Sap flow in Scots pines growing under conditions of year-round carbon dioxide enrichment and temperature elevation

Authors


S. Kellomäki Fax: + 358 13 2514444; e-mail: Seppo.kellomaki@forest.joensuu.fi

Abstract

Starting in rrrr, individual trees of Scots pine (Pinus sylvestris L.) aged 30 years were grown in closed-top chambers and exposed to normal ambient conditions (CON), elevated CO2 (Elev. C), elevated temperature (Elev. T) and a combination of elevated CO2 and temperature (Elev. C + T). Using the constant-power heat balance method, sap flow was monitored simultaneously in a total of 16 trees, four for each treatment, over a 32 d period (after the completion of needle expansion and branch elongation in 1997). An overall variation in diurnal sap flow totals (Ft) was evident during the period of measurement (days 167–198, 1997) regardless of the treatments, with a range from 0·15 to 2·82 kg tree–1 d–1. Elev. C reduced Ft by 4·1–13·7% compared with CON on most days (P varies from 0·042 to 0·108), but slightly increased it on some days (P≥ 0·131), depending on the weather conditions. Although the decrease in Ft caused by Elev. C was statistically significant on only a few days (P≤ 0·042), the cumulative Ft for the 32 d decreased by 14·4% (P = 0·047), indicating that Elev. C may have an important influence on seasonal water use of the Scots pine. Analysis of the diurnal courses of sap flow combined with corresponding weather factors indicated that the CO2-induced decrease in Ft could be largely attributed to an increase in stomatal sensitivity to vapour pressure deficit (VPD), whereas the CO2-induced increase in Ft related to an increase in stomatal sensitivity to low light levels. Elev. T increased Ft by 11·2–35·6% throughout the measuring period and the cumulative Ft for the 32 d by 32·5% (P = 0·019), which could be largely attributed to the temperature-induced increase in current-year needle area and decrease in stomatal sensitivity to high levels of VPD. There were no significant interactive effects of CO2 and temperature on sap flow, so that Elev. C + T had approximately the same Ft as Elev. T and similar diurnal patterns of sap flow, suggesting that the temperature factor played a dominant role in the case of Elev. C + T.

INTRODUCTION

The predicted climatic change in northern Europe will involve an increase in the mean annual temperature by 2–4 °C due to enhancement of atmospheric concentrations of CO2 and other greenhouse gases during the next 100 years (IPCC 1996). The rising temperature is expected to reduce soil moisture due to enhanced evapotranspiration (Mitchell et al. 1990). This will be especially true on the southern edge of the boreal zone, where the supply of water could be further limited by reduced accumulation of snow and shortening of the period of snow cover (Pastor & Post 1988; Kellomäki & Väisänen 1996). However, evapotranspiration processes depend on biologically mediated responses to aerial and soil environments that can in turn exercise feedback effects of the vegetation on radiation, energy balance, turbulence and humidity regimes and thereby determine the state of the environment itself. This implies that precise predictions of the water dynamics of the vegetation ecosystem must be based on a proper understanding of the acclimation of vegetational functions to climate changes.

Many experiments have been conducted to examine responses of plants to elevated CO2. These have proceeded from theoretical studies, to small-scale studies in chambers and greenhouses, with branch bags, open-top chambers, and up to large-scale FACE experiments (Ceulemans & Mousseau 1994; Field, Jackson & Mooney 1995). The results show that, at the leaf level, decreased stomatal conductance is observed in most plants grown under high CO2 conditions (Morison 1993; Field et al. 1995), so that it is commonly assumed that increased CO2 will tend to reduce transpiration. However, the measurements of Idso, Kimball & Mauney (1987b) and Kimball, Pinter & Mauney (1992) in cotton show that the CO2-induced decrease in stomatal conductance can lead to an increase in leaf temperature, as a result of which transpiration per unit leaf area may not change. The measurements of Kellomäki & Wang (1996) on Scots pines indicate that long-term growth under elevated CO2 conditions increased stomatal sensitivity to short-term water stress, implying that water loss from the needles will be reduced under heat stress conditions. Thus, it seems that elevated CO2 will affect several mechanisms that influence the use of water by plants, and it is not clear what the resultant changes in transpiration will be (Allen 1990). Transpiration processes are very sensitive to changes in temperature under normal conditions, and, therefore, a rise in the atmospheric temperature will undoubtedly increase water loss in plants. Few data are available, however, on how transpiration processes such as hydraulic resistance and stomatal properties will adapt to the rise in growth temperatures. At the canopy level, elevated CO2 and temperature can lead to changes in canopy structure, for example in the total leaf area index and the distributions of leaf area and the leaf-age classes within the canopy (Chaudhuri, Kirkham & Kanemasu 1990; Radoglou & Jarvis 1990; El Kohen & Mousseau 1994; Overdieck, Kellomäki & Wang 1996; Kellomäki & Wang 1997a). This could alter not only the transpirational surface area per unit ground area, but also the pattern of the microclimate within the canopy, which would increase the difficulty from the leaf scale to the canopy scale. Consequently, CO2-induced transpiration changes have often been found to be inconsistent between area-based and whole tree results, or between daily and seasonal cumulative figures (Jones et al. 1984, 1985; Rosenberg et al. 1990; Hunsaker et al. 1994; Senock et al. 1996). It thus seems that any interpretation of the effects of CO2 and temperature on transpiration must consider the spatial and temporal scales on which the observations are made.

Several methods can be useful to determine transpiration in individual trees, including a chemical tracer (Calder et al. 1986), the heat pulse technique (Devitt et al. 1993; Meinzer et al. 1993; Gutierrez & Meinzer 1994), gravimetric and porometer measurements (Knight et al. 1981; Schulze et al. 1985; Dugas et al. 1993), and a chamber system (Denmead et al. 1993). The constant-power heat balance method for sap flow measurement can nevertheless be used to monitor plant reactions to rapid environmental changes instantly and continuously without disturbing the natural functioning of the plant, and is, therefore, suitable for evaluating the effects of elevated CO2 on the use of water by plants (Dugas et al. 1994, 1997; Hunsaker et al. 1994; Senock et al. 1996).

The economic and social importance of Scots pine (Pinus sylvestris L.) in Finland has given it top priority among the forest tree species when evaluating the effects of changing climate on growth and water use in the boreal vegetation. An experiment was set up in a typical Scots pine stand in eastern Finland using closed-top chambers to evaluate the acclimation of growth and water use in Scots pine trees to elevated CO2 and temperature. This work centred only on quantifying transpiration in individual trees after 1 year of growth under conditions of enriched CO2 and temperature by means of sap flow gauges.

MATERIALS AND METHODS

Site and treatment description

The experiment was established in a naturally seeded stand of Scots pine located near the Mekrijärvi Research Station (62°47′ N, 30°58′ E, 145 m a.s.l.), University of Joensuu, in eastern Finland. Details regarding average climatic conditions and soil type have been reported previously (Wang 1996; Kellomäki & Wang 1997a). Twenty trees of approximately the same crown size and height were chosen and 16 of these were enclosed individually in closed-top chambers in 1996. To reduce the shading to the experiment tree by nearby trees, all nearby trees within 2 m of the chamber were cut down 1 year prior to the experiment running. The chamber is a cylindrical structure with eight walls, an internal volume of ≈ 19·3 m3 and a ground area of 5·2 m2. The four walls facing south and west are constructed of special heating glass (5640 W for each chamber; Eglars Ltd, Finland) and the four north- and east-facing walls of dual-layer acrylic cell glass (standard 16 mm BMMA), with three small vents located in them. Unfiltered air is fed into the chamber by a fan blower through a duct ≈ 3·5 m above the ground, and the air flow through the duct was determined periodically with a hot wire anemometer and adjusted with a butterfly valve. The air flow varied from 0·2 to 0·4 m3 s–1 depending on the season and weather conditions. A computer-controlled heat exchanger linked to a refrigeration unit (CAJ-4511YHR, L’Unit’e-Herlriet’qke, France) was installed in the top of each chamber. A view of the site is shown in Fig. 1.

Figure 1.

. Photograph of the site and its closed-top chambers at Mekrijärvi Research Station, University of Joensuu, Finland.

The combinations of CO2 and temperature resulted in four treatments: (i) ambient temperature and CO2 concentration (CON); (ii) elevated ambient concentration of CO2 (Elev. C); (iii) elevated ambient temperature (Elev. T); and (iv) elevated ambient CO2 and temperature (Elev. C + T). Each treatment was applied to four replicates.

The computer-controlled heating and cooling system, together with a set of magnetoelectric valves (controlling the pure CO2 supply), enabled the two variables of temperature and CO2 concentration to be adjusted automatically inside the chambers to conform to the ambient conditions, or to achieve a specified enrichment in CO2 and/or rise in temperature. The CO2 concentration was enriched all day throughout the year. The warming treatments were designed to correspond to the climatic scenario predicted for the site after doubling of the atmospheric concentration of CO2 (Hänninen 1995; Kellomäki & Väisänen 1997). The CO2 concentration was monitored with a CO2 sensor (GMP111, Vaisala Inc., Helsinki, Finland) located in the middle of the crown of each tree. The relative humidity and temperature within the crown were recorded separately using an RH & T probe (HMP131Y, Vaisala Inc.), in which the relative humidity was measured by a Vaisala ‘Humicap’ sensor. In addition, global solar radiation was measured (Model SKS1110 silicon pyranometer, Skye Instruments, UK) in two layers of the crown (top and middle) and volume soil water content at depths of 5 cm and 15 cm with four soil moisture probes (ThetaProbe ML1, Delta-T Devices Ltd, UK). All the sensors were connected to a data logger. Measurements were taken at 15 s intervals. During the growing season, soil water was supplied to the chambers to correspond to the water content in the ambient soil.

Measurements of sap flow and foliage area

Sap flow was monitored simultaneously over a 32 d period (days 167–198, 1997) in a total of 16 trees, four for each treatment, by the constant-power heat balance method. Sap flow gauges designed for trunks of 6–7 cm in diameter (model SGB50 and SGA70, Dynamax, Inc., Houston, TX, USA) were attached ≈ 0·3 m above the soil surface. Any rough bark and dead tissue was removed carefully, and a thin layer of silicone grease-based electrical insulating compound was spread on the trunk at the point where the gauge was to be installed. After installation, the gauges were covered with clear plastic cling film for waterproofing and aluminium foil was wrapped around both the gauges and sections of stem above and below them to minimize externally induced temperature gradients. The power supply to the gauges, ≈ 1·5 W for the SGB50 model and 2·0 W for the SGA70 model, was switched on from 2 h before sunrise until 2 h after sunset. The thermal conductivity of the stem was assumed to be 0·42 W m–1 K–1 (Van Bavel 1994) for all the trees. The first gauge conductance value (Ksh) was determined from the minimum predawn values, after which Ksh was checked and adjusted every 10 d. The gauge signals were scanned every 30 s with data logger multiplexer units (models 21 x/AM416, Campbell Scientific, Logan, UT, USA) and stored as 30 min averages. Sap flow was calculated automatically according to Eqn 2 of Van Bavel (1994). A software filter was set to eliminate spurious flow calculations at times of low or high flow conditions. As the ‘nightly power down mode’ was used in the measurements, all the analyses of daily sap flow in this paper are based only on measurements made between 0600 h and 2100 h LST (local standard time).

The foliage area of each tree was estimated in August 1997 by the sample needle method, i.e. the length, number, distance from the top, and the angle with respect to the main stem of the branches were first measured for each whorl and the numbers and mean lengths of all the needles on the tree were then determined in age classes and according to location in the crown. After that, ≈ 40 needle fascicles of each age class in each whorl branch were taken at random in order to obtain a fitting for the relationship between the projected area and the length of the needles (Eqn 1 in Kellomäki & Wang 1998). Finally, total needle areas of each whorl and the whole tree were calculated. The projected area of the needles was measured with a needle image system (WinNEEDLETM, Regent Instruments Inc. Quebec, Canada).

Statistical analyses

It was assumed that any significant responses to treatments were due to the growth temperatures and CO2 concentrations and not to unknown chamber effects. Repeated-measures analysis of variance (Moser, Saxton & Pezeshki 1990) was used to test the effects of the growth conditions (CON, Elev. C, Elev. T, and Elev. C + T) and date of measurement on sap flow. One-way ANOVA was used to test the difference in daily sap flow total between the control and enriched treatments on any specific date. To account for treatment-induced differences in needle area and branch growth, sap flow was expressed on the basis of projected needle area, the individual tree and projected area of the crown, respectively. The treatment results expressed in these ways were analysed separately.

RESULTS

Microclimate variables and crown profile

The hourly means of the 15 s readings taken from 1 September 1996 to 1 September 1997 suggest that the target CO2 concentrations and temperatures were, in general, achieved very well (see Fig. 2). The CO2 concentration was 670–730 μmol mol–1 for 86% of the time in the elevated CO2 chambers, and 330–480 μmol mol–1 in the ambient chambers, and the hourly mean air temperature was 2·0–6·2 °C higher than the true ambient temperature for 80% of the time in the elevated temperature chambers, and 0·2–1·5 °C higher in the control chambers. The hourly mean VPD was 0–0·4 kPa higher inside the elevated temperature, but 0–0·3 kPa lower in the control chambers than that outside the chambers. The hourly mean irradiance in the chambers (mean of the two sensors) was reduced by 10–24% in summer and 20–38% in winter compared with the outside levels. There was no significant difference in soil water content among the treatments.

Figure 2.

. Daily courses of incident solar radiation (Rs), ambient CO2 concentrations (Ca), air temperature (Ta), vapour pressure deficit (VPD), and volume soil water content s) in the rooting zone during sap flow measurements (25 July–25 August 1997). The plot is based on half-hourly means of 15 s readings from four replicates in each treatment. VPD was calculated by Eqn 5.13 of Jones (1992). OUT, outside chambers.

The projected needle areas of the trees growing in the chambers ranged from 1·62 to 1·85 m–2 tree–1. Taking the averages of four trees per treatment, the trees grown under Elev. T and Elev. T + C conditions had a 5·6% (P = 0·068) and 5·3% (P = 0·072) increase in total needle area, respectively, compared with the trees under CON, whereas only a 1·4% (P = 0·146) increase was observed under Elev. C conditions. The measurements also indicated that the increased needle area largely resulted from an increase in the length of the current-year needles in the case of both Elev. T and Elev. T + C, but from an increase in the numbers of older needles in the lower layers of the crown in Elev. C (Fig. 3). In addition, elevated temperature also led to a larger growth of current-year shoots than CON. The mean branch length of the whorl with the maximum branch length among all the whorls per stem (SLmax) increased by 7·9%, 6·6% and 1·4% for Elev. T, Elev. T + C and Elev. C, respectively, but the mean angle (φs) of the branches of the whorl with respect to the main stem did not change. Consequently, the projected ground area (Sg) of the crown (m2 m–2 ground), defined as Sg = π[SLmax cos(90°–φs)]2, increased by 16·7%, 13·3% and 3·2% for Elev. T, Elev. T + C and Elev. C, respectively, as compared with CON.

Figure 3.

. Changes (treatment – control) in mean (+ 1 standard error) projected needle area resulting from the Elev. C, Elev. T and Elev. T + C treatments by comparison with the control treatment, by relative height from the top of the crown. Measurements were made on 20–31 August 1997, after the period of needle expansion and branch elongation. * significant difference at the P = 0·05 level in one-way ANOVA.

Sap flow

Overall variation in the diurnal sap flow totals (Ft) was evident during the period of measurement (days 167–198) regardless of the treatments (Fig. 4), with a range from 0·15 to 2·82 kg tree–1 d–1. Statistical analysis showed that when comparison was made on the basis of Ft, temperature had a significant effect on Ft (Table 1), whereas CO2 had no effect on Ft, although the effect of CO2 enrichment was significant on some days (Fig. 4). There was no interactive effect of CO2 and temperature on Ft. As a result, Elev. T + C showed a similar effect on Ft as Elev. T. In addition, the treatment-induced changes in total needle area and crown size did not affect the statistical results above (Table 1). Although the effect of CO2 on Ft is not significant, the cumulative Ft during the 32 d showed a significant decrease under the case of Elev. C (Table 2). Compared with CON, Elev. C led to a 11% reduction (P = 0·048) in the cumulative Ft based on an individual tree, 15% (P = 0·042) on the projected needle area, and 14% (P = 0·047) on crown projected area (Table 2). Even so, the effect of CO2 on the cumulative Ft cannot be seen in Elev. C + T. Consequently, Elev. T + C showed a similar effect on the cumulative Ft as Elev. T.

Figure 4.

. Changes (treatment – control) in mean (four replicates) of daily sap flow (0600–2100 h LST) resulting from the Elev. C, Elev. T and Elev. T + C treatments by comparison with the control treatment (CON), by date of measurement. * significant difference at the P = 0·05 level in one-way ANOVA.

Table 1.  . Summary of P values for total diurnal sap flow (Ft) and cumulative Ft during the period 25 July–25 August 1997. The effects of the treatments on Ft were tested by repeated-measures analysis of variance (Moser et al. 1990) Thumbnail image of
Table 2.  . Cumulative sap flow (Ft) during the period 25 July–25 August 1997 and their statistical significance. The effects of the treatments on Ft were analysed by multifactor ANOVAThumbnail image of

Based on the needle area change, the daily mean of sap flow rates per unit needle area and daily total change in sap flow per tree, contribution of needle area to change in daily total change of sap flow was approximated. Calculations showed that the contribution of the needle area change to the change in Ft depends on treatments and measuring dates. The 2–12% of the change in Ft (kg tree–1 d–1) can be attributed to the increase in needle area under the case of Elev. C, 11–50% under Elev. T, and 10–38% under the case of Elev. C + T.

Diurnal course of sap flow

Four typical diurnal courses of sap flow, representing different combinations of weather factors, were analysed (Figs 5 & 6). Days 167 and 171 represent two typical sunny days with approximately the same total solar radiation (Rs; 5·67 and 5·52 KW m–2 d–1, respectively) and mean diurnal air temperature (17·8 and 17·2 °C in the control chambers, respectively), but with different maximum VPD values (1·3 and 2·4 kPa, respectively; see Fig. 2). In general, three ‘peaks’ can be seen in the diurnal sap flow patterns on both days, the maximum value occurring near solar noon (Fig. 5). Simultaneous measurements of sap flow and Rs indicated that the maximum sap flow rate showed a diurnal lag of ≈ 1·5 h relative to the maximum hourly mean of Rs regardless of the treatments. There were almost no differences in maximum sap flow rates among treatments. On day 171, Elev. T and Elev. T + C had markedly high sap flow rates in the morning and the afternoon hours, whereas Elev. C led to a notable depression in sap flow rates compared with CON. On day 167, the treatment-induced changes in sap flows were relatively small as compared with those on day 171. Particularly, the CO2-induced depression in sap flow rates on day 171 was not observed clearly on day 167.

Figure 5.

. Diurnal courses of sap flow (g H2O m–2 projected needle area h–1) in the CON, Elev. C, Elev. T and Elev. T + C treatments on two typical sunny days, showing means (four replicates) of half-hourly sap flow and one standard error. LST, local standard time.

Figure 6.

. Diurnal courses of sap flow (g H2O m–2 projected needle area h–1) for the CON, Elev. C, Elev. T and Elev. T + C treatments on two typical overcast days, showing means (four replicates) of half-hourly sap flow and one standard error. LST, local standard time.

Days 181 and 182 (Fig. 6) represent two typical overcast days with approximately equal mean diurnal air temperatures (14·8 and 14·6 °C, respectively) and maximum VPD (0·57 and 0·49 kPa, respectively), but slight differences in daily Rs (12·2 and 8·6 mol m–2 d–1, respectively). Comparison with the sap flow pattern on the entirely clear days (Fig. 5) showed that (i) only one ‘peak’ was observed; (ii) the maximum rate of sap flow occurred in the afternoon (about 1600–1800 h LST); (iii) Elev. T and Elev. T + C led to higher sap flow rates throughout the day; and (iv) the enriched CO2 level slightly increased sap flow rates at most hours. As a result, Ft increased by 5·7% and 8·9% as compared with CON at Elev. C on days 181 and 182, respectively, by 16·2% and 28·4% at Elev. T, and by 19·3% and 34·5% at Elev. C + T. In addition, the higher incident Rs in the afternoon on day 181 than on day 182 led to a significant increase in afternoon sap flow rates regardless of treatment, although very close values for VPD.

Responses of sap flow to VPD and Rs

In order to quantify the differences in the responses of sap flow to VPD and light and eliminate the immediate effect of temperature on the sap flow rate in the cases of Elev. T and Elev. T + C, sap flow measurements during the period of days 167–198 were plotted separately against VPD (Fig. 7) and hourly means of Rs under conditions of VPD < 1·0 kPa and an air temperature of 15–18 °C (Fig. 8). We compensated for the lag by comparing each sap flow measurement with the hourly mean VPD and the hourly cumulative Rs for the previous 1·5 h period.

Figure 7.

. Sap flow (g H2O m–2 projected needle area h–1) for the CON, Elev. C, Elev. T and Elev. T + C treatments in relation to the daytime hourly mean of vapour pressure deficit (VPD) over the previous 1·5 h. The data include all measurements made during days 167–198, 1997. The symbols indicate differences in hourly mean radiation (Rs, W m–2). The bold line is fitting a linear equation, F = A1+ B1 VPD to the data at Rs < 300 W m–2; the dashed line is fitting a polynomial equation, F = A2 VPD + B2 VPD2+ C VPD3 to the data at Rs > 300 W m–2. Estimates of parameters in both equations were given in Table 3.

Figure 8.

. Sap flow (g H2O m–2 projected needle area h–1) for the CON, Elev. C, Elev. T and Elev. T + C treatments in relation to hourly mean radiation (Rs W m–2) over the previous 1·5 h. The data include measurements made under conditions of VPD < 1·2 kPa and hourly mean temperatures ranging from 15 to 18 °C.

Plotting sap flow against VPD showed it to have a strong linear response to an increase in VPD at low VPD levels. After the maximum values it was only slightly sensitive to the change in VPD, however (Fig. 7). As the linear part showed a clear response to low light levels and the curved part to high light levels, the data can be clearly divided into two parts at a critical value for Rs of ≈ 300 W m–2. A linear fitting to the data when Rs < 300 W m–2 showed that there were no significant differences in the slopes (P = 0·14–0·43) and interceptions (P = 0·09–0·26) between the treatments (Table 3), i.e. the treatments did not lead to any change in the relationship of sap flow to VPD under low light conditions. Thus, the treatment-induced modification can be seen only at high levels of VPD, so that by comparison with that in CON, the trees growing in Elev. T and Elev. T + C still maintained high sap flow rates at high levels of VPD, whereas those of the trees growing in Elev. C were largely depressed by high VPD.

Table 3.  . Parameter estimates (± standard deviation) for F = A1+ B1 VPD (Rs < 300 W m–2), and for F = A2 VPD + B2 VPD2+ C VPD3 (Rs≤ 300 W m–2) based on the data in Fig. 7. Where F is the sap flow rate and A, B, C are three parameters. The effects of the treatments on the parameters were analysed by multifactor ANOVAThumbnail image of

The plotting of sap flow against Rs showed that when the diurnal hourly mean for VPD was greater than 1·2 kPa, Rs was usually > 300 W m–2 in all the chambers (maximum 350 W m–2). Under these conditions sap flow was relatively insensitive to changes in Rs, but when the daytime VPD was less than 1·2 kPa, there was an approximately linear relationship between sap flow and Rs, with r2 values ranging from 0·72 to 0·84 (Fig. 8). The treatments had an evident affect on the slope and interception of the lines. Elev. C increased the slope significantly compared with CON (P = 0·043) and similarly the interception (P = 0·039), but Elev. T and Elev. C + T had only a marginal influence on the slopes or interceptions.

DISCUSSION AND CONCLUSIONS

Overview

In this experiment the sap flow in single trees of Scots pine, growing in environment-controlled chambers at their natural site, was monitored and analysed. The experimental treatments were similar to those employed in our previous experiment with open-top chambers (Wang 1996; Kellomäki & Wang 1997a). The differences in the case of the new closed-top chambers are that: (i) the powerful air-conditioning system enabled the target temperature conditions to be controlled more precisely, especially on summer days; (ii) the distribution of CO2 concentrations in the chambers was more uniform, and CO2 was enriched all day (24 h) and throughout the year rather than only in the daytime and during the growing season; (iii) the soil water content in the chambers basically followed the same pattern as the ambient soil water at the site as a result of efficient watering; and (iv) light conditions inside the chambers were greatly improved by using more transparent materials. Thus, the results can be expected to reflect better the responses of Scots pines to future climate change. Some effects of the chamber on the microenvironment, such as those on the spectral distribution of solar radiation, air humidity, rate and direction of air flow, and the degree to which these effects alter the interpretation of results obtained from this experiment, were not quantified here.

Effects of CO2 enrichment

Many measurements at the leaf level show that transpiration is reduced by growth under CO2-enriched conditions on account of a CO2-induced decrease in stomatal conductance (Morison 1993; Field et al. 1995). It is commonly thought, however, that the enrichment in the atmospheric concentrations of CO2 may not reduce whole-plant or canopy transpiration (Eamus 1991; Morison 1993) because of other compensating effects under elevated CO2, such as increased leaf area (Mauney et al. 1994) or elevated leaf temperature (Idso et al. 1987b; Kimball et al. 1992; Nederhoff, Rijsdijk & de Graaf 1992), which may cause increased leaf-air vapour pressure deficits (Jones et al. 1984, 1985). Thus, it seems to be necessary to distinguish the effect of CO2 on leaf properties and crown structure in order to understand the response of transpiration on elevated CO2. Sap flow was expressed separately in our analysis on the basis of needle area and on the basis of the individual tree, the results showing that when comparison was made on the basis of needle area, the trees grown under Elev. C conditions had a lower diurnal sap flow total on most days than those growing under CON conditions, although the decrease was significant only on a few days (Fig. 4). As Elev. C conditions led to only a 1·4% increase in total needle area per tree and a negligible change in mean branch length, the sap flow reduction attributable to elevated CO2 was similar on a per tree, per m2 needle area, or per unit ground area basis (Table 1). We also noted in our previous experiment, however, that 4 years of growth under elevated CO2 conditions led to about a 20% increase in the total needle area of Scots pines (Kellomäki & Wang 1997a). In the light of this change, the CO2-induced sap flow reduction per tree could be offset by a CO2-induced increase in needle area. This provides further evidence that the CO2-induced reduction in transpiration per unit needle area may be beneficial for most trees (Ceulemans & Mousseau 1994) and crop species (Kimball 1983; Allen 1990), although the whole-plant transpiration results may be variable because the CO2-induced changes in needle area and canopy structure are more dependent on nutrition and water conditions (Norby & O’Neill 1991; Johnsen 1993), plant population density (Reekie & Bazzaz 1989), and exposure time for the tree species (Kellomäki & Wang 1997b).

If we assume that the CO2-induced depression in stomatal conductance is only a cause that leads to a reduction in transpiration, then a low sap flow rate should be recorded throughout the measurement period in the case of CO2 enrichment. The present diurnal patterns of sap flow nevertheless showed that trees growing under conditions of CO2 enrichment did not always have low rates of sap flow (see Figs 5 & 6), or a low diurnal sap flow total on all measuring dates (Fig. 4). A similar phenomenon can also be seen in the sap flow patterns recorded for Sorghum bicolor (L.) Moench and Glycine max (L.) Merr. growing under elevated CO2 conditions (Dugas et al. 1997). The translation to this variation in sap flow with time or date may be a complex process, perhaps owing to the lack of representativeness of leaf conductance measurements for the whole crown (Dugas et al. 1994), the difference in leaf temperature between treatments (Kimball et al. 1992), the accuracy of the gauges, or a combination of these factors (Senock et al. 1996). When we compared the measurements of sap flow with those of solar radiation and VPDs, however, we found that a CO2-induced increase or lack of change in sap flow was largely associated with low levels of solar radiation, whereas a CO2-induced decrease was associated with a high demand for transpiration, because the largest decrease often occurred in the afternoon hours (see Fig. 5). This was further confirmed by plotting sap flow measurements against solar radiation at VPD < 1·2 and hourly mean temperatures of 15–18 °C (Fig. 8) and against VPDs as a whole (Fig. 7). Significantly, when both plots (Figs 7 & 8) were compared with our earlier results at the needle level, in which the responses of stomatal conductance in Scots pine to light and VPD were measured (cf. figs 1 & 2 in Wang & Kellomäki 1997), the results showed that both sap flow and stomatal conductance had a good consistency in their response to light and VPD. Thus, the possible interpretations are that the CO2-induced increase in sap flow at low levels of radiation or on cloudy days could be attributed mainly to the CO2-induced increase in the sensitivity of stomatal conductance to low levels of light, whereas the CO2-induced decrease in sap flow is largely attributable to the CO2-induced increase in the sensitivity of stomatal conductance to high VPD. Similarly, the occurrence of the largest decrease in sap flow in the afternoon hours could be related to the substantial midday depression in stomatal conductance (Garcia et al. 1993; Wang 1996).

Effects of elevated temperature

The present treatments with elevated temperature led to significantly higher sap flow rates than in the control treatment regardless of weather conditions (Tables 1 & 2). This increase in sap flow can clearly be attributed to three things: the larger needle area per tree (Fig. 3), the higher immediate air temperature during monitoring (Fig. 2), and the acclimation of transpiration processes to growth temperatures.

The contribution of increased needle area to sap flow was approximated on the basis of the mean sap flow rate per unit needle area. This may underestimate the contribution of increased needle area, as the increased area mainly came from the larger growth of current-year needles, but it should be noted that the sap flow measurements were made after only 1 year of the experiment. For coniferous species bearing needles from many years, this exposure time is certainly not sufficient to obtain an understanding of the change in needle area (Kellomäki & Wang 1997b).

Depending on weather conditions, the hourly mean air temperature was 2–5 °C higher in the chambers with elevated temperature than in the control chambers during the measurements of sap flow (Fig. 2). It is difficult from the current measurements alone to quantify the effect of the immediate increase in air temperature on sap flow, but we can see from diurnal courses of sap flow that a rise of several degrees in air temperature had a larger relative effect on the increase in sap flow under conditions of low radiation and high relative humidity (e.g. in the morning hours or on a cloudy day) than on the majority of days in the growing season.

To eliminate the effect of the increase in immediate air temperature and identify the acclimation of transpiration to growth temperature, the proper methods to employ are comparison of the sap flow rates of trees growing at an elevated temperature with those in the control chambers in the presence of the same temperature variation, or by analysing the diurnal pattern of sap flow in combination with corresponding environmental variations. When hourly sap flow measurements for 32 d (days 167–198) were plotted against hourly mean air temperature, an approximate linear relationship can be seen, but the fit showed that the relationship was not significant due to large residual sum of squares (P = 0·64 for Elev. T and P = 0·77 for CON). Another alternative parameter is VPD, a complex variation of relative humidity and temperature. If it is assumed that there are no differences between the treatments in wind speed (influencing needle boundary layer conductance) or radiation inside the chambers, comparison of sap flows at the same VPD will give some significant information about the acclimation of transpiration to growth temperature. The plotting of sap flow against VPD showed that trees growing at an elevated temperature can still maintain constant transpiration when VPD is over a critical value (≈ 1·2 kPa; Fig. 7). This effect can also be observed in diurnal sap flow patterns, where sap flow rates had a relatively small decrease in the midday and afternoon hours when VPD was high (Fig. 5). The translation to this relative insensitivity of sap flow to high VPD requires a combination with other simultaneous measurements such as needle temperature (Idso et al. 1987b), needle water potential, needle conductance and hydraulic resistance (Sperry & Pockman 1993). Our early measurements on needles of Scots pine indicated that temperature elevation significantly reduced the sensitivity of stomatal conductance to decreasing needle water potential under short-term water stress conditions (Kellomäki & Wang 1996), and to increasing VPD (Wang & Kellomäki 1997). It is therefore certain, at least, that temperature-induced modifications in needle water potential and adjustments in the stomatal conductance of Scots pine needles are two important causes of transpiration in the whole tree remaining higher under short-term stress conditions.

Interaction of CO2 with temperature elevation

As discussed above, the acclimation of transpiration to a single factor, CO2 or temperature, can be largely interpreted in terms of changes in leaf area and stomal properties. Similarly, it is often observed in combined experiments that the increase in leaf area caused by growth in elevated CO2 is enhanced by elevated temperature (Idso et al. 1987a; Overdieck et al. 1996), although stomatal conductance yields inconsistent results (Beerling & Woodward 1996; Santrucek & Sage 1996; Wang & Kellomäki 1997). In an experiment conducted in a catchment area with a boreal vegetation, Beerling (1998) found that increased CO2 (560 ppmv) and elevated temperature (+ 5 °C in winter, + 3 °C in summer) together reduced stomatal conductance, which was attributed to acclimation of the stomata to CO2, but increased transpiration in the vegetation, which was attributed to greater VPD as a result of the warmer air temperatures and reduced soil moisture. It thus seems that the stomata show a significant acclimation to elevated CO2 but not to the smaller increase in air temperature. In the present experiment the increase in needle area that resulted from the combination of elevated temperature and CO2 was almost the same as that brought about by elevated temperature alone, i.e. the increase came only from the greater growth of current-year needles. Likewise, the diurnal patterns of sap flow and the responses of sap flow to environmental variations were also similar to those occasioned by elevated temperature alone. These observations are consistent with the results obtained at the individual leaf level in our previous experiment (Kellomäki & Wang 1996; Wang & Kellomäki 1997). Thus, there were no significant differences in total daily sap flow or cumulative sap flow for the 32 d (days 167–198) between the combined treatment and elevated temperature alone (Tables 1 & 2). The diversity among experiments could be related to differences in species, experimental conditions, methods, exposure time, etc., but it must be recognized that the temperature elevation used in this experiment was markedly high in winter (a mean increase of 6 °C), which may have important consequences for processes other than water loss, for example, heat damage, increased rate of development, stimulation of senescence, etc. These effects will be evaluated in later papers.

In conclusion, 1 year of growth under conditions of a doubled ambient concentration of CO2 led to a decrease in total diurnal sap flow in individual Scots pine trees on most days of the measurement period. Although the decrease in total diurnal sap flow was significant only on a few days, the reduction in cumulative sap flow over the 32 d (days 167–198) was significant. The trees growing at an elevated temperature always had higher sap flow rates, so that significant increases were observed in total diurnal sap flow and cumulative sap flow for the 32 d. There was no significant interaction between CO2 and temperature factors, and sap flow actually showed similar responses to the combination of CO2 and temperature and to elevated temperature alone. The temperature factor obviously played the dominant role in the combination of CO2 and temperature effects in the present experimental setup.

Although sap flow measurements made continuously over the main growing season can be useful for evaluating the transpiration acclimation of individual trees to growing conditions, it is often difficult to arrive at a proper interpretation of the variation in sap flow because this requires the simultaneous quantification of several other plant and environmental factors related to transpiration processes. We explored the potential of sap flow measurements as a means of inferring stomatal responses to changes in environmental conditions, and the results suggested that variations in sap flow caused by growth under conditions of elevated CO2 and temperature can be largely understood in terms of changes in stomatal properties and needle area, which is consistent with results obtained at the individual needle level in our previous experiment, although the approach adopted then was relatively simplistic.

For forest tree species, acclimation of the crown architecture and stomata to growing environments constitutes a slow but dynamic process. This implies that continual monitoring of sap flow will be needed in the future, and a more powerful approach will have to be developed in order to estimate the energy and heat flow distribution within the crown and the dynamics of crown conductance.

Acknowledgements

This work forms part of the research project ‘Likely Impact of Elevating CO2 and Temperature on European Forests’ co-ordinated by Professor Paul Jarvis, University of Edinburgh, Scotland. The research was also funded through the Academy of Finland and the University of Joensuu. We wish to express our appreciation to Jorma Aho for providing the facilities at the Mekrijärvi Research Station, and we also thank Matti Lemettinen and Alpo Hassinen for developing and maintaining the experimental infra-structure.

Ancillary