|II.||Mechanisms controlling HR||832|
|III.||Below-ground processes affected by HR||833|
|IV.||Ecological implications of HR||835|
|II.||Mechanisms controlling HR||832|
|III.||Below-ground processes affected by HR||833|
|IV.||Ecological implications of HR||835|
Hydraulic redistribution (HR) is the passive movement of water between different soil parts via plant root systems, driven by water potential gradients in the soil–plant interface. New data suggest that HR is a heterogeneous and patchy process. In this review we examine the main biophysical and environmental factors controlling HR and its main implications at the plant, community and ecosystem levels. Experimental evidence and the use of novel modelling approaches suggest that HR may have important implications at the community scale, affecting net primary productivity as well as water and vegetation dynamics. Globally, HR may influence hydrological and biogeochemical cycles and, ultimately, climate.
If you can't find a tool you're looking for, please click the link at the top of the page to "Go to old article view". Alternatively, view our Knowledge Base articles for additional help. Your feedback is important to us, so please let us know if you have comments or ideas for improvement.
Water relations are key to understanding the ecology of terrestrial plant communities, and one of the components that determines water balance is the process of hydraulic redistribution (HR; Caldwell et al., 1998). The term refers to the passive movement of water through plant roots from moist to dry soil layers; a phenomenon that improves plant transpiration and other plant processes through water provision to upper roots and its redistribution within the root system. Water efflux from roots in glasshouse experiments was first described in the 1930s (see Ryel, 2004; Caldwell et al., 1998 for a historical perspective) but it was not until 1987 that Richards & Caldwell (1987) coined the term ‘hydraulic lift’ to describe the upward movement of water from deep wet to shallow dry soil layers. Water may also move downwards (inverse hydraulic redistribution; Burgess et al., 1998) or laterally (Smart et al., 2005) depending on where dry and moist soil layers are located in the soil profile (Fig. 1). All these processes were termed HR (Burgess et al., 1998).
Hydraulic redistribution is a passive process driven by water potential gradients between soil layers as far as these are connected by an intact root system (Leffler et al., 2005). During the daytime, water is drawn up as plant transpiration forces water inflow from the soil through the stem and out to the atmosphere via stomata. When stomata are closed and transpiration is very low or suppressed at night (for all species except plants with crassulacean acid metabolism), plant water potential equilibrates with that of the soil where most active roots are found, which can result in water potential gradients between the plant and the drier soil points; hence, water flows from roots to these dry soil layers (Fig. 2). Daytime stomatal closure has also been reported for some species under very high vapour pressure deficit (VPD) conditions when radiation is high. Under such conditions, HR can also take place (Espeleta et al., 2004).
The HR process seems to be ubiquitous among plants (Caldwell et al., 1998) as long as conditions favour its occurrence. In some biomes, such as nonsaline wetlands, where soil moisture is high and rather uniform, or prairies where roots may not reach deep enough for plants to reach soil layers with different moisture, HR may not occur. However, only in a very few cases was HR not detected (Huang, 1999; Wan et al., 2000; Scholz et al., 2008; Espino & Schenk, 2009; Grigg et al., 2010); among other reasons this was because it becomes technically difficult to evidence (Caldwell et al., 1998) and serious experimental work is needed to demonstrate that, whenever conditions are favourable for HR, the process does not occur.
Hydraulic redistribution can be detected through a variety of techniques (Table 1) but these are often based on detecting daily fluctuations in bulk soil moisture around roots (Fig. 2). However, these techniques require appropriate controls in root-free soil and time-consuming calibrations to quantify the influence that temperature or other soil processes have on soil moisture (i.e. liquid and vapour water transport) (Katul & Siqueira, 2010; Warren et al., 2011). Alternative methods that do not require the use of such controls include direct measurements of xylem water movement in the root system (Brooks et al., 2002; Bleby et al., 2010) or the use of enriched water isotopes (2H and 18O) applied locally to roots or soil and monitored in other parts of the root system and soil around roots engaged in HR (Dawson et al., 2002). A more detailed description of these methods can be found in Ryel (2004). The most recent techniques include the use of radiography and radioactive isotopes (3H) but their cost and associated difficulties have largely restricted their use (Hawkins et al., 2009; Carminati et al., 2010; Moradi et al., 2011).
|Type of technique||Technique||Variable measured||References|
|Soil moisture||Soil psychrometers||Water potential (Ψ)||Nondestructive||Brown & Bartos (1982); Richards & Caldwell (1987)|
|Time domain reflectometry||Water content (θ)||Nondestructive||Topp et al. (1996); Wan et al. (2000)|
|Frequency Domain Capacitance||Water content (θ)||Nondestructive||Paltineanu & Starr (1997); Brooks et al. (2002)|
|Sap flow||Heat ratio method||Sap velocity/direction||Semidestructive||Burgess et al. (2000, 2001);|
|Heat field deformation||Sap velocity/direction||Semidestructive||Nadezhdina et al. (1998, 2006)|
|Thermal dissipation||Sap velocity/direction*||Nondestructive||Granier (1987); Brooks et al. (2002)|
|Isotopic||Deuterium (2H)||Direction/quantity of water redistributed||Destructive||Dawson (1993); Brooks et al. (2002)|
|Oxygen (18O)||Direction/quantity of water redistributed||Destructive||Querejeta et al. (2007); Lilleskov et al. (2009); Armas et al. (2010)|
|Dye||Fluorescence||Direction/velocity||Destructive||Egerton-Warburton et al. (2007)|
|Radioactive||Neutron radiography||Water content (θ)||Nondestructive||Carminati et al. (2010); Moradi et al. (2011)|
|Tritium (3H)||Destructive||Hawkins et al. (2009)|
Although we have a good understanding of the HR process and several reviews have summarized previous findings (Dawson, 1993; Caldwell et al., 1998; Horton & Hart, 1998; Jackson et al., 2000; Ryel, 2004), research over the past 10 yr has yielded a wealth of knowledge on different aspects of this process. These recent data have shifted our attention on HR from an ecophysiological curiosity to an ecologically significant process ranging from molecular to global scales.
Water movement in unsaturated soil layers depends on both soil water potential and gravity, but when gravity is no longer operating, water potentials (matric and osmotic) are the main force for water movement and act as drivers for HR. However, recent experimental evidence indicates that HR is much more complex than previously thought and a complex source–sink system in the plant–soil interface also influences the process (Scholz et al., 2002, 2008). A network of water potential gradients acting as source–sink can be established within the plant as well as between the plant and the surrounding soil, changing HR patterns and magnitude (Scholz et al., 2002; Nadezhdina et al., 2010). Water moves within this network following water potential gradients that depend on soil conditions (i.e. soil moisture), plant physiological status (water relations and differences in water potentials of a variety of tissue/plant organs), and environmental conditions (VPD) (Scholz et al., 2002). This may be the case with night-time transpiration, a common process in species able to redistribute water (Dawson et al., 2007), which may alter competitive sinks for hydraulically redistributed water. Night-time transpiration reduces leaf water potential and forces water up to the canopy, reducing flow to soil, which at night is negatively related to the difference between soil and leaf water potential (Scholz et al., 2002, 2008; Hultine et al., 2003, 2004; Prieto et al., 2010a). Howard et al. (2009) showed that night-time transpiration driven by high VPD effectively decreased HR by up to 73%, a substantial decrease that could affect total carbon gain by lowering plant transpiration rates (Ryel et al., 2002).
Other processes illustrate the source–sink interactions within the plant that affect HR patterns (Fig. 1) and evidence the complex mechanisms that control HR and its heterogeneous and patchy nature (see Bleby et al., 2010).
The bidirectional movement of water in and out of roots may imply that water does not meet differential resistance to flow through nonsuberized roots moving in both directions (a detailed discussion on this issue is given in Caldwell et al., 1998). However, most water exchange occurs in the young and distal portions of the root system (Caldwell et al., 1998), and properties of this area affect HR patterns and water flow differently (Dawson, 1997; Warren et al., 2007; Scholz et al., 2008). There is greater root resistance to water flow from roots to the soil than from soils into roots (Dawson, 1997; Schulte, 2006; Valenzuela-Estrada et al., 2009). Root resistance to water flow through roots may change on a daily basis. A large part of this temporal variation is thought to be a consequence of the activity of water-permeable proteins or aquaporins (Henzler et al., 1999). An up-regulation of aquaporins during the day in Quercus fusiformis enhanced deep root hydraulic conductivity, whereas aquaporin activity at night was reduced, decreasing root hydraulic conductivity and water flow (McElrone et al., 2007). This may imply a short-term indirect control on plant water loss in upper soil layers through a decrease in water uptake from deep roots or even a decrease in night-time water outflow from roots to dry soils. However, aquaporins are involved only in one of the three main water pathways from soil to roots, as described by the composite transport model (Steudle, 2000), contributing to only about half of the total root conductivity (Siefritz et al., 2002; McElrone et al., 2007). Moreover, McElrone et al. (2007) found that fine root hydraulic conductivities during the night associated with aquaporin activity were greater under field than under laboratory conditions, probably because of a greater demand of water for HR in the field. Thus, the degree to which, if any, aquaporins can control or contribute to HR processes still remains unknown.
Root architecture, morphological attributes such as root density and spatial distribution, and life history traits also affect HR. HR is usually detected in species with dimorphic root systems, whereas monomorphic species lacking true lateral roots in shallow soils usually do not exhibit HR (Scholz et al., 2008; Grigg et al., 2010). Hydraulic connectivity between roots is also a major factor controlling water redistribution. Water flows from a lateral root in wet soil, axially upwards through the stem and then downwards through the stem to the opposite root in dry soil provided these roots are interconnected through the stem (Burgess & Bleby, 2006; Nadezhdina et al., 2009). Radial flow through the stem interconnecting opposite roots of a tree may thus limit the amount of water that can be redistributed within a single night. In highly modular plants with a high stem physical segmentation this may represent an advantage rather than a drawback, as it would prevent water loss through roots in very dry soil (Espino & Schenk, 2009). Not surprisingly, the degree of modularity in plant species decreases from wet to dry environments (Schenk et al., 2008). Finally, suberization of deep roots (Ryel et al., 2003) and loss or death of fine roots (i.e. by low soil water potentials) can also reduce or stop HR.
Mycorrhizal fungi associated with plant roots often improve plant water relations through increased extension of root growth and increased surface absorption (Allen, 2007; Lehto & Zwiazek, 2011), playing a major role in hydraulic redistribution processes. Under dry soil conditions, water can move from roots to soils via mycorrhizal fungi tips (Querejeta et al., 2003) and may also increase the magnitude of HR. For example, efflux of water into the soil is indeed directly correlated to the abundance and viability of ectomycorrhizal fungi in Quercus agrifolia seedlings (Egerton-Warburton et al., 2008). In the same way, mycorrhizal fungi may also increase the rate of downward HR (i.e. after a rain event) through greater water absorption as a result of exploration of greater soil volumes than roots alone, potentially increasing the rate of water recharge to deep soil layers; but whether mycorrhizas play a relevant role in this process still remains unknown.
Soil type and condition may also affect HR magnitude and patterns. Soil texture determines the amount of water that a soil can retain, as well as the amount of water that can be extracted by plant roots at different soil water potentials (Hultine et al., 2004). Similarly, soil texture affects the amount of water that roots can release, with decreasing magnitudes of HR in coarser soil types (Yoder & Nowak, 1999; Aanderud & Richards, 2009; Wang et al., 2009), but the relationship between HR and soil texture is not linear (Prieto et al., 2010b) and may result in soil context-dependent HR patterns.
Two other soil properties that determine HR direction and magnitude are conductivity and salinity. The interstitial osmotic potential (Ψo) of soil can determine the rate and direction of water movement within the root system independently of the soil matric potential. Dwarf mangle trees (Rhizophora mangle) with roots connecting soil layers with different salt concentrations were engaged in HR independently of soil matric potential (Hao et al., 2009). Although matric potentials were similar in both layers, water moved upwards following the osmotic potential gradient from lower (higher Ψo) to higher salt concentrations (lower Ψo). As mangle roots are effective in excluding salt during water uptake, the authors hypothesized that freshwater movement upwards should help reduce the high salinity in shallow layers and allow dwarf mangle to avoid damage caused by high soil salinity.
Do roots benefit from HR? In general roots have longer vessels, which are more vulnerable to cavitation and hydraulic failure than stems, and therefore have narrower safety margins (Martínez-Vilalta et al., 2002). Hence, although roots may be subjected to less tension than stems during transpiration as predicted by the tension-cohesion theory (Tyree, 1997) through overnight provision of water via HR, plants can avoid drops in root water potential below the level that would cause hydraulic failure, a crucial process to withstand long drought spells (Domec et al., 2004). HR alone may be able to maintain root hydraulic conductivities at safe water potentials (Domec et al., 2006) and often root embolism can recover when HR occurs (Domec et al., 2004). Loss of root conductivity negatively affects both root water transport capacity and leaf stomatal conductance. Thus, reducing plant water stress and maintaining or recovering root conductivity overnight may help to increase stomatal conductances the next day and overall carbon gain, increasing plant fitness (Domec et al., 2006). However, maintaining root conductivity in dry soils may come at the expense of losing more water through HR. Embolism formation and losing part of the root system comprise a mechanism of preventing whole-plant failure and water loss through roots, which improves plant survival under drought conditions (Sperry & Hacke, 2002). Nevertheless, maintenance of active roots in moderately dry soil may improve plant water relations through increased use of transient soil moisture (e.g. episodic summer rain events) (Loik, 2007). In this direction, greater root resistances to cavitation may become associated with a greater capacity of the plant to lift water (Sperry & Hacke, 2002).
Caldwell et al. (1998) suggested that a reduction of soil drying rates through HR would extend fine root survival and growth; Bauerle et al. (2008) tested this hypothesis using two Vitis sp. cultivars. Plants were irrigated on one side only and application of night-time illumination in half the plants stopped HR. They observed that root water potentials were twice as negative in plants where HR was prevented than in plants where HR occurred. The latter had 50% greater root survival in dry soil, and rehydration of fine root tissues through internal HR was responsible for this effect. Since roots engaged in HR live longer, they could take up water and nutrients for an extended time. In addition, these roots do not need to be replaced at a carbon cost to the plant, which may have relevant implications for plant carbon budget (Bauerle et al., 2008). However, when roots grow old or soil becomes dry, suberization or metacutization of roots (Leshem, 1965; Ryel et al., 2003) could reduce water exchange between soil and roots, and thus reduce or stop HR.
In the soil matrix, nutrients are found both in a solid phase and in water solution. Microbial activity decomposes soil organic matter, releasing mineral ions that incorporate into the soil solution. Ions need to be transported close to the roots for their absorption and this movement occurs via mass flow, driven by transpiration fluxes, or via diffusion (Nye & Marriot, 1969). Roots are also chemically active at exuding organic carbon, organic anions (P, Fe), protons and phosphatase, fuelling decomposition and increasing nutrient availability (Cardon & Gage, 2006). All these processes depend highly on soil moisture conditions (Tinker & Nye, 2000). It could be hypothesized that increased soil moisture at the surface by HR may maintain nutrients available to plants for longer periods. In an Artemisia tridentata stand, HR maintained soil water potentials above the threshold at which nutrient diffusion stops (Ryel et al., 2002), and Matzner & Richards (1996) suggested that the ability of A. tridentata shrubs to maintain nutrient uptake under very dry soil conditions (< −5 MPa) was enhanced by release of water through HR, although they did not directly test it. Dawson (1997) showed that nutrient availability increased around roots of plants engaged in HR, nitrogen being the most affected. However, the direct effects on plant nutrient uptake were not tested. There is, however, circumstantial evidence pointing to greater nutrient uptake in dry soils linked to HR, although it was difficult to tell whether nutrient uptake was directly linked to water efflux or to other side-effects (de Kroon et al., 1998; Huang, 1999; Valizadeh et al., 2003; Leffler et al., 2004). Using a different approach, Snyder et al. (2008) increased HR rates by artificially suppressing night-time transpiration and observed, however, a trend to lower N uptake in plants with greater HR rates. Overall, a problem common to all these approaches is that either they did not include a true control (i.e. plants not engaged in HR) or nutrients were supplied as liquids, and thus readily available for uptake independently of soil moisture conditions (see Armas et al., 2012).
However, there is ample evidence supporting the idea that HR enhances plant nutrient uptake (Fig. 3). HR effectively enhances organic matter decomposition and thus nutrient mineralization rates, releasing nutrients that could be taken up by plants (Aanderud & Richards, 2009; Armas et al., 2012). Organic matter decomposition rates were 25% greater in dense- than in sparse-root patches under A. tridentata shrubs that also had lower soil water potentials (Aanderud & Richards, 2009). This decomposition increase correlated to a threefold greater HR magnitude observed in dense root patches. Organic matter decomposition rates were also 35% higher, and nitrogen mineralization tended to be greater, in soils where buffalograss (Bouteloua dactyloides) plants were engaged in HR than in those where HR was prevented (Armas et al., 2012). Enhanced soil moisture and the daily drying–rewetting cycle of HR could create a daily rhythm of root water flux and rhizodeposition that increases microbial activity and decomposition rates (Cardon & Gage, 2006). After > 2 months of drought, buffalograss plants engaged in HR also took up greater amounts of labelled 15N added to the soil as dry organic matter than those with HR prevented (Armas et al., 2012). In a similar study, I. Prieto et al. (unpublished) observed greater 15N uptake from nutrient-rich soil patches (dry organic matter) in Retama sphaerocarpa shrubs that were engaged in HR. Cladode 15N was directly correlated to the amount of water released into dry soil, indicating that HR was responsible for this effect. Hydraulic lift also enhanced root growth in nutrient-rich patches (root foraging precision), suggesting that HR could play a direct role in root foraging strategies. Interestingly, inverse HR (downward HR) has also been involved in nutrient uptake; McCulley et al. (2004) used nutrient concentration and stable isotope data to suggest that downward HR might promote the uptake of nutrients stored in deep, dry soil layers where it was otherwise unavailable to plants.
Redistributed water can also move from roots and into the soil via mycorrhizal fungal tips, allowing them to remain active in dry soil conditions (Querejeta et al., 2003). It is hypothesized that this process could indirectly enhance nutrient uptake from dry soils, extending the period of nutrient uptake by plants. Egerton-Warburton et al. (2008) tested this hypothesis in Quercus agrifolia seedlings growing in split chambers separated by a fine mesh that excluded roots but allowed ectomycorrhizal fungi to grow through. They found that water efflux into the soil in the fungal compartment was correlated to both plant N uptake and soil bacterial abundance and enzymatic activity, suggesting that water released through the extramatrical hyphal mycelia enhanced both plant nutrient uptake and soil microbial processes. Since diffusion of nutrients becomes limited in dry soil, water efflux through mycorrhizal fungi may also increase diffusion rates and thus nutrient uptake by plant roots associated with mycorrhizal fungi (Egerton-Warburton et al., 2008). However, nutrient uptake correlated with plant water uptake because of the presence of ectomycorrhizal fungi and hence the direct role of HR was not clear. These authors proposed two mechanisms by which HR may influence nutrient uptake by plants: an indirect mechanism mediated by increased soil water availability that enhances microbial and enzymatic activity, and a direct role of extramatrical hyphae in maintaining nutrient uptake, as HR in dry soils tends to increase hyphae functionality and likely nutrient diffusion rates (Querejeta et al., 2003, 2007; Lilleskov et al., 2009).
Hydraulic redistribution acts at the plant level but also influences the structure and function of plant communities (Katul & Siqueira, 2010). In this section we will focus on recent findings of global implications of HR and its importance in structuring plant communities, including its mediation in plant–plant interactions, its influence on biogeochemical cycles and ecosystem-level effects (Fig. 3).
The effect that species engaged in HR have on their neighbours is still unclear. There are reports of positive, neutral or even negative effects depending on ecosystem type, plant life form or whether donor and receiver species shared common ecto- and endomycorrhizal networks (CMNs; Table 2). There are two possible pathways by which neighbouring plants can access water redistributed overnight by a plant engaged in HR: indirectly from the soil after its release by shallow roots; or directly through CMNs linking different individuals within a plant community (Egerton-Warburton et al., 2007; Warren et al., 2008). This last pathway seems to be the most efficient, as twice as much water was transported between donor and receiver Ponderosa pine trees (Pinus ponderosa) directly via CMNs (Warren et al., 2008). In this study, the rate of water transport between large trees and seedlings was high (0.16–0.63 m d−1), which may be an indication that large amounts of water could potentially be transferred through CMNs (in this case, ectomycorrhizal CMNs). However, the relative importance of each pathway on water transport between plants still remains uncertain (Schoonmaker et al., 2007).
|Sp. engaged in HR||Target species||Growth form||Measured variable||Net effect||Reference|
|Acer saccharum||Asarum canadense||Grass||+||Dawson (1993)|
|Acer saccharum||Acer saccharum||Tree||δD/gs||+||Dawson (1996)|
|Pinus halepensis||Pistacia lentiscus||Shrub||δD/δ13C/Ψpd||+||Filella & Peñuelas (2003)|
|Acacia tortilis||Cynodon dactylon||Grass||δD/Biomass (g)||–||Ludwig et al. (2004)|
|Cajanus cajan||Zea mays||Grass||δD||ND||Sekiya & Yano (2004)|
|Markhamia lutea||Oryza sativa||Grass||gs||–||Hirota et al. (2004)|
|Prosopis velutina||Zanthoxylum fagara||Shrub||Ψl/A||+||Zou et al. (2005)|
|Pseudotsuga menziesii||Berberis nervosa||Shrub||δD||ND||Brooks et al. (2006)|
|Pseudotsuga meinziesii||Pseudotsuga meinziesii||Tree||δD/Biomass/Ψpd||Neutral||Schoonmaker et al. (2007)|
|Pinus ponderosa||Festuca pallescens||Grass||δD/Growth||–||Fernández et al. (2008)|
|Protea (Sylvia)||Leysera gnaphalodes||Shrub||δD||+||Hawkins et al. (2009)|
|Pistacia lentiscus||Juniperus phoenicia||Shrub||δD/Ψpd/gs/δ13C/A/RWC/(Fv/Fm)||–||Armas et al. (2010)|
|Retama sphaerocarpa||Marrubium vulgare||Shrub||Survival/biomass||+/−||Prieto et al. (2011)|
Water parasitism has been widely documented in the literature (Caldwell, 1990; Dawson, 1993; Filella & Peñuelas, 2003) and refers to the capacity of understorey species growing next to a species engaged in HR to take up water lifted at night (Caldwell, 1990). The first evidence was obtained by Caldwell & Richards (1989), who observed a similar isotopic signature in water released in upper soil layers through HR by A. tridentata shrubs and in stem water of neighbouring Agropyron desertorum plants. Since then, evidence of water parasitism has been found in several ecosystems and in plant species with different life forms (Table 2). Deep-rooted species may shed water overnight through HR, potentially benefiting other species, but they also compete with their neighbours for water during the daytime. Ludwig et al. (2004) reported an intense competition for water between Acacia tortilis and three grass species in its understorey. Grasses effectively took up water redistributed by A. tortilis but when tree roots were experimentally removed, preventing access to redistributed water by grasses, the biomass of the latter was greater. Thus, although grasses effectively took up redistributed water, competition between tree and grasses outweighed the potential positive effect of HR. Other studies using grass species have reported similar results (Table 2). However, in mesic ecosystems where water is not as limiting, positive effects have been reported (Dawson, 1993). Grass species usually have shallow root systems (Schenk & Jackson, 2002a) and a great capacity to extract soil water, which makes them great competitors, especially in dry ecosystems where water sources are scarce (Armas & Pugnaire, 2011). In association with a species engaged in HR, the amount of water redistributed may not be enough to meet the grass requirements, and depletion of water by the nurse during daytime would exert a greater negative effect on the grass. In this case, competition would be the predominant outcome of the interaction (Ludwig et al., 2004). On the other hand, grasses usually senesce when soil water potential becomes low and HR may benefit grass species by lengthening the period before soil moisture depletion and senescence occurs (Meinzer et al., 2004), although direct evidence of this effect is scarce (Hirota et al., 2004).
In addition, studies with shrub species living in association with a tree or shrub engaged in HR usually report positive or neutral effects (Table 2). For example, Zou et al. (2005) showed differential effects of HR by Prosopis glandulosa trees on neighbouring shrubs that ranged from positive to neutral, concluding that the effect of HR was species-specific. Shrubs are usually deep-rooted and have a less extensive shallow root system than grasses (Schenk & Jackson, 2002a), gaining access to deeper water sources and becoming less dependent on water extraction from shallow layers when these are dry. In this case, depletion of shallow soil water by the tree engaged in HR may exert a small effect on a shrub’s performance. Dawson (1993) observed a strong linear relationship between plant growth and the ratio of redistributed water taken up by different understorey species. Although individuals belonging to both life forms (grasses and shrubs) benefited from the presence of the tree, herbaceous species took up, on average, a greater proportion of redistributed water than did shrubs.
Two interesting questions arise concerning HR and plant interactions. The first of these is whether an opposite trend might be seen when the plant is engaged in downward HR. This process may reduce the availability of rainwater to neighbouring shallow-rooted species, although the net effect of such process has not yet been measured. Secondly, it raises the question as to whether water parasitism is the only outcome or whether species engaged in HR also benefit from the association with neighbours, even at the cost of losing some water. Some evidence comes from a case study in southeastern Spain where Prieto et al. (2011) observed that, although competition for water was strong among the shrub Retama sphaerocarpa, the understorey annuals and Marrubium vulgare seedlings, Retama facilitated survival of Marrubium under its canopy through HR. This positive effect may later indirectly benefit Retama shrubs, as its association with adult Marrubium plants improves nutrient uptake and water relations (Pugnaire et al., 1996). Thus, losing a fraction of water redistributed at night may bring an indirect benefit from a species that depends on HR for survival during drought periods.
In summary, neighbours associated with plants engaged in HR seem to benefit from the extra moisture and increased survival rates, and even biomass, which could potentially increase their fitness as they may establish in the community and live long enough to grow and reproduce in subsequent seasons (Prieto et al., 2011).
Plant-mediated mobilization of nutrients and mineral elements from different soil pools could have broad effects on soil development at the ecosystem scale (Verboom & Pate, 2006). For instance, HR plays an active role in soil-forming processes through translocation of metal compounds under mallee eucalypts (Eucalyptus incrassata, Eucalyptus pleurocarpa, Eucalyptus decipiens and Eucalyptus occidentalis) growing in a sand dune system in Australia (Verboom et al., 2010). Some metal compounds (Al, Si, and Fe) associated with clay formation in sand dunes were transported from deep soil layers via HR and deposited in shallow soil layers, where they were involved in soil construction around eucalyptus roots.
The downward movement of water through HR could help mobilize nutrients stored at great depths (Walvoord et al., 2003; McCulley et al., 2004; Rumpel & Kögel-Knabner, 2011), indirectly allowing nutrients to be transported upwards to surface layers. Nutrients are usually stored in shallow soil layers but in some arid and semi-arid ecosystems a reservoir of nitrate (NO3−1) is stored deeper in the soil (Walvoord et al., 2003; but see Jackson et al., 2004). The efflux of water in deep soil layers through inverse HR may help increase nutrient uptake by deep roots (McCulley et al., 2004) and incorporate deep-stored nutrients into plant tissues (i.e. leaves) in a process that may help to increase the rate of ‘nutrient uplift’ to shallow soil layers when these leaves drop and decompose (Matzner & Richards, 1996; Jobbagy & Jackson, 2004). This ‘nutrient uplift’ may remobilize nutrient pools that are otherwise inaccessible to the plant community. The mobilization and redistribution of spatially distinct pools of nutrients and other elements via HR may also have broader ecosystem-level consequences related to increased productivity (Liste & White, 2008) and changes in hydrological and biogeochemical cycles (Jackson et al., 2000; Jobbagy & Jackson, 2004; Aanderud & Richards, 2009). Evidence obtained in a semi-arid sand dune system suggests that some salt-tolerant species could release salty water in upper soil layers through HR (Armas et al., 2010). Deep-rooted, salt-tolerant Pistacia lentiscus shrubs exerted a negative effect on the performance of salt-intolerant Juniperus phoenicia shrubs, caused by the release of salty groundwater into shallow soil layers by Pistacia. During the dry season, when HR took place, xylem sap water isotope signature and osmolality were identical in Juniperus shrubs growing in close association with Pistacia and in Pistacia shrubs, whereas sap characteristics were unique in Juniperus shrubs growing alone. Thus, water-soluble compounds like small ions (K+, Na+ or Cl−) could be directly transported by the sap flow and released through HR. Similarly, Salim (1988) showed a net loss of Na+ from roots and stems tissues to the rhizosphere in Vigna radiata. If nutrients or water-soluble ions are redistributed with water, HR could have important consequences for the biogeochemical cycles of some elements.
Plant roots are most dense in the upper 2 m of soil and deep roots usually represent < 10% of the total amount of roots (Canadell et al., 1996; Schenk & Jackson, 2002a,b). Hence, moisture is usually depleted at a faster rate by evaporation and uptake in shallow soil layers. Lifted water, however, can provide a source of soil moisture, which may enhance plant transpiration rates the next day (Caldwell & Richards, 1989). Transpiration increases 10–40% as a result of HR in tropical systems (da Rocha et al., 2004), 20–25% in dry and arid environments (Ryel et al., 2002; Bleby et al., 2010), 19–40% in mesic forests (Jackson et al., 2000) and up to 81% in some Mediterranean ecosystems (Kurz et al., 2006). But what are the main ecological implications for communities? Techniques used to monitor ecosystem gas exchange (eddy covariance towers) and soil and plant physiological parameters along with implementation of results into global gas-exchange models have proved useful at elucidating the implications of HR globally. Lee et al. (2005) implemented a soil-vegetation-atmosphere water transfer model in an Amazonian forest that included the phenomenon of HR (Fig. 4). Results showed that water transfer between deep and shallow soil layers overnight increased whole-stand photosynthesis and summer transpiration by c. 40%. Moreover, the upscaling of the model to a global context showed that the cooling effect associated with the increase in plant transpiration and consequent increase in air humidity affected air temperature seasonal cycles, reducing it by > 2°C in most water-stressed regions of the world during the dry season (Lee et al., 2005). The authors concluded that this cooling effect could be underestimated in some dry ecosystems where HR might account for up to 81% of total tree transpiration during drought (Kurz et al., 2006).
Effects of HR on ecosystem carbon balance may also be potentially important. Domec et al. (2010) implemented a carbon balance model for a loblolly pine (Pinus taeda) forest in North Carolina and included both HR and nocturnal transpiration processes (Fig. 4). They observed a 30–50% HR-mediated increase in whole ecosystem transpiration, which led to an overall increase in plant carbon gain that increased annual gross ecosystem productivity (GEP) and net ecosystem exchange (NEE) by 750 and 400 g C m−2 yr−1, respectively. Increases in GEP and NEE maintained the whole forest as a sink for atmospheric carbon. Overall, the presence of HR may have important implications for ecosystem productivity and CO2 exchange processes in water-limited systems or ecosystems with marked drought periods that could otherwise act as C sources (Domec et al., 2010).
In the long term, increased transpiration rates as a result of HR may also have other ecosystem-level effects. Wang et al. (2011) showed a HR-mediated shift in vegetation composition in the Amazonian forests when modelling HR effects during extreme drought periods (e.g. El Niño years, Fig. 4). Water uptake from deep layers and its release into shallow layers accelerated depletion of deep-water reserves during the drought season, causing massive dieback among drought-sensitive evergreen species and replacement with more drought-tolerant deciduous species (Wang et al., 2011). Enhanced transpiration rates may also affect hydrological cycles in temperate forests (Jackson et al., 2000). In these forests, water available for runoff was modelled to be between 3 and 6% lower to reflect enhanced transpiration rates of trees engaged in HR, lowering by several meters the water table as a result of increased plant water consumption from deep layers.
Inverse HR, on the other hand, can affect overall ecosystem water budgets by increasing deep soil water recharge. Water movement to deep soil layers after a rain event is faster through HR than via infiltration or preferential flow and, in the absence of HR, water recharge of deep soil layers would be very low (Ryel et al., 2003). Deep-water recharge through HR accounted for c. 10% of the rainwater in a tropical rainforest (Lee et al., 2005) and between 29 and 49% in a Prosopis velutina savannah, which supplied 16–49% of drought season transpiration (Scott et al., 2008). Deep-stored water may thus extend the growing season of deep-rooted plants and increase whole ecosystem C gains during drought (Ryel et al., 2002; Scott et al., 2008).
Throughout the world, HR has been reported in over 60 plant species worldwide and in several biomes, from tropical and temperate forests to arid and semi-arid ecosystems. Recent evidence has suggested that the magnitude and frequency of HR depend on several biotic and environmental factors, resulting in a more complex and heterogeneous process than previously thought. Its effects both above- and belowground are increasingly well understood. Aboveground most of its benefits to individual plants come from increased soil moisture in dry soil layers, which in turn affects plant physiology and water relations. Increased soil moisture enhances root growth and function, and rhizosphere processes such as soil organic matter decomposition or nutrient mineralization rates, which may have relevant implications in ecosystem nutrient cycling.
Plant–plant interactions determine plant community structure and function and, although consequences of HR for neighbouring species are many, evidence shows that provision of water through HR can be a facilitative mechanism in many plant communities where recruitment is mostly restricted to the understorey of adult plants. Shared mycorrhizal networks appear to play a relevant role in water redistribution and in the redistribution of this water among plants, although more evidence is needed on the rates and magnitude of water transfer between adult plants and its ecological consequences.
This review has shown that research not only should focus on HR from an individual plant ecophysiological perspective but should also try to understand its effects on a wider scale, as this phenomenon has proved to be important at both community and ecosystem levels. Hydraulic redistribution seems to play a role in water and energy fluxes in the ecosystem, and in recent years our knowledge of community- and ecosystem-level effects has improved by incorporating HR processes into hydrological and ecological models. This puts forward the impact of HR on climate and C budgets, increasing overall productivity and maintaining whole ecosystems as carbon sinks. HR may also impact community composition and modify the hydrology of whole watersheds. In a climate change context, the Intergovernmental Panel on Climate Change has predicted a global increase in air temperatures and VPD, along with more infrequent and bigger rain events in dry ecosystems. Under this scenario, increased temperatures and VPDs would ultimately increase water demand by plants, probably leading to the depletion of deep-water reserves that might preclude the occurrence of HR in some ecosystems, reducing the potential buffering effect of HR on surface-air temperatures and C sequestration. Finally, HR should be increasingly taken into account in both hydrological and ecological models as well as in simulations under different climate-change scenarios to fully elucidate its role in net carbon exchange, gross ecosystem productivity, and water budgets and water transfer through the soil–plant–atmosphere continuum.
We thank Ronald J. Ryel, Jochen Schenk and two anonymous reviewers for their valuable comments on earlier versions of this manuscript. We thank MICINN for funding (refs. CGL2007-63718 and CGL2010-17081). I.P. was supported by a MICINN-FPI fellowship (BES2005-9280).