Investigation of spoilage in saveloy samples inoculated with four potential spoilage bacteria

Meat Science 93 (2013) 687–695

Contents lists available at SciVerse ScienceDirect

Meat Science journal homepage: www.elsevier.com/locate/meatsci

Investigation of spoilage in saveloy samples inoculated with four potential spoilage bacteria E.S. Holm a,⁎, A. Schäfer b, A.G. Koch b, M.A. Petersen a a b

Department of Food Science, Quality & Technology, Faculty of Science, University of Copenhagen, Rolighedsvej 30, 1958 Frederiksberg C, Denmark DMRI, Danish Technological Institute, Maglegårdsvej 2, 4000 Roskilde, Denmark

a r t i c l e

i n f o

Article history: Received 30 May 2012 Received in revised form 24 September 2012 Accepted 10 November 2012 Keywords: Microbial spoilage Volatile organic compounds Sensory profiling Partial least squares regression

a b s t r a c t Sliced saveloy samples were inoculated with monocultures of four potential spoilage bacteria and studied during a four week storage period. The objective was to investigate the resulting changes in the composition of Volatile Organic Compounds (VOCs) and the sensory quality of the product. Based on the sensory scores and the VOC composition Brochothrix thermosphacta, Chryseomonas luteola and Carnobacterium maltaromaticum were found to have a high spoilage potential in saveloy samples subjected to consumer simulated storage during the fourth week. Inoculation with Leuconostoc carnosum only resulted in a low level of spoilage. The sensory changes in the saveloy samples were modeled based on the VOC composition using Partial Least Squares Regression. The changes in the six sensory descriptors were closely related to the amount of diacetyl, acetoin, 2- and 3-methylbutanol, 2- and 3-methylbutanal and 2-methylpropanol found in the samples. These compounds are therefore potentially important for the shelf-life of sliced saveloy. © 2012 Elsevier Ltd. All rights reserved.

1. Introduction Microbial spoilage is the major reason for loss of consumer acceptability in cooked and sliced meat products during refrigerated storage (Borch, KantMuermans, & Blixt, 1996; Nychas, Skandamis, Tassou, & Koutsoumanis, 2008; Samelis, Kakouri, & Rementzis, 2000). The impact of microbially produced Volatile Organic Compounds (VOCs) on the quality of meat products has been demonstrated in the previous literature (Borch et al., 1996; Dainty & Hibbard, 1983; Ercolini et al., 2011; Leroy, Vasilopoulos, Van Hemelryck, Falony, & De Vuyst, 2009; Pham et al., 2008). The use of microbially produced VOCs as quality markers in cooked and sliced meat products was investigated in Holm, Schafer, Skov, Koch, and Petersen (2012). Here the microbially produced VOCs diacetyl, acetoin, 2- and 3-methylbutanol and 2- and 3-methylbutanal were found to be closely linked to the sensory evaluation of saveloy samples sliced at three different locations. The spoilage potential of the microorganisms found in cooked and sliced meat products varies greatly. The composition of the spoilage flora is therefore an important factor for the rate and extent of the quality changes observed during storage (Holm et al., 2012; Mataragas, Skandamis, Nychas, & Drosinos, 2007; Nychas et al., 2008). The microbial flora of the fresh meat is largely destroyed during the cooking process. Recontamination during slicing and handling therefore largely determines the composition of the spoilage flora (Gounadaki, Skandamis, Drosinos, & Nychas, 2008; Laursen, Byrne, Kirkegaard, & Leisner, 2009; Samelis et al., 2000). The spoilage flora of cooked and sliced meat ⁎ Corresponding author. Tel.: +45 28726541; fax: +45 72202744. E-mail address: [email protected] (E.S. Holm). 0309-1740/$ – see front matter © 2012 Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.meatsci.2012.11.016

products is typically composed of Lactic Acid Bacteria (LAB) and Brochothrix thermosphacta (Borch et al., 1996; Holm et al., 2012). The product composition and the storage conditions may favor growth of specific spoilage organisms, which will be largely responsible for the formation of spoilage (Nychas et al., 2008; Samelis et al., 2000). In this study sliced saveloy was inoculated with monocultures of the four potential spoilage bacteria B. thermosphacta, Leuconostoc carnosum, Carnobacterium maltaromaticum and Chryseomonas luteola. B. thermosphacta is an important spoilage organism in meat products well known for the production of diacetyl and acetoin under aerobic conditions (Dainty & Hibbard, 1983; Stanley, Shaw, & Egan, 1981). Pseudomonas ssp. has been associated with spoilage of fresh meat due to their rapid growth under aerobic conditions and production of sulfur compounds with low odor threshold values (Borch et al., 1996; Dainty, Edwards, & Hibbard, 1984). C.luteola is a facultative anaerobic bacterium belonging to the Pseudomonas genus. The term Lactic acid bacteria refers to a huge group of bacteria of which several genera are important for the industrial manufacturing of different foods. The spoilage potential of the different LAB varies but several studies have reported specific LAB as the main cause of spoilage in cooked meat products (Bjorkroth, Vandamme, & Korkeala, 1998; Korkeala & Bjorkroth, 1997; Samelis et al., 2000). The objective of this study was to investigate the quality changes in saveloy samples caused by the four bacteria from the point of inoculation to the end of the shelf-life period. The study was based on measurements of the VOC composition, the sensory profile and the microbial composition of the product. The sensory changes in the sliced saveloy samples were furthermore modeled based on the VOC composition using Partial Least Squares Regression (PLSR). This was done to study the

688

E.S. Holm et al. / Meat Science 93 (2013) 687–695

relation between specific VOCs and sensory evaluations of saveloy, and to study the possibility of predicting the sensory quality of saveloy samples based on changes in the VOC composition. 2. Materials and methods 2.1. Experimental setup The experiment included six different treatments of sliced saveloy samples. In four of these treatments saveloy samples were inoculated with monocultures of either B. thermosphacta, C. luteola, L. carnosum or C. maltaromaticum. One treatment involved inoculation with a 1:1:1:1 mixture of the four bacteria and the final treatment was an uninoculated control. The effect of the six treatments was studied in a four week storage period. During the first three weeks of the experiment the saveloy samples were stored at 5 °C in Modified Atmosphere Packaging (MAP) consisting of 70% N2 and 30% CO2. During the fourth week of the experiment the storage conditions were changed to Consumer Simulated Storage (CSS). Consumer simulated storage included package opening and storage at a daily repeated temperature program: 5 °C for 12.7 h, 12 °C for 9.8 h and 20 °C for 1.5 h. This temperature program has been shown to represent storage by Danish consumers (Blom-Hansen, 2004). Opening of the packages breached the modified atmosphere and allowed the oxygen to enter. However, to prevent drying of the product surface a plastic lid permeable to oxygen was applied. A group of samples was not subjected to CSS but kept for four weeks in MAP at 5 °C before analysis. During the experiment the saveloy samples were stored in MAP at 5 °C and analyzed at day 1, week 1, week 2, week 3, and week 3 + 3 days with CSS, week 3 + 7 days with CSS and after 4 weeks without. The analyses included a sensory profiling, measurement of the microbial composition and measurement of the VOC composition using dynamic headspace extraction coupled with Gas chromatography– mass spectrometry (GC–MS). Three replicate measurements for each combination of factors were performed for each of the analyses. Separate packages of saveloy were used for analysis of each replicate. A total of 396 packages were analyzed during the experiment. 2.2. Production and slicing of saveloy The saveloy was produced at the Danish Meat Research Institute (DMRI) using the same recipe and procedure as described in Holm et al. (2012). The product was seasoned with nutmeg, coriander and black pepper and was preserved with 60 ppm nitrite and 3% sodium chloride in the aqueous phase. The saveloy was cut in 1 cm slices of approximately 25 g and two slices were placed in an APET/PE-peel tray with an Oxygen Transfer Rate (OTR) of 15 mL/m 2 (24 h, 1 atm, 23 °C). The samples were inoculated according to the six treatments and the packages were sealed with a PETP12/PE-peel film, with an OTR of 5 mL/m 2. 2.3. Culture development and inoculation Frozen cultures (− 80 °C) of B. thermosphacta (4618), C. luteola (4616), L. carnosum (4010), and C. maltaromaticum (4318) were obtained from the culture collection at DMRI. The cultures were grown in Brain Heart Infusion (BHI) broth (Oxoid Ltd., Basingstoke, UK) and kept for three days at 20 °C. The cultures were then spread on BHI agar and incubated for five days at 20 °C. A single colony was transferred to BHI broth for enrichment at 20 °C for four days before inoculation of the saveloy. Each saveloy slice was inoculated with 0.1 mL broth diluted to a concentration of 50–100 cells of the specific bacteria. The mixture of all four bacteria was prepared just before inoculation and contained 50–100 cells of each of the four bacterial species.

2.4. Dynamic headspace extraction of VOCs and GC–MS analysis The extraction of VOCs from the saveloy samples was done using dynamic headspace sampling on traps packed with Tenax and Carbograph (Markes International Ltd., Llantrisant, UK). The dynamic headspace extraction and the subsequent GC–MS analysis were described by Holm et al. (2012). However, the N2 flow used to back purge the traps was set to 20 mL/min for 5 min. Measurements of the VOC composition were done at: day 1, week 1, week 2, week 3, week 3 + 3 days, week 3 + 7 days and week 4. The retention times were standardized using the Kovats Linear Retention Index (LRI) calculated from GC–MS runs of a C5–C15 alkane standard (Air Liquide, Paris, France). GC–MS runs of Tenax traps spiked with the following authentic standards were used for identification purposes: 2-methylbutan-1-ol and hexanal (Merck KGaA, Darmstadt, Germany), 3-methylbutanal and 3-hydroxy2-butanone (acetoin) (ChemService inc., West Chester, PA, USA), 1-octen-3-ol, acetic acid, 2,3-butandione (diacetyl), 2-methyl-1propanol, 2-heptanol and 1-hexanol (Sigma-Aldrich, St. Louis, MO, USA), octanal, 2-heptanone, heptanal, α-pinene, limonene, 3-carene, dimethyl disulfide and dimethyl trisulfide (Acros Organics, Geel, Belgium), and 2-pentylfuran (Lancaster Synthesis, Windham NH, USA). The GC–MS data was processed using the MSD Chemstation software (D.01.02.16, Agilent Technologies, Santa Clara, CA, USA). The peak areas were calculated based on selected target ions. Due to co-elution and similar mass spectra some peaks were not quantified individually. 2-Methylbutanal and 3-methylbutanal were quantified together at LRI 924. 2-Methylbutanol and 3-methylbutanol were quantified together at LRI 1197. Diacetyl and pentanal were quantified at LRI 970. However, diacetyl was found to be responsible for the observed increase of this peak. Three different methods were used for identification of the chromatographic peaks. The mass spectra of the peaks were compared with mass spectra of potential matching compounds using the NIST/ EPA/NIH mass spectral library (V.1.7a, Agilent Technologies, Santa Clara, CA, USA). The LRI of the chromatographic peaks was compared with the LRI of potential matching compounds using the C20M column from internet database www.flavornet.org (Acree & Arn, 2004). The maximum difference in LRI accepted for a potential match was 50 units. Finally the mass spectra and LRI were compared to GC–MS measurements of authentic standards of compounds relevant for the shelf-life of saveloy. 2.5. Sensory analysis The sensory profiling was carried out as described in Holm et al. (2012) using a trained panel of eight assessors. During training sessions the panel selected six sensory descriptors related to odor. Saveloy odor and spicy odor were positively associated with consumer acceptability. Butter-like odor, sweet odor, sour odor and old odor were negatively associated with consumer acceptability. Butter-like odor was selected based on a reference of acetoin, whereas spicy odor was selected based on a sample of the spice mixture used for saveloy production. The remaining descriptors were selected by assessment of saveloy samples at different stages of the shelf-life period. The sensory analysis was performed at: day 1, week 3, week 3 + 3 days, week 3 + 7 days and week 4. 2.6. Microbial analysis The analysis of the microbial composition of the saveloy samples was carried out as described in Holm et al. (2012). Four different growth substrates were used. BHI (Oxoid Ltd., Basingstoke, UK) was used for the total microbial count. Streptomysin and Thallous Acetate agar (STA, Oxoid Ltd., Basingstoke, UK) with STA selective supplement was used to detect B. thermosphacta. Pseudomonas agar base with Cetrimide, Fucidin and Cephalosporin agar supplement (CFC,

E.S. Holm et al. / Meat Science 93 (2013) 687–695

689

Different (HSD) tests were done to find significant differences, at a 95% level, between the levels of the tested variables. Chemometric analysis was done using the PLS Toolbox (v.5.2.2, Eigenvector Research inc., Wenatchee, WA, USA). The PLS toolbox is running in the MATLAB environment (version 7.6.0.324, The Matworks Inc., Natick, MA, USA). All PLSR-models and Principal Component Analysis (PCA) were validated using random subset validation.

Oxoid Ltd., Basingstoke, UK) was used to detect C. luteola. All Purpose medium with Tween containing polymyxin (APT, Merck KGaA, Darmstadt, Germany) with a top layer of APT was used to detect LAB. However, this substrate only inhibits growth of gram negative bacteria thus allowing growth of for example B. thermosphacta. Saveloy samples were spread on BHI at day 1, week 3, week 3+3 days, week 3+7 days and week 4 whereas the other substrates were used at day 1, week 3+7 days and week 4.

3. Results and discussion 2.7. Data processing 3.1. The VOC composition of the inoculated saveloy samples Statistical analysis was performed in JMP v.8.0.1 (SAS institute, Cary, NC, USA). Two ANOVA models were built for each analysis. The ANOVA MAP model included samples stored in MAP (day 1, week 1, week 2, week 3, week 4) whereas the ANOVA MAP + CSS included the samples stored in MAP for three weeks and subjected to CSS during the final week (week 3 + 3 days, week 3 + 7 days). The ANOVA models included the variables ‘storage time’, ‘treatment’ and ‘storage time × treatment’ as fixed effects. For the sensory descriptors the ANOVA models furthermore included ‘assessor’ and ‘repetition’ as fixed effects. Based on the ANOVA models Tukey Honest Significant

An overview of the 42 different VOCs that were identified in the headspace of sliced saveloy during the experiment is provided in Table 1. Nineteen of these VOCs were classified as terpenes which were thought to derive from the spices added to the product. Alcohols, ketones and aldehydes were the other major classes of compounds isolated from the saveloy samples. Table 1 shows the formation of VOCs by the potential spoilage bacteria inoculated on the saveloy samples throughout the experiment. The ANOVA MAP model describes the VOC formation under anaerobic conditions whereas the ANOVA MAP+CSS model

Table 1 An overview of VOCs extracted from the saveloy samples for each treatment: Control (C), Brochothrix thermosphacta (Bt), Chryseomonas luteola (Cl), Leuconostoc carnosum (Lc), Carnobacterium maltaromaticum (Cm) and the Mixture samples (M). The linear retention index (LRI) is shown along with the target ion used for quantification. Three methods were used for compound identification: The NIST mass spectral data base (N), comparison of the observed LRI with LRI from flavornet.com (L), GC-MS runs of authentic standard compounds (S). ANOVA MAP includes samples from day 1, week 1, week 2, week 3 and week 4. ANOVA MAP + CSS include samples from week 3 + 3 days and week 3 + 7 days. Letters a to d are used to indicate significant differences in the peak area of the VOCs between treatments based on a Tukey HSD test. The letter a represents the largest peak area. Nr.

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42

Code

A1 K1 K2 A2 Ac1 K3/A3 T1 T2 T3 S1 A4 T4 Ac2 T5 T6 T7 T8 T9 K4 A5 T10 T11 Ac3 F T12 Ac4 Ac5 T13 T14 K5 A6 Ac6 Ac7 S2 A7 T15 Ac8 Ca T16 T17 T18 T19

Compound name

Ethanal Acetone 2-Butanone 2- and 3-methylbutanal Isopropyl alcohol/ethanol Diacetyl/pentanal α-Pinene α-Thujene Camphene Dimethyl disulfide Hexanal β-Pinene 2-Methyl-1-propanol Sabinene 3-Carene α-Phellandrene β-Myrcene α-Terpinene 2-Heptanone Heptanal Limonene Eucalyptol 2- and 3-methylbutanol 2-Pentylfuran γ-Terpinene 3-Methylbutenol 1-Pentanol p-Cymene δ-Terpinene Acetoin Octanal 2-Heptanol 1-Hexanol Dimethyl trisulfide Nonanal α-p-Dimethylstyrene 1-Octen-3-ol Acetic acid Camphor Linalool Caryophyllene 1-Terpinen-4-ol

LRI

700 811 901 924 928 970 1012 1018 1050 1060 1073 1085 1073 1102 1131 1148 1153 1163 1173 1175 1181 1173 1197 1218 1228 1239 1239 1253 1253 1269 1273 1306 1340 1362 1383 1422 1436 1440 1447 >1447 >1447 >1447

T-Ion

44 43 43 41 45 43 93 93 93 94 44 93 43 93 93 93 93 121 43 70 68 43 55 81 93 41 42 119 121 45 43 45 56 126 43 117 57 45 95 71 133 71

ID-method

L-N N L-N S-L-N S-L S-L-N S-L-N N L-N S-L-N S-L-N L-N S-L-N L-N S-L-N L-N L-N L-N S-L-N S-L-N S-L-N L-N S-L-N S-L-N L-N N L-N L-N L-N S-L-N S-L-N S-L-N S- L-N S-L-N L-N L-N S-L-N S-L-N L-N N N N

ANOVA MAP

ANOVA MAP + CSS

C

Bt

Cl

Lc

Cm

M

C

Bt

Cl

Lc

Cm

M

a a a bc a bc a a

b b b b bc c ab ab

bc bc b bcd c a ab ab

b cd c cd ab d ab b

c bcd b a c b ab bc

c d c d ab d b c

abc b abc c c d

bc b ab b c a

a a a c c c

ab b bc c b d

b b a a c b

bc b c c a d

a a a b

ab a ab ab

a a abc ab

ab b abc ab

b a bc ab

b c c a

b a

b a

a a

b b

b a

b b

c

b

a

c

c

c

a a

ab ab

ab bc

ab c

bc bc

c c

a a a a

ab ab ab b

ab b ab bc

ab c bc c

bc b abc c

c d c d

bc ab

ab a

a ab

cd ab

ab a

d b

b a a ab ab

c b ab a bc

b b b ab bc

ab b b ab a

ab b bc ab c

a c c b ab

b

a

b

b

a

b

a

a

ab

ab

ab

b

c

bc

c

ab

ab

a

a b a b b a a

ab b a b b c a

bc a ab b b ab ab

bc b ab a a abc ab

cd b ab b b c ab

d b b a a bc b

a c ab b c b

a a b b bc b

ab b b b bc a

ab c a a a b

ab a b b b b

b c a a a b

a a a a ab a

ab bc ab a a a

cd ab ab a ab a

abc bc ab a ab a

bcd c b ab ab ab

d c c b b b

ab

a

ab

ab

ab

b

690

E.S. Holm et al. / Meat Science 93 (2013) 687–695

shows the effect of package opening and storage at temperature abuse during the fourth week of storage. All VOCs found in elevated amounts in the inoculated samples compared to the control samples were expected to derive from microbial metabolism. 3.1.1. Changes in the VOC composition during storage The microbial metabolites diacetyl, acetoin, 2- and 3-methylbutanol, and 2- and 3-methylbutanal are of particular interest in this experiment as they previously have been linked to spoilage of different types of processed meats including saveloy (Holm et al., 2012; Laursen, Leisner, & Dalgaard, 2006). The development in the peak area of these compounds is shown in Fig. 1 for saveloy samples inoculated with B. thermosphacta, C. maltaromaticum and C. luteola. The L. carnosum samples and the Mixture samples did not contain an elevated amount of these compounds and are not included in the figure. Fig. 1 shows that inoculation with B. thermosphacta, C. maltaromaticum and C. luteola resulted in a considerable increase in acetoin and diacetyl after the onset of CSS. Furthermore, the amount of acetoin and diacetyl found in saveloy samples inoculated with these bacteria was very low after four weeks of storage in MAP at 5 °C compared to week 3 + 7 days with CSS. As seen in Fig. 1, 2 and 3-methylbutanal were mainly produced by B. thermosphacta and C. maltaromaticum, whereas 2- and 3-methylbutanol were produced by B. thermosphacta, C. maltaromaticum and C. luteola. From Table 1 it is seen that the production of 2- and 3-methylbutanal and 2- and 3-methylbutanol by B. thermosphacta only increased at CSS. Fig. 1 shows that elevated amounts of 2- and 3-methylbutanal were observed both after 3 weeks and 4 weeks in MAP for samples inoculated with C. maltaromaticum. However, the production of 2- and 3-methylbutanal as well as 2- and 3-methylbutanol by C. maltaromaticum was accelerated by CSS. For the C. luteola samples the formation of these VOCs was also accelerated during CSS. The observed patterns in the formation of diacetyl, acetoin, 2- and 3-methylbutanol, and 2- and 3-methylbutanal by B. thermosphacta and C. maltaromaticum correspond well with observations made in

previous studies on raw meat, fermented meat products and shrimp (Dainty & Hibbard, 1983; Ercolini et al., 2011; Larrouture-Thiveyrat, Pepin, Leroy-Setrin, & Montel, 2003; Laursen et al., 2006). As seen in Fig. 2 samples inoculated with C. luteola contained higher amounts of DiMethyl DiSulfide (DMDS) and DiMethyl TriSulfide (DMTS) during storage at CSS compared to the other treatments. Spoilage of fresh beef by Pseudomonas spp. due to formation of various sulfur compounds, including DMDS and DMTS, during aerobe incubation has previously been reported (Dainty et al., 1984). Fig. 2 also illustrates that C. luteola and B. thermosphacta produced 2methylpropanol during CSS. From Table 1 it is seen that the samples inoculated with L. carnosum were characterized by elevated amounts of 1-hexanol and 2-heptanol compared to the control samples. The samples inoculated with a mixture of all four potential spoilage bacteria also contained elevated amounts of these straight chain alcohols. This suggests that L. carnosum was dominating the Mixture samples as the other bacteria included in the experiment did not produce these alcohols to the same extent. The formation of 1-hexanol which was the most characteristic VOC produced by L. carnosum and the Mixture samples is shown in Fig. 2. 3.2. Sensory changes of the saveloy samples during storage The results of the sensory assessments of the saveloy samples are shown in Table 2. The sensory scores were averaged over assessor and repetitions for each combination of storage time and treatment. As for the GC–MS data the ANOVA MAP and the ANOVA MAP + CSS models were used to describe the differences between the six treatments. From Table 2 it is seen that inoculation with B. thermosphacta, C. maltaromaticum and C. luteola generally had a negative effect on the sensory evaluations of the product during the experiment. The sensory scores of sweet, sour and old odor were all significantly increased during the storage at CSS compared to the control samples. The scores in spicy odor and saveloy odor were furthermore significantly decreased

120

80

40

2- and 3- methylbutanal

60

Peak area /1000

Peak area /1000

2- and 3- methylbutanal

50 40 30 20 10

0

Control

0

B. thermosphacta

C. luteola

C. maltaromaticum

Control

Peak area /1000

Peak area /1000

C. luteola

C. maltaromaticum

Diacetyl/pentanal

Acetoin 3000

2000

1000

600

400

200

0

0

Control

B. thermosphacta

B. thermosphacta

C. luteola

C. maltaromaticum

Control

B. thermosphacta

C. luteola

C. maltaromaticum

Fig. 1. The peak areas and standard deviation of: 2- and 3-methylbutanal, 2- and 3-methylbutanol, acetoin and diacetyl/pentanal. Peak areas are shown for day1 (D1), week1 (W1), week2 (W2), week3 (W3), week3 + 3 days with consumer simulated storage (CSS), week3 + 7 days with CSS and week4 (W4). Only the relevant treatments are included.

Peak area /1000 Tusinde

E.S. Holm et al. / Meat Science 93 (2013) 687–695

Peak area /1000

DMDS 60 40 20 0

Control

14

DMTS

12 10 8 6 4 2 0

C. luteola

Control

2-methyl-1-propanol

40

20

0

C. luteola

1-hexanol

30

Peak area /1000

Peak area /1000

60

691

20

10

0

Control

B. thermosphacta

C. luteola

C. maltaromaticum

Control

Leuc. carnosum

Mixture

Fig. 2. The peak area and standard deviation of: DMDS, DMTS, 2-methyl-1-propanol and 1-hexanol. Peak areas are shown for day1 (D1), week1 (W1), week2 (W2), week3 (W3), week3 + 3 days with consumer simulated storage (CSS), week3 + 7 days with CSS and week4 (W4). Only the relevant treatments are included.

during CSS in samples inoculated with these bacteria. For samples inoculated with C. luteola old odor was the most characteristic sensory descriptor. As seen in Table 2 the overall highest scores in old odor was found for the C. luteola samples measured at week 3 + 7 days with CSS. When comparing the samples stored in MAP compared to those subjected to CSS in Table 2, it is seen that that B. thermosphacta, which is a facultative anaerobic bacterium, only caused spoilage at CSS. For the samples stored in MAP there was no significant difference between the control samples and the B. thermosphacta samples in any of the sensory descriptors. However, as seen in Table 2 samples inoculated with C. maltaromaticum and C. luteola received negative sensory evaluations, compared to the control samples, both when stored in MAP for four weeks, and after 3 weeks in MAP plus seven days at CSS. This effect was more apparent when studying the storage time× treatment effect (data not shown). For the samples inoculated with L. carnosum a significant increase in sour odor compared to the control samples was observed for samples stored at CSS. However, the effect of inoculation with L. carnosum on the sensory descriptors was generally not as pronounced as for samples inoculated with B. thermosphacta, C. maltaromaticum and C. luteola. This was further highlighted when investigating the storage time×treatment effect (data not shown) and when comparing the individual scores in Table 2. As seen in Table 2 the sensory scores of the L. carnosum samples and the Mixture samples could not be separated statistically. This further indicates that the Mixture samples were dominated by L. carnosum. 3.3. Overview of relation between the sensory evaluations and the VOC composition with PCA In this study both spoilage flora, storage time and storage conditions influenced the VOC composition and the sensory profile of the saveloy samples. The effect of storage time×treatment was also found to be significant in many cases. However, this effect was not included in Table 1 and Table 2. An overview of the relation between the saveloy samples, the sensory profile and the VOC composition is therefore given in Fig. 3. In this figure the bi-plot (Principal Component (PC)1 vs. PC2)

from a four component PCA model including all the VOCs and the sensory descriptors is shown. According to the bi-plot, saveloy samples inoculated with B. thermosphacta, C. maltaromaticum and C. luteola and measured at week 3 + 3 days and week 3 + 7 days plus the C. maltaromaticum samples measured after 4 weeks in MAP are clustered with negative scores on PC1 and positive scores on PC2. These samples, which were the most spoiled in the experiment, are associated with the sensory descriptors sweet, butter-like, sour and old odor and the VOCs 2- and 3-methylbutanol (Ac3), 2- and 3-methylbutanal (A2), diacetyl (K3) and acetoin (K5) and 2-methylpropanol (Ac2). The L. carnosum samples and the Mixture samples measured at week 3 + 3 days, week 3 + 7 days and week 4 are clustered and are associated with the VOCs 1-hexanol (Ac7) and 2-heptanol (Ac6). These samples are clearly separated from the most spoiled samples on PC2. All the terpenes (T1–T19), the straight chain aldehydes ethanal (A1), hexanal (A4), heptanal (A5) and octanal (A6) and the ketones acetone (K1) 2-butanone (K2) and 2-heptanone (K3) are positively associated with saveloy odor and spicy odor with positive scores on PC1. The straight chain aldehydes and ketones are probably formed in thermal lipid oxidation processes during the cooking (Mottram, 1998). As seen in the bi-plot these lipid oxidation products together with the terpenes make up the odor of fresh saveloy. 3.4. Microbial composition of the saveloy samples The purpose of the microbial analysis was to examine the microbial growth in the six treatments and secondly to validate the inoculation step in the experiment. In Table 3 the microbial counts are expressed as the logarithm to the total number of Colony Forming Units (CFU) per grams of sample. 3.4.1. Test of the inoculation step The microbial examination of the saveloy samples showed a good match between the CFU on BHI and on the different selective growth media (CFC, STA and ATP, Table 3). Thus samples inoculated with B. thermosphacta reached comparable counts on BHI, on ATP and

692

E.S. Holm et al. / Meat Science 93 (2013) 687–695

Table 2 Includes the sensory scores of each descriptor averaged over assessor and repetitions for each treatment: Control (C), Brochothrix thermosphacta (Bt), Chryseomonas luteola (Cl), Leuconostoc carnosum (Lc), Carnobacterium maltaromaticum (Cm), and Mixture (M). ANOVA MAP includes samples from day 1, week 1, week 2, week 3 and week 4. ANOVA MAP + CSS include samples from week 3 + 3 days and week 3 + 7 days. Letters a to d are used to indicate significant differences between treatments based on a Tukey HSD test. Sensory descriptor

Saveloy odor

Spicy odor

Butter-like odor

Sweet odor

Sour odor

Old odor

Storage time

Day1 Week3 Week4 ANOVA MAP Week3 + 3 Week3 + 7 ANOVA MAP + CSS Day1 Week3 Week4 ANOVA MAP Week3 + 3 Week3 + 7 ANOVA MAP + CSS Day1 Week3 Week4 ANOVA MAP Week3 + 3 Week3 + 7 ANOVA MAP + CSS Day1 Week3 Week4 ANOVA MAP Week3 + 3 Week3 + 7 ANOVA MAP + CSS Day1 Week3 Week4 ANOVA MAP Week3 + 3 Week3 + 7 ANOVA MAP + CSS Day1 Week3 Week4 ANOVA MAP Week3 + 3 Week3 + 7 ANOVA MAP + CSS

Sensory scores C

Bt

Cl

Lc

Cm

M

5.9 4.9 4.7 a 5.0 4.4 a 4.4 3.8 3.5 – 3.8 3.5 a 0.8 0.5 0.5 b 0.6 0.2 b 1.0 0.5 0.6 b 0.7 0.9 b 0.5 0.5 1.2 b 0.6 0.6 c 0.6 0.3 1.6 c 0.8 1.2 c

5.9 4.9 4.6 a 2.3 1.6 cd 4.4 3.8 3.5 – 2.2 1.5 cd 0.5 0.3 0.7 b 4.5 7.5 a 0.9 0.5 0.9 b 3.9 4.4 a 0.7 0.6 1.5 b 4.6 5.9 ab 0.8 0.6 1.5 c 5.8 5.6 a

5.8 5.1 2.2 b 2.0 1.3 d 4.4 4.0 1.7 – 1.7 1.1 d 0.8 0.4 1.4 b 1.5 2.1 b 0.9 0.4 2.3 b 2.8 3.6 a 0.9 0.2 3.3 ab 2.3 5.3 b 1.0 0.2 5.2 ab 4.8 8.7 a

6.3 3.9 2.6 ab 3.5 2.8 b 5.1 3.1 2.2 – 2.9 2.4 b 0.4 0.7 1.1 b 0.9 0.6 b 0.9 1.0 1.3 b 1.3 0.8 b 0.5 1.6 2.7 ab 2.6 3.8 b 0.5 1.3 3.1 bc 2.4 3.4 bc

6.6 4.2 2.0 b 2.2 1.6 cd 4.9 3.3 1.7 – 2.2 1.4 cd 0.9 0.5 5.5 a 5.1 6.7 a 0.9 0.9 4.1 a 3.4 4.7 a 0.6 1.1 5.5 a 5.7 6.5 a 0.7 1.1 6.8 a 5.9 5.8 a

6.0 3.5 2.3 b 2.8 2.7 bc 4.7 3.1 2.0 – 2.6 2.1 bc 0.8 1.2 1.1 b 1.5 0.9 b 0.6 1.4 1.6 b 1.5 1.0 b 0.4 2.3 3.8 a 4.0 4.1 ab 0.5 1.9 3.8 ab 3.4 3.3 b

on the selective STA growth medium. For samples inoculated with C. luteola the BHI count corresponded well with the number of CFU of the pseudomonas selective CFC plates. Finally, the number of L. carnosum and C.maltaromaticum colonies detected on BHI was comparable to the number of colonies detected on the LAB growth medium ATP. Overall this suggests that the inoculation step was successful. As seen in Table 3 samples from all treatments, including the control, contained bacteria capable of growth at ATP-agar at day 1. This was ascribed to the background flora originating from handling of the saveloy or from spore forming bacteria surviving the cooking process. This background flora never exceeded 10 5.0 CFU/g and is therefore not expected to contribute to spoilage of the saveloy samples. 3.4.2. Microbial growth and spoilage potential Due to the varying spoilage potential of different bacteria it is difficult to evaluate the level of microbial spoilage using only the total microbial count. Nevertheless, growth to 10 7 CFU/g is often used as

an indicator of increased risk of microbial spoilage (Borch et al., 1996; Dainty & Mackey, 1992). As seen in Table 3 all the treatments inoculated with potential spoilage bacteria reached a total microbial count well above 10 7 CFU/g during the storage period. However, according to the sensory results B. thermosphacta, C. maltaromaticum and C. luteola caused heavy spoilage of the saveloy samples under CSS whereas L. carnosum only caused low levels of spoilage. The L. carnosum samples are therefore a good example showing that a high microbial count does not always equal spoilage. From Table 3 it is seen that the samples stored in MAP for four weeks reached growth levels above 10 7 CFU/g for all the tested bacteria except B. thermosphacta. This corresponds well with the VOC composition and the sensory evaluations of these samples, which showed that the facultative anaerobic B. thermosphacta did not spoil the saveloy samples packed in modified atmosphere. As seen in Table 3 the growth pattern of the Mixture samples was very similar to the growth pattern of the L. carnosum samples. At week 3 + 7 days the spoilage flora of the Mixture samples was furthermore dominated by LAB whereas the B. thermosphacta and C. luteola specific media only reached moderate plate counts. Both the sensory results and the VOC composition indicated that the L. carnosum was the dominating flora in the mixture of all four bacteria. This hypothesis is supported by the results of the microbial analysis.

3.5. Modeling the relation between the sensory evaluations and the VOC composition with PLSR PLSR-models were calculated for each of the six sensory descriptors in order to investigate the relation between the sensory descriptors and the VOC composition. The PLSR-models included all the 42 VOCs as X-variables. An overview of the characteristics of the PLSR-models is shown in Table 4. The table includes the validated Root Mean Square Error of Cross Validation (RMSECV), the number of PCs included in the model, the coefficient of determination (R 2) for the predicted vs. measured plot and the percentage of Y-variance explained by the model. The table also lists the ten VOCs which had the highest impact on the model based on their regression coefficient to the sensory descriptor. These compounds are sorted by the numeric value of the regression coefficient. The PLS-models described more than 70% of the variation in the sensory descriptors using two or three PCs. The RMSECV of the PLSR-models was generally relatively high compared to the span of the sensory scores. This suggests that the predictive ability of the PLSR-models was limited. The large variation in the VOC composition between the samples as a result of the six different treatments included in the PLSR-models is a probable explanation of this. Table 4 shows that 2- and 3-methylbutanol (Ac3), diacetyl (K3) and acetoin (K5), were among the four most important VOCs in the PLSR-models of all six sensory descriptors. 2- and 3-Methylbutanal (A2) was also among the top ten VOCs in all the PLSR-models, and the most important compound to the descriptors sweet odor and butter-like odor. 2-Methylpropanol (Ac2) was found in the top five most important VOCs in all the PLSR-models except that of butter-like odor. All the sensory descriptors used in this study therefore depend largely on the level of 2- and 3-methylbutanol, 2- and 3-methylbutanal, 2-methylpropanol, diacetyl and acetoin found in the headspace of the saveloy samples. Acetic acid (Ca), DMDS (S1), DMTS (S2) and octanal (A6) were also frequently found among the ten most important VOCs and could also be related to spoilage of saveloy samples. No terpenes were found among the most important VOCs in the PLSR-models of spicy odor or saveloy odor. This suggests that the shelf-life of the saveloy samples was limited by an increase in microbial metabolites rather than a decrease in the pleasant odors from the spices.

E.S. Holm et al. / Meat Science 93 (2013) 687–695

693

0.25

0.2

Bt3+7

K5

Cl3+7

A2

Cm3+7

PC 2 (14.4%)

T1-T19 Ac4 Ac5 A4 A5 A7 K1 K2 K4 F

K3/A3

Ac3

0.15

Ac2

Old

Ca

Sour

0.1

Cm3+3

S2 S1

C0

Bt3+3

Ac8 Cl0

0.05 Cm4

Bt0

Lc0

A1

Cl3+3

Cm0

0

Cl4 M0

Cm3

Lc3+7 Bt3 M3+3 Bt4 M3+7 M4 C4 Cl3 Lc3+3 Lc4 M3 Ac7 C3+7 Lc3 C3+3

-0.05

A6

C3

Ac1

Ac6

-0.1 -0.2

-0.1

0

0.1

0.2

0.3

0.4

0.5

PC 1 (62.9%) Fig. 3. Bi-plot (PC1 vs. PC2) from the PCA-model including all 42 VOCs and six sensory descriptors (squares) and the saveloy samples (triangles). The VOCs are denoted according to the compound code shown in Table 1. The samples are denoted according to the treatment: control (c), Brochothrix thermosphacta (Bt), Chryseomonas luteola (Cl), Leuconostoc carnosum (Lc) and Carnobacterium maltaromaticum (Cm) and the Mixture samples (M) and the time of measurement: day1 (0), week3 (3), week3+3 days (3+3), week 3+7 days (3+7) and week 4 (4).

3.6. Evaluation and practical application of the results This study shows that B. thermosphacta, C. maltaromaticum and C. luteola can cause severe spoilage of cooked and sliced meat products Table 3 An overview of the microbial counts of the six treatments on the four growth substrates given in log CFU/g. BHI indicates the total microbial count, CFC counts Pseudomonas spp., STA counts Brochothrix thermosphacta and ATP counts lactic acid bacteria and other gram positive bacteria.

Control

B. thermosphacta

Chryseomonas luteola

Leuconostoc carnosum

Carnobacterium maltaromaticum

Mixture

Time

BHI

CFC

STA

ATP

Day1 Week3 Week3 + 3 Week3 + 7 Week4 Day1 Week3 Week3 + 3 Week3 + 7 Week4 Day1 Week3 Week3 + 3 Week3 + 7 Week4 Day1 Week3 Week3 + 3 Week3 + 7 Week4 Day1 Week3 Week3 + 3 Week3 + 7 Week 4 Day1 Week3 Week3 + 3 Week3 + 7 Week 4

b1–2.5 b5.0 b5.0 b5.0 b5.0 2.5 ≤5.0 7.8 8.0 5.5 1.8 b5.0–6.0 8.6 9.1 b6.0–8.0 3.2 8.2 8.1 8.0 8.3 2.4 7.1 8.3 8.7 7.6 3.1 8.2 8.3 b6.0–8.0 8.3

b1.0 – – b1.0 b1.0 b1.0 – – b1.0 b1.0 1.6 – – 9.0 7.7 b1.0 – – 2.4 b1.0 b1.0 – – b1.0 b1.0 2.1 – – b4.0–5.1 5.8

b1.0 – – b1–1.3 b1–1.0 1.8 – – 7.9 5.3 b1.0 – – b1.0 b1.0–1.8 b1.0 – – b1.0 ≤1.0 b1.0 – – b1.0–1.7 b1.0 1.7 – – b4.0–4.9 b1

2.3 – – b5.0 b5.0 2.4 – – 8.1 5.2 2.4 – – ≤5.0 b5.0 2.4 – – 8.0 8.1 2.3 – – 8.6 7.6 3.1 – – 7.9 8.1

during storage. The volatile metabolites 2- and 3-methylbutanol, 2- and 3-methylbutanal, diacetyl, acetoin and 2-methylpropanol were associated with negative changes in the sensory profile of the saveloy samples. This result confirms the VOCs identified as chemical markers for the shelf-life of industrially sliced saveloy by Holm et al. (2012). The results are supported by other studies linking one or several of these VOCs with a decrease in the quality of cooked and sliced meat products (Dainty & Hibbard, 1983; Ercolini et al., 2011; Leroy et al., 2009; Pham et al., 2008; Stanley et al., 1981). The present results are primarily based on tests of spoilage bacteria in monocultures. However, in processed meat the initial microbial flora consists of several different bacteria. The composition of the initial microbial flora furthermore varies between different manufacturers of meat products (Gounadaki et al., 2008; Samelis et al., 2000). The competition and interactions between the different bacteria in the spoilage flora therefore play a major role in determining the rate and extent of spoilage observed in a given processed meat (Russo, Ercolini, Mauriello, & Villani, 2006). An example of this was

Table 4 Characteristics of the PLSR-models of the sensory descriptors. The table includes the RMSECV, R2 of the predicted vs. measured plot, the percentage of Y variation explained by the model and the ten most important VOCs based on their regression coefficient to the sensory descriptor. Y-variable

RMSECV

R2

PCs

% Y-explained

Top ten VOCs

Saveloy odor

1.02

0.69

2

80.1

Spicy odor

0.68

0.68

2

77.3

Butter-like odor

0.93

0.81

3

87.3

Sweet odor

0.88

0.63

2

74.0

Sour odor

1.24

0.63

2

73.1

Old odor

1.59

0.57

2

70.8

Ac3, K3, K5, Ac2, A2, A6, S1, Ca, Ac7, Ac1 Ac3, K3, K5, Ac2, A2, S1, Ca, A6, S2, Ac1 A2, Ac3, K5, K3, S1, S2, Ca, Ac8, Ac7, A6 A2, Ac3, K3, K5, Ac2, Ca, Ac6, A6, S2, S1 Ac3, A2, K5, K3, Ac2, Ca, T14, S1, T4, T11 Ac3, Ac2, K3, K5, S1, A2, S2, Ca, A6, T14

694

E.S. Holm et al. / Meat Science 93 (2013) 687–695

seen in the saveloy samples inoculated with a mixture of all four potential spoilage bacteria. In this mixture, L. carnosum became the dominating organism and, as noticed previously, this bacterium only resulted in a low level of spoilage of saveloy. This result highlights the importance of including knowledge of the composition and interactions of the spoilage flora when evaluating the shelf-life of a given processed meat product. Though some research has been made in this area, further studies investigating the development of the spoilage flora in cooked and sliced meat products from different manufacturers would be relevant (Holm et al., 2012; Samelis et al., 2000). This could provide further knowledge about the spoilage risk associated with different bacteria in different types of meat products which would be valuable for the meat industry. The present results show that the formation of VOCs during spoilage is highly dependent on the predominant spoilage bacteria and the storage conditions. As the composition of the spoilage flora varies between different manufacturers, the PLSR-models used for spoilage predictions in practice should be able to account for a large degree of variation in the formation of VOCs during storage. This was also the case for the PLSR-models calculated in this study. These models incorporated the spoilage patterns of four different bacteria. As seen in Table 4 these PLSR-models had relatively high prediction errors (RMSECV). Increasing the number of samples in the experiment would reduce the prediction error of the model. Alternatively PLSR-models could be fitted to account for the variation in saveloy samples from a specific manufacturer. This would reduce the variation included in the model and lower the prediction error. In this study storage in MAP at 5 °C represents the storage conditions recommended by the manufacturers. On the contrary, the use of CSS represents actual storage of cooked and sliced meat products in the households, and it is based on these conditions the product will be evaluated by the consumer. The storage conditions chosen for the development of a shelf-life model are therefore important in relation to how the model can be used in practice. As seen in Figs. 1 and 2 the chemical markers suggested in this study are mainly relevant at storage conditions including CSS. The possibility of predicting sensory spoilage from measurements of selected VOCs would be of great value as a quality assurance tool in the meat industry. This tool could be used to estimate the shelf-life period of a given batch of saveloy shortly after production. However, this would require that it is possible to measure the microbial production of VOCs in the meat products in the early stages after production. This was not the case in this study. According to the statistical analysis of the results in this study no significant increase in the potential chemical markers was observed between day 1 and week 1 (data not shown). This was also indicated in Figs. 1 and 2. It remains uncertain whether an increase in the sensitivity of the applied GC–MS method or the use of alternative methods for VOC measurement, such as PTR-MS, would enable early detection of spoilage in cooked and sliced meat products (Lindinger, Hansel, & Jordan, 1998). However, in the early stages after inoculation the microbial population is expected to be in the lag phase (Madigan, Martinko, & Parker, 2000). It is therefore doubtful whether a measurable amount of VOCs will be produced at this point. 4. Conclusion In this study the quality changes in sliced saveloy samples inoculated with potential spoilage bacteria were investigated using measurements of the VOC composition, the sensory profile and the microbial composition. Inoculation with B. thermosphacta, C. luteola, C.maltaromaticum was found to cause severe spoilage of the sliced saveloy samples. The VOCs 2- and 3-methylbutanol, 2- and 3-methylbutanal, diacetyl, acetoin and 2-methylpropanol were found to be related to the increase in the sensory descriptors sour odor, old odor and butter-like odor in PLSR-models. The level of the highlighted VOCs found in the saveloy samples subjected to inoculation with spoilage bacteria depended on

the storage conditions. Samples stored in MAP throughout the experiment generally contained a lower level of these metabolites and were therefore less spoiled than the samples stored in CSS during the final week of the experiment. However, particularly C. maltaromaticum was also able to spoil the product when stored in MAP. The main volatile metabolites produced by L. carnosum were straight chain alcohols such as 1-hexanol. Inoculation with L. carnosum only caused limited spoilage in the saveloy. However, this organism was found to be the dominating flora in saveloy samples inoculated with a mixture of the four tested bacteria. Overall, the results of this study has given an insight to the spoilage patterns of four potential spoilage bacteria which is relevant in relation to future attempts to control and predict the shelf-life of cooked and sliced meat products. Acknowledgements This work was partly sponsored by the Danish Pig Levy-Fund and the Danish Center for Advanced Food Studies (LMC). Lizzie Larsen, Helen Ludvigsen, Ann-Britt Frøstrup, Jens P. Teilman and Jonna Andersen are acknowledged for their technical assistance to this study. Camilla Beierholm, Tomas Jacobsen and Jakob Søltoft-Jensen are acknowledged for their help regarding the planning and execution of the present study. References Acree, T., & Arn, H. (2004). Ref type: Internet communication. http://www.flavornet.org/ Bjorkroth, K. J., Vandamme, P., & Korkeala, H. J. (1998). Identification and characterization of Leuconostoc carnosum, associated with production and spoilage of vacuumpackaged, sliced, cooked ham. Applied and Environmental Microbiology, 64, 3313–3319. Blom-Hansen, J. (2004). Vurdering af temperatur forløb fra fabrik til forbrug (internal report). Borch, E., KantMuermans, M. L., & Blixt, Y. (1996). Bacterial spoilage of meat and cured meat products. International Journal of Food Microbiology, 33, 103–120. Dainty, R. H., Edwards, R. A., & Hibbard, C. M. (1984). Volatile compounds associated with the aerobic growth of some pseudomonas species on beef. Journal of Applied Bacteriology, 57, 75–81. Dainty, R. H., & Hibbard, C. M. (1983). Precursors of the major end products of aerobic metabolism of Brochothrix thermosphacta. Journal of Applied Bacteriology, 55, 127–133. Dainty, R. H., & Mackey, B. M. (1992). The relationship between the phenotypic properties of bacteria from chill-stored meat and spoilage processes. Journal of Applied Bacteriology, 73, S103–S114. Ercolini, D., Ferrocino, I., Nasi, A., Ndagijimana, M., Vernocchi, P., La Storia, A., et al. (2011). Monitoring of microbial metabolites and bacterial diversity in beef stored under different packaging conditions. Applied and Environmental Microbiology, 77, 7372–7381. Gounadaki, A. S., Skandamis, P. N., Drosinos, E. H., & Nychas, G. J. E. (2008). Microbial ecology of food contact surfaces and products of small-scale facilities producing traditional sausages. Food Microbiology, 25, 313–323. Holm, E. S., Schafer, A., Skov, T., Koch, A. G., & Petersen, M. A. (2012). Identification of chemical markers for the sensory shelf-life of saveloy. Meat Science, 90, 314–322. Korkeala, H. J., & Bjorkroth, K. J. (1997). Microbiological spoilage and contamination of vacuum-packaged cooked sausages. Journal of Food Protection, 60, 724–731. Larrouture-Thiveyrat, C., Pepin, M., Leroy-Setrin, S., & Montel, M. C. (2003). Effect of Carnobacterium piscicola on aroma formation in sausage mince. Meat Science, 63, 423–426. Laursen, B. G., Byrne, D. V., Kirkegaard, J. B., & Leisner, J. J. (2009). Lactic acid bacteria associated with a heat-processed pork product and sources of variation affecting chemical indices of spoilage and sensory characteristics. Journal of Applied Microbiology, 106, 543–553. Laursen, B. G., Leisner, J. J., & Dalgaard, P. (2006). Carnobacterium species: Effect of metabolic activity and interaction with Brochothrix thermosphacta on sensory characteristics of modified atmosphere packed shrimp. Journal of Agricultural and Food Chemistry, 54, 3604–3611. Leroy, F., Vasilopoulos, C., Van Hemelryck, S., Falony, G., & De Vuyst, L. (2009). Volatile analysis of spoiled, artisan-type, modified-atmosphere-packaged cooked ham stored under different temperatures. Food Microbiology, 26, 94–102. Lindinger, W., Hansel, A., & Jordan, A. (1998). Proton-transfer-reaction mass spectrometry (PTR-MS): On-line monitoring of volatile organic compounds at pptv levels. Chemical Society Reviews, 27, 347–354. Madigan, M. T., Martinko, J. M., & Parker, J. (2000). Microbial growth. Brock biology of microorganisms (pp. 135–162). (9th ed.). Upper Saddle River: Prentice Hall. Mataragas, M., Skandamis, P., Nychas, G. J. E., & Drosinos, E. H. (2007). Modeling and predicting spoilage of cooked, cured meat products by multivariate analysis. Meat Science, 77, 348–356. Mottram, D. S. (1998). Flavour formation in meat and meat products: A review. Food Chemistry, 62, 415–424.

E.S. Holm et al. / Meat Science 93 (2013) 687–695 Nychas, G. J. E., Skandamis, P. N., Tassou, C. C., & Koutsoumanis, K. P. (2008). Meat spoilage during distribution. Meat Science, 78, 77–89. Pham, A. J., Schilling, M. W., Mikel, W. B., Williams, J. B., Martin, J. M., & Coggins, P. C. (2008). Relationships between sensory descriptors, consumer acceptability and volatile flavor compounds of American dry-cured ham. Meat Science, 80, 728–737. Russo, F., Ercolini, D., Mauriello, G., & Villani, F. (2006). Behaviour of Brochothrix thermosphacta in presence of other meat spoilage microbial groups. Food Microbiology, 23, 797–802.

695

Samelis, J., Kakouri, A., & Rementzis, J. (2000). Selective effect of the product type and the packaging conditions on the species of lactic acid bacteria dominating the spoilage microbial association of cooked meats at 4 degrees C. Food Microbiology, 17, 329–340. Stanley, G., Shaw, K. J., & Egan, A. F. (1981). Volatile compounds associated with spoilage of vacuum-packaged sliced luncheon meat by Brochothrix thermosphacta. Applied and Environmental Microbiology, 41, 816–818.