Plant growth‐promoting rhizobacteria mediate soil hydro‐physical properties: An investigation with Bacillus subtilis and its mutants

Plant growth‐promoting rhizobacteria and other soil bacteria have the potential to improve soil hydro‐physical properties and processes through the production of extracellular polymeric substances (EPS). However, the mechanisms by which EPS mediates changes in soil properties and processes remain incompletely understood, partly due to variations in EPS composition produced under different environmental conditions. In this study, we investigated the influence of different bacterial traits on intrinsic soil properties and processes of evaporation and infiltration using sand treated with the wild‐type Bacillus subtilis variant (UD1022) and its two mutant variants, eps−$eps^{-}$ – tasA−$tasA^{-}$ and srf AC−$AC^{-}$ . The eps−$eps^{-}$ – tasA−$tasA^{-}$ mutant suppresses EPS production through alterations in the eps and tasA genes, while the srf AC−$AC^{-}$ mutant lacks the gene for surfactin production. Experimental results confirmed that the solution viscosity of the eps−$eps^{-}$ – tasA−$tasA^{-}$ mutant was the lowest and the solution surface tension of the srf AC−$AC^{-}$ mutant was the highest among the three tested bacteria strains. The distinct intrinsic properties of EPS produced by these bacterial strains resulted in varied hydro‐physical responses in the treated sand. Key influences included modifications in wettability, hydraulic decoupling (or mixed wettability), and aggregation, which collectively led to reduced evaporation rates and heterogeneous water distribution during infiltration in the bacteria‐treated sands. Our findings advance the understanding of the role bacterial EPS play in vadose zone hydrology and offer insights for the development of sustainable strategies for increasing water retention, supporting crop production in arid regions, and facilitating land restoration.


INTRODUCTION
Plant growth-promoting rhizobacteria (PGPR) have been extensively investigated for their capacity to enhance plant tolerance to abiotic and biotic stresses, including drought (Branda et al., 2006;Kumar et al., 2012;Lakshmanan et al., Vadose Zone Journal 2015; Zheng et al., 2018). These studies suggest the potential of using PGPR to mitigate various abiotic stresses experienced by plants. Moreover, soil microorganisms like PGPR serve as valuable indicators of soil health (Fierer et al., 2021;Schimel, 2018) and play a role in maintaining and improving soil functions (Hall et al., 2018;Harris et al., 2018;Saha et al., 2020;Coban et al., 2022). Multiple mechanisms have been proposed to explain the observed soil hydrological changes mediated by PGPR and other soil bacteria. First, bacterial exudates can promote aggregate formation, leading to changes in soil structure (Rashid et al., 2016). Aggregate formation enhances soil stability and resilience to erosion (Jastrow et al., 1998;Six et al., 2004). In addition, the intra-aggregate pores tend to be smaller than those in non-aggregated soil pores, enabling soil aggregates to retain more water due to the greater capillary force within these micropore spaces. Second, bacterial extracellular polymeric substances (EPS) exhibit hygroscopic and hydrophilizing properties, creating a more hydrated environment and reducing desiccation stress (Chenu, 1993;Or et al., 2007;Roberson & Firestone, 1992). Third, PGPR can induce physicochemical changes in soil, including reduced surface tension, increased solution viscosity, and modified soil wettability, which have been found to significantly affect soil hydraulic properties (Chenu & Guérif, 1991;Lieleg et al., 2011;Stoodley et al., 2002;Hallett et al., 2013). Finally, pore clogging can reduce saturated hydraulic conductivity (Baveye et al., 1998;Rice, 1974), while under partially water-saturated conditions, water flow can bypass the clogged zone, leading to uncertain consequences (Volk et al., 2016).
Bacterial exudates, which contain polysaccharides, proteins, nucleic acids, and lipids, have complex chemical compositions (Flemming & Wingender, 2010) thus can influence soil properties through different mechanisms. Polysaccharides exhibit hydrogel effects (Brax et al., 2017), while some nucleic acids and lipids contain surface-active substances and behave as surfactants. Zheng et al. (2022) examined the individual effects of hydrogel and surfactant and found opposing trends; xanthan enhanced soil water retention because of increased solution viscosity and its hygroscopic property, while surfactants reduced soil water retention due to reduced surface tension. However, Zheng et al. (2022) used xanthan and Triton X-100 as chemical analogs for bacterial exudates. Further investigation using bacteria mutants will provide more direct evidence of the effects of various bacterial-produced substances on soil hydrological processes.
In this study, we inoculated sand using a wild-type EPSproducing strain of Bacillus subtilis UD1022 and its two mutant strains, − -− and srf − , which lacked genes for EPS and surfactin production, respectively. We hypothesized that the bacterial exudates produced by UD1022, − -− mutant, and srf − mutant would exhibit distinct physical and chemical properties, such as pellicle

Core Ideas
• Effects of wild-type Bacillus subtilis and its two mutants on infiltration and evaporation were studied. • Composition of bacterial exudates changes soil properties and thus hydrological processes. • Viscosity and surface tension are key factors responsible for the observed changes.
morphology, surface tension, viscosity, and contact angle, resulting in varied influences on soil infiltration and evaporation processes. By using mutant strains, we aimed to conduct a systematic investigation to differentiate the specific effects of different exudate components on soil intrinsic properties and hydrological processes.

Sand and bacteria strains
In this study, quartz sand with a mesh size between 40 and 50 (Unimin Corp.) was used as a model soil. Sand is commonly used in mechanistic studies like in ours because it has the essential properties of soil and thus allows targeted investigation on specific mechanisms while reducing complexities caused by processes like chemisorption and swelling that occur in soil. The sand was washed, dried, and sterilized at 121˚C for 30 min before use. The wild-type B. subtilis UD1022 (Lakshmanan et al., 2013) along with its two mutant strains, − -− and srf − were used to incubate the sand. The mutant − -− produces EPSdeficient biofilm by repressing the eps and TasA operons, while the mutant srf − lacks the srfAC gene and suppresses the production of surfactin (Bishnoi et al., 2015;Patrick & Kearns, 2009). The − -− strain was grown in Luria Bertani (LB) broth with the antibiotics of 1 μg mL −1 erythromycin and 5 μg mL −1 tetracycline, while the srf − strain was grown in LB with 1 μg mL −1 erythromycin and 25 μg mL −1 lincomycin, where the concentration of lincomycin enabled the identification, culturing, and fitness of the respective mutants. Bacterial cultures were grown to OD 600 ∼2.0, washed once in sterile water, and re-suspended to OD 600 ∼1.0 prior to application as described above. Mutant strains in the UD1022 background were generated using SPP1 phage transduction (Yasbin & Young, 1974). The sand was mixed homogeneously with deionized (DI) water and each bacterial suspension to achieve an initial water content of 5% (by weight) and a bacterial concentration of 7 × 10 7 CFU g −1 . Control samples were prepared following the same procedure but without bacteria. All samples were incubated overnight at 37˚C before packing the columns.

EPS composition
The composition of EPS produced by the different bacterial strains was analyzed at the Complex Carbohydrate Research Center of the University of Georgia. This analysis was based on the coupled gas chromatography-mass spectroscopy (GC-MS) technique. Suspensions of the three bacterial strains were centrifuged and dialyzed to separate EPS from the spent media and then freeze-dried. Samples were further prepared by acidic methanolysis processes to generate O-trimethylsilyl derivatives of the monosaccharides to be detected and quantified in the GC-MS analysis, as described in Santander et al. (2013).

Interfacial properties
Surface tensions of the three bacterial suspensions were measured using a platinum ring force tensiometer (Sigma 700, Biolin Scientific). Each measurement lasted ∼49 h until a stable surface tension value was reached (Rühs et al., 2013).
Contact angle values of sand samples, including the control and bacteria-treated samples with the three strains, were measured using the sessile drop method (Bachmann et al., 2000). Water droplets (14 μL each) were placed on a thin layer of sand (control or treated sand), prepared using doublesided tape, and photographed with a charge-coupled device (CCD)-equipped microscope. The images of droplets were analyzed with the ImageJ software (Schneider et al., 2012). Six replicates for each treatment were prepared and measured.

2.2.3
Viscoelastic behavior Prior to measuring viscosity, bacterial samples were grown in the LB broth medium as described in Section 2.1. A shear rheometer (Discovery HR-3 rheometer, TA instruments) with cone-plate geometry (60 mm ø, 1 o angle) was used for the measurements. First, amplitude and frequency sweep tests were conducted to measure the linear viscoelastic regimes of the bacteria samples ( Figure S1). For the amplitude sweep test, the applied strain was from 0.1% to 100% with an angular frequency of 1.0 rad s −1 . For the frequency sweep test, the applied frequency was 0.01-100 rad s −1 at 10% strain. Then an interfacial rheometer with a double-ring wall geometry was used to identify the storage (or elastic) modulus and loss (or viscous) modulus of the bacterial EPS samples formed at the air-water interface (Vandebril et al., 2010). Viscosity was calculated from the viscoelastic modulus, considering the geometry gap of the instrument, which in our case was 1200 μm (Naveed et al., 2019). The measurement was done at a constant deformation: the strain was set at 0.1% and the frequency was 1 rad s −1 . These definitive strain and frequency values were determined based on a linear viscoelastic regime. Each experiment was run for 2 h using bacterial suspensions that had been incubated in the sample holders for 30 h, which formed pellicles at the interface. Bacillus subtilis reaches a plateau around 30 h, which is why this incubation time was chosen for the experiment.

Pellicle assay
Pellicle assay was performed to observe morphological features of the biofilm formed by UD1022, − -− , and srf − . We followed a recommended procedure to prepare bacterial suspensions to have an optical density, measured at 600 nm wavelength, at OD 600 < 0.5 (Beauregard et al., 2013). We suspended bacterial cells from freshly cultured 1-day-old colonies in the LB growth medium and carried out multiple dilutions (at least three times) at 1:100. During each resuspension and dilution process, the bacterial cells were grown for 2 h in an incubator shaker at 160 rpm and 37˚C. For each strain, 27-μL suspension with an OD 600 = 0.09 was transferred to a 12-well plate containing 2 mL LB broth in each well. All plates were incubated at 37˚C for pellicle formation (Beauregard et al., 2013). Pictures of the pellicles formed were taken at 24 h, 72 h, and after complete desiccation to assess bacterial growth at different stages.

Infiltration and evaporation experiment
Column experiments were conducted to observe and quantify the effects of different bacterial strains on infiltration and evaporation characteristics during one wetting-drying cycle. The experiments were conducted in an enclosed Plexiglas-Al-tinted environmental chamber ( Figure 1) to maintain a constant temperature (25 ± 1˚C) and relative humidity (40 ± 8%) during the experiments. Acrylic columns with a diameter of 4.37 cm and a height of 9.66 cm were used. Each column was packed with 224 g of pretreated sand samples at 5% (by weight) moisture content, which resulted in a porosity of 41% and a bulk density of 1.55 g cm −3 . The procedure for running the wetting-drying experiment is described as follows. First, 20 mL DI water mixed with a sterile green food dye (McCormick & Company) (4% v/v) was sprinkled on top of each column using a needleheaded irrigation appendage ( Figure 1). The dye was used to Vadose Zone Journal F I G U R E 1 Experimental set-up of infiltration and evaporation experiments. enhance visual observation of wetting front propagation. The food dye is a non-ionic surfactant with propylene glycol as one of the ingredients. Any possible effect the dye might have had would be similar in all columns because they received the same amount of the dye solution. The spatial distribution of dyed water in each column was recorded continuously with a camera (Canon, DSLR EOS Rebel T7). After the infiltration experiment, these columns were placed on balances, which recorded evaporative water loss at a 10-min interval. Three replications were run for each treatment (control, UD1022, − -− , and srf − ).

Carbohydrate composition
The carbohydrate (CHO) analyses revealed that the EPS produced by all three bacterial strains were abundant in mannose, galactose, glucose, inositol, and peptidoglycan or muramic acid ( Figure 2). These compounds are major CHO polymeric substances synthesized by bacterial cells (Jones et al., 2014). Although the composition of EPS produced by the different strains was similar, the total CHO content varied, with the highest found for the srf − strain (10.3%), followed by UD1022 (7.64%) and then the − -− (4.5%). Figure 3 presents the morphological features of the pellicles formed by different bacterial strains. The UD1022 and srf − strains produced thick and continuous pellicles, while the − -− formed more discrete and thinner pellicles.

Pellicle morphology
The dried pellicles of the UD1022 and srf − strains were F I G U R E 2 Extracellular polymeric substances (EPS) carbohydrate compositions and total carbohydrate (CHO) percentages of the UD1022, − -− , and srf − strains.
more cohesive than those formed by the − -− strain.
In addition, the images indicate that the production of biofilm dry mass differed among the strains in the order of UD1022> surface tensions of 32 ± 0.36 mN m −1 and 37 ± 0.36 mN m −1 , respectively. The contact angle values of air-dried sand particles incubated with different bacteria strains and the control are shown in Figure 4b. The control had a contact angle of 30.8 ± 2.7˚, indicating a hydrophilic wetting property. All bacteria-treated samples exhibited larger contact angles compared to the control. The contact angles were 66.1 ± 2.7˚, 54 ± 3.2˚, and 54.6 ± 5.0˚for the sand treated with UD1022, − -− , and srf − , respectively. The modified wetting behavior was further demonstrated by measuring the length of time the droplets persisted on each sand surface ( Figure A2). UD1022 and srf − strains had similar effects and a more significant effect compared to the − -− strain in reducing droplet spreading time.

Viscosity
The viscosities of exudates produced by the test strains followed the order of srf − > UD1022> − -− (Figure 5a). This finding is consistent with the EPS composition analysis, which showed that the total CHO content also followed this order. Oscillatory viscoelasticity measurements quantify the elasticity (G') and the viscous nature (G'') of a material. If G' > G'', the material is considered more elastic than viscous and vice versa (Lieleg et al., 2011;Stoodley et al., 2002). Figure 5 shows that srf − and UD1022 variants exhibited more viscoelastic solid character-istics (rubber-like; Figure 5a), while the − -− strain displayed characteristics of a viscoelastic liquid (honey-like; Figure 5b).

Infiltration rate and wetting pattern
The infiltration (or wetting front) propagation into initially dry sand samples treated with different bacteria strains was recorded using dyed water (Figure 6, A1-top panel). Figure 7a,b presents the depth of wetting front and the wetting front propagation rate, respectively. The bacteria-treated columns showed slightly lower propagation rates compared to the control at the site of infiltration, then became similar after ∼1 min for all treatments. The treated columns showed more heterogeneous water distribution than the control, as indicated by the dye distribution patterns (A1-bottom panel), which were photographed immediately after the infiltration of 20 mL water was completed. Figure 7 shows the cumulative evaporative water loss and evaporation rate from the sand columns over a drying period of 12 days. The drying experiment began immediately after the completion of infiltration of 20 mL DI water into differently treated columns. Figure 7a shows that the control and UD1022-treated sand had the highest total cumulative evaporative loss, followed by the − -− and srf − treated columns. The control exhibited the highest evaporation rate until day 4, while the UD1022-treated sand maintained a higher rate from day 4 through day 12 (Figure 7b). In contrast, the − -− -treated column had the lowest evaporation rate during days 2-8 and the highest rate for the reminder of the experiment. The evaporation rate curves as a function of water saturation ( Figure 8) show a typical two-stage evaporation characteristic: stage I with a relatively high and constant evaporation rate, and stage II with a lower rate. The transition from stage I to II occurred on day 3 for the control and day 1.8 for the UD1022 and srf − treatments. The − -− treatment did not show a clear stage I, and if any, it was very brief at the highest water saturation level, then dropped to rates lower than all other treatments for most of the water saturation range. Overall, the evaporation rate for the control was higher during stage I and most of stage II, and stage I lasted longer than in the bacteria-treated samples. The transition from stage I to II was more gradual in UD1022 and srf − treated samples. In addition, the drying rate at stage II was the lowest for the control, while it was similar for the UD1022, − -− , and srf − treatments.

F I G U R E 4
Changes in the surface tension at the air-water interface and contact angle of the UD1022-, − -− -, and srf − -treated sands. The error bars indicate ± SD from the mean (n = 6).

F I G U R E 5
Viscosity profiles (a) and rheological properties (b) of the UD1022, − -− , and srf − strains. G' and G" denote storage (elastic) and loss (viscous) modulus, respectively. The error bars indicate ± SD from the mean (n = 3)

F I G U R E 6
Wetting front depth (a) and rate of wetting front propagation (b) during infiltration of 20 mL dye water into the bacteria-treated sand columns. The error bars indicate ± SD from the mean (n = 3).

DISCUSSION
In this study, our objective was to investigate the impact of different groups of bacterial exudates on the intrinsic properties of soil and the underlying mechanisms influencing hydrological processes such as infiltration and evaporation. Our results provide clear evidence that bacterial exudates modify both the interfacial properties of soil and the fluid characteristics of the soil solution. Specifically, UD1022 simultaneously reduced the solution surface tension and increased its viscosity, while the srf − variant only increased viscosity, and the − -− variant exhibited the greatest reduction in viscosity among the three strains. Moreover, our results clearly demonstrated lower infiltration rates, more heterogeneous water distribution patterns, and reduced evaporation rates in the bacteria-treated sand columns. In this section, we discuss the underlying mechanisms for the effects of different bacteria strains on soil infiltration and evaporation through the modification of surface tension and viscosity of the soil solution and the wettability of the sand matrix.

Bacterial exudates and their impact on the intrinsic properties of treated sand
Measurements of pellicle morphology, surface tension and viscosity of bacterial suspensions, and contact angle revealed distinct differences in properties and their influence on the sand samples treated with the three bacterial strains. For example, the surface tension values of srf − suspension and LB broth are similar and significantly higher than those of UD1022 and − -− strains. This can be attributed to the lack of surfactin-producing genes in the srf − strain (Patrick & Kearns, 2009), while both the UD1022 and − -− strains can produce biosurfactants. Sorption of surfactant molecules at the air-water interface leads to a reduction in surface energy and thus surface tension (Soberón-Chávez et al., 2005). It is worth mentioning that the surface tension of LB broth is slightly lower than that of DI water due to the presence of some surface-active molecules in the broth (Qi & Christopher, 2021). Biosurfactants can facilitate the spreading of biofilm, locomotion, and attachment of bacterial cells by reducing the surface energy of solid surfaces (Angelini et al., 2009;Yang et al., 2021). The reduction in surface tension can alter water retention and flow behavior in porous media by reducing water-film connectivity and the air-entry value (Carminati et al., 2017;Mueller et al., 2019;Shokri et al., 2012). According to the Young-Laplace equation, lower surface tension enables the drainage of finer pores (Read et al., 2003).

Vadose Zone Journal
Viscosity varies with the amount of polymeric substances such as mucilage or bacterial EPS (Naveed et al., 2019). The measured viscosity values of the different bacterial suspensions showed a clear correlation with the total CHO content measured for the EPS produced by each bacteria strain. The − -− strain, which had the lowest CHO content, also exhibited reduced viscosity compared to the UD1022 and srf − strains with higher CHO contents. The srf − strain produced the most viscous exudates ( Figure 5) and thus could reduce water flow by increasing hydraulic stability. All bacterial strains increased sand hydrophobicity, as indicated by the greater contact angle values of the treated samples compared to the control, although the differences among the different strains were not significant. It has been reported that soil carbon produced by plants, bacteria, or fungi can cause soil water repellency (Seaton et al., 2019). Another factor that can affect contact angle is biofilm morphology. As shown in Figure 3, the pellicles formed by the UD1022 strain had the most "wrinkles." The wrinkled morphology of the biofilm can lead to water repellency by air entrapment, which was identified as a main reason for the larger contact angle measured on pure biofilm of the wild-type variant compared to the exopolysaccharide-deficient strain in the study by Epstein et al. (2011). It should also be noted that our measurements were made using air-dried sand particles, and water repellency depends on moisture content.
The three bacterial strains induced different changes in interfacial properties, which are known factors that affect matric potential and water movement under unsaturated conditions. Below, we discuss how the distinct physicochemical properties of bacterial exudates caused the varied hydrophysical responses in the sand, thereby affecting infiltration and evaporation processes.

Effects of bacterial exudates on infiltration
The bacteria-incubated samples exhibited nonuniform water distribution during water infiltration, as indicated by dye deposition patterns, which contrasted with the control sample, where water distribution was more uniform (See Figure A3). The formation of concentrated color zones is most likely due to the nonuniform distribution of bacteria and their exudates, which may clog small pores and induce localized high water retention capabilities. The nonuniformity could also be related to the change in water repellency of sand particles upon bacterial treatment. Sand is inherently hydrophilic but can become hydrophobic when coated with nonpolar organic compounds, which orient toward air when coated on sand surfaces and lead to soil water repellency (Tschapek, 1984). Furthermore, because bacteria-incubated sand particles have greater contact angle values than untreated sand, the mixed wettability of sand particles in the columns could also cause nonuniform water distribution. As demonstrated by Bentz et al. (2022), the presence of sporadic hydrophobic deposits can induce water repellency in an otherwise well-wettable soil. Our results are consistent with their observations, suggesting that porescale heterogeneity of bacteria/exudate distribution impacts soil wettability and infiltration.
In addition to nonuniform water distribution, the sand samples incubated with bacteria had slightly lower infiltration rates than the control. The rehydration of bacteria exudates, which can create large viscous drag for flow within porous spaces or cause pore clogging, may lead to a slowdown in infiltration. This mechanism was reported as being responsible for reduced infiltration into mucilage-amended soil samples (Benard et al., 2018). However, we did not observe any significant differences in infiltration rates among the different bacterial treatments. This is consistent with the contact angle measurement, which showed similar hydrophobicity across all the bacteria-treated samples. Infiltration, as a percolation process, has a percolation threshold that is correlated with water repellency (Benard et al., 2018). Under severe water-repellent conditions, where the percolation threshold is exceeded, water penetration into soil pores is impeded. Conversely, when below the percolation threshold, the timescale for water penetration into soil pores is on the order of milliseconds to seconds. It is likely that infiltration through our samples, which were all weakly water repellent with contact angle values of <90˚ (Figure 4b), occurred below the percolation threshold. Consequently, the differences in infiltration rate among the different bacteria treatments were found to be insignificant.

Effects of bacterial exudates on evaporation
The observed reduction in evaporation in bacteria-incubated samples is consistent with previous studies that reported a deceleration of evaporation by EPS-producing strains of Pseudomonas (Roberson & Firestone, 1992;Volk et al., 2016;Zheng et al., 2018) and Sinorhizobium meliloti (Deng et al., 2015). Our study using the wild-type and two mutant strains of B. subtilis revealed two underlying mechanisms responsible for reducing the evaporation rate of bacteria exudates. The first mechanism is driven by surfactin production, which causes an earlier transition from stage I to II evaporation. This is demonstrated by comparing the drying curves of surfactinproducing − -− (surfactant-producing) treatment to the control, as well as the surfactin-producing UD1022 to the srf − treatment (Figure 8). According to Lehmann et al. (2008), the relative constant evaporation rate during stage I is maintained due to the formation of continuous flow pathways connecting the receding front with the vaporization surface. The transition from stage I to diffusion-controlled stage II is induced by the breakup of capillary continuity. Because bacterial exudates containing surfactin (e.g., − -− and UD1022) have a lower surface tension and, thus, a smaller capillary force, stage I evaporation ended earlier for the surfactin-affected samples.
The second mechanism by which bacterial exudates affect the evaporation rate is by increasing the viscosity of the fluid phase. The increase of fluid viscosity has two main consequences. First, higher viscosity can decrease evaporation flux during stage I, a capillarity-driven phase dominated by viscous flow, by reducing viscous losses (Lehmann et al., 2008). Second, higher viscosity facilitates the formation of liquid bridges or filaments between soil particles, which create an additional capillary force that retains water during the drying process Carminati et al., 2017). Studies by Benard et al. (2019Benard et al. ( , 2021 showed the formation of onedimensional polymer structures that helped maintain water film connectivity and flow. Zheng et al. (2022) found that preferential flow pathways for air were more easily formed with increasing viscosity ratios between fluid and air, leading to reduced liquid phase displacement by air and, thus, increased liquid saturation (Chen & Bonaccurso, 2014;Zhang et al., 2011). The two EPS-producing strains (UD1022 and srf − ), which increased viscosity, showed dual actions in affecting evaporation. At the initial stages of evaporation, higher viscosity led to higher water retention and lower evaporation, presumably by reducing viscous losses. However, as the viscosity continued to increase in the later stages of evaporation, it facilitated the evaporation process by maintaining a connected film flow. These observations are consistent with previous reports Carminati et al., 2017;Zheng et al., 2022).
The combined effects of surface tension and viscosity can be explained by the dimensionless Ohnesorge number that relates viscous forces to inertial and surface tension forces: , where μ is the dynamic viscosity of the liquid (Pa s), ρ is the density of the liquid (g m −3 ), σ is the surface tension of the air-water interface (mN m −1 ), and L is the characteristic length of the liquid connection (m). Carminati et al. (2017) reported that the liquid bridges formed at Oh > 1 do not break and thin filament forms, while the bridges formed at Oh < 1 break. For DI water (μ = 10 −3 Pa s; ρ = 10 3 kg m 3 ; σ = 0.07 N m −1 ), the Oh value is significantly less than 1 (Oh = 0.04 and 0.01 for a 10 and 100 μm pore size, respectively). The values of the Oh number increase as viscosity increases and surface tension reduces. Using μ = 10 −3 Pa s; ρ = 10 3 kg m 3 ; and σ = 0.04 N m −1 as the average fluid parameters for the UD1022 suspension, the calculated Oh values were 5 and 1.6 for pore sizes of 10 and 100 μm, respectively, which are much greater than 1. These effects are reflected in the different Oh values observed for the samples treated with different bacteria strains in our study. For instance, the srf − strain, which produced exudates with the highest viscosity value, gave rise to the largest Oh number, leading to the highest water retention and thus the longest water supply during evaporation ( Figure 8). Benard et al. (2019) showed formation of one-and two-dimensional structures in soil pores by mucilage as the treated soil progressively dried. Similarly, we observed such structural development by the EPS-producing UD1022 and srf − strains, as evidenced by the thread-like structures of the dry pellicles ( Figure 3). The formation of additional matrix or two-dimensional strand-like structures is further supported by the scanning electron microscope (SEM) image of the UD1022 strain ( Figure A4) and the fact that the UD1022-treated column maintained its original bulk density, whereas the sand in the control column became more compacted (i.e., bulk density increased) at the end of the evaporation experiments ( Figure A5).

CONCLUSIONS
Bacterial exudates containing different groups of chemicals induce varied changes in surface tension, viscosity, and soil wettability, consequently influencing hydrological processes such as infiltration and evaporation differently. Through the use of B. subtilis UD1022 wild-type and two mutant strains, we demonstrated the distinct influences of viscous components and surface-active components present in bacterial exudates on the intrinsic properties and hydrological processes of sand. While previous studies have highlighted the interplay between biophysical processes in regulating water supply and flux from the soil to the atmosphere through evaporation and transpiration, we emphasized in this study the role of surfactin and viscous EPS in maintaining longer hydraulic connectivity for sustained longer water supply (e.g., UD1022) during a wetting-drying cycle. The coating of bacterial exudates on substrate surfaces, such as sand, increases hydrophobicity and modifies wettability, thus facilitating heterogeneous water distribution during infiltration. The heterogeneity can contribute to reduced evaporation rates and increased water retention, in addition to the high water-absorbing capability of EPS. The capacity of microorganisms to improve both the physical and hydraulic properties of soils hold great potential in mitigating drought in crop production and restoring degraded lands. Our study advances the mechanistic understanding of bacteria-mediated changes in soil properties and hydrological dynamics. This knowledge provides a scientific basis for developing strategies that harness the beneficial functions of PGPR and other soil bacteria to promote plant growth, improve soil hydraulic properties, and potentially facilitate ecosystem restoration.

C O N F L I C T O F I N T E R E S T S T A T E M E N T
The authors declare no conflicts of interest.