Oecologia

, Volume 141, Issue 2, pp 335–345

Water conservation in Artemisia tridentata through redistribution of precipitation

Authors

    • Department of Forest Range and Wildlife Sciences and the Ecology CenterUtah State University
  • A. J. Leffler
    • Department of Forest Range and Wildlife Sciences and the Ecology CenterUtah State University
  • M. S. Peek
    • Department of Forest Range and Wildlife Sciences and the Ecology CenterUtah State University
  • C. Y. Ivans
    • Department of Biological SciencesEastern Kentucky University
  • M. M. Caldwell
    • Department of Forest Range and Wildlife Sciences and the Ecology CenterUtah State University
Pulse Events and Arid Ecosystems

DOI: 10.1007/s00442-003-1421-2

Cite this article as:
Ryel, R.J., Leffler, A.J., Peek, M.S. et al. Oecologia (2004) 141: 335. doi:10.1007/s00442-003-1421-2

Abstract

Water conservation is important for plants that maintain physiologically active foliage during prolonged periods of drought. A variety of mechanisms for water conservation exist including stomatal regulation, foliage loss, above- and below-ground allocation patterns, size of xylem vessels and leaf pubescence. Using the results of a field and simulation study with Artemisia tridentata in the Great Basin, USA, we propose an additional mechanism of water conservation that can be used by plants in arid and semi-arid environments following pulses of water availability. Precipitation redistributed more uniformly in the soil column by roots (hydraulic redistribution of water downward) slows the rate at which this water can subsequently be taken up by plants, thus prolonging water availability during periods of drought. By spreading out water more uniformly in the soil column at lower water potentials following precipitation events, water use is reduced due to lower soil conductivity. The greater remaining soil water and more uniform distribution result in higher plant predawn water potentials and transpiration rates later in the drought period. Simulation results indicate that plants can benefit during drought periods from water storage following both summer rain events (small summer pulses) and overwinter recharge (large spring pulse). This mechanism of water conservation may aid in sustaining active foliage, maintaining root-soil hydraulic connectivity, and increasing survival probability of plants which remain physiologically active during periods of drought.

Keywords

Hydraulic redistributionSoil-water recharge Artemisia tridentataSoil-water modelWater conservation

Introduction

Water availability is often limiting to physiological activity for plants in arid and semi-arid environments. Most of these ecosystems are characterized by seasonal cycles from periods of rainfall to periods of extended drought (Smith et al. 1997). Furthermore, precipitation in these systems is often erratic, resulting in only brief pulses of resources (Noy-Meir 1973). Plants that remain physiologically active during extended periods with little precipitation (drought tolerant as opposed to drought avoidant) require effective water conserving mechanisms (Turner 1986) unless sufficient groundwater is accessible.

Most responses of plants to conserve water are reactionary. That is, “water conservation” occurs after the plant experiences reduced water uptake for transpiration or senses impending water shortages in at least part of its rooted soil space. These responses include stomatal control (Comstock 2002) and drought deciduous foliage loss (e.g., Ambrosia deltoidea , Brooks and Burgess 1979; Encelia farinosa, MacMahon and Wagner 1985; Sandquist and Ehleringer 1998; Fouquieria splendens, Killingbeck 1990). Stomatal and structural changes take place once the plant directly senses declining soil water potentials (Davies and Zhang 1991; Davies et al. 1994; Tardieu et al. 1996) to reduce cavitation in the xylem vessels (Sperry et al. 2002) and reduce adverse effects on physiological activity (Tyree and Ewers 1991; Sperry and Hacke 2002). Plants may “anticipate” impending water limitations even when there is adequate water supply if soil moisture in a fraction of the rooted soil volume declines. In this case stomatal aperture or leaf growth can be reduced in response to chemical signals emanating from the portion of the root system experiencing reduced soil moisture even when other portions of the root system are well supplied with water (Blackman and Davies 1985; Zhang et al. 1987; Saab and Sharp 1989; Gowing et al. 1990). However, it is not clear whether reductions in stomatal conductance and foliage loss are simply mechanisms to reduce cavitation and effects to physiological processes (Sperry 2000; Davis et al. 2002) or proactively conserve water. Evidence indicating that plants operate at water potentials rather close to hydraulic failure when under drought stress (Tyree and Cochard 1996; Sperry et al. 1998; Kolb and Sperry 1999a; Hacke et al. 2000; Sperry et al. 2002) supports the former hypothesis.

Water conservation at longer time scales can also be the result of structural changes. Slower rates of water use can result from modification of root:leaf ratios (Sperry et al. 1998, 2002) or production of less foliage or root structure. Leaf structure can reduce water loss, particularly pubescence (e.g., Smith and Nobel 1977; Ehleringer and Mooney 1978; Sandquist and Ehleringer 1998). Plants also produce smaller xylem vessels, reducing the probability for cavitation (i.e. more drought tolerant), but increasing flow resistances (Tyree and Ewers 1991; Hargrave et al. 1994) may result in lower rates of water use (Kolb and Sperry 1999a). Also, plant storage capacity maintains water status for periods of extended drought (Tyree and Ewers 1991). All three structural adaptations result in delaying the use of soil water. However, slower use of water in relatively wet soil can result in greater water availability for other species, especially rapidly growing and shallow-rooted annuals or short-life perennials. Variable root activity can also conserve water. In this case, fine root growth into deeper, wetter soil layers is delayed until the plant experiences water limitations in the upper soil (Fernandez and Caldwell 1975). Such a strategy, however, requires that deeper soil layers become effectively recharged for the deeper roots to exploit. This may not occur when deep soil-water recharge is solely dependent on vertical infiltration of precipitation.

In systems where deep soil-water recharge is limited by infiltration, hydraulic redistribution by plant roots is an effective mechanism for moving water to deeper soil layers (Smith et al. 1999; Burgess et al. 2001; Ryel et al. 2003). The implications of this relatively rapid downward movement of water by roots have not been fully explored. Possible benefits include increasing soil water-holding capacity by increasing the infiltration capacity of surface soils (thereby reducing runoff losses), maintaining active deep roots, and reducing competition from plants with shallower root distributions (Smith et al. 1999; Ryel et al. 2003). Alternatively, deep soil redistribution of precipitation via hydraulic redistribution may simply be an artifact of root system structure and physiology and not of particular advantage for plants (Caldwell et al. 1998).

In this paper, we propose that an additional benefit of the redistribution of precipitation by roots to deeper soil layers may be the conservation of water during periods of limited water availability. This would occur in systems without rapid infiltration through the soil or subsurface water. Water uptake by plants in dry soil is limited primarily by the flow of water from the soil to the roots (Passioura 1988; Sperry and Hacke 2002). However, this relationship between soil water content (or potential) and water uptake for transpiration is very nonlinear; water uptake declines very steeply as soil water content decreases (Sperry et al. 1998; Ryel et al. 2002). Because of this relationship, the potential for plant water uptake is less when water is more uniformly distributed in the soil column than when it is concentrated in shallow soil layers. Thus, if precipitation is distributed through greater soil depths by hydraulic redistribution, the reduced average soil water content may slow the water use by the plant without changes in root activity. Here we make a case for such water conservation in a stand of Artemisia tridentata Nutt. (big sagebrush) where soil water recharge comes entirely from direct precipitation and infiltration into the soil is limited by soil characteristics. A combination of field measurements and simulation modeling was used to explore this potential mechanism for water conservation in plants following summer rain events (small summer resource pulses) and overwinter soil-water recharge (large spring resource pulse).

Materials and methods

Study site

This study was conducted in a near monoculture of Artemisia tridentata (big sagebrush) in Rush Valley, Utah, USA (112°28′W, 40°17′N, elevation 1,600 m) located in the eastern Great Basin portion of the western USA. This temperate steppe ecosystem is characterized by hot summers and cold winters. Although, precipitation events occur relatively uniformly throughout the year at this site (250–300 mm annually), soil water is recharged primarily during the period of dormancy from late autumn through early spring (Dobrowolski et al. 1990), resulting in a primary growing season in spring to early summer when temperatures are favorable for plant growth and soil moisture is abundant (Caldwell 1985). Summer precipitation occurs in sporadic rain events, but high evapotranspiration reduces the storage potential of this water in the soil. This precipitation pattern results in a large pulse of water availability in the spring and much smaller, but sometimes numerous, periods of summer pulses of water availability. Measurable groundwater at the site was at depths >10 m, far below the deepest roots. Soils are a fine, loamy, mixed, Xerollic Calciorthid of the Taylorsflat series with low shrink swell potential and minimal rock fragment, and were formed from lacustrine sediments from ancient Lake Bonneville (see Ryel et al. 2002).

Soil moisture

Soil water potential was measured hourly at nine depths (0.3, 0.45, 0.6, 0.9, 1.2, 1.5, 1.8, 2.4, and 3.0 m) using individually calibrated screen-cage thermocouple psychrometers (psychrometers: J.R.D. Merrill Specialty Equipment, Logan, Utah, USA and Wescor, Logan, Utah, USA; data logger: Campbell Scientific, Logan, Utah, USA) installed in vertical stacks during November 1998. One to three psychrometers were installed at each depth in each of three trenches surrounding a monoculture patch of A. tridentata. Soil water potential (ψ i) was converted to volumetric water content (θ i) using a relationship from Van Genuchten (1980):
$$ \theta _{{\text{i}}} = \frac{{\theta _{{\text{s}}} - \theta _{{\text{r}}} }} {{[1 + |\alpha \Psi _{{\text{i}}} |^{{\text{n}}} ]^{{\text{m}}} }} $$
(1)
where θs and θr are saturated and residual soil water contents, respectively, α , and n are parameters based on soil characteristics, and m=1–1/n (Table 1).
Table 1

Parameters for soil-water model used in simulations. Descriptions of these parameters are contained in Ryel et al. (2002). Parameters 1–4 were measured independently from this study, 5 was from Campbell (1985), 6–9 were from data presented in Fig. 2a, 10–13 were set for appropriate simulations, and 14 was measured as indicated by water uptake from the soil (see also Ryel 2002, 2003)

Parameter

Value

Units

1. Volumetric soil water at solution (θ s)

0.5

m3m−3

2. Residual volumetric soil water content (θ r)

0.02

m3m−3

3. Fitting parameter for soil water retention curve (α)

2.828

mm−1

4. Fitting parameter for soil water retention curve (n)

1.40

-

5. Soil water conductivity (Ks)

2.47

mm h−1

6. Root conductivity for water for all roots (C RT)

0.97

mm MPa−1h−1

7. Soil ψ where root conductivity reduced by 50% (ψ50)

−0.630

MPa

8. Shaping parameter (b)

1.48

-

9. Maximum transpiration rate (E RT, max)

9.6

mm day−1

10. Multiplier for day (1) or night (0) (D tran)

0 or 1

-

11. Number of soil layers

40

-

12. Layer thickness

0.10

m

13. Maximum soil depth

4

m

14. Maximum root depth

1.6

m

Plant predawn water potential

Predawn leaf water potential was measured using a Scholander pressure chamber (Scholander et al. 1965; PMS Instruments, Corvallis, Ore., USA). Measurements began one hour before and were completed prior to sunrise. One excised stem per individual plant was measured during each sampling period (n = 5 plants). Predawn water potential was then related to the average soil water content within the active rooting zone and the average soil water content weighted by the distribution of roots. These relationships were used to estimate plant predawn water potential from the soil water potential in the rooting zone.

Plant water use

Use of soil water by the stand of A. tridentata was determined by reductions in soil water potential for each of the nine soil depths containing psychrometers. Water use by plants was calculated by converting average daily water potential to volumetric water content (Eq. 1) and determining the difference in water content between consecutive days. Declines in water content were then compared to soil water potential to determine the relative soil-root conductance for water (see Ryel et al. 2002).

Model of soil water dynamics

The one-dimensional model of soil-water transport developed by Ryel et al. (2002) was used to simulate water dynamics in the soil. The model simulates soil water lost to transpiration, infiltration of soil water between soil layers, and movement of soil water among soil layers through redistribution by roots. Changes in soil water content for soil layer i are modeled as:
$$ \frac{{dW_{{\text{i}}} }} {{dt}} = \frac{{dF_{{\text{i}}} }} {{dz}} + H_{{\text{i}}} - E_{{\text{i}}} $$
(2)
where Wi is the water stored in layer i, Fi is net unsaturated flow of water into layer i, Hi is the net water redistributed by roots into layer i, Ei is the water loss through transpiration from layer i, t is time, and z is vertical thickness.
Unsaturated flow is based on Buckingham-Darcy’s law that accounts for water movement through a relatively homogeneous medium based on differences in soil water potential. Water moved by hydraulic redistribution is from wet to dry layers, and is limited by the layer with the minimum density of active roots and the water potential of the supplying (wetter) layer. Hydraulic redistribution was “turned off” by bypassing this portion of the model during simulations. The transpiration rate from a soil layer during summer months was assumed to be primarily affected by soil-root conductance as related to soil water potential (see Ryel et al. 2002) and by the relative portion of roots in that layer. The model adds precipitation events to the first soil layer. As in Ryel et al. (2002), surface evaporation was not included and was considered to have little importance in the comparative simulations conducted since evaporated water would come from the surface and very shallow soil layers where roots are minimal. Simulations are conducted on an hourly basis. Model parameters (Table 1) and the root distribution used in the simulations were from Ryel et al. (2002). Root activity to 1.6 m depth was determined by both uptake of water and from rapid recharge to depth of water following rain events (the latter could only occur if roots were active). The maximum soil depth (1.5 m) where these water dynamics were observed (Fig. 1) was considered to be the maximum depth of active roots with a buffer of 10 cm added.
Fig. 1a, b

Soil water potential (ψ) for the study stand of A. tridentata. a Soil ψ during summer 1999 at nine depths (m). Rain events during the same period are shown. b Average soil ψ for the summer periods of 1999–2002. Averages were calculated from soil depths occupied by active roots (0.3, 0.45, 0.6, 0.9, 1.2 and 1.5 m). Vertical bars represent summer rain events of >5 mm

Simulations

Simulations were conducted to compare with measured data and to assess water use following summer rain events and following overwinter soil water recharge. Soil water content distributions from measured data were compared with simulated results for the summer period of 2000 as a validation of the model. Simulations were initiated with water potential in the soil column as measured at the start of the simulation period.

A second set of simulations was conducted to assess the importance of hydraulic redistribution following summer rain events. We compared plant water use and estimated plant predawn water potential from our model with and without hydraulic redistribution following two actual summer rain events during 2000. The simulations were initiated with measured soil water potential at the start of the simulation period.

A third set of simulations was conducted to assess the importance of hydraulic redistribution of winter recharge soil water on plant water potential during the summer. Plant water use and estimated plant predawn water potential were compared in model simulations with and without hydraulic redistribution for a four-month summer period (1 May–31 August) without rain; 15 September 2000 to 30 April 2001 was used as the period for overwinter soil-water recharge. The simulation with hydraulic redistribution was initiated with measured soil water potential at the start of the summer simulation period (1 May). For the simulation without hydraulic redistribution, the initial distribution of soil water was estimated from simulations with the soil-water model over the winter recharge period assuming only infiltration affected soil water movement. Transpiration was assumed to be negligible during the simulated recharge period. Total water content in the soil layers were slightly adjusted to make initial total soil water content in the rooting zone and to 3 m depth identical at the start of both summer simulations (with and without hydraulic redistribution).

A final set of simulations was conducted to assess the differences in deep versus shallow root distributions on cumulative transpiration and plant water potential as affected by hydraulic redistribution. Simulations with and without hydraulic redistribution assuming roots were distributed 0–0.8 m (shallow) were compared to the previous set of simulations which had roots distributed from 0 to 1.6 m (deep). The shallow root distribution was created by using the upper 0.8 m of the original 0–1.6 m root distribution (deep) and was made to have the same total quantity of roots by adding a constant amount of roots to each layer. The initial distribution of soil water for each of the four situations was simulated over the winter using the precipitation from 15 September 2000 to 30 April 2001. Simulations were then conducted for the same 4-month summer period without rain. The deepest roots of the shallow root distribution were below the deepest infiltration of precipitation during the overwinter recharge period.

Results

Soil water potential

Soil water potential in the A. tridentata community during the summer declined relatively consistently except following rain events when soil water was recharged to all layers containing active roots. Water potentials were quite similar in all soil layers from 0.3 to 1.2 m (Fig. 1a) with similar recharge in each layer following rain events. The water recharge following rain events from 0.3 to 1.5 m depth resulted primarily from hydraulic redistribution by roots (see Ryel et al. 2002) and resulted in the relatively uniform soil water recharge between 0.3 and 1.2 m depth. Roots were active to at least 1.5 m, but no change in water potential at depths 1.8 m and greater indicated no activity at those depths and below (Fig. 1a). The rate of decline in average soil water potential within the active rooting zone (surface to 1.5 m depth) for this community was quite similar among years (Fig. 1b). Differences in the soil water dynamics among years was related primarily to rain events. Also, a cool spring in 1999 delayed the soil drying from early May to early June.

The daily change in soil water content declined rapidly with soil water potential (Fig. 2a). Loss of soil water was attributed to transpiration through root uptake of water since infiltration was much too slow (weeks to months vs days) to result in the observed changes in soil water content (see also Ryel et al. 2002). The loss of soil water was not linearly related to soil water potential (Fig. 2a) or volumetric soil water content (Fig. 2b); the water uptake rate declined rapidly and nonlinearly with decreasing soil water potential or content. For example, the potential rate of water extraction by plants from soil with volumetric water content of 0.2 was five times that with soil volumetric water content at 0.1 (Fig. 2b). The function in Fig. 2a was relativised to a maximum of 1.0 and used to calculate relative root conductance as a function of water potential in the model of Ryel et al. (2002).
Fig. 2a

Changes in soil water content due primarily to plant uptake as related to soil water potential (ψ) during periods of soil drying, 1999–2002. Non-linear regression was used to fit the function shown. b The relationship between changes in soil water content and volumetric soil water content for relationship developed above. Vertical dotted lines represent equivalent soil water potential using relationship in Eq. 1 and parameters in Table 1. Horizontal dashed lines indicate plant water uptake at soil water contents 0.2 and 0.1 m3 m−3. Changes in soil water content are expressed as millimeter water per meter soil depth

Soil and predawn plant water potential

The pattern of plant predawn leaf water potential and the measured average soil water potential were quite similar except following a moderate (1.2 mm) rain event during the summer of 2000 (Fig. 3). Both average soil water potential within the rooting zone (Fig. 3a) and a weighted average based on the distribution of roots (Fig. 3b) closely followed the pattern of predawn leaf water potential. Including additional measurements from 1999 and 2001, a highly significant relationship (R 2=0.93; P <0.00001) was found between soil water potential (both unweighted and weighted by roots) and plant predawn water potential (Fig. 4), indicating that plant predawn water potential was highly reflective of soil moisture conditions within the rooting zone.
Fig. 3a

Average soil water potential (ψ) measured in the active rooting zone (to 1.6 m) during summer 2000, and measured predawn plant water potential (mean ± SE, n =5) for A. tridentata. Average measured values were calculated from psychrometers at 0.3, 0.45, 0.6, 0.9, 1.2 and 1.5 m depth. Soil ψ calculated by model simulations were averaged for the same range of soil depths (0.3–1.5 m) and for the entire depth of active roots (0–1.6 m). b Average soil ψ measured in the active rooting zone (to 1.6 m) weighted by the measured root distribution, and measured predawn plant ψ for A. tridentata. Weighted average measured values were calculated from psychrometers at 0.3, 0.45, 0.6, 0.9, 1.2 and 1.5 m depth. Weighted average soil ψ calculated by model simulations were calculated for the same range of soil depths (0.3–1.5 m) and for the entire depth of active roots (0–1.6 m)

Fig. 4

Relationship between measured predawn plant water potential (ψ) and average measured soil ψ in the rooting zone for measurements conducted during summers of 1999–2001 for the stand of A. tridentata. Averages were calculated from psychrometers at 0.3, 0.45, 0.6, 0.9, 1.2 and 1.5 m depth with the weighted average weighted by the root distribution

Comparing measured and simulated soil water potential

Simulations using the model of Ryel et al. (2002) were conducted to compare with measured soil water potential in the rooting zone and with predawn leaf water potential and included 19 rain events (ranging from 0.3 to 16.0 mm). Good correspondence was found between the model simulations and the measured soil water potential for both unweighted (Fig. 3a) and weighted, by root distribution (Fig. 3b), average soil water potential for depths measured by the psychrometers within the active rooting zone (0.3–1.5 m). Simulated soil water potentials averaged over the entire active rooting zone (surface to 1.6 m) and were quite similar to the average of psychrometer measurements (0.3–1.5 m) for unweighted averages. However, for averages weighted by the root distribution the correspondence was not as good. Simulated weighted average water potential over 0–1.6 m was less than the weighted average of measurements over 0.3–1.5 m during periods without rain events, and greater during periods following rain events, reflecting the water status of the root-laden shallow soil layers (0–0.3 m) which were not measured by psychrometers.

Simulated average soil water potentials also corresponded well with measured predawn plant water potentials (Fig. 3), indicating that the simulated average soil water potential in the rooting zone was a good estimate of predawn plant water potential. The relationship between predawn plant water potential and unweighted average soil water potential (surface to 1.6 m depth) was best during periods with little or no rain, but this fit was less following the sizeable 16-mm rain event in early July. For the average soil water potential weighted by the root distribution (0–1.6 m), the relationship with predawn plant water potential was not as good during the periods with little or no rain, but was able to capture the increase in predawn plant water potential following a rain event. This likely occurred due to the greater weight placed on the shallow soil layers wetted by the rain event to which the plants apparently responded. The similar patterns for the weighted and unweighted average soil water potentials (except following larger rain events) was primarily due to the roots tending to homogenize soil moisture within the rooting zone through hydraulic redistribution (Fig. 1a).

Simulated water potential following summer rain events

Simulations were conducted to assess the effect of downward hydraulic redistribution on plant water use following two summer rain events (two summer pulses). Estimated predawn plant water potential was higher following the two rain events when hydraulic redistribution was included in the simulations. This was the case whether predawn plant water potential was estimated either as unweighted (Fig. 5a) or weighted (Fig. 5b) average soil water potential in the rooting zone. The differences in estimated plant predawn water potential were due to more soil water resulting from less water lost to transpiration after the rain events (Fig. 5d) and to the more uniform distribution of soil water (the higher average water potential resulting from the nonlinear relationship between water content and water potential—see Eq. 1). However, by the end of August cumulative transpiration was nearly the same with and without hydraulic redistribution, but plant water potential was still higher with hydraulic redistribution due to the more uniform distribution of water in the soil. While transpiration rates after the rain events (Fig. 5c) were initially higher without hydraulic redistribution, this situation changed 11 days after the second rain event when transpiration became higher with hydraulic redistribution, and within a month (early August), transpiration had increased to 30% higher in the simulation with hydraulic redistribution.
Fig. 5a–d

Estimated predawn plant water potential (ψ) and transpiration rates for the stand of A. tridentata before and after two rain events during summer 2000. The two rainfall events are shown as vertical bars (total rainfall is shown in mm). a Predawn plant ψ for the stand with and without hydraulic redistribution was estimated as the average of simulated soil ψ over the entire rooting zone (0–1.6 m). Measured predawn plant ψ was estimated as the average of soil psychrometer measurements at 0.3, 0.45, 0.6, 0.9, 1.2 and 1.5 m. b Predawn plant ψ for the stand of A. tridentata with and without hydraulic redistribution estimated as the average of simulated soil ψ over the entire rooting zone (0–1.6 m) weighted by the distribution of roots. Measured predawn plant ψ was estimated as the average of soil psychrometer measurements at 0.3, 0.45, 0.6, 0.9, 1.2 and 1.5 m weighted by the distribution of roots. c Simulated daily transpiration rate for the stand during the same period with and without hydraulic redistribution. d Simulated cumulative transpiration for the stand during the same period with and without hydraulic redistribution

Simulated water potential during summer following winter recharge

Simulations were conducted to assess the effects of hydraulic redistribution on over-winter soil water recharge (large spring pulse) and predawn plant water potential during the subsequent summer. Simulation results (Fig. 6a, b, deep root distribution) were similar to the situation following summer rainfall events (Fig. 5a, b) for both the unweighted and weighted average soil water potential as estimated predawn plant water potential was higher when hydraulic redistribution was included throughout the entire summer period. As with the situation following summer rain events, the higher estimated predawn plant water potential resulted from lower transpiration rates early in the summer (Fig. 6c) and because water recharged over winter was more evenly distributed into the soil column. Cumulative transpiration was less with hydraulic redistribution until the end of August when the both simulations indicated similar total water usage. Transpiration rates were higher for the first three weeks without hydraulic redistribution, but throughout the rest of the summer, transpiration was higher with hydraulic redistribution, becoming 27% higher from mid-July through August. The initial predawn plant water potential estimated from the average soil water potential was less for the situation without hydraulic redistribution (Fig. 6a) as soil water was concentrated in shallow soil layers while the rest of the soil was quite dry. The weighted average did not show this pattern (Fig. 6b) due to the high root volume in the wet shallow soil layers.
Fig. 6a–d

Estimated summer predawn plant water potential (ψ) and transpiration rates for the stand of A. tridentata following overwinter recharge (15 September 2000–30 April 2001) for two root distributions (see text) of depth 0–1.6 (deep) or 0–0.8 m (shallow). a Predawn plant ψ for the stand estimated from simulated soil ψ with and without hydraulic redistribution averaged over the entire rooting zone. b Predawn plant ψ for the stand estimated from simulated soil ψ with and without hydraulic redistribution over the entire rooting zone weighted by the distribution of roots. c Simulated daily transpiration rate for the stand during the same period with and without hydraulic redistribution for the two root distributions (note logarithmic scale). Curves for both “shallow” simulations are nearly identical except during initial days in May and thus are mostly superimposed. d Simulated cumulative transpiration for the stand during the same period with and without hydraulic redistribution for two root distributions

Effect of root distribution

The final set of simulations assessed the importance of deep (0–1.6 m) versus shallow (0–0.8 m) root distributions on summer cumulative transpiration and plant water potential following winter recharge as affected by hydraulic redistribution. Estimated plant water potential throughout the summer was highest with the deep root distribution and hydraulic redistribution (Fig. 6a, b). This resulted from lower initial transpiration rates (Fig. 6c), more uniform distribution of water in the soil and effective capture of infiltrating recharge water during the summer. Both simulations with the shallow root distributions eventually had lower cumulative transpiration than the simulations with the deep root distribution (Fig. 6d) due to water which infiltrated below the rooting zone during summer and became unavailable for plant uptake. Estimated plant water potential with the shallow root distribution was higher with hydraulic redistribution than without, throughout all but the first two weeks of the summer, due to more uniform distribution of soil moisture. Depending on whether it was weighted according to root distribution or not, the estimated predawn plant water potential with the shallow root distribution with hydraulic redistribution was either higher (Fig. 6b) or lower (Fig. 6a) than the estimate with the deep root distribution without hydraulic redistribution.

Discussion

This research indicates that the downward redistribution of precipitation by roots to depth in the soil column can contribute to water conservation following recharge from pulsed precipitation events. By spreading out recharge water deeper and more uniformly into the soil column, the rates of potential water use are initially reduced by lower hydraulic conductance of the soil. Subsequently, plants can have higher transpiration rates due to less initial water consumption and a more favorable distribution of water in the soil (more uniform) for transpiration. The phenomenon occurs when soil water recharge comes primarily from surface recharge, and infiltration is limited by soil characteristics. Water conservation associated with hydraulic redistribution is a physical process that does not rely on changes in the quantity of roots and leaves or use of hormonal signals (e.g., ABA) to reduce the rate of water use, but likely occurs in conjunction with these mechanisms. A. tridentata is a species which limits water loss through reductions in stomatal aperture and production of some leaves which are ephemeral (Evans and Black 1993), and has variable cavitation resistance based on the growing environment (Kolb and Sperry 1999b).

This work also indicates that hydraulic redistribution of water following pulses, both over the winter and following summer rain events, can also help maintain higher predawn plant water potential during subsequent drought periods. Higher predawn plant water potential results from higher total soil water content due to slower initial water use and the more uniform distribution of soil moisture. Overwinter redistribution may delay the time during the subsequent summer when plant water potential declines to levels where plant physiological function is affected (e.g., cavitation in xylem vessels, loss of leaves, loss of hydrologic connectivity with soil). Redistributing water from summer rain events may help to keep the existing plant physiological function maintained for an extended period at a time where more rapid water use would have little benefit to the plant.

The mechanism is also consistent with the hypothesis of Jones (1980) that plants in unpredictable environments should make maximal use of available water (operating close to hydraulic failure) but also have a mechanism to prevent death when water is very limiting. Since A. tridentata maintains physiologically active foliage throughout the summer (Evans and Black 1993), maintaining higher plant water potentials and some water for transpiration may be critical for survival during drought periods. For species which senesce during the dry portions of the summer (e.g., annuals, tussock grasses), rapid water use in the spring may be more important and redistribution and conservation of water for later use is perhaps not so beneficial. These species, therefore, may be expected to redistribute less precipitation during the overwinter recharge period and following summer rain events. In support of this contention, lower redistribution rates were found for the perennial tussock grass Agropyron desertorum and the annual grass Bromus tectorum in sites adjacent to our A. tridentata stand with similar soils (R.J. Ryel, unpublished data).

By exploiting the non-linear relationship between soil water content and hydraulic conductivity, A. tridentata can reduce the rate at which water is used. However, this method of water conservation might be viewed as providing additional water to competitors or slowing the growth rate of A. tridentata (Cohen 1970), thereby reducing the competitiveness of A. tridentata and not being an adaptive strategy. In the situation described here, however, lowered competitiveness is unlikely. First, all plants in the system would be limited in the rate of water uptake due to the lower hydraulic conductance. To more rapidly extract this water and gain a competitive advantage, other plants would have to increase their root density in relatively deep soil layers and maintain sufficient transpirational surface. This would require a reallocation of resources to roots in deeper soil layers which would result in less competitive ability with A. tridentata in shallow soil layers which contain most of the nutrients. In addition, potential reduced growth by A. tridentata during the spring related to less water availability may be offset by the ability to maintain physiological function throughout the summer. This may allow A. tridentata to be more responsive to resource pulses, particularly nutrients, when they become available during periods when other plants have senesced or are just beginning to become physiologically active with limited active biomass.

At least four benefits to species like A. tridentata may derive from this process of water storage. A significant benefit of the slowed initial rate of water use is that plant water potential can remain higher throughout the summer period where temperatures and evapotranspiration rates are high and water use is limited by soil hydraulic conductance. Since plant water potential is related to the potential for cavitation within the xylem (Tyree and Sperry 1989), higher plant water potential reduces the risk that hydraulic failure will occur. This is important as plants trying to maximize leaf area and stomatal conductance while minimizing root biomass often operate close to their hydraulic limits (Sperry et al. 1998). Such hydraulic failure can result in the subsequent death of branch units or whole plants and can limit the distribution of plants (Tyree and Ewers 1991; Portwood et al. 1997; Davis et al. 2002; Sperry and Hacke 2002). Although A. tridentata has a relatively high capacity to tolerate drought and low water potentials (Kolb and Sperry 1999a, 1999b), redistribution of precipitation to depth and the subsequent conservation of water and maintenance of higher plant water potential may permit this species to persist in more arid areas or on soils with limited permeability and unavailable groundwater. In a similar situation of hydraulic redistribution of rainwater to depth, Burgess et al. (2001) suggested that redistributed water had little effect on drought avoidance for Eucalyptus camaldulensis. However, slower use of this water redistributed to depth may have similarly helped this species keep plant water potentials higher during the dry summer.

A second benefit to A. tridentata is reduced potential for loss of hydraulic connectivity between the roots and soil that can result from the more uniform distribution of water in the soil. This benefit would be enhanced by greater quantities of soil water throughout much of the period following summer rain events or overwinter recharge. Since hydraulic contact with the soil is directly related to soil water potential (Sperry et al. 2002), more of the roots may maintain hydraulic connectivity with access to soil with higher water potential than if water was not hydraulically redistributed.

Another benefit is related to physiological activity of the leaves. In response to limited water availability, A. tridentata is effective in reducing stomatal aperture (Evans and Black 1993) and when water becomes increasingly limited, sheds some leaves to reduce transpirational surface (Black and Mack 1986). However, by reducing water use early in the growing season, this species can have higher transpiration rates later in the summer and perhaps maintain higher foliage area and photosynthesis rates throughout the summer. Of course by using more water in spring or early summer, greater photosynthesis rates during optimum conditions may occur and more tissue can perhaps be produced. However, the downside to this production is that the plant must then sustain this additional tissue or risk losing foliage and stems through severe cavitation when soil water potential becomes low.

A final benefit of water conservation through hydraulic redistribution of water to depth is to store water while at the same time reduce the availability of water to other species (Smith et al. 1999). Co-occurring species of A. tridentata are generally grasses and forbs (West and Young 2000) which are largely drought avoiders. These species have little or no physiological activity when water potentials in shallow soils become dry. By storing water at depth and at relatively low water potentials, these plants which may compete for water have limited opportunity to access this redistributed water. Some of the water would be stored below the root systems of the shallow rooted plants and the low water potential would limit the ability of these drought-avoiding plants to take up this water. This is in contrast to the increased competition for water hydraulically lifted by Acacia tortilis trees that occurs between these trees and grasses in African savannahs (Ludwig et al. 2003).

Water conservation through redistribution of water to depth was more effective with the deeper root distribution in the simulations. Deeper roots redistributed water throughout a greater soil volume, and the resulting lower soil water potential slowed water use to a greater extent in the early season. The deeper roots had the additional advantage of accessing water that had infiltrated beyond the reach of the shallower root distribution. Both the deeper and shallower root distributions were more effective in maintaining higher plant water potentials with hydraulic redistribution indicating that even for shallow-rooted plants, redistribution may be of some benefit in conserving water.

Our findings appear to be in direct contrast to the markedly higher transpiration rates due to upward hydraulic redistribution (hydraulic lift) found by Caldwell and Richards (1989), Emerman and Dawson (1996) and Ryel et al. (2002). In these studies, increased transpiration commenced as soon as the upper soil layers began to dry. The difference between their findings and those of this study is that the higher transpiration rates in their study were driven by deeper soil layers that were wet due to a supply of subsurface moisture. However, in situations such as described here there is no subsurface water and rainfall and soil infiltration are limited. Thus, recharge to deep soil layers is minimal and soils will remain at low water potentials unless recharged by hydraulic redistribution (Ryel et al. 2003). Even with this redistribution, the deep soil water occurs at water potentials as low or lower than in the shallow layers (see Fig. 1, upper). Thus, there is little potential for this deep water to be redistributed to the shallow layers to augment transpiration until later in summer. In both situations, however, hydraulic redistribution of water by roots appears to be important in keeping plant water potential higher and the plant at a higher level of physiological function during the summer.

Predawn plant water potential of A. tridentata was found in this study to correspond closely to average soil water potential within the zone of active roots. This was the case for average soil water potential calculated either weighted or unweighted by the root distribution. The weighted average of soil water potential fit best following a rain event indicating roots rapidly accessed this shallow soil water. The strong relationship between predawn plant and average soil water potentials suggested that water status of A. tridentata was in direct response to the soil that contained active roots. Similar correspondence has been found for other species (Slatyer 1967; Schulze 1991; Leffler et al. 2002). This is in contrast to the hypothesis that plant predawn water potential is in response to the water potential of the wettest layer accessed by roots (see Ritchie and Hinckley 1975; Hinckley et al. 1978; Richter 1997), and the disequilibrium between predawn plant and soil water potentials that can apparently occur under well watered conditions (Donovan et al. 2001), or has been observed for some plants with nighttime transpiration or putative leaf apoplastic solutes (Donovan et al. 2003).

In conclusion, this work illustrates an additional effect of hydraulic redistribution on plant-water relations. Water can be conserved by redistributing water concentrated in shallow soil layers to deeper soil layers at lower water potentials. These water movements by roots may aid in maintaining function of foliage and increase survival probability of plants which remain physiologically active during periods of drought.

Acknowledgements

This work was funded by the National Science Foundation (DEB-9807097) and the Utah Agricultural Experiment Station. We thank Ann Mull for her excellent field assistance, Charles Ashurst for his electronics expertise, and Darrell Johnson for allowing us to establish our study site on his property in Rush Valley, Utah.

Copyright information

© Springer-Verlag 2003