By continuing to browse this site you agree to us using cookies as described in About Cookies
Notice: Wiley Online Library will be unavailable on Saturday 7th Oct from 03.00 EDT / 08:00 BST / 12:30 IST / 15.00 SGT to 08.00 EDT / 13.00 BST / 17:30 IST / 20.00 SGT and Sunday 8th Oct from 03.00 EDT / 08:00 BST / 12:30 IST / 15.00 SGT to 06.00 EDT / 11.00 BST / 15:30 IST / 18.00 SGT for essential maintenance. Apologies for the inconvenience.
•Despite the importance of rhizosphere properties for water flow from soil to roots, there is limited quantitative information on the distribution of water in the rhizosphere of plants.
•Here, we used neutron tomography to quantify and visualize the water content in the rhizosphere of the plant species chickpea (Cicer arietinum), white lupin (Lupinus albus), and maize (Zea mays) 12 d after planting.
•We clearly observed increasing soil water contents (θ) towards the root surface for all three plant species, as opposed to the usual assumption of decreasing water content. This was true for tap roots and lateral roots of both upper and lower parts of the root system. Furthermore, water gradients around the lower part of the roots were smaller and extended further into bulk soil compared with the upper part, where the gradients in water content were steeper.
•Incorporating the hydraulic conductivity and water retention parameters of the rhizosphere into our model, we could simulate the gradual changes of θ towards the root surface, in agreement with the observations. The modelling result suggests that roots in their rhizosphere may modify the hydraulic properties of soil in a way that improves uptake under dry conditions.
How water moves from the soil into roots controls plant water relations and nutrient uptake. While it is well known that roots induce changes in the chemical, biological and physical properties of the soil in their immediate vicinity (the rhizosphere), these changes have received little attention in modelling of root water uptake. Common models are based on the assumption that soil hydraulic properties (soil water retention curve and hydraulic conductivity) do not change with distance to the root surface. However, there are hints in the literature that this is not true, as roots release polymeric substances such as mucilage, rearrange soil particles during root growth, induce shrinking and swelling by diurnal change in transpiration, and contribute to soil aggregation (Dexter, 1987; McCully, 1999; Read et al., 2003; Whalley et al., 2005; Darrah et al., 2006; Hinsinger et al., 2009). Investigations of hydraulic properties in the rhizosphere need to be conducted in situ as physical integrity is crucial for the parameters in question. In addition, such investigations have to take into account the dynamic nature of root growth and the interaction with soil.
The acquisition of data on water dynamics at small distances from the roots is technically challenging, because multidimensional, highly resolved data are required. Recently developed nondestructive, noninvasive methods such as neutron radiography and tomography (NT), X-ray computed tomography (CT) and magnetic resonance imaging (MRI) have great potential as they provide high spatial and temporal resolutions. Ferromagnetic and paramagnetic materials in the soil hinder MRI and result in poor signal-to-noise ratio (Hall et al., 1997; Carlson, 2006; Moradi et al., 2008, 2010). Thus, this technique is limited to carefully selected media such as pretreated sand and soil, agar and glass beads. Compared with NT, X-ray CT is insensitive to water, and poor at decoupling differences in soil bulk density from differences in water content. With NT, small changes in soil water content (θ) can clearly be quantified while soil minerals are nearly transparent. Hence NT is a highly attractive tool with which to explore water dynamics in the root zone of plants, in combination with information on soil structure derived from X-ray CT.
Previous studies have used imaging techniques to investigate the distribution of water around the roots of transpiring plants. In soil with uniform properties, depletion of water around the roots is expected to occur as a result of the gradients in soil water potential (h) needed to move water to the root. Because of the nonlinearity of the soil hydraulic conductivity and the radial geometry of flow to a root, local water depletion around the roots should extend outwards as the overall water content in the bulk soil decreases (Gardner, 1960). Experimental evidence for water depletion around roots was first provided by Hainsworth & Aylmore (1989). Macfall et al. (1990) observed that water depletion occurred first around the tap root and then extended to the laterals. Similarly, Segal et al. (2008) reported that the soil in the immediate vicinity of the roots of 2-wk-old barley (Hordeum vulgare) plants had a lower water content than the bulk soil. In contrast to the above findings and to the water profiles predicted by models of root water uptake, accumulation of water in the vicinity of roots has also been observed. High-contrast data on water gradient profiles were obtained by Nakanishi et al. (2005) using thermal NT. They reported increasing neutron signal towards roots at distances of 1–2 mm from the root surface and related that to water content, although they were not able to quantify water contents from the tomography data. Heterogeneous patterns of θ around roots were reported by Esser et al. (2010), and Tumlinson et al. (2008), who quantified θ profiles around the roots of lupin (Lupinus albus) and corn (Zea mays) seedlings. They called for more detailed investigations of the temporal dynamics of θ in the rhizosphere. Carminati et al. (2010) quantitatively demonstrated that there was more water in the immediate vicinity of the roots of a 3-wk-old lupin plant than in the bulk soil during a drying cycle. Based on the measured water content in the rhizosphere and bulk soil, they derived a water retention curve for the rhizosphere, which was different from that of the bulk soil and strongly hysteretic. However, their results were based on two-dimensional neutron radiographs and, consequently, the calculated θ profiles were smoothed because of the approximations resulting from two-dimensional projections. However, the study did not investigate differences along the root length and among the roots of different plants.
Possible reasons for the discrepancies among the above-mentioned studies include a lack of sufficient spatial resolution, approximations resulting from the use of two-dimensional projection images, differences among plant species and rooting systems, and differences among parts of the root structure, that is, the effects of age and root order. The aim of the current study was therefore to overcome the limitations of previous studies by conducting three-dimensional investigations of the distribution of water around different roots of different plants with sufficient spatial resolution. Our objectives were:
• to obtain three-dimensional and highly spatially resolved data on the water distribution around roots, reducing the partial volume effect and avoiding any smoothing effect resulting from approximations in two-dimensional projections;
• to compare the water distribution around the tap root, lateral roots, and the upper and lower parts of the root systems of various plants species;
• to monitor θ profiles around the roots over time as the bulk soil θ decreases, and
• to use a conceptual model to explain the observed θ profiles around the roots and in the bulk soil based on estimated hydraulic properties of the rhizosphere.
Materials and Methods
Experimental set-up and plant growth
We used a sandy soil that was collected from the artificial catchment Chicken Creek located near Cottbus, Germany. The soil (sieved to < 2 mm) consisted of sand : silt : clay at a ratio of 92 : 5 : 3 and had a bulk density of 1.4 g cm−3, a pH of 7.5, a total carbon content of 0.3%, and an electrical conductivity of 0.105 dS m−1. For more details on the soil characteristics, see Gerwin et al. (2009). We used graphite columns with a diameter of 27 mm and a length of 100 mm for the root growth studies. For each plant species, three columns were filled with the sandy soil at a bulk density of 1.4 (g cm−3) to a height of c. 90 mm. Seeds of chickpea (Cicer arietinum L.), white lupin (Lupinus albus L.), and maize (Zea mays L.) were soaked in a solution of CaCO4 overnight and then sown directly on the soil surface (one per column); 2–3 d were subsequently required for germination. The soil was initially supplied with 100 mg N kg−1 (NH4NO3), 80 mg P kg−1 (CaHPO4), 100 mg K kg−1 (K2SO4), 100 mg Ca kg−1 (CaSO4 × H20), 50 mg Mg kg−1 (MgCl2) and a micronutrient solution (for details, see Vetterlein et al., 2007). Each column had a porous plate at the bottom connected to a water reservoir through a tube. By adjusting the water level in the water reservoir, we were able to set the soil water potential at the bottom of the column to the desired value of −15 cm. The soil surface was covered with a 1-cm layer of coarse gravel (5 mm diameter) to minimize evaporation. The plants were grown for 12 d (starting from the date on which the seeds were sown in the columns) with a daily light cycle of 16 h light : 8 h darkness, a light intensity of 300 μmol m² s−1, and a controlled temperature of 23 : 19°C (day : night) before measuring water dynamics using NT. The samples transpired between 5 and 7 ml d−1 during the NT experiment. This was measured gravimetrically by weighing the samples just before and after the daily transpiration cycle.
The two NT stations NEUTRA and ICON at the Paul-Scherrer Institute, Switzerland, were used for the NT experiments. NEUTRA uses thermal neutrons, while ICON uses cold neutrons with higher sensitivity to water and higher spatial resolution than NEUTRA. In addition, to allow comparison of the two imaging stations, this set-up provided more efficient use of the available beam time. Imaging parameters for the tomography reconstruction are listed in Table 1. For more details of the two imaging set-ups, see Moradi et al. (2009a,b), and Supporting Information Methods S1. During the tomographic measurements, light inside the imaging station was provided by a lamp identical to that used in the growth chambers.
Table 1. Neutron tomography parameters at the NEUTRA and ICON stations (at the Paul-Scherrer Institute, Villigen, Switzerland)
Camera array (pixels)
Pixel size (μm)
Scintillator thickness (μm)
L : D ratio*
Effective resolution (μm)
Rotation degree (°)
Exposure time (s)
*For details see Supporting Information Methods S1 (Neutron imaging setup).
1024 × 1024
2048 × 2048
Initial tomographic scans
After the 12-d growth period, the root systems of all nine columns (three replicates for each of the three plant species) were scanned, with each scan taking c. 6 h. All of these scans were analysed for volumetric water content in the soil as a function of radial distance from the root system based on the procedure explained in the next section. Whereas we scanned the root columns of the chickpea using the NEUTRA imaging station, the columns with maize and lupin were scanned using the ICON station.
Time-series tomographic scans
Comparing the root systems of all nine columns in the neutron scans obtained, we selected the column with the largest root system for each of the three plant species and conducted additional tomographic scans to determine the dynamics of water content in the soil and in the rhizosphere over a drying and rewetting period. For this purpose, we interrupted the water supply at the lower boundary, causing continuous drying of the columns as a result of root water uptake and transpiration over the next 2 d. On the third day, the columns were rewetted by reconnecting the lower boundary of the column to the water supply for 5 h, after which a final tomographic scan was obtained. For time-series tomographic scans, we were limited by the number of replicates for each plant species that could be scanned within the same photoperiod day. However, each individual scan provided a complete three-dimensional image of the variations in soil water content in each column. Therefore, the three-dimensional images provide a complete characterization of the soil heterogeneities, with corresponding volumetric water content gradients in the rhizosphere of the complete root system in each column.
Roots could be easily segmented from the soil as a consequence of their sharp contrast resulting from the high concentration of hydrogen compared with their surroundings. The edges were detected in places where the second derivative of the intensity crossed the zero axes that highlighted areas with rapid change in pixel intensity values (Matlab, 2007). To avoid subjective root segmentation, we used an algorithm developed by Schluter et al. (2010). In this method, a combination of edge detection and bi-level segmentation is used to preserve the boundaries of large objects while excluding fuzzy objects. Therefore, gradient masks ensure an objective and fully automatic determination of the required thresholds. The tap root and lateral roots were distinguished based on their diameter, which was determined by morphological opening operations using Matlab (Matlab, 2007).
As soil heterogeneities might affect the calculated soil water content profiles calculated from the two-dimensional cross-sections, we took a different approach for calculating water content profiles that includes all soil heterogeneities in the soil columns. With an effective resolution of 200 μm (NEUTRA) and 56 μm (ICON), each three-dimensional neutron image includes c. 10 125 000 (150 × 150 × 450) spatially resolved voxels with an equal number of corresponding volumetric water content values. From this data set, thousands of volumetric water content profiles, representing volumetric water content as a function of radial distance from the root surface for a specific rooting system, were obtained. This type of spatially distributed measurement of root architecture, soil heterogeneity and rhizosphere water content distribution can only be obtained using three-dimensional noninvasive measurements as presented for this study.
The water content in the images was accurately calculated from the measured neutron attenuation coefficients in the neutron images (see Supporting Information Methods S1 (Reconstruction of tomography data) and Fig. S1), whereas the volumetric water content distance map calculations were performed using QuantIm (Vogel, 2010).
Modelling soil water potential and soil water content
To interpret the observed θ profiles around the roots, we used a simplified analytical model describing the radial flow of water to a cylindrical root. Our hypothesis is that soil hydraulic conductivity and water retention parameters vary as a function of distance to the roots. We modelled the rhizosphere as a series of concentric cylinders around the roots (see Supporting Information Methods S1 (Model rhizosphere) and Fig. S2). The cylinders had equal widths but the widths varied depending on the total width of the rhizosphere. We based our parameterization on the results of Carminati et al. (2010). They estimated a water retention curve for the rhizosphere of lupin plants grown in the same soil and under the same conditions as here. They compared the θ values of the rhizosphere and of the bulk soil during drying. Assuming that differences in soil water pressure potential between the rhizosphere and bulk soil were small overnight, when transpiration ceases, and under wet soil conditions, they derived the water retention curve of the rhizosphere (see Carminati et al., 2010 for details).
For this purpose, we fitted the Brooks and Corey equation (Brooks & Corey, 1964) to their water retention data and derived a set of soil water retention parameters for the rhizosphere and bulk soil separately. Assuming an effective average soil water retention curve for the rhizosphere soil, we generated soil water retention parameters for each of the individual soil cylinders in the rhizosphere. Among the Brooks and Corey parameters, the residual water content, θr, and the saturated water content, θs, were kept the same for all rhizosphere cylinders, while the parameters ksat, the hydraulic conductivity at saturation, and the fitting parameters λ and τ varied among the individual cylinders (Table 2). These parameters were varied in such a fashion that the soil water retention parameters of the rhizosphere domains differed from those of the bulk soil as a function of the inverse of their radial distance from the root surface, while the average of the five domains yielded the measured effective average rhizosphere water retention parameters. Hence, differences between the hydraulic properties of the rhizosphere and the bulk soil were largest for the rhizosphere domain closest to the root surface. The intention of our modelling was not to fit observations, but to explore whether the observed water profiles towards roots can be explained by including the hydraulic properties of the rhizosphere in the model. As θ gradually increased towards the root surface, we hypothesized that this gradual change in θ might be a result of the gradual change in the water retention parameters of the soil around the roots. The water retention curve of the bulk soil was obtained using the hanging-column method (Klute, 1986). The saturated conductivity was calculated with the falling head method (Klute, 1986).
Table 2. Brooks and Corey water retention and hydraulic parameters for bulk soil and rhizosphere layers
ksat (cm s−1)
θr, residual water content; θs, water content at saturation; h0, the air entry value; ksat, hydraulic conductivity at saturation; λ and τ, fitting parameters.
The water content profiles towards roots were calculated as a succession of steady-state profiles (Cowan, 1965). The steady-state approach assumes that the flux (cm3 d−1) across a cylindrical surface around the root is constant for all radii of the cylinder. Under this assumption, h and θ profiles can be calculated by solving Richards’ equation in terms of the matric flow potential (de Willigen & van Noordwijk, 1987). We imposed a flux of 0.375 cm d−1 at the root surface, which was equal to the measured water uptake of 6 ml in 16 h of photoperiod divided by the estimated root surface of 16 cm2 obtained from the measured root geometry. Assuming the continuity of water potential and flow at the interfaces between the concentric domains, the matric flux potential can be calculated for each soil domain. Then θ and h profiles towards the roots are obtained from the hydraulic conductivity and the water retention curve of each domain. Thus, the impact of the rhizosphere domain and its extent on profiles of θ and h could be evaluated through numerical simulations. Based on the tomography data, we assumed that the rhizosphere extends to 2 and 4 mm from the root surface for upper roots and lower roots, respectively, and calculated the θ profiles at various soil water contents during water uptake by plants.
As a consequence of the marked difference in the water contents of roots and soil, there was a clear contrast between roots and soil (Fig. 1). The contrast appeared much stronger than that in two-dimensional radiographs of roots of the same plants grown under the same conditions (Moradi et al., 2009a,b; Carminati et al., 2010; Esser et al., 2010), where the inherent averaging over the sample thickness made the roots slightly less visible. Fig. 2 shows a three-dimensional visualization of the root system of chickpea after segmentation from the soil. The main root is shown in dark brown and the laterals in purple. For a video file of the root structure, see Video S1.
Water content maps and profiles in initial tomographic scans
Fig. 3 shows the soil water content in one of the chickpea replicates. On the left side of Fig. 3 we show the water content map in the horizontal cross-sections at depths of 4, 6 and 8 cm below the soil surface (Fig. 3a,d,g). The two-dimensional θ maps were produced by contour plotting the cross-sections through the three-dimensional data. The white circular patches represent the roots. The tap root is distinguishable by its considerably larger cross-section compared with the other roots. Compared with the bulk soil, the higher water content around both the main root and the lateral roots is clearly visible at all three depths. This was quantified by evaluating the local water content as a function of the distance to the tap root (Fig. 3b,e,h) and to the laterals (Fig. 3c,f). In these figures, we plotted the θ of each pixel (yellow dots) and the mean value (lines). In the lowest cross-section no laterals were developed. While the bulk-soil θ averaged c. 0.2 far away from the roots, it increased to > 0.3 near the root surface.
Comparable results were found for lupin and maize plants. Fig. 4 shows the corresponding two-dimensional contour plots of θ at three different soil depths (4, 6 and 8 cm). Similar to chickpea, there was an increase of θ close to the roots compared with the bulk soil. This increase was visible around the roots of all the plant species and all the replicates at all depths. Similar to chickpea, no noticeable difference was observed in this respect between the tap root and lateral roots of lupin (data not shown). As maize had a fibrous root system without a tap root, all roots were treated in the same way. The lupin and maize samples were scanned at the ICON station, which had a higher spatial resolution (56 μm) than the NEUTRA station (200 μm). This resolution down to the grain size is the reason why the small-scale heterogeneity of the bulk soil, due to soil particle arrangement and pore space, is more pronounced compared with the samples scanned at NEUTRA.
The increase in soil water content with increasing proximity to the root surface seemed to be a general trend in all three replicates of all three plant species. As an example, we show the average volume-based soil water content profiles as a function of radial distance from the root system of lupin plants in Fig. 5. The soil water content increased with proximity to the root surface for tap roots and lateral roots of all three lupin replicates and no noticeable difference was observed between them.
Time-series of water content profile
We present the water content profiles around the roots of the chickpea sample here, as this column was more frequently scanned (at least once a day) than those of the other plants. Fig. 6(a) shows the θ profiles around the main root and lateral roots of chickpea over 3 consecutive days. The θ profiles were calculated according to the volume-based distance maps and were averaged over the entire soil volume in the upper part of the root system (soil depths of 2–6 cm). On day 1, the θ profiles showed an increase towards the roots at a distance of c. 2 mm from the root surface. On day 2, there was a decrease in θ (water depletion) at a distance of 4–5 mm and the θ profiles became steeper close to the roots. After rewetting of the sample (day 3), the θ profiles showed a similar shape to those of day 1, except that they reached higher water contents. Comparison of the θ profiles of the two-dimensional cross-sections (Fig. 3) with the three-dimensional time-series data averaged over the whole volume (Figs 5, 6) shows that the θ profiles in the cross-sections are very rough and bumpy, while the volume-based θ profiles are very smooth without much variation. This confirms that the soil heterogeneities in the soil volume can be accounted for by volume-based calculation of the θ profiles.
Fig. 6(b) shows the same information as Fig. 6(a) but for the bottom part of the root system (soil depth of 6–10 cm). The results were equivalent to those for the upper part except that no depletion zone at a distance of 4–5 mm was observed. While in the bottom part the water content increased monotonically towards the roots, it showed a decrease followed by an increase towards the root surface of the upper part of the root system. Similar results were obtained for the other two plant species.
Fig. 7 compares the zone of water content increase around the upper part and the lower part of the tap root of chickpea on day 3 when the soil was wet. While the zone of water content increase was extended to c. 4 mm away from the root surface in the lower part of the main root, it was limited to a distance of < 2 mm in the upper part of the main root. We hypothesize that this difference was caused by a larger rhizosphere extent in the lower part of the sample – that is, where roots were younger. According to this observation, we have chosen rhizosphere extents of 2 and 4 mm for the modelling approach.
Fig. 8 shows θ and h profiles around the roots simulated by the model assuming a rhizosphere thickness of 2 mm (a, b) and 4 mm (c, d). The water content profiles were calculated for different bulk soil water contents (0.2, 0.15, 0.1, 0.07 and 0.05). The gradual increase of θ towards the roots in the rhizosphere could be reproduced by assuming gradual change of the Brooks and Corey water retention parameters with decreasing distance from the root. Over a wide range of bulk soil water contents, the θ profiles showed an increase towards the root surface. Similar to the experimental data, the θ gradient became very steep when the bulk soil θ dropped below 0.1. Water potential profiles showed no visible gradients as long as the bulk soil θ was above 0.1. Below this point, for a rhizosphere extent of 2 mm, the water potential dropped from −90 hPa in the bulk soil to a minimum of −179 hPa in the vicinity of the roots. This was because the hydraulic conductivity of the soil became limiting and could not meet the transpirational demand. For a rhizosphere extent of 4 mm, the θ profiles were much gentler, even for bulk soil θ < 0.1. The water potential decreased to a minimum of only −114 hPa adjacent to the root surface compared with a water potential of −179 hPa in the case of a 2-mm extent of the rhizosphere.
Taking advantage of recent advances in NT, we successfully visualized the three-dimensional root structure and quantified the distribution of water around the roots with high spatial resolution. To the best of our knowledge, this study represents the highest spatial resolution and the most detailed quantification of the water content distribution in the rhizosphere of plants. During plant transpiration, we consistently observed an increase in the water content towards the root surface for all investigated plant species and for both tap roots and lateral roots. As roots act as a sink for water and water flows from bulk soil towards the root surface, a gradient of water potential towards the root is needed to drive the water flow. Therefore, one would expect that θ should be decreasing, and not increasing, towards the root surface. The current observations, showing a different behaviour, can be explained in two ways: what we observe as an increase in water content is an artefact from NT and image analysis; or the water retention curve of the rhizosphere is different from that of the bulk soil, as hypothesized by Carminati et al. (2010).
First, we would like to discuss the possibility of artefacts during image analysis.
We obtained a neutron attenuation coefficient for each pixel in the image after image corrections. Then we related the neutron attenuation coefficients to water content based on our calibration curve (see Supporting Information Fig. S1). One possibility is that larger neutron attenuation coefficients in the rhizosphere were caused by soil compaction as a result of root growth and misinterpreted as larger water contents. However, the increase in soil bulk density alone cannot explain the observed attenuation coefficient near roots. This is a result of the large difference in neutron attenuation coefficient between dry soil (0.2 cm−1) and water (3.5 cm−1). Even an increase in soil bulk density from 1.4 to 2.65 g cm−3, which is very unlikely, could explain only 30% of the observed increase in water content in the rhizosphere. A similar argument can be made for the elevated contents of carbohydrates, for example, acids or sugars, which may appear as an increased amount of water. Furthermore, partial volume effects cannot explain the high water contents in proximity of roots, because of the high effective spatial resolution in the images of the rhizosphere, which has been confirmed using test objects of known geometry (Supporting Information Methods S1 (neutron imaging setup)). The partial volume effect is usually significant in the first 2–3 pixels (in two-dimensional data) or voxels (in three-dimensional data) on the two sides of the segmentation line. In our study, the zone of water content increase extended to at least 2 mm (and often > 3 mm) from the root surface, which corresponds to at least 25 pixels wide, thus 10 times larger than the zone affected by partial volume effects. Finally, knowing that any kind of segmentation based on different thresholding could be subjective, we used a very robust algorithm for root segmentation that combines an edge-detection algorithm and a bi-level segmentation to preserve the boundaries of the objects while excluding the fuzzy areas around them. Root hairs, extending from the root surface into the soil, might to some extent contribute by their hydrogen content to the calculated θ in the root-hair region of the roots, but their total volume is too small to account for the observed differences (Carminati et al., 2010).
As artefacts have been rationally ruled out, we will discuss the ways in which the soil water retention curve may be altered by the presence of roots. Soil compaction around roots, caused by roots pushing through the soil, could result in a change in pore size distribution and therefore a shift in the water retention curve. This is expected to occur especially around thick roots (Dexter, 1987, 2004; Stange & Horn, 2005; Assouline, 2006). However, we observed a stronger rhizosphere effect around the younger, thinner roots than around the older, thicker upper part of the roots (Fig. 7). We could not detect any change in the bulk density of the columns used in the current experiment using X-ray tomography with a spatial resolution of 160 μm (data not shown). A higher spatial resolution such as used by Aravena et al. (2011) is needed in order to confirm this possibility.
Another possibility is the excretion of mucilage into the soil by roots, which might explain the increased neutron attenuation coefficient of the rhizosphere, as mucilage in the soil may retain substantial water at moderate suctions and increase the local water-holding capacity around the roots (Dexter, 1987; Watt et al., 1994; Young, 1995; McCully & Boyer, 1997). This is consistent with the θ profiles around the roots in the current study, which clearly show an increasing trend towards the root surface even in relatively dry soil.
Our numerical simulations demonstrated that explicit representation of a rhizosphere domain with gradually changing hydraulic properties (see Supporting Information Methods S1 (Model Rhizosphere) and also Fig. S2) closely reproduced the observed θ profiles around the roots (Fig. 8a). Hence, the observed θ profiles around the roots do not violate the physical principles of root water uptake, but can be explained solely by assuming hydraulic properties in the rhizosphere differing from those of the bulk soil.
The area in which θ increased compared with the bulk soil was larger in the lower part of the root system than in the upper part. In the upper part, the θ profiles were steeper and there was a depletion of water at a distance of 3–5 mm from the root surface at low soil water contents. By contrast, the lower roots did not show such steep θ gradients and the θ smoothly increased towards the roots even when the soil dried out. The difference in the spatial extent to which roots influence the surrounding soil might explain these observations. Our numerical simulations corroborate this hypothesis, as roots with a more extended rhizosphere experience much milder negative water potentials (and therefore less depletion of water) at their interface as the soil dries. The difference in the extent of the rhizosphere might be related to the degree of soil compaction around the roots, the rate of mucilage release and its location along the root (root tip), its diffusion into the soil, microbial mucilage degradation and the rate of root growth. Watt et al. (2006) have outlined the basic method for modelling the extent of the rhizosphere; however, there is a lack of experimental data for roots growing in soil and the actual rate of mucilage production and degradation. Similarly, true data on root growth rates in situ, that is, not from rhizoboxes, gel chambers or hydroponics, with high temporal resolution are lacking and could only be obtained with noninvasive methods, as is the case in our study.
However, differences in the extent of the rhizosphere might not be the only reason for the difference in the water profiles around different parts of the roots. Heterogeneous water uptake along the root length (Hinsinger et al., 2009) could also explain the difference in θ profiles. Differences in xylem maturation and in the number and diameter of xylem vessels as well as differences in the formation of endodermis and exodermis and aquaporin formation/activity with root development have been frequently demonstrated to influence water uptake by roots in pot and field conditions (Steudle & Frensch, 1996; Barrowclough et al., 2000; Watt et al., 2008). All these parameters affect root hydraulic conductivity and thus root water uptake in a given environment (Draye et al., 2010).
The time-series measurements of soil water content profiles were based on measurements of a single root system from the three species and there were no replications at the species level. Therefore, there was no possibility of assessing whether differences among the species and individual plants exist. An NT study of water content profiles with emphasis on the differences between individual plants of the same species is needed to answer this question. However, there are practical difficulties at the present time as a consequence of the long scan time for each single tomography (5–6 h) and the associated measurement costs.
The present experimental results and our modelling approach highlight the importance of the rhizosphere in water flow from soil to roots. The rhizosphere acts as an interface between roots and soil which improves the hydraulic conductivity of the soil and therefore helps to maintain the transpirational demand as the soil water content decreases. Whether altered rhizosphere hydraulic properties are a result of compaction, the presence of mucilage or both is still an open question and requires further investigation. As rhizosphere hydraulic properties affect not only water uptake and water distribution around the roots but also the uptake of nutrients and their diffusion and transport in the root zone, such investigations are also highly relevant for understanding solute movement.
NT of the root–soil interface showed an increase in soil water content close to the roots, that is, in the rhizosphere. Although counterintuitive, this observation was consistent for all three plant species investigated: chickpea, white lupin and maize. Both main roots and lateral roots showed higher water contents in their rhizosphere compared with bulk soil. Moreover, the spatial extent of the rhizosphere effect was even larger in the lower part of the root system than in the upper part.
The rhizosphere held more water than the bulk soil over a range of soil water potentials, even when the soil dried up. However, a small depletion of water was observed around the older roots at low water contents. Our modelling results showed that the spatial extent of the rhizosphere shapes the water profiles around the roots, with potential consequences for water uptake efficiency.
Our work highlights the importance of soil hydraulic properties in close proximity to roots for water uptake by plants. A rhizosphere formed by the roots can improve the hydraulic properties of the soil around the roots and therefore facilitate the root uptake of water and nutrients, especially at low soil water contents.
This study was funded by the EU Marie Curie Project ‘Water Watch’ via contract MTKD-CT-2006-042724. We thank the Transregional Collaborative Research Centre 38 ‘Structures and processes of the initial ecosystem development phase in an artificial water catchment’ (SFB/TRR 38) for providing soil and soil data from the Chicken Creek catchment.