1. Introduction
T. matsutake grown in different areas has different qualities. In China,
T. matsutake in Yunnan Province is the most famous, and the quality of
T. matsutake in Tibet is the best [
1]. Tibet has rich species, a variety of vegetation types, and superior animal and fungi sources because of its serene nature. The average altitude of the Tibet Autonomous Region is over 4000 m. It is located in the borderland, and has transport problems, leading to long logistics paths and high cost. The development of edible fungi industry in Tibet is faced with the challenge of extending the shelf life and minimizing post-harvest food and economic losses [
2].
T. matsutake has a high reputation in Japan and South Korea [
3], and it has always been admitted as a treasure in the edible mushroom family, and has been called the king of mushrooms [
4]. It was a kind of precious wild edible fungi with its strong fragrance, stout body, delicate flesh, and special flavor, and took the fancy of consumers for its high nutritional and medicinal value [
5]. As a kind of wild edible fungus,
T. matsutake has a relatively short life when compared with other edible funguses, such as
mushroom, white
Hgpsizygus marmoreus and
Pleurotus ostyeatus [
6,
7,
8], due to high respiration rate and transpiration, which significantly decrease its marketing period [
9,
10]. There is no protective structure on the surface of
T. matsutake, which makes it highly susceptible to infection and then decay [
9]. The shelf life of
T.matsutake is about 2 days after harvest, when it is preserved under the temperature of about 20 °C [
11]. Then it happens cap opening, browning, autolysis, and other phenomena, and loses its commodity and edible value [
3]. Its high fragility and its very short shelf life is the main obstacle to its transportation and consumption, leading to considerable waste and loss during the post-harvest steps [
2].
The cold chain can postpone the quality loss of edible fungi to increase its shelf life, but it is far from being able to meet the industry needs [
12]. Modified atmosphere packaging (MAP) technology has been widely used for the shelf life extension of edible funguses [
13,
14,
15], fruit, and vegetables [
16], which is a promising technique to delay the deterioration of
T. matsutake in the post-harvest chain. MAP can achieve an optimal gas composition in the close environment of the product [
16,
17,
18]. The modified atmosphere condition slows down the respiration rate, ethylene production, and water loss, thus reducing the enzyme activities and metabolic rate of fresh product [
19,
20,
21,
22]. In recent years, MAP has proved an effective method to modify the physiology and prolong the shelf-life of fresh food by flushing the desired initial gas into the packages in a cold chain [
23,
24].
T. matsutake is living organism; it continues to respire after harvest by absorbing oxygen from air and producing carbon dioxide. After a transition phase, a gas equilibrium is established around the product, of which the composition must be as close as possible to the optimal one to reduce respiration, prevent ripening, senescence, and fermentation, and thus increase shelf life [
16,
25].
Although significant benefit in terms of food loss reduction is expected from shelf life extension, this direct positive effect is difficult to anticipate for the post-harvest step [
26]. This is due to the lack of generalized quality indicators to quantify the shelf life of edible fungi products in general, especially in
T. matsutake. The shelf life of
T. matsutake is complex, influenced by product characteristics (weight loss rate, browning index, pH, soluble solids content, etc.), surrounding environmental conditions (temperature, gas composition, and relative humidity of the atmosphere), and spoilage development [
4,
9,
27]. Many quality parameters, such as firmness, chromatic aberration, pH, respiratory intensity, soluble solids content (SSC), browning index, weight loss rate, and sensory evaluation have been considered to be the main shelf life influence factors of
T. matsutake [
3,
11]. It is not certain which indicators directly affect its quality and shelf life. Therefore, it is essential to analyze the shelf life parameters of
T. matsutake and select the main factors to quantify its shelf life quality. Correlation analysis is a statistical method used to evaluate the strength of relationships between different variables [
28]. Zhang et al. found that the hardness, protein, total sugar, chromatic aberration (L*), and pH of
Pleurotus eryngii mushroom displayed a significantly positive correlation with sensory quality score using the method of correlation analysis [
14]. Fu et al. used correlation analysis to study the relationship between gas and physical and chemical indexes on shelf life of blueberry [
29]. In this paper, we adopt correlation analysis to optimize the quality indicators on shelf life of
T. matsutake by analyzing the relationship between the quality indexes data with its remaining shelf life.
Research has been done on predicting the shelf life of edible fungi. Researchers discussed the shelf life prediction models of edible mushrooms, such as
Shiitake mushroom, by using a kinetic model [
26,
30] based on temperature parameters, or statistics models [
6,
14]. However, there are few studies predicting the shelf life of edible fungi based on the quality indicators under the MAP condition. With the development of science and technology, the back propagation (BP) neural network has become the most widely used prediction model, which is a multi-layer feedforward one-way propagation network based on error feedback and using a backward propagation algorithm [
31]. Compared with the traditional statistical methods, the BP neural network can correct the equation according to the quality index of the food and obtain the mathematical equation of the appropriate product to predict the shelf life [
32]. The BP neural network has been applied to the shelf life prediction of fruit and vegetables, such as blueberry [
29], fresh egg [
33], and some food like quick-frozen dumpling [
34].
Based on the above discussion, this paper aims to study the quality indicators of fresh Tibetan T. matsutake changes and build its remaining shelf life prediction model through optimizing the quality characteristics of T. matsutake in different MAP under the temperature condition of 4 °C and relative humidity of 90%. The data in the experiment of quality indicators were analyzed to determine the variation tendency. The remaining shelf life prediction model of T. matsutake was established by the BP neural network method. Correlation analysis was used to optimize the indicators affecting the shelf life by the relativity between quality indicators and the remaining shelf life of T. matsutake. It could provide a reference and help for the preservation, transportation and sales of T. matsutake.
2. Materials and Methods
2.1. Raw Material
Fresh
T. matsutake were obtained from the forest area of Lingzhi City, Tibet Autonomous Region.
T. matsutake were harvested in the morning of the experiments, cooled to about 5 °C in the incubator, then transported to the laboratory under refrigerated conditions (4 ± 1 °C) and 90% relative humidity (RH) around 2 h after harvest. Harvested
T. matsutake were sorted according to the following standards: Maturity was 80–90%. The cap was smooth and unopened, and the color was shining chestnut-colored. The stipe was not shedding, and its color was ivory. Sorting was done also according to shape. Oversized
T. matsutake or very small ones compared to the batch
T. matsutake sizes were eliminated. Finally, damaged or rotten
T. matsutake were removed [
35]. They were cleaned by a soft brush, which was used to remove debris from the
T. matsutake surface and soil from the root.
After pre-cooling, sorting, and cleaning treatments, the T. matsutake were packaged in vacuum fresh keeping bags made of polyethylene (PE) (Weide New-material Corp., Xuzhou, Jiangsu, China) with dimensions of 0.15 × 0.20 m. Each bag contained 2 pieces of T. matsutake of about 0.2 kg. The total amount of fresh T. matsutake was 10 kg for this experiment.
2.2. Packaging, Storage, and Atmosphere Composition
Four packaging conditions were considered in these experiments, and the components are exemplified in
Table 1. The temperature of these tests was set at 4 °C and the relative humidity was 90%. The air group was used as a controlled trial, and the other three groups were set with different ratios of O
2 and CO
2 in the atmosphere space to do the experiments.
Before closing the fresh-keeping bags, which contained the
T. matsutake filled with the mix standard gases, we needed to draw the air out of the bags and seal up these bags by a vacuum packaging machine (Deli, Ningbo, Zhejiang, China). A small hole was cut in the corner of each bag, which could only accommodate a gas supply pipe about 8 mm wide. Mixed standard gases bought from a standard gas factory in local (Linzhi, Tibet), were used to fill the fresh-keeping bags.
T. matsutake with different modified atmosphere packaging are shown in
Figure 1.
2.3. T. Matsutake Quality Indicators Measurement
In this paper, the data of sensory indicators and physical and chemical indicators of four groups of T. matsutake were obtained. Sensory indicators included the hardness (cap, stipe), color, and odor. Physical and chemical indicators included pH, soluble solids content, and moisture content. We measured the quality indicators of T. matsutake once every two days. Each time we took 2 pieces and tested them separately, and each indicator was measured three times for the mean value.
Sensory evaluation included the hardness (cap, stipe), color, and odor of fresh Tibetan
T. matsutake. A trained assessment team panel, including the local gatherer, acquirer, cook, and 2 consumers all of these 5 people who were invited to make the sensory evaluation, observed and filled a form to record the scores separately. In this study, the score of sensory evaluation was divided into four grades; the certain specific standards required to evaluate them are listed in
Table 2. In the data processing of sensory score, the final sensory score was the median of evaluation scores by the evaluation team. For example, the odor sensory scores of
T. matsutake were 3, 3, 2, 3, 2, and then the final odor sensory score was 3.
The pH and soluble solids content (SSC) test procedures were as follows: Put 5 g T. matsutake into a mortar that was surrounded with ice. Add 20 mL of phosphate buffer (PB) containing 1% polyvinyl pyrrolidone (PVP) with a pH of 6.8 and a concentration of 0.05 mol/L. The mixture was ground with a pestle in low temperatures, and then filtered the mixture with 4 layers of gauze. The pH value was accessed using a pHB-4 pH meter (INESA.CC, Shanghai, China) by measuring the filtered clear liquid. The content of SSC was gotten by measuring the filtered clear liquid using a glycometer refractometer (LH-B55, Hangzhou, Zhejiang, China).
Moisture content was measured using a moisture meter (Shenzhen crown and moisture meter technology co., LTD., Shenzhen, China) [
13]. The weight of each test piece model was 0.05 kg to 0.10 kg.
Correlation analysis was carried out between the data of factors and the remaining shelf life of T. matsutake. Experimental data and remaining shelf life data were normalized first, and then Microsoft Excel mathematics software was used for correlation analysis. We selected and recorded the correlation data between the remaining shelf life and various experimental indicators. The greater the correlation coefficient, the more significant relationship between these indexes and their remaining shelf life. The representative indexes were selected as the independent variables and the remaining shelf life was the dependent variable of the shelf life prediction models.
2.4. The Remaining Shelf Life Prediction Method—BP Neural Network
BP neural network is a kind of multilayer feed-forward neural network [
36], whose main characteristics are the forward propagation of signals and back-propagation of errors. In forward propagation, the input signal is processed layer by layer from the input layer to the hidden layer and then to the output layer. Neurons of each layer only affect the state of neurons of the next layer. If the desired output is not obtained in the output layer, the signal shifts to back-propagation, and the network weight and threshold are adjusted according to prediction error, so that the predicted output of the BP neural network gradually approaches its desired output [
37]. The topology of the BP neural network is shown in
Figure 2.
As shown in
Figure 2, a BP neural network consists of many neurons connected to each other and has the structure of input, hidden, and output layers. In this paper, the input layer parameters were the quality indicators of
T. matsutake, and the output layer was its remaining shelf life.
and
were weights of the BP neural network. The BP neural network is a multi-layer and feed-forward neural network based on an erroneous reverse transmission algorithm [
31,
38]. The learning rule of the BP neural network is to use a steepest descent algorithm to continuously adjust the weights and thresholds by the back-propagation network to obtain the minimum sum of squared errors of the network [
34]. The BP neural network has to train the network before prediction. The network has association and prediction ability through training. The process of training the network includes network initialization, output calculation of the hidden layer, output calculation of the output layer, error calculation, weight updating, threshold updating, and judgment of whether algorithm iteration has come to an end. It is necessary to go back to step 2 if the algorithm iteration has not come to an end [
39,
40]. Based on the BP neural network, signals are predicted in the following three steps: construction of the BP neural network, training of the BP neural network, and prediction of the BP neural network.
In general, the input and output layer parameters of the BP neural network have different specifications. In order to reduce the errors of the remaining shelf life prediction model, the input and output layer parameters were normalized according to Formula (1):
In Formula (1), is normalized data, p is original data, is the minimum of original data, and is the maximum of original data.
The number of neurons in the input and output layers of the BP neural network was determined by the input and output layer variables. The calculation of the optimal number of hidden layer nodes was affirmed as follows:
In Formula (2), l is the number of hidden layer nodes, n is the number of input layer nodes, m is the number of output layer nodes, and is a constant between 0 and 10.
In this study, the training parameters of the BP neural network were set as follows: The input layer function was logsig function. The output layer function was purein function. The training function was trainrp function. The learning function was learndm function. The learning rate was set as 0.001. The momentum factor was 0.01. The training error was 1 h. The maximum step size was 10,000 [
29]. Matlab R2016b mathematics software (version 9.1, MathWork Inc., Natick, MA, USA) was used to establish the models and analyze the date.
4. Conclusions
This paper aims to study the shelf life characteristics of fresh Tibetan T. matsutake in the MAP conditions under the temperature condition of 4 °C and relative humidity of 90% for improving the shelf life of fresh Tibetan T. matsutake and ensuring its quality and safety in shelf life. Correlation analysis was applied to optimize the quality characteristics obtained from the experiments for building the remaining shelf life predicting models of T. matsutake. A BP neural network was used to build these models in different MAP conditions to quantify the relationship between quality changes and preservation conditions during shelf life, which provided the research basis for transportation and preservation methods of fresh Tibetan T. matsutake in a cold chain.
The experimental results showed that the shelf life of fresh T. matsutake reduced with increasing oxygen concentration, when the O2 content was between 1% and 21% (the O2 content of air), and its shelf life could extend to 18 days under the MAP condition of 1% O2 + 21% CO2 + 78% N2. Sensory characteristics and physical and chemical characteristics were analyzed. Sensory evaluation scores of four kinds of indicators decreased with the extension of storage time. The odor indicator in sensory characteristics was more sensitive to the freshness of T. matsutake.
The changes of physical and chemical characteristics can be divided into three periods, S1, S2, and S3, respectively, to interpret the physiological changes of T. matsutake. The changes of pH dramatically increased in the beginning during period S1, lasting about two days. This was because the cap of T. matsutake would spread after harvest, like many mushrooms. The changes of SSC and MC decreased in the period S1, lasting about four days. In period S2, the changes of pH, SSC, and MC were stable. The pH and MC all slowly decreased, and the SSC maintained stable over time. In the period S3, the fluctuation of pH suggested a great deal more volatility, which also appeared in the changes of SSC and MC. With the preservation time flowing, individual quality differences of T. matsutake greatly increased.
The three sensory characteristics—color, odor, and stipe hardness—showed great significance with the remaining shelf life, thus becoming the variables of prediction models in this paper by the correlation analysis. By analyzing the BP neural network prediction model results, the prediction data was over the error range in the fourth day. This was because there were only four levels of sensory evaluation, which led to the same scores on several consecutive days in our experiments. However, there was the possibility of dramatic changes on postharvest characteristics and quality of T. matsutake from the second day to the fourth day because of its cap spread process, which showed in our analysis on the changes of physical and chemical characteristics. Future research should focus on the more specific changes of physical and chemical characteristics and even microbial characteristics of T. matsutake after harvest, and the time interval for measuring these indicators should be shortened to once or twice a day.