Isoprene emission from leaves is temperature dependent and may protect leaves from damage at high temperatures. We measured the temperature of white oak (Quercus alba L.) leaves at the top of the canopy. The largest short-term changes in leaf temperature were associated with changes in solar radiation. During these episodes, leaf temperature changed with a 1 min time constant, a measure of the rate of temperature change. We imposed rapid temperature fluctuations on leaves to study the effect of temperature change rate on isoprene emission. Leaf temperature changed with a 16 s time constant; isoprene responded more slowly with a 37 s time constant. This time constant was slow enough to cause a lag in isoprene emission when leaf temperature fluctuated rapidly but isoprene emission changed quickly enough to follow the large temperature changes observed in the oak canopy. This is consistent with the theory that isoprene functions to protect leaves from short periods of high temperature. Time constant analysis also revealed that there are two processes that cause isoprene emission to increase with leaf temperature. The fastest process likely reflects the influence of temperature on reaction kinetics, while the slower process may reflect the activation of an enzyme.
The rate of isoprene (2-methyl-1,3-butadiene) emission from leaves is strongly dependent on leaf temperature. Previous work has determined the Arrhenius activation energy (Ea) for isoprene emission to be between 54 and 164 kJ mol–1 (Tingey et al. 1979, 1987; Monson & Fall 1989; Loreto & Sharkey 1990; Sharkey & Loreto 1993). These numbers correspond to temperature coefficients (Q10, the change in isoprene emission rate with a 10 °C increase in leaf temperature) between two and eight. The temperature dependence of isoprene emission over the short term – minutes to hours – is formalized in models by a three-parameter equation developed by Guenther, Monson & Fall (1991). Longer term temperature effects on isoprene emission have only recently been studied (Sharkey et al. 1998). The basal rate of isoprene emission, defined as the rate of emission at standard conditions of 30 °C and 1000 μmol photons m–2 s–1, varied with air temperature and photosynthetic photon flux density (PPFD) experienced by the plants over the previous 2 d. To our knowledge, no data exist on the response of isoprene emission to very rapid temperature changes that leaves may experience in forest canopies due to rapid fluctuations of wind or PPFD.
Leaves in forest canopies experience rapid light fluctuations, termed sunflecks, which occur frequently throughout the day (Pearcy 1990). Sunflecks are quite variable in duration (Pearcy, Krall & Sassenrath-Cole 1996). The shortest sunflecks are caused by leaf fluttering during small breezes and occur with a frequency of 3 Hz (Roden & Pearcy 1993). The more important sunflecks are those caused by gaps between clouds or by the swaying of the treetops in the wind, and last for several minutes. These represent 75% of sunfleck PPFD and are often when leaves experience the highest light levels. These bursts of full sunlight can cause leaves to heat up for short periods many times throughout the day. This phenomenon is known as temperature ramping (Shaw, Paw & Gao 1989; Roden & Pearcy 1993).
We have theorized that isoprene may function specifically to protect leaves from damage by transient heating events, or heatflecks, caused by sunflecks in forest canopies (Sharkey & Singsaas 1995; Singsaas et al. 1997). This is because isoprene emission can quickly respond to a heatfleck while other known thermotolerance mechanisms – changes in xanthophyll epoxidation state, synthesis of heat shock proteins, and changes in membrane lipid composition – can take many minutes, hours, or days to respond to a temperature episode (Pearcy 1978; Vierling 1991; Havaux & Tardy 1996; Nilsen & Orcutt 1996; Heckathorn et al. 1998). This hypothesis fits with the observation that many forest canopy plants make isoprene while plants that experience high temperatures continuously, such as desert plants, do not use isoprene emission as a thermotolerance mechanism.
To further support this hypothesis, we studied the natural leaf temperature fluctuations experienced by canopy trees and the effects of rapid fluctuations on isoprene emission. We analysed canopy micrometerological data collected as part of field research during July 1995 to determine the frequency, amplitude, and duration of leaf temperature fluctuations at the top of an oak canopy. Isoprene emission and leaf temperature were measured in the laboratory while leaf temperature fluctuated rapidly. We measured the time constants for isoprene emission and leaf temperature changes and compared them with time constants for leaf temperature fluctuations in the field data.
MATERIALS AND METHODS
Temperature and PPFD were recorded at the surface of a leaf at the top of a 30 m tall white oak (Quercus alba L.) tree in Duke University Research Forest (35° 58’ 25" N latitude and 79° 06’ 05" longitude) in Chapel Hill, North Carolina. A 40 m walk up tower was used to access the canopy. The forest is a mature, second growth, uneven aged stand with the oldest trees exceeding 180 years of age.
Micrometerological measurements were logged every 5 s with a data logger (model CR10, Campbell Scientific, Logan, UT, USA). PPFD was measured with galium– arsenide photodiodes (Hamamatsu Corp., Bridgewater, NJ, USA) stuck through the leaf blade and held against the leaf surface. Two photodiodes were used per leaf to simultaneously measure PPFD above and below the leaf. All connections were made with 0·07 mm diameter copper wire (Omega Engineering, Stamford, CT, USA) so the leaf remained in its natural orientation and was allowed to move freely. The current output from the photodiodes was bridged with a 1% 100 Ω resistor so the signal, in mV, was equal to PPFD (μmol m–2 s–1) divided by 10. Leaf temperature was measured with chromel–constantan thermocouples welded from 0·07 mm diameter wire. Three pieces of wire, two chromel and one constantan, were welded to make two thermocouple junctions. Wires were threaded through the veins on the underside of the leaf and one junction was pressed against the bottom of the leaf. The other thermocouple junction was suspended 1 cm below the leaf. This thermocouple measured the difference between leaf and air temperature (ΔT). Air temperature was measured in the vicinity of the leaves (not more than 0·5 m away) with another chromel–constantan thermocouple shaded by aluminium foil.
Gas exchange measurements
Single, detached leaves of a 40 year old red oak (Quercus rubra L.) tree growing outdoors in Madison, WI, USA were used in all gas exchange experiments. Leaves from a fully sun-exposed portion of the tree were selected. The leaf was removed from the tree by immersing the petiole in deionized water and cutting with a sharp razor blade. The leaf was brought indoors with the cut end under water and placed inside a gas exchange cuvette. The petiole remained immersed in water during all measurements. Detached leaves from this tree, collected in this manner, have previously been used in gas exchange and isoprene emission measurements with no discernable effects on photosynthesis or isoprene emission responses (Loreto & Sharkey 1990).
Leaves were clamped in an aluminium gas exchange cuvette with a 2·5 cm2 window. Air flow was pulled from a clean air source through the leaf cuvette and into a reaction chamber by a pump for isoprene measurement. Isoprene was measured using a real time chemiluminescent isoprene detector (Hills & Zimmerman 1990; Hills, Fall & Monson 1992). Data were logged once per second. The isoprene analyser was calibrated daily with a four point standard drawn from a 6 ppmv (parts per million by volume) isoprene standard in nitrogen (Scott-Marin Specialty Gases, Riverside, CA, USA). The response of the analyser has been shown to be linear over three orders of magnitude bracketing the concentrations used in these experiments (Hills & Zimmerman 1990). The sensitivity and time constant for isoprene detection of the system were determined by injecting isoprene standard gas into the empty leaf cuvette with the system running. Aliquots of 6 ppmv isoprene standard ranging from 0·1 to 10 cm3 were injected through the rubber gasket of the cuvette with a gas-tight syringe (Hamilton Co., Reno, NV, USA). The time between injection and detection and the total isoprene detected were recorded. The difference between detected and injected isoprene was less than 2% in all cases. Leaf thickness was measured with a Vernier caliper.
Incident light was provided by using a pair of 2·5 kW xenon-arc lamps. To allow rapid changes in leaf temperature, one light beam was filtered through 700 nm cut-off infrared filters (cold lamp) while the other was unfiltered (hot lamp). Light from both lamps was attenuated with neutral density filters so the PPFD was 1000 μmol m–2 s–1 from each lamp. Switching from the cold lamp to the hot lamp with shutters increased the infrared heat load on the leaf five-fold while the PPFD remained constant (E.L. Singsaas, unpublished results). PPFD was measured with a quantum sensor (Model LI-190, Li-Cor Inc, Lincoln, NE, USA). Spectral density was identical for both lamps as measured with a spectroradiometer (Model LI-1800, Li-Cor). Total radiant energy output from the lamps was measured by a solarimeter (Kipp Instruments, Amsterdam, The Netherlands).
Finite difference model
The time dependence of isoprene emission was simulated with a finite difference model of leaf temperature and isoprene emission. The time dependence of leaf temperature was modelled for a specific temperature jump with a time constant equation, describing the temperature at time t (s) following the imposed change:
where Tt is the leaf temperature at time t, T0 is the initial temperature, T∞ is the final temperature, and τ is the time constant for leaf temperature change.
Isoprene emission was a composite function, dependent on both temperature and time. The temperature dependence of isoprene emission, normalized such that it takes on a value of 1 at 30 °C, was simulated with an Arrhenius function:
where Ea is activation energy (kJ mol–1), R is the gas constant, TS is the normalizing temperature (K) and T is leaf temperature. This equation is a modified version of the standard temperature correction equation for isoprene emission (Guenther et al. 1993). This was used because the standard equation for the temperature response of isoprene emission contains a parameter describing the temperature at which isoprene emission decreases with increasing temperature. Isoprene was not observed to decrease at high temperatures in our experiments. This was combined with the time constant equation to give:
where It* is the equilibrium-normalized isoprene emission rate at time t, predicted by Eqn 2, and Δt is the time interval of the model (s). When isoprene emission rates were predicted, Eqns 2 and 3 were scaled by multiplying by a basal emission rate, IB, defined as the isoprene emission rate at 30 °C and 1000 μmol photons m–2 s–1 (Guenther et al. 1993). To express isoprene emission on a leaf area basis, IB was set to 30 nmol m–2 s–1. This model was fitted to isoprene emission data using least-squares regression. When isoprene emission was simulated using leaf micrometerology data, the light dependence of isoprene emission was included. The equation which describes the light dependence of isoprene emission, CL, was simulated using the equation of Guenther et al. (1993):
where α and CL1 are empirically fitted constants, and L is PPFD in μmol m–2 s–1. For simulations, the constants were set to α = 0·0027 and CL1 = 1·066 (Guenther et al. 1993). This was multiplied by the temperature response equation (Eqn 2) to predict isoprene emission response to temperature and light.
Isoprene emission rates were calculated as flow rate of air through the leaf cuvette times the isoprene concentration in the air (Hills et al. 1992). The Arrhenius equation was fitted by non-linear regression. To determine the time constant for temperature or isoprene emission changes, Eqn 1 was rearranged to:
Plotting ln [(Tt–T0)/(T0–T∞)] versus t yields a straight line, the slope of which is – 1/τ. Time constant data are also reported as half life (T1/2) such that T1/2 = τ ln(2).
Leaves experience fluctuating temperatures continuously throughout the day. One hour of data collected on 29 July 1995 are shown in Fig. 1 as an example. Most of the fluctuations are rapid, low amplitude oscillations of 1–3 °C occurring even when light on the leaf is constant. The larger amplitude changes in leaf temperature occur during large changes in PPFD (Fig. 1). These episodes of high temperature usually lasted between 15 s and 20 min.
To determine the time constant for leaf temperature changes during high temperature episodes experienced by the leaf, an episode where leaf temperature increased by more than 8 °C was analysed. PPFD and leaf temperature data were normalized to the exponential time constant equation (Eqn 1) and plotted on a logarithmic scale (Fig. 2). The slope was – 0·0174 for PPFD and – 0·0111 for leaf temperature which correspond to τ of 57·5 s (T1/2 of 39·9 s) and 90·1 s (T1/2 = 69·7 s). When all occurrences throughout the day were analysed, the time constants averaged 31·2 s for PPFD and 56·1 s for leaf temperature (T1/2 = 21·7 and 38·9 s, respectively).
Isoprene emission and temperature fluctuations
The response of isoprene emission to rapid temperature fluctuations was studied by monitoring leaf temperature and isoprene emission rate continuously from leaves while switching from the hot lamp to the cold lamp. Switching between the two lamps allowed instantaneous changes in the radiant energy load on the leaf, resulting in leaf temperature changes of 6–7 °C for most leaves. Leaf temperature changes followed an exponential curve generally lasting 30–60 s (Fig. 3). The time constant, τ, of temperature changes was 16 s (T1/2 = 11 s). The time constants for leaf temperature increases and decreases were not significantly different (P = 0·23, n = 4).
Two other time constants in the leaf–cuvette system were also characterized. The time constant of the cuvette for changes in isoprene concentration was determined by injecting known quantities of isoprene standard gas into the cuvette through the rubber gasket. The isoprene reached the detector within 1 s and cleared the system with a time constant of 0·9 s. The time constant for isoprene in the leaf airspaces was calculated using transfer function analysis (Stephanopoulos 1984) where the time constant equals the product of storage capacitance and resistance. The storage capacitance of a leaf equals the total airspace volume of the leaf. The thickness of the red oak leaves used in the experiments averaged 0·14 mm. The airspaces were assumed to occupy 25% of leaf volume, average for deciduous trees (Turrell 1936; Dengler & Mackay 1975), resulting in a leaf airspace volume of 0·036 m3 m–2. At 25 °C and 99 kPa, the air volume of an oak leaf is 1·4 mmol m–2. Stomatal conductances to water were between 0·05 and 0·2 mol m–2 s–1 during our experiments; stomatal conductance to isoprene is equal to stomatal conductance to water divided by 2·83 (Singsaas et al. 1997). Assuming this range of stomatal conductances, the leaf airspace time constant ranges between 0·02 and 0·08 s.
To test whether isoprene emission responds fast enough to follow leaf temperature, rapid changes in leaf temperature were imposed on red oak leaves (Fig. 4). A single leaf was held in a gas exchange cuvette under the cold lamp until leaf temperature and isoprene emission rate were constant. The leaf was then rapidly heated and cooled by switching between the cold and hot lamps every 20 s. Because the temperature bursts lasted only slightly longer than the time constant for leaf temperature change, leaf temperature did not reach steady-state during any single heating event. Instead, the peak temperature reached during each temperature burst increased slightly over the peak of the previous burst so that, following the initial temperature increase when the hot lamp was first applied, the average leaf temperature continued to increase by ≈ 2 °C over the experiment period (Fig. 4). Isoprene emission rate did not follow leaf temperature precisely. Isoprene emission rate increased by only 1–2 nmol m–2 s–1 during each heating event after the first, and did not decrease to its original rate when leaf temperature dropped; integrating across the rapid leaf temperature fluctuations. Because of this, the total isoprene emitted during the experiment was only 15% lower than was predicted by the temperature response (Eqn 2) even though the peak emission rate was 6–10 nmol m–2 s–1 lower than was predicted from Eqn 2 (data not shown).
To more accurately study the kinetics of isoprene emission during a step-change in leaf energy balance, a leaf was held in a gas exchange cuvette while the light source was switched from the cold to the hot lamp. Leaf temperature (not shown) and isoprene emission rate climbed until they reached their saturation points (Fig. 5). After 8 min, the light source was switched back to the cold lamp and leaf temperature and isoprene emission fell to their original levels. The increase in isoprene emission was delayed relative to the increase in leaf temperature. This time delay also affected the isoprene emission decrease as leaf temperature decreased.
The time constant for the change in isoprene emission was determined from these data by normalizing the data using Eqn 1. Figure 6 shows the increase in isoprene emission as leaf temperature increases (top panel). The log-linear plot (bottom panel) of these data shows a response made up of two intersecting lines. The two lines with different slopes indicate that there were two separate processes with different time constants controlling the response of isoprene emission. The faster process had a time constant of 37 s while the slower process had a time constant of 116 s. Because the data collected during the first 25 s resulted from both the slow and the fast processes, the straight line fitted to the slow process was subtracted from the transformed data. These data are shown on the inset graph (Fig. 6). The slope of this line gave a time constant of 8·2 s.
A finite difference model was used to determine the time constant for isoprene emission separately from the time constant for leaf temperature change. The model was fitted empirically to temperature and isoprene emission data from Fig. 5. The time constant for leaf temperature change was fitted first, and then the time constant for isoprene emission was fitted. The results of one fit are shown in Fig. 7. Because the model only accounts for a single process controlling the increase in isoprene emission rate, the combined time constant was determined. This was close to the average of the two time constants and represented the time constant for the total change in isoprene emission. The total time constant for isoprene emission was 64·2 s.
To investigate the effect the time constant of isoprene emission has on the emission rate under natural conditions, isoprene emission rate was simulated from micrometerological data. The data used for the model input were taken from micrometerological measurements. Isoprene emission was simulated using the standard light correction (Eqn 4) and the finite difference model (Eqn 3). The simulated isoprene emission rate did not fluctuate as quickly as the fastest leaf temperature fluctuations, but did change rapidly during the highest magnitude heatflecks (Fig. 8).
Leaf temperature dynamics
It is important to know the relevant timescale for leaf temperature changes in the field in order to study their effects on isoprene emission. Our leaf temperature data reveal leaf temperature changes on several scales. The most frequent temperature fluctuations lasted between 5 and 10 s. These fluctuations happen much faster than the timescale over which isoprene emission changes (Figs 4 & 5). In addition, the amplitude of these fluctuations is 1–3 °C – small enough that isoprene emission will not be much affected. The larger magnitude leaf temperature fluctuations are usually associated with large changes in incident PPFD (Fig. 1). These result in temperature changes as high as 10 °C. They are less frequent, occurring on average once per hour. In addition, the leaf temperature change is relatively slow, changing with a time constant of nearly 1 min. These phenomena, called temperature ramps, have been seen in Populus tremuloides and mixed Populus and Acer canopies (Shaw et al. 1989; Roden & Pearcy 1993). These characteristics suggest that these fluctuations are caused by clouds passing overhead and obscuring all or part of the solar disc (Pearcy et al. 1996). It is these large amplitude fluctuations in leaf temperature associated with changes in incident PPFD that are likely to have the greatest effect on isoprene emission rate.
Time constants for isoprene emission
When the leaf energy balance is changed, the leaf temperature changes, causing changes in the rate of isoprene synthesis, and ultimately changing the rate of isoprene emission. Each of these processes responds with its own time constant. The measured time constant for the increase in isoprene emission (Fig. 6) is a composite of the time constant for leaf temperature change and the time constant for isoprene synthesis change. The finite difference model separated these time constants so each could be empirically fitted independently. In principle, there are two more time constants that can be separated from these measurements: the time constant of the gas exchange cuvette and the time constant of the leaf airspace. The cuvette time constant was measured to be 0·9 s at the flow rates used in the experiments. The leaf airspace time constant was calculated. It varied with stomatal conductance, but was less than 0·1 s. As both of these time constants are less than 3% of the time constant of isoprene emission, they are not detectable with gas exchange measurements and are safely ignored.
The averaged time constant for isoprene emission is slower than the most rapid changes in leaf temperature, but faster than the time constants of the temperature ramps seen in forest canopies (Figs 1 & 2). Because of this, isoprene emission closely follows the larger temperature fluctuations while damping the noise of the small, rapid temperature fluctuations (Fig. 8). However, the simulation in Fig. 8 underestimates the fluctuation in isoprene emission due to the short-term temperature fluctuations because it was not possible to take the fast and slow processes (Fig. 6) into account with the finite difference model. Because the rapid changes in leaf temperature are small, they likely fall within the range over which isoprene emission follows leaf temperature instantaneously. The larger changes in leaf temperature happen more slowly so that isoprene emission will follow these changes quite closely. Thus, the finite difference model represents a worst-case example of how well isoprene emission tracks leaf temperature.
We have previously determined that isoprene can protect leaves from damage at high temperatures (Singsaas et al. 1997). Thermotolerance was proportional to the concentration of isoprene inside the leaf airspaces. Because there is virtually no storage of isoprene gas inside leaves, the isoprene emission rate from a leaf equals the rate of isoprene synthesis [see Sharkey (1991)]. Increasing the isoprene synthesis rate increases the isoprene concentration inside the leaf airspaces. In addition, if the stomates close at high temperatures, the concentration of isoprene increases further. The changes in leaf airspace isoprene concentration calculated here are sufficiently large to significantly alter leaf thermotolerance (Singsaas et al. 1997).
The micrometerological data indicate that the highest leaf temperatures are reached frequently during the day, but last only for a few seconds to a few minutes. Isoprene-induced thermotolerance may be most important during the temperature ramps (Fig. 1), during which leaf temperature increased by more than 8 °C within 1 or 2 min. Isoprene emission changes with a fast enough time constant that it can increase as rapidly as temperature during ramping events. This supports our theory that isoprene protects leaves from the large temperature fluctuations in leaf canopies. Other known thermotolerance mechanisms, changes in xanthophyll cycle intermediates, induction of heat shock protein synthesis, and changes in thylakoid membrane lipid composition, change over a period of minutes, hours, or days (Pearcy 1978; Havaux et al. 1996; Tardy & Havaux 1997; Heckathorn et al. 1998). None of these is rapid enough to protect leaves from temperature ramps.
Because isoprene emission rate decreases rapidly when leaves are cooled, isoprene concentration in the leaf airspaces also decreases. Thus, isoprene concentration is high only when increased thermotolerance is needed, when leaf temperature is high. This may be beneficial if isoprene functions by dissolving into the thylakoid membranes and modifying their properties. Modifications to thylakoid membranes that increase thermotolerance often result in lower electron transport rates (Webb & Green 1991; Tsvetkova, Brain & Quinn 1994). In addition, less assimilated carbon will be lost as isoprene when its thermotolerance effect is not needed. Thus, the fast time constant of isoprene emission allows a rapid increase in thermotolerance at high temperature, and an equally rapid decrease in thermotolerance when it is not needed.
Control of isoprene synthesis
Isoprene is synthesized by a temperature-regulated process. There appear to be two processes controlling the increase in isoprene emission rate when leaf temperature increases. The first has a very fast time constant, 8·2 s. This process caused isoprene emission to change essentially instantly with leaf temperature. Processes this fast are likely to be physical, such as the influence of temperature on reaction kinetics. The second process has a time constant that was slower; 116 s. This process is slow enough to reflect the activation of an enzyme. A likely candidate for this is isoprene synthase. Because the finite difference model was unable to resolve the difference between the two, the fit of the model to the data is not perfect (Fig. 7). In the first 25 s, isoprene emission increases faster than is predicted by the model. After 25 s, there is a slower induction. The time constant determined from this fit was a composite of both processes.
In conclusion, the time constant for isoprene emission is fast enough to allow large changes in isoprene emission that follow temperature ramps. Ramps occur frequently throughout the day and result in changes in leaf temperature of 8 °C or more. The change in isoprene emission is due partially to a physical process – likely changes in reaction kinetics due to changes in temperature – and partially due to a regulation of the biochemistry of isoprene emission. Regulation of isoprene emission on this timescale indicates that isoprene emission may function to protect leaves from the high temperatures reached during temperature ramp events.
This research was supported with United States National Science Foundation grant IBN-9317900. Micrometerological measurements were funded by the United States Environmental Protection Agency through Cooperative Research Agreement CR-823791. Micrometeorological methods were developed in collaboration with Prof. John Norman.
Present address: Department of Plant Biology, University of Illinois at Urbana-Champaign, 134 Morrill Hall, 505 S. Goodwin Ave, Urbana, IL 61801-3707, USA