– Volume 20 ( 2014 ) No . 2 33 IDENTIFICATION OF THE MAIN PHYSICO-CHEMICAL PROPERTIES INFLUENCING THE QUALITY OF THYME ( SATUREJA HORTENSIS )

Thyme (Satureja hortensis) is a popular spice for food, which is also often used as a medicine for various ailments. This paper presents an artificial intelligence method applied for the objective determination of the most important physico-chemical variables affecting the quality of thyme, i.e. Principal Component Analysis (PCA). The results show that the main properties which significantly influence the nutritional value of thyme are moisture (MOIST), dry matter content (DRYM), protein content (PROT) and, to a lesser extent, carbohydrate content (CARB). Humidity is strongly and negatively correlated with the latter three variables. The main variable that ensures the similarity between the thyme samples having the same geographical origin is the monosodium glutamate content, which generates its delicious (umami) taste.


INTRODUCTION
Thyme (Satureja hortensis) is a perennial plant of the Satureja genus, Labiates family (Lamiaceae).It is a short plant, of 50 -70 cm in height.Thyme prefers long, hot and dry summers, although it survives cold winters.Thyme flowers are small and have a white or pale pink color.The leaves, of oval shape, are smooth and have a distinct odor [1].Thyme is used for the preparation of various dishes, especially for beans, lentils, and the preparation of pork or game.Whole thyme leaves and stems are used as a valuable spice for conserved cabbage and cucumbers.Besides its characteristic aroma, it also has a pungent taste, thyme being also a valuable substitute to black pepper that is successfully used in the diet of people with intolerance to pepper [2].In addition thyme contains many nutrients, oils, calcium, iron, and manganese [3].
The genus Satureja constitutes about 200 species-of herbs and shrubs, widely distributed in Mediterranean area, Asia and boreal America.They are traditionally consumed for their antiseptic, antifungal, analgesic and diuretic effects [4].Recent phytochemical and bioactivity evaluations have validated the traditional uses in food and cosmetics, confirming the significant antimicrobial and antioxidant activity exhibited by these oils.The evaluation of the antibacterial activity of thyme oil, tested against clinical bacterial strains of Staphylococcus, Enterococcus, Escherichia and Pseudomonas genera, indicated that thyme oil demonstrates a better efficacy against antibiotics resistant strains of the tested bacteria than lavender oil [5].The analysis of Thymus serpyllum, Thymus algeriensis and Thymus vulgaris, performed by Gas-Chromatography -Mass Spectroscopy (GC-MS), has indicated thymol as the major component of these essential oils [6].High percentages of oxygenated monoterpenes, in particular thymol and carvacrol, have also been reported for the oil of Shirazi thyme, which also exhibits excellent anti-microbial properties [7].
Due to these properties, as well as to its exquisite flavor and taste, thyme is a valuable spice that can also be used for preservation purposes in food industry.For example, the antimicrobial activity of thyme oil was tested against two fish spoilage bacteria (Pseudomonas fluorescens and Aeromonas hydrophila/caviae) and Listeria innocua, and its potential utilization in the preservation and safety of minimally processed fish products was assessed [8].The positive effect of the thyme essential oil on some qualitative and quantitative traits and storage life of Jonagold apple was also reported [9].Investigations involving irradiation in combination with other preservation treatments indicated that essential oils in thyme are very potent anti-microbial agents, which are able to prevent the deterioration of the quality of stored foods such as mushrooms, shrimps and marinated poultry by bacteria [10].
However, not only the yield [11], but also the properties of thyme are influenced by the pedologic and related climatic conditions (climate, rainfall, mean maximum and mean minimum temperatures, mean humidity, relative heliophany and frosts) of their geographical area of origin.The growing performances of these plants in different places were recorded and the content of essential oil was determined [12].The analysis of essential oils of Satureja khuzestanica obtained by hydro-distillation, performed by high performance liquid chromatography (HPLC), indicates that there is a positive and linear relationship between the height and content of carvacrol as major component of thyme with latitude [12,13].
Thus, the identification of the main physico-chemical properties that define the quality of thyme is very important for both food and pharmaceutical industry.This paper presents an artificial intelligence method applied for the objective identification of the most important physico-chemical variables affecting the quality of thyme, i.e.Principal Component Analysis (PCA) [14].The score plots have been analyzed in conjunction with the loading plots in order to determine the type of correlations between the main physico-chemical properties.The results show that the main properties which significantly influence the nutritional value of thyme are moisture, dry matter content, protein content and, to a lesser extent, carbohydrate content.Humidity is strongly and negatively correlated with the latter three variables.The main variable that ensures the similarity between the thyme samples having the same geographical origin is the monosodium glutamate content, which generates its delicious (umami) taste [15].

EXPERIMENTAL
A number of 10 samples of thyme, originating from Galati County, Romania (sample code CGLx, x=1, 2,…, 10) have been analyzed in order to identify the main physico-chemical properties that affect their quality.
Moisture content was determined with a MF 50 AND thermo balance.The determination of protein, lipids, carbohydrate and ash was performed by standard AOAC methods.Determination of monosodium glutamate was effectuated by using the K 629 -100 enzymatic kit produced by Biovision.
The Glutamate Enzyme Mix kit recognizes a specific substrate, yielding a color reaction.Thus, the amount of glutamate can be measured by using colorimetry or spectrophotometric determinations at λ = 450 nm.Measurements were made at levels of nanomol/100 mg and the results have been converted and expressed in g.g -1 .Determination of 5'-inosine monophosphate or 5'-guanosine monophosphate was performed by using the 760 Prime Technologies capillary electrophoresis system.
The spectral analysis was performed with a Cintra UV-VIS spectrophotometer.Extracts for the spectral analysis took into account the solubility of the components responsible for umami (delicious) taste of the products.The extracts were obtained following the protocol proposed by Taylor, Hershey, Levine, Coy and Olivelle (1981): 5 g product was extracted with 25 mL of deionized water.The suspension was brought to boil 1 minute, cooled and centrifuged to 11,800 g (3,000 rot/ min -15 min).The extraction was repeated with 20 mL of deionized water.

RESULTS AND DISCUSSION
The spectra of the thyme samples were recorded between 190.3387 and 1099.992nm, by using a digitization step of 0.426829 nm (see Figure 1).Although the intensity of absorption bands has generally a low variation, significant differences are recorded in the 400 -500 nm region.The main physico-chemical properties that are generating the differences between various samples of thyme were objectively identified by using a multivariate method, i.e.Principal Component Analysis (PCA).In this regard, a hybrid training database was constructed [16], by concatenating the UV-VIS spectra of the thyme samples (spectral absorptions in the 250 -550 nm region), their humidity (variable code -MOIST), content of dry matter (variable code -DRYM), content of carbohydrates (variable code -CARB), protein content (variable code -PROT), fat (variable code -FAT), ash (variable code -ASH) and monosodium glutamate content (variable code -MSG).The analysis of the residual X-variance (see Figure 2) confirms that there are no outliers among the thyme samples.It is worth underlining that although the spectral signal is affected by an important amount of noise in the 250 -400 nm region, PCA is capable to detect and discard these random variations.Fig. 2. Analysis of the residual X-variance for the thyme samples.The PC1 vs. PC2 score plot (see Figure 3) indicates the formation of a single cluster.Its high density underlines the similarity of the thyme samples.The associated loading plot (see Figure 4) indicates that the first principal component (PC1), which explains 59 % of the variance, discriminates the samples according to their monosodium glutamate content (MSG).The second principal component (PC2), which explains 23 % of the variance, distinguishes among the thyme samples according to their moisture (MOIST) and protein content (PROT).However, these two variables have very low loadings, i.e. 0.03 and 0.01 respectively.
Although the third principal component (PC3) explains only 4 % of the variance (see Figure 5), the PC3 vs. PC1 score plot yields a less dense cluster.The associated loading plot (see Figure 6) indicates that the relatively higher dispersion of the points associated with the thyme samples is due to the differences in the moisture (MOIST), dry matter content (DRYM), protein content (PROT) and content of carbohydrates (CARB) of the samples.These variables have much larger loadings for PC3 than for PC2.It is important to emphasize that the monosodium glutamate content (MSG) is independent of the latter variables.On the other hand, the moisture (MOIST) of the samples is strongly and negatively correlated with the dry matter (DRYM), protein (PROT) and of carbohydrates (CARB) content of the samples.The fourth principal component (PC4), which explains a variance of 3 %, leads to a significant dispersion of the points associated with the thyme samples.They form a cloud located in quadrants II and III of the PC4 vs. PC1 score plot (see Figure 7).The associated loading plot (see Figure 8) indicates that dispersion is mainly due to the variations in the absorptions of the samples in the 400 -500 nm region.The same behavior is also recorded for PC5, which explains only 2 % of the variance (see Figure 9 and Figure 10).However, the monosodium glutamate content (MSG) is independent of these variables as well.

CONCLUSIONS
A hybrid database was built in order to identify the main physico-chemical properties that affect the quality of thyme having the same geographical origin.The UV-VIS spectral properties of thyme samples were concatenated with the values determined for moisture (MOIST), dry matter content (DRYM), carbohydrate content (CARB), protein content (PROT), fat (FAT), ash (ASH) and monosodium glutamate content (MSG).PCA has indicated that this training database can be compressed to a new hyperspace defined by only 4 latent independent variables, as the first four principal components cumulate an explained variance of 89 %.
The analysis of score plots and of the loading plots obtained for the first three PCs indicated that the main physico-chemical properties that define the quality of the thyme samples are the monosodium glutamate content (MSG), as well as the moisture (MOIST), the protein (PROT), carbohydrates (CARB) and dry matter (DRYM) content of the samples.A remarkable finding is that the MSG content is practically independent of the latter variables.This result is particularly important, given that the delicious (umami) taste of a vegetable, and therefore of thyme as well, is directly influenced by the content of MSG.
The fourth principal component (PC4), which explains a variance of only 3 %, leads to the formation of a more dispersed cluster, the thyme samples forming a cloud located in quadrants II and III in the PC4 vs. PC1 score plot.The associated loading plot indicates that this dispersion is mainly due to the variations recorded in the intensity of the absorption bands present in the spectra of samples in the 400 -500 nm region.The same behavior is also recorded for PC5, although it explains only 2 % of the variance.