1 INTRODUCTION

Terrestrial plants exposed to drying soils suffer from a reduced water potential which can lead to cell turgor loss (Bartlett et al., 2012), hydraulic failure of xylem vessels (Urli et al., 2013) and impaired phloem transport (Thompson, 2006). To mitigate this, plants close stomata at the cost of reduced CO₂ gain, up to carbon starvation (McDowell, 2011). In the first place, the reduced plant water potential is caused by the drop in soil water potential. However, the plant water potential additionally depends on how the hydraulic conductance of the root system is affected by soil drying (Bourbia et al., 2021; Nobel & Cui, 1992; Rodriguez-Dominguez & Brodribb, 2020; Saliendra & Meinzer, 1989). In a recent study, we showed that even moderate soil drying can induce an exponential decline of the total root system conductance (Müllers, Postma, Poorter, & van Dusschoten, 2022). Considering the distribution along the root system of such a net decline is crucial to understand its impact on the plant water balance.

Soil drying is a highly nonuniform process and usually results in a pronounced vertical gradient with relatively dry shallow soil layers and relatively wet deep soil layers (Hillel et al., 1976; Kondo et al., 2000; Markesteijn et al., 2010). Under such conditions, water in deeper layers is much easier to extract, that is, a deep root requires a less negative xylem water potential to realize a certain water uptake rate compared with an otherwise identical shallow root. Following this rationale, effectively using deep water resources is a key to withstand droughts (Wasson et al., 2012). However, several studies emphasize that plants often fail to do so and experience severe drought stress despite a relatively high water availability in deeper root zones (Gessler et al., 2022; Passioura, 1983; Prechsl et al., 2015; Rasmussen et al., 2020). A major reason is that the root conductance (K_root, see Table 1 for abbreviations) in deeper layers usually is low since roots are less abundant (Haberle & Svoboda, 2015; Kemper et al., 2020; Righes, 1980) and less conductive compared with shallow roots (Dara et al., 2015; Müllers, Postma, Poorter, Kochs, et al., 2022; Zarebanadkouki et al., 2013). Therefore, a more effective acquisition of deep water would require to locally increase root conductance over time. This can be achieved by increasing root length via additional root growth in deeper layers, which has been often observed upon soil drying (Alsina et al., 2011; Asseng et al., 1998; Dubrovsky et al., 1998; Rodrigues et al., 1995; Sharp & Davies, 1985). Nevertheless, as root hydraulic traits vary among individual roots and entire root systems (Ahmed et al., 2016, 2018; Clément et al., 2022; Müllers, Postma, Poorter, Kochs, et al., 2022; Rewald et al., 2012; Steudle & Peterson, 1998), local root length only partly determines a plant's ability to use deep water. Additionally, plants can increase the intrinsic root conductivity (conductance per length) by, for example, increasing the amount of active aquaporins (Johnson et al., 2014; McLean et al., 2011).

While in deeper, wetter layers, increasing the root conductance is a reasonable strategy for an effective water usage, it might be the other way around in upper, drier soil layers. Unregulated water uptake from drying soil can result in a severe water depletion zone around the roots, strongly reduce the local soil conductivity, and thus restrict water flow from the bulk soil towards the roots (Carminati & Javaux, 2020). The drop of the local soil conductivity scales with the water potential of the bulk soil and the water uptake rate per unit root length. Therefore, locally decreasing root conductance, and thus water uptake rates, in drier soil layers might be beneficial to avoid an interruption of the hydraulic pathway between remaining water in the bulk soil and the root surface. Mechanisms like enhanced root suberization (Barrios-Masias et al., 2015; Cruz et al., 1992; Lo Gullo et al., 1998; North & Nobel, 1991), or reducing the amount of open aquaporins (Martre et al., 2001; Rodríguez-Gamir et al., 2019) enable plants to reduce root hydraulic conductance during droughts.

Following these considerations, we analyzed how previously measured changes in the total root system conductance (Müllers, Postma, Poorter, & van Dusschoten, 2022) are distributed over depth. Although this spatial aspect is highly relevant for the plant water balance, it has been poorly investigated. The few studies measuring a spatial component of soil drying-induced alterations of root hydraulic traits used single roots, sampled at different depths (Johnson et al., 2014; Wan et al., 1994). Additional information comes from studies on partial root zone drying in which root hydraulic traits were either measured on single, sampled roots (McLean et al., 2011) or entire root systems divided into halves (Hu et al., 2011). These studies generally suggest that the conductance of roots in rather wet soil increases relatively to that of roots in rather dry soil. However, how the full vertical distributions of root hydraulic conductance change in response to soil drying is unknown, probably due to limitations in the available measurement technologies. Here, we used a recently developed, highly precise soil water sensor, called soil water profiler (SWaP; van Dusschoten et al., 2020), to noninvasively measure root water uptake profiles which reflect the distribution of root hydraulic conductance. We hypothesize that the root conductance decreases in upper soil layers while it increases in deeper layers as an early response to soil drying by means of active regulation by the plant. Such a response would reduce water stress by facilitating deep water usage and sustaining a hydraulic connection to remaining water resources in drier layers. In Figure 1, the hypothesized effect of soil drying on the spatial distribution of root conductance is summarized. The hypothesis was tested on pot-grown faba bean and maize plants during 4 days of soil drying. These two species were chosen because they differ in root system architecture, and water uptake rates per unit root length. Faba bean has a taproot system with comparably great water uptake rates per unit root length, whereas maize has a fibrous root system with comparably low water uptake rates per unit root length. Both factors potentially affect the distribution of remaining soil water under drought, and thus the resulting alterations of the local root conductance.

2 MATERIALS AND METHODS

Data used in this study were obtained during an experiment described previously in Müllers, Postma, Poorter, and van Dusschoten (2022). Below, we shortly summarize the experimental design and explain the determination of the distribution of root conductance and root length in more detail.

3 RESULTS

Withholding water for 4 days resulted in a progressive, significant reduction in both the soil water content ($$\bar{{\theta }}$$) and soil water potential ($$\bar{{{\rm{\Psi }}}_{\mathrm{soil}}}$$) averaged over depth (Table 3). For both species, $$\bar{\theta }$$ decreased from 16 mL cm⁻³ at Day 1 to 8 mL cm⁻³ at Day 4. $$\bar{{{\rm{\Psi }}}_{\mathrm{soil}}}$$ decreased from −0.01 MPa at Day 1 to −0.09 MPa (faba bean), and −0.07 MPa (maize) respectively, at Day 4.

Figure 2 shows how these reductions were distributed over depth. At Day 1 after withholding water, for both species, there was a vertical gradient in θ with drier soil layers in the top and wetter layers in the bottom (Figure 2a,c). θ ranged from 12 mL cm⁻³ (at 5 cm depth) to 20 mL cm⁻³ (44 cm) for faba bean and from 12 mL cm⁻³ (6 cm) to 23 mL cm⁻³ (44 cm) for maize (Table 4). In each individual soil layer, θ was significantly (p < 0.01) reduced from each day towards the next one for both species. At Day 4, θ ranged from 5 mL cm⁻³ (7 cm) to 11 mL cm⁻³ (38) cm for faba bean and from 5 mL cm⁻³ (5 cm) to 10 mL cm⁻³ (38 cm) for maize. Like θ, Ψ_soil in each individual soil layer decreased significantly (p < 0.01) from each to the next day for both species (Figure 2b,d). At Day 1, Ψ_soil was almost uniformly distributed, with only a slight vertical gradient (around −0.02 MPa in the upper half of the pot, and −0.01 MPa in the lower half for both species). However, proceeding soil drying led to an increase in the vertical gradient: At Day 4, Ψ_soil ranged from −0.23 MPa (at 7 cm depth) to −0.03 MPa (38 cm) for faba bean, and from −0.14 MPa (5 cm) to −0.04 MPa (38 cm) for maize. In conclusion, although the vertical gradient in θ even decreased during soil drying, the gradient in Ψ_soil increased strongly. This is due to the nonlinear character of the water retention curve: in the dry regime, a small reduction in θ causes a strong reduction in Ψ_soil.

In the next step, we analyzed how the changes in Ψ_soil affected root water uptake patterns measured as the normalized plant-driven root water uptake distributions ($${\hat{U}}_{{\rm{P}}}$$), using the SWaP. At Day 1, for both species there was a vertical gradient in $${\hat{U}}_{{\rm{P}}}$$ with greater values in upper soil layers and lower values in deeper layers (Figure 3). $${\hat{U}}_{{\rm{P}}}$$ was not constant over time: Generally, $${\hat{U}}_{{\rm{P}}}$$ decreased in the drier, upper soil layers, while it increased in relatively wetter, deeper layers. For the parameters described below, in deeper layers we observed a consistent trend from Days 1 to 3, reversing from Days 3 to 4. Therefore, we primarily tested for significant differences between Days 1 and 3. As indicated by a different background colour in Figure 3, we observed the following significant changes: For faba bean, $${\hat{U}}_{{\rm{P}}}$$ decreased in the top 8 cm (p < 0.01) and increased between 11 and 38 cm depth (p < 0.05 at 11–13 cm depth, <0.01 at 13–38 cm depth). For maize, $${\hat{U}}_{{\rm{P}}}$$ decreased in the top 10 cm (p < 0.05 at 10 cm depth, <0.001 else) and increased below 20 cm depth (p < 0.05 at 20–23 cm depth, <0.01 below).

Alterations of $${\hat{U}}_{{\rm{P}}}$$ are caused by a shift in the distribution of root hydraulic conductance (K_root). To derive the distribution of K_root over depth, we multiplied the daily average of the total conductance, as determined in a previous study (Müllers, Postma, Poorter, & van Dusschoten, 2022) (Table 2), with the daily $${\hat{U}}_{{\rm{P}}}(z)$$, according to Equation (5). The total conductance at the 1st day was similar for faba bean and maize (8–9 mL h⁻¹ MPa⁻¹ Table 2). At Day 1, for both species, there was a vertical gradient in K_root with greater values in upper soil layers (54% in the upper 10 cm for faba bean, 47% for maize) and lower values in the bottom (13% in the bottom half for faba bean, 21% for maize) (Figure 4). Withholding water for 4 days resulted in a strong reduction of K_root in upper soil layers with more negative Ψ_soil: For faba bean, K_root significantly decreased from Days 1 to 3 in the entire upper half of the pot (p < 0.01) by, on average, 72% (Figure 4a). In the lower half, K_root did not change significantly. For maize, K_root significantly decreased from Days 1 to 3 in the upper 16 cm of the pot (p < 0.01 above 12 cm depth, <0.05 else) by, on average, 66% (Figure 4b). However, between a depth of 31 and 41 cm, K_root increased significantly (p < 0.01 at 35–37 cm depth, <0.05 else) by, on average, 107% (Figure 4c). With proceeding soil drying from Days 3 to 4, K_root decreased again in these layers, such that it was not significantly different compared with Day 1.

The temporarily increased K_root in deeper parts of the maize root systems supports the uptake of deep soil water resources. We estimated the relevance of this facilitated deep water uptake by calculating how much it contributed to the total root water uptake (Figure 5). For maize, the measured increases in K_root contributed 5% to the total root water uptake rate at Day 2, 11% at Day 3, and 21% at Day 4. In different words, without the enhanced K_root, only 80% of the measured water uptake rate might have been realized at Day 4. Note that the hypothetical uptake rates (grey bars in Figure 5) were calculated by setting all measured increases in K_root to zero and assuming that this would not affect the leaf water potential, implying a strict stomatal control of the plant water status. More realistically, a lack of the increases in K_root would not be fully reflected in a reduction in total water uptake rate, but also partially in leaf water potential. However, mitigation of both, the drop in total root water uptake rate and in leaf water potential comes with reduced plant water stress. For faba bean, measured increases in K_root at each day contributed significantly less to the total root water uptake rate compared with maize (p < 0.05).

The determination of K_root, along with the associated increases in deeper layers for maize (Figures 4 and 5), are based on a correct separation of U_P and U_S by the SWaP measurements (Equations 1-3). Otherwise, the local increases might also be explained by an altered soil water distribution and the resulting compensated root water uptake. Although theoretically, U_P and U_S were separated using the fluctuating light intensity, we evaluated the potential effect of compensated root water uptake on U_P by comparing the changes in U_P and U_S between Days 1 and 3 in deeper layers for maize (Supporting Information: Appendix 2 and Figure 4). Below 20 cm depth, the measured U_P increased between Days 1 and 3 (Supporting Information: Figure 4A,D). Below 30 cm depth, this increase was of the order of 0.2 mL h⁻¹ (Supporting Information: Figure 4C,F). At the same time, the estimated U_S also increased in these layers (Supporting Information: Figure 4B,E) indicating enhanced compensated root water uptake. However, U_S at Day 3 was only of the order of 0.02 m h⁻¹, and thus a factor 10 smaller than the increase in U_P. Calculation of U_S assuming no local increases in K_root (U_S′, see Supporting Information: Appendix 2) gave even smaller values (Supporting Information: Figure 4B,E). In conclusion, even if U_P and U_S were not separated correctly by our approach, contribution of U_S to the increased U_P in deeper layers was marginal, such that the increase in K_root (Figures 4 and 5) remained.

We analyzed whether the locally enhanced K_root in maize was caused by additional root growth or an increase in root conductivity (conductance per length, k_root). Root length distributions at each day were derived from MRI images before and after the 4 days, making use of exponentially interpolating the data (Figure 6a–c). Then we derived k_root profiles at each day by dividing K_root by the root length distributions (Figure 6d–f). Initially, faba bean roots had a much greater conductivity (0.19 mL h⁻¹ MPa⁻¹ m⁻¹ on average over depth) than maize roots (0.05 mL h⁻¹ MPa⁻¹ m⁻¹ on average over depth). Like for the other parameters analyzed above, at Day 1, there was a vertical gradient in k_root. Shallow roots of both species had a greater conductivity than deeper roots. Withholding water for 4 days resulted in a decreasing k_root of shallow roots, in the faster drying, upper soil layers. In faba bean, k_root decreased significantly from Days 1 to 3 in the upper 29 cm of the pot (p < 0.01) by, on average, 76% (Figure 6d). Below 30 cm, k_root remained mostly constant. In maize, k_root decreased significantly from Days 1 to 3 in the upper 15 cm of the pot (p < 0.01 above 12 cm depth, <0.05 else) by, on average, 66% (Figure 6e). Between a depth of 27 and 35 cm, however, k_root increased significantly (p < 0.05) by up to 81% (Figure 6f). Like K_root, k_root from Days 3 to 4 decreased again in these layers. Distributions of k_root for each individual maize plant during the 4 days are shown in Supporting Information: Figure 5. For each plant, there were local increases in k_root, indicated by green dots, during the 4 days. The depth and day at which these increases occurred, however, varied strongly among the different replicates, causing the limited significance observed in Figure 6e. Using Equations (8-10), we calculated how much changes in root length, and k_root contributed to the significantly increased K_root at 31–41 cm depth for maize (Table 5). In each of these soil layers, the increase in K_root was primarily caused by an increase in k_root (relative contribution >0.75 in 8 out of 10 layers). Note that in some layers, root length even decreased slightly, causing the negative values for the contribution of root length and values >1 for the contribution of root conductivity in Table 5.

4 DISCUSSION

The spatial distributions of root hydraulic traits and how they are affected by soil drying has important implications for the plant water balance. The strong reduction of soil water potential in upper soil layers led to a significant reduction of the local root hydraulic conductance. In maize, this was partly compensated by an increased conductivity (conductance per length) in deep roots. This enabled plants to maintain an estimated 20% greater total water uptake rate compared with a scenario without local increases in root conductance.

Initially, root conductance in upper soil layers was much greater (87% in upper half of the pot for faba bean, 79% for maize) than in deeper soil layers (Figure 4). This is typical for well-watered conditions since shallow roots are usually more abundant (Haberle & Svoboda, 2015; Kemper et al., 2020; Righes, 1980) and more conductive than deeper roots (Dara et al., 2015; Müllers, Postma, Poorter, Kochs, et al., 2022; Zarebanadkouki et al., 2013). Consequently, withholding water resulted in a considerable vertical gradient in soil water potential with more negative values in the top and less negative values in the bottom (Figure 2). In response to soil drying, we observed a significant decrease in K_root in upper, relatively dry soil layers for both species (Figure 4). One plausible reason for this decrease is a partial loss of soil root contact due to root shrinkage, which was shown to be initiated at a soil water potential of around −0.02 MPa for faba bean (Koebernick et al., 2018) and maize (Duddek et al., 2022). Therefore, in upper soil layers, the relatively low (more negative) Ψ_soil might have induced root shrinkage in our study, resulting in the observed reduction in K_root, whereas in deeper layers, Ψ_soil was still sufficiently high (less negative). Nevertheless, the critical Ψ_soil at which we observed significant reductions in K_root was around −0.04 MPa (compare Figures 2 and 4), and thus slightly more negative than reported by the two studies mentioned above. Besides a loss of soil-root contact, K_root can also decrease due to enhanced root suberization and reducing the amount of active aquaporins, two mechanisms which are under biological control of the plant. Soil drying reportedly led to a decreased expression of aquaporin genes in shallow roots (Johnson et al., 2014), and an enhanced suberization of, especially, basal root parts (Kreszies et al., 2019) which predominantly reside in the top soil. Both processes might have contributed to the observed reduction in K_root in upper soil layers (Figure 4).

We assumed that the soil conductance was much greater than the root conductance to approximate K_root by K_SL (Equations 4-6 and 5). Estimations of K_soil (Supporting Information: Appendix 1) showed that this approximation was justified for a large part of the measurement (Supporting Information: Figure 3). Only in upper soil layers for faba bean at Day 4, K_soil was of the same order of magnitude as K_root (Supporting Information: Figure 3I). There, up to 8% of the decline in K_root from Days 3 to 4 as shown in Figure 4 were caused by a reduction in K_soil. Nevertheless, this effect was of minor importance, as K_root was already reduced by 72% on average between Days 1 and 3 in the upper half of the pot.

Theoretically, unregulated water uptake from drying soils can lead to steep water potential gradients in the soil around roots and thus interrupt the hydraulic pathway from the bulk soil towards the roots, and force stomatal closure (Carminati & Javaux, 2020). Our estimations of the water potential at the root surface (Supporting Information: Figure 3A,C) emphasize that such a severe water depletion zone did not occur during the measurements presented here. Most likely, the early local reduction in K_root, due to the mechanisms discussed above, and the resulting stomatal closure, could prevent an incisive drop in the local K_soil. These conclusions on the effect of K_soil initially apply only to the soil type used in this study, a loamy sand. In this context, a recent study showed that the total conductance between soil and plant decreased at a less negative Ψ_soil in a sandy soil compared with loamy soil, probably because the loamy soil sustained a greater conductivity at a given Ψ_soil (Cai et al., 2022). Additionally, the effect of a reduced soil-root contact depends on soil texture, and might be more pronounced in the loamy sand used here, compared with a sandy soil (Carminati et al., 2009).

Despite the potential benefit of preventing a severe water depletion zone around the roots, the strong reduction in K_root in upper soil layers (76% reduction in faba bean, 66% reduction in maize [Figure 4]) comes with different impairments for the plant: A reduction in K_root hampers the hydraulic redistribution of soil water from relatively wetter towards relatively drier soil layers through the roots (Neumann & Cardon, 2012). Moreover, as demonstrated in a previous study, the reduction in K_root amplifies the drop in plant water potential (Müllers, Postma, Poorter, & van Dusschoten, 2022). One possibility to partly compensate for this is deep-water uptake. Up until Day 3, water in the bottom third of the pots was much easier to access (Ψ_soil less negative than −0.03 MPa) than water in shallow soil layers (Ψ_soil locally as low as −0.15 MPa [faba bean] and −0.08 MPa [maize] [Figure 2]). However, in agreement with other studies (Clément et al., 2022; Dara et al., 2015; Zarebanadkouki et al., 2013), effectively taking up deep water was initially limited by a low local K_root (Figure 4). With proceeding soil drying, maize, but not faba bean was able to significantly increase K_root in deeper, comparably wet soil layers. The deep-water uptake, facilitated in this way, contributed up to 20% to the total root water uptake rate in maize (Figure 5) and thus, at least temporarily, alleviated the drop in plant water potential.

Deep-water uptake under drought can increase due to a local increase in root conductance, but also due to compensated root water uptake, quantified as U_S. This happens when the local Ψ_soil drops less than the global Ψ_seq which was the case in deeper soil layers in our study (compare Figure 2 and Table 2). Theoretically, we had separated U_P and U_S using the fluctuating light intensity which, on short time scales, only changes U_P but not U_S. To not fully rely on the success of this separation, we additionally estimated which impact it had for our conclusions on K_root if compensated water uptake had influenced the measured U_P (Supporting Information: Appendix 2). Since the estimated U_S at Day 3 in deeper layers in maize was a factor 10 smaller than the measured increase in U_P (Supporting Information: Figure 4), we conclude that the measured local increases in K_root largely remained, even if the separation of U_P and U_S had failed.

Normalization by root length indicated that the increase in K_root was mainly caused by a significantly increased root conductivity (conductance per length) (Figure 6). One potential reason for this phenomenon is directed, enhanced aquaporin gene expression, as observed by Johnson et al. (2014) or McLean et al. (2011). By relating empirical data on Ψ_soil, ABA concentration in roots, and its impact on aquaporin expression, Couvreur et al. (2015) estimated a 250% increase in k_root between a local Ψ_soil of −0.005 and −0.1 MPa. This effect would lead to a net increase in k_root as long as it is not outweighed by a reduced soil-root contact and other forementioned processes reducing k_root. The fact that increases in k_root only occurred above a Ψ_soil of −0.04 MPa in our study (compare Figures 2 and 6e,f) fits these considerations. Furthermore, the different abilities of faba bean and maize to increase K_root might be explained by root aquaporin regulation as well: Under nonstressed conditions, radial water uptake in bean occurs predominantly via the cell-to-cell pathway, mediated by aquaporins, whereas in maize, the apoplastic pathway contributes considerably (Javot & Maurel, 2002; Steudle & Brinckmann, 1989; Steudle & Frensch, 1989). This fits the initially greater conductivity of faba bean roots compared to maize roots (Figure 6a,b at Day 1). To compensate this, maize had a greater root length (Figure 6b,c), thus achieving a similar initial conductance (compare Figure 4a and 4b at Day 1) with a greater flexibility and potential for local increases via a facilitated cell-to-cell water transport.

However, the observed increase in k_root could have also been caused by xylem maturation. As summarized by M. McCully (1995), maturation of the late metaxylem can occur far behind the root tip (up to 10–50 cm) for various species, including maize (St. Aubin et al., 1986) and soybean (M.E. McCully, 1994). Since maturation of the metaxylem results in an increase in axial conductivity, the measured increase in k_root in maize possibly reflects the development of xylem vessels. This would imply that the axial conductivity had initially limited root water uptake in deeper soil layers, as suggested by previous studies (Clément et al., 2022; Sanderson et al., 1988; Strock et al., 2021). However, the strong increase in local k_root (up to 82%) indicates a simultaneous response of all roots of various growth stages within the respective soil layers, rather than a continuous developmental process. Additionally, Steudle & Peterson (1998) reported for maize that even though the late metaxylem is not fully developed within 25 cm from the root tip, axial conductance is orders of magnitude greater than radial conductance, except for a small apical region without developed early metaxylem. We conclude that an increased amount of active aquaporins is a plausible explanation for the measured local increases in k_root, which could have been enhanced by other mechanisms, such as xylem development.

5 CONCLUSION

We analyzed the effect of soil drying on vertical profiles of root hydraulic conductance in faba bean and maize. Withholding water for 4 days resulted in a pronounced vertical gradient in soil water potential with drier layers in the top and wetter layers in the bottom. In drier, shallow soil layers, the reduced soil water potential caused a strong decrease in root conductance, which on the one hand prevented a severe drop in soil hydraulic conductivity around the roots, on the other hand amplified water stress and forced stomatal closure. To partly compensate for this, maize, in contrast to faba bean, was able to facilitate deep water uptake by locally increasing its root conductivity. This increase improved the overall water uptake rate, and thus is an effective plant strategy to reduce water stress during soil drying.