A model coupling foliar monoterpene emissions to leaf

Transcription

1 Research A model coupling foliar monoterpene emissions to leaf Blackwell Science Ltd photosynthetic characteristics in Mediterranean evergreen Quercus species Ülo Niinemets 1, Günther Seufert 2, Rainer Steinbrecher 3 and John D. Tenhunen 4 1 Department of Plant Physiology, Institute of Molecular and Cell Biology, University of Tartu, Riia 23, EE Tartu, Estonia; 2 Joint Research Centre of the European Commission, Environment Institute, Ispra (Va), Italy; 3 Fraunhofer-Institut für Atmosphärische Umweltforschung, Kreuzeckbahnstraße 19, D-82467, Garmisch-Partenkirchen, Germany; 4 Department of Plant Ecology, University of Bayreuth, D Bayreuth, Germany Summary Author for correspondence: Ülo Niinemets Tel/Fax: ylo@zbi.ee Received: 22 May 2001 Accepted: 24 October 2001 A model was developed and parameterized to simulate monoterpene emission rates per unit leaf area (E) dependent on foliar photosynthetic potentials and activity of monoterpene synthases in the evergreen sclerophylls Quercus coccifera L. and Q. ilex L. Assuming that total activity of monoterpene synthases controls the pathway flux, the product of leaf dry mass per unit area E was culculated as the fraction of total electron flow used for monoterpene synthesis (ε), the rate of photosynthetic electron transport (J) per unit leaf dry mass and the reciprocal of the electron cost of monoterpene synthesis. In the model, light effects on E result from light responses of J, and the temperature relationship of E combines temperature dependencies of J and the specific activity of monoterpene synthase (S S ). Having determined J from leaf photosynthesis data and deriving an estimate of S S from in vitro laboratory measurements, good fits to diurnal time-courses of monoterpenoid emission rates were obtained using a single leaf-dependent coefficient, the total monoterpene synthase activity in the leaves. Our analysis demonstrates that using J as a surrogate of E leads to a realistic description of E, especially under stress conditions where the previous models fail. However, analysis of daily time courses of E indicated that storage of monoterpenoids in nonspecific leaf compartments might alter the correspondence between monoterpenoid synthesis and emission rates, especially after rapid environmental change. Key words: emission model, foliar morphology, monoterpenoid emission, monoterpenoid synthase, photosynthetic electron transport, volatile compounds. New Phytologist (2002) 153: Introduction Because in Mediterranean Quercus species foliar monoterpene emission rates (E) respond to both light and temperature (Bertin et al., 1997; Ciccioli et al., 1997; Staudt & Bertin, 1998; Kesselmeier & Staudt, 1999) similarly to isoprene emission rates (Monson & Fall, 1989; Harley et al., 1996), environmental influences on their E have been modelled using an empirical isoprene emission algorithm (Guenther et al., 1991). The phenomenological model scales the normalized emission rates (E S ) to various incident quantum flux densities and leaf temperatures employing an hyperbolic equation to describe the light dependence of E, and an Arrhenius-type relationship to simulate the temperature response (Guenther et al., 1991). In addition to the standardized emission rate, six leaf-dependent coefficients are required for parameterisation of light and temperature dependencies of emission. Because the leaves of the Mediterranean Quercus species do not possess specific storage compartments like resin ducts in conifers (Tingey et al., 1991), the empirical model assumes no monoterpene storage in the foliage. Although the values of E S are often considered species-specific, New Phytologist (2002) 153:

2 258 Research they may vary by more than an order of magnitude within the canopy of the same species (Niinemets et al., 2002). Such variability in E S along the long-term canopy environmental gradients may partly be associated with increases in leaf dry mass per unit area, which increase the amount of enzymes involved in monoterpenoid synthesis per unit foliar area (Niinemets et al., 2002), as well as the content of enzymes involved in photosynthetic carbon acquisition that provide the reduced carbon, and ATP and NADPH for monoterpenoid synthesis. There is also important evidence of stress- and seasonalityrelated variability in E S in monoterpene emitting species (Yani et al., 1993; Bertin & Staudt, 1996; Bertin et al., 1997; Peñuelas & Llusià, 1999; Llusià & Peñuelas, 2000; Staudt et al., 2000), but the empirical emission models lack the means to describe such physiological responses. The empirical models completely neglect the potential effects of environmental stress factors on the emission rates, and link the seasonality-related changes in E S to the amount of foliage present in various months during the season (Guenther et al., 2000). Yet, enzyme activities responsible for foliar isoprenoid synthesis may change independently of foliar area development (Lehning et al., 2001). In Mediterranean strongly seasonal environments, evergreen sclerophylls often suffer from severe leaf water stress developing during the day and season (Tenhunen et al., 1987). Commonly, the stomata close at midday, leading to lower intercellular CO 2 concentrations and decreased foliar net assimilation rates (Tenhunen et al., 1985, 1987). Although the empirical emission algorithm would predict the highest emission rates at midday at high light and temperature, both leaf-level cuvette measurements (Peñuelas & Llusià, 1999) and canopy level flux measurements by micrometeorological techniques (Moncrieff et al., 1997; Valentini et al., 1997) indicate that monoterpene emission rates may decline at midday from the water stressed foliage of Q. ilex. Photosynthesis provides carbon, NADPH and ATP for monoterpenoid synthesis, and coupling of monoterpene emission rates to foliage assimilation characteristics may provide a valuable basis for simulation of stress effects on the emission rates. Recently, isoprene emission rates in stressed and nonstressed conditions were successfully modelled by coupling foliage photosynthetic characteristics to isoprene synthase activity (Niinemets et al., 1999c; Martin et al., 2000; Zimmer et al., 2000). In a companion paper, we demonstrated strong positive correlations between foliage monoterpenoid emission and photosynthetic electron transport rates in Quercus coccifera and Q. ilex (Niinemets et al., 2002). These correlations provide encouraging evidence that foliage photosynthetic characteristics may also be employed for simulation of monoterpenoid emission from the species lacking specific terpene storage compartments within the leaves. Until lately, there was a broad consensus that plant isoprenoids are exclusively produced from acetyl-coa via mevalonic acid (MVA) pathway (Croteau, 1987; Kleinig, 1989; Gershenzon & Croteau, 1993; McGarvey & Croteau, 1995). Yet, a novel 1-deoxy-D-xylulose-5P (DXP) pathway for isoprenoid synthesis has been recently discovered in certain bacteria, algae and in plastids of higher plants (Lichtenthaler et al., 1997b; Sprenger et al., 1997; Disch et al., 1998; Lichtenthaler, 1999), and the investigations have unequivocally demonstrated the formation of isoprene and monoterpenes via the DXP pathway in plant plastids (Arigoni et al., 1997; Lichtenthaler et al., 1997a; Lichtenthaler et al., 1997c; Lichtenthaler, 1998). Given the crucially differing carbon and coenzyme costs for isoprenoid synthesis via the two pathways, current estimations of plant carbon and energy investments for emission of terpenoids need revision. In the current study, we calculate first carbon and metabolic energy requirements for production of various monoterpenoids emitted, and second develop a novel algorithm for estimation of monoterpene emission rates in dependence on environmental parameters for evergreen Q. coccifera and Q. ilex. The emission algorithm constructed accounts for long-term variability in leaf structure (leaf dry mass per unit area, LMA), photosynthetic electron transport rates and total monoterpene synthase activity. Theory: a model of terpene emission based on the electron cost for monoterpene synthesis Coenzyme requirement for monoterpene synthesis Based on the available biochemical information, Niinemets et al. (1999c) suggested that the minimum cost for synthesis of 1 mol isoprene via the DXP pathway is 17 mol ATP and 14 mol NADPH, and in addition, 1 mol carbon is lost as CO 2 from the 6 mol carbon present in the immediate intermediates of DXP pyruvate and glyceraldehyde 3-phosphate. These calculations assumed that 3 mol NADPH and 1 mol ATP are required for a stoichiometric conversion of 1 mol DXP to dimethylallylpyrophosphate (DPP), which is the immediate precursor of isoprene (Silver & Fall, 1995). Recent findings on the intermediate compounds and coenzyme requirements of the DXP pathway support our estimation of the NADPH cost (Kuzuyama et al., 1998; Takahashi et al., 1998; Fellermeier et al., 1999; Lange & Croteau, 1999b) as well as the requirement of 1 mol ATP for phosphorylation of 1 mol isopentenyl monophosphate (Lange & Croteau, 1999a). However, the latest investigations have demonstrated that at least 3 mol of ATP equivalents are additionally required for DPP formation: 1 mol pyrophosphate is released from CTP in the reaction catalyzed by 4-diphosphocytidyl-2C-methyl-D-erythritol (CMEP) synthase (Rohdich et al., 1999, 2000a), and 1 mol ATP is necessary to convert CMEP to 4-diphosphocytidyl-2C-methyl- D-erythritol-2-phosphate by CMEP kinase (Lüttgen et al., 2000; Rohdich et al., 2000b). Thus, the revised cost of 1 mol isoprene production is 20 mol ATP and 14 mol NADPH. New Phytologist (2002) 153:

3 Research 259 Condensation of isopentenylpyrophosphate and dimethylallylpyrophosphate resulting in formation of geranylpyrophosphate (GPP, Gershenzon & Croteau, 1993), and further isomerization and cyclization of GPP by terpene synthases leading to formation of various monoterpenes (Croteau & Cane, 1985; Croteau et al., 1988) do not require extra NADPH and ATP. Thus, the coenzyme costs of nonmodified monoterpenes are twice the cost of isoprene synthesis, that is, monoterpene synthesis via DXP pathway requires fixation of 12 mol CO 2 and this carbon is reduced to the level of that in monoterpenes with the expense of 40 mol ATP and 28 mol NADPH per mol product. Although a wide range of monoterpenes are emitted by plant species (Schönwitz et al., 1987), most of the terpene synthases produce alkenes, synthesis of which from GPP does not require extra metabolic energy. However, these costs may be different in monoterpenes which undergo enzymatic modifications after cyclization steps, and also for the ether 1,8-cineole. In the formation of 1,8-cineole, both the carbocyclization and heterocyclization steps are catalyzed by 1,8-cineole synthase, and it is conceivable that an enzyme thiol-bound intermediate is involved in the reaction (Croteau & Karp, 1977). Given that the reduced thiol released at the end of the heterocyclization is possibly oxidised by NADP +, production of 1 mol 1,8-cineole requires 1 mol NADPH less (i.e. 27 mol mol 1 ) than the formation of the other monoterpenes. In addition, a number of plant monoterpene hydroxylases are currently known. These enzymes catalyze formation of specific monoterpenoid alcohols at the expense of 1 mol NADPH per mol product produced (Mihaliak et al., 1993). The monoterpenols may further be oxidized by group-specific pyridine nucleotide-dependent monoterpenol dehydrogenases, each of which is able to catalyze ketonization of a narrow range of structurally similar monoterpenols (Croteau et al., 1978; Croteau & Felton, 1980). According to the reaction stoichiometry, dehydrogenation of 1 mol monoterpenol to monoterpenone results in net production of 1 mol NADPH. For the monoterpenoids emitted by the Mediterranean Quercus species, only the NADPH cost of 1,8-cineole, camphor and p-cymene differs from the basic estimate of 28 mol mol 1 (Table 1). However, given that the enzymatically modified terpenes are emitted in large quantities in many other important emitting species, we suggest that a weighted average NADPH and ATP requirement should generally be computed to model the total emissions (Table 1). Fraction of photosynthetic electron transport used for the monoterpene synthesis Because the exact electron requirement for ATP synthesis and the mechanisms by which the balanced production of ATP and NADPH for CO 2 assimilation is achieved in photosynthetic electron transport are still uncertain, recent formulations of the photosynthesis model of Farquhar et al. (1980) have assumed that the rate of ribulose-1,5- bisphosphate (RUBP) regeneration is NADPH-limited (von Caemmerer, 2000). Accepting this, and considering further that 2 mol NADPH per mol CO 2 are necessary to reduce carbon to the level of that in glyceraldehyde-3-phosphate, and that the production of 1 mol NADPH requires 2 mol electrons, the photosynthetic electron transport rate ( J CO, µmol m 2 s 1 2+ O 2 ) required for net carbon fixation with a rate A (µmol m 2 s 1 ) is given by (Brooks & Farquhar, 1985): J CO + O = 2 2 ( A + Rd)( 4Ci + 8Γ*), C Γ* i Eqn 1 where (R d (µmol m 2 s 1 ) is rate of mitochondrial respiration continuing in the light (µmol m 2 s 1 ), Γ* (µmol mol 1 ) is CO 2 compensation point in the absence of R d (Laisk, 1977) and C i is intercellular CO 2 concentration (µmol mol 1 ). Given the tight coupling of sugar and monoterpenoid production, and the similar ratio of NADPH to ATP requirements in both monoterpene (0.70) and sugar (0.67) synthesis, we further adopt the assumption of NADPH-limitation for monoterpenoid synthesis. Because synthesis of 1 mol monoterpene requires assimilation of 12 mol CO 2, the rate of photosynthetic electron transport (J C, µmol m 2 s 1 ) required to produce enough sugars for a monoterpene emission rate of E (µmol m 2 s 1 ) is equal to: J C 12E( 4Ci + 8Γ*) = C Γ* i Eqn 2 Increase of the reduction state of 12 mol carbon from CO 2 to sugars consumes 24 mol NADPH. Additional NADPH is needed to increase the reduction state of carbon in sugars to the level in monoterpenes. Given that 1 mol NADPH costs 2 mol electrons, the extra photosynthetic electron transport rate (J E, µmol m 2 s 1 ) for sugar reduction may be expressed as: JE = 2E( ϑ 24), Eqn 3 where ϑ is the NADPH cost of monoterpenes (mol mol 1 ) found as a weighted average of the costs of all terpene species emitted (Table 1). From Eqns 1 3, the fraction of photosynthetic electron transport rate used for terpene emission, ε, is determined as: JC + JE ε =, Eqn 4 J T where J T is the total photosynthetic electron transport rate. Because a number of other NADPH and ATP consuming reactions like nitrate reduction (Bloom et al., 1989; Holmes et al., 1989) are competitive with respect to carbon fixation, J T is larger than J CO (Eqn 1). Although J may be down-regulated to 2+ O 2 fit the overall reaction capacity in conditions when Ribulose- 1,5-bisphosphate carboxylase/oxygenase (Rubisco) limits New Phytologist (2002) 153:

4 260 Research Table 1 Molar percentage composition (average ± SD) of the monoterpenes emitted and the average NADPH costs of total emissions in Quercus coccifera and Q. ilex Field measurements, site code (Table 2) Laboratory measurements, study Compound NADPH cost (mol mol 1 )* Loreto et al. (1996a)* 2 Quercus coccifera Quercus ilex Q. ilex Loreto et al. (1998a)* 2 New Phytologist (2002) 153: Camphene ± ± ± ± ± Camphor Carene ± ± ± 0.7 1,8-Cineole ± ± 1.6 p-cymene ± 4.7* ± ± 2.5* Isoprene ± 0.60 Limonene ± ± ± ± ± 1.8* ± ± ± Linalool ± ± Myrcene ± ± ± ± ± ± ± cis-β-ocimene trans-β-ocimene α-phellandrene ± ± 0.17 β-phellandrene ± α-pinene ± ± ± ± ± ± ± ± β-pinene ± ± 6.8* ± ± ± ± ± 5.5* ± Sabinene ± ± ± α-terpinene ± ± γ-terpinene ± ± ± α-terpineol ± ± 1.6 α-terpinolene ± 0.4 α-thujene ± ± ± ± Tricyclene ± 0.04 Average NADPH cost (mol mol 1 )* * 1 ATP cost is 40 mole per mole product for monoterpenoid synthesis and 20 mol mol 1 for isoprene synthesis; * 2 potted plants, seed source: Rome, Italy; * 3 sum of p-cymene and 1,8-cineole; * 4 sum of β-pinene and sabinene; * 5 sum of 1,8-cineole and limonene; * 6 average NADPH cost was calculated as the weighted average of all monoterpenes emitted.

5 Research 261 assimilation (Sharkey et al., 1988; Pammenter et al., 1993), leaves carrying out photosynthesis at saturating light often possess a capacity of photosynthetic electron transport, which is higher than that directly required for carbon assimilation (Stitt, 1986; Bloom et al., 1989). Moreover, concurrent measurements of foliar CO 2 and O 2 exchange indicate that the electron transport rates are generally larger than necessary for sugar synthesis also at nonsaturating light intensities (Laisk et al., 1992). Use of a significant fraction of photosynthetic electron flow in nitrate reduction concurrently with carbon assimilation has frequently been observed (Robinson, 1988; Holmes et al., 1989). It has been suggested that this explains the discrepancies between foliar CO 2 and O 2 exchange (Laisk et al., 1992). However, electrons may also be used for further reduction of sugars. We assume that this excess electron transport rate provides the extra electrons (J E, Eqn 3) required to increase the carbon reduction state from the level of sugars (Eqn 2) to that in monoterpenes. In a previous formulation (Niinemets et al., 1999c), the extra electron transport was assumed to originate from mitochondrial catabolism of the photosynthetically fixed carbon. Substituting J C (Eqn 2) and J E (Eqn 3) into Eqn 4 and rearranging gives E (µmol m 2 s 1 ) as: E = ε J T ( C i Γ*). 12( 4C + 8Γ*) + 2( C Γ*)( ϑ 24) i Eqn 5 The total electron transport rate per unit area, J T, is calculated as the product of dry mass per unit area (g m 2 ), and the rate of photosynthetic electron transport per unit leaf dry mass (J m, µmol g 1 s 1 ). In the current model, the light dependence of monoterpene emission (Staudt & Seufert, 1995; Loreto et al., 1996c; Bertin et al., 1997; Ciccioli et al., 1997; Kesselmeier et al., 1997; Schuh et al., 1997; Staudt & Bertin, 1998) results from the response of total electron flow (J T ) to incident quantum flux density. In the photosynthesis model used, J vs light response is fitted by an empirical equation (Harley & Tenhunen, 1991; Harley et al., 1992). Without adequate knowledge of the excess electron transport rate, we calculated here the total rate of photosynthetic electron transport from foliar photosynthesis and monoterpene emission rates as J CO 2 + O (Eqn 2) + J E (Eqn 3). In the previous study, we demonstrated that E is strongly correlated with foliage photosynthetic electron transport rates in the Mediterranean Quercus species (Niinemets et al., 2002). Although there is conclusive evidence indicating that monoterpenes are emitted through the stomata in Q. ilex (Loreto et al., 1996c), stomatal aperture does not appear to directly control the emission rate (Loreto et al., 1996c; Kesselmeier & Staudt, 1999). Sharkey (1991) explained the stomatal insensitivity of isoprenoid emission by low, nonsaturated isoprenoid vapour pressures in leaf intercellular air space, which allows the flux to be maintained by counterbalancing stomatal closure with increases in leaf to air isoprenoid vapour pressure gradient. Thus, we assume that there are no i direct stomatal effects on monoterpene emission. However, there may be indirect effects through influences of stomatal conductance on J T, and C i (Eqn 5, Flexas et al., 1998; Ott et al., 1999; Martin et al., 2000). With decreasing intercellular CO 2 concentrations, rates of photorespiration increase, and more electrons are required to assimilate the same amount of carbon than under higher C i. As in the case of the isoprene emission model (Niinemets et al., 1999c), we assume that ε, which characterizes the competitive strength of volatile isoprenoid synthesis pathway for electrons, depends on the total activity of the enzymes controlling the pathway flux. For isoprene, the activity of isoprene synthase was assumed to determine the flux (Martin, 1997; Niinemets et al., 1999c; Silver & Fall, 1995; Wildermuth & Fall, 1996). Given that the monoterpene synthase reactions are slow with turnover numbers in the range of s 1 (Croteau & Cane, 1985; Alonso & Croteau, 1993) and that monoterpene synthases operate at metabolic branch points and catalyze the first steps leading to various monoterpenes, monoterpene synthase activities may control the pathway flux (Bohlmann et al., 1998). Studies indicate that the monoterpene cyclase activity is closely correlated with monoterpene synthesis at the leaf level (Croteau et al., 1981; Lewinsohn et al., 1991). Therefore, we assume that the total activity of leaf monoterpene synthases is the primary determinant of ε, that is, the ability of monoterpene synthesis pathway to compete for electrons. Temperature dependence of monoterpene emission In addition to the genetic and developmental factors (Burbott & Loomis, 1967; Lerdau et al., 1995; Staudt et al., 1997, 2000; Llusià & Peñuelas, 2000), leaf temperature exerts a strong control over monoterpene synthase activity (Melle et al., 1996; Fischbach et al., 2000). In Mediterranean monoterpene emitting Quercus species, the temperature response of isoprenoid emission is a curve with a maximum for all monoterpenes emitted, except for cis- and trans-β-ocimene (Loreto et al., 1998c; Staudt & Bertin, 1998), indicating that temperature-related changes in enzymatic activities control the emission of most monoterpene species. Analogously to isoprene emission (Niinemets et al., 1999c), we suggest that the temperature dependence of monoterpene emission is controlled both by the temperature responses of monoterpene synthase activities and photosynthetic electron transport. We describe the temperature relationship of the specific activity of monoterpene synthase [S S, nmol monoterpene (g monoterpene synthase) 1 s 1 ] as: S s c e = + 1 e H / RT a ( ST H )/ RT d, Eqn 6 where c is scaling constant, H a (J mol 1 ) is activation energy, H d ( J mol 1 ) is deactivation energy, S ( J mol 1 K 1 ) is entropy New Phytologist (2002) 153:

6 262 Research term, T (K) is leaf temperature, and R (8.314 J mol 1 K 1 ) is gas constant. A similar equation is used to fit the activity of light saturated photosynthetic electron transport rate ( J max, Niinemets & Tenhunen, 1997). Total leaf monoterpene synthase activity (nmol m 2 s 1 ) is the product of leaf monoterpene synthase content (C T, g m 2 ) and S S. To describe the shape of the specific activity of monoterpene synthase vs the temperature response curve, we use the data of total monoterpene synthase activity in crude leaf extracts of Q. ilex (data of Fischbach et al., 2000, see Fig. 2a). The activity expressed per total protein content in crude leaf extract in Fischbach et al. (2000) was converted to specific activity by using a representative value of S S of nmol g 1 s 1 at 30 C obtained for a mixture of monoterpene synthases from Citrus limonum, which catalyze synthesis of α- and β-pinene, and limonene (Cori & Rojas, 1985). Assuming that under light saturation, the leaf monoterpene synthases operate at maximal activity, the light saturated emission rate, E, is equal to S S C T, and an estimate of ε (Eqn 4) is given as: η ε = C T S s, J max Eqn 7 where J max is light-saturated value of total electron transport, and η is electron cost of terpene emission [mol electrons in terpene synthesis (mol monoterpenes emitted) 1 ], converts from terpene units to electron transport units, and is equivalent to (C i Γ*)/[12(4C i + 8Γ*) + 2(C i Γ*)(ϑ 24)] (cf. Eqn 5). Given that C T is a constant for a given leaf, and η is essentially constant with temperature, Eqn 7 supports the argument that the shape of the temperature relationship of ε depends on both the temperature responses of S S (Eqn 6) and J max. However, η may vary because of differences in intercellular CO 2 concentration as well as in CO 2 compensation point in the absence of mitochondrial respiration (Γ*, see Eqn 5). For practical purposes, Eqn 7 allows one to scale ε determined at a certain temperature of T 1 (ε T1 ) to another temperature T 2 (ε T1 ) by the following expression: ε T T T εt max 1 2 J SS = 2 1 T T 2 1 JmaxSS Eqn 8 where the superscripts T 1 and T 2 denote J max and S S values calculated for temperatures T 1 and T 2. Materials and Methods Data sources Most of the experimental work for model parameterisation was conducted in the Mediterranean communities near to Castelporziano, Italy, and Montpellier, France (Table 2). Details of the experimental sites, foliage photosynthetic measurements and monoterpenoid sampling and analysis techniques are reported in Niinemets et al. (2002) and Seufert et al. (1997). The foliage monoterpenoid emissions were monitored during six intensive field campaigns in Castelporziano and during three campaigns in Montpellier To gain insight into the response of emission rates to long-term variability in environment, the measurements were conducted at different heights in the canopy of the sclerophyll forest in Castelporziano, and in both south- and north-exposed twigs in the garrigue ecosystem in Montpellier. Overall, there was a strong positive effect of light availability on area-based foliage photosynthetic capacity, and maximum rates of photosynthetic electron transport and monoterpene emission (Niinemets et al., 2002). This mainly resulted from increases in physiologically active biomass per unit area with increasing light availability in the canopy. Table 2 Data sources for foliage photosynthesis (A) and terpene emission (E) characteristics Study code Site name Community Location Study period Data available (study) 1 Barcelona, Spain* 1 Maquis N, 2 7 E Nov 1996 July 1997 E (Peñuelas & Llusià, 1999; Llusià & Peñuelas, 2000) 2 Castelporziano, Rome, Italy* 2 Sclerophyll forest N, E June 6 11, 1993; May 26, 1994 E (Kesselmeier et al., 1996, 1997) 3 Castelporziano, Rome, Italy* 3 Sclerophyll forest N, E June 1993 October 1994 A and E (Bertin et al., 1997; Seufert et al., 1997; Niinemets et al., 2002) 4 Castelporziano, Maquis N, E May ; E (Street et al., 1997) Rome, Italy ( dunes site) Oct 24, Montpellier, France Garrigue N, 3 45 E July 1 6, 1995; A and E (Niinemets et al., 2002) Oct 16 26, Montpellier, France Garrigue N, 3 45 E October 8 9, 1995 A and E (Niinemets et al., 2002) 7 Quintã São Pedro, Portugal Macchia N, 9 0 W October 1982 May 1983 A (Tenhunen et al., 1984, 1985) * 1 Estimates of seasonal average monoterpenoid composition in sun-exposed twigs; * 2 measurements at a height of 6 m in midday; * 3 estimations from the upper and the lower canopy averaged over the season. New Phytologist (2002) 153:

7 Research 263 For comparison of foliage monoterpenoid compositions between Q. coccifera and Q. ilex as well as between various chemotypes of the same species, a number of additional reports was analysed (Tables 1, 2). Laboratory determinations of temperature dependencies of monoterpenoid emission by four independent studies as well as field observations of temperature responses of E in Castelporziano and Montpellier (Fig. 2) were analysed to generalize the temperature relationships. These measurements were conducted at cuvette CO 2 concentrations around 350 µmol mol 1 and with variable quantum flux densities (µmol m 2 s 1 ), which were, nevertheless, close to saturating in most studies: 400 in Staudt & Seufert (1995), 700 in Loreto et al. (1998c), 1000 in Bertin & Staudt (1996) and 2000 in Staudt & Bertin (1998). For the Castelporziano and Montpellier measurements, only the monoterpene emission rates measured at quantum flux densities larger than 700 µmol m 2 s 1 were included. The emission rates were standardized to the value at 25 C, allowing direct comparison of the shape of the temperature response curves. The emission data of Loreto et al. (1998c) were standardized to 25 C using the same value for the activation energy between 25 C and 30 C as in the other data. Derivation of photosynthesis model parameters Having determined the shape of monoterpene synthesis vs temperature response curve (Eqn 6, Fig. 2a), monoterpenoid emission rates in dependence on incident quantum flux density and leaf temperature may be simulated by Eqns 5 and 7. Thus, both the rates of photosynthetic electron transport ( J ) as well as the temperature response curve of J must be determined. Photosynthetic electron transport rates were calculated by Eqn 1 as described previously (Niinemets et al., 1999b,c, 2002). In simulations of the diurnal variabilities of photosynthetic electron transport and monoterpenoid emission rates in Montpellier and Castelporziano, daily timecourses of mitochondrial respiration rate, R d, were obtained using an activation energy of 55.1 kj mol 1 (Niinemets et al., 2002) and calculating the scaling constant from the night measurements of R d (Niinemets & Tenhunen, 1997). For the measurements of Quintã São Pedro (Table 2, Fig. 1a) the maximum electron transport rate, J max, was calculated from net assimilation rates measured at a cuvette CO 2 concentration of 2470 µmol mol 1 and incident quantum flux densities (Q) between 600 and 2000 µmol m 2 s 1 (on average 1670 µmol m 2 s 1 ) according to Niinemets et al. (1999b). For the Castelporziano and Montpellier measurements (Fig. 1b), J max was obtained from field daily timecourses of net photosynthesis estimated at current ambient CO 2 concentrations of c. 350 µmol mol 1 using late-morning and midday data, for which Q exceeded 700 µmol m 2 s 1 (A max ). Values of J max were calculated from A max according to Niinemets et al. (1999b), and provide an estimate of electron Fig. 1 Review of observed temperature responses of light-saturated values of photosynthetic electron transport rate standardized to 25 C (J max ) in Quercus coccifera (a) and Q. ilex (b). The measurements depicted in (a) were conducted at Quintã São Pedro, Portugal (Table 2). Data were fitted by a temperature equation (see Eqn 6 for the definition of temperature constants), and the optimum temperatures were calculated from the temperature parameters according to Niinemets et al. (1999a). transport rate required to achieve the measured net photosynthesis rates at light-saturation. Because under current ambient CO 2 concentrations A max is generally limited by Ribulose-1,5-bisphosphate carboxylase/oxygenase (Rubisco) rather than by electron transport, these estimates of J max should be interpreted as the lower limit of the actual capacity. In simulations of the carbon cost of monoterpenoid emission (Fig. 4e,f), the instant rates of net assimilation at 25 C were modelled according to Niinemets & Tenhunen (1997) using the Michaelis-Menten constants of Rubisco and the specificity factor of Rubisco from Jordan & Ogren (1984), the photosynthetic electron transport capacity ( J max ) of 67.9 µmol m 2 s 1, the maximum carboxylase activity of Rubisco of 23.3 µmol m 2 s 1, the rate of mitochondrial respiration of 0.48 µmol m 2 s 1 and the intercellular oxygen concentration of 198 mmol mol 1. These parameters provide New Phytologist (2002) 153:

8 264 Research a realistic estimate of photosynthetic capacity for high light acclimated leaves of Q. coccifera and Q. ilex. Further details regarding the photosynthesis model parameterisation, as well as morphological and physiological determinants of foliage photosynthetic capacity for the Mediterranean Quercus species are given in Niinemets et al. (2002). Results Composition of emitted monoterpenoids in Q. coccifera and Q. ilex Overall 21 monoterpenes and their derivatives have been observed in the emission patterns of Q. coccifera and Q. ilex (Table 1). At the Montpellier and Castelporziano sites, the most important monoterpene emitted by both species was generally α-pinene followed by β-pinene and sabinene (Table 1). These three monoterpenes made up 65 90% of total emissions. However, limonene was emitted in largest quantities in a Q. ilex chemotype studied in Montpellier in October 1995, and in Q. coccifera and Q. ilex chemotypes studied in Barcelona (Table 1), indicating that a broad intraspecific variability may exist in the composition of emitted monoterpenes. The average coenzyme cost of most monoterpenes emitted was 28 mol NADPH and 40 mol ATP per mol product (Table 1). Minor amounts of oxygenated and reduced monoterpenes were always emitted, slightly altering the weighted average NADPH cost of monoterpene synthesis (Table 1). However, the NADPH requirement for monoterpene synthesis varied only by c. 0.5% across the studies, indicating that the contribution of oxygenated and reduced monoterpenoids was very low in these species. Temperature responses of photosynthetic electron transport, in vitro monoterpene synthase activity, and monoterpene emission Recalculation of the data of Tenhunen et al. (1985) measured under saturating intercellular CO 2 concentrations demonstrated that the optimum temperature (T opt ) of photosynthetic electron transport lies between 33 and 36 C in Q. coccifera, but also that it may vary during the season (Fig. 1a). However, the light-saturated photosynthetic electron transport rates calculated from the net assimilation rates measured under current ambient CO 2 concentrations (Eqn 1) did not exhibit appreciable decline at high temperatures for the leaf temperature range observed during the Castelporziano and Montpellier measurement campaigns either in Q. coccifera (data not shown) or in Q. ilex (Fig. 1b). Although there was some evidence of decreases in J max at leaf temperatures over 30 C in Q. ilex sampled in July 1995 (Fig. 1b), the overall T opt for J max was seemingly above 40 C (Fig. 1b). A T opt -value of 40.6 C (Fig. 2a) was calculated for the temperature dependence of in vitro monoterpene synthase activity (Fischbach et al., 2000). Similar optimum temperatures were obtained for laboratory (Fig. 2b) and field (Fig. 2c,d) light-saturated monoterpene emission rates. However, the shapes of the in vivo and in vitro temperature dependencies differed (Fig. 2a,b d). At temperatures above 30 C, the standardized in vivo monoterpene emission rates were higher, and at temperatures below 25 C lower than expected on the basis of temperature response of in vitro monoterpene synthase activity. The fraction of electrons in monoterpene synthesis in relation to temperature The temperature relationships of both the in vitro monoterpene synthase activity (Fig. 2a) and in vivo monoterpene emission rates (Fig. 2b d) were steeper than the temperature responses of photosynthetic electron transport. The standardized photosynthetic electron transport rate (averaged response in Fig. 1a) increased from 0.77 at 20 C to 1.17 at 30 C, and the ratio of the standardized rate at 30 C to 20 C (Q 10 ) was For the same temperature range, monoterpene synthase activity (Fig. 2a) varied from 0.66 to 1.48 (Q 10 = 2.25), and the in vivo monoterpene emission rates from 0.57 to 1.65 (Q 10 = 2.88) in the data of Staudt & Bertin (1998, Fig. 2b). These differences in temperature response suggest that the monoterpene synthesis pathway becomes an increasingly stronger electron sink with increasing leaf temperature. Our data demonstrate that the fraction of electrons used for monoterpene synthesis (ε) is greater at higher temperature (Fig. 3). Using the temperature responses of in vitro monoterpene synthase activity and photosynthetic electron transport rate, reasonable fits between measured and calculated (Eqn 8) values of ε (Fig. 3) were obtained. However, the modelled temperature response curve of ε was substantially affected by the specific temperature responses of J max used in simulations (Fig. 1a,b), and generally did not fit the data well at low temperatures. The lack of fit at low temperatures mostly resulted from a general discrepancy between in vitro activity of monoterpene synthase, and in vivo monoterpene emission rate at low temperature (cf. Fig. 2a,b d). Dependence of the carbon and electron cost of monoterpene emission on intercellular CO 2 concentrations For all data pooled (Niinemets et al., 2002), where primarily leaf exposure, seasonality and temperature altered the monoterpene synthase activity in the leaves, the daily average proportion of electrons used for monoterpene production (mol mol 1 ) varied from to with an average ± SD for individual leaves of ± Corresponding daily average percentage of assimilated carbon lost because of the emission ranged from 0.019% to 11.8%, and averaged (± SD) 1.7 ± 2.1%. Thus, under certain conditions, plants New Phytologist (2002) 153:

9 Research 265 Fig. 2 Comparison of temperature dependencies of in vitro total monoterpene cyclase activities (a), and foliar monoterpene emission rates (b d) standardized to 25 C in Q. ilex (a c) and in Q. coccifera (d). The maximum monoterpene cyclase activities (a) were measured in crude leaf extracts, and characterise the total activity of all foliar monoterpene cyclases (Fischbach et al., 2000). The values depicted (Fischbach et al., 2000) are averages of two separate measurement series. Monoterpene emission rates (b d) were found as the sum of all monoterpenes emitted, except for the data of Staudt & Bertin (1998), where the emission rates of cis- and trans-β-ocimene were excluded. In (b d), the standardized temperature response curve of monoterpene cyclase activity (a) has also been depicted. Data were fitted by a temperature equation (see Eqn 6 for the definition of temperature constants), and the optimum temperatures were calculated from the temperature parameters according to Niinemets et al. (1999a). may lose a large fraction of the assimilated carbon via monoterpene emission (see Staudt & Bertin, 1998 for an observation of up to 20% of assimilated carbon lost due to monoterpene emission in Q. ilex). The fraction of electrons used for monoterpene synthesis varied c. 350-fold, and the fraction of carbon lost c. 600-fold, indicating that the relationship between monoterpene emission and photosynthetic electron transport is more conservative than that between net assimilation and monoterpene emission rate (Niinemets et al., 2002). Because seasonal variability in leaf physiology as well as in environmental conditions was the major factor altering the fraction of electrons and carbon going into isoprene synthesis, we calculated daily integrated rates of monoterpenoid emission (E int ) and net assimilation (A int ) to gain insight into the effects of seasonal changes in average day-time intercellular CO 2 concentrations (C i av ). Daily integrated rates of monoterpene emission were negatively correlated with C i av (Fig. 4a). However, the effects of C i av on A int were variable (Fig. 4b), depending on the covariation of A int with the environmental variables. The correlation was positive in mid-season for strongly water-stressed leaves under high daily irradiances. The correlation tended to be negative in late season in well-watered plants. In those plants, the assimilation rates were mainly constrained by low temperature and light, and the values of A int and the reduction in C i av were greater on days with higher irradiances. Despite varying A int vs C i av relations, the daily average percentage of assimilated carbon lost as monoterpenes, 12E int /A int 100, was consistently negatively related to C i av (Fig. 4c). Simulations demonstrate that the increase of carbon losses with decreasing C i av may be explained by a stronger control the intercellular CO 2 concentrations exert over net photosynthesis than over monoterpene synthesis rates (Fig. 4e). Given that a fixed fraction of electrons is available for monoterpene synthesis, the rates of monoterpene New Phytologist (2002) 153:

10 266 Research Fig. 3 Fraction of photosynthetic electron transport used for monoterpene production (ε, Eqn 4) in relation to leaf temperature at quantum flux densities greater than 700 µmol m 2 s 1. The curves represent theoretical fits to the data, and were calculated by Eqn 7 using either (dashed line) the temperature dependence of J max for Q. coccifera with an optimum temperature (T opt ) of 35.2 C (Fig. 1a), or (solid line) the temperature dependence of J max obtained for the field data in Q. ilex with a T opt of 45.3 C (Fig. 1b). Separate fits to the data correspond to various maximum monoterpenoid synthase activities in the leaves: (1) late-october measurements in Q. coccifera showing extremely low values of E (solid squares), (2) field measurements in Q. ilex in the Castelporziano site (open triangles), and (3) mid- and late-season data for Q. coccifera (solid diamonds) and Q. ilex (open diamonds) from the Montpellier site (Table 2). For each set of data, emission should increase with increasing C i (Fig. 4d f ), because the electron cost of monoterpene emission decreases with increasing C i (Fig. 4d, cf. Eqn 5 and Eqn 7). Although the correlations between monoterpene emission rates and C i av were negative rather than positive (cf. Figure 4a vs Figure 4e), this was attributable to higher values of ε in the upper canopy leaves that were under higher irradiances and possessed a lower C i (data not shown; Niinemets et al., 2002). The simulations in Fig. 4 were conducted with daily average variables to include all data sampled over the season and to obtain the largest achievable range in the electron cost of monoterpenoid synthesis and C i. Given that daily average C i values are integrating diurnal variations in temperature and irradiance, the use of integrated values may potentially lead to an important loss of information. Because of inherent limitations of monoterpene analysis that result from the need to concentrate the samples before the gas-chromatographic analysis, the timeresolution of daily data was not appropriate for a more detailed analysis (Niinemets et al., 2002). However, the simulations in Fig. 4e are equally valid for a diurnal variability in monoterpenoid the electron cost of terpene emission (η) was calculated from the data as the average of all measurements. Values of 93 mol mol 1 (upper fits), 180 mol mol 1 (middle) and 98 mol mol 1 (lower) were obtained for the current data. Having determined J max, S S (Eqn 6) and η, the content of monoterpene synthase (C T, see Eqn 7) was fitted to give the best agreement with the data. Fig. 4 Measured (a d) and simulated (e,f) effects of intercellular CO 2 concentrations on monoterpene emission (a,e) and net assimilation rates (b,e), on the percentages of assimilated carbon lost as monoterpenes (c,f), and on the electron cost of monoterpene synthesis [η; (d,f), s. Eqn 7]. In (a d), the intercellular CO 2 concentrations were averaged over the light period, but the rates of monoterpene emission and net assimilation were integrated, and the percentage of assimilated carbon lost as monoterpenes, and the values of η were averaged over the entire day. In calculations of the CO 2 -percentage lost via monoterpene emission, the emission of one mole of monoterpene was assumed to cost 12 mol of carbon (see the Theory section). In (e) and (f), the fraction of electrons in monoterpene synthesis was fixed at 0.03, and the modelled rate of monoterpene emission was calculated by Eqn 5. Further details of these simulations are reported in the Material and Methods. Solid triangles measurements in Q. ilex in Castelporziano during August 1 5, 1994, solid diamonds measurements in Q. coccifera in Montpellier, October 22 26, 1996 (Table 2). New Phytologist (2002) 153:

11 Research 267 Fig. 5 Daily time-courses of incident quantum flux density and temperature (a,d), net carbon assimilation and photosynthetic electron transport rates (b,e) and measured and modelled monoterpene emission rates (c,f) in a branch from the upper canopy in the Castelporziano forest site (a c) and in a south-exposed branch in the Montpellier garrigue site (d f, Table 2). The monoterpene emission rates were modelled by Eqn 5, scaling ε to different temperatures (Fig. 3) using Eqn 8. The temperature responses of J max from Fig. 1a (February data), and S S from Fig. 2a were employed in the calculations. The initial estimate of ε was computed from light-saturated monoterpene emission rates as an average of all estimates between 20 and 25 C. In both cases (c and f), a single parameterisation was used for all days. emission, and agree qualitatively with the experimental observations (Loreto et al., 1996c). Because of the diurnal and seasonal variability in C i, which more strongly controls photosynthesis than monoterpene synthesis (Fig. 4e,f), the fraction of carbon lost as monoterpenes also varies during the day and between the days (Fig. 4a d). Thus, we conclude that under fluctuating environmental conditions, there should be no general relationship between monoterpene emission and net assimilation rates, even though the fraction of electrons going into monoterpene synthesis may remain constant (Fig. 4e,f ). Model application: measured and modelled daily time-courses of monoterpene emission Having determined the temperature relationships of J max and the specific activity of monoterpene synthase (Eqns 6 and 7), which are considered as essentially constant for all leaves in the canopy, daily time-courses of monoterpene emission rate can be modelled with only one leaf dependent coefficient, the total foliar monoterpene synthase activity (C T S S, Eqn 7). These simulations yielded a good fit to the experimental data (Fig. 5) with a few important systematic discrepancies. First, there was evidence of a slower increase in measured monoterpene emission rates than modelled emission rates (Fig. 5c,f ). Second, monoterpene emission did continue with low rates during the night with average ± SD of ± nmol m 2 s 1 for the data in Fig. 5f (night rates of monoterpene emission were not measured during the May 1994 Castelporziano campaign). Finally the total monoterpene synthase activity calculated as an average over the three modelled days overestimated monoterpene emissions for days with low temperature, and underestimated for days with high temperature (Fig. 5f ). All these inconsistencies are in accord with the hypothesis that the leaves of Q. coccifera and Q. ilex possess a certain capacity for monoterpene storage, and that the emission rates under low temperature conditions (in the morning, during days with low temperature) are not limited by monoterpene synthesis, but by volatilization rates (Niinemets et al., 2002). There were strong positive correlations between foliar monoterpene emission and photosynthetic electron transport rates during the day (Fig. 6). For the data of three modelled days pooled, New Phytologist (2002) 153:

12 268 Research Fig. 6 Correlations between photosynthetic electron transport and monoterpene emission rates during the day. Same data as in Fig. 5. Regression lines and r 2 values are calculated by linear regression and are all significant at P < r 2 was 0.78 for May 1994 data, and r 2 = 0.79 for October 1996 data (P < for both). Compatible with the discrepancies observed between modelled and measured emission rates (Fig. 5f ), the slopes of these relationships varied with average daily temperature, being larger on days with higher temperature (Fig. 6b). Possibly because of the terpene storage at low temperatures, the monoterpene emission vs electron transport relationship tended to have a negative intercept for the May 1994 data (P < 0.03 for all data of the three days pooled in Fig. 6a). Within single day measurements, the intercept was significantly different from zero only for May 26 data (P < 0.05). Significant night emissions resulted in a positive intercept for October 24 data, but the intercepts did not differ significantly from zero on other October days modelled (Fig. 6b). Discussion Fractional composition of emitted monoterpenoids Because the reactivity of various plant monoterpenes in atmospheric ozone forming and consuming reactions varies by several orders of magnitude (Fehsenfeld et al., 1992; Benjamin & Winer, 1998; Neeb, 2000), it is important to identify the range of variation in and the controls over the monoterpene composition. Large differences in the composition of emitted monoterpenes were found between the sites and among the studies (Table 1). In most cases, we identified α- and β- pinene, and sabinene as the most important monoterpenes emitted by Q. ilex and Q. coccifera. This is compatible with the majority of previous studies (Bertin & Staudt, 1996; Loreto et al., 1996b; Staudt & Bertin, 1998). However, limonene is emitted in largest quantities in specific chemotypes of Q. coccifera and Q. ilex in Montpellier as well as in Barcelona (Peñuelas & Llusià, 1999; Llusià & Peñuelas, 2000, Table 1). Studies indicate that the stoichiometry of foliar volatile isoprenoids emitted (Tobolski & Hanover, 1971; Hayashi & Komae, 1974; Gleizes et al., 1982; Merk et al., 1988; Janson, 1993; Canard et al., 1997; Tognetti et al., 1997; Loreto et al., 1998b; Staudt et al., 2001) as well as the maximum total emission rates (Evans et al., 1982; Guenther et al., 1994; Owen et al., 1997) may be under genetic control, possibly because of species- and genotype-specific activities of various monoterpene cyclases. The composition of emitted monoterpenoids may be a relevant variable affecting the coenzyme cost of terpene synthesis, in particular, in species or chemotypes emitting mostly monoterpenoid 1,8-cineole (Hayashi & Komae, 1974; Guenther et al., 1994; Hansen et al., 1997; Owen et al., 1997, Table 1) or synthesizing large amounts of monoterpene derivatives (König et al., 1995). In the current study, the oxygenated and reduced monoterpenoids were emitted only in trace amounts. Although the relative differences in modified monoterpenoids were large between the sites and studies (Table 1), part of this variability certainly resulted from the limitations of the equipment to resolve and accurately integrate the peaks of modified monoterpenoids emitted in trace quantities (Larsen et al., 1997; Niinemets et al., 2002). Cis- and trans-β-ocimene have occasionally been found in large amounts in the emissions from Q. ilex, especially under high leaf temperatures (Staudt & Bertin, 1998), but we did not observe these acyclic compounds in detectable quantities in any of our field campaigns, and only minor amounts of these monoterpenes have been measured in other studies (Table 1, Loreto et al., 1998c). Given that acyclic monoterpenes may be the intermediates leading to cyclic monoterpenes (Gleizes et al., 1982) or may be potentially formed nonenzymatically from the immediate precursor of all foliar monoterpenes geranylpyrophosphate (GPP, Loreto et al., 1998c), we suggest that high rates of β-ocimene emissions may be indicative of a situation with high foliar GPP levels and inhibition of monoterpene cyclases, for example, under heat stress. Alternatively, one may speculate that the GPP-isomerization activity of monoterpene synthases is less affected by temperature than the cyclization activity. New Phytologist (2002) 153: