shoot, branch and canopy

Tree Physiology 20, 713–723 © 2000 Heron Publishing—Victoria, Canada

Measuring and modeling conductances of black spruce at three organizational scales: shoot, branch and canopy M. B. RAYMENT,1 D. LOUSTAU2 and P. G. JARVIS1 1

IERM, University of Edinburgh, Edinburgh EH9 3JU, U.K.

2

INRA-Forêts, BP 45, 33611 Gazinet, France

Received November 4, 1999

Summary To investigate the extent to which the energy balance of a globally important ecosystem is controlled by biological and environmental processes, measurements of water vapor flux were made on individual black spruce (Picea mariana [Mill.] B.S.P.) shoots, branches, and a whole canopy at the BOREAS Southern Study Area Old Black Spruce (SSA OBS) site. These measurements were used to estimate stomatal, branch boundary layer and canopy boundary layer conductances to water vapor. On a projected needle area basis, stomatal conductances varied between 14 and 92 mmol m –2 s –1, and total branch conductance varied seasonally between zero and about 35 mmol m –2 s –1. On a ground area basis, total canopy conductance varied between 24 and 105 mmol m –2 s –1. Total canopy conductance was partitioned into aerodynamic and physiological components by using shoot-scale measurements scaled by leaf area index. Good agreement was found with an independent estimate of aerodynamic conductance measured when the canopy was wet. Compared with most coniferous forests, the canopy was relatively uncoupled from the atmosphere, and at the ecosystem scale, the control of water vapor flux was approximately equally divided between physiological and abiotic conductances. Two widely used steady-state models of stomatal conductance were parameterized from the shoot and branch measurements. Parameters varied considerably throughout the growing season. A time-constant term was added to these static models to construct dynamic models of stomatal conductance under naturally varying environmental conditions. The dynamic versions of the models outperformed the static versions in explaining stomatal response to rapidly changing environmental conditions. The length of the time-constant term, derived using the dynamic models, suggested that stomata were slow to respond to changing environmental conditions, and that the speed of the response was strongly temperature-dependent. Keywords: BOREAS, boundary layer conductance, canopy conductance, dynamic model, Picea mariana, stomatal conductance, time-constant.

Introduction A major objective of the BOREAS project was to develop and parameterize simulation models of the processes controlling the exchange of gases and energy between the biosphere and the lower atmosphere in the boreal region (Sellers et al. 1995). Of fundamental importance in describing the energy balance of the land surface is a description of the water vapor flux. For vegetated surfaces, this implies a description of the entire pathway encountered by a molecule of water vapor escaping from the liquid phase at the cell wall until it reaches a point outside the influence of the canopy. This pathway comprises a stomatal section, i.e., through the intercellular spaces, substomatal cavity, stomatal pore and stomatal antechamber, and passage through various boundary layers, e.g., the leaf boundary layer between the leaf surface and some point outside the influence of the leaf, or the shoot boundary layer between a point among the leaves to some point outside the influence of the shoot (Figure 1). Models describing the function of stomata in regulating water vapor flux are usually parameterized from response curve studies, where individual environmental variables are varied and their effects on stomatal conductance measured. Although this is feasible for small-scale structures such as leaves or shoots, it is impractical for larger structures such as whole

Figure 1. Schematic showing the pathway of a molecule of water vapor from the stomatal cavity to the free atmosphere. Also shown are the resistances encountered and the groupings of resistances measured by each technique.

714

RAYMENT, LOUSTAU AND JARVIS

branches or canopies. It is, however, precisely with effects at these larger scales that we are most interested. A way of overcoming this difficulty is to measure water vapor flux, together with those environmental variables that influence conductance, under a range of ambient conditions. The same fitting procedures can then be used to extract model parameter values using the naturally occurring variation in the driving environmental variables. This approach, however, gives rise to two fundamental problems. First, there is the problem of correlations between the driving variables. For example, because the saturation water vapor content of air varies with temperature, a strong correlation exists between the temperature and the humidity deficit of ambient air. A less direct, but equally important correlation exists between incident irradiance and surface temperature. The limitations imposed by these correlations can be minimized by making measurements over a very long period (and therefore including a wider range of natural variation), and by extracting parameters from carefully selected subsets of the total data set. In this way, although the driving variables may not be strictly independent of each other, the effects of each can be studied in isolation. Second, there is the problem of correlation between one measurement and the next. Thus the stomatal conductance at one moment is likely to depend not only on current environmental conditions, but also on what the conductance was in the recent past. In part, this is because, on a short time scale, the microclimatic environment to which stomata respond may be decoupled from changes in driving environmental variables, but it is also a consequence of the sometimes long response time of some biological processes. For example, in Scots pine (Pinus sylvestris L.), stomata have been shown to continue reacting to a previous change in photon flux density (Q) sometime after it had been reversed (Ng and Jarvis 1980), and in Scots pine and Sitka spruce (Picea sitchensis (Bong.) Carr.), there is evidence that stomata might make 66.7% of the steady-state response to a sizeable step change in environmental conditions in about 20 minutes, and might need about 45 minutes to achieve 90% of the steady-state response at typical temperatures (Ludlow and Jarvis 1971, Neilson et al. 1972). Because the extent of this problem depends on the speed of the plant’s response to changing environmental conditions, it can be avoided by making measurements at sufficiently long time intervals that each measurement is effectively independent of the previous one. However, the rate at which most biological processes respond to changing environmental conditions remains largely unknown and it is impossible to say with certainty when, if ever, a process is fully adjusted to current conditions. Moreover, if the models generated from these data are to describe accurately the behavior of stomata in the natural environment, it is essential that the time-dependent nature of stomatal response is captured and described. Automated branch cuvettes (Dufrêne et al. 1993, Rayment and Jarvis 1999) provide a convenient way of monitoring the variation in gas exchange of whole branches as they respond to changing environmental conditions over extended periods of

time. Furthermore, by making measurements at an organizational level between that of the shoot and the whole canopy, we can obtain essential information about how processes scale between simple structures and more complex ones. From measurements of the water vapor flux and the difference between the water vapor pressures (or mole fractions) at the source (i.e., the leaf) and at the sink (i.e., the atmosphere), it is possible to estimate the total conductance to water vapor. By making measurements of the total conductance to water vapor at several organizational scales, it is possible to break down the total conductance into its constituent components, i.e., the conductances of the stomata and the various boundary layers. This paper describes a study of conductance to water vapor in the most extensive Canadian boreal forest tree species, black spruce (Picea mariana [Mill.] B.S.P.), and draws on measurements made at three organizational scales; individual shoots, whole branches and the whole ecosystem. Stomatal conductance is estimated from measurements made on individual shoots, and this estimate is used to partition the total conductance measured on whole branches into the component stomatal and branch boundary layer conductances. The stomatal component is then extracted from the total canopy conductance, generating estimates of the aerodynamic conductance of the whole canopy. These estimates are then compared with independent estimates of canopy aerodynamic conductance obtained through micrometeorological methods. Dynamic models are derived to describe how stomatal conductance varies in response to environmental variation.

Materials and methods The study was made at the BOREAS Southern Study Area Old Black Spruce (SSA OBS) site, near the southern limit of the boreal forest in northern Saskatchewan, Canada, during 1996. A full description of the BOREAS project is given by Sellers et al. (1995) and a full description of the site is given by Jarvis et al. (1997). Shoot gas exchange measurements Gas exchange measurements were made on shoots from the upper (six shoots) and lower (three shoots) crowns of four trees in July 1996 with an open gas exchange system equipped with climate control (Compact Minicuvette System, Heinz Walz GmbH, Effeltrich, Germany). Each shoot comprised four to seven needle age classes and had a total needle area of approximately 70 cm 2. Within the cuvette, shoots were placed in a horizontal plane and received bilateral illumination from two fiber-optic illuminators (Heinz Walz GmbH). Curves describing the response of gas exchange to intracellular CO2, incident Q and needle temperature were determined for each shoot. Full descriptions of the methodology used and the results obtained are given by Rayment (1998). After the gas exchange measurements were made, the shoots were cut off and the dry mass of each needle age class of each shoot was determined following drying at 65 °C in a ventilated oven. Specific needle areas for each age class and

TREE PHYSIOLOGY VOLUME 20, 2000

CONDUCTANCES OF BLACK SPRUCE

715

crown location were estimated from 30 samples of five needles whose projected areas were calculated from their length and width. The total projected needle area of each shoot was then estimated from needle dry masses and specific needle areas.

gW, was calculated from the water vapor flux, E, and the needle to air humidity difference, Dia, according to:

Branch gas exchange measurements

where Dia was calculated as the saturation humidity deficit of the surrounding air with respect to air within the leaf (assumed to be saturated with water at the leaf temperature) and expressed as a mole fraction. Shoot conductance to water vapor was considered to be made up of stomatal conductance (gsW), needle boundary layer conductance (ganW) and shoot boundary layer conductance (gasW) components, according to:

Branch gas exchange measurements were made from early April until late November 1996 with the ventilated closed-system methodology developed and described by Dufrêne et al. (1993). Modifications to their methodology and a full description of the system used are given by Rayment and Jarvis (1999). Two branch cuvettes (bags) were installed on each of two trees, one positioned in the upper canopy (about 8 m height) and one in the lower canopy (about 5.3 m height). Modal tree height was about 10.2 m. Measurements were made on each branch every 20 min. The bags were open and ventilated for 15 min, after which they were closed for 5 min and CO2 and water vapor exchange were measured on each branch, together with leaf temperature, within-bag air temperature, within-bag incident Q and within-bag relative humidity. Branch gas exchange was calculated from the slope of the regression of gas concentration against time, the volume of the bag and the needle area of the branch. Data from the start of each 5-min measurement (40 s for CO2, 80 s for H2O vapor) were excluded from the calculation to ensure that the IRGA was entirely flushed of air from the previous branch. Thus each data point represented an average over approximately 4 min. At the end of the year, the branches were cut down whole. Needles were divided into two classes: current-year needles and all previous years’ needles. Needle dry mass, specific needle area and total projected needle area were determined for each branch following the methodology described for shoots. Canopy gas exchange measurements Eddy covariance measurements of net ecosystem fluxes were made from late March until early December 1996 with the Edinburgh EdiSol system described in detail by Moncrieff et al. (1997). The ultrasonic anemometer (Solent A1012R, Gill Instruments Ltd., Lymington, England) and gas analyzer intake were positioned at 27 m height. Fluctuations in CO2 and water vapor concentration were measured with a closed path infrared gas analyzer (LI-6262, Li-Cor Inc., Lincoln, NE), and from these data 30-min mean fluxes were calculated by the Edinburgh EdiSol software (Moncrieff et al. 1997). An automatic weather station mounted close to the eddy flux system recorded a full suite of meteorological data. A complete description of the equipment installation is given by Jarvis et al. (1997). The hemisurface leaf area index for the site, based on optical measurements of canopy transmittance, was 4.4 (Chen et al. 1997). Conductance calculations For the shoots and branches, total conductance to water vapor,

gW =

E , Dia

1 1 1 1 . = + + gW gsW ganW gasW

(1)

(2)

For simple structures such as needles, boundary layer conductance can be estimated from wind velocity (u) and the characteristic dimension (d) (Monteith and Unsworth 1990). For cylinders (analogous to conifer needles) the boundary layer conductance was approximated by: ganW ≈ 4.03 (u 0.6 / d 0.4 ).

(3)

The characteristic dimension of conifer needles is small (d ≈ the width of the foliage elements; Monteith and Unsworth 1990), and the windspeed within the shoot cuvette was high (1.5 to 2.5 m s –1). Consequently, for the needles in the shoot cuvette, the needle boundary layer conductance was large, relative to the expected stomatal conductance, and could be ignored. Although needles shelter one another (Landsberg and Thom 1971) so that the boundary layer conductance of a needle grouped together with others into a shoot is approximately half that of an isolated needle exposed to a similar wind speed (Jarvis et al. 1976), we assumed that for the well-ventilated shoots in the shoot cuvette, the shoot boundary layer conductance was also large enough, relative to the stomatal conductance, to be ignored. Therefore, the total conductance to water vapor measured on shoots within the shoot cuvette was assumed to equal their stomatal conductance. For the branches, total conductance to water vapor was considered to comprise stomatal conductance (gsW) and the sum of the conductances through the needle, shoot and branch boundary layers (gabW), in a manner analogous to that described in Equation 2. The branch bag conductance data were first screened to exclude measurements in which the slope of the regression of the water vapor concentrations against time (from which the water vapor flux, E, was calculated) was not significantly different –2 –1 from zero (typically when E < 0.025 mmol m s ). Measurements for which needle to air humidity difference was less –1 than 1 mmol mol were also excluded to avoid calculating spuriously high values of conductance when the humidity gradient was small.

TREE PHYSIOLOGY ON-LINE at http://www.heronpublishing.com

716


A steady-state data set was then extracted from the branch bag data, taking only those measurements where the net CO2 flux (A), the water vapor flux, the incident PFD, the needle to air humidity difference and the calculated stomatal conductance were all within 10% of their previous values, and the needle temperature was within 1 °C of its previous value, i.e., those values measured 20 minutes before. This eliminated any potential problems that would arise if stomata were slow in responding to changing environmental conditions, and ensured that only those data in which stomatal aperture was fully adjusted to current conditions were included in the estimation of the steady-state response parameters. The steady-state data set contained 3.7% of the data from the full data set. To partition the total branch conductance into its component conductances, subsets of the shoot and branch steady-state data were extracted. These subsets included only those data collected in July, where Q was saturating (Q > 1350 µmol m –2 –1 s ), needle temperature (tn ) was at typical daytime values (20 °C < tn < 30 °C), and the humidity deficit of the air surrounding the branch with respect to saturated air at the needle temperature was high enough to ensure that the water vapor flux was well coupled to total conductance (Dia > 10 mmol mol –1). Thus, these data sets contained shoot and branch conductance data collected at the same time of year, under the same environmental conditions. The assumption was made that, under the same environmental conditions, the stomatal conductance was the same for the shoots and the branches. Therefore, the stomatal conductance measured on the shoots was subtracted from the total conductance measured on the branches, yielding the branch boundary layer resistance. This analysis was performed separately for shoots and branches in the upper and lower canopies. The boundary layer conductance of an object is mainly a function of its size and shape and of the wind velocity (Monteith and Unsworth 1990). The size and shape of each branch did not change significantly during the study period, and the wind velocity within each branch cuvette was held constant by the internal mixing fan. Therefore, the boundary layer conductance of each branch was considered to be constant, and the total conductances measured on each branch throughout the study period could be partitioned into their stomatal and boundary layer components. For the whole canopy, total conductance to water vapor, GW , was calculated by inverting the Penman-Monteith equation (Monteith and Unsworth 1990), assuming that, in the turbulent conditions under which the eddy covariance measurements of whole-canopy water flux were made, the canopy boundary layer conductances to water vapor and heat transfer were equal. Parameter GW was considered to comprise canopy boundary layer conductance (gacW) and total stomatal and total branch boundary layer conductance components by assuming that each branch was representative of an equal amount of the foliage in the canopy and that the projected leaf area index, L, was equivalent to the conducting area index, thus:

1 1 1 1 = + + . GW L gsW L gabW gacW

(4)

In addition, the canopy’s aerodynamic conductance was estimated independently from eddy covariance measurements of wet canopy vapor flux, as follows. After rainfall, when the canopy is wet, evaporation of intercepted rainfall dominates the transpiration water flux, needle conductance can be considered to present an insignificant limitation to water vapor flux, and the Penman-Monteith equation simplifies to the Penman equation (Monteith and Unsworth 1990). Furthermore, when the canopy is wet, the atmosphere turbulent and net radiation is small after rain, the canopy surface temperature approaches the wet bulb temperature (Jarvis 1993), and the Penman equation then becomes: E = gacW D / ( ε + 1),

(5)

where ε is the increase of latent heat content per increase of sensible heat content of saturated air (given by s/γ, where s is the slope of the saturation vapor pressure–temperature curve and γ is the psychrometric constant) and gacW can be simply derived. This calculation was applied to those eddy covariance data obtained when the canopy was wet, i.e., when the total evaporation during the 30-min measurement period was less than the total rainfall during the previous 30-min period. Measurement periods when rain fell, or when the average atmospheric saturation humidity deficit was very low (relative humidity > 75%) were excluded. Model parameterization—steady-state models Two types of models were fitted to the branch cuvette stomatal conductance data: the BWB-type model (Ball et al. 1987) where stomatal conductance is described as a function of the net CO2 flux, and the Jarvis-type “response surface” model, (Jarvis 1976) where stomatal conductance is described as a set of responses to climatic variables. The form of the BWB model was: gsW = gsW0 + a 1

A f ( Ds ), ( cs − Γ *)

(6)

where gsW0 is the residual conductance remaining when stomata are fully shut, a1 is a parameter, A is the net CO2 flux, cs is the CO2 mole fraction at the needle surface and Γ* is the CO2 compensation mole fraction in the absence of dark respiration. The term f(Ds) represents the response of stomata to the saturation humidity deficit of air at the needle surface, Ds, and replaced the relative humidity response as included in the original form of this model. The value of cs was calculated from: cs = ca −

A gabC

,

(7)

where ca is the CO2 mole fraction of the air surrounding the



branch and gabC is the sum of the conductances to CO2 through the needle, shoot and branch boundary layers (gabC = 0.607 gabW). * The value of Γ was given by: Γ * = 0.105 K c O 2 / K 0, where Kc (µmol mol –1) and Ko (mmol mol –1) are the Michaelis constants for Rubisco for CO2 and oxygen, respectively, and O2 is the intracellular oxygen mole fraction (210 mmol mol –1) (Harley et al. 1985). The temperature dependence of the parameters Kc and Ko were described according to Leuning (1995) in terms of their value at 20 °C (Kc20 = 302, Ko20 = 256) as follows:  Hv  29315 .  parmT = parm20 exp   , 1 − . R Tn    29315

(9)

where parmT and parm20 are the values of either parameter at needle temperature T and 20 °C, respectively, Hv is the activation energy (Kc: Hv = 59430 J mol –1 K –1; Ko: Hv = 36000 J mol –1 K –1), Tn is the needle temperature in degrees Kelvin and R is the gas constant (8.314 J mol –1 K –1). Following Jarvis (1993), Ds was calculated as: Ds =

Da + s( Tn − Ta ) , 1 + gsW / gabW

(10)

where Da and Ta are the saturation humidity deficit and absolute temperature of air surrounding the branch, respectively. Two forms of the stomatal response to the saturation humidity deficit of air at the needle surface, f(Ds), were fitted to the data; a hyperbolic response (rearranged from Lohammer et al. 1980): f ( Ds ) =

1 , 1 + D1Ds

(11)

and a linear response (Jarvis 1976): f ( Ds ) = 1 − D2Ds ,

717

 Q   gsW = gsW0 + ( gsWmax − gsW0 )    Q + KQ   (t − t )(t − t ) β (1 − D2 Ds )  n min max n β  (t opt − t min)(t max − t opt )

 ,  

(13)

where gsW0 and Ds are as described above, and gsWmax is the maximum stomatal conductance when the stomata are fully open. The PFD response was described as a Michaelis-Menten function where parameter KQ is the PFD at which gsW is half its maximum value (Massman and Kaufmann 1991). The humidity response was described as a linear decrease in conductance with increasing Ds, where the parameter D2 is as described above. The temperature response was described as a beta function with three parameters, the temperature at which conductance is at a maximum, topt, and the high and low temperatures (tmax and tmin, respectively) at which conductance would be equal to gsW0. Parameter β is given by (tmax – topt)/(topt – tmin). The parameters tmax and tmin were fitted by eye from boundary analysis of the data (i.e., the values describing an envelope containing all the data). The other parameter values were estimated iteratively. Initial estimates for the parameters gsWmax, KQ and topt were obtained from boundary analysis of the data. An initial estimate for the parameter D2 was obtained by fitting the model (Equation 13) to the steady-state data using the initial estimates of the other parameters. New values for gsWmax, KQ and topt were found by fitting the model to the steady-state data using the initial estimate of D2. A new value for D2 was found by fitting the model to the steady-state data using the new values for gsWmax, KQ and topt. This process was repeated until there was negligible change in parameter estimates. Model parameterization—time constant To construct a dynamic model of stomatal conductance in the field, a time-constant term, τ, was added to the models described above to accommodate the time dependence of the response of stomatal conductance to changing environmental conditions. The transient response was given by: gsW = gsW, t − 20 + ( gsW, t − gsW, t − 20 )(1 − exp{−20 / τ}),

(14)

(12)

where D1 and D2 are both parameters that approach zero when there is no response to needle surface humidity deficit (i.e., D1 equals 1/D0 as used by Lohammer et al. 1980). The residual conductance, gsW0 (effectively, the conductance through the closed stomata plus cuticular conductance), was determined for each branch as the mean conductance observed in a subset of data including only those measurements where needle temperatures were within 1 °C of their previous values (eliminating non-steady-state data) and Q was below –2 the value required to induce stomatal opening (Q < 5 µmol m –1 s ). The empirical coefficients a1, D1 and D2 were determined through least squares fits of the models to the steady-state data set of branch stomatal conductances. The functions included in the Jarvis model were:

where gsW is the instantaneously measured stomatal conductance, gsW,t is the modeled steady-state value of conductance (BWB-type or Jarvis-type) under current environmental conditions, and gsW,t–20 is the modeled steady-state value of conductance under the conditions 20 min previous. The time-constant, τ, is the time taken for the stomata to achieve 63.2% (i.e., 1 – exp{–1}) of its total response to a step change in environmental conditions (Fischbeck and Fischbeck 1987). Timeconstant τ was determined by fitting Equation 14 (using the parameterizations estimated from the steady-state data set) to a subset of data, containing only those measurements made when environmental conditions were changing rapidly. Net assimilation rate, A, was used as an indicator of integrated environmental conditions, such that the data set used for finding τ contained only those measurements where A was less than


718


half or more than twice the previous measurement. To investigate the extent to which the transient response of stomata differs between opening and closing, the data set was subdivided into two subsets, one containing measurements where stomatal conductance was higher than the previous value, the other containing measurements where stomatal conductance was lower than the previous value. The temperature dependence of τ was examined by combining the data into 5 °C needle temperature classes and fitting τ to each temperature class separately. Model parameters were extracted from the data by least squares analysis with the SAS NLIN or REG procedures. Significance was tested with the SAS GLM procedure.

Results Stomatal conductance measured on the shoots varied between 14 and 92 mmol m –2 s –1 (projected needle area basis), with a mean maximum value under normal daytime conditions of 25 mmol m –2 s –1. Seasonally, total branch conductance varied from zero to about 35 mmol m –2 s –1 (projected needle area basis), with a mean value under normal daytime conditions of 22 mmol m –2 s –1. Total canopy conductance varied between 24 and 105 mmol m –2 s –1 (ground area basis), with a mean value under normal daytime conditions of 58.5 mmol m –2 s –1. Partitioning total conductance between component conductances Table 1 presents the total and component conductances to water vapor measured, estimated or assumed for shoots, branches and canopy under similar steady-state environmental conditions at the same time of year. At all scales, the stomatal component of conductance was the most important in controlling the water vapor flux, although at increasing organizational

scale, needle boundary layer conductance and canopy aerodynamic conductances became increasingly important. Canopy aerodynamic conductance calculated from the Penman-Monteith equation using the estimate of canopy needle conductance scaled-up from the branch cuvette measurements was significantly (P < 0.05) lower than the estimate obtained from the total canopy conductance measured when the canopy was wet (Table 2). Steady-state model parameters The residual conductance (gsW0) determined for each branch is given in Table 3. The parameter values for the BWB-type model of stomatal conductance derived from the steady-state branch measurements are given in Table 4. Independently of whether a hyperbolic or linear function was used, the fitted parameter was in no case significantly different from zero (i.e., from the null hypothesis that stomatal conductance showed no response to Ds other than that mediated through the response of net assimilation rate to Ds). Figure 2 illustrates how the fitted parameter a1 varied through the growing season. Parameter a1 increased from low

Table 2. Canopy boundary layer conductances (and standard errors) calculated from the Penman-Monteith equation, under dry canopy conditions using the estimated canopy physiological conductance and under wet canopy conditions. Measurements are expressed on a ground area basis and were made above the canopy at the BOREAS SSA OBS site in 1996. Estimate

gacW (mol m –2 s –1)

n

Windspeed (m s –1)

Dry canopy Wet canopy

0.940 (0.239) 1.051 (0.104)

41 39

4.5 4.0

Table 1. Mean values (and standard errors) of the total and component conductances to water vapor measured {m}, estimated {e} or assumed {a} for three organizational scales: shoots, branches and whole canopy. Measurements were made on six shoots and two branches in the upper canopy, on three shoots and two branches in the lower canopy, and above the canopy at the BOREAS SSA OBS site in July 1996. Environmental conditions were: Q > 1350 µmol m –2 s –1; 20 °C < tn < 30 °C; Dia > 10 mmol mol –1. Mean windspeed at which the canopy measurements were made was 4.5 m s –1. Units of conductance are mmol m –2 s –1, shoot and branch scale conductances are expressed on a projected needle area basis, canopy scale conductances are expressed on a ground area basis. Projected leaf (i.e., needle) area index for the site was 3.0 (calculated from the hemisurface leaf area index of 4.4 and a projected area:hemisurface area ratio of 0.69). Scale

Location

Shoot Upper Lower All Branch Upper Lower All Canopy

Total conductance

Component conductances

GW {m}

gsW {e}

ganW {a}

gasW {a}

25.1 (5.2) 25.3 (0.8) 25.2 (3.0)

25.1 (5.2) 25.3 (0.8) 25.2 (3.0)

∝ ∝ ∝

∝ ∝ ∝

GW {m}

gsW {e}

gabW {e}

22.1 (1.9) 22.5 (4.6) 22.3 (2.0)

25.1 (5.2) 25.3 (0.8) 25.2 (3.0)

186.5 204.5 194.1

GW {e}

gsW {e}

gabW {e}

gacW {e}

58.5 (3.1)

77 (9)

591

940 (239)


CONDUCTANCES OF BLACK SPRUCE Table 3. Mean values (and standard errors) of the residual conductance to water vapor, gsW0, measured on two branches in the upper canopy and two branches in the lower canopy at the BOREAS SSA OBS site in 1996. Environmental conditions were low Q (< 5 µmol m –2 s –1) and constant needle temperature (see text).

719

Table 4. Parameter values (and standard errors) of the BWB model of stomatal conductance, estimated from measurements made under steady-state environmental conditions on two branches in the upper canopy and two branches in the lower canopy at the BOREAS SSA OBS site in 1996.

values in the first month of measurement to a maximum in the second month. Thereafter values declined. Parameter a1 calculated for Branch 2 in the last month was high, with a correspondingly high standard error. Values for the parameters of the Jarvis model of conductance applied to the steady-state branch measurements are given in Table 5, together with standard errors where appropriate. No standard errors are given for the envelope parameters Tmin and Tmax because these were fitted by eye and were therefore not amenable to objective statistical analysis. Although the sample size was small and a rigorous statistical treatment was not possible, it is noteworthy that the estimates for the fitted parameters KQ and Topt were higher in upper branches than in lower branches, indicating that maximum stomatal conductance was achieved at lower PFD and needle temperature in the lower branches than in the upper branches. Similarly, D2 estimated for lower branches was higher than for upper branches, indicating that lower branches were more sensitive to Ds than upper branches.

Branch Location

a1

D1 (Hyperbolic)

D2 (Linear)

Dynamic model parameters

1 2 3 4 All

1171 (63) 1824 (98) 1298 (68) 1401 (73) 1346 (37)

–0.003 (0.002) .0.002 (0.002) .0 .0.004 (0.002) .0.0009 (0.001)

–0.003 (0.002) .0.002 (0.002) .0 .0.004 (0.002) .0.001 (0.001)

Branch

Location

gsW0 (mmol m –2 s –1)

1 2 3 4 All

Upper Lower Upper Lower

4.66 (0.44) 6.28 (0.51) 9.10 (0.69) 5.88 (0.80) 7.06 (0.42)


Tables 6 and 7 present the values of the time-constant, τ, found from the measurements of stomatal conductance made under rapidly changing environmental conditions. Table 8 gives statistics describing how well the static and dynamic forms of the BWB and Jarvis models fitted the stomatal conductances in the rapidly changing environment data set. In every case, the inclusion of a time-constant term improved the fit of the model to the data. The fitted time-constant term was longer when stomatal

Table 6. Values (and standard errors) of the time-constant, τ (min), found by fitting the BWB model of stomatal conductance to measurements made under rapidly changing environmental conditions on two branches in the upper canopy and two branches in the lower canopy at the BOREAS SSA OBS site in 1996.

Figure 2. Parameter values (and standard errors) of the BWB model of stomatal conductance, estimated from measurements made under steady-state environmental conditions on two branches in the upper canopy and two branches in the lower canopy at the BOREAS SSA OBS site through 1996.

Branch Location τ (all data)

τ (increasing gsW)

τ (decreasing gsW)

1 2 3 4 All

112.3 (139.7) 45.3 (11.5) – 30.7 (11.1) 55.0 (11.8)

22.0 (2.9) 26.1 (2.5) 25.2 (2.7) 17.4 (1.5) 22.1 (1.1)


37.6 (7.9) 33.3 (3.5) 39.6 (6.0) 20.0 (2.3) 29.6 (1.9)

Table 5. Parameter values (and standard errors) of the Jarvis model of stomatal conductance, estimated from measurements made under steadystate environmental conditions on two branches in the upper canopy and two in the lower canopy at the BOREAS SSA OBS site in 1996. Branch

Location

gsWmax (mmol m –2 s –1)

KQ (µmol m –2 s –1)

D2

Topt (°C)

Tmin (°C)

Tmax (°C)

1 2 3 4 All


21.1 (1.9) 33.7 (5.7) 34.1 (3.3) 29.4 (3.8) 28.6 (0.7)

141 (46) 62 (17) 102 (26) 45 (11) 88 (12)

0.012 (0.001) 0.024 (0.003) 0.011 (0.001) 0.015 (0.003) 0.009 (0.0009)

32 (3.8) 24 (6.4) 30 (4.0) 21 (1.2) 27 (1.9)

.2 –4 –5 .2 –5

52 53 52 50 55


720


Table 7. Values (and standard errors) of the time-constant, τ (min), found by fitting the Jarvis model of stomatal conductance to measurements made under rapidly changing environmental conditions on two branches in the upper canopy and two branches in the lower canopy at the BOREAS SSA OBS site in 1996. Branch Location τ (all data) 1 2 3 4 All


τ (increasing gsW)

72.9 (23.6) 169.2 (252.8) 38.3 (5.3) 40.0 (10.7) 47.1 (9.3) 115.6 (88.9) 33.6 (5.0) 28.6 (6.6) 39.8 (3.4) 44.3 (7.5)

τ (decreasing gsW) 40.4 (9.2) 36.8 (5.7) 32.3 (5.8) 38.6 (7.7) 36.5 (3.4)

Figure 3. Values (and standard errors) of the time-constant, τ, found by fitting the BWB model of stomatal conductance to measurements made under rapidly changing environmental conditions for different needle temperatures. Data were grouped into 5 °C classes, and the midpoint value of each class is plotted. Measurements were made on two branches in the upper canopy and two branches in the lower canopy at the BOREAS SSA OBS site in 1996.

conductance was increasing than when it was decreasing, for all of the branches with both the BWB model and for three of the four branches with the Jarvis model. The fitting procedure did not converge for one of the branches when using the BWB model to describe increasing stomatal conductance. Figure 3 shows the value of the time-constant τ fitted to the data classed by needle temperature. A pronounced minimum occurred around 20 °C, indicating that, at this temperature, stomata responded most rapidly to changing environmental conditions.

should have ensured good coupling between the canopy and the atmosphere (and consequently a low Ω; Jarvis and McNaughton 1986), the lack of turbulent penetration to the foliage in the densely packed tree crowns probably meant that the branches were effectively decoupled from the atmosphere, and that control of evaporation was more or less equally partitioned between physiological and aerodynamic controls. Moreover, the small values of the decoupling coefficient commonly quoted for coniferous forest have largely been determined in windy climates such as in the British Isles and New Zealand. In continental interiors, such as this study site, it is not uncommon to have days when the net radiation is large, even though windspeed is very small. Such conditions will tend to produce large boundary layers, and consequently large values for the decoupling coefficient.

Discussion The shoot, branch and canopy conductances measured in this study are compared with literature values in Table 9. Canopy conductance The similarity between the two estimates of canopy aerodynamic conductance (Table 2) is striking, given the simple way in which the physiological conductances were scaled to the whole canopy. The small difference is easily accounted for by the error associated with the optically based measurement of LAI, especially given the highly clumped nature of the tree crowns (Chen et al. 1997). The stand-scale decoupling coefficient, Ω (Jarvis 1985), for this forest stand varied between 0.23 and 0.73, with a mean of 0.52. Typically, Ω for forest canopies is around 0.1 (Jarvis and McNaughton 1986). In the Portuguese maritime pine stand referred to above, Ω varied between 0.05 and 0.15 (Loustau et al. 1996). Although the wide spacing of the black spruce trees

Stomatal conductance Schulze et al. (1994) derived an empirical relationship between maximum stomatal conductance and nitrogen concentration, and predicted that, for the nitrogen concentration found in the trees at this site (6.9 mg g –1, Rayment and Jarvis 1999), maximum stomatal conductance would be 2.07 mm s –1, over twice the maximum stomatal conductance observed. Schulze et al. 1994 also investigated the relationship between

Table 8. Values of R 2 describing the goodness of fit of the BWB and Jarvis models of stomatal conductance to measurements made under rapidly changing environmental conditions in their static and dynamic forms. Measurements were made on two branches in the upper canopy and two branches in the lower canopy at the BOREAS SSA OBS site in 1996. Branch

Location

n

BWB Static R

1 2 3 4 All


148 316 207 272 943

0.41 0.15 0.01 0.39 0.29

Jarvis 2

Dynamic R 0.62 0.53 0.46 0.52 0.55


2

Static R 2

Dynamic R 2

0.17 0.09 0.00 0.02 0.12

0.55 0.43 0.39 0.34 0.45


721

Table 9. Comparison of shoot, branch and canopy conductances from this and similar studies. Shoot and branch conductances are given on a projected leaf area basis, canopy physiological conductance is given on a ground area basis. Measurement scale

Species

Conductance

Reference mmol m –2 s –1

Reported units Shoot Shoot

Picea mariana P. mariana

25.3 mmol m –2 s –1 22.3–38.2 mmol m –2 s –1 a

25.3 28.3–48.5 c

Shoot Shoot Branch Branch Branch Branch Branch Canopy physiological Canopy physiological Canopy physiological Canopy physiological Canopy physiological Canopy aerodynamic

P. mariana P. mariana P. mariana P. mariana Pinus banksiana Lamb. Picea abies (L.) Karst. Pinus sylvestris Picea mariana P. abies Pinus pinaster Ait. Populus tremuloides Abies amabilis Picea mariana

Canopy aerodynamic Canopy aerodynamic Canopy aerodynamic

Various Coniferous forests P. sitchensis

0.1–3.0 mmol g –1 s –1 ~40 mmol m –2 s –1 d 34.1 mmol m –2 s–1 62 mmol m –2 s –1 40–80 mmol m –2 s –1 a 5.4 mm s –1 b 1.4 mm s –1 b 2.5 mm s –1 10 mm s –1 1.7–10 mm s –1 8.4 mm s –1 0.57–7.20 mm s –1 23.5 mm s –1 (dry canopy) 26.3 mm s –1 (wet canopy) 100–1000 mm s –1 22 mm s –1 80–500 mm s –1

29.4–882 c ~ 64 c 34.1 62 51–102 c 216 c 56 c 98 400 c 68–400 c 330 22.8–288 c 940 1051 4000–40000 880 3200–20000

a

This study Middleton et al. 1997 Sullivan et al. 1997 Lamhamedi and Bernier 1994 Dang et al. (1998) This study Rayment 1994, unpublished data Saugier et al. 1997 Morén 2000 Morén 2000 This study Cienciala et al. 1994 Loustau et al. 1996 Blanken et al. 1997 Martin et al. 1997 This study Jarvis et al. 1976 Kelliher et al. 1993 Landsberg and Jarvis 1973

Values reported on a half total needle area basis. b Values reported on a total needle area basis. c Approximate values. d Values reported for conductance to CO2.

maximum stomatal conductance and maximum surface conductance for different land surface covers, and found that surface conductance was typically three times stomatal conductance. The conductances found in this study (gsW = 25.2 mmol m –2 s –1; GW = 72.4 mmol m –2 s –1) follow this prediction. In attempting to partition the total branch conductance between the stomatal and needle boundary layer components, the assumption was made that the stomatal conductances of the shoots and the branches were equal. Because the branch cuvettes contained needles up to 11 years old, whereas the shoot measurements were made on shoots with needles at the most 7 years old, and stomatal conductance declines with needle age (e.g., Ludlow and Jarvis 1971), the total stomatal conductance of the branches was almost certainly lower than that measured on the shoots. Therefore, this assumption probably led to an underestimation of the needle boundary layer conductance. This error would not, however, have been propagated to the estimated physiological conductance of the canopy, because this was based directly on the total conductance measurements made on the branches. Model parameters The large variation over time observed in the BWB model parameter a1 highlights the fact that if models such as these are to be used for modeling purposes, there is a need to make measurements of the processes controlling conductance throughout the growing season. The variation described between branches on the same tree, and between adjacent trees, serves to illustrate that extensive spatial sampling is also required to

describe canopy function adequately. Because conductance in the BWB model is primarily a function of assimilation rate, the finding that the parameter D0 was not significantly different from zero in the BWB model does not necessarily imply that stomata showed no response to the humidity deficit of the air at the needle surface. This is because in this study, which lacked environmental controls, any direct humidity response of stomatal conductance would have been indistinguishable from the consequent change in assimilation rate. In fact, the stomatal response to the humidity deficit of air at the needle surface has been shown to explain most of the variations observed in stomatal conductance for this species throughout the growing season (Dang et al. 1997). The modified BWB model accounted for more variation in measured stomatal conductances than did the Jarvis model, but in their static forms, neither model accounted for more than 41% of the variation when environmental conditions were changing rapidly. The inclusion of a simple transient response improved the fit of both models to the data. The Jarvis environment-driven model benefited most from the addition of a dynamic response, but was outperformed by the dynamic BWB model. Dynamic model The size of the time-constant term, τ, indicates that stomatal conductance in these trees responded to changing environmental conditions slowly. If stomata respond to step changes in environmental conditions in an exponential fashion, as assumed here, it takes about three time constants to achieve 95%


722


of the final response. Taking the estimate of τ generated for all branches combined and increasing and decreasing conductances combined, this means that for the branches, stomata take about 2 h to respond fully to environmental changes. The time-constant term found for the branches in the lower canopy was generally smaller than for branches in the upper canopy. In part, this might be because in the lower canopy, needles are less densely packed on the branches, and are likely to be better coupled to changing environmental conditions. Because branches lower in the canopy experience more rapidly changing conditions as the result of sunflecks, we speculate that the shorter time-constant, together with a higher sensitivity to light, represent an adaptation whereby they can more quickly adjust their conductance when environmental conditions are favorable for photosynthesis. The finding that stomatal response to changing environmental conditions was most rapid around the average daytime temperature conditions (about 20 °C) also suggests that stomatal function has been tuned to local environmental conditions. In conclusion, a combination of controlled environment shoot gas exchange measurements, whole-branch gas exchange measurements, and whole-ecosystem gas exchange measurements provided the means by which total conductance to water vapor for a forest canopy could be partitioned into its component conductances. Continuous gas exchange measurements made with the branch cuvettes allowed us to describe the transient nature of the response of one of these components, stomatal conductance, to changing environmental conditions. We assume that the relatively poor capacity of the models used to describe the stomatal behavior of this species in the field indicates that other factors play an important role in controlling the stomatal function of this species. For example, frost and soil temperature have been shown to have delayed impacts on gas exchange in black spruce (Lamontagne et al. 1998). Long-term continuous measurements of whole-branch gas exchange will allow us to understand the effects of these factors, and should improve our ability to model the land–atmosphere interactions in this globally important biome.

Acknowledgments Eddy covariance data were collected by Jonathan Massheder, University of Edinburgh. The authors thank Steve Scott, Pete Levy, Yadvinder Malhi, Mike Perks, Ford Cropley and John Moncrieff, all at the University of Edinburgh, for keeping the measurement systems running during absences from the field. This work was made possible through funding provided through the NERC TIGER programme, and through the BOREAS project, part of the NASA Mission to Planet Earth.

References Ball, J.T., I.E. Woodrow and J.A. Berry. 1987. A model predicting stomatal conductance and its contribution to the control of photosynthesis under different environmental conditions. In Progress in Photosynthetic Research. Ed. I. Biggins. Martinus Nijhoff Publishers, Netherlands, pp 221–224.

Blanken, P.D., T.A. Black, P.C. Yang, H.H. Neumann, Z. Nesic, R. Staebler, G. denHartog, M.D. Novak and X. Lee. 1997. Energy balance and canopy conductance of a boreal aspen forest: partitioning overstorey and understorey components. J. Geophys. Res. Atm. 102:28915–28927. Chen, J.M., P.M. Rich, S.T. Gower, J.M. Norman and S. Plummer. 1997. Leaf area index of boreal forests: theory, techniques, and measurements. J. Geophys. Res. Atm. 102:29429–29443. Cienciala, E., A. Lindroth, J. Èermák, J.E. Hallgren and J. Kuèera. 1994. The effects of water availability on transpiration, water potential and growth of Picea abies during a growing-season. J. Hydrol. 155:57–71. Dang, Q-L, H.A. Margolis, M.R. Coyea, S.Y. Mikailo and G.J. Collatz. 1997. Regulation of branch-level gas exchange of boreal trees: roles of shoot water potential and vapor pressure difference. Tree Physiol 17:521–535. Dang, Q-L., H.A. Margolis and G.J. Collatz. 1998. Parameterization and testing of a coupled photosynthesis–stomatal conductance model for boreal trees. Tree Physiol. 18:141–154. Dufrêne, E., J.-Y. Pontailler and B. Saugier. 1993. A branch bag technique for simultaneous CO2 enrichment and assimilation measurements on beech (Fagus sylvatica L.). Plant Cell Environ. 16: 1131– 1138. Fischbeck, H.J. and K.H. Fischbeck. 1987. Formulas, facts and constants for students and professionals in Engineering, Chemistry and Physics, 2nd Edn. Springer-Verlag, Berlin, 251 p. Harley, P.C., J.A. Weber and D.M. Gates. 1985. Interactive effects of light, leaf temperature, CO2 and O2 on photosynthesis in soybean. Planta 165:249–263. Jarvis, P.G. 1976. The interpretation of the variations in leaf water potential and stomatal conductance found in canopies in the field. Phil. Trans. R. Soc. Lond. B Biol. Sci. 273:593–610. Jarvis, P.G. 1985. Coupling of transpiration to the atmosphere in horticultural crops: the omega factor. Acta Hortic. 171:187–205. Jarvis, P.G. 1993. Water losses of crowns, canopies and communities. In Water Deficits. Plant Responses from Cell to Community. Eds. J.A.C. Smith and H. Griffiths. Bios, Oxford, pp 285–315. Jarvis, P.G. and K.G. McNaughton. 1986. Stomatal control of transpiration: scaling up from leaf to region. Adv. Ecol. Res. 15:1–49. Jarvis, P.G., G.B. James and J.J. Landsberg. 1976. Coniferous forest. In Vegetation and the Atmosphere. Vol. II, Case studies. Ed. J.L. Monteith. Academic Press, London, pp 171–240. Jarvis, P.G., J.M. Massheder, S.E. Hale, J.B. Moncrieff, M. Rayment and S.L. Scott. 1997. Seasonal variation of carbon dioxide, water vapor, and energy exchanges of a boreal black spruce forest. J. Geophys. Res. Atmos. 102:28953–28966. Jeffree, C.E., R.P.C. Johnson and P.G. Jarvis. 1971. Epicuticular wax in the stomatal antechamber of Sitka spruce and its effects on the diffusion of water vapor and carbon dioxide. Planta 98:1–10. Kelliher, F.M., R. Leuning and E.-D. Schulze. 1993. Evaporation and canopy characteristics of coniferous forests and grasslands. Oecologia 95:153–163. Lamontagne, M., H.A. Margolis and F. Bigras. 1998. Photosynthesis of black spruce, jack pine, and trembling aspen after artificially induced frost during the growing season. Can. J. For. Res. 28:1–12. Lamhamedi, M.S. and P.Y. Bernier. 1994. Ecophysiology and field performance of black spruce (Picea mariana): a review. Ann. Sci. For. 51:529–551. Landsberg, J.J. and P.G. Jarvis. 1973. A numerical investigation of the momentum balance of a spruce forest. J. Appl. Ecol. 10:645–655. Landsberg, J.J. and A.S. Thom. 1971. Aerodynamic properties of a plant complex structure. J. R. Meteorol. Soc. 97:565–570.


CONDUCTANCES OF BLACK SPRUCE Leuning, R. 1995. A critical appraisal of a combined stomatal–photosynthesis model for C3 plants. Plant Cell Environ. 18:339–355. Leverenz, J.W., J.D. Deans, E.D. Ford, P.G. Jarvis, R. Milne and D. Whitehead. 1982. Systematic spatial variation of stomatal conductance in a Sitka spruce plantation. J. Appl. Ecol. 19:835–851. Lohammar, T., S. Larsson, S. Linder and S.O. Falk. 1980. FAST-Simulation models of gaseous exchange in Scots Pine. Ecol. Bull. 32:505–523. Loustau, D., P. Berbigier, P. Roumagnac, C. Arruda-Pacheco, J.S. David, M.I. Ferreira, J.S. Pereira and R. Tavares. 1996. Transpiration of a 64-year-old maritime pine stand in Portugal. 1. Seasonal course of water flux through maritime pine. Oecologia. 107: 33–42. Ludlow, M.M. and P.G. Jarvis. 1971. Photosynthesis in Sitka spruce (Picea sitchensis (Bong.) Carr.). i. General characteristics. J. Appl. Ecol. 8:925–953. Martin, T.A., K.J. Brown, J. Èermák, R. Ceulemans, J. Kuèera, F.C. Meinzer, J.S. Rombold, D.G. Sprugel and T.M. Hinckley. 1997. Crown conductance and tree and stand transpiration in a second-growth Abies amabilis forest. Can. J. For. Res. 27:797–808. Massman, W.J. and M.R. Kaufmann. 1991. Stomatal response to certain environmental factors: a comparison of models for subalpine trees in the Rocky Mountains. Agric. For. Meteorol. 54:155–167. Middleton, E.M., J.H. Sullivan, B.D. Bovard, A.J. Deluca, S.S. Chan and T.A. Cannon. 1997. Seasonal variability in foliar characteristics and physiology for boreal forest species at the five Saskatchewan tower sites during the 1994 boreal ecosystem–atmosphere study. J. Geophys. Res. Atmos. 102:28831–28844. Morén, A.-S. 2000. Modelling branch conductance of Norway spruce and Scots pine in relation to climate. Agric. For. Meteorol. 98–99: 579–593. Moncrieff, J.B., J.M. Massheder, H. deBruin, J. Elbers, T. Friborg, B. Heusinkveld, P. Kabat, S. Scott, H. Soegaard and A. Verhoef. 1997. A system to measure surface fluxes of momentum, sensible heat, water vapour and carbon dioxide. J. Hydrol. 189:589–611.

723

Monteith, J.L. and M.H. Unsworth. 1990. Principles of environmental physics, 2nd Edn. Edward Arnold, London, 291 p. Neilson, R.E. and P.G. Jarvis. 1975. Photosynthesis in Sitka spruce (Picea sitchensis (Bong.) Carr.). iv. Response of stomata to temperature. J. Appl. Ecol. 12:879–891. Ng, P.A.P. 1978. Response of stomata to environmental variables in Pinus sylvestris L. Ph.D. Thesis, Univ. Edinburgh, 200 p. Ng, P.A.P. and P.G. Jarvis. 1980. Hysteresis in the response of stomatal conductance in Pinus sylvestris L. needles to light: observations and a hypothesis. Plant Cell Environ. 3:207–216. Rayment, M.B. 1998. Carbon and water fluxes in a boreal forest ecosystem. Ph.D. Thesis, Univ. Edinburgh, 145 p. Rayment, M.B. and P.G. Jarvis. 1999. Seasonal gas exchange of black spruce using an automatic branch bag system. Can. J. For. Res. 29:1528–1538. Saugier, B., A. Granier, J.-Y. Pontailler, E. Dufrêne and D.D. Baldocchi. 1997. Transpiration of a boreal pine forest measured by branch bag, sap flow and micrometeorological methods. Tree Physiol. 17:511–519. Schulze, E.D., F.M. Kelliher, C. Korner, J. Lloyd and R. Leuning. 1994. Relationships among maximum stomatal conductance, ecosystem surface conductance, carbon assimilation rate, and plant nitrogen nutrition—a global ecology scaling exercise. Annu. Rev. Ecol. System. 25:629–660. Sellers, P.J., F.G. Hall, H. Margolis, R.D. Kelly, D.D. Baldocchi, J. den Hartog, J. Cihlar, M. Ryan, B. Goodison, P. Crill, J. Ranson, D. Lettenmaier and D.E. Wickland. 1995. The Boreal Ecosystem–Atmosphere Study (BOREAS): an overview and early results from the 1994 field year. Bull. Am. Meteorol. Soc. 76: 1549–1577. Sullivan, J.H., B.D. Bovard and E.M. Middleton. 1997. Variability in leaf-level CO2 and water fluxes in Pinus banksiana and Picea mariana in Saskatchewan. Tree Physiol. 17:553–561.
