, the X-matrix being the experimental design.
52 63 57 61 61 53 56 59 65 53 54 68 60 51 66 64 49 61
f f f f f f f f f f f f f f f f f f
2.6 2.3 4.5 1.0 15 7.4 2.0 7.9 3.1 3.4 8.1 2.2 5.4 9.2 0.1 13 8.6 5.6
Intensity
76 79 58 64 85 61 70 66 83 66 63 49 63 55 79 54 52 60
f f f * f f f f f f f f f f f f f f
1.8 2.6 9.5 6.6 2.5 3.9 11 18 8.5 8.1 8.8 13 13 0.3 1.5 7.9 25 1.7
Salami
0.3 10 15 5 6 9 9 10 8 21 7 37 29 22 31 30 19 16
f + f f f f f f f + f f f f f f f f
0.3 0.6 7.8 2.2 4.4 3.3 5.0 1.4 3.2 7.5 1.1 11 16 1.5 16 19 13 12
Cheese
29 33 26 33 39 21 34 27 38 37 39 18 33 26 45 19 21 30
f 0.4 f 5.0 f 6.3 f 2.5 f 16 It 0.5 If: 6.5 f 3.2 f 2.2 f 5.8 f 3.4 f 3.2 f 6.9 f 8.9 * 16 f 0.9 f 5.3 l l-7
Sourdough 17 22 18 20 27 29 17 19 17 23 12 22 17 17 30 14 28 15
f 0.4 f 1.4 f 1.9 f 1.4 zk 4,2 zk 11 f 0.7 f 13 f 0.7 f 1.2 f 4.3 f 1.2 f 3.0 l 12 & 1.0 f 3.0 f 6.7 f 1.8
Sourish 22 23 23 16 25 17 19 20 20 21 20 17 19 26 18 26 17 20
f 2.5 f 0.6 f 5.5 f 2.5 zk 0.6 f 0.7 f 0.4 f 0.8 f 3.2 f 5.0 f 3.5 f 0.6 f 5.4 f 4.1 + 1.3 f 14 f 1.8 f 4.0
Fatty
DescriptoP
YScores in mm on a line scale of 150 mm, mean and standard deviation of duplicates. %x content of sausage types in Table 1.
A B C D E F G H I J K L M N 0
fYP@
Sausage
TABLE 2 Sensory Scores
2 2 5 2 5 4 4 1 1 3 5 10 6 16 5 17 5 5
f f f f f f f f f f f f f f f f f f
0.8 2.6 0.5 0.1 1.4 4.0 3-3 O-8 0.6 0.5 1.2 2.5 0.8 11 1.0 17 6.9 7.5
Rancid 4 9 12 11 7 9 9 15 7 13 12 16 21 36 8 31 13 11
* f f f f f f f f f f f f f f f f f
2.2 1.2 6.5 1.1 3.6 6,4 2.8 4.2 1.2 4.0 2.4 9.3 5.7 8.4 1.2 21 10 0.6
Nauseous 0.1 5 9 30 3 1 0.5 7 2 21 10 6 6 4 1 2 0.2 3
f f f f f f f f f f f f f f zt f f f
0.1 3.0 7.7 6.5 1.2 1.6 0.3 9.2 1.0 0.9 3.3 2.7 2.5 0.3 0.7 0.4 0.2 1.7
Burned
* f f f f f f f f f f f f f f f f f
13 17 33 13 16 45 26 26 13 15 16 48 38 40 20 35 25 27
6.3 4.1 2.9 6.7 15 16 11 20 7.7 2.9
5.1 0.9 14 6.1 0.1 30 0.3 10
Solvent
? s: k 6 g 6
Dried sausages -
RESULTS
IZZ. Sensory evaluation
AND
215
DISCUSSION
Relationship between factors and sensory data Table 2 shows the sensory scores obtained for the 18 different sausage types and Table 3 the factors that have a significant influence on the strength of the different aroma notes. The size of the individual regression coefficients is proportional to the importance of the factor in question. Figure 1 shows a loading plot from the PLS analysis on the factors and the sensory scores. Factors placed close to the centre of the plot affect the sensory properties to a lesser degree than factors far from the centre. The first principal component (PC) describes 36% of the variation in the data set, the second PC 14%. This means, that factors pulling descriptors apart horizontally are of more significance to the descriptors than factors pulling them apart vertically. Due to the nature of an experimental design it is not possible to test the significance of the principal components by crossvalidation. The Sirius program does not include any other testing abilities. There are small discrepancies between the results in Table 3 and Fig. 1 as the mathematics behind the results are different and the plot only explains 50% of the total variance of the experiment. Table 3 displays the results from a multivariable linear regression on each descriptor separately, the PLS plot arises from an algorithm working on all descriptors and factors simultaneously. The loading
FATTY
NAUSEOUS RANCID
SALAMI SOURDOUGH
tem*nat
nat*glu
BURNED
principal component 1 (36%)
Fig. 1. Loading plot from PLS analysis on sensory scores and experimental design. Percentage on axes indicates the variation explained by the principal component. tern = temperature, ped = P. pentosaceus, nit = nitrite (NaNO& nat = nitrate (KNO& glu = glucose, sal = salt. Two-factor interactions are confounded, cf. Table 2 in Part I.
PED
+ 4*
_1** + 3*** + 0.2* + 5***
+ 8***
__fj++* q*** 4*
TEA4
-3*
-
GLU
SAL
Two-factor interaction.@
-3*** -PED.GLU***, +‘NAT.GLU***, +PEDSAL* + NAT.GLU* + TEM.GLU**, + NIT.NAT* + TEM.GLU* _2*** -l** +TEM*NAT**, +TEM*PED*, -PEDSAL*, -TEM.GLU* +2*** _1*** -l*** +TEM.NAT***,-TEMSAL***,-TEM.GLU***, +PED.NAT**, +4*** _2** -1* + TEM.NAT**, -TEMSAL*, -TEM.GLU*, -NITSAL* -2* +4** -NITSAL* + 9*** + GLU.TEM**
NAT
_1* _5*** -3. -4* -2* +4*
NIT
Regression coeficients of main effects
Significant effects”
+TEM.NIT**
a*** = p 5 O$lOl, ** = P < 0.01, * = P < 0.05, l = P 5 O-10. TEM = fermentation temperature, PED = Pediococcuspentosuceus, NAT = KNOJ, NIT = NaNO*,%LU = glucose, SAL = salt. + or - in front of a factor indicates whether the factor increases or decreases the sensory score. The unit of the coefficients is mm, see Table 2. bTwo-factor interactions are confounded cf. Table 2 in Part I.
Intensity Salami Sourdough Sourish Cheese Fatty Rancid Nauseous Burned Solvent
Sensory descriptor
TABLE 3 Main Effects and Two-Factor Interactions
Dried sausages -
ill. Sensory evaluation
217
plot is included since it gives an overview that describes the coherence between the descriptors. It appears that none of the factors significantly affect the ‘intensity’ of the sausage. At the sensory sessions the ‘intensity’ was described as the overall odour sensation experienced at first on opening the glass jar with the sausage sample. The descriptor is probably too coarse to be of value in profiling the aroma of this type of sausage. The salami and sourdough odours seem to express the same attribute of the sausages since they are placed side by side in the plot (Fig. 1). However, salami aroma is decreased by more factors than sourdough smell (Table 3), this effect appears if a third dimension is included in the loading plot. In fact, salami aroma is decreased by high amounts of all of the ingredients except glucose, although the interaction effect, -PED*GLU, shows that P.pentosaceus has a more negative influence on salami aroma if glucose is added (and if salt concentration is low: +PED*SAL). One should keep in mind that those effects are valid only in the ingredient span shown in Table 1. For example salt and nitrite may very well be essential to salami aroma although their effects are negative according to the results. The results simply state that they will influence negatively if added in excessive amounts above 1.5% and 50 ppm, respectively (cf. Table 1). Concerning the direct influence of glucose on salami aroma, this is one of the cases where the results from the two analyzing methods differ. The loading plot in Fig. 1 suggests that glucose addition reduces the salami aroma, while that effect is not significant according to the linear regression model (Table 3). Rancidity and nauseating smell coincide (Fig. 1). Except for some of the twofactor interactions their ingredient dependence are the same (Table 3). This means that, in future sensory sessions, it is not necessary to include both descriptors. High temperature and addition of nitrate in particular, increase the rancidity of the sausages. It seems reasonable that ‘rancid’/‘nauseous’ and ‘salami’/‘sourdough’ are placed opposite in the plot. A proper salami aroma should not include rancid notes. ‘Salami’ is also opposed to ‘sourish’, ‘cheese’ and ‘solvent’. Burned odour is affected by salt, especially, but also by nitrite and nitrate (Table 3). This is not evident from the loading plot. However, if the principal components 3 and 4 are included, those effects clearly show. Cheesy smell is especially favoured by high temperatures. The previous paper showed that the amount of cheese-smelling components, like 2-methylpropionic, butanoic and 3-methylbutanoic acid, were strongly increased by high temperature. Relationship between volatile components and sensory data
Figures 2 and 3 display two loading plots from the PLS analysis of the volatile components and the sensory data. The quantified volatiles were shown in Table 3 of the preceeding paper. Table 4 shows the variance explained by the principal components (PC). To facilitate a better explanation of the results four PC’s have been included, though the degree of explanation of PC 3 especially is small. The first PC explains 3 1% of the variation in the sensory data set. This means that about one third of the sensory profile in the direction of PC 1 is accounted for by the volatile components. It is seen, in Figure 2, that ‘nauseous’, ‘rancid’ and ‘cheese’, in particular, and their correlation with the aldehydes decanal, pentanal, hexanal and 3-methylhexanal, and the acids methylbutanoic, 2-methylpropionic
218
L. H. Stahnke
BURNED
-
heptanonitrile
*-m*thylfura”
acetic acid + I-penten-3-01 P-propanol
‘OLVENT
6J
- NAUSEoUSde%~kyl-butanoic
F
-
s
_ RANCID
P-
3 .a
-
2 ‘C a
2-methyl-propionic acid P-methylpropanol l-propanol pentanal hexanal
aid
3-methylbutylacetata
Isopropyl PH
xetate
s.xylosus
P-pentanone bmeth lbutanal Bmethylgutanal
diawl
acetoin
+ n ethyl-butanol
FFAFAm *-Penten
INTENSITY
2-heptanone
s”~Y@fouGH
2-methyl-3-buten-2-01
I-butanol
butanal -
4-methyl-2-pentanone acetaldehyde
2-methylprzpanal
3methylhexanal
CHEESsuta”oic
8
2.3butandiol acid
2.nonanone ethyl acetate
3-methyl-2-buten-l-01
ropaoate ethyl bmethylbutan
SALAMI
ethanol
e$~~~~~“~~
te
ethyl P-methyl butanoate
principal component Fig. 2.
1
Loading plot from PLS analysis on sensory scores and content of volatile compounds. Table 4 shows the variation explained by the principal components.
and butanoic acid are responsible for this. ‘Salami’, ‘sourdough’ and ‘sourish’ are opposed to the above descriptors, pulled towards the right by the 2-alkanones and the numbers of S. xylosus and towards the bottom of the plot by the ethyl esters. Also an earlier study showed that certain ethyl esters are essential to salami aroma (Stahnke, 1994). It seems that, if S. xylosus is important in producing salami aroma, this is not caused merely by producing ethyl esters, since S. xylosus is placed on the first principal component. The ethyl esters are placed close to zero of the first PC (Fig. 2). ‘Rancid’, however is placed opposite to the numbers of S. xylosus. It may be that this is caused by the catalase-producing capability of S. xylosus. Catalase catalyzes the decomposition of peroxides involved in autoxidative reactions leading to rancidity. Lactic acid bacteria, other than P. pentosacew, are present in the sausages. Their presence is not included in the statistical analysis, but they may also be responsible for the formation of ethyl esters. There is evidence that several lactic acid bacteria are capable of performing esterification of acids (Hosono et al., 1974; Bills et al., 1965). According to Fig. 3 the sourish note to salami odour is especially correlated with acetic acid, but perhaps also with butanoic acid. 2-Heptanone does not have a sour note (cf. below and Table 5) but may be of importance in the total perception of ‘sourish’. As to be expected, ‘sourish’ and acetic acid are strongly opposed to pH. The same is the case for ‘sourdough’. Figure 3 also indicates that the short-chain aldehydes 2- and 3-methylbutanal may be important to ‘salami’
Dried sausages -
III. Sensory evaluation
219
PH
3.methylbutanal 2,:methy’b”%%thylpropanal s. .ylosus z-pen&I 101
2,3-butandiol
FATTY
BURNED SOURDOUGH
ethyl octawate 1-butanol
SALAMI ethyl P-methylpropanoate
isopropyl acetate p~~~ialhe,~~l~g~~
2-methy’propano’
NAUSEOUS
methylfuran
2-i-manol i-propanol
RANCID
ethyl 3.methylbut; ‘le 2.methyl-3.buten-2.olethy
2-pentanone acetaldehyde
3.methyl-2.butenol
ethyl proplonate
noate 3-methylbytyl 2_methylbutanoate
dwetyl 4.methyl-2.pentanone eth
2-nona%GvENT ethanol
P-methylproplonic acid _~______. methylbutanolc acid 3.methylhexanal acetate methylbutanol
I acetatk-Penten-got
+ acetoin
+ acetic acid butanoic
CHEl!SE
acid
FFA 2.heptanone
SOURISH
I
I
I
I
I
I
principal component 3 Fig. 3.
Loading plot from PLS analysis on sensory scores and content of volatile compounds. Table 4 shows the variation explained by the principal components.
and ‘sourdough’. Both aldehydes have high volatilities with characteristic smells. Sourdough aroma seems to be influenced by the cheesy smelling acids as well (2-methylpropionic and methylbutanoic acid). Solvent odour may be caused by a combination of all of the alcohols and the acetaldehyde and isopropyl acetate, pulling ‘solvent’ towards the left (Fig. 3). It is interesting that ‘fatty’ and diacetyl coincide in Fig. 2. Diacetyl is a highly volatile compound with a buttery and perhaps fatty smell in connection with other compounds. Most likely the aldehydes pentanal, hexanal and 3-methylhexanal. The amount of free fatty acids (FFA, 16 : 0 + 18 : 3) also coincides with ‘fatty’ odour. It is doubtful, though, if FFA have any direct influence on odour since their volatility are almost zero. The burned odour does not seem to correlate with the presence of any single component, but is opposed to the ethyl esters (Fig. 2). Ethyl esters may mask TABLE 4
PLS Analysis of Relative GC Peak Areas and Sensory Scores % Variance explained by principal components
X-matrix (GC peak areas) Y-matrix (sensory scores)
PC1
PC2
PC3
PC4
19 31
12 8
14 5
9 10
Total 54 54
220
L. H. Stahnke
TABLE 5
Results from GC-Effluent Sniffing RT-index” Description of odour 486 520 603 647 681 711 731 742 773 774 777 790 788 811 831 845 857 863 866 874 882 886 901 902 908 931 932 943 951 958 961 966 976 979 985 988 992 1000 1013 1017 1024 1043 1044 1049 1058 1067 1079
rotten green green, weak green, weak butter, very strong dry, green, pelargonium, weak sour, nailpolish, cheese, bit fruity breath of garlic burned rubber, coffee, cooked meat rotten toast, bread vinegar butter, weak fruit bubblegum, strong fruit, weak sour, weak fruit, strong dry, exhaust coffee, toasted onion, bush green leaves sharp, metallic fruit old cooked potatoes, nauseous chutney, fruity cheesy, weak boiled milk/meat, vitamins solvent green leaves fruit, weak cheesy, very intense pelargonium, nettle, fruity green leaves coffee, toasted bread, toasted rancid unpleasant bread breath of garlic sweaty socks, very intense fungi, mouse, popcorn, strong chutney?, cucumber?, weak green cooked, toasted potatoes, pleasant metallic, sharp, soil chutney, fruit, weak mushroom nettle, pelargonium, plant
Sausage type@
Component’
all all
acetaldehyde
A, E, L J, K M 0, Q
F, K L all 2, 3-butadione = diacetyl A, C, F, J, K all 2- and 3-methylbutanal all - 0, Q A C, JA L K, N, 0, Q all L, M, P, Q all acetic acid 2, 3pentadione B, D, E, J, M all ethyl-2-methylpropanoate all 2-methylpropylacetate all propanoic acid all ethylbutanoate all - N, Q A, F, L L, M, Q
D, E, G, M 0 all all - P, Q all all all
J, K M, P all A, C, D, L all all - F, L, Q all A L J, K M all A, & C, E, F, G K, M, 0, Q A, B, F, Q N, P, Q all all -J, K, L, M, 0 all all A B, K, N E, J, M, N all all B, J% L, Q all all
hexanal ethyl-2-methylbutanoate (cis-x-hexenal) ethyl-3-methylbutanoate 2-methylpropanoic acid N-camp. (amine) 3-methylhexanal ethylpentanoate butanoic acid 2-heptanone heptanal dimethylpyrazine dimethylpyrazine c&t-heptenal 3-methylbutanoic acid 2-methylbutanoic acid
(ethylhexanoate) (1octen-3-one)
Dried sausages -
221
III. Sensory evaluation
TABLE 5-continued 1084 1091 1106 1139 1145 1154 1165 1171 1177 1186
nettle, dry, pelargonium orange nettle, plastic, weak hot soil, mushroom, sharp, metallic popcorn, fungi, mouse, weak plastic, nettle, plant fruity, weak mushroom deepfried, weak honey
all octanal all - P, Q all-C,D,F,I A, C, D, E, J, L, M, 0 3-octen-2-one all - A, D, E, F all - D, E, F, I K N, 0, P (x-nonen-y-one) all - I, 0 trans-2-octenal G I, K Q B, E, G I, K, L, M, 0, Q
1187
nettle, dry, pelargonium
B, C F, G I, K M N I-’
1190 1193 1196 1199 1202 1219 1233 1250 1255 1266 1282 1296 1308 1324
coffee, burned, cress orange fresh, nauseous Jerusalem artichoke, burned plastic orange, weak cooked meat bandage, cooked frankfurter, strong deepfried, weak popcorn vitamins, cooked meat cucumber cucumber staple, horse dried flowers
1330 1339 1357
bandage, cooked frankfurter pelargonium deepfried
all - A, B, M, P, 0 all - J, L, Q all - 0 all - A, C, J, L, N, P K L, M, N, 0, P C, F, G N all E, I, K M, Q all - D, F all - C all D, E, I, K, M, N, P all - F B, C, E, C. K, M, N, 0, Q all - A, D, I, L C, K L, M all
nonanal
methoxyphenol
= guiaicol
trans-2-nonenal methylphenol
= cresol
‘Kovats indices of the used GC-method. bOnly odours detected in more than two sausages are included; cf. Table 1 for explanation of sausage types. ‘Identified components that may be responsible for odour. Compounds in parenthesis are suggestions based on odour and RT-index, cf. also Table 2 in Part II.
burned smell when present in larger amounts. ‘Burned’ could arise from a mixture of components or from components not identified because of their very low concentrations, but having low sensory threshold values (e.g. the dimethylpyrazines, cf. Table 5). The fact that the four principal components only explain 54% of the variance in the sensory data set (Y-matrix, see Table 4) also shows that many important aroma components may not have been detected in this experiment. Table 4 also shows that only 54% of the variation of the X-matrix is explained by the first four principal components. This could indicate that many of the volatiles are of no importance to aroma. Even if this is so, leaving out compounds like the saturated alcohols (Table 5 shows that they have little aroma value) only results in a slight increase in degree of explanation.
222
L. H. Stahnke
Effluent snifling
Table 5 shows the outcome of the effluent sniffing. The identified components which are proposed as responsible for the odours are compiled in Table 2 of the previous paper. The sniffing procedure was quite simple there being no attempt to quantify the smells, but simply detecting their presence. In order to find any correlation between the detected odours and the sensory data, attempts were made to do a PLS analysis on the results; but these were not successful, partly because the sniffed odours were not quantified. Binomial data ( = present/non-present) were used instead. However Table 5 clearly shows that there are many components with characteristic odours that have not been detected/identified by the gas chromatographic and mass spectrometric analyses. An earlier study showed this as well (Stahnke, 1994). In order to obtain a fuller picture of the aroma development in fermented sausages it would be necessary to identify all of the components with sensory value. The identified components of special importance to aroma seem to be the volatile acids, many aldehydes and several esters and ketones. Also two different dimethylpyrazines may be of importance. It appears that saturated alcohols have too high sensory threshold values to be detected individually, but perhaps they are of some importance in the complete aroma of the sausages. Many of the unidentified odours are of character which suggests that they probably arise from breakdown of sulfuric amino acids, e.g. the odours at RTindex 486, 742, 774, 1000. Other odours have notes reminiscent of long, polyunsaturated aldehydes, e.g. those at RT-index 1250, 1296, 1357; and some smell like amides, e.g. the odours at 1017, 1145, 1255. It is notable that more than half of the sausage types possessed the same odours. This indicates that differences in the overall aroma of fermented sausages are caused by differences in the ratio of aroma components and not so much by development of specific compounds in certain types of sausages. TABLE 6 Aroma Profile of Two Types of Sausages Sausage type
Sensory descriptoP Old-fashioned
Modern
Low temperature, added nitrate
High temperature, nitrite, glucose, P. pentosaceus
Salami Sourdough Sourish Cheese Fatty Rancid Nauseous Burned Solvent “Aroma strength is increased in direction of arrow.
+ t + -+ + ct t+ u +
Dried sausages -
Ill. Sensory evaluation
223
CONCLUSIONS Table 6 summarizes the aroma profile of two types of sausages. The data are based on the linear regression model, assuming independency between the sensory descriptors. Table 6 shows that salami odour is more pronounced in ‘oldfashioned’ sausages. Also, ‘old-fashioned’ sausages have more sourdough, fatty and solvent-like smells while ‘modern’ types of fermented sausages have more sourish and cheesy aroma notes. ‘Salami’ seems to correlate with the presence of ethyl esters and 2-alkanones as well as with high numbers of S. xylosus. Also 2- and 3-methylbutanal may be of importance. Sourdough odour is influenced by the same components as salami odour, mixed with cheesy-smelling acids like 2-methylpropionic and 3-methylbutanoic acid. Fatty aroma is probably related to the content of diacetyl as well as the aldehydes hexanal and 3-methylhexanal. Sourish notes to salami odour seem to correlate especially with acetic and perhaps butanoic acid. Cheesy aroma is correlated with high amounts of volatile acids like 2-methylpropionic, butanoic and 3-methylbutanoic acid. ACKNOWLEDGEMENTS This study was sponsored by the Center for Food Research at Technical University of Denmark, DK-2800 Lyngby, Denmark. REFERENCES Acree, T. E., Barnard, J. & Cunningham, D. G. (1984). Food Chern., 14, 273. Aishima, T. & Nakai, S. (1991). Food Rev. Znt., 7(l), 33. Berdague, J.-L., Monteil, P., Montel, M. C. & Talon, R. (1993). Meat Sci., 35, 275. Berger, R. G., Macku, C., German, J. B. & Shibamoto, T. (1990). J. Food Sci., 55, 1239. Bills, D. D., Morgan, M. E., Libbey, L. M. & Day, E. A. (1965). J. Dairy Sci., 48, 1168. Chien, M. 8z Peppard, T. (1993). In Flavor measurement, eds. C.-T. Ho & C. H. Manley. Marcel Dekker, Inc., New York. p. 1. Fischer, K.-H. & Grosch, W. (1987). Lebensm. Wiss. Techn., 20, 233. Grosch, W. (1993). Trends Food Sci. Techn., 4(3), 68. Hosono, A., Elliott, J. A. & McGugan, W. A. (1974). J. Dairy Sci., 57(5), 535. Johansson, G., Berdague, J.-L., Larsson, M., Tran, N. & Borch, E. (1994). Meat Sci., 38, 203. Martens, H. (1982). In Food research and data analysis, eds. H. Martens & H. Russwurm, Jr. Appl. Science Publishers, London, p. 5. Stahnke, L. H. (1994). Meat Sci., 38, 39.