Acclimation of photosynthetic capacity in Scots pine to the annual cycle of temperature

Transcription

1 Tree Physiology 24, Heron Publishing Victoria, Canada Acclimation of photosynthetic capacity in Scots pine to the annual cycle of temperature ANNIKKI MÄKELÄ, 1,2 PERTTI HARI, 1 FRANK BERNINGER, 1 HEIKKI HÄNNINEN 3 and EERO NIKINMAA 1 1 Department of Forest Ecology, P.O. Box 27, FIN-00014, University of Helsinki, Finland 2 Corresponding author (annikki.makela@helsinki.fi) 3 Department of Ecology and Systematics, P.O. Box 65, FIN-00014, University of Helsinki, Finland Received June 9, 2003; accepted September 28, 2003; published online February 2, 2004 Summary Coniferous trees growing in the boreal and temperate zones have a clear annual cycle of photosynthetic activity. A recent study demonstrated that the seasonal variation in photosynthetic capacity of Scots pine (Pinus sylvestris L.) could be attributed mainly to the light response curve of photosynthesis. The magnitude of the light response curve varied over the season while its shape remained constant, indicating that the two physiological parameters quantifying the curve the quantum yield per unit internal carbon dioxide concentration and the corresponding light-saturated rate remained proportional to each other. We now show, through modeling studies, that the quantum yield (and hence the light-saturated rate) is related to the annual cycle of temperature through a delayed dynamic response. The proposed model was tested by comparing model results with intensive measurements of photosynthesis and driving variables made from April to October in three shoots of Scots pine growing near the northern timberline. Photosynthetic capacity showed considerable acclimation during the growing season. A single model describing photosynthetic capacity as a reversible, first-order delay process driven by temperature explained most of the variation in photosynthetic capacity during the year. The proposed model is simpler but no less accurate than previous models of the annual cycle of photosynthetic capacity. Keywords: annual cycle, dynamic model, optimal stomatal control, shoot photosynthesis. Introduction Coniferous trees growing in the boreal and temperate zones have a clear annual cycle of photosynthetic activity. The rate of photosynthesis is low or zero during winter, increases during spring, peaks during summer and decreases during autumn. Part of this cycle can be explained by direct responses to photosynthetically active radiation (PAR) and air temperature, but it has long been known that the inherent photosynthetic capacity of the needles (i.e., the rate of photosynthesis achieved under stable reference conditions) also changes during the year (Pisek and Winkler 1958, Linder and Lohammar 1981). The springtime recovery of photosynthetic capacity has been attributed to a delayed effect of rising air temperatures (Pelkonen and Hari 1980, Schaberg et al. 1995, Bergh et al. 1998) or to the thawing of frozen soil (Bergh and Linder 1999). Furthermore, night frosts depress photosynthesis the following day, and the effect of a severe night frost may be visible for several days (Polster and Fuchs 1963, Bergh et al. 1998). Pelkonen and Hari (1980) developed a model to simulate the acclimation of photosynthetic capacity to air temperature in Scots pine (Pinus sylvestris L). Photosynthetic capacity followed air temperature with a delay, rapid fluctuations were filtered out, and the recovery was completely reversible. The model gave a reasonable fit to field measurements of shoot photosynthesis. Repo et al. (1990) presented a similar model for the annual cycle of frost hardiness of trees that described the development of frost hardiness as a reversible, first-order delay process driven by air temperature. Bergh et al. (1998) developed a model to analyze climatic factors controlling productivity of Norway spruce that uses different driving variables for the spring, summer and autumn seasons, including air temperature and soil frost during spring and frost occurrence in the autumn. Hänninen and Hari (2002) recently carried out a detailed comparison of the models developed by Pelkonen and Hari (1980) and Bergh et al. (1998). Hari and Mäkelä (2003) analyzed the seasonal variation of individual shoot photosynthesis in an extensive data set on boreal Scots pine with a model (Hari et al. 1986, 1999) based on the optimal stomatal control theory (Cowan and Farquhar 1977). The model used PAR, saturation deficit of water vapor, and ambient temperature as driving variables. Hari and Mäkelä (2003) attributed the seasonal variation in photosynthetic capacity to two model parameters: the quantum yield per unit plant internal carbon dioxide (CO 2 ) concentration, α (henceforth called quantum yield) and, to a lesser extent, the carbon cost of transpiration, λ, i.e., the amount of carbon invested to take up the water lost in transpiration (mol CO 2 (mol H 2 O) 1 ). The saturation level of the light response curve also varied, but the variation was proportional to that of α, such that

2 370 MÄKELÄ, HARI, BERNINGER, HÄNNINEN AND NIKINMAA the shape of the light response curve remained constant. Taking these variations into account allowed for accurate predictions of daily photosynthetic production, with a percentage of explained variance (PEV) > 95% (Hari and Mäkelä 2003). In this study, we used the measurements and analysis reported by Hari and Mäkelä (2003) to examine the relationship between the seasonal variation in temperature and the seasonal variation in photosynthetic capacity. Based on the results of Hari and Mäkelä (2003), we sought to determine if it is possible to explain seasonal changes in quantum yield, α, as a dynamic delayed response to air temperature. For this purpose, we used a modified version of earlier models of the reversible annual cycle of development (Pelkonen and Hari 1980, Repo et al. 1990). Materials and methods Measurements Empirical daily values for α were estimated based on measurements of photosynthetic rates and environmental factors reported in Hari and Mäkelä (2003), and measurements of air temperature. The primary data were collected from the end of April to late October in 1997 at SMEAR I (Station for Measuring Forest Ecosystem Atmosphere Relations) in Finnish Lapland (67 46 N, E) in an approximately 50-year-old Scots pine stand with about 1000 trees ha 1 and a dominant height of 8 m. Translucent, acrylic measurement chambers (3.6 dm 3 ) were attached to three 1-year-old shoots about 7.5 m above ground level. Measurements of CO 2 concentration, water vapor, air temperature inside the chambers and PAR immediately outside the chambers were taken 120 times per day, except when there were interruptions caused by failure of the electricity supply, giving a total of 54,000 measurements. A detailed description of the measuring system is given in Hari et al. (1999), and details of the data have been provided by Hari and Mäkelä (2003). Model of gas exchange The gas exchange model that we used (Hari et al. 1986, Mäkelä et al. 1996, Hari and Mäkelä 2003) is based on the idea of optimal stomatal control proposed by Cowan and Farquhar (1977), but the method of solving the optimization problem is different, allowing for more explicit expressions of the assumptions and results. The basic assumptions and concepts of the model are reviewed below, and detailed equations and analysis can be found in Hari and Mäkelä (2003). (1) Gross photosynthetic rate (A(t); mol CO 2 m 2 s 1 ) at time t (s) is proportional to leaf internal CO 2 concentration (C i ; mol CO 2 m 3 ) multiplied by a saturating response ( f(i); m s 1 )to PAR (I; mol m 2 s 1 ): αγit () At () = f(()) It Ci() t = Ci() t αit () + γ where γ (m s 1 ) is the saturation level of f(i) and α (m 3 mol 1 ) is the initial slope of the function. (1) (2) Carbon dioxide is taken up by the leaf by diffusion through the stomata, and net uptake equals photosynthesis minus respiration: At () = gt ()( Ca Ci ()) t + r() t (2) where g (m s 1 ) is stomatal conductance; C a (mol CO 2 m 3 )is ambient CO 2 concentration; and r (mol CO 2 m 2 s 1 ) is respiration, which depends on leaf temperature (T l ; C) and two parameters, r 0 (mol CO 2 m 2 s 1 ) and Q 10 : r(t l ) = r 0 Q 10 T l/10. (3) Transpiration (E; mol H 2 Om 2 s 1 ) is proportional to the saturation deficit of water vapor (D; mol H 2 Om 3 ) and g: Et () = agtdt () () (3) where a is the ratio of the diffusion rates of water and carbon dioxide and D is the difference between ambient water vapor concentration and saturated water vapor concentration at the prevailing leaf temperature. Leaf temperature is assumed to be greater than ambient temperature, T a, by a term proportional to irradiance: T l = T a + bi. (4) Following Cowan and Farquhar (1977), stomatal conductance is determined for all times and driving variables as the conductance that maximizes carbon gain over a specified time period: t Max At () λ Et () dt (4) 0 where λ (mol CO 2 (mol H 2 O) 1 ) is the carbon required in the long term (several days) to sustain the transpiration flow E(t). Combining Equations 1 and 2 allowed us to eliminate C i and express A(t) in terms of C a, I, T and g. The optimal stomatal conductance can then be found using methods of dynamic optimization, and it becomes a function of all the driving variables of the model. In addition, g(t) is constrained by a minimum (g min ) determined by cuticular transpiration, and a maximum (g max ) corresponding to fully open stomata (Hari et al. 1986, Hari and Mäkelä 2003). Hari and Mäkelä (2003) estimated the parameters for the model with the data set used in this study. They found that all but two parameters of the model could be held constant over the growing season (Table 1). The most significant effect of the annual cycle could be attributed to a change in the magnitude of the light response curve (Equation 1), apparently following the annual cycle of temperature. The change in the magnitude of Equation 1 takes place such that both α and γ have an annual cycle but simultaneously γ =cα, where c is a constant. It is, therefore, sufficient to consider the seasonal variation of one of these parameters only. In addition, a slight indication of a declining trend in the parameter indicating the cost of water (λ) was detected, but this was not statistically significant. Therefore, we held λ constant in this study (Table 1). TREE PHYSIOLOGY VOLUME 24, 2004

3 ACCLIMATION OF PHOTOSYNTHESIS TO TEMPERATURE 371 Table 1. Parameter values for the photosynthesis model (Hari and Mäkelä 2003). Abbreviations: PAR = photosynthetically active radiation; and α = quantum yield. Parameter Definition Units Value g min Minimum conductance (cuticular) m s g max Maximum conductance (stomata fully open) m s b Increase in leaf temperature per unit PAR C mmol 1 m 2 s 8 1 r 0 Leaf specific respiration rate at 0 C µmol m 2 s Q 10 Relative increase in respiration per 10 C 2.3 a Ratio of H 2 O to CO 2 diffusion rates 1.6 λ Cost of water in units carbon mol CO 2 (mol H 2 O) c γ = cα in the light response curve µmol m 2 s Note that the ambient air temperature was measured inside the chamber with constant mixing of air, and the transpiration rate is low and thus the proportion of latent heat of evaporation from the leaf heat balance is also low. Model of the annual cycle Based on the above results, the annual pattern of photosynthesis can be predicted if the change in α (and γ) is modeled throughout the growing season. We propose a dynamic acclimation model for predicting α from the development of ambient temperature. We define the state of photosynthetic acclimation, S, as an aggregated measure of the state of those physiological processes of the leaves that determine the current photosynthetic capacity at any moment, and assume that its development over time is driven by temperature. Because S is an abstract aggregate variable that can be normalized appropriately, S has the same units as temperature. Describing the slow process of acclimation, we postulate that S follows leaf temperature (T; C) in a delayed manner: if T is held constant, S approaches T, and if T is changed, S will move toward the new temperature with a time constant τ (Figure 1). This gives rise to the following dynamic model for S: Figure 1. Leaf temperature, T (thin line), initially at T 0, is raised to T f at time t 0. The state of photosynthetic acclimation, S (thick line), which is initially at equilibrium with T, approaches the new temperature asymptotically with the time constant τ = 12 days (dashed vertical line). ds dt = 1 T S τ ( ) (5) where τ (h) is a time constant. We calculated the predicted photosynthetic capacity, α, assuming a linear relationship between α and S: α ( S) = max{ c 1 ( S S 0 ), 0 } (6) where S 0 ( C) is a threshold value of the state of acclimation and c 1 is a coefficient of proportionality. This formulation is similar to that proposed by Pelkonen and Hari (1980). The state of acclimation was defined by these authors with a more complicated differential equation, but was essentially driven by temperature, as in the present model. In contrast, Repo et al. (1990) used the linear model of Equation 5 to derive the annual development of frost hardiness in Scots pine. Parameter estimation Predicting the seasonal course of photosynthesis with the model involves estimating three parameters: the time constant, τ, the threshold temperature, S 0, and the constant, c 1. The data consist of the daily estimates of α, denoted by α(t i ) for day t i, in the three measurement chambers (Hari and Mäkelä 2003); ambient temperatures measured by a thermocouple in each chamber just after closure (such that chamber closure does not considerably affect the temperature); and photosynthetic photon flux measured with a PAR sensor (LZ190, Li-Cor, Lincoln, NE) just outside each chamber to convert ambient temperature to leaf temperature. In addition, the mean of all chambers was analyzed. The time course of the state of development, S, was calculated for several values of τ. Each of the time developments was calculated by simulating Equation 5 at a 20-min time step from the initial state S(0) = 10 C. We chose a starting date early in the year to avoid affecting the results. To predict the daily values of S, we denote the result of each simulation at noon of day t i by S τ (t i ). The corresponding parameter esti- TREE PHYSIOLOGY ONLINE at

4 372 MÄKELÄ, HARI, BERNINGER, HÄNNINEN AND NIKINMAA mates, c τ1 and S τ0 (Equation 6), were determined by minimizing the residual sum of squares, SS τ, for each time constant τ: SS τ = min ( α ( t α i ) τ( t i )) i where α τ ( t i ) is the result of Equation 6 when the time constant has the value τ. The value of τ minimizing SS τ and the corresponding values c τ1 and S τ0 were chosen as the best estimates. The goodness of fit was measured in terms of PEV, defined as: PEV = SS tot SS SS tot res where SS tot is total variance in the explained variable and SS res is residual variance after model fitting. In a simple linear regression, PEV is equivalent to r 2. Results The best-fit time constant of the temperature response varied little between chambers, but the quantum yield of Chamber 0 Figure 2. Time course of quantum yield, α, over the growing season in three measurement chambers. The daily value of α was estimated by fitting the model of Hari and Mäkelä (2003) to measurements of photosynthesis for each day separately, keeping the rest of the parameters at constant values for the whole season (Table 1). 2 (7) (8) remained slightly lower than that of the other two chambers (Figure 2, Table 2). For the mean of all chambers, the best-fit time constant was 330 h and the regression explained 92% of the variation in mean chamber α calculated for each day. We tested the sensitivity of the results to τ by plotting predicted α τ ( t i ) for short, optimum and long response times (Figure 3). The predicted variation in α has a wider amplitude if τ is small rather than optimum. Also, the predicted change takes place earlier than the observed change. If τ is large, the opposite is true. Even with the best-fit parameterization, the empirical value (i.e., the value estimated from measurements of photosynthetic rates and environmental factors) of α manifests a lot of variation around the model prediction α, which is smoother (Figure 3b). The variation could be caused by statistical error in the observed α( t i ), or by effects of other factors not included in the model. To test if there was another significant seasonal trend not covered by the model, the residuals were plotted against time over the growing season, but no trend was detected (Figure 4a). Similarly, no trend was detected with respect to mean daily irradiance (Figure 4b). The possibility of direct temperature effects was tested in three ways. First, we analyzed the residuals against daily mean temperatures, but no systematic features were detected (Figure 4c). Second, we assessed the possibility that low nighttime temperatures combined with high radiation in the early morning would reduce photosynthesis more than predicted by the model because of damage to the photosynthetic system (Krivosheeva et al. 1996). We calculated the mean temperature on summer mornings (June 15 to August 30) when the zenith angle was between 0.1 and 0.3 rad, selected mornings with high radiation only, and plotted residuals against morning temperature. There was a slight trend in the residuals (R = 0.54, n = 37, P < 0.001) for the selected mornings (Figure 5). In the few cases when there was a night frost, the rate of early morning photosynthesis seemed to fall below that predicted by the model (Figure 6a), although the model accurately described the daily course of photosynthesis on sunny days without morning frosts (Figure 6b). Finally, we calculated the course of mean daily photosynthesis based on the photosynthesis model of Hari and Mäkelä (2003) with the parameter values used in this study (Table 1), such that the daily estimates of α(t i ) were replaced by the best-fit predictions α (Figure 7). When compared with daily photosynthesis measurements, this prediction gave a PEV = Table 2. Best fit parameter values for estimating predicted photosynthetic capacity ( α) in the three chambers and for the pooled data (Chambers 0, 1 and 2 combined). Abbreviations: τ = time constant; c 1 = coefficient of proportionality; S 0 = threshold value of state of acclimation; PEV = percentage of explained variance; and SS τ = residual sum of squares for τ. Chamber τ (h) c 1 (m 3 mol 1 C) S 0 ( C) PEV SS τ (m 3 mol 1 ) Pooled data TREE PHYSIOLOGY VOLUME 24, 2004

5 ACCLIMATION OF PHOTOSYNTHESIS TO TEMPERATURE 373 Figure 4. Residuals of the model as a function of (A) time and (B) daily mean irradiance (PAR = photosynthetically active radiation). Figure 3. Comparison of empirical (estimated on the basis of measurements (mean of three chambers)) and predicted quantum yield, α, for three different values of the time constant τ. (A) τ = 60 h, percentage of explained variance (PEV) = 0.78; (B) τ = 330 h, PEV = 0.92 (best fit); and (C) τ = 750 h, PEV = Discussion Physiological aspects of the annual cycle An annual cycle is characteristic of many physiological and morphological attributes of trees growing in cool and temperate zones. This is the case also with the photosynthetic machinery, as demonstrated by this study and earlier studies (Pisek and Winkler 1958, Pelkonen and Hari 1980, Bergh et al. 1998). The photosynthetic data of the present study were obtained with chamber measurements carried out at the shoot level (Hari and Mäkelä 2003). When analyzed with a model based on the theory of optimal control of stomata (Hari et al. 1986), a clear annual pattern was found for α, which describes the capacity of the photosynthetic machinery. The dependence of α on ambient temperature seemed to yield good predictions of photosynthesis over the entire growing season, even though the suggested physiological, biochemical and biophysical mechanisms responsible for springtime recovery and summertime fluctuations of photosynthesis differ widely. Although these mechanisms cannot be inferred from our shoot-level data, some aspects can be assessed in light of our results. The springtime recovery of photosynthetic capacity in conifers is qualitatively well understood (e.g., see Huner et al for a review), and has generally been attributed to the onset and gradual release of photoinhibition, which down-regulates photosynthesis when low temperatures are combined with high light fluxes (e.g., Krivosheeva et al. 1996). In the summer, some injuries may occur after night frosts (Bergh et al. 1998) or when temperatures just above zero are combined with high solar radiation (Lamontagne et al. 1998). We found that a combination of low temperatures and high irradiance reduced photosynthetic capacity more than expected by temperature only (Figure 5), but these occasions were relatively rare and could not explain all of the observed summertime fluctuations. Summertime variation in photosynthesis has sometimes been attributed to sink limitation (Luxmoore 1991, Turnbull et al. 2002). According to this view, the plant maintains a balance between the production and consumption of carbon; if produc- TREE PHYSIOLOGY ONLINE at

6 374 MÄKELÄ, HARI, BERNINGER, HÄNNINEN AND NIKINMAA Figure 5. (A) Residuals of the model as a function of mean daily temperature. (B) Residuals of the model for the days during June 15 August 30, when morning radiation was high, plotted against morning temperature (when solar radiation was rad ( C)). tion exceeds potential growth, photosynthesis will be downregulated. This would imply that photosynthetic production is less sensitive to temperature when carbohydrate stocks are low, i.e., in late June and early July (Sofronova and Kaipiainen 1996), and more sensitive after growth cessation (Stitt and Krapp 1999). We detected no changes in the temperature response of photosynthesis over the summer (Figure 4). Another possibility is that a single mechanism could explain most changes in the photosynthetically active system during spring and summer. We call this the dynamic acclimation hypothesis. This hypothesis assumes that a tree regulates its photosystem to minimize damage by low temperature or high light stress, or both. A mechanism reducing photosynthetic capacity at the beginning of a cold spell would considerably reduce the risk of photoinhibitory damage and is, therefore, likely to have been favored by natural selection. Such a mechanism could be driven by different temperature dependencies of the breakdown and resynthesis of photosynthetically active substances, in much the same way as has been postulated for the regulation of maintenance respiration rate (Johnson and Thornley 1985). Our analysis showed that the time constant of the delayed temperature effect on photosynthesis was of the same order of Figure 6. Comparison of predicted and measured course of mean daily photosynthesis during a normal bright day (A) and during a day when frost occurred during the previous night (B). magnitude as that of frost hardiness (Repo et al. 1990). This suggests that acclimation of the light response of photosynthesis may be associated with biochemical synthesis reactions. Preliminary analyses have shown that considerable changes take place in pigment concentrations in parallel with the springtime recovery of photosynthesis (E. Juurola, University of Helsinki, unpublished data). A closer analysis (not shown) Figure 7. Measured and predicted time course of mean daily photosynthesis. Predicted values were calculated using the photosynthesis model of Hari and Mäkelä (2003) combined with the present model of daily quantum yield. TREE PHYSIOLOGY VOLUME 24, 2004

7 ACCLIMATION OF PHOTOSYNTHESIS TO TEMPERATURE 375 indicated that the response time may be slightly shorter in the spring than in the autumn. Modeling the annual cycle The key finding of our study is that a single model describing photosynthetic capacity as a reversible, first-order delay process driven by air temperature provides predictions accounting for the main features of change in photosynthetic capacity from spring to autumn, with the exception of the rare occurrence of transient damage caused by frost or low temperature events in the summer. It is not known if the changes are controlled by a single physiological mechanism, or by a multitude of different causal chains. The proposed model is simpler, but not significantly less accurate, than the model of Bergh et al. (1998), which uses different driving variables for the spring, summer and autumn seasons, including air temperature and soil frost during spring and frost occurrence in the autumn. Even though soil temperature and frost events were not necessary as explanatory variables, this does not imply that they do not play a role in the causal events regulating photosynthetic capacity. Several previous studies have identified soil temperature (Schaberg et al. 1995) and stem temperature (Wieser 2000) as important factors affecting the spring recovery of photosynthesis. The equation describing the dynamic dependence of the state of photosynthetic acclimation, S, on temperature is equivalent to the description of temperature changes in large bodies as a function of ambient temperature. We would therefore expect a reasonable correlation of the value of α predicted on the basis of air temperature with the temperature of any large body in the forest. However, soil temperature was not a good predictor in our study because the snow cover was still about 1 m deep at the onset of photosynthetic activity. It has been possible to explain the timing of many phenological events, such as vegetative bud burst and flowering, with various temperature sum models where the rate of development is dependent on air temperature (Arnold 1959, Sarvas 1967, Häkkinen et al. 1998). But contrary to the seasonal changes in photosynthetic capacity, the ontogenetic development leading to these visible phenological events is irreversible. Because of this, the seasonal changes in photosynthesis are not developmental processes in the narrow sense of the term. However, the model structures used for ontogenetic development are also apt for describing acclimation, and provided the methodological background for the model developed by Pelkonen and Hari (1980), who called their analog of our S the state of development of photosynthesis. Because of the importance of reversibility, we have used the term state of acclimation instead. In both the present model and the model of Pelkonen and Hari (1980), S follows air temperature with a time constant. The definition of S in the model of Pelkonen and Hari (1980) differs from the present one by a linear transformation, but this does not essentially affect either the definition of S or the method of parameter estimation for the models. A more substantial difference is the assumption about the rate of change of S (Equation 5). Pelkonen and Hari (1980) defined a strongly nonlinear differential equation where the rate of change was slow while the temperature remained close to the long-term mean but increased faster than linearly if larger variations in temperature occurred. Our results show that the linear model provides predictions similar to those of Pelkonen and Hari (1980), but is more transparent and easier to fit to data. The main contribution of our study is its potential application to estimates of the carbon balance of boreal, temperate and subalpine coniferous stands, which can be overestimated by up to 40% if acclimation to the annual cycle is not taken into account (Bergh et al. 1998). An adequate, easily parameterized model is therefore crucial for accurate production estimates. On the other hand, acclimation to temperature will be less important in environments where annual variation in temperature is not pronounced, and will not be directly applicable to deciduous trees in which the cycle of leaf growth and shedding has replaced the need for photosynthetic acclimation (but see Turnbull et al. 2002). However, because of the analogy between our acclimation model and earlier models of ontogenetic development and frost hardiness, it seems likely that an approach consisting of a slow dynamic change in parameters of a fast-responding physiological process will be more generally applicable to different acclimation and development processes in trees from different vegetation zones. Acknowledgments This study was supported by the Academy of Finland, Project Nos and References Arnold, C.Y The determination and significance of the base temperature in a linear heat unit system. Proc. Am. Soc. Hortic. Sci. 74: Bergh, J. and S. Linder Effects of soil warming during spring on photosynthetic recovery in boreal Norway spruce stands. Global Change Biol. 5: Bergh, J., R.E. McMurtrie and S. Linder Climatic factors controlling the productivity of Norway spruce: a model-based analysis. For. Ecol. Manage. 110: Cowan, I.R. and G.D. Farquhar Stomatal function in relation to leaf metabolism and environment. Soc. Exp. Biol. Symp. 31: Häkkinen, R., T. Linkosalo and P. Hari Effects of dormancy and environmental factors on timing of bud burst in Betula pendula. Tree Physiol. 18: Hänninen, H. and P. Hari Recovery of photosynthesis of boreal conifers during spring: a comparison of two models. For. Ecol. Manage. 169: Hari, P. and A. Mäkelä Annual pattern of photosynthesis of Scots pine in the boreal zone. Tree Physiol. 23: Hari, P., A. Mäkelä, E. Korpilahti and M. Holmberg Optimal control of gas exchange. Tree Physiol. 2: Hari, P., P. Keronen, J. Bäck, N. Altimir, T. Linkosalo, T. Pohja, M. Kulmala and T. Vesala An improvement of the method for calibrating measurements of photosynthetic CO 2 flux. Plant Cell Environ. 22: Huner, N.P.A., G. Öquist and F.M. Sarhan Energy balance and acclimation to light and cold. Trends Plant Sci. 3: TREE PHYSIOLOGY ONLINE at

8 376 MÄKELÄ, HARI, BERNINGER, HÄNNINEN AND NIKINMAA Johnson, I.R. and J.H.M. Thornley Temperature dependence of plant and crop processes. Ann. Bot. 55:1 24. Krivosheeva, A., D.L. Tao, C. Ottander, G. Wingsle, S.L. Dube and G. Öquist Cold acclimation and photoinhibition of photosynthesis in Scots pine. Planta 200: Lamontagne, M., H. Margolis and F. Bigras Photosynthesis of black spruce, jack pine, and trembling aspen after artificially induced frost during the growing season. Can. J. For. Res. 28:1 12. Linder, S. and T. Lohammar Amount and quality of information on CO 2 exchange required for estimating annual carbon balance of coniferous trees. In Understanding and Predicting Tree Growth. Ed. S. Linder. Studia For. Suecica 60: Luxmoore, R.J A source sink framework for coupling water, carbon, and nutrient dynamics of vegetation. Tree Physiol. 9: Mäkelä, A., F. Berninger and P. Hari Optimal control of gas exchange during drought. Theoretical analysis. Ann. Bot 77: Pelkonen, P. and P. Hari The dependence of the springtime recovery of CO 2 uptake in Scots pine on temperature and internal factors. Flora 169: Pisek, A. and E. Winkler Assimilationsvermögen und Respiration der Fichte (Picea excelsa Link) in verschiedener Höhenlage und der Zirbe (Pinus cembra L.) an der alpinen Waldgrenze. Planta 51: Polster, H. and S. Fuchs Winterassimilation und Atmung der Kiefer (Pinus silvestris L.) im mitteldeutschen Binnenlandklima. Arch. Forst. 12: Repo, T., A. Mäkelä and H. Hänninen Modelling frost resistance of trees. In Modelling to Understand Forest Functions. Ed. H. Jozefek. Silva Carelica 15: Sarvas, R The annual period of development of forest trees. Proc. Finn. Acad. Sci. Lett. 1965: Schaberg, P.G., R.C. Wilkinson, J.B. Shane, J.R. Donnelly and P.F. Cali Winter photosynthesis of red spruce from three Vermont seed sources. Tree Physiol. 15: Sofronova, G. and L. Kaipiainen Dynamics of carbohydrate distribution in Scots pine. In Production Process of Scots Pine; Geographical Variation and Models. Eds. P. Hari, J. Ross and M. Mecke. Acta For. Fenn., pp Stitt, M. and A. Krapp The interaction between elevated carbon dioxide and nitrogen nutrition: the physiological and molecular background. Plant Cell Environ. 22: Turnbull, M.H., R. Murthy and K.L. Griffin The relative impacts of daytime and night-time warming on photosynthetic capacity in Populus deltoides. Plant Cell Environ. 25: Wieser, G Seasonal variation of leaf conductance in subalpine Pinus cembra during the winter months. Phyton 40: TREE PHYSIOLOGY VOLUME 24, 2004