Ann. For. Sci.

57 (2000) 755–765 755

© INRA, EDP Sciences

Original article

A generic model of forest canopy conductance

dependent on climate, soil water availability
and leaf area index

André Graniera,*, Denis Loustaub and Nathalie Brédaa

a Institut National de la Recherche Agronomique, Unité d'Écophysiologie Forestière, 54280 Champenoux, France
b Institut National de la Recherche Agronomique, Unité de Recherches Forestières, BP 45, 33611 Gazinet Cedex, France

(Received 2 June 2000; accepted 3 October 2000)

Abstract – This paper analyses the variation in tree canopy conductance for water vapour (gc) in order to derive a general expression,
including the effects of solar radiation (R), vapour pressure deficit (D), leaf area index (LAI) and extractable soil water. Canopy con-
ductance was calculated from transpiration measured in 21 broadleaved and coniferous forest stands, under different climates: tem-
perate, mountain, tropical and boreal. Common features in the dependence of gc on climate and on soil water content were exhibited.
When soil water was not limiting, gc was shown to increase linearly with LAI in the range 0 to 6 m2 m–2 and reach a plateau value.
Besides the positive effect of increasing R and the negative effect of increasing D on gc, it was surprisingly shown that a decrease in
extractable soil water induced a similar reduction in gc in various tree species, equally in coniferous and in broadleaved. Based on these
findings, a general canopy conductance function is proposed.

canopy conductance / sap flow / transpiration / species comparison / leaf area index / water stress / model / synthesis

Résumé – Un modèle générique de conductance de couverts forestiers dépendant du climat, de la disponibilité en eau dans le
sol et de l’indice foliaire. Ce travail réalise l'analyse des facteurs de variation de la conductance du couvert pour la vapeur d'eau (gc)
avec l'objectif d'en donner une expression générale, prenant en compte les effets du rayonnement global (R), du déficit de saturation
de l'air (D), de l'indice foliaire (LAI) et de la réserve hydrique extractible du sol. La conductance du couvert a été calculée à partir de
la transpiration mesurée dans 21 peuplements forestiers feuillus et résineux, sous différents types climatiques : tempéré, montagnard,
tropical et boréal. Ce travail a montré, pour ces divers peuplements, une dépendance similaire entre gc et les facteurs climatiques, ainsi
qu'avec la réserve hydrique extractible du sol (REW). En conditions hydriques non limitantes, on observe que gc augmente linéaire-
ment avec le LAI entre 0 et 6 m2 m–2, puis atteint un plateau. De façon surprenante, en dehors de l'effet positif sur gc de l'augmenta-
tion de R, et l'effet négatif de celle de D, on montre que la diminution de REW a des conséquences similaires sur gc pour diverses
espèces forestières, aussi bien feuillues que résineuses. À partir de ces observations, un modèle général de conductance de couvert est
proposé ici.

conductance de couvert / flux de sève / transpiration / comparaison inter spécifique / indice foliaire / sécheresse / modèle /
synthèse

* Correspondence and reprints

Tél. (33) 03 83 39 40 38 ; Fax. (33) 03 83 39 40 69 ; e-mail: [email protected]
756 A. Granier et al.

1. INTRODUCTION whole canopy acts as a single layer for water exchange to

the atmosphere, even if it has been demonstrated that
During the last decades, a large number of studies have multilayer models are more suitable for detailed physio-
been conducted, quantifying forest transpiration and its logical functioning of the forest canopy [39].
spatial and temporal variation, under various stand condi- The objectives of this paper are to: 1) compare canopy
tions (age, species, site, climate), involving different conductance among a large range of forest stands, differ-
techniques. High time scale resolution (hour) data can be ing in species composition or in climatic and soil charac-
obtained through sap flow measurements [28], which teristics; 2) evaluate the effect of leaf area index as a
have few requirements in term of fetch and stand topog- possible source of variation in transpiration; 3) build a
raphy as compared with the common meteorological generic model of forest stand transpiration independent of
methods. Sap flow has been shown to measure accurate- tree species.
ly stand transpiration [9, 10, 28], providing an adequate
sampling of sap flux accounting for variation in size, tree
representativeness, species and age can be performed. 2. METHODS
Thus, sap flow is scaled most usually from individual
trees to the stand, using a scaling variable, that can be tree 2.1. Sites
circumference, sapwood area or leaf area [28].
Site characteristics and tree species used in the analy-
When analysing stand transpiration, large temporal
sis are listed in table I. This data set covers a wide range
and spatial variation is generally observed. The first
of tree species, coniferous and broadleaved, under vari-
source of variation is due to climate because available
ous climate and site conditions, temperate, tropical and
energy and atmospheric deficit in vapour pressure drive
boreal. In some stands, measurements were performed
the transpiration flux from vegetation to the atmosphere.
during several years, allowing us to take into account the
The second source is the biological regulation exerted
inter-annual variation of climate (table I).
through canopy surface conductance, which is controlled
mainly by stand LAI, and stomatal conductance. In addi- In some of these experiments, soil water content in the
tion, atmospheric turbulence and stand structure deter- root zone was measured and data were converted to rela-
mines the aerodynamic transfer between the canopy and tive extractable water (REW, dimensionless), defined as:
the atmosphere. However, it is widely recognized that the
stand structure has a weak influence on variation in forest W – Wm
REW = (1)
transpiration as compared to climatic factors and surface WFC – Wm
(or canopy) conductance. Forests are found over a wide
range of climates and differ in many characteristics rele- where W is the soil water content in the root zone, Wm is
vant to stand transpiration and canopy conductance, e.g. the minimum soil water (i.e. lower limit of water avail-
their phenology, leaf life span, drought response (avoid- ability), WFC is the soil water content at field capacity.
ance vs. tolerance), canopy structure, etc. Whether some
common pattern in canopy conductance emerge across
forests is a challenging question since forest ecosystems 2.2. Calculation of canopy conductance
must also satisfy common ecological constraints such as
water conservation or xylem cavitation risk [49]. The aim Canopy conductance for water vapour (gc, m s–1) was
was here to analyse the different sources of variation in calculated from transpiration measurements and from cli-
canopy conductance between forest stands covering a mate data using the rearranged Penman Monteith equa-
wide range conditions, using a simple multivariate model, tion (see [18]):
and try to separate the influence of climate from the
intrinsic characteristics of stand. ga E λ γ
gc = (2)
Different approaches have been developed to model s A + ρ cp D g a – λ T s + γ
transpiration of forest stands. The most mechanistic mod-
els of canopy transpiration are multilayered [25]. They where E (kg m–2 s–1) is the stand transpiration, λ (J kg–1)
describe the canopy transpiration within horizontal ele- is the latent heat of water vaporisation, γ (Pa K–1) is the
mentary layers. The multilayered models must be used in psychometric constant, s (Pa K–1) is the rate of change of
the case of a two-layer vegetation as for instance to saturating vapour pressure with temperature, A (W m–2) is
describe the functioning of an overstory-understory asso- the available energy of the forest canopy, ρ (kg m–3) is the
ciation [25]. Since the work of Jarvis and Mc Naughton density of dry air, cp (J K–1 kg–1) is the specific heat of air,
(1976, [23]), many authors made the assumption that the D (Pa) is the vapour pressure deficit, and ga (m s–1) is the
 A model of forest canopy conductance 757

Table I. Main characteristics of the sites. Methods used for fluxes measurements are sap flow (SF), eddy covariance (EC) or energy
balance (EB).

Species Site Age Height Temp Rain LAI m2 Method Project / reference /
(yr) (m) (°C) (mm) (m–2) SF/EC remarks

Quercus petraea Champenoux (France) 35 15 9.6 740 6.0 SF control [2, 3]

Q. petraea Champenoux (France) 35 15 9.6 740 3.3 SF thinned [2, 3]
Q. rubra Ede (The Netherlands) 17.4 4.9 EB [38]
Fagus sylvatica Hesse (France) 30 14 9.2 820 5.7 SF/EC EUROFLUX
F. sylvatica Aubure (France) 120 22.5 6.0 1 500 5.7 SF REKLIP
F. sylvatica Kiel (Germany) 100 29 8.1 697 4.5 EB [19]
Abies bornmulleriana Champenoux (France) 25 11 9.6 740 8.9 SF plantation
Picea abies Champenoux (France) 21 11 9.6 740 9.5 SF plantation
P. abies Aubure (France) 30 13 6.0 1 500 6.1 SF REKLIP
Pinus sylvestris Hartheim (Germany) 35 12 9.8 667 2.9 SF/EC HartX [27]
Pinus pinaster Losse (France) 37 20.3 13.5 900 2.5 SF/EC HAPEX-MOBILHY [14]
P. pinaster Le Bray (France) 18 12 13.5 900 2.7 SF EUROFLUX
Tropical rainforest Paracou (French Guiana) 33 25.8 2 900 8.6 SF natural forest [16]
Simarouba amara Paracou (French Guiana) 5 4.7 25.8 2 900 3.5 SF plantation [17]
Goupia glabra Paracou (French Guiana) 11 15 25.8 2 900 4.3 SF plantation [16]
Eperua falcata Paracou (French Guiana) 11 10 25.8 2 900 10.8 SF plantation
Pinus banksiana Old Jack Pine (SA, Canada) 75-90 12.7 0.1 390 2.2 SF/EC BOREAS [44]

aerodynamic conductance. We calculated ga from Thom's Typically, discarded data correspond to early morning
[48] equation. In closed stands, available energy was and late afternoon periods. Furthermore, when D is low
assumed to be equal to the net radiation measured over during the early morning, dew is quite likely to occur and
the canopy, minus heat storage in the air and in the above affects tree transpiration and its measurement.
ground biomass. In open stands (e.g. LAI < 3), where a
significant fraction of the radiative flux reaches the soil Excluding these data has only limited consequences on
surface, heat flux in the soil should not be neglected. calibrating the gc functions, because they represent peri-
Nevertheless, in the absence of soil heat flux measure- ods of low transpiration rates. Modelling stand transpira-
ment in most of the studied stands, this term was not tion under conditions of maximum transpiration rates, i.e.
taken into account here. However, when LAI < 3.0 and when both D and gc are high (and therefore the product
canopies did not occupy the entire ground area, canopies gc.D is high), is more crucial.
likely did not absorb all the net radiation and actual tree
canopy conductance would be underestimated. A time lag between sapflow and canopy transpiration
has been often reported, even when the vapour flux above
In some experiments, E was directly measured above a stand was directly measured [11] or when it was esti-
the stand (Bowen ratio or eddy covariance technique), mated by a model [5, 15]. This phenomenon is due to
while in other studies transpiration was estimated from
water exchanges between tissues and the transpiration
sapflow measurements. In most of our experiments pre-
stream within the trees [23]. This capacitance effect was
sented here, the continuous heating technique was used
often reported in coniferous species [18, 22, 30, 31, 45],
[8], performed on 5 to 10 trees according to stand hetero-
the time lag being typically in the range of 1 to 2 h, while
geneity [28]. For computing gc from transpiration and cli-
it is much less important in broadleaved species (30 min
matic variables, some precautions were taken:
in oak, 60 min in poplar [15, 21]). Water exchanges can
• periods during rainfall and for the 2 hours following be described with RC-analogue models [20, 31]. For an
rainfall were excluded in order to avoid the discrepan- accurate calculation of canopy conductance, it is there-
cy between evaporation and tree transpiration, fore necessary to take into account this time lag in order
to improve the synchronism between sapflow and climat-
• when either global radiation, vapour pressure deficit, ic demand. When this time lag is not taken into account,
or stand transpiration were too low (< 5% of the max- this would change the relationship between calculated gc
imum value), data were also eliminated, because of the and the climatic variables changes (e.g., figure 1).
large relative uncertainties in computing gc from equa- Furthermore, excluding the time lag results in an increase
tion 2 under these conditions. of the scatter of data: in this example, correlation coeffi-
758 A. Granier et al.

Figure 1. Effect of accounting

for the time lag between sapflow
and vapour pressure deficit (D)
on the estimate of canopy con-
ductance in Pinus pinaster.

cients equalled to 0.32 with no time lag, vs. 0.67 with a for water stress is either soil water deficit or leaf water
1 h time lag. potential (see Sect. 3.3 below).
Validation can be performed in several ways: parame-
terise canopy conductance function parameters from one
2.3. The canopy conductance sub-model
year's data set, and compare estimated to measured gc and
transpiration for other years [47], compare model para-
Jarvis and Steward [23, 47] proposed a multiplicative-
meters obtained on even days to those on odd days with-
type function to relate the variation of gc to the environ-
in the same set of data [7], compare measured to comput-
mental factors. This approach is now widely used [6, 7,
ed stomatal conductances, derived from calculated
12, 15, 18, 38]. The following model, derived from Jarvis
canopy conductance and from LAI [18].
and Steward [23, 47] was used here:
In order to check if the response of one tree species
gc = gcmax ⋅ f1(R,D) ⋅ f2(LAI) ⋅ f3(Is) ⋅ f4(t) (3) could be extrapolated to other site and climate conditions,
Granier et al. [13] compared measured tree transpiration
where gcmax (m s–1) is the maximum gc, reduced by the
in an old mountain beech forest (Aubure forest) to tran-
following functions fi varying between 0 and 1 of: both
spiration estimated from canopy conductance which was
global radiation (R) and air vapour pressure deficit (D)
calibrated in another beech stand growing under plain
measured above the stand; leaf area index (LAI); a vari-
conditions (Hesse forest, see table I).
able quantifying water stress intensity (Is); air tempera-
ture (t). No interaction between the variables was Equation 3 was parameterised for each stand. First,
assumed here. According to the studies, the variable used coefficients of f1(R,D) were fitted under non-limiting
 A model of forest canopy conductance 759

temperature and soil water, in stands with high LAI (>6). Model 2: (5)
R 1
Then, each other fi function was separately parame- g c = g cmax ⋅
terised. R + R0 1 + b ⋅ D
In order to compare the stands, we calculated a stan-
dardised canopy conductance (gc*), corresponding to the Fitting of the parameters in equations (4) and (5) (and in
following set of variables: global radiation = 500 W m–2, the further functions) was based on the minimum sum of
D = 1 kPa, Relative Extractable Water = 1, and no limiting squares using the Gauss-Marquardt algorithm. In contrast
air temperature (i.e. in the range 18–30 °C). to stomatal conductance, those functions do not show a
saturation at high values of R. The parameter R0 varies
according to the species between 50 and 300 W m–2, with-
3. RESULTS out any clear relation to leaf area index. Nevertheless, the
highest R0 coefficients are found in the coniferous stands.
3.1. Effects of radiation, vpd and temperature Figure 2 shows a large scattering of gc within the low-
est radiation class (0 to 200 W m–2). This scatter is the
An example of the variation of canopy conductance in result of both the rapid increase of gc with R, but also to
beech (Fagus sylvatica) as a function of global radiation the large uncertainty in calculating canopy conductance
and vapour pressure deficit is shown in figure 2. As for at low values of transpiration, such as during early morn-
stomatal conductance, canopy conductance increases ing or late afternoon.
when incident radiation increases, and decreases when
vapour pressure deficit increases. We used Lohammar- Parameterisation of gc needs to take into account, if
type equations for describing the combined effects of possible, the effect of water exchange between tissues
both variables, expressed as follow: and sap flow, provoking a time lag between transpiration
and sap flow. The procedure to test this capacitance effect
was the following: we introduced increasing time lags (0,
Model 1: R (4) 0.5, 1.0, 1.5 and 2.0 h) in the calculation of gc, sapflow
g c = g cmax a – b ln D
R + R0 lagging behind climatic variables. At each step, the func-
tion f1 was fitted, and the regression coefficients were

Figure 2. Canopy conductance (gc) in a

beech forest (Fagus sylvatica) calculat-
ed from sapflow measurements as a
function of vapour pressure deficit (D).
Data are sorted according to radiation.
Euroflux experiment, Hesse forest 1998
(France).
760 A. Granier et al.

compared. The time lag was assumed to correspond to the the narrow range of temperatures, because most of the
highest r2 obtained. We checked if this procedure was observations were performed during summer.
correct by comparing this estimated time lag to the
observed time lag between water flux measured above the
stand and scaled up sap flow in a Scots pine forest [11]; 3.2. LAI
the same value was obtained, equal to 90 min. For our
sample species (table I), it varied between 0 and 1.5 h, Figure 4 shows the relationship between standardised
depending on tree species. We found that water stress canopy conductance gc* and LAI in 20 stands. For LAI <
increased the time lag in some tree species like Pinus 6, gc* linearly increased to a value of 1.33 cm s–1. With
pinaster or Picea abies (data not shown). In experiments LAI larger than 6.0, canopy conductance did not
where water supply varied during the season, we there- increased further.
fore applied this procedure to each soil water content
class. The following function was fitted on this data set:
Because radiation and vapour pressure deficit are cor- LAI ≥ 6 f1(LAI) = 1 [7]
related (r2 ranging from 0.2 to 0.4), the coefficients R0, a, LAI < 6 f1(LAI) = LAI / 6 .
and b are also correlated.
The variation of canopy conductance vs. D, under high
global radiation, R = 700 W m–2 (figure 3), showed a sim- 3.3. Water stress
ilar pattern in all studied stands. The negative effect of
increasing D on gc was accurately modelled with func- Many studies have demonstrated the negative effect of
tions 4 or 5. Coefficients of determination for models 1 soil water depletion on canopy conductance. Variation of
and 2 were in general close, but model 2 often gave gc can be related either to predawn water potential as in
slightly better fits than model 1. Besides this common [32], to soil water reserve or soil water deficit [18], or to
feature, some of the studied species were found to be relative extractable water in the soil (REW) as in [15]. We
more sensitive to D. Two examples are Quercus petraea, preferred to use the latter variable for extensive studies
for both the control and thinned stands, and Simarouba and for modelling purposes, because:
amara (tropical). In other tree species (Abies bornmulle- – predawn water potential, even if it a physiological
riana, temperate, and Eperua falcata, tropical), sensitivi- indicator of tree water status, and therefore has a more
ty of gc to D was lower than the average response. causal significance, is not often available in field stud-
According to the tree species, the relative variation of gc, ies;
when D passed from 1 to 2 kPa, ranged from –20% to
–60%. As reported by Oren et al. [37], gc sensitivity to D – soil water reserve is very site dependent, ranging from
is well correlated with gcmax. Fitting the coefficient b to a ca. 50 to 200 mm, according to rooting depth, soil
of equation (4) gave: b = 0.253 a (r2 = 0.92, see insert of properties, etc., while REW is varying between 0 and
figure 3). 1, whatever the site;
Absolute values of gc differed markedly among the – both predawn water potential and REW are strongly
stands. Canopy conductance appears to be higher in sites related [4].
where LAI is high (upper curves with closed symbols in Figure 5 illustrates the relationship between gc and REW
figure 3, LAI being in the range of 5.7 to 10.8), than in in five coniferous and broadleaved stands. For all these
low LAI stands. species, gc/gcmax progressively decreases when REW
varies from 1 to 0, this decrease being more pronounced
When pooling all the stands where LAI > 5.7, the fol- when REW drops below 0.4, as previously reported [12].
lowing function was obtained: When pooling all the data, the following relationship was
obtained:
R 1
g c = 4.047 (r2 = 0.76). (6) 1/2
R + 100 1 + 2.0615 D 2
p 1 + p 2 ⋅ REW – p 1 + p 2 ⋅ REW – 2.8 p 1 ⋅ p 2 ⋅ REW
In most of the data sets that we used here, when the f2 Is =
1.4
response of gc to both R and D was extracted, no signifi-
cant relationship between gc residuals and air temperature (r2 = 0.77) [8]
was pointed out. This probably results from: i) the high
correlation between air temperature and D (r2 > 0.5), ii) in which p1 = 1.154 and p2 = 3.0195.
 A model of forest canopy conductance 761

Figure 3. Canopy conductance of various forest stands as a function of vapour pressure deficit, for a global radiation of 700 W m–2,
under non-limiting soil water. Closed symbols correspond to stands with a high LAI (≥5.7), open symbols or lines are for stands with
a lower LAI (<5.7). The value of LAI is indicated in the legend. For Pinus pinaster + understorey: data of [7]. Insert, the relationship
between the coefficients a and b of the model 3 (see text).
762 A. Granier et al.

Figure 4. Standardised canopy con-

ductance gc* (R = 500 W m–2, D = 1
kPa) as a function of LAI in 20 forest
stands. Same data as for figure 3. Other
values are coming from [19] and [38].
Data in the dotted circle are for the 3
pine stands (Pinus pinaster and P.
sylvestris).

4. DISCUSSION the intercept (= coefficient a) was found to be similar

between the forest stands reported here. A few exceptions
In contrast to grasslands, gc generally controls forest were noted. Two species demonstrated a slightly higher
transpiration [26] because it is at least one order of mag- sensitivity to atmospheric drought i.e. Quercus petraea
nitude lower than ga. This is less true in poorly ventilated and Simarouba amara, two light demanding tree species.
canopies such as in tropical rainforests [34, 40], in some Finally, two species showed lower sensitivity, i.e. Abies
dense deciduous plantations [21] or during early morning
bornmulleriana and Eperua falcata, both shade tolerant
hours when windspeed (and therefore ga) is still low [33].
In most of the studies we reported here, the decoupling and high LAI species. The common response of gc to D
coefficient Ω, as defined by McNaughton and Jarvis [36], (in 13 of the 17 species in table I) contrasts strongly with
ranged between 0.1 and 0.2, demonstrating a strong cou- leaf level measurements of stomatal conductance. Larger
pling between the canopies and the atmosphere. Thus, the differential stomatal sensitivity between species to air
simplified model of transpiration proposed by vapour pressure deficit has been often reported, among
McNaughton and Black [35], derived from the Penman- conifer species (e.g. in Sandford and Jarvis [42]). Our
Monteith equation, is applicable in most forest types. In observation probably results from the averaged response
this simplified model, transpiration is proportional to D, of a whole canopy, resulting from the mixing of leaves of
gc and LAI.
different physiological properties (sun vs. shade, leaves
The dependence of gc on D, expressed as the slope of of different ages in coniferous species, etc.), submitted to
gc vs. ln(D) (= coefficient b of equation (4)), relative to differing environmental conditions [29].
 A model of forest canopy conductance 763

Figure 5. Variation of relative canopy

conductance (gc/gcmax), as a function of
relative extractable water in the soil
(REW) in 5 forest stands: oak (Quercus
petraea, LAI = 6.0), beech (Fagus sylvati-
ca, LAI = 5.8), fir (Abies bornmulleriana,
LAI = 8.9), spruce (Picea abies, LAI = 6.1)
and pine (Pinus pinaster, LAI = 2.7). In
oak, beech, spruce and pine, gc is related to
modelled gcmax. In fir, gc is related to gcmax
measured in a well-watered plot. A unique
relationship was drawn.

The effect of air temperature on gc, although being less growing conditions (plain vs. mountain). Moreover, in
investigated, seems to play an important role in the regu- this work, a comparison with the data from Herbst [19] on
lation of stomatal and hence canopy conductance. In the same species also showed very close gc function.
Scots pine, Gash et al. [7] calibrated a parabolic function These 3 stands were characterised by similar values of
with an optimum between 15 and 20 °C. In beech, Granier LAI (5.5 to 6.0).
et al. [13] found in spring a decrease in gc when air tem- Canopy conductance is nearly proportional to LAI
perature dropped below 15 °C. On the opposite, no tem- between 0 to 6, as previously shown by Granier and
perature effect was detected for oaks, neither in spring Bréda [15], in which different temperate oak stands were
nor in summer. Our attempts to derive the function f3 in compared. Similar results have been noted within the
equation (3) were not successful, and there are not same stand during leaf expansion [15]. Compared to
enough data yet available to derive a general relationship. forests, low vegetation like crops and grasslands, exhibit
Probably, different species could show a different sensi- a different response to increasing LAI, with gc and tran-
tivity to temperature and different optima, tropical spiration saturating at a much lower LAI threshold (about
species probably being more sensitive to temperature 3 to 4) [43]. The saturation of forest transpiration at LAI
than temperate and boreal species. Furthermore, Gash et higher than 6.0 can be explained by the important shad-
al. [7] calibrated different functions relating the depen- ing of low canopy strata by the upper levels when LAI
dence of gc to temperature in a same tree species (Pinus increases. For LAIs less than 6, leaf area index is therefore
pinaster) growing in two sites. a key factor for explaining between-stand variation in
A close similarity in transpiration of different forests transpiration. Nevertheless, two tree species, Pinus
was also reported by Granier et al. [13] in two beech pinaster and P. Sylvestris (figure 4, dotted circle), were
stands, differing in both age (30 vs. 120 years old), and distinguished from the average gc*(LAI) relationship,
764 A. Granier et al.

probably due to their clumped crown structure and, there- [4] Bréda N., Granier A., Barataud F., Moyne C., Soil water
fore, to their different radiation absorbing properties. dynamics in an oak stand. I. Soil moisture, water potentials and
Similarity in response of various forest types to climate water uptake by roots, Plant and Soil 172 (1995) 17–27.
has been previously highlighted by Shuttleworth [46] [5] Cienciala E., Lindroth A., Cermak J., Hallgren J.E.,
who compared time courses of canopy conductance of Kucera J., Assessment of transpiration estimates for Picea abies
various temperate and tropical forests (see his figure 10, trees during a growing season, Trees - Structure and Function 6
p. 146). He found an average value of 1 cm s–1 for most (1992) 121–127.
species. Under similar high radiation conditions, this cor- [6] Dolman A.J., van Den Burg G.J., Stomatal behaviour in
responds to the value of gc that was observed here when an oak canopy, Agric. For. Meteorol. 43 (1988) 99–108.
D equals about 1.5 kPa in forest stands with high LAI [7] Gash J.H.C., Shuttleworth W.J., Lloyd C.R., André J.C.,
(≥ 6). Goutorbe J.P., Gelpe J., Micrometeorological measurements in
Les Landes forest during HAPEX-MOBILHY, Agric. For.
The effect of soil water deficit on gc was rather sur- Meteorol. 43 (1989) 131–147.
prising. A very similar response was noted in five very [8] Granier A., Une nouvelle méthode pour la mesure du flux
different species (figure 5). For instance, Pinus pinaster de sève brute dans le tronc des arbres, Ann. Sci. For. 42 (1985)
is a drought avoider [1], whereas Quercus petraea is a 193–200.
drought tolerater [2]. The threshold 0.4 for REW, beyond [9] Granier A., Evaluation of transpiration in a Douglas-fir
which canopy conductance is linearly reduced, was pre- stand by means of sap flow measurements, Tree Physiol. 3
viously reported in a large spectrum of tree species and (1987) 309–320.
soil types [12]. [10] Granier A., Biron P., Bréda N., Pontailler J.-Y., Saugier
B., Transpiration of trees and forest stands: short and long-term
In conclusion, this work demonstrated that a generic monitoring using sapflow methods, Global Change Biology
model of canopy conductance could be proposed, as (1996) 265–274.
much for broadleaved as coniferous forest stands, even if [11] Granier A., Biron P., Köstner B., Gay L.W., Najjar G.,
physiological differences are often observed at the leaf Comparisons of xylem sap flow and water vapour flux at the
level. This probably results from the canopy approach stand level and derivation of canopy conductance for Scots pine,
that buffers the response of individual leaves forming the Theor. Appl. Climat. 53 (1996) 115–122.
canopy. For instance in the Amazonian forest, the canopy [12] Granier A., Bréda N., Biron, P., Villette S., A lumped
layers behave differentially [40, 41], the lower layers water balance model to evaluate duration and intensity of
being less ventilated and therefore less coupled to the drought constraints in forest stands, Ecol. Modelling 116 (1999)
atmosphere than the upper levels. Nevertheless, the 269–283.
whole canopy response to both R and D is not very dif- [13] Granier A., Biron P., Lemoine D., Water balance, tran-
ferent from that of any other canopies [46]. spiration and canopy conductance in two beech stands, Agric.
For. Meteorol. 100 (2000) 291–308.
We also showed that tree transpiration in open stands [14] Granier A., Bobay V., Gash J.H.C., Gelpe J., Saugier B.,
is reduced when decreasing LAI. Nevertheless, the total Shuttleworth W.J., Vapour flux density and transpiration rate
evapotranspiration is not proportionally reduced, since comparisons in a stand of Maritime Pine (Pinus pinaster Ait.) in
stand opening increases the available energy reaching the Les Landes forest, Agric. For. Meteorol. 51 (1990) 309–319.
understorey vegetation and therefore increases its transpi- [15] Granier A., Bréda N., Modelling canopy conductance
ration rate. and stand transpiration of an oak forest from sap flow measure-
ments, Ann. Sci. For. 53 (1996) 537–546.
[16] Granier A., Huc R., Barigah S.T., Transpiration of nat-
ural rain forest and its dependence on climatic factors, Agric.
REFERENCES
For. Meteorol. 78 (1996) 19–29.
[17] Granier A., Huc R., Colin F., Transpiration and stomatal
[1] Aussenac G., Granier A., Quelques résultats de cinétique conductance of two rain forest species growing in plantations
journalière du potentiel de sève chez les arbres forestiers, Ann. (Simarouba amara and Goupia glabra) in French Guyana, Ann.
Sci. For. 35 (1978) 19–32. Sci. For. 49 (1992) 17–24
[2] Bréda N., Cochard H., Dreyer E., Granier A., Water [18] Granier A., Loustau D., Measuring and modelling the
transfer in a mature oak stand (Quercus petraea): seasonal evo- transpiration of a maritime pine canopy from sap-flow data,
lution and effects of a severe drought, Can. J. For. Res. 23 Agric. For. Meteorol. 71 (1994) 61–81.
(1993) 1136–1143. [19] Herbst M., Stomatal behaviour in a beech canopy: An
[3] Bréda N., Cochard H., Dreyer E., Granier A., Field com- analysis of Bowen ratio measurements compared with porome-
parison of transpiration, stomatal conductance and vulnerability ter data, Plant Cell Environ. 18 (1995) 1010–1018.
to cavitation of Quercus petraea and Quercus robur under [20] Herzog K.M., Häsler R., Thum R., Diurnal changes in
water stress, Ann. Sci. For. 50 (1993) 571–582. the radius of a subalpine Norway spruce stem: their relation to
 A model of forest canopy conductance 765

the sap flow and their use to estimate transpiration, Trees - in a lowland tropical forest tree, Plant Cell Environ. 16 (1993)
Structure and Function, 10 (1995) 94–101. 428–436.
[21] Hinckley T.M., Brooks J.R., Cermark J., Ceulemans R., [35] Naughton K.G., Black T.A., Evapotranspiration from a
Kucera J., Meinzer F.C., Roberts D.A., Water flux in a hybrid forest: A micrometeorological study, Water Resour. Res. 9
poplar stand, Tree Physiol. 14 (1994) 1005–1018. (1973) 1579–1590.
[22] Jarvis P.G., Water Transfer in Plants, in: de Vries D.A. [36] Naughton K.G., Jarvis P.G., Predicting the effects of
and Afgan N.G. (Eds.), Heat and Mass Transfer in the Plant vegetation changes on transpiration and evaporation, in:
Environment, Part 1, Scripta Book Co., Washington D.C., 1975, Kozlowski T.T (Ed.), Water deficits and plant growth, Vol. VII,
pp. 369–374. Academic Press, New York, 1983, pp. 1–47.
[23] Jarvis P.G., The interpretation of the variations in leaf [37] Oren R., Sperry J.S., Katul G.G., Pataki D.E., Ewers
water potential and stomatal conductance found in canopies in B.E., Phillips N., Schäfer K.V.R., Survey and synthesis of intra-
the field. Phil. Trans. R. Soc. Lond. ser. B 273 (1976) 593–610. and interspecific variation in stomatal sensitivity to vapour pres-
[24] Jarvis P.G., McNaughton K.G., Stomatal control of sure deficit, Plant Cell Environ. 22 (1999) 1515–1526.
transpiration: scaling up from leaf to region. Advances in eco- [38] Ogink-Hendriks M.J., Modelling surface conductance
logical Research, Academic Press, London 15 (1986) 1–49. and transpiration of an oak forest in the Netherlands, Agric. For.
[25] Kelliher F.M., Black T.A., Price D.T., Estimating the Meteorol. 74 (1995) 99–118.
effects of undestory removal from a Douglas-fir forest using a [39] Raupach M.R., Finnigan J.J., “Single layer models of
two-layer canopy evapotranspiration model, Water Resour. Res. evaporation from plant canopies are incorrect but useful, where-
22 (1986) 1891–99. as multilayer models are correct but useless”: discuss. Aust. J.
[26] Kelliher F.M., Leuning, R., Schulze, E.D Evaporation Plant Physiol. 15 (1988) 705–16.
and canopy characteristics of coniferous forests and grasslands [40] Roberts J., Cabral O.M.R, Aguiar L. de F., Stomatal and
- Review, Oecologia 95 (1993) 153–163. boundary-layer conductances in Amazonian terra firme rain for-
[27] Köstner B., Biron P., Siegwolf R., Granier A Estimates est, J. Appl. Ecol. 27 (1990) 336–353.
of water vapor flux and canopy conductance of Scots pine at the [41] Roberts J., Cabral O.M.R., Fisch G., Molion L.C.B,
tree level utilizing different xylem sap flow methods, Theor. Moore C.J., Shuttleworth W.J., Transpiration from an
Appl. Climat. 53 (1996) 105–113. Amazonian rainforest calculated from stomatal conductance
[28] Köstner B., Granier A., Cermák J., Sap flow measure- measurements, Agric. For. Meteorol. 65 (1993) 175–196.
ments in forest stands - methods and uncertainties, Ann. Sci. [42] Sandford A.P., Jarvis P.G., Stomatal responses to
For. 55 (1998) 13–27. humidity in selected conifers, Tree Physiol. 2 (1986) 89–103.
[29] Leverenz J.W., Deans J.D, Ford E.D., Jarvis P.G., Milne
[43] Saugier B., The evapotranspiration of grasslands and
R., Whitehead D., Systematic spatial variation of stomatal con-
crops, C. R. Acad. Agric. Fr. 82 (1996) 133–153.
ductance in a Sitka spruce plantation, J. Appl. Ecol. 19 (1982)
835–851. [44] Saugier B., Granier A., Pontailler J.-Y., Dufrêne E.,
Baldocchi D.D., Transpiration of a boreal pine forest measured
[30] Loustau D., Berbigier P., Roumagnac P., Ferreira M.I.,
by branch bags, sapflow and micrometeorological methods,
Pereira J.S., Arruda-Pacheco C., David J.S., Tavares R.,
Tree Physiol. 17 (1996) 511–519.
Transpiration of a 64-year-old maritime pine stand in Portugal.
I: Seasonal course of water flux through maritime pine. [45] Schulze E.D., Cermak J., Matyssek R., Penka M.,
Oecologia 107 (1996) 33–42 Zimmermann R., Vasicek F., Gries W., Kucera J., Canopy tran-
[31] Loustau D., Domec J.C., Bosc A., Interpreting the vari- spiration and water fluxes in the xylem of the trunk of Larix and
ation in xylem sap flux density within the trunk of maritime pine Picea trees: a comparison of xylem flow, porometer and cuvette
(Pinus pinaster Ait.): application of a model for calculating measurements, Oecologia 66 (1985) 475–483.
water flows at tree and stand levels, Ann. Sci. For. 55 (1998) [46] Shuttleworth W.J., Micrometeorology of temperate and
29–46. tropical forest, Phil. Trans. R. Soc. Lond., ser. B 324 (1989)
[32] Lu P., Bréda N., Biron P., Granier A., Water relations of 299-334.
adult Norway spruce trees under soil drought in the Vosges [47] Stewart J.B., Modelling surface conductance of pine
montains: water potential, stomatal conductance and transpira- forest, Agric. For. Meteorol. 43 (1988) 19–35.
tion, Ann. Sci. For. 52 (1995) 117–129. [48] Thom A.S., Momentum, mass and heat exchange of
[33] Martin T.A., Brown K.J., Hinckley T.M., Kucera J., plant communities, in: Monteith J.L. (Ed.), Vegetation and the
Meinzer F.C., Sprugel D.G., Control of transpiration in a 220- atmosphere, Vol. 1, Academic Press, London, 1975.
year-old Abies amabilis forest, For. Ecol. Manage (in press). [49] Whitehead D., Regulation of stomatal conductance and
[34] Meinzer F.C., Goldstein G., Holbrook N.M., Jackson P., transpiration in forest canopies, Tree Physiol. 18 (1998)
Cavelier J., Stomatal and environmental control of transpiration 633–644.