www.elsevier.com/locate/plantsci

Plant Science 171 (2006) 572–584

Sap flow and daily electric potential variations in a tree trunk

Dominique Gibert a,*, Jean-Louis Le Mouel c, Luc Lambs b,Florence Nicollin a, Frederic Perrier c,d

a Geosciences Rennes (CNRS UMR 6118), Universite Rennes 1, Bat. 15 Campus de Beaulieu, 35042 Rennes cedex, Franceb Laboratoire Dynamique de la Biodiversite, CNRS, 29 rue Jeanne Marvig, 31055 Toulouse, France

c Equipe de Geomagnetisme, Institut de Physique du Globe de Paris, 4 place Jussieu, 75252 Paris cedex 05, Franced Commissariat a l’energie atomique, 91680 Bruyeres-le-Chatel, France

Received 11 January 2006; received in revised form 9 May 2006; accepted 2 June 2006

Available online 10 July 2006

Abstract

Electric potential has been monitored since December 2003 in the roots and at two circumferences and one vertical profile in a standing poplar

(Populus nigra). Electric potential is sampled using 6 mm diameter stainless steel rods, inserted 5 mm deep in the sapwood and is referenced to an

unpolarisable lead/lead chloride electrode installed 80 cm deep in the soil. Diurnal variations are observed with seasonal differences. During

winter, diurnal variations depend on the measurement point, with variable amplitudes and sometimes anti-correlations between electrodes. By

contrast, a stable and coherent organisation is established in the spring, with larger amplitudes, and lasts during summer. Dedicated experiments

have been performed to rule out a direct effect of temperature on the electrodes, and thus, demonstrate a genuine electrical source in the tree. Daily

electrical variations have been reported previously, and have been interpreted as electrokinetic effects associated with sap flow. However, a

comparison of the electrical signals with a direct measurement of the sap flow by a continuous heat flow method, shows that the electrical variation,

although clearly correlated to sap flow, is not simply proportional to it. In a living system, electrokinetic effects, in addition to thermoelectrical

effects, are probably modified significantly by additional electrochemical effects, such as membrane diffusion potentials, ion active transport by

proteins or action potentials. Electric potential variations in trees may thus, reveal physical mechanisms in living systems not accessible by other

methods. A better understanding of the electrical response of trees associated with sap flow may improve the knowledge of transfer processes

between the soil and the atmosphere. This is important for the understanding of adaptive response of trees, the modelling of water and carbon

balance in relation to climate change, and the quantification of the contribution of trees to the migration, retention and dispersion of contaminants.

# 2006 Elsevier Ireland Ltd. All rights reserved.

PACS: 02.70.Uu; 02.30.Zz; 91.25.Qi; 91.35.Pn; 91.40. � k; 93.30.Vs

Keywords: Electric potential; Electrokinetic effects; Membrane potential; Sap flow; Temperature effects

1. Introduction

Estimating the effect of global climatic changes on

ecosystems and the associated feedback mechanisms requires

a better understanding of tree transpiration and carbon

assimilation [1]. These processes depend on sap flow, which

drives the whole tree physiology, the water balance and the

hydrology of the subsurface as well (e.g., [2–5]) and they

* Corresponding author. Tel.: +33 223236091; fax: +33 223236090.

E-mail addresses: dominique.gibert@univ-rennes1.fr (D. Gibert),

lemouel@ipgp.jussieu.fr (J.-L. Le Mouel), lambs@cict.fr (L. Lambs),

florence.nicollin@univ-rennes1.fr (F. Nicollin),

perrier@ipgp.jussieu.fr (F. Perrier).

0168-9452/$ – see front matter # 2006 Elsevier Ireland Ltd. All rights reserved.

doi:10.1016/j.plantsci.2006.06.012

have been studied in dedicated large-scale field experiments

(e.g., [6,7]). The direct measurement of sap flow up to now is

based on thermal methods [8], which can be separated into

heat pulse velocity techniques and heat balance techniques

[9]. However, the calculation of the total water flux remains

difficult. At the forest level, the water is shared between trees

of different species and sizes [10]. At the tree level, several

points need clarification, such as the contribution of night

respiration [11,1], the radial variation of the sap velocity in

the sapwood [12,13] or the occasional reverse sap flow in

roots [14]. Sap flow is different for the three major xylem sap

conducting systems (non-, diffuse- and ring-porous) and does

not bear a simple relationship with tree diameter or canopy

surface. Estimating the water flux from sap velocity

D. Gibert et al. / Plant Science 171 (2006) 572–584 573

1 WGS 84 geographic coordinates: 47�5600500N and 2�5304500W, i.e. Uni-

versal Time used in figures is almost the same as the local Solar Time.

measurement, thus, remains affected by large uncertainties

[10].

Sap flow may also be accessed through measurements of the

spontaneous electric potential of trees. Early measurements in an

elm trunk [15] showed that the electric potential of the trunk with

respect to a ground electrode shows daily variations with

amplitude of the order of 20–40 mV, and a minimum of the

potential in the afternoon. These daily variations have been

associated with the sap flow. Indeed, the motion of liquids in

porous media produces electric potential variations referred to as

streaming potentials or electrokinetic effects (e.g., [16–20]). This

effect is characterised by one coupling parameter, the streaming

potential coefficient, defined as the ratio of the potential

difference to the pressure difference. Electrokinetic effects have

been observed in laboratory experiments with rocks (e.g., [21]);

they lead to remarkable spatial and temporal variations of the

spontaneous electric potential (SP) in natural conditions

[22,17,23], especially near active volcanoes (e.g., [24]).

The early observation on the tree trunk was confirmed by an

experiment on a spruce tree [25], which in addition reported a

non-linear relationship between the daily electric potential

variation and the intensity of the solar radiation. Measurements

at the surface of a chestnut tree [26] indicated that the time

variations of the electric potential depend strongly on the

season. In summer, clear daily variations of the trunk potential

are observed, with a minimum in the afternoon when upper

electrodes are referred to lower electrodes, in agreement with

the previous experiments. In the late fall, after leaf shedding,

these clear daily variations disappear, as expected if sap flow is

the dominating mechanism for electric potential generation.

However, electrical signals, with more erratic time variations,

are still observed, indicating that other mechanisms may also

contribute, possibly unrelated to sap flow. The electric potential

was also monitored at several points of a Turkey oak (Quercus

cerris) during 2 years [27,28]. This experiment indicated that

the summer daily variations are in phase at the various points,

with again a minimum of the potential with respect to the

ground in the afternoon. The amplitude can vary by more than a

factor of two from point to point. Two periods for large daily

variations were observed: end of March and summer (end of

June–July).

Streaming potentials have also been studied in Salix alba L.

sapwood samples in the laboratory and the results have been

compared with the observed electrical daily variation in a

standing tree to obtain estimates of the sap flow velocity [29].

The sap flow velocity, thus, estimated from the electrical

variations, is 15–17 m h�1, which is much higher than the

accepted values for S. alba L., which are of the order of 2–

3 m h�1[29]. The positive sign of the streaming potential

coefficient obtained in the laboratory is in agreement with the

sign reported for most rocks in usual natural conditions (e.g.,

[30]), but does not agree with the sign of the observed electric

daily variation in the tree trunk. Indeed, for a positive streaming

potential coefficient, the potential of the trunk should be

positive with respect to the ground. While the observed daily

variation could be reasonably interpreted at first in terms of

streaming potentials, the relationship between the sap flow and

the electric potential variation therefore remains puzzling.

Other effects, such as thermo-electrical effects, membrane

potential and experimental artifacts, need to be considered

before a coherent physical interpretation can be proposed. It

appears, therefore, interesting to undertake new experiments to

investigate in more details the relationship between the sap flow

and the electric potential variations in a tree.

Monitoring of the electric potential in trees and more

generally in living plants, may be interesting for other reasons.

Electric activity has been evidenced in hibiscus and maize (e.g.,

[31,32]) and electrical signals have been observed in response

to changes in transpiration and photosynthesis in willow plants

[33] or Mimosa pudica [34]. The electric potential in trees may,

therefore, reflect a combination of physical, chemical and

physiological responses in relation with water transport,

photosynthesis and adaptive feedback mechanisms. The

seasonal variation of the electrical daily variation, for example,

may be closely associated with the poorly understood seasonal

variations of the enzymes of sucrose metabolism [35].

In this paper, we present a new experiment lasting more than

2 years in a poplar tree. In this experiment, we have focused on

the spatial and temporal variations of the electric potential

distribution in a single tree, using an increased number of

electrodes compared with previous work. The comparison of

several trees will be considered in a later stage. The

experimental set-up is first presented. After a brief overview

of the annual cycles of the tree, we present specific experiments

performed to constrain the physical mechanisms. In particular,

the electrical daily variations are compared with a direct and

independent measurement of the sap flow with a thermal

method. In addition, temperature effects, which can potentially

affect the measured electric potential, are studied in details. In

the conclusion, we discuss the generation mechanisms of

electric potential in the tree trunk.

2. Experimental set-up

2.1. Electrode array

The investigated tree is a poplar (Populus nigra L.) located in

Remungol1 (Brittany, France). The first part of the experimental

set-up, with a set of 26 electrodes, was installed on August 6,

2002, 21 electrodes in the trunk and five in two emerging roots

(Figs. 1 and 2). Trunk electrodes are arranged in three groups:

two circular rings and one vertical line. The lower ring, with

eight electrodes numbered E1 to E8, is located 1.0 m above the

ground and has a circumference of 2.7 m. The upper ring, with

eight electrodes numbered E11 to E18, is located 3.4 m above

the ground and has a circumference of 2.4 m. The vertical line

comprises five electrodes numbered from E30 (0.5 m above the

ground) to E34 (2.9 m above the ground) and aligned with E6

and E16 on the northern face of the trunk. Root electrodes E01,

E02 and E03 (Figs. 1 and 2) are implanted in a root running

D. Gibert et al. / Plant Science 171 (2006) 572–584574

Fig. 1. General view (from west) of the array of electrodes shown as black dots

on a simplified sketch of the tree. The box labelled T–H indicates the position of

the outdoor temperature and humidity sensors.

Fig. 2. Plan view showing the canopy area, the trunk and the two roots equipped

with electrodes. Also shown are the two Petiau electrodes, P1 and P2.

eastward from the trunk, partially emerging from the ground.

The diameter of this root is about 0.18 m, and the electrodes

E01, E02 and E03 are at distances of 0.45, 0.90 and 1.35 m

from the eastern edge of the trunk, respectively. Electrodes E04

and E05 are located at 0.6 and 1.2 m from the north-eastern

basis of the trunk on a shallow root with a diameter of 0.15 m

(Fig. 2). Seven electrodes (E21 installed on May 31, 2003, and

E22 to E27 installed on August 18 and 27, 2004) were added

along a vertical segment whose lower (E21) and upper (E26 and

E27) ends are at 5.6 and 10.5 m above the ground, respectively

(Fig. 1).

The electrodes have a diameter of 6 mm and were cut in

stainless steel rods which were carefully degreased 1 day

before the installation. The apparent metallic part of the

electrodes is cut to 15 mm, and each electrode is connected

by wrapping a cable which is secured with a epoxied cap of

thermo-sheath (Fig. 3). The connectivity of an electrode with

its cable is tested with an ohm-meter. The electrodes are

inserted in the tree by drilling a hole with a diameter of 8 mm

through the bark and another one with a diameter of 6 mm

and a depth of 15 mm in the wood. Next, the electrodes are

gently hammered until being embedded in the wood. In this

way, the non-isolated metallic part of the electrodes is fully

stuck in the wood so that the meteorological influences, such

as rain or fast temperature changes are reduced. The contact

resistance of the electrodes has been measured with an AC

ohm-meter and varies from 4700 to 6400 V with an average

of 5400 V. The electrode array is complemented with a non-

polarisable lead–lead chloride electrode [36] which is

used as a reference for all potential measurements. This

electrode is located 5.0 m away from the tree in the eastward

direction (Figs. 1 and 2) and is buried at a depth of 0.7 m

with clay as contact material. On May 31, 2003, another non-

polarisable electrode labelled E102, was buried using the

same method at a depth of about 0.3 m near the root

electrode E01.

2.2. Data acquisition system

Data acquired at the beginning of the experiment are not

used in the present study, and only those obtained with the

digital acquisition system installed in November 2003 are

presented here. The measurement device is a Keithley 2701

digital multimeter with an input impedance larger than 100 M

V and equipped with a relay matrix having 40 measurement

D. Gibert et al. / Plant Science 171 (2006) 572–584 575

Fig. 3. Top: detailed view of an electrode. The apparent metallic part of the electrode is fully embedded in the wood. The epoxied cap secures the cable which is

wrapped around the stainless steel rod. Bottom left: view of an electrode placed on the tree. The uncovered part of the electrode is fully embedded in the wood under

the bark in order to reduce meteorological influences. Bottom right: in the Granier’s method, the sap flux is obtained by measuring the thermal flux advected by the sap

flow with one heater (H) and two thermo-couples (T).

channels controlled through an acquisition software. Measure-

ments of the electric potential are made at all electrodes with

a sampling interval of 1 min. We use a UT time base,

synchronised in real time to the Frankfurt atomic clock. Both

the computer and the multimeter are powered with a backup

generator which prevents breaks caused by short failures of the

electrical power line.

In order to check the entire electrical system, and in

particular, the high input impedance of the recorder, we

performed several injections of electrical current in the soil

near the basis of the tree. The applied voltage at the grounded

electrodes was 12 V for an electrical current of 2.5 mA.

Fig. 4 shows the electrical potential measured at several

Fig. 4. Electrical potential measured at several electrodes during an injection of

electrical current in the soil near the tree. Hours are given in Universal Time.

electrodes during two current injections of opposite signs.

The response due to the injections is clearly seen on all

channels. In addition, the electric potential response happens

within one sampling time. This indicates that our measure-

ments are not affected by artificial delay times or capacitive

effects, that could, for example, have resulted from an

insufficient input impedance of the acquisition system. Such

tests are performed regularly.

2.3. Meteorological measurements

In November 2003, a meteorological station was installed to

record the main atmospheric parameters in the neighbourhood

of the tree. The measured parameters include the speed and

direction of the wind, the atmospheric pressure, the outside

temperature and humidity, the rain events. Temperature and

humidity are also measured indoor in order to identify a

possible sensitivity of the electronic devices to the environ-

mental conditions. All meteorological parameters are recorded

every 15 min and synchronised with the time base used for the

potential measurements.

2.4. Influence of temperature on the electrode array

Stainless-steel electrodes might be affected by significant

temperature effects, and it is important to discuss purely

thermal effects as the cause of the observed diurnal variations of

the electric potential. In principle, this is possible, as the

temperature sensitivity of the electrode–wood contact may be a

D. Gibert et al. / Plant Science 171 (2006) 572–584576

Fig. 5. Thermal tests made on electrode E31 to verify that no important

potential variations are induced by temperature. Potential at nearby E6 elec-

trode and outdoor temperature are given for reference. The transient variations

of E31’s potential are observed when introducing the warm (or cold) bottle in

the box covering the electrode. Hours are given in Universal Time.

rapidly varying function of the water content in the wood and

the air humidity. Temperature influence on the tree electrodes

would introduce temporal effects depending on position. For

instance, the amplitude of the potential variations should be

larger on the south side exposed to solar heating. Note that this

is not what we observe actually; however, dedicated thermal

experiments were performed to check for a possible

temperature effect on our electrode array.

As shown by Petiau [36], the temperature sensitivity of the

grounded electrochemical electrode used as as reference is

negligible (smaller than 30 mV=�C), and our thermal experi-

ment focused on the steel electrodes. In August 2004, we

applied large artificial thermal perturbations to E31 which was

previously covered with an insulating pot. Fig. 5 shows the

results obtained during both a heating and a cooling test done by

introducing a bottle of hot water or ice in the pot. Excepted for

sharp transient variations of the potential occurring when

introducing the bottle, no perturbation of the E31 potential is

observed when compared with the potential measured on the

nearby E6 electrode. This clearly rules out a direct effect of

temperature on the measured electrode potential.

Looking at Fig. 5 more closely, we observe that potential

variations at both E31 and E6 have features similar to those

present in the curve of the outdoor temperature measured in the

vicinity of the tree (Fig. 1). This condition is, however, not due

to a temperature effect at the electrodes, but to a global effect

affecting the whole tree. This was further checked by studying

the correlation between the potential and temperature curves. In

particular, we observed that the correlation between the E6 and

E31 potential curves is maximal for zero-time delay both before

and, significantly, after the installation of the insulating pot on

E31. Conversely, when the pot is installed, the correlation

between the outdoor and E31 temperature curves is maximum

for a time-delay of 45 min. This is another manner to

demonstrate that the delayed correlation between electric

potential and outdoor temperature results from the activity of

the tree.

2.5. Sap flow measurements

During a limited period of time (from June 21 to July 15 of

2004), direct measurements were performed in addition to the

electrode measurements. The sap flow is measured with the

heat-balance technique initially proposed by Granier [9]. This

method uses a pair of probes, each equipped with a small

resistor to produce heat and a miniature thermocouple to

measure the temperature inside the wood. The diameter of a

probe equals 2 mm for a length of 20 mm. Only the upper

heater is turned on and both probes are used to make

differential temperature measurements in order to measure the

thermal imbalance due to sap flow (Fig. 3). One pair of probes

was placed in the root near E01 and two pairs were placed

10 cm apart between E31 and E32 (Fig. 1). The part of the

trunk holding the probes was covered with an insulating

blanket in order to attenuate the thermal perturbations

produced by the atmospheric variable conditions. The

temperature was measured every 5 min and these data were

later converted into sap flow current through an experimentally

derived calibration [9].

3. Presentation of data

3.1. Overview of electric potential variations

Figs. 6–9 present the data for seven electrodes from

December 1, 2003 to May 21, 2005. This choice of seven

electrodes is representative of the whole electrode array: one

root electrode (E01), E6 on the lower ring, E11 and E18 on the

upper ring, E34 located on the vertical line just beneath the

upper ring, E21 and, when available, E25 located on the vertical

line 5.50 and 9.50 m above the floor, respectively. Two

breakdowns of long duration interrupt the curves, the first from

January 12, 2004 to February 17, 2004, the second from

September 4, 2004 to September 15, 2004.

These recordings of long duration show that the electrical

activity is present on all electrodes over the whole time span

although with varying amplitudes and a high variability from

one electrode to another. Large daily variations are present on

most electrodes from spring to summer. A first important

observation is the persistence of the electrical activity during

D. Gibert et al. / Plant Science 171 (2006) 572–584 577

Fig. 6. Top: potential signals measured on six electrodes representative of the entire electrode array for the December 2003–April 2004 period. Relative potential

values. Bottom: outdoor temperature measured near the tree (see Fig. 1 for location). Tick marks fall at midday.

Fig. 7. Same as Fig. 6 for the April–August 2004 period.

Fig. 8. Same as Fig. 6 for the September 2004–January 2005 period.

D. Gibert et al. / Plant Science 171 (2006) 572–584578

Fig. 9. Same as Fig. 6 for the January–May 2005 period.

winter (Figs. 6 and 8) when the tree activity is expected to be

greatly reduced. During this period, a daily variation is

observed on all electrodes, sometimes with a high amplitude as

large as the one seen in spring and summer. This is, for

example, the case for electrode E18 whose diurnal amplitude

remains of the order of � 30–50 mVall year long. Note that in

December, the E6 curve appears to be the upper envelope of the

E18 signal. In winter, the signals appear one-sided with daily

positive jumps with respect to a smoothly varying base level. In

spring and summer, this one-sided structure is less visible and

replaced by a more symmetrical pattern. This is particularly

clear when comparing the E18 curve in September and

December 2004 in Fig. 8.

At many points in the tree, long-term and daily electric

potential variations appear anti-correlated with the variations of

the outdoor temperature. This is, for example, the case for the

long-term trends of E01 and E6 in December 2003 (Fig. 6), or

for the long-term trend and the daily variations of E6 March–

April 2004. Note, however, that in March–April 2004, there is

no clear relationship between E01 and the temperature,

Fig. 10. Enlargement of a part of Fig. 9 showing the occurrence of negative daily ju

outdoor temperature. Notice the persistence of the daily positive jumps for electro

indicating that this anti-correlation between electric potential

and temperature is complex and time-dependent. When

considering the diurnal signal of a given electrode on a

long-term basis (Figs. 6–9), larger temperature daily variations

are not systematically associated with larger electrical diurnal

amplitudes as would be expected for a dominant temperature

effect. These observations, together with the thermal tests

discussed above, rule out the hypothesis that the observed

electrical variations are due to a purely thermal artifact at the

electrical contact between steel and wood. Instead, a physical

mechanism indirectly connected to the temperature and

involving the activity of the tree must be invoked, as proposed

by previous authors.

Additional peculiar observations are made when looking at

the period from February to March 2004, a period which is also

shown with an enlarged time scale in Fig. 10. During this

period, all electrodes, except electrode E18, display large

negative peaks, for example, around February 23, March 7 or

March 13 (Fig. 10), while E18 shows a more regular daily

variation. Interestingly, these large negative peaks on the

mps of the electric potential (E6, E34, E11 and E21) corresponding to negative

de E18. Tick marks of horizontal axis fall at midday.

D. Gibert et al. / Plant Science 171 (2006) 572–584 579

Fig. 11. (a) Sap flow measured with two probes located near E32. (b) Outdoor temperature. (c) Electrical potential measured at E32. (d) First time-derivative of the

E32 potential shown in (c). Tick marks of horizontal axis fall at midday.

potential occur precisely when the temperature goes below

zero. The same phenomenon is observed during the winter of

2004–2005 (Fig. 9), especially in the second half of February

2005.

The zoomed signals shown in Fig. 10, reveal the occurrence

of dipolar transient signal with a duration of several hours on

March 3 and 6, 2004. Such signals frequently occur with a

duration varying from several minutes to several hours. The

amplitude of the dipolar signal increases with the altitude of the

considered electrode, a feature observed for all similar signals

we examined, which rules out the possibility that these signals

are caused by instabilities of the reference ground electrode.

The amplitude of these transient signals depends more strongly

on altitude than observed for the step-like signals recorded

during the test of current injection in the soil (Fig. 4). It is then

unlikely to explain these transient signals by invoking telluric

currents, either natural or anthropic, originating from the

ground. Rain or storms, which could have been a first

explanation, must also be excluded using our meteorological

data. The origin of such signals, which appear occasionally,

remains unclear and deserve further study.

A big change occurs in the recorded signal from winter to

spring in both years 2004 and 2005 with the onset of a stable

daily activity at all electrodes. This onset is particularly clear in

April and May of years 2004 (Fig. 7) and 2005 (Fig. 9) with

starting times differing of several weeks from one electrode to

another. For instance, activity begins around March 20, 2005

for E18 and around May 1, for E25 (Fig. 9). Once established,

the daily activity remains high during summer and autumn, and

the return to the low winter activity also occurs at noticeably

different dates depending on the electrode (Fig. 8).

3.2. Comparison with sap flow measurements

We now turn to the comparison between electrical signals

and sap flow measurements. The sap flow curves corresponding

to the probes located near E32 are shown in Fig. 11 a for the

period going from June 21 to July 15, 2004. The two curves

show clear diurnal variations; they have a similar appearance

but their amplitudes differ up to a factor of two on certain days.

The electrical potential (Fig. 11c) measured at E32 and its first

time-derivative (Fig. 11d) also show a diurnal variation with a

variable amplitude which does not appear to be directly related

to the amplitude of the sap flow. The same temporal variation of

the electrical signal was observed during the days preceding the

installation of both the thermal probes and the insulating

blanket. This ensures that the electrical variations are not due to

the heating device.

To look more closely at the correlation between the

electrical signal and the sap flow, a period of 4 days was

selected (Fig. 12). This detailed view allows to examine

precisely the temporal coincidence between the sap flow and

the electrical variations, either in the potential curve or in its

time-derivative. The variations goes from the early time in the

morning, when the sap flow begins to be reactivated, to the

mid-morning when the sap flow reaches its maximum value.

During this period of sap flow reactivation, the time-

derivative of the potential curve shows a conspicuous bipolar

variation with a high amplitude, beginning with a positive

lobe. The late negative lobe of the bipolar event reaches its

minimum value (black dashed line) precisely when the sap

flow stops increasing and reaches a plateau. At that time, the

potential begins to decrease and becomes minimal slightly

after the sap flow has left its plateau-level and begins to

diminish sharply. The electrical potential then returns to its

original value in about 8 h, while a zero sap flow is reached

faster, after about 4 h. The time when zero sap flow is reached

(red dashed line) or keeps decreasing slowly asymptotically

to zero, also coincides approximately with the maximum of

the time-derivative of the electrical potential. Notice that a

large electrical signal still remains while the sap flow has

completely vanished.

D. Gibert et al. / Plant Science 171 (2006) 572–584580

Fig. 12. Same as in Fig. 11 for a limited period showing the correlations between electrical and sap flow events. Hours are given in Universal Time.

4. Discussion

In the rather small number of experiments made on standing

trees, listed in the introduction, the measured electric potentials

were attributed to an electrokinetic effect forced by the sap flow

[15,26,25]. The rough agreement of the value of the

electrokinetic coupling coefficient measured in the laboratory

with the value required to account for the observed magnitude

of the potentials in the standing tree was taken as a strong clue

in favour of this hypothesis, although the sign does not appear

to be the correct one. Furthermore, these previous experiments

were generally performed over short time periods, with few

Fig. 13. Amplitude of the electric potential variation from

electrodes. Our new observations make this interpretation quite

questionable. Indeed:

(1) A

the

negative potential is observed on all the trunk electrodes

with respect to the ground.

(2) I

n the framework of electrokinetic theory, the electric field

is proportional to the pressure gradient, hence, to the sap

flow per unit surface. This implies, at least in average, for

example, taken over the eight electrodes of a ring, a linear

increase of the electric potential amplitude with height in

the trunk (recall that the reference electrode is located in the

ground). This variation is not observed at all (Fig. 13).

5:00 a.m. to 5:00 p.m. period of June 1, 2004.

D. Gibert et al. / Plant Science 171 (2006) 572–584 581

(3) T

Fig.

scale

he heterogeneity of the diurnal amplitudes of the electric

potential from electrode to electrode is difficult to explain in

terms of an electrokinetic effect. Of course, we may expect

lateral heterogeneity in the xylem vessels through which the

sap is pumped up. However, the tree is a conducting

medium, and some homogenisation of the potential at a

given height is expected.

(4) T

he strongest argument comes from the comparison of the

electric potential curves with the sap flow curves (Figs. 11

and 12). Clearly, the electric potential is not at pace with

the sap flux, assuming that the thermal method measures

the sap flux with negligible retardation. Actually, our

measurements of the sap flow at two different points

provide the same time variation and support this

assumption. In particular, we note that a strong negative

potential is still present after the sap flow has receded,

around 10 p.m. (Fig. 12); it is at this very moment that the

curve of the time-derivative of the potential reaches its

maximum value. Similarly, in the morning, during the 2-h

time interval taken by the sap flux to reach its maximum,

the electric potential hardly changes. These observations

cannot be reconciled with the electrokinetic mechanism in

its simple form.

(5) T

he electric potentials in a root and in the nearby soil are not

equal (Fig. 14), suggesting the presence of an electrical

barrier between the tree root system and the soil.

Nevertheless, the curves of Figs. 11 and 12 evidence a stable

functional relationship between the sap flow and the electric

potential. To simplify the picture, it appears that the sap flow

polarises (electrifies) negatively the whole tree trunk with

respect to the ground; but this polarisation process has a long

time constant, on the order of hours; when the sap flow

vanishes, in the evening, it is with this long time constant that

the polarisation also goes to zero. Another remarkable fact is

that the rapid onset of the sap flux in the morning generates a

specific transient variation on the electric potential, particularly

outstanding in its time-derivative (Fig. 12).

14. Example of diurnal potential variation measured on a chemical electrode (

s used for both potential curves (left scale for E01 and right scale for E102).

A well known phenomenon in geophysical prospecting,

although still poorly understood, is the so-called induced

polarisation (e.g., [37,38]). Injecting electric current into the

ground through two electrodes generates an electric field in

their neighbourhood; when switching off the current source, the

potential goes back to zero with a time constant which can

sometimes reach several seconds, depending on the nature of

the ground and of the electrodes (e.g., [39,37]). Such induced

polarisation effects can be particularly strong in the presence of

ore deposits, which is the basis of its use as an empirical

prospecting method. This observation reveals the ubiquitous

presence of electromotive forces in the ground after turning off

the external current source. In their absence, the ground would

be back instantaneously equipotential.

However, the time constant involved in the present

experiment is so large that it seems necessary to call for a

mechanism specific to biological living systems. Note that all

the curves corresponding to the different electrodes behave in a

similar fashion with respect to the sap flow curve, while the

amplitudes of this daily variation differ greatly from point to

point. This time constant does not appear specific to a given

neighbourhood but characterises the whole tree.

The mechanism that could be invoked to generate such a

trunk polarisation and time constant remains unclear. As

mentioned in the introduction, the observed electric potential

may reflect a combination of physical, chemical and

physiological responses to the sap flow, photosynthesis and

adaptive feedback controls of the tree. While we are not able to

propose a comprehensive model at this stage, we can sketch the

following scenario.

We have to find a mechanism able to accommodate the

peculiar electrical system observed in the tree trunk. The trunk

is a conductor, but the electrical potential, which depends on the

position vector r and time t, must be able to sustain a temporal

relaxation of the form:

@Vðr; tÞ@t

þ Vðr; tÞt¼ F½r; SðtÞ�; (1)

E102) placed in the soil near the root electrode E01. Note the different vertical

Major tick marks of horizontal axis fall at midday.

D. Gibert et al. / Plant Science 171 (2006) 572–584582

Fig. 15. (a) Main electrical structure of a sectional cross-section of the tree

trunk. Small longitudinal capacitive elements are composed by insulating

xylem elements embedded in the conductive trunk and containing the circulat-

ing and conducting sap (b).

where t is some relaxation time and F is some position-

dependent functional of the sap flow SðtÞ, vanishing when

the sap flow vanishes. As t is of the order of several hours,

some poorly conducting medium must be present within the

conductor, able to generate a significant capacitance. In addi-

tion, we know that F is a negative function of S, basically

constant with height z, but smoothly varying around the cir-

cumference. This imposes a configuration of the source with

translational symmetry along the tree axis, as depicted in

Fig. 15 a. The spatial configuration of the potential results

from leakage of charge accumulated within this supposed non-

perfect isolator. Note that a grounded perfect conductor with an

inner distribution of charge is characterised by constant poten-

tial and spatially varying surface charge distribution. Here, we

have a non-perfect insulator embedded in a non-perfect con-

ductor, which allows to maintain a small volume current.

As the potential varies with azimuth, we have to conclude

that the charge distribution maintained by the insulator is not

cylindrical. This probably requires a non-cylindrical sap flow as

well. In our tree, given the asymmetric distribution of branches

and roots around the axis of the trunk (Fig. 2), it may not be

unexpected. Note that the two measurements of sap flow also

support a non-cylindrical sap velocity. At each electrode, the

potential is the result of the potential distribution due to sum of

all the individual electrical sources, namely the leakage of all

the supposed elementary capacitors (Fig. 15).

Once this electrical mechanism is proposed, the role of the

sap flow can be sketched as follows. Let us consider a xylem

element (Fig. 15b). The conducting sap pervades the trunk and

makes it a conductor; when photosynthesis and transpiration

are active, it is circulating upwards in xylem elements. The

walls of such elements are made of insulating ligneous cell

walls. In the case of a solid rock surface, elementary charges are

adsorbed, creating an excess charge in the electrolyte. It is the

motion of this excess charge that creates the electrokinetic

effect. In our case, however, a different situation happens,

which completely changes the resulting distribution of currents.

Indeed, imagine that the electric charges are not just adsorbed at

the solid surface, but are now allowed to freely diffuse within

the wall. What now dominates the electrical current is not the

circulation along the axis of the trunk but the radial charge

transfer. When the sap flow is suppressed, the charge carriers

slowly diffuse back to the sap electrolyte or alternatively may

get neutralised within the wall.

This situation is analogous to a heat source in an

underground cavity [40]. The temperature in the cavity during

heating is rising linearly after some time, corresponding to a

steady state heat transfer from the cavity to the surrounding

rock. When the heat source is removed, the temperature is

going back to the initial state, but with a dynamics controlled by

the diffusion of heat in the walls. If this analogy is relevant for

the tree potential, then the relaxation should actually not be

exponential but a power law of time. The data in the tree at this

time do not allow to distinguish between the two different types

of relaxation.

To summarise, we propose that the spatio-temporal

characteristics of electrical potential in the tree trunk might

be explained by a biological mechanism allowing the diffusion

of charge carriers (e.g., heavy protein fragments or charge

carrying hormones) across the insulating gel of the xylem walls.

Maybe this charge transport is controlled actively by dedicated

ionic channels in the wall, followed by molecular diffusion

once across the membrane. Such a mechanism then suggests the

presence of electrically active structures within the xylem

channels. Thus, the xylem may play a more active role in the

electrical response than would have been expected from non-

living cells with pure conduction.

5. Conclusion

This experiment confirms the existence and largely extends

the investigation of specific daily variations of the electric

potential distribution in a tree trunk as mentioned in previous

studies [15,26–28]. These daily variations bear a definite

relation to the simultaneously sap flow in summer. The

D. Gibert et al. / Plant Science 171 (2006) 572–584 583

observations of electrical variations in winter would then imply

a sporadically varying sap flow, both in space and time, during

this season. These results, deduced from measurements on a

single tree, should be confirmed by performing analogous

studies on several trees in order to get a statistical significance.

While the order of magnitude of the observed variation is the

same as observed by previous workers, an electrokinetic

mechanism seems to be unable to account for the independence

of the potential with respect to altitude. Instead, we propose a

different mechanism based on charge diffusion from the

conductive sap flow channels into the resistive xylem walls.

This mechanism could be clarified by a dedicated experimental

program, for instance, by performing measurements on small

trees in the laboratory or by using ion-selective microelectrodes

or energy dispersive X-ray microanalysis.

Thus, electrical monitoring of a living tree can reveal both

unexpected patterns of the sap flow and new mechanisms of

charge exchange in xylem elements. These results should raise

a renewed interest in electrical measurements in trees.

Experiments with long-term monitoring using a large number

of distributed electrodes are needed to make progress.

Acknowledgments

The authors thank Frederic Conil for technical assistance and

Pierre Morat for guidance. This work is IPGP contribution no.

2149. We thank M.L. Gicquel for providing us with facilities

during the experiment. This work was financially supported by

the CNRS and ANDRA through the GdR FORPRO and corre-

sponds to the GdR FORPRO contribution number 2006/03 A.

References

[1] S. Lebaude, N. Le Goff, J.-M. Ottorini, A. Granier, Carbon balance and

tree growth in a Fagus sylvatica stand, Ann. For. Sci. 57 (2000) 49–61.

[2] L. Lambs, M. Berthelot, Monitoring of water from the underground of the

tree: first results with a new sap extractor on a riparian woodland, Plant

Soil 241 (2002) 197–207.

[3] L. Lambs, J.-P. Loudes, M. Berthelot, The use of the stable oxygen isotope

(18O) to trace the distribution and uptake of water in riparian woodlands,

Nukleonika 47 (2002) 115–155.

[4] J.D. Johnson, R. Tognetti, P. Paris, Water relations and gas exchange in

poplar and willow under water stress and elevated atmospheric CO2,

Physiol. Plant. 115 (2002) 93–100.

[5] L. Meiresonne, N. Nadezhdin, J. Cermak, J. Van Slycken, R. Ceulemans,

Agric. Forest Meteorol. 96 (1999) 165–179.

[6] A. Granier, P. Biron, D. Lemoine, Water balance, transpiration and canopy

conductance in two beech stands, Agric. Forest Meteorol. 100 (2000) 291–

308.

[7] L. Lambs, E. Muller, Sap flow and water transfer in the Garonne river

riparian woodland, France: first results on poplar and willow, Ann. For.

Sci. 59 (2002) 301–315.

[8] R.H. Swanson, Significant historical developments in thermal methods for

measuring sap flow in trees, Agric. Forest Meteorol. 72 (1994) 113–132.

[9] A. Granier, Mesure du flux de seve brute dans le tronc du Douglas par une

nouvelle methode thermique, Ann. Sci. For. 44 (1987) 1–14.

[10] A. Guevara-Escobar, W.R.N. Edwards, R.H. Morton, P.D. Kemp, A.D.

Mackay, Tree water use and rainfall partitioning in a mature poplar-

pasture system, Tree Physiol. 20 (2000) 97–106.

[11] R. Benyon, Nighttime water use in an irrigated, Eucalyptus grandis

plantation, Tree Physiol. 19 (1999) 853–859.

[12] N. Phillips, R. Oren, R. Zimmermann, Radial patterns of xylem sap flow in

non-, diffuse- and ring-porous tree species, Plant Cell Environ. 19 (1996)

983–990.

[13] S.D. Wullschleger, A.W. King, Radial variation in sap velocity as a

function of stem diameter and sapwood thickness in yellow-poplar trees,

Tree Physiol. 20 (2000) 511–518.

[14] S.S.O. Burgess, M.A. Adams, T.M. Bleby, Measurement of sap flow in

roots of woody plants: a commentary, Tree Physiol. 20 (2000) 909–913.

[15] D.S. Fensom, The bioelectric potentials of plants and their functional

significance: V. Some daily and seasonal changes in the electrical potential

and resistance of living trees, Can. J. Bot. 41 (1963) 831–851.

[16] J.J. Bikerman, Electrokinetic potentials, in: C.A. Hampel (Ed.), The

Encyclopedia of Electrochemistry, Reinhold Publishing Corporation,

New York, 1964, pp. 471–475.

[17] J.L. Thony, P. Morat, G. Vachaud, J.-L. Le Mouel, Field characterization

of the relationship between electrical potential gradients and soil water

flux, C.R. Acad. Sci. Paris 325 (1997) 317–321.

[18] A. Revil, H. Schwaeger, L.M. Cathles, P.D. Manhardt III, Streaming

potential in porous media 2. Theory and application to geothermal

systems, J. Geophys. Res. 104 (1999), 20,033–20,048.

[19] S. Marino, D. Coelho, S. Bekri, P.M. Adler, Electro-osmotic phenomena in

fractures, J. Colloid Interface Sci. 223 (2000) 292–304.

[20] F. Perrier, P. Morat, Characterization of electrical daily variations induced

by capillary flow in the non-saturated zone, Pure Appl. Geophys. 157

(2000) 785–810.

[21] V.A. Bogoslovsky, A.A. Ogilvy, The study of streaming potentials on

fissured media models, Geophys. Prospect. 20 (1972) 109–117.

[22] V.A. Bogoslovsky, A.A. Ogilvy, Natural potential anomalies as a quanti-

tative index of the rate of seepage from water reservoirs, Geophys.

Prospect. 18 (1970) 261–268.

[23] M. Trique, F. Perrier, T. Froidefond, J.-P. Avouac, S. Hautot, Fluid flow

near reservoir lakes inferred from the spatial and temporal analysis of the

electric potential, J. Geophys. Res. 107 (B10) (2002) 2239, 10.1029/

2001JB000482.

[24] J. Zlotnicki, J.L. Le Mouel, Possible electrokinetic origin of large mag-

netic variations at La Fournaise Volcano, Nature 343 (1990) 633–636.

[25] M. Schuch, R. Wanke, Die zeitlichen Variationen der electrischen Stro-

mungs-spannung in einem Fichtenstamm, verursacht durch die tagliche

anderung des Saftstromes, Oecol. Planta 3 (1968) 169–176.

[26] P. Morat, J.-L. Le Mouel, A. Granier, Electrical potential on a tree. A

measurement of the sap flow? C.R. Acad. Sci. Paris 317 (1994) 98–101.

[27] A. Koppan, L. Szarka, V. Wesztergom, Annual fluctuation in amplitudes of

daily variations of electrical signals measured in the trunk of a standing

tree, C.R. Acad. Sci. Paris 323 (2000) 559–563.

[28] A. Koppan, A. Fenyvesi, L. Szarka, V. Wesztergom, Measurement of electric

potential difference on trees, Acta Biol. Szegediensis 46 (2002) 37–38.

[29] W. Gindl, H.G. Loppert, R. Wimmer, Relationship between streaming

potential and sap velocity in Salix Alba L, Phyton 39 (1999) 217–224.

[30] B. Lorne, F. Perrier, J.-P. Avouac, Streaming potential measurements, I:

Properties of the electrical double layer from crushed rock samples, J.

Geophys. Res. 104 (B8) (1999), 17, 857–17, 877.

[31] J. Fromm, M. Hajirezaei, I. Wilke, The biochemical response of electrical

signaling in the reproductive system of Hibiscus plants, Plant Physiol. 109

(1995) 375–384.

[32] J. Fromm, H. Fei, Electrical signaling and gas exchange in maize plants of

drying soil, Plant Sci. 132 (1998) 203–213.

[33] J. Fromm, W. Eschrich, Electric signals released from roots of willow

(Salix viminalis L.) change transpiration and photosynthesis, J. Plant

Physiol. 141 (1993) 673–680.

[34] C. Koziolek, T.E.E. Grals, U. Schreiber, R. Matyssek, J. Fromm, Transient

knockout of photosynthesis mediated by electrical signals, New Phytol.

161 (2003) 715–722.

[35] S. Schrader, J.J. Sauter, Seasonal changes of sucrose-phosphate synthase

and sucrose synthase activities in poplar wood (populus � canadensis

Moench ‘‘robusta’’) and their possible role in carbohydrate metabolism, J.

Plant Physiol. 159 (2002) 833–843.

[36] G. Petiau, Second generation of lead-lead chloride electrodes for geo-

physical applications, Pure Appl. Geophys. 157 (2000) 357–382.

D. Gibert et al. / Plant Science 171 (2006) 572–584584

[37] D.J. Marshall, T.K. Madden, Induced polarization, a study of its causes,

Geophysics 24 (1959) 790–816.

[38] K.A. Titov, A. Kemna, H. Tarasov, Veresscken, Induced polarization of

unsaturated sands determined through time domain measurements,

Vadose Zone J. 3 (2004) 1160–1168.

[39] C. Schlumberger, Etude sur la prospection electrique du sous-sol, Gau-

thier-Villar, Paris, 1920.

[40] C. Crouzeix, J.-L. Le Mouel, F. Perrier, P. Richon, P. Morat, Long term

thermal evolution and effect of low power heating in an underground

quarry, C.R. Acad. Sci. Geosci. Paris 335 (2003) 345–354.