Correlation between volatile profiles of Italian fermented sausages and their size and starter culture

Food Chemistry 192 (2016) 736–744

Contents lists available at ScienceDirect

Food Chemistry journal homepage: www.elsevier.com/locate/foodchem

Correlation between volatile profiles of Italian fermented sausages and their size and starter culture Chiara Montanari a, Eleonora Bargossi b, Aldo Gardini c, Rosalba Lanciotti a,b, Rudy Magnani d, Fausto Gardini a,b, Giulia Tabanelli a,⇑ a

Centro Interdipartimentale di Ricerca Industriale Agroalimentare, Università degli Studi di Bologna, Sede di Cesena, Piazza Goidanich 60, 47521 Cesena, FC, Italy Dipartimento di Scienze e Tecnologie Agro-alimentari, Università degli Studi di Bologna, Sede di Cesena, Piazza Goidanich 60, 47521 Cesena, FC, Italy Dipartimento di Scienze statistiche ‘‘Paolo Fortunati’’, via delle Belle Arti 41, 40126 Bologna, Italy d C.l.a.i. Soc. Coop., Via Gambellara 62, 40026 Imola, BO, Italy b c

a r t i c l e

i n f o

Article history: Received 4 April 2015 Received in revised form 22 June 2015 Accepted 14 July 2015 Available online 16 July 2015 Keywords: Fermented meat Sausage size Aroma profile Linear discriminant analysis (LDA) Starter cultures

a b s t r a c t The aroma profiles of 10 traditional Italian fermented sausages were evaluated. The volatile organic compounds (VOCs) obtained by solid-phase microextraction and gas chromatograph–mass spectrometry were analysed using principal component analysis (PCA) and linear discriminant analysis (LDA). PCA allowed an acceptable separation but some sausage typologies were not well separated. On the other hand, the supervised approach of LDA allowed a clear grouping of the samples in relation to sausage size and starter culture. In spite of the extreme variability of the volatile profiles of the sausage typologies, this work showed the influence of diameter on VOC profile. The differences observed can be related to the effects that some fundamental physicochemical characteristics (such as water loss kinetics and oxygen availability) have on the results of ripening processes. Differences in VOC profiles were also observed due to the lactic acid bacteria used as starter cultures, with differences mainly attributable to compounds deriving from pyruvate metabolism. Ó 2015 Elsevier Ltd. All rights reserved.

1. Introduction European countries are the major producers and consumers of cured meats, even if these products are widespread around the world (Lachowicz, Zochowska-Kujawska, & Sobczak, 2012). Among cured meats, fermented sausages, which originated in the Mediterranean area during Roman times, have great importance (Leroy, Geyzen, Janssens, De Vuyst, & Scholliers, 2013). They are a heterogeneous group of products, with great differences in relation to raw material (type of lean meat and fat), ingredients (salt concentration, nitrate/nitrite, spices and herbs, additives), size (diameter, weight, type of casing) and ripening conditions (temperature, relative humidity, use of moulds and/or smoke) (Leroy et al., 2013; Toldrá, 2006). In Italy the impact on the food industry of fermented sausages manufacture in 2013 was very important; the total production of cured meats was 1,180,000 tonnes, with an economic value of about €7900 million and, within this production, 109,000 tonnes (€925 million) were represented by fermented sausages (ASSICA, 2013). ⇑ Corresponding author. E-mail address: [email protected] (G. Tabanelli). http://dx.doi.org/10.1016/j.foodchem.2015.07.062 0308-8146/Ó 2015 Elsevier Ltd. All rights reserved.

Because of historical, cultural and traditional practices (often different from one city to another), Italian traditional cured meats are characterised by a high variability, as evidenced by the European Protected Designation of Origin (PDO) and Protected Geographical Indication (PGI) recognitions (European Commission, 2014). In spite of their roots in tradition, the processes for Italian fermented sausages production have been modified through the centuries, by adapting to improved technological standards and modifications in the perception of food (Leroy et al., 2013). These changes have undergone an exponential acceleration in the last decades due to the risks associated with novel pathogens or the market requests for products with lower salt/fat content or functional properties (Toldrá & Reig, 2011). Nowadays, the use of starter cultures and rigorous conditions of temperature and relative humidity during ripening are commonly applied by Italian producers. This could result in a standardisation of the traditional features of different products, even if the wild ripening microbiota maintains a crucial role on the final characteristics and on flavour formation (Ravyts, De Vuyst, & Leroy, 2012). Volatile organic compounds (VOCs) are formed essentially through the metabolism of lipids and proteins, as well as being end-products of the lactic fermentation, mainly through oxidative transformations (Leroy et al., 2013; Ordóñez, Hierro, Bruna, & de la

737

C. Montanari et al. / Food Chemistry 192 (2016) 736–744

Hoz, 1999). The action of enzymes (endogenous or microbial) during maturation is responsible for the typical sausage aroma profile. The metabolism of lactic acid (derived from sugar fermentation) leads to the production of molecules such as acetic, propionic and butyric acids, acetaldehyde, diacetyl and acetoin. The products with major impact originating from the metabolism of amino acids liberated by proteases and peptidases are branched aldehydes, acids and alcohols. The free fatty acids deriving from lipids can be the substrate for the production of carbonyl compounds (aldehydes and methyl ketones), and also hydrocarbons and alcohols (Carballo, 2012; Demeyer, 2004; Ordóñez et al., 1999). These processes determine the characteristic VOC profiles of fermented sausages and depend on complex interactions among several factors, including raw material properties, mincing degree of meat batter, sausage dimension, type of casing, ripening process length, relative humidity and temperature, use of starter cultures, and characteristics of the wild microbiota (Incze, 2010; Janssens, Myter, DeVuyst, & Leroy, 2012; Leroy et al., 2013; Ravyts et al., 2010; Toldrá, 2006). In the last decades the fermented meat industry was subjected to a high degree of innovation, determined by the challenge of new safety concerns (including emerging pathogens and salt reduction) and by an improvement of process efficiency (Leroy et al., 2013). In spite of the risk of product standardisation due to these innovative practices, the industry focuses its attention on the distinctiveness of the aroma profiles of ‘‘typical’’ products to promote their recognisability and uniqueness as a fundamental key to develop market strategies. The objective of this work was to study the aroma profiles of 10 different industrially produced Italian fermented sausages, in order to highlight the presence of volatile molecules responsible for flavour formation through a solid-phase microextraction/gas chromatography–mass spectrometry (SPME/GC–MS) analysis. The same products were analysed in a previous study (Tabanelli et al., 2014) in relation to some physicochemical and microbiological attributes, as well as in relation to their biogenic amine content. In this work, some statistical exploratory tools – principal component analysis (PCA) and linear discriminant analysis (LDA) – have been used to correlate the differences observed in VOCs profiles to the size of the tested sausages and to the type of starter culture used for their manufacture. The main aim of this work was to better understand the effects on the aroma profile of important physicochemical features (such as water loss and oxygen availability), highly influenced by sausage diameter. 2. Materials and methods 2.1. Fermented sausages Ten types of fermented sausages were considered in this trial. All of them were produced by CLAI, a company located in Imola (Italy). The samples (six salamis for each typology) were taken directly from the factory at the end of ripening in May 2014. The six samples for each sausage typology were taken from two different batches produced on different days and for each batch three samples were taken from a different position in the ripening chamber. The sausage typologies were grouped in relation to their diameter. The small sausages (diameter less than 50 mm) were Cacciatore (CCT), Salsiccia passita (PST) and Salame Aquilano (AQL); the medium sausages (diameter between 50 and 100 mm) were Felino type salami (FLN), Salame Romagnolo (RMG) and Salame Napoli (NPL); the large sausages (diameter greater than 100 mm) were: Milano type MLN), Lombardo type (LMB), low fat sausage (LFS) and Sicilian type sausage (SCL). Two types of starter cultures were used: starter culture A contained Lactobacillus sakei and a mixture of Staphylococcus xylosus

strains, while starter culture B contained Pediococcus pentosaceus and the same mixture of S. xylosus strains. The choice of the starter lactic acid bacteria depended on previous experiences and trials of the sausage producer. The external mould growth was induced in all the sausages by inoculating with a selected Penicillium nalgiovense strain. All starter cultures were provided by Chr. Hansen Italia (Parma, Italy). All the sausages contained nitrates and nitrites, according to European legislation (DIRECTIVE 2006/52/ EC, 2006) and spices (black pepper, garlic). The main characteristics of these sausages are summarised in Table 1. 2.2. Aroma profile analysis Volatile organic compounds of the ten different typologies of fermented sausages were analysed using solid phase microextraction coupled with gas chromatography–mass spectrometry (SPME/GC–MS), as previously reported by Tabanelli, Montanari, Grazia, Lanciotti, and Gardini (2013). In particular, samples (3 g) were placed in 10-mL sterilised vials, sealed by PTFE/silicon septa and heated for 10 min at 45 °C. After that a fused silica SPME fibre covered with 75 lm Carboxen/polydimethylsiloxane (CAR/PDMS StableFlex) (Supelco, Steinheim, Germany) was introduced into the headspace for 40 min. Adsorbed molecules were desorbed in the gas-chromatograph for 10 min. For peak detection, an Agilent Hewlett–Packard 6890 GC gas-chromatograph equipped with a 5970 MSD MS detector (Hewlett–Packard, Geneva, Switzerland) and a Varian (50 m  320 lm  1.2 lm) fused silica capillary column were used. The conditions were as follows: injection temperature, 250 °C; detector temperature, 250 °C; carrier gas (He) flow rate, 1 mL/min. It is the ion source temperature. The MS acquisition parameters were: Low Mass: 30; High Mass: 300 and Threshold : 100; 5.1 scan/sec. The oven temperature was programmed as follows: 50 °C for 1 min, from 50 °C to 65 °C at 4.5 °C/min, from 65 °C to 230 °C at 10 °C/min, then holding for 25 min. Volatile peak identification was carried out by computer matching of mass spectral data with compounds contained in the NIST 98 and Wiley 6 mass spectral databases. For each type of fermented sausage, the volatile profile composition was expressed as relative percentage of each single peak area with respect to the total peak area. Data reported are the means of six different sausages. 2.2. Statistical analysis Data for each sausage typology are the mean of six different samples. ANOVA, principal component analysis (PCA) and linear discriminant analysis (LDA) were carried out in the statistical environment R (R Development Core Team, Vienna, Austria). ANOVA was carried out by fitting the classical one-way model:

Y ij ¼ l þ si þ eij ;

i ¼ 1; 2; 3; j ¼ 1; . . . J i

eij  Nð0; r2 Þ where l is the overall mean compound concentrations and si denotes the deviation of the response from l when the i-th group is considered. The aim of this step was simply to point out the differences in the compound concentrations with respect to the three sausage diameter sizes. Therefore the following hypothesis system was tested:



H0 : s1 ¼ s2 ¼ s3 H1 : At least two of the si ’s are different

If the null hypothesis was rejected, post hoc tests were performed to distinguish the different si’s. To correct the p-values to allow pairwise comparisons, Tukey’s procedure was applied.

738

C. Montanari et al. / Food Chemistry 192 (2016) 736–744

3. Results and discussion

a

b

B 15 0.898 5.36 B 8 0.934 5.21 Starter culturea Ripening time (week) awb pHb

Starter culture (A) contains Lactobacillus sakei and a mixture of Staphylococcus xylosus strains; starter culture (B) contains Pediococcus pentosaceus and a mixture of Staphylococcus xylosus strains. From Tabanelli et al. (2014).

A 4 0.871 6.10 B 3 0.906 5.88 A 3 0.918 6.71 A 5 0.921 5.44 A 6 0.922 5.38 B 4 0.922 5.19

Ø 65 1100 18 7 Ø 75 1400 20 10 250  100 3200 8.5 4.5

Ø 120 3600 24 30 with lard cubes B 6 0.922 5.18 Ø 105 3850 22 3 Ø 105 4000 26 3 Section

Dimension (mm) Weight (g) Fat in the meat batter (%) Mincing size (mm)

Napoli (NPL) Sicilianella (SCL) Milano(MLN)

Characteristics of the tested sausages

Table 1 Main characteristics of sausage samples grouped in relation to their size.

A 4 0.959 4.86

60  30 400 20 4 Ø 65 1100 18 3 with lard cubes

Ø 37 550 20 6

Passita (PST)

Medium

Lombardo(LMB)

Low fat sausage (LFS) Large

Felino type (FLN)

Romagnolo (RMG)

Small

Aquilano (AQL)

Ø 47 175 23.5 5

Cacciatore (CCT)

3.1. Volatile molecule profiles Six different samples for each sausage typology were analysed with a SPME/GC–MS to evaluate the VOCs composition at the end of ripening. In Fig. 1 an example of the gas-chromatographic profiles obtained is shown. The percentage of each volatile compound has been calculated after the exclusion of molecules deriving from pepper, with the aim to focus the attention on the compounds produced during sausage ripening through the action of endogenous and microbial enzymes. Terpenes and terpenoids, such as 3-carene, b-pinene, thujene, a- and b-phellandrene, b-myrcene, p-cymene, caryophyllene and limonene detected in the sausages and deriving from this spice were not considered (Menon & Padmakumari, 2005). In fact, the use of whole or coarsely ground pepper can cause a non-uniform distribution of this spice in the samples analysed and, consequently, to a non-uniform distribution of these volatile molecules. With this perspective, the compounds derived from garlic, such as methyl allyl sulfide, diallyl sulfide and dimethyldisulfide (Schmidt & Berger, 1998) have also not been taken into consideration. The results obtained for each sausage typology VOC profile are reported in Table 2, and are expressed as the mean of 6 repetitions, with relative standard deviations. The compounds are grouped into chemical families: ketones, aldehydes, alcohols, esters and acids. The total peak area is detected as an indirect measure of the total amount of VOCs extracted. The unidentified compounds are not included; in any case, they accounted for less than 3% of total peak area. Ketones represented the most important group of compounds, especially in the sausages with a small diameter, where they represented from 64.43% (AQL) to 81.42% (PST) of the total peak area. In the other sausages their relative percentage was never lower than 37%, with the exception of MLN (15.29%). The three main ketones detected (acetone, 2-butanone and 3-hydroxy-2butanone) originate mainly from pyruvate metabolism (Carballo, 2012; Ordóñez et al., 1999). Aldehydes (represented by hexanal and benzeneacetaldehyde) were present in low amounts, and their highest percentage was observed in medium-sized sausages FLN (5.14%) and NPL (7.71%), and in the large-diameter sausage MLN (8.32%). This result was surprising if compared with the aroma profiles of analogous products (Bianchi et al., 2007; Tabanelli et al., 2012; Tabanelli et al., 2013), in which aldehydes were the most important chemical group and hexanal represented 30% and more of the total peak area. The alcohols detected were mainly ethanol, 3-methyl-2-butanol and 1-propanol, and their total amount was extremely variable. Their percentage was generally lower in small-diameter sausages (from 10.23% to 16.40%) and higher in large-diameter ones (from 21.22% to 44.30%). The alcohol 1-hexanol, deriving from hexanal reduction, was present in relatively low concentrations. Esters had their maximum percentage in large-diameter sausages and were represented essentially by ethyl acetate. Acids percentage, and in the first instance acetic acid, showed a great variation among the sausages, reaching the maximum level in MLN (27.91%) and NPL (24.90%). Small size sausages were generally characterised by a low concentration of organic acids, especially PST (2.93%). The formation of flavour compounds in fermented sausages is the result of complex reactions, which are mainly oxidative (Ordóñez et al., 1999). Diameter is an important factor affecting flavour profile, since it influences water loss kinetics and oxygen availability inside the sausage (Demeyer, 2004; Ordóñez et al., 1999; Toldrá, 2006). For this reason, the concentration of the most important components detected by SPME/GC–MS was analysed in

C. Montanari et al. / Food Chemistry 192 (2016) 736–744

relation to sausage size (Table 1). Edwards, Ordóñez, Dainty, Hierro, and de la Hoz (1999) showed a correlation between the calibre of the dry-fermented sausages and the presence of carbonyl compounds and found that in ‘‘fuet’’ sausage (diameter 20– 40 mm) the levels of aldehydes and ketones were higher than in ‘ ‘salchichón’’ (diameter 40–60 mm), due to the greater diffusion of oxygen inside ‘‘fuet’’. Successively, the aroma profile data were grouped in relation to sausage size and then analysed with one-way ANOVA. All the volatile compounds showed significantly different percentages (p < 0.01) in relation to the diameter, with the exception of 2-butanone (Table 2). In addition, the data of the aroma profile components were analysed with ANOVA also in relation to the starter cultures used and the amounts of 13 out of the 22 compounds detected were significantly (p < 0.05) dependent on the starter culture used (data not shown). In particular, significant differences were observed for 2,3-butanedione, 2-pentanone, 3-octanone, 3-hydroxy-2-butanone, 2-butanol, 1-propanol, 3-methyl, 2-buten-1-ol, 2,3-butanediol, ethyl acetate, propyl acetate, ethyl butanoate, acetic acid and propanoic acid. The distributions of the main compounds, in relation to sausage size, are shown in Fig. 2 as box and whisker plots, in which the median and the 25–75 quartile interval are reported. Besides quartiles, the bars represent the value of the minimum/maximum observations falling in an interval described by the interquartile range multiplied by +1.5 or 1.5. Outside this range, the eventual observations can be considered as outliers or observations far from the distribution median and they are represented with a point. As reported in Fig. 2, acetone was present in significantly lower amounts in large-diameter sausages (mean 5.28% vs. 21.21% and 23.51% in medium and small size sausages, respectively). This ketone derives from pyruvate metabolism (Carballo, 2012) or from branched amino acid catabolism (Stahnke, 1999). Pyruvate can be metabolised through several pathways by lactic acid bacteria, which, in the absence or scarcity of fermentable sugars, can also reoxidise lactate to pyruvate in the presence of O2 (von Wright & Axelsson, 2011). Both 3-hydroxy-2-butanone and 2,3-butandione were detected in significantly higher proportion in small size sausages. These two ketones derive from pyruvate metabolism and confer cheesy,

739

buttery and snack notes to sausages (Marco, Navarro, & Flores, 2008; Olivares, Navarro, & Flores, 2009a). The decrease of such compounds has been associated with the flavour deterioration of Spanish fermented sausages (Lorenzo, Bedia, & Bañón, 2013). Some observations far from the distribution median were highlighted for these ketones; in particular, two out six samples of PST showed a 2,3-butanedione percentage significantly lower if compared with the small size sausages. Among large size sausages, all the LFS samples resulted as observations far from the distribution median, because of their higher presence of 2,3-butanedione. The same latter sausage typology was also responsible for the points far from the distribution median observed for 3-hydroxy-2-butanone. Noteworthy, the proportion of 2,3-butanediol (not shown in Fig. 2), deriving from reduction of 3-hydroxy-2-butanone, was high only in small sausages CCT and PST. Among alcohols, 3-methyl-2-butanol accumulation is related to the metabolism of the branched amino acid leucine (Carballo, 2012; Gutsche, Tran, & Vogel, 2012; Ordóñez et al., 1999). Among the sausages, the lowest percentage of this alcohol was found in the small size sausages (mean 1.64%), and the highest in the medium size products (mean 3.46%), even if these samples showed a high variability. By contrast, 3-methylbutanoic acid, deriving from the same amino acid degradation (via decarboxylative oxidation), was present only in small-diameter sausages. The alcohol 1-propanol was found at higher levels, though with a high variability, in medium and large size sausages (mean 5.44% and 6.77%, respectively). In particular, NPL showed a 1-propanol mean percentage of 13.16%, while the samples of SCL (mean content 19.50%) resulted as points far from the distribution median among large size sausages. Also 2-butanol (deriving from the reduction of 2-butanone) was present in higher percentages in medium and large-diameter sausages. The higher ethanol amount was found in large sausages, among which LFS reached the maximum percentage (mean value 24.70%), reported as observations far from the distribution media. Ethanol formation can be the result of several pathways, including pyruvate metabolism when sugars have been consumed (von Wright & Axelsson, 2011) and the metabolism of some amino acids, such as threonine (Carballo, 2012).

Fig. 1. Example of a gas-chromatogram of VOCs profile of the LFS sample. In the chromatogram are present also the compounds deriving from spices which were not considered in the statistical analyses.

740

C. Montanari et al. / Food Chemistry 192 (2016) 736–744

Table 2 Volatile compounds expressed as % of total peak area in sausage samples at the end of ripening. For each sausage typology, the data reported are the means of 6 repetitions. The table reports also the mean value of the volatile compounds grouped in relation to sausage size. For these data also standard deviation is reported within brackets together with the results of ANOVA. When ANOVA was significant (P 6 0.05) lower-case letters are reported. Values with the same letter are not statistically different within each line (P > 0.05) according to the post hoc comparisons of the ANOVA. Volatile organic compound

Sausage typology in relation to the sizea Small

Medium

Large

AQL

CCT

PST

Mean

FLN

RMG

NPL

Mean

LMB

SCL

LFS

MLN

Mean

Acetone 2-Butanone 2,3-Butanedione 2-Pentanone 3-Octanone 3-Hydroxy-2-butanone

32.75 3.91 3.30 1.93 0.00 22.54

27.12 12.06 3.89 3.11 0.00 33.04

10.68 39.64 3.05 2.12 3.41 22.52

23.52a (±9.74) 18.54 (±15.92) 3.41a (±0.69) 2.39a (±0.63) 1.14a (±1.70) 26.03a (±5.81)

24.57 22.85 3.62 0.00 0.00 8.95

28.42 16.24 2.10 3.96 0.00 14.28

10.63 32.08 0.00b 0.49 0.00 0.00

21.21a (±8.25) 23.72 (±7.38) 1.91b (±1.66) 1.48a,b (±1.87) 0.00b (±0.00) 7.74b (±6.11)

3.40 51.99 0.44 0.95 0.00 0.00

10.81 24.84 0.00 0.67 0.00 0.64

5.41 0.77 3.40 2.59 0.00 29.11

1.50 13.79 0.00 0.00 0.00 0.00

5.28b (±3.70) 22.85 (±19.43) 0.96b (±1.49) 1.05b (±1.01) 0.00b (±0.00) 7.44b (±12.79)

Total ketones Hexanal Benzenacetaldehyde

64.43 0.28 0.00

79.23 0.85 0.00

81.42 0.40 2.29

75.03a (±8.34) 0.51a (±0.28) 0.76a (±1.32)

59.99 5.14 0.00

64.99 1.03 1.85

43.20 0.10 7.51

56.06b (±10.18) 2.09a (±2.35) 3.12b (±3.31)

56.78 2.10 1.14

36.97 1.12 0.00

41.28 0.89 0.00

15.29 7.47 0.85

37.58c (±15.45) 2.90b (±2.93) 0.50a (±0.81)

Total aldehydes 3-Methyl-2-Butanol Ethanol 2-Butanol 1-Propanol 1-Pentanol 3-Methyl-3-buten-1-ol 3-Methyl-2-buten-1-ol 1-Hexanol 2,3-Butanediol

0.28 1.05 10.93 0.44 0.57 1.34 0.70 0.17 1.21 0.00

0.85 0.76 1.02 0.00 0.64 0.00 0.00 0.00 0.41 7.40

2.76 3.10 0.77 1.00 0.00 0.00 0.00 0.00 0.53 6.85

5.14 4.74 4.31 2.97 1.67 3.11 1.54 1.23 4.62 0.00

2.88 4.81 4.36 0.90 1.48 1.54 0.00 0.96 2.17 0.76

7.71 0.83 1.21 3.35 13.16 0.00 0.00 0.00 0.30 0.00

5.24b (±2.18) 3.46b (±2.04) 3.29a (±1.70) 2.41a (±1.38) 5.44b (±5.72) 1.55b (±1.43) 0.51a,b (±0.81) 0.73b (±0.57) 2.36b (±2.19) 0.25b (±0.74)

3.31 3.26 4.30 7.29 2.36 1.62 0.00 0.00 2.39 0.00

1.12 2.73 8.42 4.59 19.50 0.00 1.34 0.72 1.65 0.00

0.89 0.44 24.70 0.00 0.00 1.26 3.01 1.55 1.39 0.85

8.32 1.92 5.37 27.65 5.22 1.19 0.00 0.69 2.25 0.00

3.41b (±3.27) 2.09a (±1.21) 10.70b (±8.51) 9.88b (±10.90) 6.77b (±7.82) 1.02a (±0.82) 1.09b (±1.33) 0.74b (±0.64) 1.92b (±0.70) 0.21b (±0.39)

Total alcohols Ethyl acetate Propyl acetate Butanoic acid ethyl ester

16.40 6.57 0.00 0.00

10.23 0.77 0.00 0.00

12.24 0.00 0.00 0.00

24.21 0.76 0.00 0.00

16.97 1.23 0.00 0.00

18.85 0.82 4.23 0.00

20.01b (±3.50) 0.94a (±0.39) 1.41b (±2.11) 0.00a (±0.00)

21.22 10.62 0.00 0.00

38.94 2.71 2.08 0.00

33.26 9.12 0.00 1.75

44.30 2.31 1.05 0.00

34.43c (±9.04) 6.19b (±3.92) 0.78a (±0.99) 0.44b (±0.78)

6.57 8.53 0.00 0.41 1.70

0.77 6.23 0.00 0.00 0.00

0.00 0.00 0.00 1.35 1.58

(±3.02) (±4.00) (±0.00) (±0.64) (±0.98)

0.76 8.33 0.00 0.00 0.00

1.23 13.31 0.00 0.00 0.00

5.04 23.83 1.07 0.00 0.00

2.34a (±2.08) 15.16b (±6.90) 0.36a (±0.54) 0.00b (±0.00) 0.00b (±0.00)

10.62 6.56 0.00 0.00 0.00

4.79 16.21 1.53 0.00 0.00

10.87 10.99 0.00 0.00 0.00

3.36 24.98 1.32 1.61 0.00

7.41b (±3.68) 14.69b (±7.46) 0.71b (±0.90) 0.40a,b (±0.74) 0.00b (±0.00)

10.65 33.67

6.23 29.52

2.93 44.00

6.60a (±3.83) 37.73(±7.03)

8.33 31.65

13.31 25.50

24.90 39.80

15.51b (±7.39) 32.32 (±8.08)

6.56 23.25

17.74 31.84

10.99 42.53

27.91 44.95

15.80b (±8.72) 35.64 (±9.40)

Total esters Acetic acid Propanoic acid Butanoic acid 3-Methylbutanoic acid Total acids Total peak areac

1.30a 1.64a 4.24a 0.48a 0.40a 0.45a 0.23a 0.06a 0.72a 4.75a

(±1.30) (±1.14) (±4.96) (±0.56) (±0.37) (±0.73) (±0.39) (±0.09) (±0.45) (±3.58)

12.96a (±3.17) 2.45a (±3.02) 0.00a (±0.00) 0.00a (±0.00) 2.45a 4.92a 0.00a 0.59a 1.09a

a Salame Aquilano (AQL), Cacciatore (CCT), Salsiccia passita (PST), Felino type salami (FLN), Salame Romagnolo (RMG), Salame Napoli (NPL), Milano (MLN), Sicilian type sausages (SCL), low fat sausage (LFS), Lombardo (LMB). b Not detected under the adopted conditions. c Total peak area is expressed as Arbitrary Unit/107.

Similar pathways can be responsible for the accumulation of acetic acid, detected in higher percentages in medium and large sausages. Differently from ethanol, the higher proportions of acetic acid were found in medium size sausages (even if without significant differences with respect to large-diameter products). The medium-size sausages were also characterised by a lower percentages of ethyl acetate, reflecting the low amounts of ethanol. The esterase activity in fermented sausages can be due to bacteria (in particular staphylococci) and fungi (yeast and moulds) (Lorenzo et al., 2013; Talon, Chastagnac, Vergnais, Montel, & Berdagué, 1998) and the fruity note of esters are important for flavour; they can also mask rancid odours (Stahnke, 1995). 3.2. Principal component analysis (PCA) To explore the possible relations between the sausage typology and their VOCs, a principal component analysis (PCA) was carried out on the correlation matrix based on the relative percentage of the 24 volatile compounds detected in the samples (Table 2), including the data from as outliers (points far from the distribution media) in the previous investigation. According to the Kaiser’s rule, the scree plot and the explained cumulative variability, the first six factors were chosen as an optimal compromise. They accounted for 84.63% of the total variability (27.65%, 20.46%, 14.26%, 9.69%, 7.53% and 5.04%, respectively). Table 3 reports the Eigenvectors for the

correlation matrix of the first two factors, accounting for 48.11% of the total variability. The resulting grouping of the sausage samples in relation to the typology is shown in Fig. 3. In this figure the 95% confidence ellipses are drawn for each sausage typology. The addition of confidence ellipses can improve the interpretation because they show the degree of separation of the groups (Hammer & Harper, 2006) and outliers are easily identified. The degree of linear correlation between the variables is indicated by the eccentricity, and highly correlated variables give narrow ellipses. The small size sausages (CCT, PST and AQL) were all grouped in the right part of the graph, and particularly in the 4th quadrant. This was due to the weight of compounds such as 2,3-butanedione, 2,3-butanediol and, especially, 3-hydroxy-2-butanone on PC1, and acetic acid and 1-hexanol on PC2. AQL was shifted toward the 1st quadrant because of its higher content of ethyl acetate and ethanol. Two of the medium-size sausages, FLN and RMG, whose manufacture was very similar and differed mainly for the presence in RMG of lard cubes, were grouped mainly in the 1st quadrant and were not well separated, as shown by the partial overlapping of the respective confidence intervals. They were distinguished from small-size sausages by lower amounts of 3-hydroxy-2-butanone. The principal differences between these two types of sausages (FLN and RMG) were the higher content of hexanal, 1-hexanol, 2-butanol and 1-pentanol in FLN and the higher content of 2-pentanone and

741

C. Montanari et al. / Food Chemistry 192 (2016) 736–744

Fig. 2. Box and whisker plots of the main aroma compounds. Median, 25–75 quartiles and bars representing the value of the minimum/maximum observation falling in an interval described by the interquartile range multiplied by +1.5 or 1.5 are reported. Outside of this range, the observations were reported as a point if they are observations far from the distribution media.

acetic acid in RMG. The third medium-size sausage (NPL) was located in the 3rd quadrant because of its higher percentage of 1-propanol and propyl acetate. Concerning the large sausages, the MLN, LMB and SCL were grouped in the left part of the graph; this location was determined by the effects of 2-butanol and 1-propanol and by the lower percentage of acetone (PC1). Between MLN and LMB, differences were induced by the higher percentages of 2-butanone and ethyl acetate in LMB and the higher percentage of 2-butanol and acetic acid in MLN. The sausages characterised by a low fat concentration (LFS) were grouped in the extreme right part of the 1st quadrant, due to their high content of ethyl alcohol, 2,3-butanedione, 3-hydroxy-2-butanone and acetic acid (PC1), and the low presence of 2-butanone (PC2). 3.3. Linear discriminant analysis (LDA) in relation to the size The approach adopted by the PCA allowed an acceptable separation of the different type of sausages in relation to their VOCs. However, some sausage typologies were not well separated and no general indication characterising the aroma profiles in relation to the sausage size could be found through this statistical tool. As a successive step, a supervised classification analytical tool, which takes into account the information known a priori regarding sausage category, has been applied. The aim was to focus the attention on the possibility to highlight differences or similarities determined by the sausage sizes during the complex process, which leads to the product ripening. LDA is a class-modelling technique that classifies each sample in a specific class known a priori. To use LDA it is necessary that the number of samples is less than the number of variables (Vera et al., 2011). In this case, LDA was applied, choosing as classification class the size of sausages according to Table 1. LDA is probably the most frequently used supervised pattern recognition method and the best studied one. Many LDA applications in food research have been reported (Berrueta, Alonso-Salces, & Héberger, 2007; Zielinski et al., 2014). This technique is based on the determination of linear discriminant functions, which maximise the ratio of between-class variance and minimise the ratio of within-class variance (i.e. maximizing the F-ratio). Therefore, LDA reduces the dimensionality of the data; in addition, it maximises the covariance matrix between the groups with respect to the one within the groups. Since in this work LDA was not employed as a

Table 3 Eigenvectors of the first two factors of PCA (accounting for 48.11% of the total variability). Volatile organic compound

Factor 1

Factor 2

Acetone 2-Butanone 2,3-Butanedione 2-Pentanone 3-Octanone 3-Hydroxy-2-butanone Hexanal Benzenacetaldehyde 3-Methyl-2-butanol Ethanol 2-Butanol 1-Propanol 1-Pentanol 3-Methyl-3-buten-1-ol 3-Methyl-2-buten-1-ol 1-Hexanol 2,3-Butanediol Ethyl acetate Propyl acetate Ethyl butanoate Acetic acid Propanoic acid Butanoic acid 3-Methyl butanoic acid

0.2037 0.1549 0.3482 0.2762 0.0919 0.3521 0.1471 0.1885 0.0193 0.1368 0.2498 0.2928 0.0622 0.1259 0.0700 0.0233 0.1806 0.0693 0.2945 0.1576 0.2954 0.3059 0.0706 0.1421

0.0781 0.2031 0.0137 0.0943 0.2839 0.0618 0.1814 0.1873 0.0046 0.3314 0.0891 0.0020 0.2713 0.3314 0.3632 0.2557 0.2998 0.2241 0.0609 0.2644 0.1357 0.0234 0.1322 0.2059

classification technique but only as a dimensional reduction tool, the normality assumption of data was not required. The latent variables obtained in LDA are a linear combination of the original variables. This function is called canonical variates. Being k classes, k  1 canonical variates can be determined if the number of variables is larger than k. In this work, two canonical variates have been obtained (being 3 the classes of sausages considered) and they accounted for 77.60% and 22.40% of the variance, respectively. In Fig. 4 the scores, computed as the linear combination of the variable values and the canonical variates, were plotted. They show a clear discrimination of the observations. The simpler statistical tool used to evaluate the significance of the separation is the statistic Wilks’ lambda. The null hypothesis that it tests consists in the equality of the group means. In this case, a highly significant discrimination (Wilks’ Lambda 0.000158, p < 0.0001) was obtained. This result points out that the LDA provides a clear grouping of the sausages samples

742

C. Montanari et al. / Food Chemistry 192 (2016) 736–744

Fig. 3. PCA factor coordinates for the first two factors of the different fermented sausages. Also the 95% confidence ellipse is drawn.

in relation to their size. In Table 4 the structure matrix, i.e. the Pearson correlations between linear discriminant functions and the variables considered to obtain LDA, is reported. The Pearson correlations represent normalised indicators, which range from 1 to +1; the higher is their absolute value, the higher is the variable contribution to the discrimination. By convention, correlation in excess of 0.33 may be considered eligible while lower ones are not (Tabachnick & Fidell, 2007). The sausages were well discriminated by linear discriminant 1. The large diameter sausages were grouped mainly by the higher content of 2-butanol, acetic acid, ethyl alcohol, ethyl acetate, propanoic acid, 3-methyl-2-buten-1-ol and 1-propanol (all these compounds were characterised by a Pearson correlation >0.4 in absolute value). By contrast, the small diameter sausages differed from the other classes because of the higher content of acetone, 2,3-butanedione, 3-hydroxy-2-butanone, 2,3-butanediol and 3-methylbutanoic acid (characterised by a Pearson correlation >0.5). This reflected a more intense metabolism of lactic acid due to the moulds, whose hyphae can penetrate easily the inner part of these sausages, and to the lactic acid bacteria, which can use more efficiently pyruvate because of the major availability of oxygen determined by the low diameter (Incze, 2010; von Wright & Axelsson, 2011). Also 3-methylbutanoic acid is produced from leucine under oxidative conditions (Smit, Engels, & Smit, 2009). The linear discriminant 2 differentiated the medium-diameter sausages from the other two classes because of the presence of higher proportions of benzeneacetaldehyde, 3-methyl-2-butanol, 1-hexanol and propyl acetate. In Fig. 4, the 95% confidential ellipses are reported. The only outliers are samples of LFS, reflecting the important differences caused by the unusually low fat content of these sausages (8.5% vs. values ranging from 18% and 26% for the other sausages) on the aroma formation during fermentation and ripening. 3.4. Volatile organic compounds in relation to the starter culture The second relevant productive variable regarding the sausages was the type of starter culture used and, in particular, the commercial lactic acid bacteria starter cultures responsible for fermentation (L. sakei or P. pentosaceus according to Table 1). The choice between the two possible lactic acid bacteria species was an internal protocol of the industry, which uses these starter cultures according to its procedures, fundamentally based on empirical observations. As reported above, the ANOVA of the sausages in relation to starter cultures indicated the presence, among VOCs, of several molecules significantly associated with the lactic acid

bacteria used. The results of the PCA presented in Fig. 3 showed an interesting trend regarding the type of starter culture used (A or B). Four (NPL, MLN, LMB and SCL) of the five fermented sausages produced with starter culture B (P. pentosaceus) were located in the left part of the graph, well separated from the others. Only AQL, a small sausage produced with starter culture B, was grouped among the fermented sausages obtained with starter culture A (L. sakei). To better understand the possible role of the starter cultures in the aroma profile of the sausages, an LDA was carried out, grouping the samples according to the starter culture employed. The application of LDA on the aroma compounds of sausages in relation to the type of starter used allows a highly significant discrimination (Wilks’ Lambda 0.013163, p < 0.001). In this case only a canonical variable was obtained (LD1), due to the number of classes (2) in which the observations were subdivided. The clear discrimination indicated by the Wilks’ Lambda was determined by the factors with the higher Pearson correlations reported in Table 4. The significant Pearson coefficients (>0.33 or <0.33) are highlighted in bold in the table. The negative sign of these coefficients characterised the variables that identified the starter A., while the positive sign identified the starter B. 2,3 Butanedione, 3-hydroxy-2-butanone, 2,3 butanediol and 2-pentanone were the most relevant aroma compounds associated with starter culture A, while 1-propanol, propyl acetate, propanoic acid, acetic acid and 2-butanol had a great weight in grouping sausages produced with starter culture B. In the case of starter culture A, 2,3-butanedione, 3-hydroxy-2-butanone and 2,3-butanediol were related to the metabolism of pyruvate, while 2-pentanone was the result of lipid oxidation (Olivares, Navarro, & Flores, 2009b). Propanoic acid and 1-propanol are also derived from pyruvate, as are acetic acid and 2-butanol (Carballo, 2012). The differences imparted to the aroma profile by the two starter cultures seem to mainly rely on the different ability of L. sakei and P. pentosaceus to metabolizse the products of the lactic fermentation (Cocconcelli & Fontana, 2010). An example could be the different capability of the two lactic acid strains to re-oxidise lactate to pyruvate (Goffin et al., 2006; von Wright & Axelsson, 2011). The switch from anaerobic to aerobic metabolism of lactate in the absence of fermentable sugars can have an important impact on the relative prevalence of compounds such as ethanol, acetic acid, 2,3-butanedione, 3-hydroxy-2-butanone and 2,3-butanediol. However, other metabolic routes can be responsible for the production of these compounds. Le Bars and Yvon (2008) demonstrated that Lactococcus lactis was able to produce lactate, pyruvate, acetate, 2,3-butanedione, 3-hydroxy-2-butanone and 2,3-butanediol through the metabolism of aspartate. Given the

C. Montanari et al. / Food Chemistry 192 (2016) 736–744

743

Fig. 4. LDA plot for the classification of different sized sausages using volatile compounds. Also the 95% confidential ellipses are reported.

Table 4 Structure matrix of LDA reporting the Pearson correlation coefficient between each variable and linear discriminant functions in relation to the sausage size and the starter culture. Correlations in excess of 0.33 (in absolute value) are reported in bold and may be considered eligible. Volatile organic compound

Acetone 2-Butanone 2,3-Butanedione 2-Pentanone 3-Octanone 3-Hydroxy-2butanone Hexanal Benzenacetaldehyde 3-Methyl-2-butanol Ethanol 2-Butanol 1-Propanol 1-Pentanol 3-Methyl-3-buten1-ol 3-Methyl-2-buten1-ol 1-Hexanol 2,3-Butanediol Ethyl acetate Propyl acetate Ethyl butanoate Acetic acid Propanoic acid Butanoic acid 3-Methylbutanoic acid

LDA in relation to sausage size

LDA in relation to the starter culture used

the diameter of the sausages on the final VOCs profile. Thus, the product size is an essential parameter, which can drive, together with other important process variables, the biochemical process during ripening by modulating the activities of the endogenous enzymes of meat and the metabolism of the sausage microbiota. This influence is mainly due to the effects of the kinetic of water losses (and, in turn, the aw of the system), product acidification and to the effects that different oxygen availability can play. Differences in VOC profiles were also observed in relation to the lactic acid bacteria (Pediococcus pentosaceous or L. sakei) used as starter cultures and the differences were mainly attributable to the compounds deriving from pyruvate metabolism. The standardisation of the fermented sausage production processes applied by the industry allows, from one side, to show the effects that the main variables (such as diameter and starter cultures) play on the final characteristic of the products, independently from the typology. On the other side, an excessive standardisation could be regarded as a limit in relation to the diversity of the typical products historically developed, especially in the Mediterranean area.

Linear discriminant 1

Linear discriminant 2

Linear discriminant 1

0.7225 0.1051 0.6004 0.4039 0.4213 0.5803

0.2493 0.0962 0.1075 0.1020 0.2727 0.3633

0.3378 0.2294 0.7421 0.5799 0.3289 0.6862

0.3990 0.1123 0.0572 0.4274 0.5102 0.4012 0.1763 0.3573

0.0962 0.5007 0.4577 0.2603 0.1355 0.1550 0.3580 0.0481

0.1187 0.2378 0.2510 0.0730 0.4845 0.5884 0.1638 0.2484

0.4504

0.2815

0.3677

References

0.3087 0.6166 0.4739 0.1918 0.3676 0.4832 0.4271 0.0779 0.5853

0.3524 0.3918 0.3992 0.3476 0.1728 0.3429 0.0212 0.3755 0.3788

0.0924 0.5598 0.3027 0.5328 0.3323 0.5351 0.5683 0.1091 0.0117

ASSICA (Associazione Industriali delle Carni e dei Salumi) (2013). 2013, un anno difficile per i salumi italiani. (http://www.assica.it/UserFiles//file/Cs_ Produzione%20e%20Consumi_2013.pdf). Berrueta, L. A., Alonso-Salces, R. M., & Héberger, K. (2007). Supervised pattern recognition in food analysis. Journal of Chromatography A, 1158, 196–214. Bianchi, F., Cantoni, C., Careri, M., Chiesa, L., Musci, M., & Pinna, A. (2007). Characterization of the aromatic profile for the authentication and differentiation of typical Italian dry-sausages. Talanta, 72, 1552–1563. Carballo, J. (2012). The role of fermentation reactions in the generation of flavor and aroma of foods. In B. M. Mehta, A. Kamal-Eldin, & R. Z. Iwanski (Eds.), Fermentation, Effects on food properties (pp. 51–83). Boca Raton: CRC Press. Cocconcelli, P. S., & Fontana, C. (2010). Starter cultures for meat fermentation. In F. Toldrá (Ed.), Handbook of meat processing (pp. 199–218). Ames: WileyBlackwell. European Commission (2014). Agriculture and rural development, Door (Retrieved 27 October 2014 from http://ec.europa.eu/agriculture/quality/door/list. html?locale=en&recordStart=170&filter.dossierNumber=&filter.comboName=& filterMin.milestone__mask=&filterMin.milestone=&filterMax. milestone__mask=&filterMax.milestone=&filter.country=&filter.category= PDOPGI_CLASS_12&filter.type=&filter.status). Demeyer, D. (2004). Meat Fermentation: Principles and Applications. In Y. H. Hui, L. Meunier-Goddik, Ö. Solvejg Hansen, J. Josephsen, W-K. Nip, P. S. Stanfield, & F. Toldrá (Eds.). Handbook of Food and Beverage Fermentation Technology (chapter 20). Boca Raton: CRC Press.

abundance of this amino acid in meat (Kauffman, 2012), this metabolic pathway could explain, at least in part, the origin of these compounds in the fermented sausages. 4. Conclusions In spite of the extreme variability of the volatile profile of the ten sausage typologies, this work showed the important role of

744

C. Montanari et al. / Food Chemistry 192 (2016) 736–744

DIRECTIVE 2006/52/EC of the European parliament and of the council (5 July 2006) amending Directive 95/2/EC on food additives other than colours and sweeteners and Directive 94/35/EC on sweeteners for use in foodstuffs. Official Journal of the European Union 26.7.2006, L204, 10–22. Edwards, R. A., Ordóñez, J. A., Dainty, R. H., Hierro, E. M., & de la Hoz, L. (1999). Characterization of the headspace volatile compounds of selected Spanish dryfermented sausages. Food Chemistry, 64, 461–465. Goffin, P., Muscariello, L., Lorquet, F., Stukkens, A., Prozzi, D., Sacco, M., et al. (2006). Involvement of pyruvate oxidase activity and acetate production in the survival of Lactobacillus plantarum during the stationary phase of aerobic growth. Applied and Environmental Microbiology, 72, 7933–7940. Gutsche, K. A., Tran, T. B. T., & Vogel, R. F. (2012). Production of volatile compounds by Lactobacillus sakei branched chain a-keto acids. Food Microbiology, 29, 224–228. Hammer, Ø., & Harper, D. (2006). Paleontological data analysis. Oxford: Blackwell Publishing. Incze, K. (2010). Mold-ripened sausages. In F. Toldrá (Ed.), Handbook of meat processing (pp. 363–378). Ames: Wiley-Blackwell. Janssens, M., Myter, N., DeVuyst, L., & Leroy, F. (2012). Species diversity and metabolic impact of the microbiota are low in spontaneously acidified Belgian sausages with an added starter culture of Staphylococcus carnosus. Food Microbiology, 29, 167–177. Kauffman, R. G. (2012). Meat composition. In Y. H. Hui (Ed.), Handbook of meat and meat processing (2nd ed., pp. 45–62). Boca Raton: CRC Press. Lachowicz, K., Zochowska-Kujawska, J., & Sobczak, M. (2012). Fermented meat products. In B. M. Mehta, A. Kamal-Eldin, & R. Z. Iwanski (Eds.), Fermentation, effects on food properties (pp. 309–344). Boca Raton: CRC Press. Le Bars, D., & Yvon, M. (2008). Formation of diacetyl and acetoin by Lactococcus lactis via aspartate catabolism. Journal of Applied Microbiology, 104, 171–177. Leroy, F., Geyzen, A., Janssens, M., De Vuyst, L., & Scholliers, P. (2013). Meat fermentation at the crossroads of innovation and tradition: a historical outlook. Trends in Food Science & Technology, 31, 130–137. Lorenzo, J. M., Bedia, M., & Bañón, S. (2013). Relationship between flavour deterioration and the volatile compound profile of semi-ripened sausage. Meat Science, 93, 614–620. Marco, A., Navarro, J. L., & Flores, M. (2008). The sensory quality of dry fermented sausages as affected by fermentation stage and curing agents. European Food Research and Technology, 226, 449–458. Menon, A. N., & Padmakumari, K. P. (2005). Studies on essential oil composition of cultivars of black pepper (Piper nigrum L.)-V. Journal of Essential Oil Researh, 17, 153–155. Olivares, A., Navarro, J. L., & Flores, M. (2009a). Establishment of the contribution of volatile compounds to the aroma of fermented sausages at different stages of processing and storage. Food Chemistry, 115, 1464–1472. Olivares, A., Navarro, J. L., & Flores, M. (2009b). Distribution of volatile compounds in lean and subcutaneous fat tissues during processing of dry fermented sausages. Food Research International, 42, 1303–1308. Ordóñez, J. A., Hierro, E. M., Bruna, J. M. L., & de la Hoz, L. (1999). Changes in the components of dry-fermented sausages during ripening. Critical Reviews in Food Science and Nutrition, 39, 329–367.

Ravyts, F., De Vuyst, L., & Leroy, F. (2012). Bacterial diversity and functionalities in food fermentations. Engineering in Life Sciences, 12, 356–367. Ravyts, F., Steen, L., Goemaere, O., Paelinck, H., De Vuyst, L., & Leroy, F. (2010). The application of staphylococci with flavour-generating potential is affected by acidification in fermented dry sausages. Food Microbology, 27, 945–954. Schmidt, S., & Berger, R. G. (1998). Aroma compounds in fermented sausages of different origins. Lebensmittel-Wissenschaft und Technologie, 31, 559–567. Smit, B. A., Engels, W. J. M., & Smit, G. (2009). Branched chain aldehydes: production and breakdown pathways and relevance for flavour in foods. Applied Microbiology and Biotechnology, 81, 987–999. Stahnke, L. H. (1995). Dried sausages fermented with Staphylococcus xylosus at different temperatures and with different ingredient levels-Part II. Volatile components. Meat Science, 41, 193–209. Stahnke, L. H. (1999). Volatiles produced by Staphylococcus xylosus and Staphylococcus carnosus during growth in sausage minces: Part II. The influence of growth parameters. Lebensmittel-Wissenschaft und Technologie, 32, 365–371. Tabachnick, B. G., & Fidell, L. S. (2007). Using multivariate statistics (5th ed.). Boston: Pearson/Allyn and Bacon. Tabanelli, G., Coloretti, F., Chiavari, C., Grazia, L., Lanciotti, R., & Gardini, F. (2012). Effects of starter cultures and fermentation climate on the properties of two types of typical Italian dry fermented sausages produced under industrial conditions. Food Control, 26, 416–426. Tabanelli, G., Montanari, C., Bargossi, E., Patrignani, F., Lanciotti, R., & Gardini, F. (2014). Characterization of typical Italian fermented sausages in relation to chemico-physical and microbiological features. In: F. Toldrá & M. Flores (Eds.), Book of abstracts of the 2nd International Symposium on Fermented Meat (p. 90). Tabanelli, G., Montanari, C., Grazia, L., Lanciotti, R., & Gardini, F. (2013). Effects of aw at packaging time and atmosphere composition on aroma profile, biogenic amine content and microbiological features of dry fermented sausages. Meat Science, 94, 177–186. Talon, R., Chastagnac, L., Vergnais, L., Montel, M. C., & Berdagué, J. L. (1998). Production of esters by staphylococci. International Journal of Food Microbiology, 45, 143–150. Toldrá, F. (2006). Meat fermentation. In Y. H. Hui (Ed.), Handbook of food science, technology and engineering Volume 4 (chapter 181). Boca Raton: CRC Press. Toldrá, F., & Reig, M. (2011). Innovations for healthier processed meats. Trends in Food Science and Technology, 22, 517–522. Vera, L., Aceña, L., Guasch, J., Boqué, R., Mestres, M., & Busto, O. (2011). Characterization and classification of the aroma of beer samples by means of an MS e-nose and chemometric tools. Analytical and Bioanalytical Chemistry, 399, 2073–2081. von Wright, A., & Axelsson, L. (2011). Lactic acid bacteria: an introduction. In S. Lahtinen, A. C. Ouwehand, S. Salminen, & A. von Wright (Eds.), Lactic acid bacteria, microbial and functional aspects (4th ed., pp. 1–16). Boca Raton: CRC Press. Zielinski, A. A. F., Haminiuk, C. W. I., Nunes, C. A., Schnitzler, E., van Ruth, S. M., & Granato, D. (2014). Chemical composition, sensory properties, provenance, and bioactivity of fruit juices as assessed by chemometrics: a critical review and guideline. Comprehensive Reviews in Food Science and Food Safety, 13, 300–316.