A method for predicting the degree of cold injury of bitter gourd

Abstract

This invention belongs to the field of post-harvest quality testing technology for agricultural products, specifically relating to a method for predicting the degree of chilling injury in bitter gourd. The method uses *Bitter Gourd 'Du Ruan'* as the research object, employing segmented pre-cooling treatment to measure indicators such as chilling injury index, respiration rate, weight loss rate, firmness, chlorophyll content, malondialdehyde content, and antioxidant enzyme activity during storage. This determines suitable pre-cooling conditions, explores the impact of different pre-cooling settings on its physiological quality and chilling injury development during storage, and constructs a bitter gourd chilling injury prediction model based on key indicators. The method for predicting the degree of chilling injury in bitter gourd provided by this invention offers a theoretical basis and technical support for optimizing post-harvest cold chain processes for bitter gourd, mitigating low-temperature damage, and extending shelf life.

Description

Technical Field

[0001] This invention belongs to the field of post-harvest quality testing technology for agricultural products, specifically relating to a method for predicting the degree of chilling injury in bitter gourd. Background Technology

[0002] Momordica charantia( Momordica charantia L. Bitter melon (Momordica charantia) is the fruit of an annual climbing herbaceous plant belonging to the genus Momordica in the Cucurbitaceae family. It is rich in various bioactive components, including essential amino acids, carotenoids, ascorbic acid, and various phenolic compounds, possessing antioxidant, anti-inflammatory, antibacterial, and immune-regulating effects. However, bitter melon is a climacteric fruit, meaning its physiological metabolism is vigorous at room temperature after harvest, resulting in a short shelf life. Problems such as softening of the flesh, yellowing of the peel, and rotting easily occur, severely impacting its commercial value and storage / transportation period.

[0003] To ensure the post-harvest quality of bitter gourd, pre-cooling treatment is performed. This involves rapidly cooling the fruit to a suitable storage temperature and removing field heat, inhibiting its vigorous respiration, delaying ripening and senescence, and inhibiting metabolic activity, thus effectively extending the storage and transportation life of bitter gourd. Studies have shown that effective pre-cooling treatment can significantly reduce the loss rate of fruits and vegetables throughout the entire cold chain by up to 20%. However, improper pre-cooling temperatures can also induce chilling injury. Bitter gourd is a typical cold-sensitive fruit and vegetable, highly sensitive to low temperatures. Under improper storage temperatures, it is prone to symptoms such as sunken spots, water-soaked spots, discoloration of the peel, dehydration, and tissue collapse and rotting. Therefore, appropriate storage temperature is a key factor affecting the storage quality of bitter gourd.

[0004] Zheng Cunna et al. published a paper titled "Study on the Influence of Different Storage Temperatures on the Storage Quality of Bitter Melon," which investigated chilling injury, physiological and biochemical changes, and the effects of chilling storage on bitter melon stored at low temperatures after harvest in Chunxiao. The results showed that chilling injury occurred on the 9th day after storage at 8℃, with respiration intensity and cell membrane permeability significantly increasing with the severity of chilling injury; bitter melon stored at 12℃ rotted on the 9th day, with respiration intensity and cell membrane permeability significantly increasing with the severity of rot; and bitter melon stored at 10℃ had a marketable rate of over 90% after 15 days. Therefore, this study indicates that 10℃ is the optimal storage temperature for bitter melon. Other studies have also shown that storing bitter melon at 10-12℃ can effectively delay senescence and postpone the onset of chilling injury, making this a relatively safe and effective storage temperature range for bitter melon.

[0005] Currently, common pre-cooling technologies for fruits and vegetables include cold storage pre-cooling, forced air pre-cooling, vacuum pre-cooling, chilled water pre-cooling, and ice water pre-cooling. Among these, cold storage pre-cooling has become the most widely used pre-cooling method in cold chain transportation due to its low equipment requirements and ease of operation. However, the impact of pre-cooling processes on the storage quality of cold-sensitive bitter gourd is still unclear, especially lacking in-depth research on its physiological metabolism and chilling injury patterns under different pre-cooling conditions. Therefore, studying and exploring the impact of pre-cooling processes on the physiological quality and chilling injury development of bitter gourd during storage is of great significance for optimizing post-harvest cold chain processes for bitter gourd, mitigating low-temperature damage, and extending shelf life. Summary of the Invention

[0006] To ensure the post-harvest quality of bitter gourd, this invention provides a method for predicting the degree of chilling injury. Using *Citrus durum-judae* as the research object, this invention employs a segmented pre-cooling treatment to measure indicators such as chilling injury index, respiration rate, weight loss rate, firmness, chlorophyll content, malondialdehyde content, and antioxidant enzyme activity during storage. This determines suitable pre-cooling conditions, explores the impact of different pre-cooling settings on its physiological quality and chilling injury development during storage, and constructs a chilling injury prediction model based on key indicators. This provides a theoretical basis and technical support for optimizing the post-harvest cold chain process for bitter gourd, mitigating low-temperature damage, and extending shelf life.

[0007] To achieve the above objectives, the present invention provides the following technical solution: This invention provides a method for predicting the degree of chilling injury in bitter gourd, comprising the following steps: Step S1: Sample pre-cooling treatment: Select fruits with similar shape and size, uniform color, and no deformities, overripeness, or cracks. Store qualified bitter gourd samples separately under no pre-cooling, rapid pre-cooling, three-stage pre-cooling, and two-stage pre-cooling conditions. Step S2, Sample index determination: On days 0, 4, 8, 12 and 16 of storage, the chilling injury index, respiration rate, weight percentage, hardness, chlorophyll, malondialdehyde (MDA), superoxide dismutase (SOD), peroxidase (POD) and catalase (CAT) of bitter gourd samples were measured. Step S3: Screening key predictive factors: Conduct correlation analysis and factor analysis on the cold damage-related indicators detected in step S2, extract three common factors F1, F2 and F3, and establish a common factor score calculation model. Step S4: Construct a cold damage prediction model: Based on the analysis in step S3, perform multiple linear regression analysis to construct a prediction model.

[0008] Furthermore, the different pre-cooling treatment methods in step S1 are grouped as follows: Control group (CK): Bitter melon samples were not pre-cooled and were directly placed in a storage environment set at 11±1 ℃; Rapid pre-cooling group (T1): Bitter gourd samples were placed in a pre-cooling environment of 4±1 ℃ for 4 hours, and then transferred to 11±1 ℃ for storage. Three-stage pre-cooling group (T2): The bitter gourd samples were first pre-cooled at 4±1 ℃ for 80 minutes; then transferred to 8±1 ℃ for another 80 minutes of pre-cooling; and finally transferred to 11±1 ℃ for storage. Two-stage pre-cooling group (T3): The bitter gourd samples were first pre-cooled at 6±1 ℃ for 2 hours, and then stored at 11±1 ℃.

[0009] Furthermore, in step S2, the chilling injury index is statistically graded based on the chilling injury symptoms of the bitter gourd samples. The chilling injury symptoms of the bitter gourd samples are graded as follows: Grade 0: The surface of the melon is smooth, with no signs of chilling injury, and the flesh is firm; Level 1: Slight chilling injury, with slight pitted spots or water-soaked spots on the peel covering ≤10% of the peel area, and no obvious changes in the flesh, which loses its luster; Level 2: Moderate chilling injury, with sunken spots or water-soaked spots on the peel covering 11% to 30% of the peel area, and the flesh beginning to soften, lose its luster, and begin to wrinkle; Level 3: Severe chilling injury, with sunken spots or water-soaked spots on the peel covering 31% to 60% of the peel area, the peel turning yellow, and the flesh becoming noticeably softer; Level 4: Severe chilling injury, with sunken spots or water-soaked spots on the peel covering more than 60% of the peel area, the peel color turning significantly yellow, and the appearance of tail rot, decay, or white mold.

[0010] Furthermore, the cold damage index is calculated using the following formula: .

[0011] Furthermore, the three common factors F1, F2, and F3 in step S3 are as follows: The first common factor F1 is determined by the cold injury index, respiration rate, weight loss rate, hardness, malondialdehyde content, and catalase activity. The second common factor F2 is determined by superoxide dismutase (SOD) activity and chlorophyll content. The third common factor F3 is determined by peroxidase (POD) activity.

[0012] Furthermore, the common factor score calculation model in step S3 is as follows: The common factor score F is calculated using the component score coefficient matrix B and the standardized vector of original variable observations X: F = B′ R - ¹ X, where R -1The inverse of the correlation coefficient matrix of the original variables is given by the following formula for calculating the score of the common factors: In the formula, i represents the total number of observed variables participating in the factor analysis; c j1 ,c j2 ,…,c ji Z1, Z2, ..., Zn are the coefficients in the component score coefficient matrix corresponding to the j-th common factor; p For the original observed variables X1, X2, ..., X p The value after standardization.

[0013] Furthermore, the formulas for calculating the scores of the common factors F1, F2, and F3 are as follows: F1=-0.197 Z1+0.204 Z2-0.182 Z3+0.182 Z4-0.052 Z5-0.197 Z6+0.021 Z7-0.059 Z8+0.190 Z9; F2=-0.071 Z1-0.396 Z2-0.138 Z3+0.031 Z4+0.383 Z5+0.077 Z6+0.505 Z7-0.101 Z8-0.029 Z9; F3=0.036 Z1-0.052 Z2+0.068 Z3+0.059 Z4+0.370 Z5+0.104 Z6-0.274 Z7+0.780 Z8+0.081 Z9; Among them, Z1 to Z9 represent the standardized chilling injury index, respiration rate, weight loss rate, hardness, chlorophyll content, malondialdehyde (MDA) content, superoxide dismutase (SOD) activity, peroxidase (POD) activity and catalase (CAT) activity, respectively.

[0014] Furthermore, in step S4, the multiple linear regression analysis uses the scores of each of the m common factors F1, F2, ..., F... m Using Y as the independent variable and the measured cold damage index as the dependent variable, a model is constructed. The mathematical form of the model is: In the formula, β 0 is a constant term. β 1, β 2,…, β m , where are the regression coefficients of each common factor.

[0015] Furthermore, the chilling injury index prediction model uses the measured chilling injury index (Y) as the dependent variable and respiration rate (X2), weight loss rate (X3), hardness (X4), chlorophyll content (X5), malondialdehyde (MDA) content (X6), superoxide dismutase (SOD) activity (X7), peroxidase (POD) activity (X8), and catalase (CAT) activity (X9) as initial independent variables, and employs generalized least squares stepwise multiple linear regression analysis.

[0016] Furthermore, the model ultimately identified four independent variables: weight loss rate (X3), malondialdehyde (MDA) content (X6), catalase (CAT) activity (X9), and superoxide dismutase (SOD) activity (X7); the regression equation for the cold injury index prediction model is: Y = -2.07 + 2.667·X3+ 3.183·X6– 0.717·X9+ 0.022·X7.

[0017] The specific steps of the method for predicting the degree of chilling injury in bitter gourd provided by this invention are as follows: Step S1, Sample pre-cooling treatment: (1) Raw material screening: Du Ruan bitter gourd fruits produced in orchards in Guangzhou, Guangdong Province, China, were selected with the same maturity and harvest period. After harvesting, the fruits were immediately transported to the laboratory for manual screening. Fruits with similar shape and size, uniform skin color, no mechanical damage, no deformity, no overripeness, and no cracks were selected as experimental samples.

[0018] (2) Group design: The qualified bitter melons were randomly and evenly divided into 4 groups, with the same number in each group, and each group corresponded to a different pre-cooling treatment method, as follows: Control group (CK): No pre-cooling treatment was performed; the samples were directly placed in a storage environment set at 11±1 ℃. Rapid pre-cooling group (T1): Bitter gourd was placed in a pre-cooling environment of 4±1 ℃ for 4 hours, and then transferred to storage at 11±1 ℃; Three-stage precooling group (T2): The bitter melon was processed in the following three stages in sequence: a) First stage: Pre-cool at 4±1 ℃ for 80 minutes; b) Second stage: Transfer to an environment of 8±1 ℃ and continue pre-cooling for 80 minutes; c) Third stage: Transfer to storage at 11±1 ℃; Two-stage pre-cooling group (T3): The bitter gourd was first pre-cooled at 6±1 ℃ for 2 hours, and then transferred to 11±1 ℃ for storage.

[0019] (3) Sampling and index determination: Physicochemical index tests were conducted on days 0, 4, 8, 12 and 16 of storage. Three fruits were randomly selected from each group each time, and external indexes (such as chilling injury index, respiration rate and hardness) were measured first. After the test, the seeds and pods of each fruit were removed, chopped and thoroughly mixed to make a mixed sample, which was immediately placed in an ultra-low temperature freezer at -80 ℃ for subsequent extraction and determination of physiological and biochemical indexes.

[0020] Step S2, Sample index determination: The aforementioned multiple indicators refer to the data collection on chilling injury index, respiration rate, weight percentage, firmness, chlorophyll, malondialdehyde (MDA), superoxide dismutase (SOD), peroxidase (POD), and catalase (CAT) of bitter gourd during storage.

[0021] (1) Chilling injury index: The fruits are graded according to the chilling injury symptoms. The chilling injury symptom grading of the bitter gourd samples is as follows: Grade 0: The surface of the melon is smooth, with no signs of chilling injury, and the flesh is firm; Level 1: Slight chilling injury, with slight pitted spots or water-soaked spots on the peel covering ≤10% of the peel area, and no obvious changes in the flesh, which loses its luster; Level 2: Moderate chilling injury, with sunken spots or water-soaked spots on the peel covering 11% to 30% of the peel area, and the flesh beginning to soften, lose its luster, and begin to wrinkle; Level 3: Severe chilling injury, with sunken spots or water-soaked spots on the peel covering 31% to 60% of the peel area, the peel turning yellow, and the flesh becoming noticeably softer; Level 4: Severe chilling injury, with sunken spots or water-soaked spots on the peel covering more than 60% of the peel area, the peel color turning significantly yellow, and the appearance of tail rot, decay, or white mold.

[0022] The cold damage index is expressed by the following formula: .

[0023] (2) Respiration rate: Take one fruit, weigh it, place it in a sealed container, record the CO2 production over 1 hour, record the temperature before and after, and measure 3 parallel samples. Calculate the respiration intensity using the following formula: Where: a: carbon dioxide concentration after measurement (ppm); 44 / 22.4: conversion factor from ppm to mg / m³ under standard conditions; a0: initial carbon dioxide concentration, ppm; T: temperature after measurement, °C; T0: temperature before measurement, °C; V: total volume of container, dm³ 3 t: Time taken for measurement, h; m: Mass of fruits and vegetables used for measurement, kg.

[0024] (3) Weight loss rate: Weigh the sample before and after storage using an electronic balance and calculate the weight loss rate using the following formula: In the formula: A is the weight of the bitter melon before storage, in g; B is the weight of the bitter melon after storage, in g.

[0025] (4) Hardness: Three fruits were randomly selected, and the hardness was measured at three points at equal intervals around the equator of the fruit. The texture analyzer program used TPA (Texture Profile Analysis) to simulate the human chewing process. The probe type was a P / 5N cylindrical probe, and the parameters were set as follows: force sensing range 500 N, trigger force 0.1 N, deformation 15%, and probe pre-measurement speed 60 mm / min.

[0026] (5) Chlorophyll content: The absorbance was measured at 663 nm and 645 nm using a UV-Vis spectrophotometer via ethanol extraction, and the calculation formula is as follows: In the formula: A 645 The absorbance of the sample at 645 nm was extracted; A 663 The absorbance of the sample at 663 nm was extracted; W is the fresh weight of the sample, kg; N is the dilution factor.

[0027] (6) Malondialdehyde (MDA) content: The absorbance of the reaction product at 532 nm was determined using the thiobarbituric acid (TBA) method. Where: V: volume of MDA extract, L; W: fresh weight of plant tissue, g; A 532 The absorbance of the sample at 532 nm; A 600 =Absorbance of the sample at 600 nm; A 450=Absorbance of the sample at 450 nm.

[0028] (7) Related antioxidant enzyme activities: a) Superoxide dismutase (SOD) activity: SOD activity can be calculated by measuring the change in the reactant at 450 nm using a water-soluble tetrazolium salt (WST-1) colorimetric method. The formula is as follows: In the formula: A1: absorbance of the measuring orifice; A 1-0 A1: Absorbance of the blank well; A2: Absorbance of the control well; 2-0 : Absorbance of the control blank well; 12: Reaction volume / dilution factor, 0.24 mL / 0.02 mL; N: Dilution factor before testing; W: Weight of tissue sample, g; V 样 : Total volume of buffer solution added during sample homogenization, mL.

[0029] b) Peroxidase (POD) activity: The POD activity can be calculated by measuring the change in the reactant at 470 nm using the guaiacol method. The formula is as follows: Where: A1 is the absorbance value of the sample measured after 3 min of incubation; A0 is the absorbance value of the sample measured after adding the working solution; W: weight of the tissue sample, mg; V T V: Total volume of enzyme extraction solution, mL; S : Volume of enzyme solution used in the assay, mL; t: Reaction time, min.

[0030] c) Catalase (CAT) activity: The activity of CAT can be calculated by measuring the change in the reactant at 405 nm using the ammonium molybdate method. The formula is as follows: In the formula: V 样 : Sample volume in this system, 0.1 mL; T: Reaction time, 60 s; W: Tissue sample mass, g; V 样总 : Total volume of the sample homogenate, mL.

[0031] Step S3: Screening key predictive factors: Correlation and factor analysis were performed on the chilling injury related indicators detected in step S2: chilling injury index (X1), respiration rate (X2), weight loss rate (X3), hardness (X4), chlorophyll (X5), MDA content (X6), SOD activity (X7), POD activity (X8) and CAT activity (X9). Three common factors F1, F2 and F3 were extracted, and a common factor score calculation model was established.

[0032] The key predictor screening refers to: (1) Screening for significant differences: Using SPSS software, Waller-Duncan test was performed on all indicators measured in step two. With p<0.05 as the significance threshold, the indicators that showed statistically significant differences between different treatment groups were obtained, forming the primary screening indicator set. (2) Correlation strength screening: Spearman's rank correlation coefficient analysis method is used to calculate the correlation between each index and the cold damage index for the primary screening index set. p<0.05 is used as the significance standard, and the absolute value of the correlation coefficient |r| ≥ 0.7 is used as the strong correlation standard. The indexes that are strongly correlated with the cold damage index and are significant are screened out to form the core correlation index set. (3) Principal Component Factor Dimension Reduction: Exploratory factor analysis was performed on the core related index set, and principal component analysis was used for extraction. First, KMO and Bartlett tests were performed. The KMO sampling appropriateness measure was required to be ≥ 0.6 and the Bartlett sphericity test significance p < 0.001 before principal component analysis could be performed. The rotation method adopted was the Kaiser normalized maximum variance method. Principal component factors were extracted based on the eigenvalue greater than 1 and the cumulative variance contribution rate ≥ 80%. According to the rotated component matrix, the index variables with an absolute value of loading greater than 0.7 on each principal component factor were selected and defined as the final key predictive factors. Common factor score calculation: Based on the factor analysis results, a calculation model for standardized observed variables and common factor scores is established. The common factor score F is calculated using the component score coefficient matrix B and the standardized vector of original variable observations X: F = B′ R - ¹ X, where R -1 The inverse of the correlation coefficient matrix of the original variables is given by the following formula for calculating the score of the common factors: In the formula, i represents the total number of observed variables participating in the factor analysis; c j1 ,c j2 ,…,c ji Z1, Z2, ..., Zn are the coefficients in the component score coefficient matrix corresponding to the j-th common factor; p For the original observed variables X1, X2, ..., X p The value after standardization.

[0033] In one specific embodiment, principal component analysis extracted three common factors, whose cumulative variance contribution rate reached 86.23%. The common factors in the rotated component matrix mainly consist of: the first common factor (F1): mainly determined by X1, X2, X3, X4, X6, and X9, with loadings of -0.954, 0.732, -0.902, 0.916, -0.84, and 0.937 on the principal factors, respectively. F1 is a comprehensive factor; a higher score indicates milder chilling injury symptoms, lower weight loss, better texture retention, stronger CAT activity, and lower MDA content. It is also important to note the relatively high respiration rate. In the early stages of fruit and vegetable storage, before the rapid decline in respiration, a higher respiration rate is observed, reflecting a covariation pattern of "high respiration - low chilling injury."

[0034] The second common factor (F2) is determined by X5 and X7, with loadings of 0.705 and 0.838 on the principal factors, respectively. Higher scores indicate stronger superoxide scavenging ability and better color retention of the fruit during storage.

[0035] The third common factor (F3) was mainly determined by peroxidase (POD) activity (0.927), revealing the main enzymatic scavenging pathway of peroxides in fruits.

[0036] Based on the component score coefficient matrix obtained from factor analysis, the calculation formulas for the three common factors are established as follows: F1=-0.197 Z1+0.204 Z2-0.182 Z3+0.182 Z4-0.052 Z5-0.197 Z6+0.021 Z7-0.059 Z8+0.190 Z9; F2=-0.071 Z1-0.396 Z2-0.138 Z3+0.031 Z4+0.383 Z5+0.077 Z6+0.505 Z7-0.101 Z8-0.029 Z9; F3=0.036 Z1-0.052 Z2+0.068 Z3+0.059 Z4+0.370 Z5+0.104 Z6-0.274 Z7+0.780 Z8+0.081 Z9; Among them, Z1 to Z9 represent the standardized chilling injury index, respiration rate, weight loss rate, hardness, chlorophyll content, malondialdehyde (MDA) content, superoxide dismutase (SOD) activity, peroxidase (POD) activity and catalase (CAT) activity, respectively.

[0037] Step S4: Construct a cold damage prediction model: (1) Based on the correlation analysis and factor analysis in step S3, SPSS statistical analysis was used to calculate the scores of each common factor, F1, F2, ..., Fm. m Using Y as the independent variable and the measured cold damage index as the dependent variable, a prediction model is constructed using multiple linear regression analysis. The mathematical form of the resulting model is as follows: In the formula, β 0 is a constant term. β 1, β 2,…, β m , where are the regression coefficients of each common factor.

[0038] (2) In one specific embodiment, the measured chilling injury index (Y) was used as the dependent variable, and respiration rate (X2), weight loss rate (X3), stiffness (X4), chlorophyll content (X5), malondialdehyde (MDA) content (X6), superoxide dismutase (SOD) activity (X7), peroxidase (POD) activity (X8), and catalase (CAT) activity (X9) were used as initial independent variables. A stepwise multiple linear regression analysis was performed using the generalized least squares method. After stepwise regression screening, the final model retained four independent variables: weight loss rate (X3), malondialdehyde (MDA) content (X6), catalase (CAT) activity (X9), and superoxide dismutase (SOD) activity (X7). The resulting regression equation for the chilling injury index prediction model is: Y = -2.07 + 2.667·X3+ 3.183·X6– 0.717·X9+ 0.022·X7.

[0039] The method for predicting the degree of chilling injury in bitter gourd provided by this invention applies a segmented precooling procedure to the postharvest treatment of bitter gourd. The results show that the three-stage precooling treatment (T2) exhibits the best effect, significantly reducing the chilling injury index by 25.63% compared with the control group. This indicates that the segmented heating, through stepwise temperature stimulation, effectively activates the antioxidant defense system of bitter gourd while avoiding irreversible damage that may be caused by a single low temperature. Based on various physiological indicators, the theoretical mechanism was transformed into an applied model. With chilling injury index as the dependent variable and the other eight indicators as independent variables, a chilling injury prediction model was established using stepwise multiple linear regression analysis with generalized least squares method. The chilling injury prediction equation was obtained as Y = -2.07 + 2.667·X3 + 3.183·X6 – 0.717·X9 + 0.022·X7. This model constructs a chilling injury prediction index system covering physical characteristics, texture attributes and physiological activities from four dimensions: weight loss rate (X3), MDA (X6), CAT activity (X9) and SOD activity (X7). This provides a basis for early diagnosis of chilling injury risk during bitter gourd storage.

[0040] Compared with existing technologies, the method for predicting the degree of chilling injury in bitter gourd provided by this invention has the advantages of simple operation and strong practicality. It only requires the measurement of four easily obtainable indicators to accurately assess the risk of chilling injury, which provides convenience for industrial applications and realizes real-time quality monitoring and early warning in the storage process of bitter gourd. Attached Figure Description

[0041] Figure 1 This is a diagram showing the morphological changes of bitter melon during storage.

[0042] Figure 2 Correlation heatmap of bitter melon indicators.

[0043] Figure 3 This is a scatter plot used to validate the cold damage index prediction model. Detailed Implementation

[0044] The technical solutions in the embodiments of the present invention will be clearly and completely described below. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0045] Example 1: Establishment of a chilling injury prediction model for post-harvest pre-cooled bitter gourd during storage. 1. Materials and Methods: 1.1 Pre-cooling treatment of bitter melon: The bitter melon fruits of Duruan were harvested from local orchards in Guangzhou, Guangdong Province, China. Fruits of similar shape and size, uniform color, and free from deformities, overripeness, and cracks were selected for pre-cooling storage experiments. The bitter melons were randomly divided into four groups: no pre-cooling group (CK), where bitter melons were directly stored at 11±1 ℃; rapid pre-cooling group (T1), where bitter melons were pre-cooled at 4±1 ℃ for 4 hours, then transferred to 11±1 ℃ for storage; three-stage pre-cooling group (T2), where bitter melons were pre-cooled at 4±1 ℃ for 80 minutes, then transferred to 8±1 ℃ for 80 minutes, and finally transferred to 11±1 ℃ for storage; and two-stage pre-cooling group (T3), where bitter melons were pre-cooled at 6±1 ℃ for 2 hours, then transferred to 11±1 ℃ for storage.

[0046] On days 0, 4, 8, 12 and 16 of storage, physical indicators such as chilling injury index, respiration rate and hardness of bitter gourd were measured. Three fruits were randomly selected from each group, and after removing the seeds and pods, they were chopped, mixed and stored in a -80℃ refrigerator for subsequent physiological and biochemical index determination.

[0047] 1.2. Chilling damage index: The chilling injury index of bitter melon was determined according to the method of Lin (Lin, X., Wang, L., Hou, Y., Zheng, Y., & Jin, P.(2020). A Combination of Melatonin and Ethanol Treatment Improves Postharvest Quality in Bitter Melon Fruit. Foods, 9(10), 1376.). The fruit was graded according to the chilling injury symptoms, as described in Table 1. The chilling injury index was expressed by the following formula: Table 1. Classification of Symptoms of Chilling Injury in Bitter Melon 1.3 Respiratory rate: Respiration rate was measured using an H3055 grain and vegetable respiration meter (Zhejiang Top Cloud Agriculture Technology Co., Ltd.). One fruit was taken, weighed, and placed in a sealed container. CO2 production was recorded for 1 hour, along with the initial and final temperatures. Three parallel samples were measured. Respiration intensity was calculated using the following formula: a: Carbon dioxide concentration after measurement (ppm); 44 / 22.4: Conversion factor from ppm to mg / m³ under standard conditions; a0: Initial carbon dioxide concentration, ppm; T: Temperature after measurement, °C; T0: Temperature before measurement, °C; V: Total volume of container, dm³ 3t: Time taken for measurement, h; m: Mass of fruits and vegetables used for measurement, kg.

[0048] 1.4 Weightlessness rate: The weight of the samples before and after storage was measured using an LT10K electronic balance (Changshu Tianliang Instrument Co., Ltd.), and the weight loss rate was calculated using the following formula: A represents the weight of the bitter melon before storage, in grams; B represents the weight of the bitter melon after storage, in grams.

[0049] 1.5 Hardness Measurement: Three fruits were randomly selected, and their firmness was measured at three points at equal intervals around the equator. The fruits were pressed at a uniform speed using the probe of an FTCTMS-PILOT texture analyzer (Beijing Yingsheng Hengtai Technology Co., Ltd.). The measurement data were recorded, and the average value and standard deviation were calculated. The program used was TPA (Texture Profile Analysis) to simulate the human chewing process. A P / 5N cylindrical probe was used, with the following parameters set: force sensing range 500 N, trigger force 0.1 N, deformation 15%, and probe pre-measurement speed 60 mm / min.

[0050] 1.6 Chlorophyll content: Chlorophyll content was determined according to the method described by Wang Rui (Wang Rui, Ba Liangjie. Experimental Food Preservation Technology [M]. Beijing: China Light Industry Press, 2019.6). The absorbance was measured at 663 nm and 645 nm using an ethanol extraction method and a TU-1900 double-beam UV-Vis spectrophotometer (Beijing Purkinje General Instrument Co., Ltd.). The calculation formula is as follows: A 645 The absorbance of the sample at 645 nm was extracted; A 663 The absorbance of the sample at 663 nm was extracted; W is the fresh weight of the sample, kg; N is the dilution factor.

[0051] 1.7 Determination of malondialdehyde (MDA) content: Take 0.2 g of bitter melon homogenate tissue, add 2 mL of tissue homogenate, centrifuge at 4000 g for 10 min, and collect the supernatant for later use. Add 200 μL of tissue homogenate to a 2 mL centrifuge tube as a blank control, add 200 μL of MDA extraction solution for determination, and then add 200 μL of TBA working solution. Refer to the detection reaction system in the instructions of the Plant Malondialdehyde (MDA) Detection Kit (TBA Colorimetric Method) (Shanghai Yuanye Biotechnology Co., Ltd.) to add reagents sequentially. Mix well, cover, boil in a 95 ℃ water bath for 30 min, then remove and cool in a cold water bath to room temperature. Measure the absorbance at 532 nm using a Tecan Infinite Array full-wavelength multi-functional microplate analyzer (Tecan (Shanghai) Experimental Equipment Co., Ltd.). The MDA content calculation formula is as follows: V: Volume of MDA extract, L; W: Fresh weight of plant tissue, g; A 532 The absorbance of the sample at 532 nm; A 600 =Absorbance of the sample at 600 nm; A 450 =Absorbance of the sample at 450 nm.

[0052] 1.8 Determination of the activity of relevant antioxidant enzymes: Superoxide dismutase (SOD) enzyme activity: Take 0.2 g of bitter melon sample, add 1.8 mL of phosphate buffer (phosphate buffer: 0.1 mol / L pH 7-7.4), vortex to mix for 3 min, centrifuge at 5000 g for 10 min, collect the supernatant, and adjust the concentration (2-3 times dilution) according to the SOD inhibition rate. Add the reaction solution to the microplate sequentially, following the operating table in the instruction manual of the Superoxide Dismutase (SOD) Detection Kit (WST-1 method) from Nanjing Jiancheng Biotechnology Co., Ltd., vortex to mix, incubate at 37 ℃ for 20 min, and read the value at 450 nm using a microplate reader. The specific formula for calculating SOD activity is as follows: A1: Absorbance of the measuring well; A 1-0 A1: Absorbance of the blank well; A2: Absorbance of the control well; 2-0 : Absorbance of the control blank well; 12: Reaction volume / dilution factor, 0.24 mL / 0.02 mL; N: Dilution factor before testing; W: Weight of tissue sample, g; V 样 : Total volume of buffer solution added during sample homogenization, mL.

[0053] Peroxidase (POD) activity: Add 0.2 g of bitter melon sample to 2 mL of pre-cooled POD Lysis Buffer homogenate, centrifuge at 10000 g for 15 min at 4℃, and collect the supernatant, which is the crude POD extract. Add the reaction solution sequentially to a 96-well plate according to the reaction system table in the instructions of the Plant Peroxidase (POD) Detection Kit (Guaifenesin Microplate Method) from Shanghai Yuanye Biotechnology Co., Ltd. Immediately use a microplate reader to measure the absorbance A0 of the test solution at 470 nm. After incubation at 37℃ for 3 min, measure the absorbance A1. The specific formula for calculating POD activity is as follows: A1 is the absorbance value of the sample measured after 3 min of incubation; A0 is the absorbance value of the sample measured after adding the working solution; W: weight of the tissue sample, g; V T V: Total volume of enzyme extraction solution, mL; S : Volume of enzyme solution used in the assay, mL; t: Reaction time, min.

[0054] Catalase (CAT) activity: Take 0.4 g of bitter melon sample, add 1.2 mL of 0.1 mol / L phosphate buffer (pH 7-7.4), vortex for 1 min, centrifuge at 5000 g for 10 min, and collect the supernatant. Add the reaction solution to the centrifuge tube and mix well. Measure the absorbance at a wavelength of 405 nm. The specific formula for calculating CAT activity is as follows: V 样 : Sample volume in this system, 0.1 mL; T: Reaction time, 60 s; W: Tissue sample mass, g; V 样总 : Total volume of the sample homogenate, mL.

[0055] 1.9 Statistical Analysis: All indicators were tested in triplicate. Experimental data were processed and statistically analyzed using Excel 2022, and results are expressed as mean ± standard deviation. SPSS 27.0.1 software was used for significance, correlation, principal component analysis, factor analysis, and regression analysis. The Waller-Duncan test was used for significance analysis (p<0.05 indicates significant difference), and Spearman correlation coefficient was used for correlation analysis (p<0.05, p<0.01). Principal component analysis was used for factor extraction, and the Kaiser normalized maximum variance method was used for rotation, selecting variables with common factors for interpretation based on the variable loadings in the rotated component matrix. Multiple stepwise linear regression analysis was used for regression analysis.

[0056] 2. Results and Analysis: 2.1 Analysis of quality change indicators of bitter melon: 2.1.1 Changes in appearance and chilling injury index of bitter melon under different pre-cooling treatments: Figure 1 This diagram illustrates the morphological changes of melons during storage. Figure 1 It is evident that the appearance quality of bitter gourd fruits varied significantly with prolonged storage. From day 12 onwards, the differences in appearance color among the groups became apparent. By day 16, the CK, T1, and T3 treatments had all lost their original luster, becoming dull and exhibiting various symptoms. The CK group showed white mycelium on the surface, and the rotten area expanded. The T1 group showed obvious tail rot lesions accompanied by scorched marks. The T3 group showed significant yellowing, with slight white mycelium and obvious scorched marks. In contrast, the T2 group fruits remained bright green and plump, without lesions or mycelial infection, and their overall appearance was significantly better than the other treatments. This indicates that the gradient pre-cooling treatment used in T2 can delay color deterioration, maintain the commercial appearance of the bitter gourd, and validates its optimal storage effect under the experimental conditions.

[0057] The chilling injury index is an important indicator for evaluating the effectiveness of pre-cooling storage. Table 2 shows that throughout the storage period, the chilling injury index of bitter gourd in all treatment groups gradually increased with the extension of storage time, but the effects of different pre-cooling treatments on delaying the development of chilling injury varied significantly. From day 4, the chilling injury index of the control group (CK) reached 11.1%, indicating that it had entered the first stage of chilling injury, with slight dents or water-soaked spots appearing on the peel, accompanied by a decrease in gloss. In stark contrast, the chilling injury index of the T2 treatment group was only 5.3%, significantly lower than that of CK and other treatment groups. By day 8 of storage, the chilling injury index of the CK group was close to the upper limit of the second stage of chilling injury, meaning that the area of ​​spots on the peel expanded, and the flesh began to soften and wrinkle, while the chilling injury index of the T2 group was 35% lower than that of the CK group. In the middle and late stages of storage (days 12-16), the chilling injury in the CK group further intensified, while all treatment groups showed varying degrees of inhibition, with the T2 group consistently showing the most significant advantage. On day 16, the chilling injury indices of the T1, T2 and T3 treatment groups decreased by 6.77%, 25.54% and 10.15% respectively compared with the CK group. Among them, the T2 group showed the greatest reduction, and its value indicated that its chilling injury level may still be in the level 2 range, effectively delaying the process to level 3 severe chilling injury and avoiding obvious yellowing of the peel and severe softening of the flesh.

[0058] Table 2 Statistics on the Chilling Injury Index of Bitter Melon Note: A~D in the same row are significance markers for different groups at the same time; a~e in the same column are significance markers for the same group at different time (p<0.05).

[0059] 2.1.2 Changes in weight loss, firmness, and chlorophyll content of bitter melon under different pre-cooling treatments: During low-temperature storage, water transpiration in bitter melon led to fruit weight loss, with the weight loss rate increasing in all groups. Low temperature inhibited water loss, and throughout the entire period, the weight loss of each pre-treated fruit was consistently lower than that of the control group (CK). In the early stage of storage (4 days), the water loss rate of the control group was 1.6, 1.5, and 1.75 times that of each treatment group, respectively, but there was no significant difference in weight loss rate among the groups. From day 12 onwards, the weight loss rate of all groups except T2 increased significantly (p<0.05). By the end of storage (16 days), the weight loss rate of the T2 group was 33.9% lower than that of the CK group, 21.6% lower than that of the T1 group, and 30.1% lower than that of the T3 group. The T2 treatment had a sustained and optimal effect on inhibiting water loss and reducing weight loss in bitter melon throughout the entire storage period.

[0060] Firmness is a key quality indicator for measuring fruit softening and spoilage. Its decrease is mainly due to the combined effects of multiple factors, including water loss, respiratory metabolic consumption, and pectin degradation. Table 3 shows that the firmness of bitter melon in all groups decreased during storage, while the firmness of pre-cooled bitter melon was higher than that of the control (CK) group. The rate of decrease was faster in the mid-storage stage (8-12 days), while the rate of decrease slowed in the late storage stage (12-16 days), with the T2 group maintaining a clear advantage in firmness. By day 16, there were no significant differences among the CK, T1, and T3 groups, except for T2, which was significantly higher than the other groups, remaining above 50 N (p<0.05).

[0061] Chlorophyll content is a key indicator reflecting the fruit's ability to retain green color and its senescence process. Table 3 shows that the overall chlorophyll content of bitter gourd decreased during storage, but the groups treated with different pre-cooling methods exhibited significant retention effects, especially the T2 treatment, which showed the most significant advantage in the later stages of storage. In the early to mid-stages of storage (days 4-8), the chlorophyll content of all treatment groups remained at a high level, even exceeding the initial value at some points, possibly related to post-harvest physiological regulation or measurement fluctuations. From day 12 onwards, the chlorophyll content of the CK group continued to decline, indicating accelerated green fading and the fruit entering the senescence stage; although the T1 and T3 groups performed well in the early stages, they also began to show significant degradation at this point; the T2 group maintained a chlorophyll content of 254.5 mg·kg⁻¹. - ¹The highest level. From the end of storage to 16 days, the T2 group was still higher than other groups, 10.1%, 6.1%, and 5.3% higher than the control group, T1 group, and T3 group, respectively.

[0062] Table 3. Quality Change Indicators of Bitter Melon under Different Pre-cooling Treatments Note: A~D in the same row are significance markers for different groups at the same time; a~e in the same column are significance markers for the same group at different time (p<0.05).

[0063] 2.1.3 Changes in respiration rate and MDA content of bitter melon under different pre-cooling treatments: Respiration rate is an important indicator of the metabolic activity of fruits and vegetables. Table 4 shows that pre-cooling treatment significantly inhibited the respiration intensity of bitter melon during storage. On day 4 of storage, the respiration rates of groups T1, T2, and T3 decreased by 42.48%, 36.46%, and 24.67% respectively compared to the control group (CK), with group T1 showing the most significant inhibitory effect at this time. As storage progressed into the middle and late stages (days 8-16), the effect of treatment T2 gradually became apparent and remained superior to the other groups, maintaining the lowest respiration rate throughout. By the end of storage (day 16), the respiration rates of groups T1, T2, and T3 decreased by 41.61%, 48.03%, and 31.22% respectively compared to the control group, with group T2 showing the most significant inhibitory effect, indicating that treatment T2 can more effectively delay fruit metabolic senescence during long-term storage.

[0064] MDA, as a stable end product of lipid peroxidation, is a key indicator for measuring the degree of oxidative damage to biomembranes. MDA levels generally increased in all groups during storage, but pre-cooling significantly mitigated this process. Compared to the control group, T1 and T2 treatments significantly inhibited MDA accumulation at multiple time points (days 4, 12, and 16) (p<0.05). The T2 group maintained a consistently low MDA level throughout storage, especially at the end of storage, where the MDA content in the control group was 1.28 times that of the T2 group. This indicates that the T2 treatment has a significant advantage in reducing membrane lipid peroxidation and maintaining cell membrane structural integrity.

[0065] Table 4. Respiration rate and MDA content of bitter melon Note: A~D in the same row are significance markers for different groups at the same time; a~e in the same column are significance markers for the same group at different time (p<0.05).

[0066] 2.1.4 Changes in the activity of antioxidant enzymes in bitter melon under different pre-cooling treatments: When plants are subjected to abiotic stress, the activity of antioxidant enzymes changes accordingly to cope with cell membrane lipid peroxidation, thereby enhancing the plant's own stress resistance. Superoxide dismutase (SOD) is the first line of defense in the plant's antioxidant system, enabling superoxide to undergo dismutation reactions to generate O2 and H2O2. As shown in Table 5, during storage, SOD activity generally showed a trend of first increasing and then decreasing, and the treatment groups were generally higher than the CK group, indicating that the pre-cooling treatment improved the fruit's early adaptation to the environment, which to some extent enhanced the maintenance of SOD activity in bitter gourd. In the early stage of storage (4 days), the SOD activity of the T2 group (299.38 U / g) was significantly higher than that of other groups. After 16 days of storage, the SOD activity of the T2 group was 57.44% higher than that of the CK group, and the SOD activity continued to increase in the middle and late stages. This indicates that the T2 group can maintain SOD activity, which has a positive effect on maintaining a good appearance of bitter gourd during storage.

[0067] POD is one of the key antioxidant enzymes in fruits and vegetables under adverse environments. It plays a role in regulating plant stress resistance by regulating the balance of reactive oxygen species (ROS) metabolism and scavenging H2O2 produced during metabolism, thus enhancing the plant's ability to respond to environmental stress and inhibiting fruit senescence. Table 5 shows that the POD activity of each group changed significantly. This dramatic change demonstrates that the POD response is the most active during ROS metabolism, playing a major role in scavenging H2O2 in response to stress. During storage, each group showed different trends of increase or decrease. Specifically, the POD activity of the CK group decreased continuously from an initial 22.7 U / mg to 10.57 U / mg on day 12. Although it slightly rebounded to 18.51 U / mg at the end, it remained at a low level, indicating that its antioxidant system was not effectively activated. The activity of the T1 group increased sharply to 30.65 U / mg on day 8 and then quickly dropped to 11.94 U / mg, exhibiting a dynamically unstable defense state, suggesting that this treatment may lead to oxidative metabolic imbalance and excessive energy consumption. The T2 and T3 groups showed more stable antioxidant responses. The activity of group T2 remained between 21.07 and 20.14 U / mg during the mid-to-late storage period, with gentle fluctuations; while in group T3, the activity increased to 26.17 U / mg during the mid-to-late storage period (day 12) and remained at a high level (16.53 U / mg at the end). This indicates that both treatments can effectively activate the defense system, maintain oxidative metabolic homeostasis, thereby delaying aging and improving storage quality.

[0068] CAT is a key enzyme in plants that specifically scavenges H2O2 and degrades it into H2O and O2. As shown in Table 5, the CAT activity of all experimental groups decreased with prolonged storage time. However, the treatment groups showed different effects in delaying the decline in CAT activity. Throughout the storage period, the CAT activities of the T1 and T2 groups were consistently significantly higher than those of the control group (p<0.05), with the T2 group maintaining high levels of enzyme activity on days 8 (4.74 U / g) and 12 (3.7 U / g). Towards the end of storage (day 16), the CAT activities of the T1 group (2.57 U / g) and the T2 group (2.39 U / g) were still significantly higher than those of the CK group (1.26 U / g) and the T3 group (1.36 U / g), while the protective effect of the T3 group significantly weakened in the later stages of storage. The results indicate that the T1 and T2 treatments can effectively maintain the CAT activity of bitter gourd fruit and enhance its ability to scavenge hydrogen peroxide, thus potentially better mitigating oxidative damage.

[0069] Table 5. Enzyme activities related to chilling injury during bitter melon storage. Note: A~D in the same row are significance markers for different groups at the same time; a~e in the same column are significance markers for the same group at different time (p<0.05).

[0070] 2.2 Establishment of the statistical model: 2.2.1 Correlation analysis and factor analysis: Correlation analysis was performed on chilling injury-related indicators, such as chilling injury index (X1), respiration rate (X2), weight loss rate (X3), hardness (X4), chlorophyll (X5), MDA content (X6), SOD activity (X7), POD activity (X8) and CAT activity (X9). Figure 2 This is a heatmap showing the correlation between bitter melon indicators. (Example:) Figure 2As shown, the chilling injury index was significantly positively correlated with weight loss rate and MDA content (p<0.01), and significantly negatively correlated with respiration rate, stiffness, chlorophyll, and CAT activity (p<0.01), but not with SOD activity and POD activity. Respiration rate was significantly positively correlated with stiffness and CAT activity (p<0.01), and significantly negatively correlated with weight loss rate, MDA content, and SOD activity (p<0.01), and negatively correlated with chlorophyll (p<0.05). Weight loss rate was significantly positively correlated with MDA (p<0.01), and significantly negatively correlated with stiffness, chlorophyll, and CAT activity (p<0.01). Stiffness was significantly positively correlated with chlorophyll and CAT (p<0.01), positively correlated with POD (p<0.05), and significantly negatively correlated with MDA (p<0.01). Chlorophyll was positively correlated with SOD and POD (p<0.05), and significantly positively correlated with CAT (p<0.01). MDA content was significantly negatively correlated with CAT activity (p<0.01). POD was positively correlated with CAT (p<0.05). The main variables were highly correlated, therefore principal component analysis was necessary.

[0071] First, the suitability of the data for factor analysis was assessed using the KMO and Bartlett tests. The results are shown in Table 6. The KMO sampling suitability score for the bitter melon data was 0.732, and the Bartlett sphericity test showed a significance level of p < 0.001, indicating that this data set is suitable for factor analysis. The factor analysis results for the nine indicators of bitter melon are shown in Table 7. Initial factors were extracted using principal component analysis. Based on the Caesar criterion (eigenvalues ​​greater than 1), three common factors (F1, F2, and F3) were extracted, with a cumulative variance contribution rate of 86.23%, indicating that these three factors can explain most of the information in the original variables and have high representativeness.

[0072] To obtain a clearer factor structure, the initial factor loading matrix was orthogonally rotated using the Kaiser normalized maximum variance method. The rotated component matrix is ​​shown in Table 8. The first principal factor (F1) is mainly determined by X1, X2, X3, X4, X6, and X9, with loadings of -0.954, 0.732, -0.902, 0.916, -0.84, and 0.937, respectively. F1 is a comprehensive factor; a higher score indicates milder chilling injury symptoms, lower weight loss, better texture retention, stronger CAT activity, and lower MDA content. It is also worth noting that the respiration rate is relatively high. During the early stages of fruit and vegetable storage, before the rapid decline in respiration, a higher respiration rate is observed, reflecting a covariation pattern of "high respiration - low chilling injury." The second principal factor (F2) was determined by X5 and X7, with loadings of 0.705 and 0.838, respectively. Higher scores indicated stronger superoxide scavenging ability and better color retention in the fruit during storage. The third principal factor (F3) was determined by POD activity (0.927), revealing the main enzymatic scavenging pathway of peroxides in the fruit.

[0073] Based on the factor analysis results, the rotated factor model is expressed as a linear combination of common factors, with component score coefficients shown in Table 9. The factor scores of the observable variables are calculated using the component score coefficients and the standardized original variables, with the formula: F = B′ R - ¹ X, where B is the component score coefficient matrix, R - ¹ represents the inverse of the correlation coefficient matrix of the original variables, and X is the standardized vector of observed values ​​of the original variables. Based on the component score coefficients in Table 9, the scores of the three common factors are calculated as follows: F1=-0.197 X1+0.204 X2-0.182 X3+0.182 X4-0.052 X5-0.197 X6+0.021 X7-0.059 X8+0.19 X9; F2=-0.071 X1-0.396 X2-0.138 X3+0.031 X4+0.383 X5+0.077 X6+0.505 X7-0.101 X8-0.029 X9; F3=0.036 X1-0.052 X2+0.068 X3+0.059 X4+0.370 X5+0.104 X6-0.274 X7+0.78 X8+0.081 X9.

[0074] Table 6. KMO and Bartlett's Test Table 7 Principal Component Variance Extraction method: Principal component analysis.

[0075] Table 8. Rotated Component Matrix a Extraction method: Principal component analysis. Rotation method: Caesar normalized maximum variance method. Rotation α converged after 4 iterations.

[0076] Table 9 Component Score Coefficient Matrix Extraction method: Principal component analysis. Rotation method: Kaiser normalized maximum variance method.

[0077] 2.2.2 Stepwise linear regression analysis: To construct a predictive model for chilling injury index in bitter gourd after refrigeration, based on correlation and factor analysis, a multiple linear regression model was constructed using chilling injury index (Y) as the dependent variable and respiration rate (X2), weight loss rate (X3), firmness (X4), chlorophyll (X5), MDA content (X6), SOD activity (X7), POD activity (X8), and CAT activity (X9) as independent variables. The model construction and testing results are shown in Tables 10-11. After stepwise selection, the final model retained four independent variables: X3, X6, X9, and X7. The resulting predictive regression equation is as follows: .

[0078] The overall variance analysis of the regression model reached a highly significant level (F=442.36, P<0.001), indicating that the independent variables included in the model have sufficient statistical explanatory power for the cold damage index. The model has a high goodness of fit, with a correlation coefficient (R) of 0.985 and a coefficient of determination (R²) of [missing value]. 2 The adjusted R² was 0.97, and the adjusted R² was 0.968, indicating that the model could explain 96.8% of the variance in the dependent variable. Multicollinearity analysis showed that the variance inflation factor (VIF) of each variable in the model ranged from 1 to 3.3, all much less than 10, indicating that the model did not have a serious multicollinearity problem. Residual analysis showed that the standardized residuals had no extreme outliers, satisfying the basic assumptions.

[0079] The standardized coefficients (Beata) reflect the relative influence of each independent variable on the dependent variable; the larger the absolute value, the stronger the explanatory contribution of that variable to the dependent variable. From largest to smallest, they are: weight loss rate (0.78), MDA content (0.161), CAT activity (-0.141), and SOD activity (0.082). The results show that in the constructed multiple linear regression model, weight loss rate is the primary positive factor predicting chilling injury, followed by MDA content. CAT activity exhibits a weak negative protective effect, while SOD activity has the smallest predictive contribution. Therefore, this invention successfully constructs a statistically significant and robust predictive model, clearly indicating that weight loss rate and MDA content are the most critical indicators for predicting chilling injury during bitter gourd storage, providing a basis for subsequent in-depth discussion and application.

[0080] Table 10 Model Summary e a. Predictor variables: (constant), weight loss rate. b. Predictor variables: (constant), weight loss rate, MDA. c. Predictor variables: (constant), weight loss rate, MDA, CAT. d. Predictor variables: (constant), weight loss rate, MDA, CAT, SOD. e. Dependent variable: Cold damage index.

[0081] Table 11 Coefficients a a. Dependent variable: Cold damage index.

[0082] Table 12. Analysis of Variance for Multiple Linear Regression a a. Dependent variable: Cold damage index. b. Predictor variables: (constant), weight loss rate, MDA, CAT, SOD.

[0083] 3. Discussion and Conclusion: This invention applies a segmented precooling procedure to the postharvest treatment of bitter gourd. Results show that the three-stage precooling treatment (T2) exhibits the best effect, significantly reducing the chilling injury index by 25.63% compared to the control group. This indicates that the segmented heating, through stepwise temperature stimulation, effectively activates the antioxidant defense system of bitter gourd while avoiding irreversible damage that might be caused by a single low temperature. Based on various physiological indicators, the theoretical mechanism is transformed into an applied model. With the chilling injury index as the dependent variable and the other eight indicators as independent variables, a chilling injury prediction model is established using stepwise multiple linear regression analysis with generalized least squares method. The resulting chilling injury prediction equation is Y = -2.07 + 2.667. X3+3.183 X6-0.717 X9+0.022 The X7 model constructs a chilling injury prediction index system covering physical characteristics, texture properties and physiological activities from four dimensions: weight loss rate (X3), MDA (X6), CAT activity (X9) and SOD activity (X7), providing a basis for early diagnosis of chilling injury risk in bitter gourd storage.

[0084] Example 2: A method for predicting the degree of chilling injury in bitter gourd Step S1, Sample pre-cooling treatment: (1) Raw material screening: Du Ruan bitter gourd fruits produced in orchards in Guangzhou, Guangdong Province, China, were selected at the same maturity and harvest period. After harvesting, the fruits were immediately transported to the laboratory for manual screening. Fruits with similar shape and size, uniform skin color, no mechanical damage, no deformity, no overripeness, and no cracks were selected as experimental samples.

[0085] (2) Group design: The qualified bitter melons were randomly and evenly divided into 4 groups, with the same number in each group, and each group corresponded to a different pre-cooling treatment method, as follows: Control group (CK): No pre-cooling treatment was performed; the samples were directly placed in a storage environment set at 11±1 ℃. Rapid pre-cooling group (T1): Bitter gourd was placed in a pre-cooling environment of 4±1 ℃ for 4 hours, and then transferred to storage at 11±1 ℃; Three-stage precooling group (T2): The bitter melon was processed in the following three stages in sequence: a) First stage: Pre-cool at 4±1 ℃ for 80 minutes; b) Second stage: Transfer to an environment of 8±1 ℃ and continue pre-cooling for 80 minutes; c) Third stage: Transfer to storage at 11±1 ℃; Two-stage pre-cooling group (T3): The bitter gourd was first pre-cooled at 6±1 ℃ for 2 hours, and then transferred to 11±1 ℃ for storage.

[0086] (3) Sampling and index determination: Physicochemical index tests were conducted on days 0, 4, 8, 12 and 16 of storage. Three fruits were randomly selected from each group each time, and external indexes (such as chilling injury index, respiration rate and hardness) were measured first. After the test, the seeds and pods of each fruit were removed, chopped and thoroughly mixed to make a mixed sample, which was immediately placed in an ultra-low temperature freezer at -80 ℃ for subsequent extraction and determination of physiological and biochemical indexes.

[0087] Step S2, Sample index determination: The aforementioned multiple indicators refer to the data collection on chilling injury index, respiration rate, weight percentage, firmness, chlorophyll, malondialdehyde (MDA), superoxide dismutase (SOD), peroxidase (POD), and catalase (CAT) of bitter gourd during storage.

[0088] (1) Chilling injury index: The fruits are graded according to the chilling injury symptoms. The chilling injury symptom grading of the bitter gourd samples is as follows: Grade 0: The surface of the melon is smooth, with no signs of chilling injury, and the flesh is firm; Level 1: Slight chilling injury, with slight pitted spots or water-soaked spots on the peel covering ≤10% of the peel area, and no obvious changes in the flesh, which loses its luster; Level 2: Moderate chilling injury, with sunken spots or water-soaked spots on the peel covering 11% to 30% of the peel area, and the flesh beginning to soften, lose its luster, and begin to wrinkle; Level 3: Severe chilling injury, with sunken spots or water-soaked spots on the peel covering 31% to 60% of the peel area, the peel turning yellow, and the flesh becoming noticeably softer; Level 4: Severe chilling injury, with sunken spots or water-soaked spots on the peel covering more than 60% of the peel area, the peel color turning significantly yellow, and the appearance of tail rot, decay, or white mold.

[0089] The cold damage index is expressed by the following formula: .

[0090] (2) Respiration rate: Take one fruit, weigh it, place it in a sealed container, record the CO2 production over 1 hour, record the temperature before and after, and measure 3 parallel samples. Calculate the respiration intensity using the following formula: Where: a: carbon dioxide concentration after measurement (ppm); 44 / 22.4: conversion factor from ppm to mg / m³ under standard conditions; a0: initial carbon dioxide concentration, ppm; T: temperature after measurement, °C; T0: temperature before measurement, °C; V: total volume of container, dm³ 3 t: Time taken for measurement, h; m: Mass of fruits and vegetables used for measurement, kg.

[0091] (3) Weight loss rate: Weigh the sample before and after storage using an electronic balance and calculate the weight loss rate using the following formula: In the formula: A is the weight of the bitter melon before storage, in g; B is the weight of the bitter melon after storage, in g.

[0092] (4) Hardness: Three fruits were randomly selected, and the hardness was measured at three points at equal intervals around the equator of the fruit. The texture analyzer program used TPA (Texture Profile Analysis) to simulate the human chewing process. The probe type was a P / 5N cylindrical probe, and the parameters were set as follows: force sensing range 500 N, trigger force 0.1 N, deformation 15%, and probe pre-measurement speed 60 mm / min.

[0093] (5) Chlorophyll content: The absorbance was measured at 663 nm and 645 nm using a UV-Vis spectrophotometer via ethanol extraction, and the calculation formula is as follows: In the formula: A 645 The absorbance of the sample at 645 nm was extracted; A 663 The absorbance of the sample at 663 nm was extracted; W is the fresh weight of the sample, kg; N is the dilution factor.

[0094] (6) Malondialdehyde (MDA) content: The absorbance of the reaction product at 532 nm was determined using the thiobarbituric acid (TBA) method. Where: V: volume of MDA extract, L; W: fresh weight of plant tissue, g; A 532 The absorbance of the sample at 532 nm; A 600 =Absorbance of the sample at 600 nm; A 450 =Absorbance of the sample at 450 nm.

[0095] (7) Related antioxidant enzyme activities: a) Superoxide dismutase (SOD) activity: SOD activity can be calculated by measuring the change in the reactant at 450 nm using a water-soluble tetrazolium salt (WST-1) colorimetric method. The formula is as follows: In the formula: A1: absorbance of the measuring orifice; A 1-0 A1: Absorbance of the blank well; A2: Absorbance of the control well; 2-0: Absorbance of the control blank well; 12: Reaction volume / dilution factor, 0.24 mL / 0.02 mL; N: Dilution factor before testing; W: Weight of tissue sample, g; V 样 : Total volume of buffer solution added during sample homogenization, mL.

[0096] b) Peroxidase (POD) activity: The POD activity can be calculated by measuring the change in the reactant at 470 nm using the guaiacol method. The formula is as follows: Where: A1 is the absorbance value of the sample measured after 3 min of incubation; A0 is the absorbance value of the sample measured after adding the working solution; W: weight of the tissue sample, mg; V T V: Total volume of enzyme extraction solution, mL; S : Volume of enzyme solution used in the assay, mL; t: Reaction time, min.

[0097] c) Catalase (CAT) activity: The activity of CAT can be calculated by measuring the change in the reactant at 405 nm using the ammonium molybdate method. The formula is as follows: In the formula: V 样 : Sample volume in this system, 0.1 mL; T: Reaction time, 60 s; W: Tissue sample mass, g; V 样总 : Total volume of the sample homogenate, mL.

[0098] Step S3: Screening key predictive factors: Correlation and factor analysis were performed on chilling injury related indicators: chilling injury index (X1), respiration rate (X2), weight loss rate (X3), hardness (X4), chlorophyll (X5), MDA content (X6), SOD activity (X7), POD activity (X8) and CAT activity (X9). Three common factors F1, F2 and F3 were extracted, and a common factor score calculation model was established.

[0099] The key predictor screening refers to: (1) Screening for significant differences: Using SPSS software, Waller-Duncan test was performed on all indicators measured in step two. With p<0.05 as the significance threshold, the indicators that showed statistically significant differences between different treatment groups were obtained, forming the primary screening indicator set. (2) Correlation strength screening: Spearman's rank correlation coefficient analysis method is used to calculate the correlation between each index and the cold damage index for the primary screening index set. p<0.05 is used as the significance standard, and the absolute value of the correlation coefficient |r| ≥ 0.7 is used as the strong correlation standard. The indexes that are strongly correlated with the cold damage index and are significant are screened out to form the core correlation index set. (3) Principal Component Factor Dimension Reduction: Exploratory factor analysis was performed on the core related index set, and principal component analysis was used for extraction. First, KMO and Bartlett tests were performed. The KMO sampling appropriateness measure was required to be ≥0.6 and the Bartlett sphericity test significance p<0.001 before principal component analysis could be performed. The rotation method adopted was the Kaiser normalized maximum variance method. Principal component factors were extracted based on the eigenvalue greater than 1 and the cumulative variance contribution rate ≥80%. According to the rotated component matrix, the index variables with an absolute value of loading greater than 0.7 on each principal component factor were selected and defined as the final key predictive factors. Common factor score calculation: Based on the factor analysis results, a calculation model for standardized observed variables and common factor scores is established. The common factor score F is calculated using the component score coefficient matrix B and the standardized vector of original variable observations X: F = B′ R - ¹ X, where R -1 The inverse of the correlation coefficient matrix of the original variables is given by the following formula for calculating the score of the common factors: In the formula, i represents the total number of observed variables participating in the factor analysis; c j1 ,c j2 ,…,c ji Z1, Z2, ..., Zn are the coefficients in the component score coefficient matrix corresponding to the j-th common factor; p For the original observed variables X1, X2, ..., X p The value after standardization.

[0100] In one specific embodiment, principal component analysis extracted three common factors, whose cumulative variance contribution rate reached 86.23%. The common factors in the rotated component matrix mainly consist of: the first common factor (F1): mainly determined by X1, X2, X3, X4, X6, and X9, with loadings of -0.954, 0.732, -0.902, 0.916, -0.84, and 0.937 on the principal factors, respectively. F1 is a comprehensive factor; a higher score indicates milder chilling injury symptoms, lower weight loss, better texture retention, stronger CAT activity, and lower MDA content. It is also important to note the relatively high respiration rate. In the early stages of fruit and vegetable storage, before the rapid decline in respiration, a higher respiration rate is observed, reflecting a covariation pattern of "high respiration - low chilling injury."

[0101] The second common factor (F2) is determined by X5 and X7, with loadings of 0.705 and 0.838 on the principal factors, respectively. Higher scores indicate stronger superoxide scavenging ability and better color retention of the fruit during storage.

[0102] The third common factor (F3) was mainly determined by peroxidase (POD) activity (0.927), revealing the main enzymatic scavenging pathway of peroxides in fruits.

[0103] Based on the component score coefficient matrix obtained from factor analysis, the calculation formulas for the three common factors are established as follows: F1=-0.197 Z1+0.204 Z2-0.182 Z3+0.182 Z4-0.052 Z5-0.197 Z6+0.021 Z7-0.059 Z8+0.190 Z9; F2=-0.071 Z1-0.396 Z2-0.138 Z3+0.031 Z4+0.383 Z5+0.077 Z6+0.505 Z7-0.101 Z8-0.029 Z9; F3=0.036 Z1-0.052 Z2+0.068 Z3+0.059 Z4+0.370 Z5+0.104 Z6-0.274 Z7+0.780 Z8+0.081 Z9; Among them, Z1 to Z9 represent the standardized chilling injury index, respiration rate, weight loss rate, hardness, chlorophyll content, malondialdehyde content, SOD activity, POD activity and CAT activity, respectively.

[0104] Step S4: Construct a cold damage prediction model: (1) Based on the correlation analysis and factor analysis in step S3, SPSS statistical analysis was used to calculate the scores of each common factor, F1, F2, ..., Fm. m Using Y as the independent variable and the measured cold damage index as the dependent variable, a prediction model is constructed using multiple linear regression analysis. The mathematical form of the resulting model is as follows: In the formula, β 0 is a constant term. β 1, β 2,…, β m , where are the regression coefficients of each common factor.

[0105] (2) In one specific embodiment, the measured chilling injury index (Y) was used as the dependent variable, and respiration rate (X2), weight loss rate (X3), stiffness (X4), chlorophyll content (X5), malondialdehyde (MDA) content (X6), superoxide dismutase (SOD) activity (X7), peroxidase (POD) activity (X8), and catalase (CAT) activity (X9) were used as initial independent variables. A stepwise multiple linear regression analysis was performed using the generalized least squares method. After stepwise regression screening, the final model retained four independent variables: weight loss rate (X3), malondialdehyde (MDA) content (X6), catalase (CAT) activity (X9), and superoxide dismutase (SOD) activity (X7). The resulting regression equation for the chilling injury index prediction model is: Y = -2.07 + 2.667 X3+3.183 X6-0.717 X9+0.022 X7.

[0106] Example 3: Validation of a method for predicting the degree of chilling injury in bitter gourd 1. Experimental Method: The correlation coefficient (R) of the model obtained in Example 2 on the training set is 0.985, and the coefficient of determination (R²) is... 2 The adjusted R is 0.97. 2 The value was 0.968. In the experimental data, the leave-one-out cross-validation method was used for verification (Li Bo, Sun Xianglong, Yao Mingze, et al. Validation of the DSSAT-CROPGRO-Tomato model under different amounts of straw returned to the field in greenhouse [J]. Journal of Ecology, 2021, 40(03): 908-918.).

[0107] 2. Experimental Results: Experimental results are as follows Figure 3 As shown.

[0108] Figure 3 This is a scatter plot validating the cold damage index prediction model. Figure 3 It can be seen that the correlation coefficient (R) between the predicted value and the experimental value obtained by the method for predicting the degree of chilling injury in bitter gourd provided by the present invention is 0.974, and the coefficient of determination (R²) is... 2 The value is 0.949, and after adjustment (R) 2 The value is 0.948.

Claims

1. A method for predicting the degree of chilling injury in bitter gourd, characterized in that, Includes the following steps: Step S1: Sample pre-cooling treatment: Select fruits with similar shape and size, uniform color, and no deformities, overripeness, or cracks. Store qualified bitter gourd samples separately under the following conditions: no pre-cooling, rapid pre-cooling, three-stage pre-cooling, and two-stage pre-cooling. Step S2, Sample index determination: On days 0, 4, 8, 12 and 16 of storage, the chilling injury index, respiration rate, weight percentage, hardness, chlorophyll, malondialdehyde, superoxide dismutase, peroxidase and catalase of bitter gourd samples were measured. Step S3: Screening key predictive factors: Conduct correlation analysis and factor analysis on the cold damage-related indicators detected in step S2, extract three common factors F1, F2 and F3, and establish a common factor score calculation model. Step S4: Construct a cold damage prediction model: Based on the analysis in step S3, perform multiple linear regression analysis to construct a prediction model.

2. The method for predicting the degree of chilling injury in bitter gourd as described in claim 1, characterized in that, The different precooling treatments in step S1 are grouped as follows: Control group: Bitter melon samples were not pre-cooled and were directly placed in a storage environment set at 11±1 ℃; Rapid pre-cooling group: Bitter gourd samples were placed in a pre-cooling environment of 4±1 ℃ for 4 hours, and then transferred to 11±1 ℃ for storage; Three-stage pre-cooling group: The bitter gourd samples were first pre-cooled at 4±1 ℃ for 80 minutes; then transferred to 8±1 ℃ for another 80 minutes; and finally transferred to 11±1 ℃ for storage. Two-stage pre-cooling group: The bitter gourd samples were first pre-cooled at 6±1 ℃ for 2 hours, and then stored at 11±1 ℃.

3. The method for predicting the degree of chilling injury in bitter gourd as described in claim 1, characterized in that, In step S2, the chilling injury index is statistically graded based on the chilling injury symptoms of the bitter gourd samples. The chilling injury symptoms of the bitter gourd samples are graded as follows: Grade 0: The surface of the melon is smooth, with no signs of chilling injury, and the flesh is firm; Level 1: Mild chilling injury, with slight sunken spots or water-soaked spots on the peel covering ≤10% of the peel area, no obvious changes in the flesh, and loss of luster; Level 2: Moderate chilling injury, with sunken spots or water-soaked spots on the peel covering 11%~30% of the peel area, the flesh begins to soften, lose luster, and begin to wrinkle; Level 3: Severe chilling injury, with sunken spots or water-soaked spots on the peel covering 31%~60% of the peel area, the peel color turns yellow, and the flesh softens significantly; Level 4: Extreme chilling injury, with sunken spots or water-soaked spots on the peel covering >60% of the peel area, the peel color turns significantly yellow, and tail rot or rot or white mold appears.

4. The method for predicting the degree of chilling injury in bitter gourd as described in claim 3, characterized in that, The cold damage index is calculated using the following formula: 。 5. The method for predicting the degree of chilling injury in bitter gourd as described in claim 1, characterized in that, In step S3, the three common factors F1, F2, and F3 are respectively: The first common factor F1 is determined by the cold injury index, respiration rate, weight loss rate, hardness, malondialdehyde content, and catalase activity. The second common factor F2 is determined by superoxide dismutase activity and chlorophyll content. The third common factor F3 is determined by peroxidase activity.

6. The method for predicting the degree of chilling injury in bitter gourd as described in claim 5, characterized in that, The common factor score calculation model in step S3 is as follows: The common factor score F is calculated using the component score coefficient matrix B and the standardized vector of original variable observations X: F = B′ R - ¹ X, where R -1 The common factor score is calculated using the inverse matrix of the original variable correlation coefficient matrix. In the formula, i represents the total number of observed variables participating in the factor analysis; c j1 ,c j2 ,…,c ji Z1, Z2, ..., Zn are the coefficients in the component score coefficient matrix corresponding to the j-th common factor; p For the original observed variables X1, X2, ..., X p The value after standardization.

7. The method for predicting the degree of chilling injury in bitter gourd as described in claim 6, characterized in that, The formulas for calculating the scores of the common factors F1, F2, and F3 are as follows: F1=-0.197 Z1+0.204 Z2-0.182 Z3+0.182 Z4-0.052 Z5-0.197 Z6+0.021 Z7-0.059 Z8+0.190 Z9; F2 = -0.071 Z1-0.396 Z2-0.138 Z3+0.031 Z4+0.383 Z5+0.077 Z6+0.505 Z7-0.101 Z8-0.029 Z9; F3=0.036 Z1-0.052 Z2+0.068 Z3+0.059 Z4+0.370 Z5+0.104 Z6-0.274 Z7+0.780 Z8+0.081 Z9; Among them, Z1 to Z9 represent the standardized chilling injury index, respiration rate, weight loss rate, hardness, chlorophyll content, malondialdehyde content, superoxide dismutase activity, peroxidase activity and catalase activity, respectively.

8. The method for predicting the degree of chilling injury in bitter gourd as described in claim 1, characterized in that, In step S4, the multiple linear regression analysis uses the scores of each common factor, F1, F2, ..., F... m Using Y as the independent variable and the measured cold damage index as the dependent variable, a model is constructed. The mathematical form of the model is: In the formula, β 0 is a constant term. β 1, β 2,…, β m , where are the regression coefficients of each common factor.

9. The method for predicting the degree of chilling injury in bitter gourd as described in claim 8, characterized in that, The chilling injury index prediction model uses the measured chilling injury index (Y) as the dependent variable and respiration rate (X2), weight loss rate (X3), hardness (X4), chlorophyll content (X5), malondialdehyde content (X6), superoxide dismutase activity (X7), peroxidase activity (X8), and catalase activity (X9) as initial independent variables. It employs generalized least squares stepwise multiple linear regression analysis.

10. The method for predicting the degree of chilling injury in bitter gourd as described in claim 9, characterized in that, The model ultimately identified four independent variables: weight loss rate (X3), malondialdehyde content (X6), catalase activity (X9), and superoxide dismutase activity (X7); the regression equation for the cold injury index prediction model is: Y = -2.07 + 2.667·X3+ 3.183·X6– 0.717·X9+ 0.022·X7.

Images