*2.1. Chemicals*

Carvacrol (purity 99%) purchased from Shanghai Aladdin Co., Ltd. (C804847), and because of its insolubility in water, it was previously dissolved in ethanol (50%, *w*/*w*) to obtain a stock solution with a proper concentration of 10 mg· mL−1.

#### *2.2. Preparation of P. Digitatum Spores*

The highly pathogenic fungus *P. digitatum* was isolated from an infected orange with typical green mold symptoms and maintained on potato dextrose agar (PDA: 200 g peeled potatoes, 20 g glucose, 18 g agar powder and 1 L distilled water) medium plates at 25 ◦C. The preparation of *P. digitatum* spore suspension was based on a previous method [22], and the tested pathogen was incubated at 25 ◦C for

7 days. The seven-days-old plate was washed with sterile water and then gently dispersed by spread bacteria beads to release spores. Finally, the spore suspensions were filtered through a sterile cotton ball in a funnel to remove mycelia and PDA fragments and adjusted to the suitable concentration of 1 × 10<sup>6</sup> spores mL−<sup>1</sup> with the aid of a hemocytometer.

#### *2.3. Antifungal E*ff*ects of Carvacrol against P. Digitatum*

The inhibitory effect of carvacrol on the mycelial growth of *P. digitatum* was determined as previously reported [22]. Briefly, the 0, 0.0625, 0.125, 0.25, 0.5, and 1 mL of carvacrol stock solution was diluted with 2, 1.9375, 1.875, 1.75, 1.5, 1 mL of sterile 0.5% Tween 80 and mixed with 18 mL of PDA for obtaining the final concentrations of 0 (control), 0.03125, 0.0625, 0.125, 0.25, and 0.5 mg·mL−1. The mycelial disks (5 mm in diameter), cut from the periphery of a seven-days-old culture using a stainless-steel punch, was placed in the center of each Petri dish (90 mm in diameter). Then, all plates were incubated at 25 ◦C for seven days. Four replicates were used per treatment and the experiment was carried out at two separate times. Mycelial growth inhibition (MGI) of carvacrol treatment against control was calculated using the following equation:

$$\text{MGI} \left( \% \right) = \frac{Dc - Dt}{Dc - 5} \times 100$$

where *Dc* and *Dt* were the mean colony diameter of control and treated sets, respectively.

The minimum inhibitory concentration (MIC) was defined as the lowest carvacrol concentration that completely inhibited the growth of *P. digitatum* after 48 h of incubation at 25 ◦C. The minimum fungicidal concentration (MFC) was considered the lowest concentration of carvacrol with no visible fungal growth on a PDA plate after a following 5 days incubation at 25 ◦C [22].

#### *2.4. The E*ff*ect of Carvacrol on Mycelial Weights and Water-Retention Rate*

The effects of carvacrol on the wet and dry weights as well as the water-retention rate of *P. digitatum* mycelial were determined by the method described by Tian et al. with some modifications [23]. Briefly, 100 μL of *P. digitatum* containing 10<sup>6</sup> spores mL−<sup>1</sup> was inoculated into 50 mL of potato dextrose broth (PDB) medium and then was incubated in a rotary shaker (150 rpm) at 25 ◦C. After shake incubating for 48 h, the carvacrol solution at final concentrations of 0 (control), 0.03125, 0.0625, 0.125, 0.25 and 0.5 mg·mL−<sup>1</sup> were added into the above-mentioned PDB and then incubated for 24 h at 25 ◦C in a rotary shaker. The mycelia from the carvacrol treated and control PDB was collected by filtering used a Buchner funnel and washed three times with sterile water. The wet weights of the mycelia were measured, the mycelia were dried at 70 ◦C for 12 h, and the dry weights were then measured using an analytical balance (Ms105, Mettler Toledo, Greifensee, Switzerland). The water-retention rates of the carvacrol-treated and control groups were calculated using the following equation:

$$\text{Water-retention rate } (\%) = \frac{\text{W}w - \text{Wd}}{\text{W}w} \times 100$$

where *Ww* and *Wd* were the mean of wet and dry weights in carvacrol treated and control sets, respectively.

#### *2.5. Sample Preparation for 1H NMR Spectroscopy*

The collected mycelial treated with MIC carvacrol for 4, 8 and 12 h respectively were washed three times with pre-cooled PBS buffer solution and then added with 3.8 mL pre-cooled methanol-water mixture (1/0.9, v/v). The mixture was placed on the ice for 4 min of sonication bathing, and then added with 4 mL trichloromethane. After full oscillation, the mixture was centrifuged at 4 ◦C and 10,000 rpm for 10 min. The upper methanol water phase was placed in the nitrogen blowing instrument (NBI, HSC-24B, Tianjin Hengao Technology Development Co. Ltd, Tianjin, China) to blow off the methanol, and then the supernatants were freeze-dried, and then stored under −80◦C until NMR analysis [11].

In the NMR measurements, the samples were dissolved in 550 mL 99.8% D2O phosphate buffer (0.2 M, pH = 7.0), which contained 0.05% (w/v) 3-(trimethylsilyl) sodium propionate -2, 2, 3, 3-d4 (TSP). After rotating for 15 s and centrifuging for 10 min at 12,000 rpm and 4 ◦C, the supernatant was transferred to a clean nuclear magnetic resonance tube (5 mm) for analysis. The 1H-NMR spectrum was recorded in the 298 K at 500 MHz nuclear magnetic resonance spectroscopy (Bruker Avance III, Bruker Instruments, Darmstadt, Germany). Field-frequency locking with D2O and TSP was used as a reference for chemical shift (1H, D 0.00). The carrpurcell-meiboom-gill sequence [90 (t-180-t) n-acquisition], which is edited by lateral relaxation, has a total spin-echo delay (2 n t) of 40 ms. The spectra were recorded in 64K data points with 128 scans, ranging from −5–15 ppm. By multiplying FIDS with exponential weighting function (corresponding to line broadening at 0.5Hz), Fourier transform was performed on the spectrum.

#### *2.6. Spectral Pre-Processing and Data Analysis*

All 1H NMR spectra were phase and baseline corrections, and the peak was manually calibrated using Topspin software (Bruker BioSpin GmbH, version 3.5, Rheinstetten, Karlsruhe, Germany). Then the peak was exported to ASCII file using Mestrec (version 4.9.6, Mestrelab Research SL, Santiago de Compostela, Spain) and imported into R software (https://www.r-project.org/) for further analysis. The region containing residual water signal was removed. The spectrum was combined with an adaptive intelligent algorithm. Before multivariate data analysis, the remaining trash cans were the probability quotient normalization and Pareto scale.

Orthogonal signal correction partial least squares discriminant analysis (OSC-PLS-DA) was used to reveal the metabolic differences among groups. Score charts are used to show clustering between categories, and load/s charts are used to identify different metabolites between two groups. Differential metabolites are metabolites in the upper right quadrant and the lower left quadrant of the S diagram. According to the correlation coefficient from blue to red, the loading graph is color coded. The model was validated internally by repeated two cross-validations and externally by a permutation test.
