Isolation of antagonistic bacteria was performed by serial dilution method as previously described (Heo et al., 2022). Briefly, rhizosphere soil samples of various plants were collected from Agi mountain, Andong, Republic of Korea (36°32’22.9”N 128°56’14.3”E). One gram of samples was dissolved in 10 ml of sterilized water and mixed by using a shaker (DLAB Scientific co., Ltd., China) at room temperate for 2 h. The mixture was filtered using cheese cloths. One milliliter of the mixture was transferred sterilized 1.5 ml micro-tube and serially diluted up to 10^-7. Subsequently, 0.1 ml of the diluted suspension was seeded on King’s B agar plates (KB, MB cell, Korea), then incubated at 28 °C for 48 h. All single colonies were transferred to fresh KB broth using sterile toothpicks.

Acidovorax citrulli strains KACC17909, KACC17000, KACC17001, and KACC17005 were generously provided by Korean Agricultural Culture Collection (KACC; http://genebank.rda.go.kr/). Antagonistic activity of unknown bacterial isolates against Acidovorax citrulli was screened as previously described with minor modifications (Bauer et al., 1966). Briefly, each bacterial isolate was grown in a KB broth medium at 180rpm 28 °C for 72 h. To prepare the agar plug, 100 μl of bacterial suspension was spread on KB agar plates and incubated at 28°C for 72h. After incubation, plates were punctured with a sterile cork borer (5.5 mm in diameter). Agar plugs of bacterial isolates were placed in the middle of KB agar plates seeded with A. citrulli strain KACC17909. Among 240 unknown bacteria isolated and tested, only 1 bacterial isolate named YM002 showed significant antagonistic activity against A. citrulli KACC17909. After the screening, antagonistic activity of YM002 is further tested against another A. citrulli strains KACC17000, KACC17001, and KACC17005. The plates were incubated at 28°C for 72h and the clear zone (total inhibition zone diameter – YM002 growth diameter) was measured. Swimming motility of YM002 was also measured by colony diameter during the agar plate assay in the presence of different strains of A. citrulli. All bacterial isolates were preserved in 20% glycerol (v/v) at -80°C. The strain YM002 was deposited to the KACC (accession number: KACC92433P).

To identify bacterial strain showing antagonistic activity against A. citrulli, 16s rRNA sequencing and phylogenetic analysis were performed. Genomic DNA was extracted by using a Genomic DNA prep kit (BIOFACT, Daejeon, Korea), according to the manufacturer’s instructions. 16s rRNA region was amplified using a universal primer set, 27F (5′-AGA GTT TGA TCC TGG CTC AG-3′) and 1492R (5′-GGT TAC CTT GTT ACG ACT T-3′) (Lane, 1991; Weisburg et al., 1991). PCR reaction was performed with a mixture containing 0.5 μl Pfu-X (2.5 U/μl), 5 μl 10X Pfu-X reaction buffer, 1 μl 10 mM dNTP mix, 2 μl DNA template, and 2 μl of each primer. The PCR cycle was performed at an initial denaturation (2 minutes at 95 °C) and followed 33 cycles with denaturation at 95 °C for 30 seconds, annealing at 56 °C for 30 seconds, extension at 72 °C for 30 seconds, and a final extension at 72 °C for 5 minutes. Amplified PCR product was purified using Gel & PCR purification system (BIOFACT, Daejeon, Korea), according to the manufacturer’s instructions. The 16s rRNA gene of antagonistic bacteria was sequenced by Solgent sequence analysis services (Solgent co., Ltd, Korea). The sequence data were aligned by using Lasergene sequence analysis software (Burland, 2000). The resulting sequence of the 16s rRNA gene was analyzed with NCBI’s Gen-Bank sequence database (http://www.ncbi.nlm,nih.gov) to identify the closest species relatives. The phylogenetic tree was constructed by Maximum likelihood methods with bootstrap values of 1,000 replications using MEGA X (Kumar et al., 2018).

The extracellular amylase production ability of YM002 was determined as previously described (Cappuccino and Sherman, 1996) with slight modifications. Briefly, YM002 agar plug was inoculated on the center of starch agar plates (5.0 g peptone, 3.0 g yeast extract, 2.0 g soluble starch, and 15.0 g agar/L, pH adjusted to 7.0) for 5 days at 28 °C. After incubation, 3% iodine solution was poured into the plates and incubated for 20 minutes at room temperature. The diameter of the yellow-colored clear zone was measured.

To measure the ACC deaminase activity, minimal Dworkin and Foster (DF) salt media was prepared as previously described (Penrose et al., 2001). Briefly, the trace elements (10mg H₃BO₃, 11.19mg MnSO₄·H₂O, 124.6mg ZnSO₄.7H₂O, 78.22mg CuSO₄·5H₂O, and 10mg MoO₃) were dissolved in 100 ml sterile distilled water (DW). FeSO₄ solution was prepared by dissolving 100 mg of FeSO₄·7H₂O in 10 ml sterile DW. Medium compounds (4.0g KH₂PO₄, 6.0g Na₂HPO₄, 0.2g MgSO₄.7H₂O, 2.0g glucose, 2.0g gluconic acid, 2.0g citric acid, and 15g agar) and 0.1 ml of each solution were dissolved in 1L DW (pH adjusted to 7.2) and autoclaved. After autoclave, 3.0 mM 1-aminocyclopropane-1-carboxylic acid (ACC) was supplemented as a sole nitrogen source. YM002 agar plug was inoculated on the center of minimal DF agar medium for 7 days at 28°C. After incubation, the colony diameter was measured.

The extracellular protease production of YM002 was determined by monitoring the clear zone formation on skim milk agar plates (Chu, 2007). Briefly, 10g of skim milk was dissolved in 400 ml DW, and 1g yeast extract and 15g agar were dissolved in 600 ml DW, separately. Each reagent was autoclaved and mixed on a clean bench. YM002 agar plug inoculated on the center of skim milk medium. YM002 was incubated for 5 days at 28°C, and the diameter of the clear zone was measured.

To check siderophore production activity, CAS agar plate assay was performed as previously described (Schwyn and Neilands, 1987; Arora and Verma, 2017). Briefly, 60.5mg chrome azurol S (CAS) was dissolved in 50 ml sterile deionized water and added to 40 ml hexadecyl trimethyl ammonium bromide (HDTMA). HDTM solution was prepared by dissolving 72.9 mg HDTMA in 40 ml sterile DW. Then, mixed 90 ml CAS-HDTMA solution with 900 ml NA agar medium. One millimolar ferric chloride solution was prepared by dissolving FeCl₃·6H₂O in 10 mM HCl. NA agar medium and Ferric chloride solution were autoclaved separately. After autoclaving, both reagents were mixed on a clean bench. YM002 agar plug inoculated on CAS agar plate for 7 days at 28°C. After incubation, the color change from blue to yellow was observed (Louden et al., 2011).

The zinc solubilization activity of YM002 was determined as previously described (Bapiri et al., 2012). PKV medium (10.0 g glucose, 1.0 g (NH₄)SO₄, 0.2 g KCl, 0.2 g K₂HPO₄, 0.1 g MgSO₄, 0.2 g yeast extract and 15 g agar/L, pH 7.0) supplemented separately with either 1% ZnO or 1% ZnCO₃. YM002 agar plug inoculated on the center of PKV agar medium for 5 days at 28°C. After incubation, the diameter of the clear zone was measured.

For ammonia (NH₃) production assay, YM002 was grown in KB broth medium for 72 h and adjusted OD₆₀₀ of 0.05 in sterile DW. One milliliter of cell suspension was inoculated at 95 ml of peptone water broth for 48 h at 48 °C. One milliliter of supernatant was mixed with 1 mL of Nessler’s reagent and 8.5 mL of sterile DW (Goswami et al., 2015). The mixture color change from brown to yellow indicates NH₃ production, which is observed by absorbance at 450 nm by using a Nanoready Touch microvolume spectrophotometer (Genomic Base, Korea). The NH₃ concentration was estimated using the standard curve generated by ammonia sulfate ranging from 1 to 10 μmol/mL. The experiment was performed with three replicates.

The quantification of indole-3-acetic acid (IAA) production was estimated using Salkowski’s reagent (Bric et al., 1991). YM002 was grown in KB broth medium for 72 h and adjusted to OD₆₀₀ of 0.05 in sterile DW. One milliliter of YM002 cell suspension was inoculated into 95 ml of KB broth medium containing 3 mM L-tryptophan. The culture was incubated in a shaking incubator at 180 rpm at 28 °C for 72 h. One milliliter of supernatant was mixed with two milliliters of Salkowski’s reagent (0.5 M FeCl₃·6H₂O in 35% perchloric acid), then incubated at room temperature for 30 minutes under dark conditions. The mixture color change from brown to pink indicates IAA production, which is observed by absorbance at 530nm by using a Nanoready Touch microvolume spectrophotometer (Genomic Base, Korea). The standard curve of IAA concentration range from 10 to 100 μg/mL was prepared by using commercial IAA (Sigma, USA). The experiment was performed with three replicates.

Biofilm formation inhibitory activity of culture filtrate (CF) of YM002 against A. citrulli was measured as previously described (O’Toole et al., 1999; Masum et al., 2017). To prepare cell-free CF, YM002 was inoculated from a single colony into 100 ml of KB broth and incubated in a shaking incubator at 180 rpm at 28 °C for 72 h. Cultures were transferred to a 50 mL falcon tube and centrifuged at 10,000 rpm for 10 min. The supernatant was filtered through the membrane filter with a pore size of 0.22 μm (PALL Life Sciences, USA). One milliliter of A. citrulli strain KACC17001 (O.D₆₀₀ = 0.05) was mixed with 9 mL of fresh KB broth with different concentrations of YM002 CF (1, 3 and 10%) with five replicate wells per treatment, then seeded in 6-well cell culture plate. Plates were incubated at 28 °C without agitation for 24 h and 48 h, then bacterial cells were carefully discarded. Ten milliliters of 0.1% crystal violet in sterile DW were added to each well for 30 minutes. The plates were then washed with sterile DW, air dried for 30 min, and resuspended in 10 mL of 95% ethanol for 20 minutes. The absorbance was measured at 590 nm.

The effect of YM002 culture filtrate against swimming motility of A. citrulli was determined using methods previously described with slight modifications (Bahar et al., 2009; Masum et al., 2017). A. citrulli KACC17001 was grown in KB broth for 24 h, centrifuged at 10,000 rpm for 10 min, and the pellet was washed twice with sterile DW. The pellet was resuspension in sterile DW and its concentration was adjusted to OD₆₀₀ of 0.05. Two microliters of bacterial suspension were inoculated on the center of a soft NA (0.3% agar) plate supplemented with 0, 25, and 50% YM002 CF. After incubating the plates for 48h at 28 °C, the colony diameter of A. citrulli was measured. This experiment was done with 8 replicates for each.

The plant growth promoting (PGP) activity of YM002 was tested by using root dipping methods (Orhan et al., 2006). Two-week-old cucumber (Cucumis sativus L. cv. Joeunbaekdadaki) roots were submerged into 10⁵ and 10⁶ cfu/mL of YM002 cell suspension for 30 min. Control cucumbers were dipped into sterile tap water. Cucumber plants treated with or without YM002 were transferred into the pot (10 × 9 cm), filled up with commercial potting soil, and their growth was monitored at 25 °C for 14 days. To measure the PGP activity of YM002, different growth parameters, including height, number of leaves, root length, and fresh and dry weights of leaves, stems, and roots were measured.

To test the biocontrol activity of YM002, cucumber plants were inoculated with spontaneous mutant of A. citrulli KACC17001 which shows Rifampicin resistance in the presence or absence of YM002 by using the spray inoculation methods (Song et al., 2020). To generate spontaneous mutant of A. citrulli KACC17001, 0.5 ml of 10⁹ cfu/ml of bacterial cell suspension was inoculated onto KB agar plate containing 10 µg/ml of rifampicin (Rif), and incubated at 28 °C for 3 days. The mutant colony survived on KB agar plate containing Rif was obtained and used for bacterial growth assay. YM002 was inoculated from a single colony into 100 ml of KB broth and incubated shaking incubator at 180 rpm at 28 °C for 72 h. Two-week-old cucumber plants were sprayed with either 50 ml of 10% CF of YM002 or 1 × 10⁶cfu/ml of cell suspension with 0.01% tween 20. Control plants were sprayed with sterile DW with 0.01% tween 20. After 24h, 50 ml of A. citrulli strain KACC 17000, 17001, 17005 and 17909 (1 × 10⁷ cfu/ml) were sprayed. Eight leaf discs per plant were collected by using a cork borer (diameter = 5.5 mm) for real-time qPCR and bacterial titer measurement. After 5 days of inoculation, disease severity was measured by monitoring the percentage of the necrotic area by using ImageJ software (Abramoff et al., 2004).

Defense-related gene expression was determined by using real-time quantitative polymerase chain reaction (RT-qPCR). The cucumber leaves were ground by using SILAMAT S6 (Ivoclar Vivadent Inc., Austria). Total RNA was extracted by using the APure™ Plant RNA extraction kit (Genomic Base, Korea). Complementary DNA was constructed by using BioFACT™ RT-Kit (BIOFACT Co., Ltd., Korea), according to the manufacturer’s instructions. RT-qPCR was performed in Linegene 9600 Fluorescent Quantitative detection system (Bioer Technology, China). A reaction mixture of 20 μl was prepared by mixing 10 μl SYBR Green Real Time PCR master Mix (TOYOBO CO., LTD., Japan) with 2 μl cDNA (1:20 diluted form), 6.4 μl sterilized distilled water and 0.8 μl of 10 pmole primers. Actin is used as a reference gene to normalize the expression of CONSTITUTIVE TRIPLE RESPONSE 1 (CTR1), Pathogenesis-related protein 1-1a (PR1-1a), Phenylalanine ammonia-lyase 1 (PAL1), Ascorbate peroxidase (APOX) and Lipoxygenase 1 (LOX1) in RT-qPCR. Primers used in this experiments are provided in Supplementary Table 1 . The RT-qPCR cycle was performed at an initial denaturation (20 second at 95 °C), 40 cycles of denaturation (95 °C for 20 sec), annealing (56 °C for 20 sec) and extension (72 °C for 20 sec), and followed final extension (72 °C for 2 min). The RT-qPCR data were calculated by using the 2^{− ΔΔCt} method. All RT-qPCR experiment was performed with 3 replicates for each sample.

For all experiments, analysis of variance (ANOVA) and Duncan’s multiple range test were performed by using IBM SPSS version 28 software (SPSS Inc., USA). Data were expressed as mean ± standard deviation (SD) from triplicates of each experiment. Significant differences were considered at P ≤ 0.05.

We hypothesized that rhizosphere of various plants growing in mountain area contains useful and distinct microflora, and some of them exhibit antagonistic effect against plant pathogenic bacteria, such as Acidovorax citrulli. A total of 240 unknown bacterial strains were isolated from rhizosphere soil samples of various weeds and screened for their antibacterial activity against different strains of Acidovorax citrulli (KACC 17000, 17001, 17005, and 17909) in an agar plate assay. Only one unknown bacterial strain showed distinct antagonistic activity against all tested A. citrulli strains ( Figure 1A ). This isolate was named YM002 and used for further study. YM002 showed a significant inhibition zone against four different strains of A. citrulli on KB agar plates. Inhibition zone diameters induced by YM002 against A. citrulli strains 17000, 17001, 17005, and 17909 were 1.95 ± 0.13, 8.47 ± 0.49, 9.97 ± 0.5 and 7.57 ± 0.55 mm, respectively ( Figure 1B ). Interestingly, YM002 showed enhanced swimming motility in the presence of A. citrulli strain 17000 ( Figure 1C ). To identify YM002, the 16s rRNA gene sequence was amplified, sequenced, and subjected to phylogenetic analysis. From the NCBI BLAST (https://blast.ncbi.nlm.nih.gov/Blast.cgi) search, the 16s rRNA gene sequence of YM002 showed the highest sequence similarity with bacteria belonging to the genus Paenibacillus. Among different Paenibacillus spp. analyzed, YM002 showed the highest homology at 99.0% with P. tianmuensis strain B27 (Accession no. NR_116789.1) ( Figure 1D ). Based on the results, YM002 was identified as P. tianmuensis.

To test whether YM002 can suppress disease symptoms, fully expended cucumber leaves were spray-inoculated with 4 different strains of A. citrulli (1 × 10⁷ cfu/ml) at 24 h after treatment with cell suspension (CS) or culture filtrate (CF) of YM002. As shown in Figure 2A (upper panel), all A. citrulli strains induced necrotic water-soaking lesions in cucumber leaves at 5 days post inoculation (dpi); however, different strains showed different levels of virulence in cucumber leaves. The A. citrulli strain KACC17001 showed the highest virulence, and necrotic lesion area of 49.05 ± 12.19% (Mean ± SD) in cucumber leaves ( Figure 2B ). On the other hand, the necrotic lesion area induced by KACC17000, KACC17005 and KACC17909 were 16.11 ± 2.78, 4.50 ± 2.47 and 6.35 ± 4.12% (Mean ± SD), respectively, suggesting KACC17001 is the most virulent strain on tested cucumber plants. Thus we used A. citrulli KACC17001 for further experiments. Bacterial growth of A. citrulli KACC17001 was measured by using the spontaneous mutant to rifampicin ( Figure 2C ). Treatment of cucumber plants with CS or CF of YM002 significantly reduced growth of A. citrulli KACC17001.

Importantly, cucumber leaves pretreated with either the CS or CF of YM002 showed significantly reduced necrotic symptom area development after the infection with A. citrulli strains ( Figures 2A, B ). Pretreatment of cucumber leaves with CS of YM002 significantly decreased necrotic lesion area by 97.56% after inoculation with A. citrulli strain KACC17001 ( Figure 3B ), but not with KACC17000, KACC17005, and KACC17909. Necrotic lesion area in cucumber leaves by KACC17000, KACC17001, and KACC17005 was also significantly decreased by 93.57, 98.34, and 91.27% upon pretreatment with CF of YM002. This suggests that CF of YM002 is more effective than CS of YM002 for controlling the disease. Bacterial growth of A. citrulli strain KACC17001 was counted from the inoculated cucumber leaves ( Figure 3C ). Both CS and CF of YM002 treatments significantly lowered bacterial growth of A.citrulli at 2 dpi. Similar to the observed disease severity, CF of YM002 showed a higher inhibitory effect on bacterial growth compared to those of CS of YM002. This suggests CF of YM002 show significant in vivo antagonistic activity against A. citrulli.

In agar plate assay, YM002 showed growth inhibition activity against all tested A. citrulli strains ( Figures 1A, B ). Thus direct bactericidal/bacteriostatic activity of YM002 was tested by agar well diffusion assay ( Supplementary Figure 1 ). Unexpectedly, CF of YM002 did not show bactericidal/bacteriostatic activity against A. citrulli. Thus, we further tested whether CF of YM002 can alter pathogenesis-related phenotypes of A. citrulli. Reportedly, biofilm formation and swimming motility are considered as essential factors in the pathogenesis of A. citrulli (Bahar et al., 2009; Bahar et al., 2010). First, the biofilm formation by A. citrulli strain KACC17001 was measured at OD590nm every 24 h in the presence of 0, 1, 3, and 10% of YM002 culture filtrate ( Figures 3A, B ). Biofilm formation of A. citrulli strain 17001 was significantly decreased by 55.16, 46.36 and 64.27% at 48h after incubation with 1, 3, and 10% CF compared to the negative control, which contained 0% CF ( Figure 3B ). Swimming motility of A.citrulli strain KACC17001 was also measured on a soft nutrient agar (NA) plate supplemented with 0, 25 and 50% of YM002 culture filtrate. The growth diameter was significantly (P < 0.05) decreased by 23.34 and 86.91% in the presence of 25 and 50% of CF of YM002 compared to soft NA plate without CF ( Figures 3C, D ). These findings suggest that CF of YM002 can suppress the biofilm formation and swimming motility of A. citrulli.

To test whether YM002-induced enhanced resistance of cucumber plants could be due to its induced resistance (IR)-inducing activity, the expression of defense-related genes was tested by using the real-time qRT-PCR. For gene expression analysis, two-week-old cucumber leaves were pretreated with either CS (1 × 10⁶ CFU ml^-1) or 10% CF of YM002 at 24h before the inoculation with A. citrulli strain KACC17001. Defense-related gene expression levels were measured in cucumber leaves at 12 and 24 hours post inoculation (hpi) with A. citrulli strain KACC17001 ( Figure 4 ). Among the different genes tested, CONSTITUTIVE TRIPLE RESPONSE 1 (CTR1) and Pathogenesis-related protein 1-1a (PR1-1a) were significantly increased at 12 hpi with A. citrulli strain KACC17001 by 2.03- and 8.63-fold, respectively, in CF of YM002-treated leaves compared to mock-treated ones. Expression of Phenylalanine ammonia-lyase 1 (PAL1) was significantly increased at 24 hpi with A. citrulli strain KACC17001 by 3.82-fold in CF of YM002-treated leaves compared to mock-treated ones. On the other hand, expression levels of Ascorbate peroxidase (APOX) and Lipoxygenase 1 (LOX1) were not significantly different between the treatments.

Many Paenibacillus spp. are known to have different PGP activities (Grady et al., 2016). Thus the different PGP activities of YM002 were tested ( Figure 5 ). Notably, YM002 showed different PGP activities. The yellow halo zone surrounding the agar plug of YM002 on CAS agar plates indicates a positive result for siderophore production. The halo zone diameter was 3.46 ± 0.21 mm (Mean ± SD) at 7 dpi ( Figure 5A ). Zinc solubilization activity was observed on YM002-inouclated PKV medium supplemented with zinc oxide (ZnO), and zinc carbonate (ZnCO₃) ( Figures 5B, C ). Halo zone diameter was measured 1.67 ± 0.39 and 1.13 ± 0.26 mm (Mean ± SD) at 5 dpi, respectively. The extracellular protease production ability of YM002 was confirmed through the clear zone formation on skim milk agar plates ( Figure 5D ). The clear zone diameter induced by YM002 was 11.61 ± 0.26 mm (Mean ± SD) at 5 dpi. YM002 showed successful growth in minimal DF agar plates with ACC as the sole nitrogen source, suggesting it has ACC deaminase activity ( Figure 5E ). The growth diameter of YM002 was 4.18 ± 0.10 mm (Mean ± SD) at 7 dpi. YM002 showed a starch hydrolysis activity as it developed clear zone surrounding the colony at the center of the starch agar plate ( Figure 5F ). YM002-induced halo zone diameter was 20.22 ± 2 mm (Mean ± SD) at 7 dpi. To confirm the IAA production activity of YM002, the supernatant of YM002 was collected up to 72 hpi and reacted with salkowski’s reagent. The color of the mixture was observed positive results development of yellow to pink ( Figure 5G ). Quantitative estimation of IAA production from YM002 revealed that the maximum IAA production of 87.85 ± 7.45 μg/ml (Mean ± SD) was measured at 24 h after incubation. Lower levels of IAA production were observed from 36 to 72 hpi. In this experiment, the bacterial cell number of YM002 was almost saturated within 24 hpi, but its level was maintained up to 72 hpi ( Figure 5H ). The ammonia (NH₃) production ability of YM002 was also analyzed. Mixing YM002 supernatant with Nessler reagent induced a color change from brown to yellow, suggesting successful generation of ammonia ( Figure 5I ). The production of ammonia was observed initially at 36 hpi, and its level was increased by 15.18 ± 0.77 μg/ml at 48 hpi. Similar to the IAA production assay, the bacterial cell number of YM002 was almost saturated within 24 hpi, but its level was maintained up to 45 hpi ( Figure 5J ). Taken together, YM002 showed various PGP activities, such as siderophore production, zinc solubilization, extracellular protease production, ACC deaminase, starch hydrolysis, IAA production, and NH₃ production activities.

To test whether the PGP activity of YM002 can actually promote the growth of cucumber plants, two-week-old cucumber seedling plants were inoculated with 1 × 10⁵ and 1 × 10⁶ CFU/ml of YM002. As shown in Figure 6A , the cucumber inoculated with CS of YM002 showed enhanced plant growth 14 dpi. Most of the growth parameters, except the height of plants, were significantly enhanced ( Figures 6B–J ). In cucumber plants treated with 1 × 10⁶ and 1 × 10⁵ CFU/ml of YM002, root length (52.58 and 38.03%), number of leaves (23.81 and 19.05%) were significantly increased. In addition, 1 × 10⁶ and 1 × 10⁵ CFU/ml of YM002 treatments also increased fresh weight of leaves (64.48 and 94.69%), stems (25.20 and 18.57%) and roots (85.04 and 67.72%), as well as dry weight of leaves (48.72 and 61.54%), stems (22.1 and 7.5%) and roots (96.93 and 85.38%) compared to mock-treated plants. This suggests that YM002 not only enhances the defense against A. citrulli but also promote the growth of cucumber plants.