3.1. ABTS•+ Radical Scavenging Capacity and Phenolic Content of Spice Extracts
The ABTS
•+ radical scavenging activity of 7 plant extracts is shown in
Table 1. Among tested spices, cloves were characterized by the highest antioxidant activity with the TEAC value of 2496 μM/g. The radical scavenging capacity of extracts from allspice and bay leaf was 3-fold and 9-fold lower, respectively, when comparing to cloves. Nutmeg, caraway, black seed and cardamom with the TEAC values ranged from 45 to 13 μM/g formed the group of spices with the lowest antioxidant activity. Our results are in agreement with the study of [
32] who showed that the ABTS
•+ radical scavenging capacity of the extracts from cloves and allspice were very high with the TEAC values of 2071 and 719 μM/g, respectively. In this study, the rank of radical scavenging capacity of extracts provided by the TEAC assay was as followed: Cloves > allspice > bay leaf > nutmeg ≥ caraway ≈ black seed ≥ cardamom and was the same as the rank of DPPH radical scavenging activity and similar to the rank provided by the FRAP assay, both showed in our previous paper [
13].
High positive Pearson’s correlation coefficients were observed between the results of TEAC and DPPH methods (r = 0.997 at
p = 0.000, Student’s t test) and TEAC and FRAP methods (r = 0.998 at
p = 0.000, Student’s t test). As it was shown previously [
13] the antioxidant activity of spice extracts resulted from their phenolic content. For the correlation of ABTS
•+ radical scavenging activity of spice extracts with their total polyphenol content (from [
13]) the r Pearson’s coefficient was 0.988 (at
p = 0.000, Student’s t test).
Table 2 includes the results of UPLC analysis of phenolic acids and some polyphenols in spice extracts. The highest content of phenolic acids was observed for clove extract. Surprisingly, levels of compounds in allspice extract quantified by UPLC was very low despite showing high total polyphenol content measured by Folin–Ciocalteu’s methods [
13]. However, significant correlation was noticed between total phenolic acids by UPLC (
Table 2) and the TEAC values (
Table 1) of extracts, with r Pearson’s coefficient equaled to r = 0.81 (
p = 0.027, Student’s t test).
Generally, the direct comparison of phenolic content of extracts with the literature data was difficult, since huge variations were observed in the qualitative and quantitative characteristics of spice extracts reported by others [
33,
34,
35]. These differences could be attributed to the genotypic and environmental differences within species, sample preparation and determination methods.
3.4. Color and pH
The pH value could be used as the indicator of meat quality which is affected by microbiological changes and various chemical reaction including oxidation. Both factors: Time and treatments had significant effects on the pH values (
Table 4). The values of pH of all samples decreased (up to 5th day for control sample and up to 7th day of storage for extract-treated samples), thereafter increased sharply showing the maximum on day 10, and decreased again. This increase was the most pronounced in the control sample (maximum pH = 7.32). The sharp increase of pH could be an effect of protein oxidation, and the production of ammonia (by utilization of amino acids by spoilage microorganisms) as storage progresses [
9]. The initial level of pH in the control samples was similar to that reported by [
41]. However others [
9] observed an increase of pH values of raw chicken meat over storage period. Generally, the addition of extracts decreased pH values when comparing to control sample, and the lowest pH values were observed for the sample containing cardamom (
Table 4). Similar observations have also been made by [
9] who reported that the addition of clove extract decreased pH values in comparison to control samples. The effect was ascribed to the antimicrobial activity of compounds found in clove extract which could inhibit the growth and proliferation of spoilage microorganisms that metabolize basic nitrogen compounds [
9].
Table 4 contains the results of the instrumental color measurements using L*a*b* color space. All color parameters (L*, a*, b*) were significantly affected by the treatments (
p = 0.00, F test) and storage time (
p = 0.01, F test). The lightness of all treated samples (except cardamom) was lower than the control one at each day of the analysis. The highest decrease of L* parameter was observed for clove- and black seed-treated chicken meat. This affected the sensory evaluation of color of these meat samples by the panelist (see below
Section 3.6). Similar results were obtained by [
41] who reported lower L* values of raw chicken meat samples containing acorn when comparing to control sample. However, others [
9] reported significant higher lightness of raw chicken samples with clove extracts in comparison to control sample which is contrary to the results obtained in this study.
The significant reduction in redness (a*) of raw chicken meat was observed after the addition of cardamom, bay leaf and black seed (
Table 4). Similar effect was reported previously in raw chicken containing acorn [
41]. The highest values of a* parameter were noticed for clove-treated sample. During the first days of storage the value of a* parameter increased slightly, peaked on the 3rd day and decreased thereafter with time in all tested samples. The decrease of a* values with time, and the lowest a* values for control sample were reported by [
9]. The changes of a* parameter of raw chicken meat could be due to the formation of metmyoglobin as a result of the interaction between products of pigment oxidation and lipid oxidation [
44].
The changes in yellowness (b* values) were very small during storage (
Table 4). Apart from bay leaf- and caraway-treated samples with unchanged b* values, all other samples showed slight decrease of the parameter with time. Yellowness of raw chicken meat containing allspice and black seed were significantly lower, whereas nutmeg, bay leaf, and cloves significantly higher when comparing to control sample. The effect of natural extract addition on the color parameters depends strongly on the type of the treatment [
13].
3.5. Micobiological Analysis
The results on the number of microorganisms in all analyzed samples were shown in
Table 5. Based on data analysis, significant differences were found in TVC during storage and between treatments. The starting TVC was approximately 5 log cfu/g for all samples with a value of 5.33 log cfu/g for the control one. The amount of microorganisms increases with storage time. Similar results on TVC values in raw chicken patties during storage at 4 °C were reported by others [
12,
36]. Generally, the TVC for the samples with plant extract addition was lower than the control sample, and bay leaf was the most potent antimicrobial extract (based on ANCOVA).
Pseudomonas spp. and
Enterobateriaceae counts were significantly affected by the time and treatments. The counts increased with increasing time, and except the sample containing cardamom, all other samples showed lower total mean
Pseudomonas spp. and
Enterobateriaceae counts than the control sample. The increases of these Gram negative bacteria with time were consistent with the results reported by [
9].
LAB is able to grow both in aerobic and anaerobic conditions and is a substantial part of the natural microflora of meat [
9]. The factors of both time and treatments had significant effects on LAB count. The initial LAB counts were from 3.50 for caraway to 3.85 log cfu/g for black seed samples. For all samples the LAB counts increased during the whole storage period reaching the highest level for control sample. The lowest total LAB count was reported for caraway sample (4.32 log cfu/g). Among tested extracts, black seed inhibited the growth of LAB to the least extent. Similar values for control sample and raw chicken meat containing clove extract were reported by [
9].
3.6. Sensory Analysis
Results of sensory evaluation (total mean values by ANCOVA) were presented on the
Figure 5a–c. Time and treatments had significant effects on sensory scores. The values of all evaluated attributes decreased with time, and the decrease was the most pronounced in the control sample (
Figure S1). The highest scores of all attributes were ascribed for the sample with cardamom followed by cloves.
All extracts inhibited the odor alteration of meat during storage when comparing to control (
Figure 5a,
Figure S1). Similar results were reported by [
11], who noticed that sensory scores for control sample differed significantly from pork nuggets with black seed addition, which might result from inhibition of lipid oxidation and off-flavor development during storage by phenolic compounds found in black seed extract. The odor alteration of cooked chicken meat with the addition of wine industry residues extract was also lower comparing to control, as shown by [
45]. In accordance to the results of this study, the protective effect of clove extract on odor attribute of raw chicken was also noticed by [
9].
In regards to color attribute no significant differences were observed between bay leaf-, black seed-treated, and control samples (
Figure 5b). This could result from the dark color of the extracts (
Table 4), as was previously observed in chicken meat with grape extract [
45].
The addition of spice extract to chicken meat improved the texture of the products (
Figure 5c). Others reported lack of statistically significant differences between texture scores of control and black seed-treated pork nuggets [
11], and control and plant extract-treated goat meat [
46].
3.7. Multivariate Analysis
To put more insight into the data, multivariate analyses were performed, namely: PCA, CA, and GDA. Prior to statistical analysis r Pearson’s correlation coefficients were calculated (
Table S1) and the data matrix was standardized. Based on the data mining algorithm: Selection and elimination of variables, the following variables were characterized by the highest importance (expressed as Chi-squared) and chosen for further analysis: TBARS, hexanal, CD, and SH contents and L*, a*, b* parameters (
Figure S2).
All parameters of oxidation were significantly correlated (
supplementary materials Table S1). The highest r Pearson’s correlation coefficients were determined between SH content and TBARS (−0.70). Significant correlations were also reported between lightness and: Hexanal, TBARS, or SH content. The latter was equal to −0.67. Mir et al. [
47] reported in the review that oxidation changes of proteins in chicken meat affected the light scattering and thus L* values. At minimal protein denaturation, low light scattering is observed and the chicken meat has dark color. Thus, it could explain the significant correlation between lightness and protein oxidation observed in the study. The color coordinates: a* and b* were also significantly correlated. The correlations observed were proved by PCA (
Figure 6a). PCA gives the possibility to compress the data to a few principal components (PCs) that can be later used to reconstruct data without any preliminary assumptions about their distribution and almost without any loss of information [
31,
48]. This helps to understand the relationships between variables.
Using graphical criterion, the first four PCs were extracted (three of them with eigenvalue greater than 1) and together they explained 87.3% of the total variance. The PC1 explained 42% of variance and the PC2 the 23% of variance. Based on the loaded values, showing the correlations between components and variables, it could be stated that TBARS, hexanal content, SH content, and lightness (L*) determined sample distribution along PC1 (
Figure 6a). Similar results were obtained by [
49], who also reported high correlations between lipid, protein and myoglobin oxidation parameters in PCA. Moreover based on the PCA, the authors concluded that myoglobin, lipid, and protein oxidation reactions in rabbit meat were interlocked with each other in a certain manner, and each process seemed to facilitate one another [
49]. The same could be stated based on the results showed in this study, since the projection of variable on the factor plane (
Figure 6a) indicated complex and inseparable interrelationship among them. When oxidation of lipid occurred the oxidation reactions could be transferred to proteins fraction [
40,
49] and meat’s pigment [
50]. An inverse interaction could be also carried out [
51].
Figure 6b showed the projection of the scores on the factor plane. PC1 enabled to separate control sample (on the right of PC1) with the highest lipid oxidation and lightness and the lowest SH content from other samples. Clove-, black seed-, and allspice-treated samples characterized with the lowest protein oxidation (the highest SH groups) among all tested extracts, thus they were on the left of PC1. PC2 was mainly correlated with a* and b* color parameters and it grouped well the samples of raw chicken meat with cloves (with the highest values of a* and b* among samples) and black seed (with the lowest values of a* and b*).
The PCA is less sensitive to differentiate all samples (with allspice, bay leaf, cardamom, caraway and nutmeg). Thus, the CA was applied as another unsupervised pattern recognition technique of multivariate analysis to group samples on the basis of similarities in clusters. The Ward method and Euclidean distance between centroids were applied. The CA dendrogram enabled to separate eight clusters (
Figure 6c). All samples with the plant extracts were stacked in one cluster, whereas control sample with high TBARS, hexanal and L* values was stacked in a separate cluster. Moreover, clove-, black seed-, and allspice-treated samples with low lipid/protein oxidation heaped in one cluster different from the rest of treated samples. Allspice and black seed, both with low values of b* coordinates, were stacked in another cluster than cloves showing high b* value. Also bay leaf (with low a* value) was heaped in clusters other than cardamom, caraway, and nutmeg. The lowest Euclidean distance was measured between nutmeg- and caraway-treated samples. These samples were the most similar according to the tested parameters, which disrupted to discriminate them well. The small distance was also noticed between caraway and cardamom samples.
GDA is supervised pattern recognition technique enabling obtainment of discriminant functions, which maximize the ratio of between-class variance and minimize the ratio of within-class variance [
31]. The first 5 discriminant functions were statistically significant and explained 99.9% of total variance. The first discriminant function accounted for 49.7% of total variance, and the second one for 33.6%. Based on standardized canonical discriminant function coefficients the importance of the independent variables in the canonical function was discussed (
Table 6). The higher the absolute value of standardized coefficient the higher the discrimination power of the related variable is.
SH content contributed the most to the first canonical function, CD content to the second one, whereas a* parameter to the third one. The discrimination of meat samples with spice extracts was noticeable (
Figure 6d). Based on the results it could be concluded that meat samples showing very high oxidative stability of proteins, namely: Clove-, allspice-, and black seed-treated samples, were excellent discriminated with 100% of the classification’s propriety. The propriety of classification was also 100% for the samples treated with bay leaf and caraway. Only 2 control samples, 3 samples with nutmeg, and 2 samples with cardamom extracts were incorrectly classified. Summarizing, according to the classification matrix 95.1% of the samples were classified correctly which proved very strong classification ability of the constructed discrimination models. Discriminant functions were successfully used previously to classified samples of raw ground pork with garlic and marjoram according to the fatty acid composition [
10].