ÜLO NIINEMETS, 1 WOLFGANG BILGER, 2 OLEVI KULL 1 and JOHN D. TENHUNEN 3

Transcription

1 Tree Physiology 19, Heron Publishing----Victoria, Canada Responses of foliar photosynthetic electron transport, pigment stoichiometry, and stomatal conductance to interacting environmental factors in a mixed species forest canopy ÜLO NIINEMETS, 1 WOLFGANG BILGER, 2 OLEVI KULL 1 and JOHN D. TENHUNEN 3 1 Estonian Institute of Ecology, Riia 181, Tartu EE-51014, Estonia 2 Department of Biology and Nature Conservation, Agricultural University of Norway, P.O. Box 5014, N-1432 Ås, Norway 3 Department of Plant Ecology, University of Bayreuth, D Bayreuth, Germany Received August 18, 1998 Summary We studied limitations caused by variations in leaf temperature and soil water availability on photosynthetic electron transport rates calculated from foliar chlorophyll fluorescence analysis (ϑ) in a natural deciduous forest canopy composed of shade-intolerant Populus tremula L. and shadetolerant Tilia cordata Mill. In both species, there was a positive linear relationship between light-saturated ϑ (ϑ max ) per unit leaf area and mean seasonal integrated daily quantum flux density (S s, mol m 2 day 1 ). Acclimation of leaf dry mass per area and nitrogen per area to growth irradiance largely accounted for this positive scaling. However, the slopes of the ϑ max versus S s relationships were greater on days when leaf temperature was high than on days when leaf temperature was low. Overall, ϑ max varied 2.5-fold across a temperature range of C. Maximum stomatal conductance (G max ) also scaled positively with S s. Although G max observed during daily time courses, and stomatal conductances during ϑ max measurements declined in response to seasonally decreasing soil water contents, ϑ max was insensitive to prolonged water stress, and was not strongly correlated with stomatal conductances during its estimation. These results suggest that photorespiration was an important electron sink when intercellular CO 2 concentration was low because of closed stomata. Given that xanthophyll cycle pool size (VAZ, sum of violaxanthin, antheraxanthin, and zeaxanthin) may play an important role in dissipation of excess excitation energy, the response of VAZ to fluctuating light and temperature provided another possible explanation for the stable ϑ max. Xanthophyll cycle carotenoids per total leaf chlorophyll (VAZ/Chl) scaled positively with integrated light and negatively with daily minimum air temperature, whereas the correlation between VAZ/Chl and irradiance was best with integrated light averaged over 3 days preceding foliar sampling. We conclude that the potential capacity for electron transport is determined by long-term acclimation of ϑ to certain canopy light conditions, and that the rapid adjustment of the capacity for excitation energy dissipation plays a significant part in the stabilization of this potential capacity. Sustained high capacity of photosynthetic electron transport during stress periods provides an explanation for the instantaneous response of ϑ to short-term weather fluctuations, but also indicates that ϑ restricts potential carbon gain under conditions of water limitation less than does stomatal conductance. Keywords: dynamic adaptation, photosynthetic acclimation, Populus tremula, seasonality, Tilia cordata, water stress, xanthophyll cycle. Introduction Along light gradients in tree canopies, there is a positive relationship between irradiance and air temperature, and a negative relationship between irradiance and humidity (Chiariello 1984, Shuttleworth et al. 1985). Therefore, vapor pressure deficits also scale positively with irradiance. Focusing on these correlations is important for understanding tree canopy carbon gain because of the interactive effects of high light and water stress on photosynthetic reactions (Björkman et al. 1981, Björkman and Powles 1984, Ögren and Öquist 1985, Valladares and Pearcy 1997), and high light and low or high temperature (Powles et al. 1980, Ludlow and Björkman 1984, Oberhuber and Bauer 1991, Brugnoli et al. 1994, Valladares and Pearcy 1997), and because plant responses to multiple stresses cannot be predicted from single-factor analyses (Björkman 1987, Valladares and Pearcy 1997). Understanding the interactive influences of multiple stress factors on photosynthetic reactions is further complicated in natural habitats because of seasonal fluctuations in light, air temperature and soil water regimes. The interactive effects of high light and water stress on photosynthetic electron transport have been studied previously using a range of fixed irradiances, but none of these studies included a broad light gradient comparable to that occurring in a natural forest canopy. Studies on the effects of rapidly developing, high intensity water stress have demonstrated that foliar water limitations result in a decline in photosynthetic electron transport rates (Björkman et al. 1981, Björkman and Powles 1984). However, recent work designed to examine the effects of a more realistic development of natural drought conditions

2 840 NIINEMETS, BILGER, KULL AND TENHUNEN indicates that photosynthetic electron transport rates are relatively insensitive to water stress (di Marco et al. 1988, Cornic and Briantais 1991, Epron and Dreyer 1992, 1993, Epron et al. 1992, Cornic 1994). Possibly, plants avoid damage through the use of highly effective mechanisms for non-radiative quenching of excitation energy as well as high foliar acclimation potentials to gradually developing stresses. In the xanthophyll cycle, conversion of violaxanthin by way of antheraxanthin to zeaxanthin is thought to play an important role in excitation energy dissipation (see Pfündel and Bilger 1994, Gilmore 1997 for review). There is a good correlation between the sum of foliar zeaxanthin and antheraxanthin contents and leaf capacity for non-photochemical quenching of excitation energy (Gilmore 1997). Studies demonstrate that the xanthophyll cycle pool size flexibly adjusts to altered levels of excitation energy. The pool of xanthophyll cycle carotenoids (VAZ, sum of violaxanthin, antheraxanthin, and zeaxanthin), whether expressed per unit total carotenoids, per unit total chlorophyll or per unit leaf area, is consistently larger in leaves exposed to high irradiances than in leaves exposed to low irradiances (Thayer and Björkman 1990, Adams et al. 1992, Lovelock and Clough 1992, Brugnoli et al. 1994, Bilger et al. 1995a, Königer et al. 1995, Logan et al. 1996). There is also evidence for greater VAZ pools in leaves acclimated to low temperature than in leaves acclimated to high temperature (Oberhuber and Bauer 1991, Adams and Demmig-Adams 1994, Haldmann et al. 1996) as well as in leaves subjected to water stress (Kyparissis et al. 1995). Despite the potential importance of the xanthophyll cycle in altering the capacity for thermal dissipation of excitation energy, the dynamic tuning of the xanthophyll cycle during periods of water stress has not been studied extensively within natural vegetation stands. A field study was carried out in the intermixed canopy of two coexisting, temperate deciduous forest species of contrasting shade tolerance. Populus tremula L. is a shade-intolerant, early-successional species and Tilia cordata Mill. is a shadetolerant, late-successional species. In the studied forest stand, P. tremula occupied upper canopy positions above the foliage of T. cordata. The primary objectives of our study were to determine (1) how potential leaf electron transport rate changes in response to drought, and (2) whether there are interactions between day-to-day fluctuations in irradiance and temperature in their influences on foliar photosynthetic electron transport rate. We also investigated (3) to what extent and how quickly the xanthophyll cycle pool size adjusted to increased excitation energy during periods of drought and low temperature, and (4) whether changes in pigment composition and total pool sizes, and in capacities of photosynthetic electron transport occur in a coordinated manner. Material and methods Study area The study was conducted at Järvselja (58 22 N, E, elevation m), Estonia during July and August The deciduous tree canopy with total leaf area index of about 6 consisted of two layers. The upper layer ( m) was dominated by P. tremula and Betula pendula Roth., and the lower layer (4--17 m) by T. cordata. The woody understory was mainly composed of Corylus avellana L. and a coppice of T. cordata. The conifer Picea abies (L.) Karst., which comprised 18% of total stand basal area, was an important component in all forest layers. Mean (± SE) water-saturated hydraulic conductivity of the upper soil layer was relatively high at (3.0 ± 2.0) 10 3 cm s 1 at 7.5 cm, but the deeper layers had higher bulk density and lower hydraulic conductivities----(5.2 ± 4.5) 10 4 cm s 1 at 20 cm, (3.8 ± 3.0)10 4 cm s 1 at 30 cm, and (9.4 ± 5.4) cm s 1 at 50 cm. Further details of the stand are reported in Niinemets et al. (1998a). Estimation of integrated quantum flux densities in the canopy We used measurements of photosynthetically active quantum flux density (Q), made with quantum sensors in 18 canopy locations during the entire growing season, together with estimates of fractional penetration of irradiance at the sensor locations, obtained at weekly intervals by hemispherical photography, to estimate mean integrated quantum flux densities incident to the leaves (Niinemets et al. 1998a). Multiple linear regressions were developed between daily integrated Q (Q i, mol m 2 day 1 ), which was calculated from sensor readings taken at 1-min intervals, and the fractions of penetrating diffuse (I dif, diffuse site factor) and potential penetrating direct solar radiation of open sky (I dir, direct site factor). Factors I dif and I dir were calculated from hemispherical photographs. Hemispherical photographs were also taken from the sample points immediately after foliage collection for measurements of chlorophyll fluorescence parameters and leaf conductances. Values of Q i for the sample locations were calculated from regression equations in the form of Q i = ai dif + bi dir. Mean seasonal total integrated quantum flux density (S s, mol m 2 day 1 ) was defined as the mean Q i between the completion of lamina expansion growth (about June 3, 1995) and the cessation of growth in leaf thickness (about July 21, 1995). For short-term Q i, Q i was averaged over 3 days preceding foliar sampling (S 3d ). Mean integrated values for direct irradiance (D s = mean seasonal and D 3d = direct light averaged over 3 days preceding the measurements) were calculated as the products of b and I dir. Measurement of site weather variables and soil water potential The difference between total precipitation and canopy interception was measured with a rain gauge (interception area 0.05 m 2, Type Plyuviograf P-2, Zavod No. 491, Russia) placed just under the canopy. Air temperatures in five canopy horizons were measured continuously with thermistors (TS5, Motorola Semiconductor, Inc., Austin, TX) and logged at 1-min intervals. Fish-eye photographs were taken just above the sensors at weekly intervals, so that estimates of integrated Q could also be calculated for the temperature sensors. Soil surface minimum and maximum temperatures were measured with standard liquid-in-glass minimum and maximum thermometers that were read each day at noon. To estimate soil water content (Θ, g g 1 ), duplicate soil samples were taken at weekly intervals TREE PHYSIOLOGY VOLUME 19, 1999

3 FOLIAR PHOTOSYNTHESIS IN RELATION TO MICROCLIMATE FACTORS 841 and weighed just after sampling and after drying at 105 C to a constant mass. Values of Θ were used to calculate soil matrix potential (Ψ s ), based on a simple empirical equation developed by Gardner et al. (1970) that was parameterized for each soil horizon (Niinemets et al. 1999a). In situ chlorophyll fluorescence measurements Mean (± standard error) height of the sampled trees was 25.0 ± 1.7 m in P. tremula (n = 4) and 15.1 ± 0.7 m in T. cordata (n = 4). The canopy was accessed from permanent scaffoldings (height 25 m). Steady-state fluorescence yield (F s ) and fluorescence yield after application of a saturating pulse of white light (F m ) were measured with a portable pulse-modulation fluorometer (PAM-2000, Heinz Walz GmbH, Effeltrich, Germany) equipped with a leaf clip holder (Model 2030-B) as detailed in Bilger et al. (1995b) and Niinemets et al. (1998a). Saturating pulse kinetics were checked before each measurement series at a given canopy height, and the quantum flux density and length of the saturated pulse adjusted to close all photosystem II (PS II) reaction centers, but to avoid photo-inhibitory damage of the analysis area (Anonymous 1993). We assumed that both Photosystems I and II absorbed equal amounts of light and calculated photosynthetic electron transport rate (ϑ, µmol e m 2 s 1 ) according to Genty et al. (1989). For the calculation of absorbed Q, we used an estimate of leaf absorptance computed from leaf chlorophyll concentration per unit area (Evans 1993); mean values (± SD) obtained for leaf absorptance were ± for P. tremula and ± for T. cordata. Natural light was used during the F s measurements on July 19, and steady-state values were sampled after the leaves had been exposed to direct beam irradiance for at least 10 min. In the same leaf, F s and F m were measured at three to four locations along the midrib. During other measurement campaigns, artificial light provided by an external halogen lamp (Decostar 51-S, Osram GmbH, München, Germany; measurements on July 21) or the internal halogen lamp of PAM-2000 (Bellaphot, Osram GmbH; on all other measurement days) was used. Starting from dim light ( µmol m 2 s 1 ), Q was increased in steps until no further increases in the steady-state values of ϑ were observed. Values of ϑ measured at quantum flux densities higher than those assumed to be saturating for electron transport ( µmol m 2 s 1 for the upper canopy leaves and µmol m 2 s 1 for the lower canopy leaves) were used as estimates of photosynthetic electron transport capacity (ϑ max ). According to laboratory measurements under controlled conditions, these threshold irradiances are close to saturating for both species (Niinemets and Oja, unpublished observations 1997, Niinemets et al. 1998b). Use of these irradiances enabled us to measure the effective quantum yield of PS II (φ PSII ) with sufficient accuracy for all leaves in the canopy (φ PSII during measurements of ϑ max was 0.20 ± 0.02). However, we occasionally observed that ϑ increased with increasing Q over a light range widely exceeding the saturating range; e.g., ϑ for the upper canopy leaves of P. tremula increased to Q values as high as 3500 µmol m 2 s 1. We attributed the increase in ϑ with increasing quantum flux density beyond that limiting to photochemistry to the increase in leaf temperature with increasing measurement Q. According to laboratory estimation, the optimum temperature for ϑ is around 35 C in P. tremula and 40 C in T. cordata (Niinemets and Oja, unpublished measurements). Because maximum leaf temperatures recorded during the measurements were below 36 C, ϑ scaled positively with leaf temperature in most of our measurements. Although the thermal effect of light accounted for about a 2 C change in leaf temperature and a 5--15% change in ϑ max within single leaf measurements, it was possible to use these single-leaf values of ϑ max to develop the ϑ max versus leaf temperature response curves over a wide temperature range (cf. Figure 3). Chlorophyll fluorescence was measured between 1300 to 1600 h on July 19 and 21, and between 0800 and 1430 h on other measurement days. Measurements of ϑ along the canopy light gradient were always started at the top of the canopy to decrease the influence of differences in air temperature during a measurement campaign on the ϑ max estimates within canopy profiles. Estimation of foliar stomatal conductances (G) in the canopy Leaf conductances to water vapor of the leaves used for estimation of fluorescence characteristics were measured just after completion of all fluorescence measurements with a steadystate porometer (LI-1600, Li-Cor, Inc., Lincoln, NE) at ambient temperature and humidity. Natural illumination was used, but care was taken to hold the leaf in direct beam irradiance until steady-state values were reached. In all cases, leaf conductances were corrected for boundary layer conductance (> 2700 mmol m 2 s 1 ) to obtain a measure of stomatal conductance. The day after the fluorescence measurements, G of the same leaves was measured at hourly intervals between 0730 and 2130 h. The highest value of G per leaf recorded during the two consecutive days was defined as maximum leaf conductance (G max ). Generally, G max was observed during morning measurements. Because weather conditions were similar on both days, and fluorescence measurements were time-consuming, a paired value of G for each ϑ max was taken based on the hourly value of G corresponding to the time of the ϑ estimations. Nevertheless, because the canopy gradient in G was strong, values of G measured just after completion of all fluorescence measurements were correlated with the paired values taken from the daily time courses (r 2 = 0.85, P < for a linear relationship across the entire set of data). Gas-exchange measurements in the laboratory Carbon assimilation characteristics of the species were measured over the entire growing season. On each measurement date, two twigs were taken from each species, one from the upper canopy and the other from the lower canopy. The twigs were cut under water, transported to the laboratory within an hour of collection, recut under water and held in dim light before measurement of net assimilation versus intercellular CO 2 (C i ) response curves (A--C i curves) as described in Niinemets et al. (1998b). Photosynthetically active quantum flux density was held at 1000 µmol m 2 s 1, mean ± SE leaf temperature was 25.7 ± 0.1 C and leaf-to-air vapor pressure TREE PHYSIOLOGY ON-LINE at

4 842 NIINEMETS, BILGER, KULL AND TENHUNEN deficit varied from to mol mol 1 across all measurements. The capacities for photosynthetic electron transport (J max ) and carboxylase activity of ribulose-1,5-bisphosphate carboxylase/oxygenase (Rubisco; V cmax ) were calculated as described in Niinemets et al. (1999b). A different symbol was used for foliar photosynthetic electron transport capacity calculated from foliar gas-exchange measurements at high CO 2 concentration (J max ) to distinguish it clearly from the estimate (ϑ max ) obtained from chlorophyll fluorescence measurements at current ambient CO 2 concentrations. In general, the ratio of quantum yields of PS II and photosynthetic O 2 evolution (Seaton and Walker 1990, Öquist and Chow 1992) or CO 2 absorption at high light and saturating C i (Oberhuber et al. 1993, Valentini et al. 1995) is conservative. Therefore, chlorophyll fluorescence analysis is a reliable tool for rapid assessment of foliar electron transport potentials (Schreiber et al. 1994). The capacity calculated from gas-exchange measurements (range of variation µmol m 2 s 1 ) was occasionally lower than ϑ max (range of variation µmol m 2 s 1 ), especially for high-light grown leaves. This discrepancy may be related to the effects of different CO 2 concentrations, e.g., feedback inhibition of photosynthetic electron transport at high C i (cf. Pammenter et al. 1993). It may also be partly related to higher quantum flux densities as well as greater leaf temperatures during some of the chlorophyll fluorescence measurements in the field. Sampling and analysis of foliage for pigments Leaf samples for analysis of pigments were taken between 1200 and 1400 h. Discs of 1.03 cm 2 were punched from the leaves with a cork-borer, put in labeled vials and plunged in liquid nitrogen. The samples were stored in airtight sealed vials at 18 C until analyzed. Pigments were analyzed by HPLC as described in Niinemets et al. (1998a). Morphological and nitrogen analyses Each leaf used in in situ ϑ max and G max determinations and laboratory gas-exchange analyses, was further analyzed for leaf dry mass per area and nitrogen concentration. Projected leaf area was calculated from a trace of leaf circumference made with a computer digitizer (QD-1212, QTronix, Taiwan). Petioles were removed, and the leaflets weighed after ovendrying at 70 C for 48 h. Lamina nitrogen concentrations were measured with an elemental analyser (CHN-O-Rapid, Foss Heraeus GmbH, Hanau, Germany). Results Site weather variables and soil water content during the study There were substantial rains during the study period (Figure 1A); however, because of low infiltration capacity of the loamy soil, soil water potential of sampled mineral soil horizons ( cm) was low over most of the growing season, except at the beginning of the season (end of May). The rains affected the water potential of the top soil layer more than that of the deeper soil layers which had higher bulk density and lower hydraulic conductivity (Figure 1B). Day-to-day variations in both soil surface (Figure 1C) and air temperature at 15 m height in the canopy (Figure 1D) were considerable. Mean daily air temperature over the study period was 15.5 ± 0.3 C, but daily minimum temperature varied fourfold, and daily maximum temperatures varied twofold during the same time period (Figure 1D). Seasonal mean integrated Q just above the temperature sensors correlated positively with mean seasonal daily maximum (r 2 = 0.95, P < 0.01) and mean daily air temperatures (r 2 = 0.83, P < 0.05), and negatively with mean daily minimum temperature (r 2 = 0.90, P < 0.01). Analogous correlations were sometimes weaker Figure 1. Summary of local weather and soil data during the growing season of 1995 at the study site at Järvselja, Estonia. A: The part of precipitation reaching the soil surface (canopy throughfall). B: Soil water potential determined from volumetric soil water content by empirical relationships in different soil horizons. C: Soil surface minimum and maximum temperatures measured over the day with liquid-inglass minimum and maximum thermometers and recorded each day at noon. D: Mean daily, minimum, and maximum canopy air temperatures at a height of 15 m measured with thermistors and logged at one-minute intervals. The study period is marked as a rectangle on panel B. TREE PHYSIOLOGY VOLUME 19, 1999

5 FOLIAR PHOTOSYNTHESIS IN RELATION TO MICROCLIMATE FACTORS 843 over the short-term (with the values of r 2 varying from 0.67 to 0.99, averaging 0.91) than over the long term. These correlations were taken into account when calculating mean daily, maximum and minimum air temperatures at leaf locations from actual temperature measurements and estimations of integrated light above the sensor locations (cf. Material and methods). Scaling of foliage electron transport potentials and stomatal conductances with nitrogen Positive linear relationships of light-saturated photosynthetic electron transport (ϑ max ) calculated from chlorophyll fluorescence measurements (Figure 2A) and daily maximum stomatal conductance (G max ; Figure 2C) with leaf nitrogen per unit area (N a ) were observed on all measurement days. However, the slope of the relationship between ϑ max and N a was steeper on days when leaf temperature was high. Over a temperature range of about 15 C, ϑ max per unit leaf dry mass varied more than threefold in P. tremula (Figure 3A). When ϑ max was converted to 25 C, based on standardized ϑ max versus leaf temperature response curves (Figures 3B and 3C), all data points fit a single linear relationship (Figure 2B). The slope of the G max versus N a relationship was high at the beginning of the measurement period when soil water availability was high and decreased with decreasing soil water availability (cf. Figures 1B and 2C). The major source of variation in lamina nitrogen concentration per area was the variability in leaf dry mass per area (Figure 2D). Nitrogen concentration per unit dry mass (N m ) varied only from 1.33 to 1.84 mmol g 1 (mean ± SE = 1.60 ± 0.02 mmol g 1 ) in P. tremula, and from 1.63 to 2.22 (mean ± SE = 1.92 ± 0.02 mmol g 1 ) in T. cordata (means are significantly different at P < according to a separate samples t-test). Acclimation to canopy light profile as a dominant factor in N-related variability in ϑ max and G max In both species, there was a single linear relationship between seasonal integrated mean daily quantum flux density (S s ) and ϑ max per unit area standardized to 25 C (Figures 4A and 4B). Partly because of greater nitrogen concentration per unit dry mass in T. cordata, the slope of this relationship was higher in T. cordata (Figure 4A) than in P. tremula (Figure 4B). There was a positive correlation between ϑ max per unit dry mass at 25 C and S s in T. cordata (P < 0.02), but not in P. tremula (P > 0.2). Maximum stomatal conductances also depended on longterm integrated light (Figures 4C and 4D); however, the slope of this relationship depended on soil water availability, and was lower during periods of lower soil water potential. There were also indications of increased water stress in upper canopy leaves of P. tremula at low soil water potentials (Figure 4D). In Figure 2. Correlations of light-saturated values of linear photosynthetic electron transport rates calculated from chlorophyll fluorescence measurements (ϑ max, µmol m 2 s 1 ; A and B) and maximum values of stomatal conductance to water vapor observed during daily time courses (G max, mmol m 2 s 1 ; C) with foliar nitrogen content per unit area (N a, mmol m -2 ), and the correlation between N a and lamina dry mass per unit area (M A, g m 2 ; C) over the season in Populus tremula ( in insets) and Tilia cordata ( in insets). A: ϑ max at ambient leaf temperatures versus N a on different dates during the study period. Each point corresponds to the mean of the two highest ϑ max values per leaf. Leaf temperatures (T L ) depicted are the means for the whole series of ϑ max measurements on a given day. The slope of the linear regressions between ϑ max and N a was higher on days where T L was higher during the measurements. The data of both Populus tremula and Tilia cordata were fit by the same linear regression lines (see the inset to relate the data points to species): r 2 = 0.74, P < for the measurements during July (mean ± SE leaf temperature was ± 0.34 C), and r 2 = 0.93, P < for the measurements during August (T L = ± 0.19). B: ϑ max standardized to a common temperature of 25 C using the empirical equations depicted in Figure 3 versus N a. C: Maximum stomatal conductance in relation to N a. Linear regressions were calculated for the points measured during July (upper line) and during August (lower line). D: N a versus M A. Separate linear regressions for each species (see the insets to associate the data points with species) were calculated through all values sampled over the season. TREE PHYSIOLOGY ON-LINE at

6 844 NIINEMETS, BILGER, KULL AND TENHUNEN Figure 3. Light-saturated photosynthetic electron transport rates estimated by chlorophyll fluorescence in relation to leaf temperature during the measurements. A: The dependence of ϑ max per unit leaf dry mass on leaf temperature in P. tremula (n = 390): ϑ max was expressed per unit dry mass to account for leaf nitrogen and dry mass per area related variability in leaf photosynthetic electron transport potentials (Figure 2). B and C: ϑ max in standardized units in P. tremula and T. cordata. Because a limited range in temperature was available in measurements within single leaves, the curves on panels B and C were obtained according to the following routine: (1) ϑ max values for a leaf were standardized with respect to the highest light-saturated value measured; (2) the leaves were ranked according to the maximum leaf temperatures observed during the measurements; (3) the standardized values of ϑ max of different leaves were brought to the same scale by calculating linear regressions between ϑ max and T L over the range of overlapping temperatures in a leaf preceding the leaf with the next lower ranking, and the standardized ϑ max of the latter leaf was adjusted to be compatible with the value expected for the next higher leaf. The procedure (3) was started with the leaf measured at the highest temperatures. Thus, each curve obtained in the way described has ϑ max = 1.0 at the optimum temperature. The temperature dependence of ϑ max was modeled according to Niinemets and Tenhunen (1997). All model parameters in panel A were taken from B, except for the scaling constant, c, which was found by nonlinear regression. Inset in panel C demonstrates the results of the comparison of the shapes of temperature response curves of ϑ max in P. tremula and T. cordata. For this comparison, the scaling constant (c) was decreased in T. cordata to give an equal ϑ max value at T L = 25 C for both species. most drought-stressed leaves at the end of the study period, G max was independent of S s in P. tremula (filled circles in Figure 4D). Leaf nitrogen content per unit area, which was mainly determined by leaf dry mass per area (Figure 2D), did not change during the study period (Figures 4E and 4F). The constancy in N a probably accounts for the conservative relationships between ϑ max and S s (Figures 4A and 4B). Relationships between ϑ max and stomatal conductance Although ϑ max and G during the estimations of ϑ max were correlated (Figure 5A), the slope of the relationship depended on measurement date. The slope was greater for July and August measurements than for July and August measurements (Figure 5A). The difference in mean leaf temperature during ϑ max estimations provided one explanation for the different slopes. Mean leaf temperature was higher during the late July and late August measurements than during the mid-august measurements (cf. Figure 2A). Nevertheless, the slope was also low for the July measurements, when leaf temperatures were high. Because of this discrepancy, and because ϑ max standardized to a common temperature was time-independent during the study (Figures 4A and 4B), we conclude that water-stress related changes in G were also responsible for different slopes in Figure 5A. At a common electron transport capacity, stomatal conductance was relatively higher in well-watered conditions (July ) than under water stress conditions. When ϑ max was standardized to 25 C and plotted against morning values of G---- likely to represent an estimate of stomatal conductances under low stress conditions----there was a single linear relationship between G and ϑ max (Figure 5B), indicating that the light-acclimation of stomatal conductances and ϑ max proceeds in a coordinated manner, but also that stomatal conductance is more responsive to foliar water stress that develops during the day and with soil drought. Variation in foliage photosynthetic potentials calculated from gas-exchange measurements during the season The capacity for photosynthetic electron transport calculated from net assimilation versus intercellular CO 2 concentration curves (J max ) and maximum Rubisco carboxylase activity (V cmax ) was lower at the beginning and end of the growing season than in mid-season (Figure 6). At each measurement date, values of J max and V cmax were more scattered in T. cordata than in P. tremula, because both J max and V cmax per unit dry mass were positively correlated with S s in T. cordata but not in P. tremula (data not shown). Despite the scatter, these measurements demonstrate that J max and V cmax were relatively constant or slightly increased during the period of increasing soil water limitation. Both J max and V cmax decreased at the end of the study, even though soil water content started to recover (cf. Figures 1B and 6). Changes in xanthophyll cycle pool size: short- versus long-term light effects The pool of violaxanthin cycle carotenoids (VAZ, sum of zeaxanthin, antheraxanthin and violaxanthin) per unit total leaf chlorophyll increased with increasing integrated mean daily quantum flux density (Figures 7A--D). The data were more TREE PHYSIOLOGY VOLUME 19, 1999

7 FOLIAR PHOTOSYNTHESIS IN RELATION TO MICROCLIMATE FACTORS 845 Figure 4. Acclimation of photosynthetic electron transport rates estimated by chlorophyll fluorescence (A, B), stomatal conductances (C, D) and leaf nitrogen contents (E, F) to mean seasonal incident quantum flux density on different days during the study. Maximum measured values are given for both photosynthetic electron transport rate and stomatal conductance, and ϑ max has also been standardized to 25 C (cf. Figure 3). Linear regression lines on panels A, B, E and F are for all values; the upper lines on panels C (linear regression) and D (second-order polynomial regression) are for July , and the lower curves for August measurements. Insets demonstrate the relationships with both species combined (symbols as in Figure 2). The values of r 2 for the relationships across all data in the inset of panel D are 0.14 (P < 0.02) for P. tremula and 0.33 (P < 0.001) for T. cordata. The decline in G max during the season is related to decreases in soil water availability. Figure 5. Stomatal conductance to water vapour (G s ) in relation to ϑ max. A: Correlation of ϑ max with G s estimated just after the measurements of ϑ max. Linear regressions are calculated for the points measured during July (upper line) and during August (lower line). B: Correlation of ϑ max with G max. The ϑ max was standardized to 25 C based on the shapes of ϑ max versus temperature curves depicted in Figure 3, and all data were fitted by a single regression line. Figure layout as in Figure 2. dispersed when total seasonal integrated light (S s ) was used as the explaining variable (r 2 = 0.59 for all data pooled), and data collected on days with high integrated irradiances had greater slopes than data collected on cloudy days (Figures 7A and 7B). By contrast, VAZ/Chl fit to a single regression line with both total (r 2 = 0.69, P < 0.001) and direct light (r 2 = 0.68, Figures 7C and 7D) integrated over three days preceding foliar sampling. There was a strong autocorrelation between total and direct light averaged over three days (r 2 = 0.96, P < 0.001), and thus, no inferences of whether VAZ/Chl was adjusted to total or direct integrated short-term light could be made. Contrary to VAZ/Chl, VAZ per total carotenoids (VAZ/Car) was better correlated with seasonal average integrated Q (r 2 = 0.73, P < 0.001; Figures 7E and 7F) than with short-term Q (r 2 TREE PHYSIOLOGY ON-LINE at

8 846 NIINEMETS, BILGER, KULL AND TENHUNEN correlation was very weak (r 2 = 0.12, P < 0.001). When T min and short-term integrated light were used as explaining variables in the multiple linear regression analysis with VAZ/Chl, r 2 was 0.73, and both variables were highly significant (P < 0.001). The results were qualitatively similar with VAZ/Car (data not shown). Discussion Figure 6. Seasonal variability in the capacity for photosynthetic electron transport (J max, closed symbols) and carboxylase activity of ribulose-1,5-bisphosphate carboxylase/oxygenase (Rubisco; V cmax, open symbols) calculated from net assimilation versus intercellular CO 2 response curves measured in the laboratory with detached twigs. At each measurement date, two twigs were measured in both species, one from the upper canopy and the other from the lower canopy. The rates expressed per unit dry mass are given to eliminate the effects of the variability in leaf dry mass per area on calculated rates. The V cmax per unit area varied from 9 to 51 µmol m 2 s 1, and J max per unit area from 23 to 148 µmol m 2 s 1 across all measurements. The data points surrounded by the rectangle were measured over the time period compatible with available data on in situ chlorophyll fluorescence characteristics and stomatal conductances. = 0.61, P < 0.001), suggesting that carotenoid stoichiometry responds less flexibly to short-term light fluctuations. Analysis of light effects on total chlorophyll and carotenoid pools (Figure 8) indicated that no chlorophyll photo-destruction occurred during the study period (Figures 8C and 8D), and that the major effect of changes in daily integrated light receipt was the alteration of total carotenoid pool size (Figures 8A and 8B). Thus, we conclude that the positive relationship between VAZ/Chl and short-term integrated light (Figures 7C and 7D) resulted from light-induced increases in total carotenoid content per unit of chlorophyll (r 2 = 0.55, P < for P. tremula and r 2 = 0.37, P < for T. cordata for light averaged over three days before sampling), rather than changes in carotenoid stoichiometry or chlorophyll photo-destruction. Air temperature effects on VAZ pool size Because of the strong correlations between seasonal estimates of air temperature and S s, we used mean daily (T), minimum (T min ), and maximum (T max ) air temperatures averaged over three days preceding foliar sampling to test the effect of air temperature on VAZ. The VAZ/Chl ratio was not related to T (r 2 = 0.03, P > 0.08), but was negatively correlated with T min (Figure 9). However, the correlation with T max was slightly positive (r 2 = 0.21, P < 0.001). The latter weak correlation was attributable to a positive scaling of T max with integrated Q averaged over three days preceding the sampling (r 2 = 0.48, P < for all data). Although T min, which was generally observed at night, was negatively related to integrated light, the Acclimation of potential electron transport rate and stomatal conductance to the canopy light profile We observed a positive scaling of the light-saturated value of photosynthetic electron transport with leaf nitrogen concentration per unit area (N a, Figures 2A and 2B) and with mean integrated light (Figures 4A and 4B) (cf. Kull and Niinemets 1998, Niinemets et al. 1998a, 1998b). The primary determinant of these positive relationships was the dependence of N a (Figures 4E and 4F, Niinemets 1997, Kull and Niinemets 1998) and leaf dry mass per area (M A ; Niinemets 1997, Niinemets and Tenhunen 1997, Niinemets et al. 1998b) on long-term integrated light. In the studied canopy, ϑ max per unit leaf area varied more than 8-fold across the light gradient (Figures 4A and 4B), but ϑ max per unit leaf dry mass varied only two-fold in mid-season (Niinemets et al. 1998a). Moreover, the positive relationship between N a and ϑ max per unit area also resulted from light-related changes in M A as may be concluded from the relatively low variability in ϑ max per unit foliar N across the canopy (data not shown, cf. Niinemets and Tenhunen 1997, Niinemets et al. 1998a, 1998b). In the current study, ϑ max of both species fit to the same line with N a (Figures 2A and 2B), but T. cordata (Figure 4A) had a steeper slope with integrated light than P. tremula (Figure 4B). This was attributable to a greater nitrogen content per dry mass in T. cordata as well as to a slight but positive correlation between ϑ max per unit foliar N and integrated light (data not shown, cf. Niinemets et al. 1998a). Species differ in the extent to which the capacity for photosynthetic electron transport acclimates to rapid changes in irradiance. For example, Gossypium hirsutum L. increased photosynthetic oxygen evolution threefold in 7 days in response to a 20-fold step change in irradiance, but photosynthetic capacity decreased in Monstera deliciosa Liebm. in response to the same treatments (Demmig-Adams et al. 1989). Very limited capacity for adjustment of ϑ max was also observed on transfer of young potted trees of Fagus sylvatica L. from shade to sunlight (unpublished data of W. Bilger). In the current study, there was no evidence of changing ϑ max as a result of light fluctuations (Figures 2B, 4A and 4B). In general, there is a positive correlation between maximum leaf conductance to H 2 O (G max ) and leaf photosynthetic capacity (cf. Wong et al. 1979) as was confirmed in our study (Figures 2C and 5). Yet, the relationships of stomatal conductance with ϑ max and N a were not universal and depended on the date of foliar sampling. Both the decline in stomatal conductances in response to decreasing soil water availability (Figures 2C, 4C and 4D) and the relative insensitivity of ϑ max to stomatal conductances (Figure 5A) were responsible for the variability TREE PHYSIOLOGY VOLUME 19, 1999

9 FOLIAR PHOTOSYNTHESIS IN RELATION TO MICROCLIMATE FACTORS 847 Figure 7. Effects of total daily integrated quantum flux density averaged over the season (A, B, E, F) or direct daily quantum flux density averaged over the 3 days preceding sample collection (C, D) on the pool of xanthophyll cycle carotenoids (sum of violaxanthin, antheraxanthin and zeaxanthin, VAZ) expressed per unit chlorophyll (A--D) or as a fraction of total carotenoids (E, F). Linear regressions were calculated for single measurement campaigns (A, B) or for all data pooled (C--F). Insets demonstrate the relationships with both species pooled (symbols as in Figure 2). in G max versus ϑ max relationships. Nevertheless, when stomatal conductances measured in the morning in relatively low water stress conditions were plotted against ϑ max values standardized to 25 C (Figure 5B), all data fit a single positive relationship. Thus, we conclude that both the potential electron transport rate and stomatal conductance are acclimated to canopy light gradient, and that long-term canopy light conditions are the major determinant of the maximum stomatal conductances and electron transport potentials. Effects of leaf temperature on ϑ max Besides light, air temperature is another strongly fluctuating environmental variable (Figure 1D). Although potential ϑ max (Figures 2B and 6) did not change remarkably during the study period, actual ϑ max was lower on days with lower air temperature (Figure 2A). When all data were pooled, ϑ max varied by a factor of three over the temperature range of C (Figure 3). This strong control of ϑ max exerted by leaf temperature is compatible with observed temperature response curves of J max assessed from leaf net gas-exchange in conditions where carbon assimilation is likely to be limited by photosynthetic electron transport (e.g., Berry and Björkman 1980, Harley and Tenhunen 1991). Influences of progressive water limitation on stomatal conductances and photosynthetic electron transport Though there was some effect of seasonality on J max and V cmax determined from net photosynthesis versus intercellular CO 2 response curves (Figure 6), no decline in foliar photosynthetic potentials or nitrogen content (Figures 4E and 4F) was observed during the drought cycle. Moreover, the light-saturated rates of photosynthetic electron transport (Figures 4A and 4B), which were measured in conditions where stomata were likely to limit carbon gain (Figures 4C and 4D), were also unaffected by soil water depletion and stomatal closure. These results are in line with other observations demonstrating that PS II photochemistry is little affected by drought under natural conditions (Epron and Dreyer 1992, 1993, Epron et al. 1992, Cornic 1994, Valentini et al. 1995, Valladares and Pearcy 1997). In contrast, other experiments have demonstrated a dramatic decline in the optimal quantum yield of PS II (calculated as the TREE PHYSIOLOGY ON-LINE at

10 848 NIINEMETS, BILGER, KULL AND TENHUNEN Figure 8. Correlations between mean total daily integrated quantum flux density over 3 days preceding foliar sampling and (A, B) total leaf carotenoids, and (C, D) leaf chlorophyll (a + b). Linear regressions were calculated over all measured values. Symbols and insets as in Figure 7. Figure 9. Relationship between daily minimum air temperature and VAZ/Chl. Data were fitted by a third-order polynomial regression. Symbols and inset as in Figure 7. Daily minimum temperatures at leaf locations were found from air temperature measurements in different canopy horizons, and for each leaf were corrected for a weak correlation between daily integrated mean quantum flux density and air temperature (linear regressions between air temperatures and integrated light values just above sensor locations). ratio of variable to maximum fluorescence ratio of the darkadapted sample, F v /F M ) and whole chain photosynthetic electron transport (Björkman et al. 1981, Powles and Björkman 1982, Björkman and Powles 1984). However, strong and rapidly developing (predawn leaf water potentials lower than 5 MPa were reached in less than 10 days) water limitations were applied in these studies. Soil water availability in temperate forest ecosystems declines over a longer time period (e.g., Figure 1B), possibly inducing acclimation phenomena. In Quercus petraea L. ex Liebl., no photo-damage of PS II was found in leaves exhibiting predawn leaf water potentials as low as 3.0 to 3.4 MPa (Epron et al. 1992, Epron and Dreyer 1993). We did not measure leaf water relations in the 1995 growing season. However, leaf water potential at midday did not drop below 2.8 MPa in either P. tremula or T. cordata during the 1996 growing season, which was characterized by a longer drought period leading to lower soil water contents than in 1995 (Niinemets et al. 1999a). Thylakoid energization and quantum yield of PS II depend on intercellular CO 2 and O 2 concentrations; both low CO 2 and O 2 concentrations result in decreased photochemical efficiency of PS II (Dietz et al. 1985, Sharkey et al. 1988, Peterson 1991, Harbinson 1994, Ishibashi et al. 1997). This is because the dark reactions of photosynthesis are the major sinks of excitation energy, and the rate of PS II photochemistry is matched with the consumption of reductive equivalents and ATP in these processes (Sharkey et al. 1988, Harbinson 1994). Insofar as V cmax and J max calculated from CO 2 response curves (Figure 6) did not change in response to water limitation, declining stomatal conductances with advancing soil water depletion are indicative of lower intercellular CO 2 concentrations. Thus, given that the same maximum rate of PS II photochemistry was reached in both low and high foliar stomatal conductances (Figure 5A), the results of the current study suggest that photorespiration played an important role as an electron acceptor in water stress conditions. This is compatible with previous observations indicating that the ratio of photorespiration to carbon assimilation increases under conditions of water stress (Cornic 1994). It is also in agreement with previous findings that carbon assimilation rates are more re- TREE PHYSIOLOGY VOLUME 19, 1999

11 FOLIAR PHOTOSYNTHESIS IN RELATION TO MICROCLIMATE FACTORS 849 sponsive to stomatal closure than the rates of photosynthetic electron transport (Epron et al. 1992, di Marco et al. 1994, Valladares and Pearcy 1997). There are interactions between irradiance and water stress (Valladares and Pearcy 1997), and irradiance, water stress and photoinhibition (Björkman and Powles 1984, Ludlow and Björkman 1984, Valladares and Pearcy 1997). In the current study, there was evidence of more rapid development of water stress in the upper canopy (cf. Figures 2C, 4C and 4D), but also that foliage electron transport capacities (Figures 4A and 4B) were not down-regulated to match the lowered intercellular CO 2 concentrations. Thus, we suggest that photorespiration was a more important sink of excitation energy in the upper canopy, and that the fraction of foliar carbon lost because of photorespiration also increased with increasing integrated irradiance. From another perspective, high oxygen concentration under circumstances where photorespiration is low may promote photoinhibition (Van Wijk and Krause 1991). This may be especially significant at low temperatures when the rate of photorespiration relative to ribulose-1,5-bisphosphate carboxylation is low (e.g., Jordan and Ogren 1984). However, the activity of several scavenger systems increases as a result of acclimation to low temperatures (Schöner and Krause 1990, Leipner et al. 1997). A similar response of the studied species may be inferred from the current data, because low air temperature did not result in any sustained changes in the capacity for photosynthetic electron transport. Changes in xanthophyll cycle pool size: short- versus long-term light effects Mechanisms of excess excitation energy dissipation come into play when the potential supply of ATP and reductive equivalents exceeds the demand in photorespiration and carbon assimilation, e.g., under low temperature conditions or in high light. We obtained the best correlation between VAZ/Chl and integrated Q averaged over three days preceding foliar sampling rather than with the light estimate averaged over the season (Figures 7A--D). This result is in accord with the former work indicating rapid adjustment of xanthophyll cycle pool size to step changes in light environment (Björkman and Demmig-Adams 1994, Bilger et al. 1995a). Variability in daily light receipt altered both the carotenoid stoichiometry (VAZ to total carotenoid ratio, Figures 7E and 7F) and carotenoid to chlorophyll ratio. Changes in VAZ/Chl ratio may be attributable to the decline in foliar chlorophyll in response to high light and water stress (e.g., Kyparissis et al. 1995) or to step changes in irradiance (Demmig-Adams et al. 1989). In the current study, there was no evidence of declining chlorophyll concentrations during the drought cycle (Figures 8C and 8D). Instead, rapid changes in VAZ/Chl were mostly attributable to increased total carotenoid accumulation (Figures 8A and 8B; cf. also Figures 7A--D) not changing carotenoid stoichiometry. Interestingly, long-term integrated mean light was a better estimate of carotenoid stoichiometry than light integrated over three days preceding foliar sampling. Quick alteration, in 4--5 days, of Car/Chl ratio in response to a step increase in irradiance has been confirmed in other experiments (Demmig-Adams et al. 1989). However, in the latter study, rapid changes in carotenoid stoichiometry were more important determinants of VAZ/Chl than the alteration of total carotenoid pool size. In general, VAZ/Chl is more variable than VAZ/Car. For example, for a 30-fold change in growth irradiance VAZ/Chl and VAZ/Car varied by a factor of 2 and 0.2, respectively, in cell suspensions of Chenopodium rubrum (recalculated from Schäfer and Schmidt 1991); and VAZ/Chl varied by a factor of four and VAZ/Car by 2.5 across the tropical overstory--understory light gradient (recalculated from Königer et al. 1995). In the current study, there was evidence that VAZ/Car was determined by a long-term light signal or by irradiance during thylakoid development (Figure 7E and F), and that it was the total carotenoid pool size that was flexibly adjusted to fluctuating light conditions. Given that fixed pigment ratios are frequently observed for various thylakoid light harvesting complexes (e.g., Ruban et al. 1994, Iue-chih Lee and Thornber 1995), the difference between VAZ/Chl and VAZ/Car may reflect an overall inflexibility in thylakoid modification in the studied species. Interactions between air temperature and light in determining VAZ pool size In general, high light effects are more damaging at lower temperature (Powles et al. 1980) because the fraction of excitation energy used in photosynthesis decreases. According to laboratory experiments, the rates of violaxanthin de-epoxidation and development of non-radiative energy dissipation decrease with decreasing leaf temperature (Bilger and Björkman 1991, 1994, Adams et al. 1995) potentially further aggravating the stress. However, the final extent of energy dissipation increases with decreasing temperature (Bilger and Björkman 1991, 1994, Adams et al. 1995). Also, the rate of de-epoxidation may not be a relevant factor under field conditions, where leaf temperatures change over a longer time period allowing adequate adjustment of xanthophyll cycle epoxidation state (e.g., Adams et al. 1995). Leaves acclimated to low temperatures also have a large pool of VAZ (Oberhuber and Bauer 1991, Adams and Demmig-Adams 1994, Brugnoli et al. 1994, Haldmann et al. 1996, Thiele et al. 1996, Leipner et al. 1997, but see also Adams et al. 1995), and they are also more resistant to photoinhibition (Haldmann et al. 1996, Thiele et al. 1996, Leipner et al. 1997). In agreement with these observations, we found a negative correlation between daily minimum temperature and VAZ/Chl (Figure 9), and at a common light receipt, VAZ/Chl was larger on days with lower temperature according to a multiple linear regression analysis. Because the excitation energy quenching by photosynthetic carbon assimilation and photorespiration was low on these days (Figure 3), these results support the hypothesis that xanthophyll cycle pool size is regulated by the amount of excess excitation energy (Bilger et al. 1995a, Verhoeven et al. 1997). TREE PHYSIOLOGY ON-LINE at