Abstract
Twelve physicochemical parameters of 15 traditional wine samples of Msalai were completely analyzed with data mining methods to understand the interaction between physicochemical attributes and the homogeneity among the Msalais samples. Although the physicochemical differences were significant among the Msalais samples, there was also significant correlation between chromaticity and turbidity, between total soluble solids and dry extract, between total phenol and chromaticity, volatile acid content, dry extract, as well as iron content. With multivariable dimension scale analysis, 12 physicochemical parameters were clustered according to their contribution to color (iron, color), flavor (volatile acids, alcohol content, total acid, phenol, total sugar), and body (dry extract, turbidity, total soluble solids, density) of Msalais. Homogeneity was observed in the physicochemical parameters of the samples from one modern factory; however, there was strong heterogeneous in the physicochemical parameters of the samples from different traditional workhouses. Data mining with the help of a variety of analytical methods provides a good foundation for the quality control of Msalais, especially for its grade estimation.
Introduction
Msalais is a kind of grape wine that is different from conventional wine in production technology and quality. The local grapes from Southern Xinjiang are compressed and grape juice and grape residues were boiled and cooled down to room temperature to naturally ferment into Msalais, the specific technology was reported by Lixia et al.[Citation1] Generally, water is added to submerge grape residues to boil it, and then it is filtered, the filtrate is mixed with grape juice, the mixture is boiled again to form the initial fermenting substance that is finally spontaneously fermented into Msalais with indigenous microbiomes. The name Msalais refers to the one-third, which indicates the remaining one-third volume of grape juice that is obtained after removing two-thirds volume of the mixture by boiling and evaporating. Sometimes, enologists add pigeon blood, pilose antler, wolfberry fruit, and so on to increase the health benefits of Msalais. It is said that Msalais has the advantage of anti-impotence for men and age-defying properties for women in the local production region, and it is a good preventive agent for rheumatism in people who drink Msalais often. Msalais is rich in catechin, quercetion, and resveratrol. Testing in rats at the Xinjiang Medical University has proved the therapeutic effects of Msalais for cardiovascular disease (data has not been published).
Due to the unique production technology, Msalais possesses different quality characteristics of brown color, turbidity (TUB), and mellow rustic taste with strong burnt flavor.[Citation2] With the rapid economic development, Msalais has been recently produced on a large-scale based due to the demand in the market. However, the quality of Msalais varies to a large extent, at different levels, due to the arbitrariness in the production of Msalais in small workhouses, the expansion of production without quality control, and the change in traditional product characteristics without a standard criterion.[Citation3] It is essential to construct an imminent Msalais quality control system to serve the market, and thus, enhance the uniform production as well as the consumption of Msalais.
Based on computer data mining technology, such as principal component analyses (PCA) and multivariable dimension scale (MDS) analyses, experiments with limited data are used to comprehensively analyze the class of wine, to locate its originality, and to understand the relationship between the various kinds of quality indicators.[Citation4–Citation10] Physicochemical characteristics, an important part of food quality, are directly or indirectly determined by raw materials and technology.[Citation11,Citation12] Chromaticity (Chro), one of the important organic characters of the final Msalais product, is directly affected by the boiling process as well as storage with oxygen. Flavor characteristics including alcohol content (AC), total sugar (TS), total acidity (T acid), volatile acidity (V acid), and total phenol content are directly or indirectly determined by the quality of raw materials and the degree of fermentation. TUB, another important organic characteristic of Msalais, is due to the absence of a filter process, as observed in wine production. Since superfluous iron in wine will cause iron deposits, which is influenced by wine technology,[Citation12] iron content maybe an important safety index for the quality of Msalais. The basic physicochemical characteristics including relative density (RD), dry extraction (DET), and total soluble solid density (SSD) are influenced directly by both the boiling and the fermentation process. The pH value affects the quality of Msalais in many ways such as color, flavor, and even the shelf life. In the current study, we aim to provide a good standard to compare the criteria for quality control and the classification of all kinds of Msalais. Twelve physicochemical parameters including RD, DET, SSD, AC, pH, TS, T acid, V acid, TUB, Chro, total iron content, and total phenol content, were chosen to analyze the correlation of these physicochemical characteristics of Msalais and to analyze the difference between 15 Msalais samples from different brewing houses or factories.
Materials and methods
Samples of Msalais
Samples of Msalais
15 samples of Msalais were collected from seven workhouses and one factory. The name of the workhouse or factory and the code of the samples used in the study were shown in . The background of these samples was not very clear, but there is some general knowledge about them. First, the main grape variety to brew Msalais the local wine grape is named Vitisvinifera L. The Uygur herb was not extracted in the Msalais samples. The specific brewing process and the question of whether other grape varieties were mixed to brew the Msalais samples, is not clear. There are some differences between the factory and the workhouses in the production of Msalais. The main difference is in the equipment used for compression, boiling, and fermentation. In the factory, there is a continuous screw presser with a production capacity of 1000 t/d, a fermentation tank and a storage tank with a designed volume of 50 and 80m3, respectively, and a steam boiler with a designed production capacity of 6000–8000 kg at one time. On the other hand, in the traditional workhouses the equipment is much smaller and more manual. In general, grape juice is manually squeezed, boiled in an earth oven, and fermented in ceramic jars. The operation parameters in factories remain unchanged, while the parameters changed based on the enologists in traditional workhouses.
TABLE 1 The samples of Msalais and their experimental code
Physicochemical Analysis
The physicochemical analysis was carried out according to the analytical methods of wine, which fruit wine established by the National Standards of People’s Republic of China,[Citation13] the physicochemical characteristics of RD, DET, AC, pH, TS, T acid, and V acid were determined. Phenol content was determined by the Folin-Ciocalteu method.[Citation14] Chro was determined as a sum of the absorbance at 420, 520, and 620 nm using a V-530 spectrophotometer (JASCO, Tokyo, Japan) and a 1-mm path-length quartz cell.[Citation15,Citation16] Total iron content was measured with an atomic absorption spectrophotometer.[Citation17] SSD was measured with a hand-held Brix meter, and TUB was determined using photoelectric turbidimeter (WGZ-2000 type) based on the instrument operation, the 1000 range was adjusted with purified water as zero Nephelometric Turbidity Units (NTU).
Statistical Analysis
Statistical analysis, including means, one sample t-test, correlation, and PCA was performed using the PC software package SPSS9.0 (IBM Corporation, NY, USA). MDS. Classification was performed using the software package Orange Data Mining 2.7. that was downloaded from the website http://orange.biolab.si/. Statistical significance was set at p < 0.05.
Result and Discussion
Comparative Analysis of the Physicochemical Parameters of Msalais
From , a one sample t-test demonstrated a significant difference in the 12 physicochemical parameters of 15 the Msalais samples (p < 0.01). It was also proved that the physicochemical characteristics fluctuated greatly after carrying out the local standard[Citation18,Citation19] that were established by the Administration of Quality and Technology Supervision of the Xinjiang Uyghur Autonomous Region, China. TUB with a mean value of 316.87 NTU and dry extract with a mean value of 63.04 g/L were representative characteristics for the quality of Msalais, and were much higher than the common value observed in wine.[Citation16] This further confirmed the unique quality of Msalais. At the same time, AC, total acid, volatile acid, RD, and total soluble solids density meet the requirement of the standard of GB 15037-2006 for wine.[Citation20] Thus, Msalais may be regarded as a special wine from a scientific point of view.
TABLE 2 The value of physicochemical parameters from 15 Msalais samples and the significant difference analysis (one sample t-test)
The difference between the maximum and the minimum values of SSD was 27.3 Brix, the average value was 16.48 Brix, and the minimum value is 10.1 Brix. The average value and most of the individual SSD value were just over the limit for white wine in GB 15037-2006.[Citation20] TS content that is the main indicator of the quality of Msalais was in the range of 7.5 ~ 12.5 g/L, and it was also between the high value zone observed in dry wine and the low value zone observed in semi-sweet type wine. The TS content of the samples GuZuoFang (GZF), PaLiHati (PLH), and WuSiman (WS) was within the range of TS content in natural dry and dry sparkling wine, it was interesting that these samples contained air like sparkling wine. However, AC ranges between 3.62 ~ 11.5%, with 7.85% average in Msalais. In the four Msalais samples of XieHui (XH), XJ, PLH, and DP, the AC is much lower than 7.5 g/L, the lowest value in GB 15037-2006 for wine.[Citation20] TS and AC values in Msalais are not high. The low value maybe due to the natural fermentation that is mostly influenced by temperature fluctuation, the yeast community structure, and enological characters.
The total phenols in the 15 samples was a minimum of 6.12 mg/L and a maximum of 73.73 mg/L with a difference of 57.61 mg/L and a mean value of 32.71 mg/L. However, on the whole, the value range of total phenols in Msalais was still far below that of the red and white wine.[Citation16,Citation21–Citation24] Different processing methods for grapes greatly influence total phenols content and its species.[Citation25] The low phenol content in Msalais can be explained by the brewing materials from the local edible varieties of grape with low tannin and anthocyanins. In addition to this, boiling of grape juice or grape residues could lead to the loss of phenol.
TUB is the basic feature of Msalais, and is an important index for experienced winemakers to evaluate the quality of Msalais. In the 15 samples, TUB was between 64.1 and 1015 NTU. There was excessively clarified sample, for example, Ali (AL) and an excessively cloudy sample, GZF. Artificially adding clarifier or transferring Msalais from one tank to another tank too often excessively clarifies Msalais (about 100 NTU). The flavor of artificially clarified Msalais was different from the traditional Msalais with the natural sedimentation. To meet the demand of some consumers in the southeast part of China, clarified Msalais has appeared in the market. However, we are faced with the problem of classifying Msalais while ensuring the flavor characteristics with yeast mud and the TUB of liquor body. Excessive cloudy with a lot of yeast mud may influence the sensory characteristics of Msalais.
Factors that impact wine chromaticity, includes phenols, raw materials in the flower pigment, or even the concentration of sugar.[Citation21,Citation26] The chromaticity value of Msalais mainly comes from the boiling process with the intense Maillard reaction and the polymerization of phenols with an aerobic thermal reaction. The chromaticity is due to decomposed by yeast during fermentation and is influenced by the absorption of yeast mud during the aging stage. The effects of precipitation cannot be ignored. Chromaticity has the mean value of 5.50 with the range of 1.62 ~ 27.1 in Msalais. Chromaticity in Msalais was not obviously different from the Chinese grape wine of the Helan Mountain and the Jingyang region,[Citation16,Citation27] but it was low compared to other wine.[Citation21,Citation22] The sample of GZF had the highest chromaticity value and the highest TUB value. There is correlation between the two parameters of chromaticity and TUB.
Total acid and volatile acid of most Msalais samples were within the range of GB 15037-2006 (less than 8 and 1 g/L, respectively).[Citation20] However, the total acid of DW and H samples was much higher than 10 g/L and the volatile acid was much higher than 1 g/L, this was abnormal for wine. Volatile acid could be used as one of the indicators of microbial contamination during the whole process and the hygiene index for environmental brewing Msalais.
Traditionally, boiling grape juice in an earthen pot with high iron content to ferment Msalais would lead to a high total iron content in Msalais. Total iron content in the sample of PLH researches 15.68 mg/L, much higher than the 8 mg/L standard in GB 15037-2006.[Citation20] It is also reasonable to assume that the iron content of sample DL2 was the minimum value of 0.635 mg/L because of the stainless container used to boil and ferment grape juice. Due to varied grade of brown color for Msalais, the organic quality, rather than wine color, was affected less by iron content. If Msalais was exposed to air, iron content may have an impact on the color of Msalais, which changes from red brown to dark brown.
Correlation Between the Physicochemical Parameters
In , at the p < 0.05 level, total phenol content had a significant correlation with chromaticity, volatile acid, dry extract, and total iron content. At the p < 0.01 level, the correlation was also shown to be significant between volatile acid and chromaticity, between chromaticity and volatile acid and TUB, between total soluble solids and dry extract, and RD as well as AC. The similarity between these physicochemical characteristics were further to analyze by MDS, as shown in . There was an interesting observation that these physicochemical parameters were clustered according to their contribution to the color, the flavor, and the body of Msalais. V acid, T acid, AC, TS, and phenol were obviously classed together, this proves there is a close relationship between the five physicochemical characteristics, and they all contribute to the flavor of Msalais. Chro had a good similarity with iron content. It conformed with a significant correlation between the two characteristics at p < 0.05. This indicates iron may influence on the color of Msalais. RD, SSD, DET, and TUB, representing the body of Msalais, did not belong to the group of flavor and color characteristics, DET, and TUB did not grouped with any other characteristic, while RD had a good similarity with SSD. This proved DET and TUB were both indispensable to evaluate the quality of Msalais.
TABLE 3 Correlation between the physicochemical parameters
FIGURE 1 Multi-dimensional scaling (MDS) of physicochemical characters. A low average signed Sammon stress level of 0.1238 was obtained, meaning that the 2D representation of the physicochemical characters involves a low loss of information. Gray lines connect similar physicochemical characters. The width of line shown the distance between physicochemical characters. Symbol size shown the stress of each physicochemical characters. The green characters represented the flavor of Msalais, the red characters represented the color of Msalais, the blue represented the body of Msalais.
Classification Analysis
For PCA analyses, Kaiser-Meyer-Olkin (KMO) and Barlett’s test was carried out initially and shown to be significant. The four principal components were extracted with 84.52% sum of cumulative contribution rate (). The components, Comp1 and Comp2 were mainly made up of phenol, iron, TUB, T acid, V acid, and AC, the Comp2 and Comp4 components were mainly made up of pH, TS, Chro, and RD. Comp1 and Comp4 mainly explained the factors of DET and SSD (). Based on these principal compounds, the factor score of each sample was obtained (data not shown here). The 15 samples were clustered into four groups, including the group of XH and PLH, the group of DW and H, the group of WS and GZF, and the group of the remaining nine samples except MaiTuDi (MDT) from Daolang Co. (). We obtained similar results by the MDS analyses. All the nine samples from the modern factory of Daolang Co. were classified together, while the six samples from the traditional workhouses were distributed into three clusters with the group of MTD, WF, WS, and GZF, the group of DW and H, and the independent PLH (). MDS had the ability to differentiate the samples from the factory and the traditional workhouse. This proved that the quality of Msalais from the same plant with a large production scale, displayed the homogeneity, while the smaller traditional workhouse displayed heterogeneity in the Msalais samples. The heterogeneity in the quality of Msalais from different traditional workhouse, may be related to the brewing microbiomes in the different workhouses, and the different operation parameters that is depended on the experience of the enologists. A set of parameters control the brewing process in modern plants. The variety and the maturity of the grape was another factor that affects the heterogeneity, but it may not be as strong as the effect of boiling and fermentation processes that affect Msalais quality, since boiling may partly destroy the difference in the grape varieties and mature to form an even quality of the initial fermenting substances. The results demonstrate the challenge in the production of Msalais in order to keep increasing the production scale without the loss of the traditional diversity and high quality. The clustering results were another evidence of the use of MDS, PCA analyses in tracing a regional sample.[Citation4,Citation9,Citation28]
TABLE 4 Total variance explained for PCA
FIGURE 2 The physicochemical parameter distribution in the 3D space constructed by Comp1, Comp2, and Comp4, with an 84.52% sum of cumulative contribution rate.
FIGURE 3 The classification of 15 Msalais samples based the score of factors extracted from the PCA.
FIGURE 4 Multi-dimensional scaling (MDS) of samples. A low average signed Sammon stress level of 0.2063 was obtained, meaning that the 2D representation of the samples involves a low loss of information. Gray lines connect similar physicochemical characters. The width of line shown the distance between physicochemical characters.
Further analyses together with demonstrate that the Chro of XH and PLH was significantly higher than that of the other samples, while the AC of DW and H was significantly lower than that of the other samples. There is no obvious common characteristic between WS and GZF and the rest of the samples. Thus, the analyses based on PCA and MSD, not only identify immediately the main factors affecting the quality of the samples, but are also capable of scientifically clustering the multiple samples. Data mining may provide a powerful scientific tool or comprehensive analysis and classification of multiple samples.
Conclusion
The analyses of the 12 physical and chemical parameters of 15 the samples of Msalais demonstrate that the value of physicochemical parameters among different Msalais samples was are significantly different. There is a significant correlation between color and TUB, between total soluble solids and dry extract, between RD and AC, and between total phenol content and chromaticity, volatile acid content, dry extract, and iron content. With MDS analysis, the 12 physicochemical parameters were clustered according to their contribution to color (iron, color), flavor (volatile acids, AC, total acid, phenol, TS), and body (dry extract, TUB, total soluble solids, density) of Msalais. Most of the samples from a modern Msalais plant were clustered together and proved the homogeneous quality among the samples; however, the samples from different traditional Msalais workhouses were distributed among four different groups that explained the heterogeneity among these samples. Based on PC software, combination of a variety of analysis methods such as one sample t-test, correlation analysis, PCA, and MSD may be used to completely analyze the product quality of Msalais. In the future, the analysis of sensory characteristics and some key specific compounds including volatile components in Msalais may be taken to construct the method system for discriminating and grading Msalais by data mining.
FUNDING
This article was supported by the National Natural Science Foundation projects of Xinjiang wine yeast high-throughput analysis of genetic diversity, its fermentation characteristics (31260393), and the enological characteristics of Sacccharomy cescerevisiae associated with Msalais (13060023). Xinjiang Construction Corps Industry Science and Technology Project (2014BA026) also supported the study for sample collection.
Additional information
Funding
REFERENCES
- Lixia, Z.; Mingfu, G.; Dongqi, G.; Christensen, M.; Xujie, H.; Hongmei, L.; Shu, F. Preliminary Analysis of Yeast Communities Associated with the Spontaneous Fermentation of Musalais, a Traditional Alcoholic Beverage of Southern Xinjiang, China. South African Journal of Enology and Viticulture 2012, 33(1), 95–104.
- Lixia, Z.; Wen, Z. Quantitative Descriptive Sensory Analysis of Msalaisi from A’wati, Xinjiang. Food Science (in Chinese) 2013, 34(1), 38–44.
- Lixia, Z.; Feng, S.; Dongqi, G.; Xujie, H.; Jingcheng, Y. The Technology of Musalais, a Traditional Alcohol Beverage, in Xinjiang, 7th Proceedings of the Seventh International Symposium on Viticulture and Enology, Yangling, China, 169–171 pp., 2011; Li, H.; Ed.; The Peoples Press of Shannxi.
- Greenough, J.D.; Longerich, H.P.; Jackson, S.E. Element Fingerprinting of Okanagan Valley Wines Using ICP-MS: Relationships Between Wine Composition, Vineyard, and Wine Colour. Australian Journal of Grape and Wine Research 1997, 3(2), 75–83.
- Gawel, R.; Iland, P.G.; Francis, I.L. Characterizing the Astringency of Red Wine: A Case Study. Food Quality and Preference 2011, 12(1), 83–94.
- Ouyang, Q.; Zhao, J.; Chen, Q.; Lin, H. Classification of Rice Wine According to Different Marked Ages Using a Novel Artificial Olfactory Technique Based on Colorimetric Sensor Array. Food Chemistry 2013, 138(2), 1320–1324.
- Green, J.A.; Parr, W.V.; Breitmeyer, J.; Valentin, D.; Sherlock, R. Sensory and Chemical Characterization of Sauvignon Blanc Wine: Influence of Source of Origin. Food Research International 2011, 44(9), 2788–2797.
- González, Á.M.; González, B.C.; Cancho, G.B.; Simal, G.J. Relationships Between Godello White Wine Sensory Properties and Its Aromatic Fingerprinting Obtained by GC–MS. Food Chemistry 2011, 129(3), 890–898.
- Roullier, G.C.; Boutegrabet, L.; Gougeon, R.D.; Schmitt-Kopplin, P. A Grape and Wine Chemodiversity Comparison of Different Appellations in Burgundy: Vintage Versus Terroir Effects. Food Chemistry 2014, 152, 100–107.
- Johnson, T.E.; Hasted, A.; Ristic, R.; Bastian, S.E. Multidimensional Scaling (MDS), Cluster and Descriptive Analyses Provide Preliminary Insights into Australian Shiraz Wine Regional Characteristics. Food Quality and Preference 2013, 29(2), 174–185.
- Aday, S.; Karagul Yuceer, Y. Physicochemical and Sensory Properties of Mihalic Cheese. International Journal of Food Properties 2014, 17(10), 2207–2227.
- Soto Vázquez, E.; Rio Segade, S.; Fernández Gomez, E. Incidence of the Winemaking Technique on Metal Content and Phenolic Composition of Red Wines. International Journal of Food Properties 2013, 16(3), 622–633.
- GB/T 15038:2006. Analytical methods of wine and fruit wine. National Standards of People’s Republic of China,Standards Press of China, Beijing, 2006.
- Ivanov, V.; Vojnoski, B.; Stefova, M. Effect of Winemaking Treatment and Wine Aging on Phenolic Content in Vranec Wines. Journal of the Science of Food and Agriculture 2012, 49(2), 161–172.
- Hanzhang, Q. Analytical Chemistry of Wine; Light Industry Press of China: Beijing, China, 1992.
- Xiaoyong, H. Influences of phenols on the color of the new red wine produced at the eastern food of Helan Mountain. Thesis (Masters), Ningxia University, China, 2013.
- Karadjova, I.; Izgi, B.; Gucer, S. Fractionation and Speciation of Cu, Zn, and Fe in Wine Samples by Atomic Absorption Spectrometry. SpectrochimicaActa Part B: Atomic Spectroscopy 2002, 57(3), 581–590.
- DB65/T 2924:2008. Msalais alcoholic beverage. Xinjiang Bureau of Quality and Technical Supervision, Xinjiang, 2008.
- DB65/T 2925:2008.The technics criterion of Msalais produce machining. Xinjiang Bureau of Quality and Technical Supervision, Xinjiang, 2008.
- GB /15037:2006. Wine. National Standards of People’s Republic of China, Standards Press of China, Beijing, 2006.
- Dongmei, L.; Weiqiang, Z. Factors Influencing the Chromaticity of Tinto. Liquor Making Science Technolgy 2006, 4, 55–57.
- Teng, M.; Li, Z.; Jun, L. Statistical Analysis of Physical and Chemical Properties of Wine Produced in 2008 in Changli. Journal of Hebei Normal University of Science & Technology 2012, 1, 5–11, 36.
- Somers, T.C.; Verette, E. Phenolic Composition of Natural Wine Types. In Wine Analysis; Hans, F.L.; John, F.J.; Eds.; Springer Heidelberg, Berlin, 1988; 219–257 pp.
- Kassara, S.; Kennedy, J.A. Relationship Between Red Wine Grade and Phenolics, Tannin Composition, and Size. Journal of Agricultural and Food Chemistry 2011, 59(15), 8409–8412.
- Garrido, J.; Borges, F. Wine and Grape Polyphenols—A Chemical Perspective. Food Research International 2013, 54(2), 1844–1858.
- Figueiredo, G.M.; Cancho, G.B.; Simal, G.J. Effects on Colour and Phenolic Composition of Sugar Concentration Processes in Dried-on-Or Dried-off-Vine Grapes and Their Aged or Not Natural Sweet Wines. Trends in Food Science & Technology 2013, 31(1), 36–54.
- Ling, L. Study on the quality of grape and wine in Jing Yang. Thesis (Masters), Northwest Agriculture and Forest University, China, 2010.
- Díaz, C.; Conde, J.E.; Estévez, D.; Pérez Olivero, S.J.; Pérez, T.J.P. Application of Multivariate Analysis and Artificial Neural Networks for the Differentiation of Red Wines from the Canary Islands According to the Island of Origin. Journal of Agricultural and Food Chemistry 2003, 51(15), 4303–4307.