Introduction

Water movement in soil-plant-atmospheric continuum (SPAC) is a major component of hydrologic cycle and main driver of water balance in terrestrial ecosystems (Manzoni et al. 2013). The passive movement of water in SPAC influences the behaviour of many biophysical processes including, for instance, ecosystem productivity (i.e., plant photosynthetic activity (Boyer 1970; Jarvis 1976)), climate regulation (Lee et al. 2005; Novick et al. 2022) and microbial metabolism (Ghezzehei et al. 2019). On account of these essential roles, measuring water transport in SPAC has been a subject of active research in hydrology, plant and agricultural sciences, and meteorology (Deng et al. 2017; Norman and Anderson 2005). From the origin of the cohesion-tension theory, in order to quantify the SPAC hydraulic status, the hydraulic resistances that are encountered in individual domains of soil, plant and atmosphere (Sperry et al. 1998; Tyree and Zimmermann 2002) are required to be understood. Among the other constituents, plant resistance is a dominant component of the total hydraulic resistance in SPAC as it poses a complex mechanism for continuous water transport (Hillel and Hatfield 2005; Cornelis et al. 2010). Since stomata in the plant leaves are also tightly coupled with carbon cycle by simultaneous exchange of H2O and CO2 and their responses are relatively less complex to measure, the leaves are regarded as an effective biotic determinant (Díaz 2013) for interpreting plant hydraulic status. Thus, leaf-centric measurements such as changes in gross leaf or canopy area and canopy conductance are often conducted (Zhang et al. 2021; Tuzet et al. 2003; García-Tejera et al. 2017) to gauge the plant hydraulic status. But hidden behind the visible above-ground plant mechanism, plant roots need to respond adaptively to all biotic and abiotic SPAC resistances (Tyree and Ewers 1991; Johnson et al. 2016; Novick et al. 2022).

Given the significance of roots in a plant hydraulic system, their primary mechanism and functioning are encountered during water uptake from soil. Other than the optimisation of water uptake due to fluctuating demand of transpiration (Javaux et al. 2013; Couvreur et al. 2014), roots also play a critical role in self-regulation of plants under stress. For instance, upon drought, roots either regulate the water loss by stomatal closure in leaves (abscisic acid (ABA) signalling; Aroca et al. 2001; Jackson et al. 2003; Mehra et al. 2022) or experience tissue dehydration/embolism which could lead to plant mortality (Dobra et al. 2010; Choat et al. 2012). Therefore, understanding the hydraulics of roots during water uptake, especially amid its regulation to overcome water stress, is crucial to link the hydraulic statuses between plants and SPAC (Tang et al. 2018). A wealth of research, from the early resistance-based analogues (Cowan 1965; Anderson et al. 2003; Fatichi et al. 2016) to excised root measurements (Zarebanadkouki et al. 2016; Tardieu et al. 2017) and recent tomographic imaging methods (Carminati et al. 2009; Zarebanadkouki et al. 2019; Pascut et al. 2021), are hitherto improving the conceptualisation and parametrisation of such root hydraulic responses. All these studies referred root hydraulics to various parameters (Antmann et al. 2022); in particular, water potential (ψ) and hydraulic conductance (k) are predominantly determined to quantify root water uptake and to find key relations with soil and plant functions (Cai et al. 2022). Whilst undeniably both ψ and k are significant, the root water potential (ψR) could be more informative when root hydraulics are measured at the whole root level without explicitly determining the resistance in root geometry (Maurel et al. 2010). Indeed, it is the changes in ψ gradients that drive water flows through SPAC; thus, the information on ψR (i.e., the energy state of water in roots) is imperative to fundamentally link the ψ of soil (ψS) with that of leaves (ψL), both of which are much more convenient and straightforward to measure experimentally. Notwithstanding the necessity for obtaining this relevant information, the measurement on ψR is, however, discrete and not ascertained properly to aggregate with other ψ variables.

Water potential of roots, ψR, historically was interpreted from the Darcy-Richard’s equation (or referred to as microscopic root water uptake models), which has provided a conceptual framework for theoretical understandings of root water uptake. The foundational water uptake model derived by Gardner (1960) provided a mechanistic relationship of ψ between a single rootlet and the neighbouring soil across the soil-root interface. The models followed expanded the Gardner schema to incorporate detailed root geometry and transpiration functions using a water extraction term (or known as sink term; Whisler et al. 1968; Feddes et al. 1974; Herkelrath et al. 1977). Despite facilitating several outputs in the study of root water uptake, the Gardner’s model and their extended versions of root water uptake models have major shortcomings on determining the ψR. First, the governing model predicts the mere matric water potential (ψm) of individual roots which does not comprise the dominant ψR components, i.e., the osmotic water potential (ψo) and the pressure water potential (ψp) (Papendick and Campbell 1981; Tyree and Jarvis 1982). Second, even when ψR was included in the water extraction term of the coupled models, the values inputted are either simulated or alternated with ψL or stem ψ (Passioura 1980; Cai et al. 2022); thus, the water uptake models are yet to be verified with any actual ψR measurement (Zhuang et al. 2014) and to some extent the models could not be validated against the root measurements because the process of water uptake is more than the general assumption of simple linear flow (Dalton et al. 1975; Yu et al. 2007). Similarly, the measurement techniques like pressure chamber (Scholander et al. 1965) and psychrometer (Wullschleger et al. 1988) determined ψ drop between soil and leaf xylem, which were indispensable for understanding the hydraulic vulnerability of plants to several abiotic stresses. However, because methods like pressure chamber only observe the exudation point at the leaves but not in the roots, the value of ψR cannot be obtained. Moreover, even if the techniques are extended to measure the ψR in excised roots (Adeoye and Rawlins 1981; Oosterhuis 1987; Ball & Oosterhuis 2005; Bauerle et al. 2008; Puértolas et al. 2015), it is limited to short temporal scales that do not contribute a relationship with ψS and ψL. Because of the dearth of knowledge on the actual value of ψR, any observations of coordinated responses with soil and leaves are limited. Notwithstanding, previous studies have already demonstrated a nonlinear correlation between ψS and ψR based on the ψR measurements made by different techniques other than direct root fragment investigation (Dodd et al. 2010; Cai et al. 2022). It is important and necessary to fill the data gap to better inform different relevant applications, for examples, failure protection of engineered soil slopes where soil-root reinforcement changes with varying root hydraulic status (Boldrin et al. 2018; Wu et al. 2021) and plant water use strategies and calibrating root water models (Novick et al. 2022). In this study, we test the following hypotheses: (1) ψR obtained from excised root fragments, which is less commonly measured, has correlation with ψS as the soil dried (i.e., increase water stress); and (2) Morphological leaf traits, such as specific leaf area, readily change at varying soil and root water potential. We addressed these hypotheses by testing the model species Chrysopogon zizanoides L. under induced drought and at different growth stages. We specifically chose this plant species for investigation because they have a taproot system, and the roots would typically grow sub-vertically reaching up to 5 m (Hellin and Haigh 2002); this monocotyledon grass root growth pattern makes it ideal to measure ψ gradients at different soil depths and is also favourable for use of slope stabilisation (National research council 1993). We also experimented the soil-plant pots at three different growth periods to test the same hypotheses at different plant ages.

Methods and data processing

Sample preparation and measurement protocol

The accession used in the experiment to test the hydraulic traits is Chrysopogon zizanoides L. (vetiver bunch grass). We grew the grass in 54 cylindrical pots (1 m tall and 0.1 m diameter) filled with clayey sand soil (ASTM D2487-17 n.d.) at 1.4 Mg/m3 dry density. To avoid cluster of roots impeding the ψ measurements, only single grass was fed in each pot. The 54 soil–plant pots were then equally split into three sets and allowed to propagate for three different growth periods of 3-, 4- and 5-months (i.e., 18 pots per 1 growth period; Fig. 1). All the pots were regularly irrigated to the field capacity and sufficient natural light was provided during the entire propagation period. A control pot was also established with same soil and density without any plants. The soil water availability was continuously monitored at three equally distanced elevations in a single pot using moisture content sensors (SM-150 T, Decagon devices) and the raw data was logged continuously for every 5 min.

Fig. 1
figure 1

Test program and plant replicates. a The experiment prepared 54 soil–plant pots, that was equally split into 3 sets of 18 representing 3 plant growth stages (3-, 4- and 5-months). The pots were experimented at different time intervals during the monotonic drying from the soil water status of field capacity (VWC ≈ 35%) to near wilting point (VWC ≈ 10%). b Illustration of pots prepared and soil water status monitoring for the experiment

Full size image

After the plants were grown to the targeted growth period, irrigation was ceased and drought was imposed to each pot. Plants were then sampled at different time intervals representing soil water status from the field capacity to near wilting point (Fig. 1). Before experimentation, the pots were kept in a constant environmental chamber (24 °C temperature; 50% humidity and 50 \(\mathrm{\mu mol }{{\text{m}}}^{-2}{{\text{s}}}^{-1}\) light intensity) overnight to normalise the transpiration demand (vapour pressure deficit of 1.3 kPa) for all the plant samples, irrespective of the available soil water. During experimentation, firstly, the leaves were pruned from the pots and the corresponding leaf biomass was measured. The leaves were then scanned to obtain the surface leaf area using an image analysis software, Image J (NIH, USA). The scanned leaves were kept in a 55 °C oven until a constant weight was achieved to obtain dry leaf biomass. The specific leaf area (SLA) was then calculated by dividing the measured leaf area and dry leaf biomass. After measuring the leaf traits, the 1 m tall pot was then cut into three sections for root extraction. The roots in each section were segregated and the soil adhered to the root was cleaned. The weight (biomass) of the cleaned root was measured and subsequently placed the roots into a potentiometer (WP4C dew-point potentiometer, Meter group) for the direct measurements of ψR. Thereafter, the root diameter and dry root biomass were determined. The root diameter was obtained by the average values of five root fragments that were used to measure ψR in the potentiometer. In addition to plant traits, soil parameters such as volumetric water content (VWC) and gravimetric water content, bulk and dry density in all three sections of all pots were measured after obtaining the plant hydraulic properties.

Soil water release curve and soil water potential mapping

The VWCs from each section of the 54 pots were related to the electric permittivity measured in the moisture sensors (Fig. 2a). The electric permittivity was plotted against theoretical VWC (= gravimetric water content × dry density) to derive calibration curves specific to pots of different plant growth periods and soil depths (Fig. 2b). A unique calibration curve was not found, and we applied linear regression to best-fit the individual curve (all showing P < 0.05 and R2 > 0.9). The regression equation was then inputted with the electric permittivity to derive a calibrated VWC (Fig. 2c). The calibrated VWC ensured the experimental water content is close to the theoretical estimates.

Fig. 2
figure 2

Calibration of volumetric water content from electric permittivity. a Electric permittivity values from moisture sensors plotted against experimental time (in days) from the day of transplanting (May 7 – November 10, 2021). The values were obtained at different stages of drought simulation after 3-, 4- and 5- month grass propagation, measured at three different mean depths in 1-m-high cylindrical pots. b Relationship between electric permittivity value and volumetric water content (regression P-value < 0.05) estimated based on gravimetric soil water content and soil dry density (y-axis). (c) Volumetric water content derived from linear regression equations with electric permittivity value inputs. The leap between 120 and 160th day (September 11 – October 9, 2021) in X-axis at (a) and (c) is due to the experimental halt during monsoon that hindered the drought progression

Full size image

Soil water release curve (SWRC) of a control soil was obtained by measuring the ψS and the corresponding calibrated VWC. The value of ψS of the control soil was measured from 0 to -0.1 MPa using tensiometer and for ψS lower than -1 MPa, WP4C dew point potentiometer was used. From the measured datasets of ψS and VWC, nonlinear regression was made by the mathematical model proposed by Van Genuchten (1980) to procure a complete SWRC for the control soil (Fig. 3a). This measurement could be, however, uncertain in experimental setups where ψS approached the value of cavitation for the case of tensiometers. Besides, continuous measurements of a wide range of ψS between -0.1 MPa and 1 MPa, which are essential to understand both water release characteristics and plant water relations, could not be accurately obtained using any existing measurement technology. To address this information gap, the ψS was mapped from the water content measured in the pots and the SWRC of the control soil. Given that only a single grass was transplanted in each pot, we assumed the rather small root volume existed in the soil would not affect the SWRC of the soil from the planted pots. Hence, the values of ψS of the planted pots were estimated by mapping the calibrated VWC and SWRC of the control soil (Fig. 3b).

Fig. 3
figure 3

Mapping soil water potential. a Soil water release curve (SWRC) measured for clayey sand soil with applied van Genuchten fit (fitting parameters of n = 1.52, m = 0.3421, α = 0.018, air entry value = 13 kPa). b Soil water potential mapped from SWRC of bare soil for the planted pots at all mean depth, plotted against experimental time

Full size image

Measurement of root water potential

The values of ψR for all the root samples was measured using the WP4C dew-point potentiometer. Whilst the thermodynamically principled instrument (similar to psychrometric technique) is commonly used to measure ψS above -1 MPa, its characterisation to measure in plant samples are limited mostly because of the difficulties in maintaining the transient water status and inability to subjectively view the xylem exudation or dew-point formation. This was however, overcome after verifying the sensitivity of ψL measurements and its consensus with existing measurement tools like pressure chamber (Martínez et al. 2013). More so, it is also recently mentioned that the potentiometer could be extended beyond its siloed soil discipline application (Novick et al. 2022), which encourages its exploration in measuring ψ among plant samples. For this reason and other advantages such as direct sample feeding, we attempted to measure the ψR using the dew-point potentiometer. The value measured by the potentiometer is a combination of ψm and ψo components in ψR and is accurate after -1 MPa (because of the nearly linear relationship between ψ and relative humidity before 1 MPa), which is usually the magnitude of ψ in plant cells during soil drying (Donovan et al. 2003).

As placing the entire root length into the potentiometer chamber will overload the capacity of the sampling cup and might not consider the root xylem into account, we conducted a trial experiment (1) to find the number of root fragments required to measure ψR from the potentiometer; and (2) to verify whether the transient water status was maintained during the measurement interval. The roots were cut into smaller fragments that could fit into the sampling cup and could excise the xylem vessels along the root length. In the trial, 12 small-scale soil-plant pots of same accession and soil type were prepared and grown for a month. Thereafter, the roots were excavated from the pots at three different soil water statuses at ψS = -0.5, -0.75 and -1 MPa (four replicates per one ψS value). From the first replicate of each soil water status, the excavated roots were cleaned from the adhered soil and three 1 cm fragments were cut. The biomass of these fragments was measured and were placed into the potentiometer chamber for ψR measurement. After the completion of the measurement, the biomass was measured again to calculate any root water loss during testing. The experimental sequence was similarly followed for other pots, except the number of fragments placed inside the chamber was five, seven and 10 from the second, third and fourth replicates, respectively. No significant difference in the value of ψR was observed when the fragment number was more than five, and the water loss during the measurement interval was as low as 5% from its initial weight. This ensured that there is no water loss during the measurement interval inside the potentiometer chamber and only five root fragments were found enough for each measurement of ψR in the experimental pots.

For the measurements made in the experimental pots, each individual root in the soil section were sampled and cut into fragments for ψR measurement (i.e., the roots were not grouped together to avoid the variability). We also followed several precautions prior and during the measurement to maintain the transient water status: (1) as the root samples were excavated from three cut sections of a single pot, the cut sections were immediately wrapped and kept in a dry ice chamber to avoid any moisture loss from the external environment; (2) the potentiometer was calibrated before each ψR measurement using a standard KCl solution (-2.2 MPa at 24 °C); (3) the roots were not excavated from the soil until it was ready to place inside the potentiometer chamber and was handled only using forceps to avoid any natural contact; and (4) the whole root excavation and cutting into small fragments were carried out within 60 s to minimise its exposure to the environment.

Statistical analysis

All statistical analysis was conducted using the SPSS 25.0 statistical software. The correlations were tested using nonlinear regression (except SLA vs ψR: tested using linear regression) and determined the corresponding goodness of fit, R2. The significance of test results were analysed by one-way analysis of variance (ANOVA), followed by a Bonferroni post-hoc analysis for parameters with same sample numbers and Tukey post-hoc analysis for different sample numbers.

Results

Responses of soil and root water potential under drought stress

To demonstrate the feasibility of direct measurements of ψR in a pressure chamber, the ψR response of Chrysopogon zizanoides L. (vetiver grass, the model species in this study) was obtained using a dew point potentiometer (a thermodynamic principled instrument similar to psychrometer; WP4C dew-point potentiometer, Meter devices). An apparent decline in ψR was observed as the available soil water depleted and correspondingly ψS decreased (Fig. 4a). As the plants progressed towards the permanent wilting point (i.e., -1.5 MPa of ψS), the value of ψR approached an asymptote of -4 MPa. Our results acquired ψR measurements in excised roots from field capacity to near wilting moisture status, and marked a nonlinear (logarithmic) ψS–ψR correlations, regardless of the plant growth periods (three to five months). The non-linearity of ψS–ψR response was observed to originate at the ψS ranged between -0.3 and -0.5 MPa. Further adding to the general response in drought, the coordinated ψS – ψR measurements was compared at different soil depths and plant growth periods. The roots in shallower soil depth consistently exhibited less negative ψR than those found in deeper depths, which essentially should drive water along the root system. The ψR variation at these two soil depths, however showed no significant difference from the one-way ANOVA results (p-value > 0.05). The ψR turned nonlinear faster at greater plant growth period (Fig. 4b), causing earlier wilting.

Fig. 4
figure 4

Water potential measurements connecting soil and root hydraulic continuum. a Relationship between soil and root water potential under monotonic drying. The complete dataset was obtained from 54 soil-plant pots, which are split into 3 equal sets of 18 representing 3 different plant growth periods. The plot also includes the ψ measurement at 2 different soil depths from the 1-m pot. The dataset from each growth period is fitted with nonlinear (logarithmic) regression line that are generated from the 18 pots. b The regression lines generated from the dataset are focused to demonstrate the onset of ψR nonlinearity at different plant growth periods. The pointed marks represent the evident increase in slope and faster approach to ψR threshold for the 5-month grown plants

Full size image

Leaf responses to varying root water potential

The significance of direct measurements of ψR was verified against SLA to substantiate the hydraulic continuum between roots and leaves. SLA showed an apparent decline with ψS (Fig. 5b) and displayed a similarly significant nonlinear (logarithmic) correlation depicted in Fig. 4a. The onset of the nonlinearity of SLA was also observed at the values of ψS ranged between -0.3 and -0.5 MPa, consistent with the range obtained for the onset of the nonlinearity of ψR. This consistency confirms the continuity between the root and leaf hydraulics during soil drying. Besides the general drought response, SLA did not show any significant difference among different plant growth periods (p-value > 0.05). To examine the linkage between root and leaf hydraulics, we observed SLA reducing with reducing ψR, meaning that the dry matter in the leaves was increasing as ψR progressed to drop. The linearity was found between SLA and ψR was well perceived within the 90% confidence interval (p-value < 0.05) from the onset of the nonlinearity of ψR.

Fig. 5
figure 5

Plant leaf area and correlations of specific leaf area with soil and root water potentials. a A box plot of vetiver (Chrysopogon zizanoides L.) bunch grass leaf area at 3- (n = 18), 4- (n = 18) and 5-month (n = 18) growth periods. The decrease in leaf area at 4-month growth period compared to 3-month and 5-month was presumably attributed to the negative influence of humid weather in rainy seasons on leaf area development. One-way ANOVA test showed statistically significant difference (P-value < 0.05) among the leaf area in different growth periods, marked by mean discrimination letters from post hoc Tukey analysis. b Relationship between ψS and specific leaf area fitted with the generalized nonlinear (logarithmic) fitting from all the pots (c) Relationship between ψR and specific leaf area fitted with a moderate linear regression (straight dotted line). A 90% confident interval (CI) was calculated and plotted from the linear regression generated by the dataset (positive and negative CI plotted at either side of the straight dotted line). Both the plots include ψ measurements only from the deeper soil depths

Full size image

Root water potential at varying root diameter

To identify a trait-based relationship with the measured ψR, a correlation between root diameter and ψR was obtained. It should be emphasized again that the trait and hydraulics of the roots were measured individually, and the roots were not grouped together. The root diameter was obtained by the average values of root fragments that were measured for ψR in the potentiometer. The diameter measured in all our pots ranged between 0.3 mm and 0.6 mm (Fig. 6a) and showed no significant difference among the plant growth periods (p-value > 0.05). In general, the ψR was observed to be less negative for larger root diameters and more negative for smaller roots (Fig. 6b). Upon fitting, the dataset displayed a negative power correlation.

Fig. 6
figure 6

How influencing is root diameter to root water potential?. a A box plot signifying variation in root diameter at two different soil-depths of 1-m pot (n = 54 (18 in each growth period) for 0.1 – 0.4 m; and n = 29 (9 in 3-month; 11 in 4-month and 9 in 5-month). One-way ANOVA showed no statistically significant difference (P-value > 0.05) among the plant growth periods. b Relationship between ψR and root diameter from all the soil-plant pots fitted with a generated nonlinear (second-order power) regression

Full size image

Discussion

Water potential response interprets soil-root hydraulic pathway

From a physical point of view, our results allow to visualise the gradient between ψS and ψR, which are known to be linear in wet soils and gradually becomes nonlinear as the soil dries (Javaux and Carminati 2021; Koch et al. 2018). Although any complementary results of the direct ψR measurements could not be juxtaposed because of the data limitations, our results may be used to interpret the existing findings of root hydraulic conductance (kroot; determined from the relationship between transpiration and xylem ψL) and ψ at the soil-root interface (ψS-R). Cai et al. (2022) reported that kroot markedly declined with decreasing ψS for six different crop species, attributable to several reasons including the downregulation of aquaporin (Caldeira et al. 2014), ABA signalling (Maurel and Nacry 2020), formation of lacunae in the root cortex (Cuneo et al. 2016) and/or a root shrinkage (Rodriguez-Dominguez and Brodribb 2020). We similarly conceptualise that all or one of these underlying mechanisms could have also resulted in our observed steady decline in ψR. In fact, the drop in ψ was captured in the immediate vicinity of the soil-root interface (Cai et al. 2022), that closely resembles with the drop of ψR in our results. The continuous decline in ψR later approached to an asymptote of -4 MPa after prolonged soil drying (Fig. 4a). This near flattening of ψR responses could be ascertained to the ceasing of transpiration leading to gradual stoppage of root water uptake (Sperry et al. 2002; Garg et al. 2020). While the actual mechanics or physiological changes in roots behind this transpiration limiting root water uptake is still unclear, studies indicate this response might be the influences of structural modification in roots (Drew 1987), root xylem embolism (Steudle 2001) and development of vapour gaps (Faiz and Weatherley 1982).

The nonlinearity of ψS–ψR correlation demonstrates the regulation of water uptake as the plants experienced water stress. The nonlinear correlation of the total ψ difference between soil and roots under the influence of applied root pressure is well known since the theoretical and experimental conclusions drawn from Fiscus (1975). Steudle (2000) furnished the integration of root anatomical views from the observations of kroot that have definitively demonstrated nonlinear flow relations in water transport pathways of apoplast or symplast. More recently, Cai et al. (2022) found similar nonlinear response for their findings of both kroot and ψS-R during soil drying. Apart from the encompassing of transpiration and stomatal regulation in different plant species (Abdalla et al. 2021; Hayat et al. 2020), the investigation conducted by Cai et al. (2022) incorporated the observations from different soil textures to identify the effects of ψS on the nonlinearity in root hydraulics. The study also defined a critical value of ψS that corresponds to the onset of the nonlinearity of kroot, where the soil water fluxes start to limit in the soil-plant continuum and triggers stomatal closure (Carminati and Javaux 2020). On this basis, we integrated the critical ψS concept with our results for the onset of the nonlinearity of ψR and found the value of critical ψS ranged between -0.3 to -0.5 MPa (Fig. 4a). The critical range of ψS observed was found to be similar to the findings of Cai et al. (2022), for the crops that were grown in sandy loam soil (same soil type used in our study) and was consistent with where the values of kroot started to decline.

When interpreting the ψR observations along the root length, the root in deeper soil depths exhibited less negative ψR (Fig. 4a), which is possible because of the differentiation of xylem vessels during the developmental stages of plant growth (Varney and Canny 1993; Bramley et al. 2009). The ψR variation at two different soil depths, however, deemed no significance (p-value > 0.05), indicating a smaller gradient along the root xylem. This observation again concurs with root anatomical findings of applied root pressure and drought induced difference in axial kroot (i.e., water movement in the xylem), where a significantly greater difference was observed only at root tip (apical meristem) and not along the entire root length (Frensch and Steudle 1989; Huang et al. 1995). Steudle and Peterson (1998) inferred this difference in axial kroot by showing the maturity of both late and early metaxylem vessels at root tip and the maturity of only the early metaxylem vessels beneath the root tip. Huang and Nobel (1992) reported similar results for the rootstocks of Agave deserti L., a monocotyledonous plant species resembling the accession experimented in this study. All this anatomical evidence implies that the difference in measured ψR might be because of the xylem vessel maturity at different soil depths, which were possibly close enough to be read from the direct measurements of potentiometer. Whilst this root anatomical changes and its influence in ψR might not be often intrusively determined in transpiring plants, existing high-resolution neutron radiography showed similar observation of non-uniform root water uptake along the root system (Dara et al. 2015; Zarebanadkouki et al. 2016). Having connoted that, it should be emphasised that such uniformity or non-uniformity of water uptake along root length may vary with different plant functional groups (such as dicots) and its root anatomical development (Roumet et al. 2016). We also deem that the insignificant difference in ψR at the two soil depths could be potentially affected by the cease of plant transpiration due to the cut-off of root segments, which is always inevitable for any kind of direct, intrusive ψR measurements. The other observation of faster progression to non-linearity in higher plant growth period signifies the influence of root age in measured ψR (Fig. 4b). This is probably because of the significant increase in leaf area (Fig. 5a) over the plant growth period, which might have demanded more water uptake in the plants grown for 5-months. The results were consistent with the findings of Lopez and Nobel (1991), where the kroot first increased and later decreased with increase in plant age.

Exiting water at leaf concurs with measured root water potential

Water exiting from plant leaves by means of transpiration completes the soil-plant hydraulic continuum, which over the years, has been thoroughly deciphered by the quantitative features of actual transpiration and stomatal conductance (gS). Using these leaf hydraulic parameters, typical responses of plant functioning and regulation to nearly all abiotic stresses can be captured (Negin and Moshelion 2016; Mittler 2006). Indeed, one of these leaf hydraulic parameters would have been suitable if we sourced the continued response of ψ from one single pot, but because our measurements were obtained destructively from different pots and the leaf area in each pot varied (p-value > 0.05; Fig. 5a), the transient response of gs or actual transpiration to drought would be deficient to correlate with the measured ψR. We addressed this methodological gap by measuring SLA to determine the root-leaf hydraulic continuum. SLA is a normalised leaf parameter that is defined by the ratio of surface area of the leaf and its dry mass (expressed in m2 kg−1). SLA has been found to have strong links with several environmental conditions including drought than other leaf traits (Messier et al. 2010). With respect to leaf water relations, low SLA means more dry matter per leaf area and often correlates with low rates of stomatal water loss (Shipley et al. 2005); high SLA, on the other hand, means less dry matter per leaf area and adopts a more “disposable” stomatal water loss strategy (Dwyer et al. 2014).

Notwithstanding its limited interpretation, firstly, SLA displayed the apparent decline with negative ψS (Fig. 5b) and a similar nonlinear (logarithmic) correlation as depicted in ψS–ψR response (Fig. 4a). Secondly, the onset of the nonlinearity of SLA was also observed at the values of ψS ranged between -0.3 and -0.5 MPa, consistent with the onset of the nonlinearity of ψR. As grass species (used in this study) are known for their lack of strong stomatal control (Bollig and Feller 2014), the decline in SLA may represent the phenotypic adjustment to the applied drought (Wellstein et al. 2017). The reasoning, however, cannot be interpreted with gS or other transpiration parameter, existing studies showed strong correlation between the reduced SLA and decline of ψL (Pérez-Ramos et al. 2013) and also demonstrated a enhanced water-use efficiency under the water stress conditions (Ackerly 2004; Wright et al. 2001). Despite well capturing the general response, SLA did not show any significant difference among different plant growth periods (p-value > 0.05). The correlation between measured ψR and SLA (Fig. 5c) explained the hydraulic continuum between roots and leaves by displaying a negative linear relationship. The linearity was especially well perceived within the 90% confidence interval (p-value < 0.05) from the onset of the nonlinearity of ψR. Given that the reduction of SLA likely corresponds to the phenotypic adaptation of the grass species (Bollig and Feller 2014), we conceptually presume that the negative linearity identified could be attributable to stomatal regulation after the plants experienced the water stress. Whilst there is no direct experimental proof to test the hypothesis, the findings of kroot have linked to the control of stomatal closure from the onset of its nonlinearity (Abdalla et al. 2021, 2022). These existing findings support the fact that both the decline and nonlinearity of the measured ψR were linked to the leaf hydraulics from our measurements of SLA.

Root water potential driven by root traits

It is well established that root water uptake varies with factors like root architecture (Doussan et al. 2006; Javaux et al. 2008), root hairs (Carminati et al. 2017; Marin et al. 2021), root types (Ahmed et al. 2016, 2018) and other physical traits, and is usually challenging for both modelling and experimental investigations. Existing experimental studies, however, considered the aforementioned factors collectively as a measure of root hydraulics and partially addressed the complexity of root distribution (Cai et al. 2022). Some existing studies have also qualitatively demonstrated the effects of root length and root distribution on the process of root water uptake by imaging such as radiographs (Zarebanadkouki et al. 2016). However, the influence of root diameter which signifies the capacity of root xylem in radial root water uptake and extent in axial root water uptake (Kirfel et al. 2017) is underreported. In our results, no statistically significant difference was observed between the root diameters of different soil depth and among different growth periods (p-value > 0.05; Fig. 6a). The ψR was observed to be less negative for larger root diameters and more negative for smaller roots (Fig. 6b), displaying a negative power correlation. The observed trend of reduction in ψR for roots with smaller diameter could be because of (1) the shrinkage of roots during soil drying (Rodriguez-Dominguez and Brodribb 2020) (2) the exceedance of the water uptake demand at greater plant growth periods due to increased leaf area (Fig. 5a) and underdeveloped root diameter (Fig. 6a) and (3) thicker roots tend to have more xylem vessels and could be the reason larger root diameters measured less ψR (Aloni and Zimmermann 1983; Tyree et al. 1994). Although the first reason could not be illustrated without the information on root growth development over time, the latter coincides with our observation in Fig. 4a where the nonlinearity of ψR and plant wilting reached earlier at greater plant growth periods. A similar observation was reported in Abdalla et al. (2022), where the nonlinearity of transpiration was found at shorter root systems. Overall, our results from the measured root diameter and its significant correlation with ψR explain the decline in ψR (Fig. 6b) and discover a new experimental information on the correlation between root diameter and root water uptake.

Conclusion

Our study on direct ψR measurement encapsulated the ψ gradient between soil and roots during root water uptake to the significant logarithmic correlation between them under the imposed drought stress. Our subsequent exploration on the correlation of leaf hydraulics using SLA observed mirroring patterns of nonlinearity. The strong association with other existing information on root hydraulics and root traits supports the observations and calls for future studies to address the stumbling block of direct ψR measurement using potentiometer. Given all the positive outcomes, it is also important to specify some of the limitations such as the difficulties of obtaining transpiration parameters and the inability to source wider ranged root diameters and lengths for a complete correlation with exiting water flux and trait-based understanding, respectively.