Tree Physiology 28, 959–970
© 2008 Heron Publishing—Victoria, Canada
Comparative measurements of transpiration and canopy conductance
in two mixed deciduous woodlands differing in structure and species
composition
MATHIAS HERBST,1–4 PAUL T. W. ROSIER,2 MICHAEL D. MORECROFT 2 and
DAVID J. GOWING1
1
The Open University, Department of Biological Sciences, Walton Hall, Milton Keynes MK7 6AA, U.K.
2
Centre for Ecology and Hydrology, Crowmarsh Gifford, Wallingford OX10 8BB, U.K.
3
Present address: Department of Geography and Geology, University of Copenhagen, Øster Voldgade 10, DK-1350 Copenhagen K, Denmark
4
Corresponding author ([email protected])
Received June 12, 2007; accepted November 16, 2007; published online April 1, 2008
Summary Transpiration of two heterogeneous broad-leaved
woodlands in southern England was monitored by the sap flux
technique throughout the 2006 growing season. Grimsbury
Wood, which had a leaf area index (LAI) of 3.9, was dominated
by oak (Quercus robur L.) and birch (Betula pubescens L.)
and had a continuous hazel (Corylus avellana L.) understory.
Wytham Woods, which had an LAI of 3.6, was dominated
by ash (Fraxinus excelsior L.) and sycamore (Acer pseudoplatanus L.) and had only a sparse understory. Annual canopy
transpiration was 367 mm for Grimsbury Wood and 397 mm
for Wytham Woods. These values were similar to those for
beech (Fagus sylvatica L.) plantations in the same region, and
differ from one another by less than the typical margin of uncertainty of the sap flux technique. Canopy conductance (gc),
calculated for both woodlands by inverting the PenmanMonteith equation, was related to incoming solar radiation
(RG ) and the vapor pressure deficit (D). The response of gc to RG
was similar for both forests. Both reference conductance (gcref),
defined as gc at D = 1 kPa, and stomatal sensitivity (–m), defined as the slope of the logarithmic response curve of gc to D,
increased during the growing season at Wytham Woods but not
at Grimsbury Wood. The –m/gcref ratio was significantly lower
at Wytham Woods than at Grimsbury Wood and was insufficient to keep the difference between leaf and soil water potentials constant, according to a simple hydraulic model. This
meant that annual water consumption of the two woodlands
was similar despite different regulatory mechanisms and associated short-term variations in canopy transpiration. The
–m/gcref ratio depended on the range of D under which the measurements were made. This was shown to be particularly important for studies conducted under low and narrow ranges
of D.
Keywords: broad-leaved woodland, heterogeneous forest, potential evaporation, sap flux, seasonality, stomatal sensitivity.
Introduction
Changes in woodland distribution, forestry practices and
global climate have created a need for information about forest
water use. A comparative study by Roberts (1983) suggested
that transpiration by temperate forests is relatively uniform.
However, subsequent studies have shown that broad-leaved
woodlands tend to use more available energy for transpiration
than do coniferous stands (Komatsu 2005). Moreover, it is
now known that variation in annual water use by temperate
broad-leaved forests can be large, even within small regions
such as the lowlands of southern England (Roberts and Rosier
2006).
Although such variation in water use can be explained, in
part, by woodland fragmentation (Herbst et al. 2007a), the role
of woodland heterogeneity remains unclear. Micrometeorological methods are unsuitable for the study of water use by
heterogeneous woodlands in which structure and species composition change over short distances. The sap flux method provides an alternative source of information, which can be scaled
up to provide estimates of canopy transpiration thereby allowing calculation of canopy conductance (gc) (Granier et al.
1996). Use of gc is more appropriate than energy-based concepts, such as potential evaporation, to interpret variations in
transpiration from vegetation that is well coupled to the atmosphere (Jarvis and McNaughton 1986, Dolman et al. 2003).
The magnitude of gc and the sensitivity of its response to the
vapor pressure deficit above the forest (D) can vary considerably among plant stands (McNaughton and Jarvis 1991).
To facilitate a comparison of the response of gc to D among
sites, Oren et al. (1999) introduced an empirical description
based on a reference conductance (gcref), i.e., gc at D = 1 kPa. It
can be shown from theoretical considerations that the slope of
the logarithmic decrease of gc with increasing D, i.e., the
stomatal sensitivity to D, is proportional to gcref if the stomatal
sensitivity is such as to keep a constant difference between leaf
960
HERBST, ROSIER, MORECROFT AND GOWING
and soil water potentials, irrespective of changes in D. This
proportionality is influenced by both the range of D and the ratio of boundary layer to stomatal conductance (Oren et al.
1999), but in practice, it is often close to 0.6 (Oren et al. 1999,
Ewers et al. 2005). The precisely defined gcref is better suited
for comparisons among studies than maximum conductance
(Körner 1994), which is difficult to define. Theoretically, it
makes the model of Oren et al. (1999) more widely applicable,
but in practice, it has rarely been used for forest canopies in humid temperate regions.
The first aim of our study was to quantify and compare transpiration and canopy conductance of two heterogeneous
woodlands: a type of woodland that has been neglected in previous hydrological research. Our second aim was to investigate the universal applicability of the model of Oren et al.
(1999), and to develop a more general interpretation of this
concept that takes the observed range of D into account and facilitates comparisons of stomatal sensitivity across sites subject to different meteorological conditions.
Materials and methods
Study sites
Two mixed woodlands in southern England were selected for
study. The locations and some stand characteristics are given
in Table 1. Wytham Woods (hereafter called Wytham) is located near Oxford and is mostly ancient, semi-natural woodland covering 415 ha. Two edges of this woodland have been
the subject of a complementary study (Herbst et al. 2007a).
The study reported here was carried out in the central part of
the woodland where the canopy is dominated by sycamore
(Acer pseudoplatanus L.), ash (Fraxinus excelsior L.) and oak
(Quercus robur L.) trees. The age range of the trees in Wytham
is large, and the understory is heterogeneous and does not form
a continuous layer.
Grimsbury Wood (hereafter called Grimsbury) is located
about 38 km south of Oxford near the town of Newbury.
Grimsbury covers about 350 ha and is made up of a mosaic of
different plantations including both broad-leaved and coniferous stands. The site chosen for this study has a canopy domi-
nated by oak and birch (Betula pubescens L.). The stand has
not been managed for many years and has developed a continuous understory canopy layer consisting mainly of multistemmed hazel (Corylus avellana L.) shrubs. Thus, both
Wytham and Grimsbury are highly heterogeneous in age structure and species composition.
Biometric data
A survey of all stems larger than 2 cm in diameter, was carried
out on 60 × 60 m and 35 × 55 m plots at Wytham and
Grimsbury, respectively. The circumferences of 288 stems at
Wytham and 616 stems at Grimsbury were measured, and
basal area (BA, m2 ha – 1) calculated for each species and for
each canopy layer (Table 2). In addition, total cross-sectional
sapwood area per unit ground area (SWA; cm2 m – 2) for the different groups was estimated on the basis of the stem survey.
Some empirical functions relating SWA to stem diameter at
breast height (DBH; cm) were from earlier studies (Vincke et
al. 2005, Herbst et al. 2007a; see Table 3). However, for birch
and hazel, original data were used. In birch, the boundary between sapwood and heartwood is often visible on cross sections and logs of six felled trees from the same forest were
examined to derive the relationship between SWA and DBH.
For hazel, the stained logs from the calibration experiment
(see below) were used.
Cumulative leaf area index (LAI) was measured about once
a month with an optical analyzer (LAI-2000, Li-Cor). The
above-canopy readings were taken from the top of towers providing access to the canopy, and the below-canopy readings
were taken in a grid consisting of 20 measurement points in
Wytham and 30 in Grimsbury. The readings for the leafless
canopy (1.06 in Wytham and 1.57 in Grimsbury) were subtracted from the other readings. In Grimsbury, half of the measurement points were located in an area where the sensor saw
no evergreen holly trees, and this subset of 15 readings was
used to determine the zero value in winter.
Table 2. Basal area (BA; m2 ha – 1) distribution among the most abundant tree species of the overstory (o/s) and understory (u/s) in Wytham
Woods and Grimsbury Wood. Total BA was 27.00 m2 ha – 1 in Wytham
Woods and 43.46 m2 ha – 1 in Grimsbury Wood.
Species
Table 1. Locations and stand characteristics of the two broad-leaved
woodlands. Tree densities are based on a survey of stems larger than
2 cm in diameter on plots covering 3600 m2 in Wytham Woods and
1925 m2 in Grimsbury Wood.
Wytham Woods
Grimsbury Wood
Location
Elevation (m)
Mean canopy height (m)
Maximum leaf area index
51°47′ N, 1°20′ W
105
21
3.6
51°27′ N, 1°16′ W
115
22
3.9
Tree density (ha –1)
Overstory
Understory
306
494
390
2810
Acer campestre L.
Acer pseudoplatanus
Betula pubescens
Corylus avellana
Crataegus monogyna L.
Fagus sylvatica L.
Fraxinus excelsior
Ilex aquifolium L.
Quercus robur
Salix caprea L.
Other
Total
TREE PHYSIOLOGY VOLUME 28, 2008
Wytham BA
Grimsbury BA
o/s
u/s
o/s
u/s
–
16.22
–
–
–
–
4.66
–
2.62
0.35
–
0.22
1.40
–
0.32
0.60
–
0.57
–
–
–
0.04
–
–
15.16
–
–
–
0.56
–
22.58
–
–
–
–
0.83
3.39
–
0.40
–
0.21
0.22
–
0.11
23.85
3.15
38.30
5.16
TRANSPIRATION AND CANOPY CONDUCTANCE IN WOODLANDS
Table 3. Allometric relationships between cross-sectional sapwood
area (SWA) and stem diameter at breast height (DBH) for the tree
species present at Wytham Woods and Grimsbury Wood.
Species
Equation
Source
Fraxinus excelsior SWA = 2.85DBH
Quercus robur
SWA = 10.7DBH – 97.8
Betula pubescens
Corylus avellana
Herbst et al. (2007a)
SWA = 16.3DBH – 100.9 This study
SWA = 0.526DBH2.092
Other diffuse-porous species
SWA = 0.565DBH2
Vincke et al. (2005)
Microclimate and potential evaporation
Meteorological data for Wytham have been measured routinely at a nearby grassland site (altitude 160 m a.s.l., distance
to nearest trees = 100 m) since 1992 (Morecroft et al. 1998).
The data are recorded as hourly means by an automatic
weather station (AWS) and include incoming solar radiation,
dry and wet bulb temperatures, wind speed and direction, and
gross rainfall. Net radiation (RN ) above the forest was measured with two Q-7 net radiometers (REBS, Seattle, WA)
mounted on a tower. During parts of the investigation, wind
speed was measured above the forest with a sonic anemometer, and a regression with wind speed from the AWS was used
to estimate the wind speed above the forest for the remaining
periods.
Meteorological variables in Grimsbury were measured
above the forest canopy from a tower. The data included incoming solar radiation and air temperature, net radiation, relative humidity, wind speed and rainfall measured with a tipping
bucket gauge installed near the base of the tower and connected to a funnel mounted on top of the tower. Data were recorded by a logger.
Soil water contents from four access tubes in Grimsbury and
six access tubes in Wytham were measured about twice a
month with a soil moisture neutron probe. The access tubes at
Wytham are a few hundred meters from the sap flux site in an
ash-dominated part of the woodland. Readings were taken every 0.1 m from the soil surface down to 1.4 m in Wytham and
every 0.1 m down to 0.6 m, then every 0.2 m down to 2 m and
then every 0.3 m down to 3.8 m in Grimsbury.
Potential evaporation, according to Priestley and Taylor
(1972) (EPT ), was calculated from RN and temperature as:
λEPT = α
s
RN
s+γ
(1)
where is latent heat of vaporization of water (J g – 1), s is the
slope of the curve relating saturated vapor pressure to temperature (kPa K –1), is the psychrometric constant (kPa K – 1) and
α = 1.26 is the Priestley-Taylor coefficient.
Sap flux density and transpiration
Sap flux density (Fd ) of individual trees was measured with
961
Granier type thermal dissipation probe (TDP) sets (Dynamax,
Houston, TX). Each TDP set comprised two metal probes with
a diameter of 1.2 mm and a length of 30 mm (TDP30), except
those used with large sycamore and birch trees which were
80 mm long (TDP80). The probes contained fine thermocouples, and the upper probe included a heating element covering its entire length. The output of the probes, which were installed at a mean height of 1.5 m above the soil surface, was recorded as hourly means from measurements made at 10-s intervals. At both sites, the data were recorded by a data logger
configured to read the signals of 30 TDP probes per site.
A standard calibration is widely used for the TDP method
(Granier 1985) relating Fd (kg m – 2 s – 1) to the difference in
temperature between a pair of probes (∆T, °C):
. K 1. 231
Fd = 0119
(2)
where K is a parameter calculated as:
K=
∆Tm − ∆T
∆T
(3)
and ∆Tm is the value of ∆T when there is no sap flux. This was
determined as the upper envelope of the nocturnal ∆T maxima
over periods of about 10 days (Granier 1987, Oliveras and
Llorens 2001). If part of the probe is out of contact with sapwood, ∆T must be replaced by the actual temperature difference in the sapwood, ∆TSW, which is calculated as:
∆TSW =
∆T − b ∆Tm
a
(4)
where a is the fraction of the probe in contact with sapwood
and b = 1 – a (Clearwater et al. 1999). The average fraction of
probe lengths in our experiment that were in contact with conducting wood, rather than with heartwood or phloem, is given
in Table 4.
The validity of Equations 2–4 was tested for all study species by a calibration procedure with freshly cut logs with installed TDP sensors measuring the volume of stained water
passing through the sapwood (and thereby determining its
cross sectional area and depth) (Herbst et al. 2007b). The
method also determined the fraction of the probe in contact
with hydroactive wood. Equations 2–4 successfully described
the measured Fd in all species except ash. In ash, the TDP30
probes showed an erratic temperature response to sap movement, because the major part of the flux occurred in only a few
small sections along the probe needles representing the vessels
in the early wood of the last four annual rings (Herbst et al.
2007a). Therefore an independent, empirical response curve
was used for the application of TDP30 probes in ash trees:
Fd = 2.023 K 2 + 0.428 K
(5)
which was derived from calibration runs with several logs of
different diameters (Herbst et al. 2007a) and was used in combination with Equation 3.
TREE PHYSIOLOGY ONLINE at http://heronpublishing.com
962
HERBST, ROSIER, MORECROFT AND GOWING
Field measurements were carried out from April 25 to November 7 (days of year (DOY) 115–311), 2006 in Wytham
and from April 27 to November 22 (DOY 117–326), 2006 in
Grimsbury. About once a month, half of the sap flux probes
were moved to new trees. In this way, 8–9 trees in Wytham and
9–10 trees in Grimsbury were equipped with sap flux sensors
at any time. Over the season, 25 trees in Wytham and 29 trees
in Grimsbury, representing the most abundant species, were
investigated (Table 4). To account for possible azimuthal variations in Fd (Leuzinger et al. 2005), the largest trees had four
probes installed around their circumference and the smaller
ones had two or three probes. In the hazel shrubs, one probe
per stem was installed.
For each woodland, the Fd values were averaged by species
and canopy layer and multiplied by SWA of the respective
group to obtain transpiration values in mm. The SWA of all
trees on the surveyed plots was estimated from DBH according to the empirical functions listed in Table 3. Total stand
transpiration was the sum of the transpiration of all groups corrected for the few trees that belonged to species omitted from
the sap flux measurements (Tables 2 and 4) by increasing the
stand totals in proportion to the contribution of these species to
the total BA per canopy layer.
Two types of errors can occur when using the sap flux technique to measure forest canopy transpiration: systematic errors due to the application of TDP probes in heterogeneous
wood (Clearwater et al. 1999) and random errors related to
sampling density and scaling (Granier et al. 2000). Systematic
errors in Fd were avoided through the calibration procedure.
The probability of random errors was estimated from the sample size in relation to the magnitude of tree-to-tree variations
in Fd (Kumagai et al. 2005a) and to the SWA versus DBH relationship (Kumagai et al. 2005b) as follows. Based on the observed azimuthal and tree-to-tree variability in Fd, a potential
error of 50% in Fd per probe was assumed. Thirty probes per
woodland were used at any time, and on four occasions, half of
them were moved to new trees which brought the number of
measurement positions per year and woodland to 75. If the errors per probe are treated as random and summed quadratically, then the potential error in the average Fd per woodland
was 9% at any particular time and 6% for the annual totals. The
allometric functions for calculating SWA were based on data
from at least 30 trees in most of the species, except birch and
hazel (see above). Based on the typical variability in these
allometric relationships (Kumagai et al. 2005b), an uncertainty of 10% for birch and hazel and 4% for all other species,
or a mean of 7%, was assumed. Because Fd and SWA are multiplied, the resulting overall error in stand transpiration was
11% for instantaneous rates and 9% for annual totals. An earlier study based on the same methodology in a beech (Fagus
sylvatica L.) forest in southern England (Roberts et al. 2001,
2005) found a deviation of less than 10% between canopy transpiration rates measured by the sap flux and eddy covariance
techniques.
No sap flux data were obtained in Grimsbury between
DOY 272 and 278 because of equipment failure. This gap was
filled by estimating the transpiration rates from EPT based on
the mean ratio between Fd (per species) and EPT for the 5 days
before and 5 days after the missing period. These estimated
rates were used for the annual totals, but not in the gc analysis.
For the totals, it was further assumed that there was no transpiration before DOY 115 or after DOY 326. This means that any
transpiration from the few evergreen holly shrubs in the understory at Grimsbury during the winter was neglected.
Canopy conductance
Canopy conductance (m s – 1) was calculated from the hourly
canopy transpiration and meteorological variables by rearranging the Penman-Monteith equation (Monteith 1965):
gc =
λ E γ ga
s RN + ρ cp D ga − λ E( s + γ )
(6)
where E is transpiration rate (converted from mm h – 1 to g m – 2
s – 1 by dividing by 3.6), ga is aerodynamic conductance between the forest canopy and the atmosphere at the reference
height (m s – 1), is the density of dry air (g m – 3) and cp is the
specific heat of air (J g – 1 K – 1). We assumed that Fd measured
in the tree trunks lagged one hour behind actual canopy transpiration (Herbst et al. 2007a).
Aerodynamic conductance was calculated from the wind
Table 4. Number (n), species and range in diameter at breast height (DBH) of trees fitted with sap flux sensors at each site. The mean fraction of the
TDP probe area in contact with active sapwood ( fsw ) is also given for each species and sensor type.
Species
Acer campestre
Acer pseudoplatanus
Betula pubescens
Corylus avellana
Crataegus monogyna
Fraxinus excelsior
Ilex aquifolium
Quercus robur
Wytham Woods
Grimsbury Wood
fsw
n
DBH (cm)
n
DBH (cm)
TDP30
TDP80
2
8
–
–
2
7
–
6
12–29
11–58
–
–
11–29
12–33
–
44–83
–
–
9
9
–
–
3
8
–
–
11–46
6–9
–
–
7–18
14–81
0.96
0.90
0.83
0.95
0.81
0.37
0.95
0.70
–
0.95
0.71
–
–
–
–
–
TREE PHYSIOLOGY VOLUME 28, 2008
TRANSPIRATION AND CANOPY CONDUCTANCE IN WOODLANDS
speed above the forest according to Businger (1956):
ga =
κ 2 u( z)
z−d
ln 2 

 z0 
(7)
where κ is von Kármán’s constant (dimensionless), u(z) is
wind speed at reference height (m s –1), z is reference height
(m), d is zero plane displacement height (m) and z0 is roughness length (m). The latter two values were estimated from
canopy height (h) as d = 0.8h and z0 = 0.1h.
For periods of 28 days, which were not identical to the periods during which the sensors remained in one set of trees, all gc
data obtained during hours with incoming solar radiation (RG )
> 400 W m – 2, D > 0.05 kPa, u > 0.5 m s –1 and less than 0.5 mm
cumulative rainfall over the previous 8 h were plotted against
D. The upper envelopes of the data clouds were calculated as
the mean plus one standard deviation of all gc values obtained
over D ranges of 0.2 kPa, starting at 0.5 kPa (Schäfer et al.
2000, Ewers and Oren 2000). The response function fitted to
the upper envelopes was:
gc ( D) = gcref − m ln D
(8)
where gcref is canopy conductance at D = 1 kPa and m is sensitivity of the gc response to D (Oren et al. 1999).
The response of gc to RG was plotted for all hourly datasets
with D between 0.8 and 1.2 kPa and described as:
gc ( D, RG) = gcmin + ( gc ( D) − gcmin )(1 − e− qR G )
(9)
where gc(D) is maximum gc for the chosen D range which was
assumed to equal the gcref of the respective period as fitted by
Equation 8. Minimum conductance in the dark, gcmin, and an
empirical parameter defining the slope (q) was fitted with
curve fitting software (Sigma Plot, SPSS, Chicago, IL). The
responses of gc to D and RG could not be analyzed for the final
part of the growing season because D never exceeded 0.8 kPa
after October 10.
If stomatal regulation is sufficient to keep the difference between leaf and soil water potentials constant, then the ratio
–m/gcref, for a given range of D and a given ratio of ga /gc,
can be predicted from the theoretical relationship between
stomatal conductance, E, water potential and saturation deficit
(Oren et al. 1999).
 λ γ  k  1
   ∆ΨS − L
gsu = 
 cp ρ   A  D
(10)
In contrast to Oren’s original formulation based on leafscale observations and molar units, Equation 10 refers to the
canopy scale and to micrometeorological units. Surface conductance (gsu; m s –1), represents ga and gc in series (gsu = (ga–1 +
gc–1 ) –1), k is hydraulic conductance of the soil-to-leaf pathway,
defined as water flux per time and water potential gradient
963
(g s – 1 MPa – 1), A is surface area (leaf area if applied at the leaf
scale or ground area if applied at the canopy scale; m2 ) and
∆ΨS–L is the difference between soil and leaf water potential
(MPa), assumed to be constant. Equation 10 does not allow for
diurnal changes in trunk water storage or for limitations in the
water supply from the soil to the roots.
If k is water flux (F; g s – 1) per unit of ∆ΨS–L and E is defined
as F per unit surface area, then k/A can also be written as
E/∆ΨS–L and Equation 10 reduces to:
λγ E

gsu = 
 cp ρ  D
(11)
which is identical to the formula of Whitehead and Jarvis
(1981). This implies that, if k and ∆ΨS–L are constant, E remains constant with changing D and gsu declines hyperbolically with increasing D. If D decreases, there will be a point
where the stomata are fully open and canopy conductance is
maximal (gcm ). From this point, E decreases linearly with decreasing D.
Equation 11 was used to predict the –m/gcref ratio in Equation 8 for a realistic range of potential gcref values that were
generated with different values of E (0.16, 0.24 and 0.32 mm
h – 1, converted to g m – 2 s – 1) and measured mean monthly ga
(ranging from 0.09 to 0.15 m s – 1), leaving all other variables
constant (D = 1 kPa, = 2460 J g – 1, = 0.066 kPa K – 1,
cp = 1.005 J g – 1 K – 1 and = 1200 g m – 3 ) and setting gcm to
0.025 m s – 1. For each value of E, and the corresponding gcref,
D was then varied between the observed monthly minimum
and maximum and the corresponding values for –m were approximated as –∆gc /∆lnD for the respective range of D (Oren
et al. 1999) and plotted against gcref. The slope of the resulting
relationship equals the theoretical –m/gcref ratio. No minimum
D values below 0.6 kPa were considered (Ewers and Oren
2000). The theoretical –m/gcref ratio was calculated for each of
the 28-day periods and compared with the observed ratios.
For the sensitivity analysis presented in the Discussion, a
constant ga of 0.12 m s – 1 was used, and the ranges of D were
chosen arbitrarily. All other variables remained as described
above.
Results
Weather patterns and leaf area development
The seasonal courses of mean daily air temperature and mean
daytime D above the two woodlands were similar (Figure 1A).
Midday maximum D (not shown) reached 4 kPa on only one
day (DOY 200) and exceeded 2.5 kPa only in July. May and
October were the wettest months, but there was no extended
drought during the growing season of 2006 (Figures 1B
and 1C). Leaf area index was always higher in Grimsbury than
in Wytham (Figure 1D). For days on which LAI was measured
at both sites, the difference was significant (P < 0.01, paired
t-test). In Grimsbury, LAI remained above 0 all year round because of the presence of evergreen holly (Ilex aquifolium L.)
TREE PHYSIOLOGY ONLINE at http://heronpublishing.com
964
HERBST, ROSIER, MORECROFT AND GOWING
Figure 1. (A) Mean daily air temperature (T ) and mean daytime vapor
pressure deficit (D), (B, C) daily (bars) and cumulative (line) rainfall,
(D) cumulative leaf area index (LAI) and (E) soil water content
(SWC; at 0–1 m) at Wytham Woods (dotted lines) and Grimsbury
Wood (solid lines) from April 25 to November 22, 2006 (DOY = day
of year).
trees in the understory. In early June, LAI decreased temporarily in Grimsbury because the oak trees were partially defoliated by herbivorous insects. To a lesser extent, the sycamore
trees in Wytham were also affected by insect infestations. The
sites had similar top soil water contents (Figure 1E). Because
of regular rainfall events in 2006, soil water content did not
show large fluctuations at either site.
Canopy transpiration
Total transpiration per unit ground area was similar for both
woodlands during the first half of the growing season (Figure 2A). Between mid-August and early October, stand E was
higher in Wytham than in Grimsbury. However, many trees in
Grimsbury remained foliated well into November when trees
at Wytham were already leafless. Annual E was 397 mm in
Wytham and 367 mm in Grimsbury, the difference being less
than the measurement accuracy of the sap flux method. In
Wytham (Figure 2B), the ash and sycamore trees contributed
to a similar extent to stand E during most of the growing season, but the ash trees came into leaf more than 3 weeks after
the sycamore trees. The contribution of understory trees to total stand E was about 9%. In Grimsbury (Figure 2C), canopy E
was dominated by oak, which showed a strong seasonal variation with more than 75% of annual E occurring from June to
Figure 2. Cumulative transpiration (E), scaled up from hourly sap flux
densities, for (A) each entire stand and for the main species in (B)
Wytham Woods and (C) Grimsbury Wood from April 25 to November
22, 2006 (DOY = day of year). Annual E totals were 397 mm for
Wytham and 367 mm for Grimsbury. In Wytham, sycamore transpired
178 mm, ash 162 mm and oak 21 mm, and the understory trees (u/s)
added a further 35 mm to total E. In Grimsbury, oak transpired
195 mm and birch 105 mm. Understory E was 67 mm of which 50 mm
came from hazel shrubs.
August. Birch used less water than oak during this period, but
used more in the spring and about the same amount in the autumn. In Grimsbury, the contribution of understory trees to total stand E was 18%, about twice as high as in Wytham.
Figure 3 compares daily canopy E with the Priestley-Taylor
formulation (Equation 1) of daily potential evapotranspiration. Seasonal EPT was 535 mm in Grimsbury and 557 mm
in Wytham. The seasonal totals of E corresponded to mean
Priestley-Taylor coefficients of 0.86 for Grimsbury and 0.90
for Wytham. However, there was strong seasonal variation in
the ratio between E and EPT. In both woodlands, E was constantly, and often considerably, lower than EPT until early August. In Wytham, E was similar to EPT in the second half of August (around DOY 230) and exceeded EPT on many days in
September and early October. In Grimsbury, a similar trend
was observed, but it occurred about a month later than in
Wytham, with E and EPT becoming similar in the second half
of September (around DOY 260) and E exceeding EPT between
mid October and mid November.
TREE PHYSIOLOGY VOLUME 28, 2008
TRANSPIRATION AND CANOPY CONDUCTANCE IN WOODLANDS
965
Figure 3. Daily sums of canopy transpiration (E, bars) and potential
evaporation according to the Priestley-Taylor formula (bold line) for
(A) Grimsbury Wood and (B) Wytham Woods from April 25 to November 22, 2006 (DOY = day of year).
Canopy conductance
The response of gc to D (Figure 4) was plotted separately for
six consecutive 4-week periods for each woodland starting at
the beginning of their respective growing season. The logarithmic curves, which were fitted to the upper envelopes of the
data clouds, appeared steeper for Grimsbury than for Wytham
during most of the season. For D > 2 kPa (and during most of
the time even for D > 1.5 kPa), gc was lower in Grimsbury than
in Wytham. Values of gc were highest under saturating light
conditions, at low D, in the middle of the growing season in
Grimsbury and after mid-August in Wytham. Higher maximum conductances were observed in Grimsbury than in
Wytham during the first half of the season, whereas lower values were found in Grimsbury in the late season. The logarithmic curves better represented the shape of the upper envelopes
for the Wytham data than for the Grimsbury data. The decrease
in gc with increasing D in Grimsbury was generally stronger
than predicted by the logarithmic curve for D < 1.5 kPa and
weaker for higher D values. The reference conductance (at D =
1 kPa) and the slope for each dataset are given in Table 5. Both
gcref and –m showed a seasonal increase in Wytham (r 2 = 0.97,
P = 0.007 for gcref ; and r 2 = 0.68, P = 0.043 for –m) but not in
Grimsbury (r 2 = 0.12, P = 0.51 for gcref ; and r 2 = 0.24, P = 0.36
for –m). The values of gcref from Table 5 were used as a fixed
input for the curve fits describing the response of gc to RG
(Equation 9). The resulting parameters are listed in Table 5.
For both woodlands and during the whole growing season, the
increase in gc with increasing RG was almost linear for RG <
400 W m – 2 but leveled off at higher RG (data not shown).
A closer inspection of the control of E, reflected in different
responses of gc to D in the two woodlands, is presented in Fig-
Figure 4. Canopy conductance (gc) as a function of vapor pressure
deficit (D) for Wytham Woods and Grimsbury Wood at different time
periods during the 2006 growing season. Only data for incoming solar
radiation > 400 W m –2 are shown (solid circles). The upper envelopes
(large open circles) were the basis for the logarithmic curve fits
(lines). The numbers on the right denote the day of year (DOY) range
of the observations graphed and serve as data labels in Figure 5A. Details of the calculations are given in the text, and the fitted parameters
are shown in Table 5.
ure 5. In Figure 5A, the slope of the response function (–m) is
plotted against gcref. At the peak of the growing season (represented by Period 3, which covers most of July) the –m/gcref ratio was closest to the universal ratio of 0.6 (Oren et al. 1999) at
both sites. At Grimsbury, gcref was lower and –m/gcref higher in
all other periods of the season. At Wytham, gcref was lower before Period 3 but higher thereafter. The –m/gcref ratio was
lower in the early season than in Period 3 and higher after
mid-August (Periods 5 and 6). The mean –m/gcref ratio for all
periods was 0.78 for Grimsbury and 0.61 for Wytham, and the
difference between sites was significant (P = 0.028, paired
TREE PHYSIOLOGY ONLINE at http://heronpublishing.com
966
HERBST, ROSIER, MORECROFT AND GOWING
Table 5. Empirical parameters fitted to the functions relating gc to D (Equation 8, Figure 4) and RG (Equation 9). The significance of individual parameters is indicated by asterisks (ns, P > 0.05; *, P < 0.05; **, P < 0.01; and ***, P < 0.001), and the theoretical ratio of the two parameters describing the gc response to D was calculated with Equation 11.
Time period
(DOY)
D range
(kPa)
Mean daytime D
(kPa)
Theoretical ratio
–m/gcref
gcref
(mm s – 1)
–m
(mm s – 1 kPa – 1)
gcmin
(mm s – 1)
Initial slope
of gc versus RG
Wytham Woods
115–142
143–170
171–198
199–226
227–254
255–282
0.5–2.1
0.7–2.5
0.7–3.2
0.5–4.0
0.5–1.9
0.5–1.7
0.52
0.80
1.08
1.00
0.61
0.45
1.09
0.80
0.72
0.66
1.14
1.21
5.62***
9.31***
11.30***
11.25***
16.65***
15.03***
2.95**
4.22***
6.61***
5.95***
14.54***
10.03*
2.64***
1.89***
1.50***
2.63***
6.84***
5.19***
0.91 ns
2.95**
3.18***
3.15***
1.92 ns
2.82 ns
Grimsbury Wood
117–144
0.5–2.5
145–172
0.5–2.4
173–200
0.5–4.3
201–228
0.5–3.0
229–256
0.5–2.5
257–284
0.5–1.9
0.58
0.92
1.29
1.10
0.86
0.60
0.97
0.98
0.78
0.90
0.97
1.09
3.92***
10.29***
12.92***
11.25***
8.57***
9.22***
3.34**
7.84***
8.62***
8.59***
7.22***
7.29***
1.45**
1.20*
1.22*
1.31***
1.32**
1.73 ns
0.60 ns
2.45*
2.40*
3.54***
4.25***
3.12 ns
t-test). However, a comparison among periods, and with the
universal ratio of 0.6, is difficult when the range of D used to
fit the curves also differs. Because the range of D influences
the theoretical value of –m/gcref, the observed –m/gcref ratios
must be compared with the theoretical values for similar
ranges of D as calculated from Equation 11 rather than with
the universal ratio of 0.6. The ratio of the observed to the theoretical –m/gcref for each 4-week period is shown in Figure 5B. The ratio was always higher in Grimsbury than in
Wytham, indicating that stomatal regulation was more sensitive and better able to maintain a relatively stable leaf water
potential in the oak- and birch-dominated forest. In Wytham,
this sensitivity was particularly low in the spring and autumn.
On average, the ratio of actual to theoretical –m/gcref was 0.83
in Grimsbury and 0.66 in Wytham, and the difference was significant (P = 0.030, paired t-test). The variation in this ratio
over the growing season did not correlate with relative soil water content (r 2 = 0.46, P = 0.14 for Wytham; and r 2 = 0.01, P =
0.84 for Grimsbury) or with mean daytime D (r 2 = 0.55, P =
0.09 for Wytham; and r 2 = 0.09, P = 0.57 for Grimsbury). The
best (albeit not significant) correlation was found with mean
daytime D of the previous month (r 2 = 0.74, P = 0.06 for
Wytham and r 2 = 0.41, P = 0.24 for Grimsbury), hinting at a
possible acclimation of the canopy to mid-term weather patterns.
Discussion
Control of transpiration
Figure 5. (A) Slope of the logarithmic curve describing the response
of canopy conductance (gc) to vapor pressure deficit (D) plotted
against the reference conductance (gcref ). The universal ratio of 0.6
suggested by Oren et al. (1999) is indicated by a broken line. The
number beside each symbol refers to the day of year (DOY) range for
which the slopes were calculated, according to Figure 4. (B) Ratio of
actual –m/gcref to the theoretical –m/gcref value necessary to maintain
leaf water potential for increasing D from April 25 to October 11,
2006. Symbols: 䊉, Grimsbury Wood; and 䊊, Wytham Woods.
There was no significant difference in annual E totals between
the studied woodlands, or when compared with reported results for two even-aged beech plantations (having no understory) in the U.K. and in northern Germany that transpired on
average 377 and 389 mm per year, respectively (Herbst et al.
1999, Roberts et al. 2001). Based on comparisons of annual E
relative to equilibrium E, or Priestley-Taylor coefficient (α),
these four woodlands fall well within the range of 0.82 ± 0.16
for 14 temperate broad-leaved woodlands reviewed by Komatsu (2005). However, this interpretation holds only for mean
annual and masks any large changes during the growing sea-
TREE PHYSIOLOGY VOLUME 28, 2008
TRANSPIRATION AND CANOPY CONDUCTANCE IN WOODLANDS
son. Shuttleworth and Calder (1979) were among the first to
question the use of a constant for forests, and Monteith
(1995) showed that is closely linked to mean daily gc. Therefore, the fraction of available energy dissipated by transpiration is poorly described by a constant factor. Instead it is controlled by gc which, in Oren’s model, depends on gcref, –m and
micrometeorological conditions.
One reason for the large seasonal changes in the fraction of
RN used for transpiration (Figure 3) is that, in temperate forests, the development of LAI in the spring and early summer
lags considerably behind the seasonal increase in available energy. A further time lag exists between LAI development
and the physiological differentiation of leaves (Morecroft and
Roberts 1999, Morecroft et al. 2003). This, as well as the onset
of leaf senescence, is species-specific and has a strong influence on mean gc and . Ash trees in wet woodlands, for example, reach peak E later than co-occurring tree species (Callaghan 2007), and oak-dominated forests increase the fraction
of available energy used for transpiration late in the season
provided there is sufficient water in the soil (Rasmussen and
Rasmussen 1984, Poyatos et al. 2007).
Response of gc to the environment
Regulation of water loss at the leaf scale is species-specific.
Ash has been shown to be exceptionally tolerant (Guicherd et
al. 1997) of low leaf water potentials while maintaining high gc
and E at high D (Besnard and Carlier 1990, Marigo and Peltier
1996). This behavior changes only on sites that experience soil
water deficits (Carlier et al. 1992). Our study is the first in
which similar characteristics were observed directly at the
stand scale of an ash- and sycamore-dominated woodland.
Oak stands, however, are usually characterized by a strong reduction in gc with increasing D that is unaffected by soil water
as long as the relative extractable soil water content remains
above 50% (Granier and Bréda 1996). This characteristic behavior of gc, observed in a Quercus petraea (Matt.) Liebl.
stand in France, was confirmed in our study for the Q. roburdominated woodland at Grimsbury. Under non-limiting light
and at the peak of the growing season, gc decreased from about
13 mm s –1 at D = 1 kPa to about 4 mm s –1 at D = 3 kPa in both
stands. The –m/gcref ratio of 0.67 observed during the period
with the largest D range (Figure 5A) is similar to ratios found
in two evergreen oak species (0.62 for Q. ilex L. and 0.63 for
Q. suber L.) for a similar range of D (David et al. 2007). However, in the long term and at the stand scale, differences in
stomatal behavior between the ash and sycamore forest and the
oak and birch forest partly balanced each other out and were
compensated for by differences in LAI and the length of the
growing season, resulting in similar annual E of the two woodlands.
A reanalysis of gc data for an even-aged, monospecific
beech plantation in Germany (Herbst 1995) by the same standards as used in this study, yielded –m/gcref ratios of 0.71 and
0.64 for two consecutive years, for D ranges of 0.5–2.5 and
0.5–3.5 kPa, respectively. Therefore, stomatal sensitivity in
the beech forest was slightly higher than in the ash and sycamore forest but lower than in the oak and birch forest. Oren et
967
al. (1999) reported a tendency for ring-porous trees to have
higher stomatal sensitivity than diffuse-porous trees. Both
Wytham and Grimsbury are mixed stands that contain both
ring- and diffuse-porous species, but the site with the highest
proportion of ring-porous trees (the oak trees in Grimsbury)
showed the highest stomatal sensitivity, in agreement with
Oren et al. (1999).
Interannual variations in stomatal sensitivity may be described through acclimation of the stomata to prevailing
weather patterns rather than through simple phenological
models (Kutsch et al. 2001). Our study supports this suggestion, because the ratio of actual to theoretical sensitivity of gc
was at least weakly correlated with the mean D experienced by
the canopies during the previous weeks. The seasonal course
of gcref might have been influenced by partial defoliation by insects, because regrowth foliage (after defoliation) can develop
a higher stomatal conductance than primary foliage in some
deciduous tree species (Turner and Heichel 1977).
Range of vapor pressure deficit
The theoretical decrease in gc with increasing ln D (Equation 8) is not constant over the entire range of D (Oren et al.
1999). The theoretical proportionality factor between –m and
gcref changes if the upper limit of the range of D varies while
the lower limit remains at D = 1 kPa (Oren et al. 1999, Ewers et
al. 2007). For the mean ga observed at Wytham and Grimsbury,
this behavior of –m/gcref is evident in Figure 6A. However, if
the lower limit of D is altered (Figure 6B), the theoretical
slope calculated as –∆gc /∆lnD changes significantly, even
with only small changes in the range of D. Applying the same
concept for a range of D of only 1 kPa (Figure 6C) causes an
even higher variability in the theoretical –m/gcref, depending
on mean D. The situations illustrated in Figures 6B and 6C
have not been the focus of research so far, because they are
mainly relevant to humid temperate climates for which Oren’s
model is not yet well established. However, they imply that an
apparent agreement between –m/gcref ratios from different
sites may be meaningless if the range of D over which gc was
measured differed substantially.
For example, the mean –m/gcref ratio of 0.61 observed for
Wytham could wrongly suggest that this ash-dominated forest
maintained a constant leaf water potential through sensitive
stomatal regulation, which would contradict all physiological
evidence from earlier studies. Therefore, it seems more appropriate to assess relative sensitivity, by the quotient of the actual
and the theoretical –m/gcref for a specific D range, across sites.
For Wytham, such a calculation clarified that stomatal regulation was not sufficiently sensitive to keep the difference between leaf and soil water potentials constant.
The effect of using a narrow range of D when fitting the response curve was highlighted by Oren et al. (1999) and Oren
and Pataki (2001), but apart from the recent study by Herbst et
al. (2007b), to our knowledge, no others have calculated the
theoretical –m/gcref ratio necessary to regulate the leaf water
potential specifically for their sites and climates. Ewers et al.
(2005) acknowledged the variability of the theoretical –m/gcref
ratio explicitly in their introduction, but then nevertheless
TREE PHYSIOLOGY ONLINE at http://heronpublishing.com
968
HERBST, ROSIER, MORECROFT AND GOWING
Figure 6. Slope of the logarithmic curve describing the response of
canopy conductance (gc) to vapor pressure deficit (D) plotted against
the reference conductance (gcref ) for different ranges of D. The lines
show the relationships predicted from Equation 11 assuming a constant aerodynamic conductance (ga ) of 0.12 m s –1. The –m/gcref ratio
changes with (A) the upper boundary of the D range, (B) the lower
boundary and (C) the mean value of a constant D range.
compared their results simply to the universal slope of 0.6, despite using a D range of 0.6–3 instead of 1– 4 kPa. Oren et al.
(2001) seem to have used a range of D of 0.2–1.8 kPa while
showing that –m/gcref remained robustly at 0.6 in their study of
a Taxodium forest. From the figures in Schäfer et al. (2000),
their –m/gcref ratio observed for a German beech forest over a
D range of 0.2–2.2 kPa is about 1.1 which, given the low and
narrow range of D, may well be close to the theoretical slope
for their site rather than reflecting an exceptionally sensitive
stomatal behavior.
Such low ranges of D are particularly problematic for the
calculation of the theoretical stomatal response. This can be
deduced by comparing the theoretical hyperbolic response of
gsu to D in Equation 11 with the empirical logarithmic response in Equation 8. For high values of D, Equation 8 can always be parameterized in such a way as to track the theoretical
hyperbolic curve closely over a considerable range of D. However, as D decreases and finally approaches the point where
gc = gcm, the shapes of the two functions deviate substantially,
and for every small change in D, a large change in the parameters in Equation 8 is necessary to approximate the logarithmic
curve with the hyperbolic curve. Therefore, the theoretical
–m/gcref ratio changes substantially if the range of D is extended to values below 1 kPa (cf. Figures 6 B and 6C). If only
the range of D above 1 kPa had been used in the calculations
for Figure 5B, for example, then the data would have been
shifted upward by a considerable margin to give a mean of
1.19 for Grimsbury and 0.86 for Wytham. This implies that,
while D > 1 kPa, the oak and birch forest reduced stomatal
conductance even more than necessary to maintain a constant
difference between leaf and soil water potentials, whereas the
ash and sycamore forest still responded too weakly to achieve
this. A similar calculation for the beech forest data of Herbst
(1995) brings the ratio between the actual and theoretical sensitivity to 1.00 in the first year of observations and to 1.09 in
the second year. This illustrates the need for careful data interpretation if the approach of Oren et al. (1999) is used in cool
and humid climates where a large fraction of the data is obtained at D around or below 1 kPa.
In conclusion, differences in woodland structure and species-specific physiological responses to the environment can
cause differences in the partitioning of forest E between canopy layers and seasons. In practice, many of the differences
will cancel each other out if the total annual water use of a
woodland is considered and if special cases such as hedgerows
(Herbst et al. 2007b), woodland edges (Herbst et al. 2007a)
and wet woodlands (Callaghan 2007) are excluded. This
prompted Roberts (1983) to describe forest E as a conservative
hydrological process. Despite the considerable progress made
during the last 25 years in understanding the underlying regulatory mechanisms, this statement remains valid; however, if
there are rapid changes in climate, woodland structure or species composition, then this statement can no longer be taken
for granted. Methods have yet to be established to predict forest water use reliably while allowing for such changes. Data
availability permitting, gc-based approaches have the greatest
potential to produce robust and transferable conclusions about
forest E in changing environments. The concept introduced by
Oren et al. (1999) has the advantage that it enables a physiological comparison across sites by standardizing the analysis
of the response of gc to D in a simple manner. It can be applied
to both leaf-scale and canopy-scale data, and the –m/gcref ratio
is unaffected by the scale (Oren et al. 1999). The model, therefore, has potentially wide applicability (Ewers et al. 2007).
However, our results indicate that, in comparative studies or
extrapolations, it is essential to take the actual range of D into
account, because it often varies considerably across sites and
seasons.
Acknowledgments
We are grateful to Michèle Taylor and Dave McNeil who looked after
the meteorological data at Wytham Woods and Grimsbury Wood, respectively, and to Nathan Callaghan for contributing to the calibration
work. The study was proposed and initiated by John Roberts of CEH
Wallingford who died on February 22, 2007. Without his input, this
TREE PHYSIOLOGY VOLUME 28, 2008
TRANSPIRATION AND CANOPY CONDUCTANCE IN WOODLANDS
paper would not have been possible. The work was part of the Lowland Catchment Research (LOCAR) programme and funded by the
National Environmental Research Council (NERC) through Grant
no. NER/T/S/2001/00939. We thank Ram Oren for valuable comments on an earlier version of the manuscript.
References
Besnard, G. and G. Carlier. 1990. Potentiel hydrique et conductance
stomatique des feuilles de frêne (Fraxinus excelsior L.) dans une
forêt alluviale du Haut-Rhône français. Ann. Sci. For. 47:353–365.
Businger, J.A. 1956. Some remarks on Penman’s equation for the
evaporation. Neth. J. Agric. Sci. 4:77–80.
Callaghan, N.C. 2007. Water loss from wet woodlands of the English
lowlands. Ph.D. Thesis, The Open University, Milton Keynes,
U.K., 295 p.
Carlier, G., J.P. Peltier and L. Gielly. 1992. Comportement hydrique
du frêne (Fraxinus excelsior L.) dans une formation montagnarde
mésoxérophile. Ann. Sci. For. 49:207–223.
Clearwater, M.J., F.C. Meinzer, J.L. Andrade, G. Goldstein and
N.M. Holbrook. 1999. Potential errors in measurement of nonuniform sapflow using heat dissipation probes. Tree Physiol. 19:
681–687.
David, T.S., M.O. Henriques, C. Kurz-Besson et al. 2007. Water-use
strategies in two co-occurring Mediterranean evergreen oaks: surviving the summer drought. Tree Physiol. 27:793–803.
Dolman, A.J., E.J. Moors, T. Grünwald, P. Berbigier and C. Bernhofer. 2003. Factors controlling forest atmosphere exchange of water, energy and carbon. In Fluxes of Carbon, Water and Energy of
European Forests. Ed. R. Valentini. Springer-Verlag, Heidelberg,
pp 207–223.
Ewers, B.E. and R. Oren. 2000. Analyses of assumptions and errors in
the calculation of stomatal conductance from sap flux measurements. Tree Physiol. 20:579–589.
Ewers, B.E., S.T. Gower, B. Bond-Lamberty and C.K. Wang. 2005.
Effects of stand age and tree species on canopy transpiration and
average stomatal conductance of boreal forests. Plant Cell Environ.
28:660–678.
Ewers, B.E., D.S. Mackay and S. Samanta. 2007. Interannual consistency in canopy stomatal conductance control of leaf water potential across seven tree species. Tree Physiol. 27:11–24.
Granier, A. 1985. Une nouvelle méthode pour la mesure du flux de
sève brute dans le tronc des arbres. Ann. Sci. For. 42:193–200.
Granier, A. 1987. Evaluation of transpiration in a Douglas-fir stand by
means of sap flow measurements. Tree Physiol. 3:309–320.
Granier, A. and N. Bréda. 1996. Modeling canopy conductance and
stand transpiration of an oak forest from sap flow measurements.
Ann. Sci. For. 53:537–546.
Granier, A., P. Biron, N. Bréda, J.-Y. Pontailler and B. Saugier. 1996.
Transpiration of trees and forest stands: short and long-term monitoring using sapflow methods. Global Change Biol. 2:265–274.
Granier, A., P. Biron and D. Lemoine. 2000. Water balance, transpiration and canopy conductance in two beech stands. Agric. For.
Meteorol. 100:291–308.
Guicherd, P., J.P. Peltier, E. Gout, R. Bligny and G. Marigo. 1997. Osmotic adjustment in Fraxinus excelsior L.: malate and mannitol accumulation in leaves under drought conditions. Trees 11:155–161.
Herbst, M. 1995. Stomatal behaviour in a beech canopy: an analysis
of Bowen ratio measurements compared with porometer data.
Plant Cell Environ. 18:1010–1018.
Herbst, M., C. Eschenbach and L. Kappen. 1999. Water use in neighbouring stands of beech (Fagus sylvatica L.) and black alder (Alnus
glutinosa (L.) Gaertn.). Ann. Sci. For. 56:107–120.
969
Herbst, M., J.M. Roberts, P.T.W. Rosier, M.E. Taylor and D.J. Gowing. 2007a. Edge effects and forest water use: a field study in a
mixed deciduous woodland. For. Ecol. Manage. 250:176–186.
Herbst, M., J.M. Roberts, P.T.W. Rosier and D.J. Gowing. 2007b.
Seasonal and interannual variability of canopy transpiration of a
hedgerow in southern England. Tree Physiol. 27:321–333.
Jarvis, P.G. and K.G. McNaughton. 1986. Stomatal control of transpiration: scaling up from leaf to region. Adv. Ecol. Res. 15:1–49.
Komatsu, H. 2005. Forest categorization according to dry-canopyevaporation rates in the growing season: comparison of the
Priestley-Taylor coefficient values from various observation sites.
Hydrol. Proc. 19:3873–3896.
Körner, C. 1994. Leaf diffusive conductances in the major vegetation types of the globe. In Ecophysiology of Photosynthesis.
Eds. E.-D. Schulze and M.M. Caldwell. Springer-Verlag, Heidelberg, pp 463–490.
Kumagai, T., S. Aoki, H. Nagasawa, T. Mabuchi, K. Kubota, S. Inoue,
Y. Utsumi and K. Otsuki. 2005a. Effects of tree-to-tree and radial
variations on sap flow estimates of transpiration in Japanese cedar.
Agric. For. Meteorol. 135:110–116.
Kumagai, T., H. Nagasawa, T. Mabuchi, S. Ohsaki, K. Kubota,
K. Kogi, Y. Utsumi, S. Koga and K. Otsuki. 2005b. Sources of error
in estimating stand transpiration using allometric relationships between stem diameter and sapwood area for Cryptomeria japonica
and Chamaecyparis obtusa. For. Ecol. Manage. 206:191–195.
Kutsch, W.L., M. Herbst, R. Vanselow, P. Hummelshøj, N.O. Jensen
and L. Kappen. 2001. Stomatal acclimation influences water and
carbon fluxes of a beech canopy in northern Germany. Basic Appl.
Ecol. 2:265–281.
Leuzinger, S., G. Zotz, R. Asshoff and C. Körner. 2005. Responses of
deciduous forest trees to severe drought in Central Europe. Tree
Physiol. 25:641–650.
Marigo, G. and J.P. Peltier. 1996. Analysis of the diurnal change in osmotic potential in leaves of Fraxinus excelsior L. J. Exp. Bot.
47:763–769.
McNaughton, K.G. and P.G. Jarvis. 1991. Effects of spatial scale on
stomatal control of transpiration. Agric. For. Meteorol. 54:279–301.
Monteith, J.L. 1965. Evaporation and environment. Symp. Soc. Expt.
Biol. 19:205–234.
Monteith, J.L. 1995. Accommodation between transpiring vegetation
and the convective boundary layer. J. Hydrol. 166:251–263.
Morecroft, M.D. and J.M. Roberts. 1999. Photosynthesis and stomatal conductance of mature canopy Oak (Quercus pubescens) and
Sycamore (Acer pseudoplatanus) trees throughout the growing
season. Funct. Ecol. 13:332–342.
Morecroft, M.D., M.E. Taylor and H.R. Oliver. 1998. Air and soil microclimates of deciduous woodland compared to an open site.
Agric. For. Meteorol. 90:141–156.
Morecroft, M.D., V.J. Stokes and J.I.L. Morison. 2003. Seasonal
changes in the photosynthetic capacity of canopy oak (Quercus
pubescens) leaves: the impact of slow development on annual carbon uptake. Int. J. Biometeorol. 47:221–226.
Oliveras, I. and P. Llorens. 2001. Medium-term sap flux monitoring in
a Scots pine stand: analysis of the operability of the heat dissipation
method for hydrological purposes. Tree Physiol. 21:473–480.
Oren, R. and D.E. Pataki. 2001. Transpiration in response to variation
in microclimate and soil moisture in southeastern deciduous forests. Oecologia 127:549–559.
Oren, R., J.S. Sperry, G.G. Katul, D.E. Pataki, B.E. Ewers, N. Phillips
and K.V.R. Schäfer. 1999. Survey and synthesis of intra- and
interspecific variation in stomatal sensitivity to vapour pressure
deficit. Plant Cell Environ. 22:1515–1526.
TREE PHYSIOLOGY ONLINE at http://heronpublishing.com
970
HERBST, ROSIER, MORECROFT AND GOWING
Oren, R., J.S. Sperry, B.E. Ewers, D.E. Pataki, N. Phillips and
J.P. Megonigal. 2001. Sensitivity of mean canopy stomatal conductance to vapor pressure deficit in a flooded Taxodium distichum L. forest: hydraulic and non-hydraulic effects. Oecologia 126:
21–29.
Poyatos, R., J. Èermák and P. Llorens. 2007. Variation in the radial
patterns of sap flux density in pubescent oak (Quercus pubescens)
and its implications for tree and stand transpiration measurements.
Tree Physiol. 27:537–548.
Priestley, C.H.B. and R.J. Taylor. 1972. On the assessment of surface
heat flux and evaporation using large scale parameters. Monthly
Weather Rev. 100:81–92.
Rasmussen, K.R. and S. Rasmussen. 1984. The summer water balance in a Danish oak stand. Nordic Hydrol. 15:213–222.
Roberts, J.M. 1983. Forest transpiration: a conservative hydrological
process? J. Hydrol. 66:133–141.
Roberts, J.M. and P.T.W. Rosier. 2006. The effect of broadleaved
woodland on Chalk groundwater resources. Q. J. Engineer. Geol.
Hydrogeol. 39:197–207.
Roberts, J.M., P.T.W. Rosier and D.M. Smith. 2001. Effects of afforestation on chalk groundwater resources. Final Report, DEFRA
project CWO823, CEH Wallingford, 81 p.
Roberts, J.M., P.T.W. Rosier and D.M. Smith. 2005. The impact of
broadleaved woodland on water resources in lowland U.K.: II. a
comparison of evaporation estimates made from sensible heat flux
measurements over beech woodland and grass on chalk sites in
Hampshire. Hydrol. Earth Syst. Sci. 9:607–613.
Schäfer, K.V.R., R. Oren and J.D. Tenhunen. 2000. The effect of tree
height on crown level stomatal conductance. Plant Cell Environ.
23:365–375.
Shuttleworth, W.J. and I.R. Calder. 1979. Has the Priestley-Taylor
equation any relevance to forest evaporation? J. Appl. Meteorol.
18:639–646.
Turner, N.C. and G.H. Heichel. 1977. Stomatal development and seasonal changes in diffusive resistance of primary and regrowth foliage of red oak (Quercus rubra L.) and red maple (Acer
rubrum L.). New Phytol. 78:71–81.
Vincke, C., N. Bréda, A. Granier and F. Devillez. 2005. Evapotranspiration of a declining Quercus robur (L.) stand from 1999 to
2001. I. Trees and forest floor daily transpiration. Ann. For. Sci.
62:503–512.
Whitehead, D. and P.G. Jarvis. 1981. Coniferous forests and plantations. In Water Deficits and Plant Growth. Vol. VI. Ed. T.T. Kozlowski. Academic Press, London, pp 49–152.
TREE PHYSIOLOGY VOLUME 28, 2008