A root is a root is a root? Water uptake rates of Citrus root orders


B. Rewald. Fax: +972 8 659 6757; e-mail: brewald@rootecology.de


Knowledge about the physiological function of root orders is scant. In this study, a system to monitor the water flux among root orders was developed using miniaturized chambers. Different root orders of 4-year-old Citrus volkameriana trees were analysed with respect to root morphology and water flux. The eight root orders showed a broad overlap in diameter, but differences in tissue densities and specific root area (SRA) were clearly distinguishable. Thirty per cent of the root branch biomass but 50% of the surface area (SA) was possessed by the first root order, while the fifth accounted for 5% of the SA (20% biomass). The root order was identified as a determinant of water flux. First-order roots showed a significantly higher rate of water uptake than the second and third root orders, whereas the fourth and fifth root orders showed water excess. The water excess suggested the occurrence of hydraulic redistribution (HR) as a result of differences in osmotic potentials. We suggest that plants may utilize hydraulic redistribution to prevent coarse root desiccation and/or to increase nutrient acquisition. Our study showed that the novel ‘miniature depletion chamber’ method enabled direct measurement of water fluxes per root order and can be a major tool for future studies on root order traits.


Water uptake is a critical mechanism to replace water lost by transpiration. Knowledge of water uptake is key to understanding plant function under current and future climates (Feddes et al. 2001) and is required for developing sound models for ecological and agricultural purposes (Schröder et al. 2009).

While plant function aboveground is rather clearly distinguishable (i.e. leaves absorb carbon, and stems and branches provide transport, storage and structural functions), our understanding of functional differentiation belowground is hampered by root systems' inaccessibility and complexity. Development and ageing are well known to affect the physiological characteristics of roots (see Wells & Eissenstat 2003 and references therein), including water uptake and conductance of certain root sections. For example, small, ephemeral fine roots are thought to be of utmost importance for water uptake, while coarse, woody roots mainly provide transport and structural functions (Steudle 2000). A practical assumption has been that all roots of a given size class (e.g. non-woody, ephemeral ‘fine roots’ defined as roots with diameters ≤2 mm) can be treated as a coherent mass; that is, they have the same construction and maintenance respiration costs and the same rates of water and nutrient uptake (Persson 1980; Jackson, Mooney & Schulze 1997; Rewald & Leuschner 2009). In contrast to the current paradigm, an increasing number of studies suggest that root functions vary according to the position of an individual root on the branching root system (‘root order’; Pregitzer et al. 2002; Hishi 2007; Guo et al. 2008; Pregitzer 2008; Valenzuela-Estrada et al. 2008). Thus far, mostly indirect, specifically morphological and anatomical analyses have been used to provide evidence of differences in root order functionalities. Such analyses were considered valid because root anatomy and physiology are often found to be tightly linked (Eissenstat & Achor 1999; Hishi 2007) and direct measurements are difficult to achieve (for nutrient uptake, see Lucash et al. 2007, but see Waisel, Zilberstaine & Eshel 1990). To date, hydraulic measurements of certain root orders are limited to lateral root ends by the classical use of ‘potometers’ (Zwieniecki & Boersma 1997 and references therein) or to root junctions (Schulte 2006). Thus, the development of a method to measure the uptake rates of different root orders would be of great help to explore the physiological traits of root systems in general.

To assess the influence of root order-specific functional traits on the whole root system, the root branching structure has to be known. The morphological/anatomical properties and frequencies of the first few root orders have been determined on three dozen woody species (e.g. Pregitzer et al. 2002; Guo, Mitchell & Hendricks 2004; Wang et al. 2006, 2007; Guo et al. 2008). Determinations of the total number of root orders have been made in even fewer species (e.g. Valenzuela-Estrada et al. 2008). This dearth of studies is surprising because root architecture, classically described by the branching pattern only (e.g. ‘herringbone’ or ‘dichotomous’; see Dunbabin, Rengel & Diggle 2004 and references therein), is a fundamental aspect of plant productivity, determining the resource-foraging strategy of a plant (Lynch 1995). A broader knowledge about the frequency and morphology of root orders will deepen our understanding of plant foraging strategies.

In this study, we assessed the importance of different root orders for water uptake in Citrus volkameriana rootstocks. To overcome technological limitations, we constructed ‘miniature depletion chambers’, allowing water flux measurements on the level of root orders. More specifically, the aims were (1) to quantify the abundance and morphology of specific root orders and (2) to provide evidence that water flux rates of Citrus plants growing under homogeneous environments are primarily related to the place of a root segment within the root branching system (i.e. the root order).


One-year-old Citrus sinensis Osbeck cultivar Newhall (Navel orange) shoots were grafted on adequately sized C. volkameriana Ten. & Pasq. (Volkamer lemon) rootstocks in the year 2006 (Gilat Experimental Station, Israel). C. volkameriana rootstocks are of economical importance and wide use in citriculture, to enhance either fruit yield or plant resistance against the Citrus tristeza virus (Georgiou & Georgiou 1999; Ramin & Alirezanezhad 2005).

The drip-fertigated plants were grown for three more years in a greenhouse in soil-filled 15 L pots. In June 2009, four equal-sized plants were selected; pots were removed, and roots were rinsed carefully using a soft water jet. Plants were moved into opaque 20 L pots filled with ∼17 L of 1.0 strength Long Ashton solution (LAS; Ottow 2005). The osmolality of the solution was 24 ± 1 mmol kg−1 (mean ± SE; Vapro 5520; Wescor, Logan, UT, USA). The LAS was aerated constantly and transpired water was refilled every other day with demineralized water (DW); the entire nutrient solution was exchanged every other week. The pots, tightly covered to reduce evaporation and to prevent light penetration, were placed in a temperature- and light-controlled growth room. A 12.5 h day was established (5:30 a.m.–6:00 p.m.), with a photosynthetic photon flux density of 300–400 µmol m−2 s−1. Air temperature and relative humidity (RH) were logged close to the plant canopies; maxima/minima were 28 °C, 31% RH during the day and 21 °C, 73% RH at night (HOBO U10; Onset Computer Corporation, Pocasset, MA, USA). Leaf transpiration at noon (12:00 a.m.–1:00 p.m.) was 81.6 ± 7.4 mmol m−2 s−1 (mean ± SE, n = 24; SC-1 Porometer; Decagon, Pullmann, WA, USA). Temperature and RH during water uptake measurements (11:00 a.m.–2:00 p.m.) were 26.60 ± 0.04 °C and 35.66 ± 0.08% RH, respectively (mean ± SE). After 2 weeks of acclimation, we conducted one uptake measurement per day.

Analysis of root morphology and biomass

Our root order designation follows the ‘classical’ stream classification approach (see Pregitzer et al. 2002 for details). In brief, most distal roots (root tips) were named first-order roots; roots that possess only first-order side roots were named second-order roots; root segments bearing first- and second-order side roots were named third-order roots, and so on.

The rootstocks of two similar-sized Citrus plants (biomass ratios: leaf : stem 0.53 ± 0.01, root : shoot 0.51 ± 0.03; mean ± SE) were dissected from the stem right above the highest side root. The stems and leaves were oven-dried (70 °C, 48 h) and dry biomass was recorded. Because of the time-consuming dissection process, an up-scaling approach was used to receive the proportion of root orders within the root systems (∼70% of the root systems were analysed in detail). In doing so, 12 ‘large root branches’ (i.e. branches including fifth- or fifth and sixth root orders) attached to the tap root were randomly chosen and detached. Subsequently, six side-root branches (≤5 root orders present) of each large root were randomly selected for dissection. Additionally, 4 ‘small root branches’ (≤4 root orders present) per plant, growing directly at the tap root, were harvested. The ratio between analysed large and small branches was set according to visual observations of the root system structure. In total, 152 root branches (12 large branches × 6 smaller branches × 2 plants plus 4 small branches from the tap root × 2 plants) were analysed. Any particles adhering to the roots were carefully removed with forceps and the root branches were rinsed clean with water. The 152 root branches were dissected, starting with the first root order. The remaining roots on the 12 large branches and roots left at the tap root (i.e. root order 8) were sorted up to the seventh root order; the first through third root order were combined to reduce processing time. Live roots were distinguished from dead roots using methods as described in Rewald & Leuschner (2009); only living roots were measured. Roots were kept moist during the whole dissection process.

Root segments were stored, separated by order and branch, in tap water (4 °C) until digital images of root orders 1–6 were made on a flat-bed scanner to determine root surface area (SA), tissue density (TD) and diameter (WinRhizo 2005c; Régent Instruments Inc., Québec, QC, Canada). Analyses of the morphology of the seventh and eighth root order were not feasible due to limited samples. Root samples were oven-dried (70 °C, 48 h) to allow for biomass determination.

Chamber construction and positioning

We constructed miniaturized chambers to measure water fluxes among the different root orders (Fig. 1 & Supporting Information Fig. S1). The chamber body was manufactured from plastic storage tubes [polypropylene (PP), 15 mm inner diameter], which were cut after 15 mm in length (both sides open). The side of the tubes were cut open lengthwise to allow for root installation; in doing so, approximately 3 mm of the plastic was removed with a circular saw to reduce the tube diameter. On both front ends, septa (diameter d = 12.5 mm, IceBlue; Restek, Bellefonte, PA, USA) were glued in place (ethyl-2-cyanoacrylate, LocTite Super Glue-3; Henkel, Boulogne, France). Both septa and plastic tube were cut open on one side, allowing for root insertion by spreading the chamber open along the section. The septa were pre-drilled (0.3–3.0 mm diameter; Fig. 1b) to enable sealing while also preventing over-excessive squeezing of inserted roots (0.55–3.53 mm diameter; see Fig. 3a).

Figure 1.

Schematic side (a) and front view (b) of a ‘miniature depletion chamber’ attached to a second-order root. The uptake chambers (15 mm in length) were crafted out of a plastic tube, pre-cut at the longitudinal side and with glued septa on either front side. Super glue was used for sealing the cuts in the septa and hot glue for sealing the cut in the plastic tube after root embedding. Oxygen-enriched solutions, either demineralized water (DW) or 0.5 strength Long Ashton solution (LAS), were delivered using a syringe through one of the septa (drawing not at scale; see Materials and Methods and Supporting Information Fig. S1 for details).

Figure 3.

Properties of C. volkameriana root orders. (a) Diameter distribution; diameters of 1 and 2 mm are marked by lines to indicate commonly used size classes. The y-axis is log transformed due to presentability (box plot; Mann–Whitney U-test; P < 0.01, n = 41–122). The diameter distributions of root segments used for water flux measurements are indicated by arrows (root orders 1–4); the diameter of the single fifth-order root segment is indicated by a plus symbol. (b,c) Root tissue density (TD) and specific root area (SRA; mean + SE; Mann–Whitney U-test; P < 0.01, n = 41–122). Distribution of root surface area (SA) (d) and root biomass (e) within root branches with five root orders (mean + SE; Mann–Whitney U-test; P < 0.05, n = 30). Different letters denote significant differences between root orders.

For chamber placement, we lifted the root system of one of four trees out of the hydroponics and fixed it in mid-air (<10 min). A root segment of a specific order was chosen out of preselected segments. Selection criteria were the lack of side branches on a length of >17 mm, no signs of decay (i.e. dark-coloured, shrivelled) and an undamaged epidermis/rhizodermis. The chosen segment was blotted dry using a paper towel before being placed between septa. For measuring the water uptake of the first root order, chambers with only one pre-drilled septum were used; the root tip ended within the chamber. After root insertion, a small clamp was used to close the chamber and cuts were sealed with either hot glue (plastic tube) or super glue (septa); the root–septa interfaces were sealed by the pressure of the septa and a small amount of super glue (Fig. 1 & Supporting Information Fig. S1).

The glue was allowed to dry for 4–5 min before ∼1.3 mL of either DW (osmolality: below measurement threshold) or 0.5 strength LAS (osmolality: 15 ± 1 mm kg−1) was injected into the chamber with a syringe; a second needle was used in parallel to release air from the sealed chamber. Solutions were aerated for 18–24 h before use to prevent oxygen deficiency. DW or 0.5 strength LAS was used to induce higher than normal, and thus measurable, water uptake rates due to the lower osmotic potentials of these solutions in comparison with the 1.0 strength LAS surrounding the rest of the root system. At the time of solution injection, improperly sealed (i.e. leaking) chambers were easily detected by solution outflow at other places than the second needle; leaking chambers were removed. In tests during method development, using stain-filled (fast green, 2%) and air-filled chambers, it was shown that properly sealed chambers stayed tight for several days (>7 d) under experimental conditions (i.e. submerged in 1.0 strength LAS at 20 °C). Furthermore, a piece of electrical wire was installed in the chamber instead of a root to test for artificial fluxes caused even by inanimated objects; neither water uptake nor excess was measured (Supporting Information Fig. S2). Examples of raw water fluxes (i.e. not related to the inserted root SA and biomass) for root orders 1–4 and a control (i.e. chamber with wire) are displayed in Supporting Information Fig. S2. Chamber installation took place at 9:00 a.m. to allow for a 2 h equilibration period before measurements.

Water flux measurement

After chamber installation, a thin, transparent plastic tube (inner diameter di = 0.7 mm) was connected to the chamber using a hollow needle (length = 13 mm, di = 0.4 mm). The needle, connected to the tube via a Luer lock, was pinned through one of the septa (Supporting Information Fig. S1). The second end of the tube (approx. 60 cm in length) was placed within a storage container (35 mL volume) on an analytical scale (CP225D; Sartorius, Göttingen, Germany; Fig. 2). Both the tube and the storage container were filled with DW or 0.5 strength LAS, respectively; the tubing system was kept air bubble free. To prevent measurement biases by gravimetric forces, the solution levels in the storage container and hydroponics were brought to the same height. The temperature of the hydroponic system was kept at 20.0 ± 0.1 °C with a radiator coil connected to a cooling bath (BL-30; MRC, Holon, Israel). The weight of the storage container was recorded every minute with the PC programme Sartorius Connect v. 1.0 (Sartorius AG, Göttingen, Germany).

Figure 2.

Schematic drawing of the experimental set-up. Plants were placed in opaque 20 L pots filled with aerated, 1.0 strength Long Ashton solution (LAS) (20 °C). The pots were covered to reduce evaporation and to prevent light penetration. The miniature depletion chamber was connected to a storage container filled with either demineralized water (DW) or 0.5 strength LAS. The storage container was placed on an analytical scale; its weight was logged once a minute (see Supporting Information Fig. S2 for raw data). The solutions in the storage container and the pot were held at the same level by adjusting the height of a lift.

After taking the measurement, the position of the root segment within the root system (i.e. top, side, centre or bottom) was noted before the root branch with the chamber was cut at the basal side and removed for analysis. The root order of the segment and its distance to the lateral end of the branch was determined. The chamber was cut open on one side, without removing septa, and the root segment was cut at the inner side of the septa to determine the SA and the dry mass (see above). Five to nine root segments were measured per root order and solution type. Because of the high density of the side branches, a fifth-order root was only accessible on one plant; thus, the fourth and fifth root orders were pooled. The diameter distributions of root segments used for flux measurements are given in Fig. 3a.

For measuring the daily ‘maxima’ of fluxes (Fm, ‘mass flow’; gram per hour), the period between 11:00 a.m. to 2:00 p.m. was chosen because of high transpiration rates (see above). We performed linear regressions (R2adj. = 0.76–0.99, P < 0.001; Xact 7; SciLab, Hamburg, Germany) to determine mean hourly mass flow rates of the 3 h measurement period. Mass flux rates were either correlated with the SA of root segments (Js, ‘water flux density’; gram per square centimetre per hour) or with their dry mass (Jm, ‘water flux efficiency’; gram per gram dry weight per hour) where water flux density refers to the physiological uptake capacity under a given environment (e.g. root temperature, water potential gradient and transpiration). Data of SA of root orders per branch were used to calculate the relative water flux densities per root order and the mean water flux densities of branches. Scaling the flow by root segment biomass (Jm) is justified by considering the cost of resource allocation (although respiration per root order is unknown); consequently, Jm provides information of ecological rather than physiological importance.

Determination of chamber solution and root cortex osmolality

To assess possible root exudation, five miniature depletion chambers each were installed on root orders 1–4 as described above. Before chamber placement, the root system was washed for 1 h in DW. As a control, five chambers were sealed without inserting a root. All 25 chambers were filled with 1.1 mL 0.5 strength LAS. Subsequently, the root system was placed in a controlled growth room and in hydroponics as described before. After 5 h of incubation (9 a.m.–2 p.m.), the root system was lifted out of the 1.0 strength LAS (20 °C) and solutions were extracted with a syringe from each chamber. The osmolality (millimole per kilogram) of a 10 µL aliquot was determined immediately on a vapour pressure osmometer (Vapro 5520, Wescor) at 20 °C room temperature.

Fifteen randomly selected root branches of two plant individuals were dissected to determine the osmolality of the root cortex of different root orders. One root segment of root orders 1–4 was collected per branch and was stored (<20 min) in 1.0 strength LAS. Subsequently, equal-sized pieces (approx. 4 mm2) of the root cortex were carefully removed under a dissection microscope (40×) using a scalpel and forceps. The osmolality of cortex fragments was determined on an osmometer (as above). The dry weight (70 °C, 48 h) of each sample was used to determine the osmolality (millimole per kilogram).


Calculations were conducted with the PC programme SAS v. 9.1 (SAS Institute, Cary, NC, USA). Data sets were tested for Gaussian distribution with the Shapiro–Wilk test and for homogeneity of variances with the Levene test. The non-parametric Mann–Whitney U-test was used on the ‘morphological analysis’ data set to test for significant differences in root morphology, root cortex osmolality and water flux rates. A Scheffé test was used to test for significant differences in the osmolality of the chamber solutions.

The procedure PROC GLM, including cross-effects, was used on the ‘depletion chamber experiment’ data set to test for the influence of root order, segment distance to the end of the branch, spatial arrangement within the root system and root diameter on water flux rates. Because cross-effects were non-significant, interactions were removed from the GLM model before final calculation took place.


Morphology of root orders

The C. volkameriana rootstocks possessed eight different root orders. The eighth root order formed the tap root. While intra-order variance in diameter was high (e.g. first root order: d = 0.46–1.28 mm), mean diameter differed significantly between root orders (P < 0.01; Fig. 3a). The mean diameter of the first root order (root tips) was 0.51 ± 0.01 mm (mean ± SE) and increased slowly to 0.74 ± 0.02 mm in the third root order; a sharp increase in root diameter up to 4.50 ± 0.57 mm (sixth root order) was found in root orders ≥4.

The TD of the different root orders increased significantly up to the fifth root order (P < 0.01; Fig. 3b). The TD of first-order roots was 0.20 ± 0.01 g d.wt cm−3 and 0.39 ± 0.01 g d.wt cm−3 in fourth order roots; TD of the fifth and sixth root orders were 0.53 ± 0.03 and 0.61 ± 0.04 g d.wt cm−3, respectively.

The specific root area (SRA) decreased significantly with increasing root order (P < 0.01; Fig. 3c). While the first root order had a SRA value of 430 ± 12 cm2 g d.wt−1, the SRA value steeply decreased to 27 ± 9 cm2 g d.wt−1 in sixth order roots. The coefficient of variation (CV) of morphological parameters was higher between root branches than between tree individuals; thus, no significant differences in root morphology were found between tree individuals (data not shown).

Proportions of root orders on SA and biomass

Analysed on the level of branches, approximately 50% of the root SA (48.4 ± 1.6%; Fig. 3d) was provided by first-order roots. Root orders 2 and 3 accounted for 27.1 ± 0.9% and 13.9 ± 0.6%, the fourth and fifth root orders for 5.8 ± 0.7% and 4.8 ± 0.9%, respectively (P < 0.01).

Root orders were found to possess significantly different fractions (P < 0.01) of the root biomass, analysed either on the levels of root branches (Fig. 3e) or on the level of whole root systems (Fig. 4). The analysis of root branches revealed that approximately 70% of the root branch biomass was used to build root orders 1–3 (31.4 ± 1.6%, 21.6 ± 0.9% and 16.5 ± 0.9% for root orders 1, 2, and 3, respectively). For the fourth root order, only 11.5 ± 1.2% of the biomass was allocated and 19.0 ± 2.2% was allocated to the fifth root order. Considering the whole root system, the tap root (eighth root order) holds 32.7 ± 2.2% of the root biomass (Fig. 4). Biomass distributions to the sixth and seventh root orders were 7.3 ± 3.1% and 6.0 ± 2.6%, respectively.

Figure 4.

Relative root biomass of root orders within the whole, eight-order C. volkameriana root system (mean + SE). Root order 8 is the tap root.

Osmolality measurements

After 5 h, the osmolality of the chamber solution incubated with fourth-order roots was higher (28 ± 2 mmol kg−1) than the osmolalities of both control measurements (0.5 strength LAS in empty chambers) and in hydroponic pots (1.0 strength LAS); no significant differences were found between the osmolality of the control and of root orders 1–3 (Fig. 5a).

Figure 5.

(a) Osmolality of 0.5 strength Long Ashton solution (LAS) in ‘miniature depletion chambers’ after being installed on different root orders for 5 h (mean + SE; Scheffé test; P < 0.05, n = 5). The osmolality of 1.0 strength LA in pots is indicated by a dotted line (24 mmol kg−1); Control (Ctrl) measurements were obtained by incubating 0.5 strength LAS in empty chambers under the same conditions (n = 5). (b) Osmolality of tissue outside the stele in different root orders (mean + SE; Mann–Whitney U-test; P < 0.05, n = 11). Different letters denote significant differences within root orders and between root orders and Ctrl.

The osmolality of the root tissue outside of the stele was found to be very heterogeneous within root orders (Fig. 5b). Although there was a general tendency (P < 0.1, data not shown) to a higher osmolality in the fourth root order compared to root orders ≤3, a significant difference (P < 0.05) was only found between the second root order (297 ± 26 mmol kg−1) and the fourth root order (671 ± 143 mmol kg−1).

Influence of root order on water flux

Mass flux rates between 11:00 a.m. and 2:00 p.m. were found to differ significantly between C. volkameriana root orders (P < 0.05; Fig. 6). Water uptake out of 0.5 strength LAS declined significantly from root order 1 to 3 (water flux efficiencies: 277 ± 61, 41 ± 10 and 7 ± 1 g g d.wt−1 h−1, respectively), while the fourth/fifth root order class possessed water excess (−202 ± 61 g g d.wt−1 h−1). The water flux density was five times higher (Fig. 6a), and the water flux efficiency seven times higher (Fig. 6b) in the first root order than in the second root order. While the water flux efficiency of the second and third root orders differed significantly, no differences were found between water flux densities (0.155 ± 0.053 and 0.112 ± 0.016 g cm−2 h−1, respectively).

Figure 6.

Water flux density (a) and water flux efficiency (b) of C. volkameriana root orders. fourth/fifth root orders were pooled due to a limited sample size. Either demineralized water (DW) or 0.5 strength Long Ashton solution (LAS) was used for flux measurements. Different letters denote significant differences between root order-specific flux rates (mean ± SE; Mann–Whitney U-test; P < 0.05, n = 5–9).

Additionally, a significant difference between the water uptake of the first root order out of either demineralized water (DW) and 0.5 strength LAS was found (Fig. 6). Water flux density out of DW was ∼30% higher than out of 0.5 strength LAS (1.085 ± 0.139 and 0.707 ± 0.148 g cm−2 h−1, respectively).

Both the water flux density and the dry mass-related water flux were significantly related to the root order (P < 0.01; Table 1), but neither other parameters (distance of the root segment to the distal end of the root branch, the position of the measured segment within the whole root system and the root diameter) nor cross-effects had a significant influence on the water fluxes in the homogeneous hydroponic environment.

Table 1.  GLM analysis for the effects of root order, root segment distance to the end of the root branch, segment position within the root system and root segment diameter on the water flux density and the water flux efficiency of C. volkameriana roots
Sourced.f.Water flux densityWater flux efficiency
  • a

    Distance of basal end of the measured root segment to the end of the root branch (e.g. root tip = 0 cm).

  • b

    Position of the branch within the root system (i.e. top, side, centre or bottom).

  • Only fluxes out of 0.5 strength LAS were included in the analysis.

  • d.f., degrees of freedom; SS, sum of squares.

Error2144.7595  0.74713  

Up-scaled to the SA of root orders within a root branch, a branch with four root orders (SAs of 50.84, 28.5, 14.6 and 6.1% of root orders 1–4, respectively) possessed a total water uptake of 0.268 g cm−2 h−1. Root orders 1–4 contributed 134, 17, 6 and −67% to this total water uptake (0.268 g cm−2 h−1 = 100%).


A paradigm shift in the understanding of root system organization was initiated by Kurt Pregitzer (2002); however, studies that characterize root systems by orders and not by arbitrary size classes are scant. The maximal number of root orders in trees was speculated to be seven (Pregitzer et al. 1997; Wells & Eissenstat 2003). In this study, we have shown that young C. volkameriana root systems can hold eight different orders (Fig. 4).

In roots of woody species analysed to date, morphology changed systematically within different root orders, with the first root order having the thinnest diameters and the highest specific root lengths (Pregitzer et al. 2002; Guo et al. 2004, 2008; Wang et al. 2006, 2007; Valenzuela-Estrada et al. 2008). Accordingly, the first root order in C. volkameriana was the thinnest root order (Fig. 3a). Mean Citrus root diameters increased modestly until the third root order (13 and 27% in relation to the next lower root order, respectively), while a steeper increase occurred in higher root orders (52, 80 and 124% related to the next lower root order, respectively). We hypothesize that ‘initiated’ secondary growth is responsible for the higher diameter increase in root orders ≥4. Interestingly, 90% of the roots in the first three orders, nearly 50% of the fourth root order and even some roots of the fifth order had diameters <1 mm (see dotted line in Fig. 3a) and possessed a great overlap in diameters between orders. Applying the classical ≤2 mm diameter threshold for defining ephemeral, non-woody fine roots (see dotted/dashed line in Fig. 3a) would only account for 50% of the fifth and higher root orders as structural woody roots, while in this study, root orders ≥4 have been found to be lignified and thus must be considered woody roots. Our results confirmed the limitations of diameter class-based analyses, especially for analyzing traits within root systems or between species (Comas & Eissenstat 2009). TD increased and SRA decreased in C. volkameriana with increasing root order (Fig. 3b,c), as shown before (e.g. Valenzuela-Estrada et al. 2008). In consequence, lower root orders in C. volkameriana seem to have higher carbon use efficiencies in providing SA for uptake processes (see also Wells & Eissenstat 2003 and references therein).

The finest two to three orders are considered to primarily serve for water and nutrient absorption, and the next higher root orders are transitional between the functions of absorption and transport (Guo et al. 2008; Valenzuela-Estrada et al. 2008). In this study, we found that water flux was specifically dependent on the position of a root segment within the branching system (‘root order’) rather than other morphological or spatial characteristics (Table 1). Other root characteristics such as the position of a root segment within the whole root system (e.g. at the top or bottom), its distance to the lateral end of the root branch (as a rough measurement of age) and the root diameter were only of minor importance and were not significant in determining the water fluxes (Table 1).

The first three root orders of C. volkameriana possessed water uptake (Fig. 6), which is in accordance with the previously predicted function of these orders (Guo et al. 2008; Valenzuela-Estrada et al. 2008) and with other studies considering developmental and age effects on radial root hydraulic properties (see Steudle 2000 and Wells and Eissenstat 2003 and references therein). Changes in the development of Casparian bands and suberin lamellae (Peterson, Murrmann & Steudle 1993), in the development of a suberized exodermis or peridermis (Moon et al. 1986), and/or in aquaporin expression (Kaldenhoff et al. 2008) might have resulted in the observed differences. Interestingly, the water flux density (i.e. flux related to SA) in the first root order was 4.5 times higher, but the water flux efficiency (i.e. flux related to dry mass) was 6.8 times higher than those of the second root order. It might be concluded that the physiological differences between these orders are less important for root system functioning than the ecological differences, that is, a higher water uptake of the first root order in respect to the invested biomass (water flux efficiency). Further studies, including respiration and turnover rates, will be needed to determine water uptake efficiencies among root orders.

Surprisingly, a significant water efflux was measured in fourth/fifth root orders, representing ‘hydraulic redistribution’ (HR). The water efflux occurred in all measured fourth root order segments and is thus believed to be a stable phenomenon in hydroponically grown C. volkameriana rootstocks. Although it was found that HR could occur at soil water potentials close to zero (Warren et al. 2007), slight potential differences must have initiated the excess flow. Our analysis of the osmolality of root tissue outside of the stele revealed a significantly higher osmolality in non-stele tissues of the fourth root order than in the cortex of the second root order (Fig. 5b); however, differences between the first and third root orders and the fourth root order remained marginally significant (P < 0.1). Another possible cause is the release of root exudates to increase nutrient uptake (e.g. Hinsinger 1998; Traoréet al. 2000), subsequently lowering the water potential in the miniature depletion chamber. Indeed, the osmolality of the solution within the chambers attached to fourth root order roots was found to be higher than the 1.0 strength LAS in hydroponics, indicating a strong release of exudates (Fig. 5a). While the purposes of the exudation remain undetermined, we hypothesize that the measured HR might be useful for plants to keep the surrounding of structural roots moist either to increase nutrient uptake or to prevent root desiccation (Comas, Bauerle & Eissenstat 2010; Hodge 2010). It is vital to keep high-order roots intact since they play a major role in transport even if they possess some efflux. Our measurement with an inanimate object revealed that no artificial fluxes occurred. Furthermore, besides the differences in fourth root order cortex and chamber solution osmolalities, no gravimetric (i.e. by equal solute levels in the storage container and in the pot) or osmotic [i.e. by higher osmolality of the LAS in the pot than (initially) in the chamber and the storage container] gradients were present, which could explain the observed water efflux.

A number of different methods have been used in the past for studying traits of specific root regions. However, to our knowledge, none of them was able to distinguish water uptake rates between root orders. In our miniature depletion chamber approach, like in many other studies, roots were measured in aqueous solution to increase control of the solute composition and to eliminate the effect of soil hydraulic conductivity (Tyree 2003) and root temperature (Aston & Lawlor 1979). In this study, the flux density of first-order roots in DW was found to be 30% higher than in 0.5 strength LAS (Fig. 6a), likely related to the lower osmolality of DW as compared with the 0.5 strength LAS.

Considering the distribution of root orders within a branch, the fourth and fifth root orders possess ca. 10% SA versus 90% SA build by root orders 1–3; the water uptake of the whole branch remained positive. Up-scaling the measured flux densities (Fig. 6a) according to the SA per root orders (Fig. 3d) resulted in a flux density of 0.268 g cm−2 h−1 per C. volkameriana root branch (with four root orders). Because the hydroponic pots were filled with 1.0 strength LAS but the measurement chambers contained 0.5 strength LAS and the root exudation of fourth order roots occurred within the limited volume of the chambers, we assume that influx and efflux rates were higher than in the rest of the root system. Comparable data sets about biomass- or SA-related uptake rates of roots are rare. Observed root SA-related uptake rates in central European tree species in situ range between 0.02 and 1.4 g cm−2 d−1 (Coners & Leuschner 2002; Korn 2004; Leuschner, Coners & Icke 2004; Burk 2006). Kramer & Bullock (1966) calculated for young, white pine root segments water flux densities up to 0.178 g cm−2 h−1. Thus, the measured flux densities in C. volkameriana seem to be within a realistic range but should be rather interpreted in relation to each other than as absolute numbers because of the artificial set-up.

The method we describe here appears promising for relatively non-invasive measurements of root water uptake rates. The low cost of the method and its simplicity and ease of use should permit its wide use in laboratories as a complementary method to other techniques. In addition, we have described a novel phenomenon in which HR is related to high-order roots of trees, which exude water even when grown in hydroponics. Future studies should seek to identify the mechanistic role of the exudation, study differences in anatomy, physiology and morphology that underlie the observed uptake rates and compare various species subjected to different water potential gradients.


The authors wish to thank E. Raveh (Gilat Experimental Station) for providing the Citrus plants, M.D. McCue, L. Rose and two anonymous reviewers for helpful comments on a previous version of this manuscript, and T. Gendler, Liron Summerfield and O. Shelef for their help regarding root dissection and image analysis. B.R. was partially supported by a postdoctoral fellowship awarded by the Jacob Blaustein Center for Scientific Cooperation (BCSC).