Journal of Experimental Botany, Vol. 57, No. 12, pp.
3283–3291, 2006
doi:10.1093/jxb/erl085 Advance Access publication 22 August, 2006
RESEARCH PAPER
Redistribution of soil water by lateral roots mediated by
stem tissues
S. S. O. Burgess1,2,* and T. M. Bleby1,†
1
School of Plant Biology, University of Western Australia, 35 Stirling Highway, Crawley, WA 6009 Australia
2
Cooperative Research Centre for Plant-Based Management of Dryland Salinity, 35 Stirling Highway, Crawley,
WA 6009, Australia
Received 23 February 2006; Accepted 20 June 2006
Abstract
Evidence is increasing to suggest that a major activity
of roots is to redistribute soil water. Roots in hydraulic
contact with soil generally either absorb or lose water,
depending on the direction of the gradient in water
potential between root and soil. This leads to phenom-
ena such as ‘hydraulic lift’ where dry upper soil layers
drive water transfer from deep moist layers to the
shallow rhizosphere and, after rain or surface irriga-
tion, an opposite, downward water transfer. These trans-
port processes appear important in environments where
rainfall is strongly seasonal (e.g. Mediterranean-type
climates). Irrigation can also induce horizontal trans-
fers of water between lateral roots. Compared with
transpiration, the magnitudes, pathways, and resist-
ances of these redistribution processes are poorly
understood. Field evidence from semi-arid eucalyptus
woodlands is presented to show: (i) water is rapidly
exchanged among lateral roots following rain events,
at rates much faster than previously described for
other types of hydraulic redistribution using sap flow
methods; (ii) large axial flows moving vertically up or
down the stem are associated with the horizontal
transfer of water between roots on opposite sides of
the stem. It appears that considerable portions of the
stem axis become involved in the redistribution of
water between lateral roots because of partial sector-
ing of the xylem around the circumference of these
trees.
Key words: Hydraulic redistribution, root systems, sap flow,
soil water, xylem.
* To whom correspondence should be addressed. E-mail: ssb@cyllene.uwa.edu.au
y
Present address: Department of Biology, Duke University, Durham, NC 27708, USA.
ª The Author [2006]. Published by Oxford University Press [on behalf of the Society for Experimental Biology]. All rights reserved.
For Permissions, please e-mail: journals.permissions@oxfordjournals.org
Downloaded from http://jxb.oxfordjournals.org by on May 13, 2010
Introduction
The idea that plant root systems can act as ‘equalisers’ of
soil moisture dates back more than 70 years (Breazeale,
1930; Caldwell et al., 1998). Early evidence that roots
redistribute water within the rhizosphere came from green-
house and laboratory measurements of water transfer
across root systems that spanned physically separated
soil compartments, or so-called ‘split-root’ experiments
(Baker and Van Bavel, 1986; Sakuratani et al., 1999; and
see Caldwell et al., 1998, for a review).
The first observations of roots acting as conduits for
water transfer within the rhizosphere of natural ecosystems
were made by Richards and Caldwell in 1987 (Richards and
Caldwell, 1987). These authors used thermocouple psy-
chrometers to measure increases in the water potential of
shallow soil layers during the night (i.e. in the absence of
transpiration to drive water absorption and soil moisture
depletion). They concluded that water was being absorbed
by deep roots with access to moist soil and then lost from
shallow roots into dry surface soil. This phenomenon,
which they termed ‘hydraulic lift’, substantially improved
our understanding of the ecological importance of water
transfer by roots. Since then, a succession of field studies,
using an ever-broadening range of techniques, has added
considerable detail to our understanding of hydraulic re-
distribution, including the role of species (Espeleta et al.,
2004), soil type (Yoder and Nowak, 1999), root morphol-
ogy (Hultine et al., 2003b), and weather (Williams et al.,
1993), to name just a few examples.
Stable isotope tracing (Dawson, 1993) and sap flow
measurements in roots (Burgess et al., 1998) have also
brought important advances to our understanding of the
dynamics of water transfer by root systems. Burgess et al.
3284 Burgess and Bleby
(1998) used sap flow measurements in roots and Schulze
et al. (1998) used stable isotope labelling techniques to
demonstrate that hydraulic lift reverses direction if surface
soil layers become wetter than deep soil layers following
rain or irrigation (Smith et al., 1999). To reflect the multi-
directional nature of water transfer by root systems, the
term hydraulic redistribution was coined to include both hy-
draulic lift and downward transfer of water following rain
(Burgess et al., 1998).
Underscoring the multidirectional nature of water trans-
fer by roots, Brooks et al. (2002) provided evidence of both
vertical and horizontal transfer of water by roots of two
conifer trees. Deuterium-labelled water was detected hor-
izontally from a partial irrigation of the root system and
provided a field-based example to indicate horizontal
redistribution. Similarly, a recent study by Smart et al.
(2004) demonstrated both downward and horizontal (or
‘transverse’) redistribution following partial irrigation of
grapevines with deuterated water. The results of this study
and that of Brooks et al. (2002) suggest that horizontal
redistribution of water may be more limited than vertical
redistribution owing to lateral roots on opposing sides of
stems being connected by pathways of relatively high
resistance at the stem base.
Little is know about the horizontal redistribution of water
under field conditions and, more generally, how water
moves within plants in the presence of multiple sources and
sinks joined by pathways of varying resistances. For
example, it is not clear whether roots growing in dry soil
are equal competitors for water with above-ground organs,
given that the axial orientation of xylem vessels favours
water movement from root to shoot, rather than from root
to root. Smart et al. (2004) recently provided further in-
formation on this topic by using a stable isotope tracer to
show that water moves ‘around’ or ‘across’ the stem from
roots on an irrigated side of a grape vine to roots on
a non-irrigated side. Although deuterium-labelled irriga-
tion water did not appear in soils on the non-irrigated side
of the vine until one week after application, deuterium-
labelled water injected directly into the stem of the vine
moved rapidly around/across the stem and into roots on
all sides of the stem.
Taken as a whole, the broad suite of experimental data
clearly indicates that water can be redistributed by root
systems in any direction, but always according to the
gradient in water potential. An important caveat is that the
hydraulic connectivity and conductivity between various
roots at the stem base will probably determine the rates of
transfer.
A clear understanding of the root–stem and root–root
pathways requires information on a number of aspects of
wood anatomy. For example, in angiosperms, long vessels
and perforated end-plates of comparatively low resistance
yield hydraulic conductivities that are much greater along
vessels than between vessels (where water must move via
pits). Since xylem vessels are typically aligned axially, flow
along vessels usually means flow along the axis of a plant,
whereas flow between vessels is ‘radial’ or ‘circumferen-
tial’. The relative importance of these flow vectors will vary
according to different vessel lengths and diameters (e.g. as
found in roots versus stems; Pate et al., 1995), different pit
and end-plate resistances, and a range of other anatomical
factors (Orians et al., 2005). Spiralling of xylem vessels
(Howard, 1932) will also alter flow distributions (Brooks
et al., 2002). Flow paths may be further altered by the
presence of heartwood, latewood, or rays, which may partly
or completely interrupt flow in radial or circumferential
directions. Zwieniecki et al. (2003) also shows that flows
via pits will be strongly ion-mediated.
In the present study, field measurements of xylem sap
flow were made to gain an insight into how water may be
exchanged between lateral roots connected to different sides
Downloaded from http://jxb.oxfordjournals.org by on May 13, 2010
of a tree. Given that wood is essentially a porous material,
the guiding hypothesis in this study was that water might
flow in practically any direction within wood tissues, but
the dominant directions will be dictated by the orientation
and physical properties of xylem vessels, including the
location of ‘non-conductive’ tissues such as heartwood.
The aim was to elucidate patterns of water uptake and
redistribution in lateral roots by carefully measuring flow
vectors through roots and stems.
Materials and methods
Site description
The research was conducted at the Corrigin Water Reserve approx-
imately 2 km west of Corrigin, Western Australia (328199 S, 1178529
E). This reserve consists of 1096 ha of remnant vegetation containing
a range of plant communities typical to Western Australia’s wheat-
belt region. Low-lying areas have heavy soils and are sparsely
wooded. The study site was chiefly a mixture of wandoo (Eucalyptus
wandoo Blakely) and salmon gum (Eucalyptus salmonophloia F.
Muell.) trees ranging in height from 10 m to 30 m, with a scattered
dwarf scrub understorey (e.g. Olearia, Grevillea spp.). This type of
woodland is widespread throughout the semi-arid region of south-
western Australia (Beard, 1990), but little information is available on
rooting habit (Hobbs and O’Connor, 1999). Lamont (1985) recorded
that E. wandoo saplings have root systems dominated by a taproot,
whereas mature trees are dominated by an extensive system of lateral
roots extending out approximately 10 m at 20–40 cm depth. These
findings agree with our excavations of both wandoo and salmon gum
saplings and observations of wind-thrown mature specimens.
Surface soils at the site were a light brown sandy loam, which
altered rapidly to a heavy clay–loam subsoil. The long-term (since
1910) average annual rainfall for Corrigin is 376 mm, with an annual
pan evaporation of 1817 mm (Bureau of Meteorology, Australia).
A Mediterranean-type climate prevails, with most rain falling in the
cooler winter months.
Sap flow measurements
The heat ratio method (Burgess et al., 2001a, b) was used to measure
sap flow in stems and roots of three wandoo and three salmon gum
trees of similar size (;40–60 cm diameter at breast height). In

December 2003, 16 probe sets were installed to instrument the six
trees in a fashion typical for transpiration estimates. Three probes
were inserted 1208 apart at ;1.3 m height around the circumference
of each tree. Only two probes 1808 apart were available for the third
individual of each species. In all cases, probes were sent via AM
16/32 multiplexers to CR10X dataloggers (Campbell Scientific
Inc. Logan, Utah, USA). Burgess and Dawson (2004) provide
a detailed description of a similar system.
In July 2004 a further 40 probe sets were progressively installed to
monitor specific patterns of sap flow in individual trees; 35 of the 40
probes were installed in two salmon gums. Soil was excavated around
the base of these two trees, exposing 7–8 major lateral roots fairly
evenly distributed around the circumference of each tree. Although
all roots could be considered ‘lateral’, some roots had a steeper angle
of descent than others. Due to the size of the trees and the extremely
hard soil, it was only possible to excavate to a depth of approximately
60 cm. As a result, information on the existence of a taproot or sinker
roots could not be obtained for these specimens, but the findings of
Lamont (1985) for the closely related E. wandoo and our own
observation of wind-thrown E. salmonophloia suggest that a large
taproot was unlikely. A single sap flow sensor was placed in each
Redistribution of soil water by lateral roots 3285
Soil moisture measurements
On one occasion in June 2005 following a period of considerable
rainfall, soil moisture samples were collected to a depth of 40 cm
depth using a hand auger. Three replicate samples were collected
5 m from the base of the tree depicted in Fig. 3 at all four
cardinal positions. Gravimetric soil water content was calculated
by weighing and then drying samples to a constant weight at 70 8C.
Results
The measurements of sap flow in roots showed that
immediately following rain, the whole course of flow in
certain lateral roots would rapidly elevate, whilst in other
roots, flows would rapidly decrease and even reverse. These
contrasting patterns were most obvious at night: uptake
of water continued in some roots despite the absence of
transpiration, whilst flows in other roots completely re-
versed. Figure 1A shows average night-time flow rates in

Downloaded from http://jxb.oxfordjournals.org by on May 13, 2010

lateral root and a ring of eight further probes was installed in the stem seven different lateral roots around the circumference of
of each tree (;1 m above the roots), evenly spaced at 458 intervals
around the circumference.
a single tree. After each rain event, this pattern of elevated
Overall, a total of 56 probe sets were used during this experiment, flows in some roots and reverse flows in other roots would
connected to five separate and electrically isolated dataloggers. For slowly attenuate over a number of days until ‘normal’
reference purposes, spot checks of the behaviour of the instruments (i.e. flows corresponding only to the typical course of
were also made using commercially available devices, including transpiration, i.e. no reverse or elevated flows at night)
a HRM sensor (HRM-30, ICT International, Armidale, Australia) and
compensation heat pulse sensor (SF-100, Greenspan Technology,
patterns of flow resumed. The amount of time taken for this
Warwick, Australia). The latter instrument, which can only measure
flow unidirectionally (Burgess et al., 2000a), was installed upside-
down with the aim of confirming possible flow reversals measured
with the HRM devices.
During the course of the sap flow measurements, a number of
experimental treatments were initiated to test their effects on data
collected using the HRM. Each treatment was generally made on one
probe for each of the salmon gum trees being monitored (n=3) and
included (i) completely removing all bark down to the xylem from
a few cm around the site of probe insertion, (ii) completely removing
all bark and then sealing around the probe with waterproof putty,
(iii) affixing a plastic rain shield to protect probes from rainfall and
stem flow, and (iv) completely severing the xylem around indi-
vidual probes. This last treatment was done by making a U-shaped cut
15 cm wide, 15 cm high, and 10 cm deep into the wood of the trunk
around the probe.
Aside from five probe sets subject to the xylem severing treatment
above, the majority of sap flow data were not corrected for zero
bias using the methods outlined by (Burgess et al., 2001a), which
again would involve cutting the xylem. This decision was made to
permit further collection of data from most probes beyond the course
of the present investigation. Our method of calculating a zero
offset for calibration purposes was to use night-time values at least
3 weeks after any rain event, when the analysis of flow trajectories
indicated rain-induced redistribution processes had attenuated com-
pletely. Since night-time transpiration can affect night-time flow
rates (Benyon, 1999; Snyder et al., 2003, note two spikes at approx-
imately day 11 and day 31 of Fig. 4 due to nights of very low humid-
ity), only night-time values from periods of high humidity were used.
Fig. 1. (A) Night-time rates of sap flow (averages of midnight to 04.00 h
in seven lateral roots radiating out in different directions from the base
Meteorological measurements of a single Eucalyptus salmonophloia tree ‘S1’, growing at Corrigin
Western Australia. Thirty five mm of rain fell at the end of day 294
A tipping bucket rain gauge (CS701), leaf wetness sensor (model (Spring 2004). (B) Night-time rates of sap flow measured at eight
237), and temperature/relative humidity sensor (CS500) were also positions around the stem of the same tree at 1 m height. The filled circles
attached to a CR10X datalogger (all devices Campbell Scientific that are surrounded by larger open circles are interpolated due to the
Inc.) to monitor weather conditions. intermittent failure of one sensor during these observations.
3286 Burgess and Bleby
to occur scaled approximately with the size of each rain by each probe in the first tree are summarized diagram-
event (see Fig. 4, which indicates recovery times ranging matically in Fig. 3.
from >10 d for large rain events and ;2 d for small events). A long-term record of night-time sap flow in the stem
The locations of roots with elevated positive flows and of the first salmon gum tree (S1) is shown in Fig. 4.
roots with reverse flows were more or less restricted to two Throughout the year, rainfall always produced the same sap
distinct ‘hemispheres’ around the circumference of the tree. flow dynamics as documented in Figs 1–3. Again, the
Roots in northern and eastern sectors of the tree tended to magnitude of changes to sap flow patterns in stems scaled
have positive flows in response to rain, whilst roots in with the size of the rain event. Consecutive rain events
southern and western sectors showed reverse flows (see tended to produce an additive effect. The large, isolated
Fig. 3). Between these two ‘hemispheres’ were transitional
regions where flows were not altered by rain.
Measurements of sap flow using eight probe sets spaced
around the circumference of stems showed an almost
identical pattern to that measured in roots (Fig. 1B). For
all sensors, there was broad correspondence between the
flow patterns in the various sectors of the stem and those of
the individual roots radiating out, not directly beneath, but

Downloaded from http://jxb.oxfordjournals.org by on May 13, 2010

adjacent to each stem sector. For example, flows in the N
sector of the stem corresponded well with flows measured
in a lateral root in the NE sector (compare Fig. 1A, B).
Similarly, flows in the W sector of the stem flows showed
good agreement with flows in a lateral root in the NW
sector (Fig.1A, B). Data for the other salmon gum tree, that
had a similar arrangement of instruments, followed a
broadly similar pattern and for simplicity are not discussed
in further detail (Fig. 2A, B). Probe locations, and the
approximate magnitude and direction of flows measured

Fig. 3. Schematic diagram showing positions of sap flow measurement

in a single Eucalyptus salmonophloia tree (S1). Measurement positions
are indicated by the locations of arrows, which also denote, by direction
and length, the relative direction and magnitude of night-time sap flow
as depicted on day 296 in Fig. 1. Eight probe sets were installed equi-
distantly around the main stem at 1 m height, and seven probe sets were
installed in lateral roots (excavated to 0.6 m depth) approximately 0.3 m
from the base. The tree was divided into the eight indicated sectors, based
on cardinal direction, with one stem probe and one root probe installed
per sector (except no root measurement for the SE sector).

Fig. 2. (A) Night-time rates of sap flow (averages of midnight to

04.00 h) in seven lateral roots radiating out in different directions from
the base of a single Eucalyptus salmonophloia tree ‘S2’, growing
at Corrigin Western Australia. Thirty-four mm of rain fell over day
362/day 363 (Summer, 2004). (B) Night-time rates of sap flow meas-
ured at eight positions around the stem of the same tree at 1 m height.
Fig. 4. Long-term record of night-time rates of sap flow in response to
rain in the stem of a single Eucalyptus salmonophloia tree (S1) growing
at Corrigin, Western Australia. Flows were measured on the south side of
the tree; the pattern on the north side of the tree mirrored that shown here.

summer rainfall event seen in the early part of the record
in Fig. 4 gives an indication of the maximum duration
an individual rain event altered sap flow patterns in stems
(10–12 d).
Experiments to exclude the possibility that flow changes
were due to factors other than xylem sap flow showed that
neither complete removal of all bark tissues, nor water-
proofing around probes with putty, nor adding plastic rain
shields altered the phenomenon described above (data not
shown). The final treatment completely severed the active
xylem surrounding four probe sets. Cutting the xylem
immediately stopped all flow dynamics, including the
changes caused by rain (Fig. 5; days 177 onwards). Spot
measurements with both the HRM-30 and SF-100 devices
also broadly confirmed the magnitude and direction of
flows (including large reverse flows) measured with the
HRM devices used here (data not shown).
Soil moisture measurements
Soil moisture was significantly greater at the north and
east sampling positions compared with the south and west
positions around the tree depicted in Fig. 3. Values were
11.6% for N, 11.6% for E, 8.8% for S, and 8.5% for W.
Samples at the N position differed significantly from S
and W at the 0.1 level (Student’s t test), whereas samples
at the E position differed significantly from S and W at
the 0.01 level.
Discussion
Patterns of sap flow among lateral roots on opposing sides
of the salmon gum trees following rain showed a donor–
recipient pattern similar to hydraulic redistribution be-
tween taproots and lateral roots as extensively documented
in the literature using a range of thermometric techniques
(Burgess et al., 1998; Smith et al., 1999; Scholz et al.,
2002; Hultine et al., 2003a). Although hydraulic redis-
tribution between lateral roots has been demonstrated in
Fig. 5. Multiple changes to diurnal sap flow patterns in the stem of
Eucalyptus salmonophloia in response to rain, and the effect of severing
xylem vessels around the site of measurement (day 177 onwards).
Redistribution of soil water by lateral roots 3287
greenhouse-based split-root experiments (Sakuratani
et al., 1999) or induced by irrigation in field-based studies
(Brooks et al., 2002), the authors are not aware of similar
reports of lateral redistribution resulting from rain.
What drives water movement between lateral roots?
Hydraulic redistribution in any form requires that a root
system span soil layers (or compartments/patches) of dif-
ferent water potential. Our simple survey of soil moisture
contents on different sides of a single tree demonstrated that
considerable variability in water contents (and by inference
soil water potentials) of the upper soil layers is possible,
and this variability was in agreement with the direction of
hydraulic redistribution observed here. The reasons for soil
moisture variability around single trees remain unclear, but
may include rain shadows, patches of non-wetting soil, or
varying soil textures. Soils in this region typically exhibit
Downloaded from http://jxb.oxfordjournals.org by on May 13, 2010
strong texture contrasts from the A to B horizon and
variable subsurface relief (Tennant et al., 1992) can alter
at what depth a wetting front encounters heavy subsoils.
During excavations, pockets of extremely dry soil were
often encountered even after months of heavy winter rain.
In addition to soil variability, lateral rooting depth may
vary in the vertical plane. It is possible that some lateral
roots may support sinker roots that have penetrated the
heavier subsoils by following macropores. Such lateral
roots would then function partly as a taproot, and ‘similar-
looking’ laterals on different sides of a tree may be func-
tionally quite different. Given the case where the direction
of flow in the SW-oriented root (Fig. 3) is at odds with roots
on immediately adjacent sides, it could be surmised that
patterns of redistribution in each individual tree would
be a complex interaction of soil moisture distribution and
root distribution. Despite the exception of the SW-oriented
root of the tree depicted in Fig. 3, the distinct ‘sidedness’
seen for this and the second salmon gum measured
similarly suggests hypotheses such as rain shadow effects
are worth further investigation.
Sap flow patterns in stems
Installation of eight probes around the circumference of
two salmon gum trees demonstrated substantial spatial varia-
tion in night-time water flows in response to rain, which
followed a distinct temporal progression. The distribution
of different flow vectors followed a regular pattern around
the stem. Nearly every 458 sector of the stem had a night-
time rate of sap flow that was slightly different to adjacent
sectors. Although considered beyond the scope of this pre-
sent investigation, it was noted that the direction and
magnitude of flows measured deeper in the xylem (data
not shown) did not always match those measured in the
outer xylem, adding a complexity which suggests radial
sectoring in addition to circumferential sectoring. Both
circumferential and radial sectoring of flow pathways have
3288 Burgess and Bleby
previously been demonstrated using sap flow techniques
(Nadezhdina, 1999).
The overall pattern of night-time sap flow detected in
stems following rain consisted of (at least for the outer
xylem) large positive flows in sectors on one side of the tree
grading smoothly via sectors of neutral-flow to maximum
negative flows in sectors on the other side of the tree. To
our knowledge, such night-time flow dynamics have not
been recorded for stems before. Possible disturbance to the
measurements from such influences as stem flow, changes
to bark moisture content, water entry into probe drill-holes,
or datalogger grounding problems was ruled out experi-
mentally. Critically, xylem-severing experiments demon-
strated that flow patterns were solely dependent on the
continuity of xylem tissues surrounding the probes (bark
removal had no effect), confirming that xylem water flux
was responsible for the measured flow patterns. Further
confirmation of the magnitudes and directions of flow
measured by these sap flow systems by two commercial
sap flow devices is significant, since the SF-100 uses a dif-
ferent principle of measurement (see a description of the
compensation heat pulse method by Burgess et al.,
2000a), while the HRM-30 has the same measurement
principle but an entirely different electronic platform.
The tight coupling of the simultaneous positive and
negative flows in stems during the night strongly suggests
a donor–recipient relationship between flows of the same
type measured in the roots of these trees. In fact, the direc-
tion and magnitude of flows in each stem sector closely
corresponded to those of roots, not immediately below each
sector, but at approximately 458 rotation to the stem. Given
the similarity of flow patterns in the stem with those of
the roots, and allowing for some spiralling of stem tissues
(Howard, 1932), it is argued that the flows measured in the
stem were the result of exchange of water between roots,
as described below.
What is the mechanism for transfer of water among
poorly-connected lateral roots?
Under the assumption that a specialized ‘junction zone’
(Burgess et al., 2000b) does not exist at the stem base to
facilitate water exchange among roots, a number of other
pathways might be considered. Firstly, the possibility of
a radial exchange of water between different sides of these
mature trees can be excluded: tests with dye infiltration
readily demonstrate that heartwood blocks the majority of
this pathway. Secondly, flows around the circumference
of the tree may be possible via intervessel pits. If this path-
way were extremely conductive, one might expect rapid
flows in a narrow circumferential band immediately con-
necting the insertion points of each root. However, com-
pared with axial flow, it is more difficult for water to move
sideways through stem tissues, since it will encounter
a dense succession of vessel walls (and fibre cells) and must
in each case traverse pit membranes of comparatively high
resistance (Sperry and Hacke, 2004).
A third possibility is thus raised: if circumferential water
movement is largely limited by pit membrane resistance,
one means of increasing conductivity in this direction is to
increase the cross-sectional area over which water can flow
through pits. Given the comparatively high axial conduc-
tivity of xylem, this can readily be achieved via the axial
movement of water. For example, assume one lateral root
acts as a sink for water, exerting, for example, a hydrostatic
tension of 3 MPa on the rest of the tree. Because of low
axial resistance, there is only a small gradient in the water
potential along the paths connecting this root to the rest
of the tree. Hence a tension of 3 MPa exerted at the base
of the stem that might induce circumferential water move-
ment will produce a comparable driving force for further
circumferential water movement a number of metres higher
Downloaded from http://jxb.oxfordjournals.org by on May 13, 2010
up the stem. The same rules apply to the other side of a
tree, where a lateral root acting as a source of water (e.g.
a tension of only 1 MPa) is connected. Thus, circumferen-
tial flow will be driven across a plane that extends a number
of metres above soil level. Ultimately water that moves
circumferentially must also move axially from the ‘source’
to the ‘sink’, both of which are at the base of the tree.
In other words, a net circumferential transport of water
ought to induce a considerable accompanying axial flow.
Figure 6 is a simple cartoon of this concept of circumferential
Fig. 6. Conceptual diagram of the flow paths permitting exchange of
water between lateral roots attached to opposite sides of the same stem,
where the most direct radial pathway is blocked by heartwood. Thickness
and direction of arrows indicates the relative magnitudes and direction
(axial and circumferential) of water transported from root to root. This
model suggests that the magnitude of axial flows should diminish with
increasing stem height depending on the ratio of axial to circumferential
hydraulic conductivities.

transport of water between poorly-connected roots on oppo-
site sides of a stem as mediated by substantial axial flows.
As support for this idea that circumferential transport
must occur over a considerable axial plane, it is worth
remembering that the survival of trees in double saw cut
experiments depends on the cuts being made a considerable
vertical distance apart (Orians et al., 2005). Further support
comes from the findings of Schulte (2006), who was only
able to draw water from one branch of a root to another if
a considerable length of the parent root was retained to
facilitate the water transfer. Just how large the axial plane
along which circumferential transport occurs is in E.
salmonophloia and E. wandoo remains to be measured;
preliminary sap flow measurements at 4 m height suggest
tissues at this height are still partly involved. The amount
of axial flow relative to circumferential transport should
reflect the ratio of axial to circumferential hydraulic con-
ductivity of xylem, which depends on a range of anatom-
ical factors including vessel diameter, frequency of contact
between vessels and the number and conductivity of
intervessel pits (Domec et al., 2005; Orians et al., 2005).
The fact that E. salmonophloia demonstrates distinct flow
paths over a number of metres of stem tissue suggests
xylem of this species is highly sectored rather than
integrated.
How widespread is this phenomenon?
To date, stem-mediated hydraulic redistribution has been
observed in five species: Eucalyptus salmonophloia, E.
wandoo, E. astringens Maiden, E. horistes L. Johnson &
K. Hill (K Brooksbank, unpublished data), and E. polybractea
RT. Baker (J Carter, unpublished data). E. astringens has
a ‘mallet’ growth form (slender-stemmed tree with steeply
angled branches), while E. horistes and E. polybractea have
the ‘mallee’ growth form (woody species with multiple stems
arising from a lignotuber). These three species and the two
tree-form eucalypts described in the present study are all
endemic to semi-arid areas of Western Australia, and all grow
on heavy soils.
In terms of this study’s work, sap flow has been
measured in a number of other species using the HRM,
including other eucalypts, but this phenomenon has not
been detected before. The absence of this phenomenon in
gymnosperm species such as Sequoia sempervirens D. Don
(S Burgess, unpublished data) might be expected, since
tracheid structures are unlikely to give axial flow such
a strong advantage over flow in other directions. Its absence
in other Eucalyptus species will need to be explained in
terms of rooting architecture, soil properties, and xylem
structure.
In terms of the broader context of research on sap flow
using a range of techniques in a range of species and
ecosystems, it is not certain as to whether this phenomenon
is more widespread, but in some cases has not been detected
with the methods employed. Many of the previous studies
Redistribution of soil water by lateral roots 3289
on Eucalyptus species (Barrett et al., 1995) have used
commercially available compensation heat pulse techni-
ques that have poor resolution at the lower end of the
measurement range (Becker, 1998) and cannot detect flow
reversals (Burgess et al., 2000a). These techniques have
been used previously in E. salmonophloia by Farrington
et al. (1994) who noted rapid responses to rain, but pro-
vided no information on night-time sap flow rates. Simi-
larly, the thermal dissipation probe or ‘Granier method’
(Granier, 1985), which has been popular in many North
American studies, is sometimes used in a mode that resets
minimum night-time flows to zero every night. This ap-
proach would obscure any night-time dynamics as reported
in the present study. Thus, the question remains as to how
widespread this phenomenon is, and whether species or
soil-type is the most important factor generating this type
of water flow in tree stems.
Downloaded from http://jxb.oxfordjournals.org by on May 13, 2010
Conclusions
Although Brooks et al. (2002) and Smart et al. (2004) have
demonstrated the possibility of lateral redistribution of
water under field conditions following irrigation, the
findings presented here are, as far as is known, the first
example of the transfer of water between lateral roots as
a natural phenomenon following rain. Furthermore, the
magnitudes of sap flow resulting from the exchange of
water between roots in these large trees were considerably
greater than those in any other studies the authors are pre-
sently aware of that used comparable techniques. Speeds of
water flow in roots at night approached rates more typical
of those found in stems during peak transpiration. (Here
we mean peak flows in the absence of this phenomenon.
Since flows caused by transpiration are additive to flows
caused by hydraulic redistribution, peak rates under such
conditions can be higher than is ‘typical’ and will always
be greater than any night-time flow rates.) These rapid
flows suggest fairly large volumes of soil water are being
redistributed, with potentially large ecological and hydro-
logical implications for these water-limited ecosystems.
Secondly, it is believed that this is the first time that large
axial flows involving many metres of stem tissue have been
implicated as part of the transport pathway that permits
water to be transferred between roots. Night-time flow rates
in the stems were not much slower than those measured
in roots; clearly, exchange of water between roots is a
non-trivial transport role for stem tissues in some spe-
cies. Further work is needed to determine its prevalence in
woody plants, i.e. does it reflect certain root architectures,
growth forms, soil types or climatic factors? Subsequent
research should focus on the relative hydraulic conductiv-
ities of xylem in the radial, circumferential, and axial
planes. Such knowledge should provide further insights
into flow paths and source–sink relationships within a plant.
3290 Burgess and Bleby
Finally, a broader survey of species/ecosystems where
this process may be relevant would be beneficial to assess
its implications for typical transpiration studies that utilize
sap-flow gauges installed in stems. Where axial water move-
ment that is not the result of transpiration is present, either
very careful circumferential sampling strategies, or proto-
cols to separate the two processes, will be needed in order to
estimate total plant water use.
Acknowledgements
We wish to thank the Cooperative Research Centre for Plant-Based
Management of Dryland Salinity and the Australian Research
Council for financial assistance. We thank Perry Swanborough and
Patrick Mitchell for assistance in the field, Hans Lambers, Erik
Veneklaas, Mark Adams, Jenny Carter, and Kim Brooksbank for
helpful discussions. Thanks also to Robin Campbell at the Corrigin
Land Care Office and the Ecosystems Research Group (UWA
School of Plant Biology) for general assistance.
References
Baker JM, Van Bavel CHM. 1986. Resistance of plant roots to
water loss. Agronomy Journal 78, 641–644.
Barrett DJ, Hatton TJ, Ash JE, Ball MC. 1995. Evaluation of
the heat pulse velocity technique for measurement of sap flow in
rainforest and eucalypt forest species of south-eastern Australia.
Plant, Cell and Environment 18, 463–469.
Beard JS. 1990. Plant life of Western Australia. Kenthurst:
Kangaroo Press.
Becker P. 1998. Limitations of a compensation heat pulse velocity
system at low sap flow: implications for measurements at night and
in shaded trees. Tree Physiology 18, 177–184.
Benyon RG. 1999. Nighttime water use in an irrigated Eucalyptus
grandis plantation. Tree Physiology 19, 853–859.
Breazeale JF. 1930. Maintenance of moisture-equilibrium and
nutrition of plants at and below the wilting percentage. Arizona
Agricultural Experimental Station Technical Bulletin 29,
137–177.
Brooks JR, Meinzer FC, Coulombe R, Gregg J. 2002. Hydraulic
redistribution of soil water during summer drought in two
contrasting Pacific Northwest coniferous forests. Tree Physiology
22, 1107–1117.
Burgess SSO, Adams MA, Bleby TM. 2000a. Measurement of sap
flow in roots of woody plants: a commentary. Tree Physiology 20,
909–913.
Burgess SSO, Adams MA, Turner NC, Beverly CR, Ong CK,
Khan AAH, Bleby TM. 2001a. An improved heat pulse method
to measure low and reverse rates of sap flow in woody plants. Tree
Physiology 21, 589–598.
Burgess SSO, Adams MA, Turner NC, Ong CK. 1998. The
redistribution of soil water by tree root systems. Oecologia 115,
306–311.
Burgess SSO, Adams MA, Turner NC, Ong CK, Khan AAH,
Beverly CR, Bleby TM. 2001b. Correction: an improved heat
pulse method to measure low and reverse rates of sap flow in
woody plants. Tree Physiology 21, 1157.
Burgess SSO, Dawson TE. 2004. The contribution of fog to the
water relations of Sequoia sempervirens (D. Don): foliar uptake
and prevention of dehydration. Plant, Cell and Environment 27,
1023–1034.
Burgess SSO, Pate JS, Adams MA, Dawson TE. 2000b. Sea-
sonal water acquisition and redistribution in the Australian
woody phreatophyte, Banksia prionotes. Annals of Botany 85,
215–224.
Caldwell MM, Dawson TE, Richards JH. 1998. Hydraulic lift:
consequences of water efflux from the roots of plants. Oecologia
113, 151–161.
Dawson TE. 1993. Hydraulic lift and water use by plants: impli-
cations for water balance, performance and plant–plant interac-
tions. Oecologia 95, 565–574.
Domec J-C, Meinzer FC, Gartner BL, Woodruff D. 2005.
Transpiration-induced axial and radial tension gradients in trunks
of Douglas-fir tree. Tree Physiology 26, 275–284.
Espeleta JF, West JB, Donovan LA. 2004. Species-specific patterns
of hydraulic lift in co-occurring adult trees and grasses in a sandhill
community. Oecologia 138, 341–349.
Farrington P, Bartle GA, Watson GD, Salama RB. 1994. Long-
term transpiration in two eucalypt species in a native woodland
estimated by the heat-pulse technique. Australian Journal of
Ecology 19, 17–25.
Granier A. 1985. A new method of sap flow measurement in tree
Downloaded from http://jxb.oxfordjournals.org by on May 13, 2010
stems. Annales des Sciences Forestieres 42, 193–200.
Hobbs RJ, O’Connor MH. 1999. Designing mimics from in-
complete data sets: salmon gum woodland and heathland ecosys-
tems in South West Australia. Agroforestry Systems 45, 365–392.
Howard NF. 1932. Twisted trees. Science 75, 1323.
Hultine KR, Cable WL, Burgess SSO, Williams DG. 2003a.
Hydraulic redistribution by deep roots of a Chihuahuan Desert
phreatophyte. Tree Physiology 23, 353–360.
Hultine KR, Williams DG, Burgess SSO, Keefer TO. 2003b.
Contrasting patterns of hydraulic redistribution in three desert
phreatophytes. Oecologia 135, 167–175.
Lamont B. 1985. Gradient and zonal analysis of understorey
suppression by Eucalyptus wandoo. Vegetatio 63, 49–66.
Nadezhdina N. 1999. Sap flow index as an indicator of plant water
status. Tree Physiology 19, 885–891.
Orians CM, Babst B, Zanne AE. 2005. Vascular constraints and
long-distance transport in dicots. In: Holbrook NM, Zwieniecki
MA, eds. Vascular transport in plants. Oxford: Academic
Press.
Pate JS, Jeschke WD, Alyward MJ. 1995. Hydraulic architecture
and xylem structure of the dimorphic root systems of South-West
Australian species of Proteaceae. Journal of Experimental Botany
46, 907–915.
Richards JH, Caldwell MM. 1987. Hydraulic lift: substantial
nocturnal water transport between soil layers by Artemisia
tridentata roots. Oecologia 73, 486–489.
Sakuratani T, Aoe T, Higuchi H. 1999. Reverse flow in roots of
Sesbania rostrata measured using the constant power heat balance
method. Plant, Cell and Environment 22, 1153–1160.
Scholz FG, Bucci SJ, Goldstein G, Meinzer FC, Franco AC.
2002. Hydraulic redistribution of soil water by neotropical savanna
trees. Tree Physiology 22, 603–612.
Schulte PJ. 2006. Water flow through junctions in Douglas-fir roots.
Plant, Cell and Environment 29, 70–76.
Schulze ED, Caldwell MM, Canadell J, Mooney HA, Jackson
RB, Parson D, Scholes R, Sala OE, Trimborn P. 1998.
Downward flux of water through roots (i.e. inverse hydraulic lift)
in dry Kalahari sands. Oecologia 115, 460–462.
Smart DR, Carlisle E, Goebel M, Núñez BA. 2004. Transverse
hydraulic redistribution by a grapevine. Plant, Cell and Environ-
ment 28, 157–166.
Smith DM, Jackson NA, Roberts JM, Ong CK. 1999. Reverse
flow of sap in tree roots and downward siphoning of water by
Grevillea robusta. Functional Ecology 13, 256–264.

Snyder KA, Richards JH, Donovan LA. 2003. Night-time
conductance in C3 and C4 species: do plants lose water at night?
Journal of Experimental Botany 54, 861–865.
Sperry JS, Hacke UG. 2004. Analysis of circular bordered pit
function. I. Angiosperm vessels with homogenous pit membranes.
American Journal of Botany 91, 369–385.
Tennant D, Scholz G, Dixon J, Purdie B. 1992. Physical and
chemical characteristics of duplex soils and their distribution in the
south-west of Western Australia. Australian Journal of Experi-
mental Agriculture 32, 827–843.
Redistribution of soil water by lateral roots 3291
Williams K, Caldwell MM, Richards JH. 1993. The influence of
shade and clouds on soil-water potential: the buffered behavior of
hydraulic lift. Plant and Soil 157, 83–95.
Yoder CK, Nowak RS. 1999. Hydraulic lift among native
plant species in the Mojave Desert. Plant and Soil 215,
93–102.
Zwieniecki MA, Orians CM, Melcher PJ, Holbrook NM.
2003. Ionic control of the lateral exchange of water between
vascular bundles in tomato. Journal of Experimental Botany 54,
1399–1405.
Downloaded from http://jxb.oxfordjournals.org by on May 13, 2010