Coupled surface–atmosphere models are being used with increased frequency to make predictions of tropospheric chemistry on a ‘future’ earth characterized by a warmer climate and elevated atmospheric CO2 concentration. One of the key inputs to these models is the emission of isoprene from forest ecosystems. Most models in current use rely on a scheme by which global change is coupled to changes in terrestrial net primary productivity (NPP) which, in turn, is coupled to changes in the magnitude of isoprene emissions. In this study, we conducted measurements of isoprene emissions at three prominent global change experiments in the United States. Our results showed that growth in an atmosphere of elevated CO2 inhibited the emission of isoprene at levels that completely compensate for possible increases in emission due to increases in aboveground NPP. Exposure to a prolonged drought caused leaves to increase their isoprene emissions despite reductions in photosynthesis, and presumably NPP. Thus, the current generation of models intended to predict the response of isoprene emission to future global change probably contain large errors. A framework is offered as a foundation for constructing new isoprene emission models based on the responses of leaf biochemistry to future climate change and elevated atmospheric CO2 concentrations.
The emission of isoprene (2-methyl-1,3-butadiene) from terrestrial ecosystems provides one of the principal controls over oxidative photochemistry in the lower atmosphere, especially above continental regions (Fehsenfeld et al. 1992; Crutzen et al. 1999; Monson & Holland 2001; Monson 2002). Global isoprene emissions may be in the range 450–600 Tg C yr−1 or approximately 40% of the global biogenic non-methane volatile organic compound (NMVOC) emissions (Guenther et al. 1995; Potter et al. 2001; Levis et al. 2003; Naik et al. 2004; Tao & Jain 2005; Guenther et al. 2006). Increases in isoprene emissions have occurred over the past few decades in certain geographical areas, while at the same time legislative actions have reduced anthropogenic NMVOC emissions (Chameides et al. 1988; Constable et al. 1999; Purves et al. 2004; Tao & Jain 2005). This has caused the production of isoprene by urban and suburban forests to dominate controls over local and regional air quality. The oxidation of isoprene in the troposphere occurs in a matter of hours and results in the production of a variety of more stable, but pernicious compounds, including ozone (O3), organic nitrates and organic acids (Fehsenfeld et al. 1992; Fuentes et al. 2000). Globally, isoprene emissions from terrestrial ecosystems are thought to cause an increase in O3 concentration, a decrease in hydroxyl radical (OH) concentration and an increase in the tropospheric lifetime of methane (CH4) (Wang et al. 1998; Poisson et al. 2000; Roelofs & Lelieveld 2000). Thus, the biogenic emission of this one compound has far-reaching effects on air quality and global tropospheric chemistry.
Once in the atmosphere, isoprene and other reactive NMVOCs contribute to O3 production or destruction, depending on the concentration of nitrogen oxides (NOx). When NO is present at sufficiently high concentrations (more than 5–30 pptv), the oxidation of NMVOCs produces NO2 and, following photolysis of NO2, supports the net production of O3. At low NO concentrations, isoprene has the potential to react directly with O3 (Brasseur et al. 1999). The capacity for a mole of NOx to catalyse the formation of O3 in the presence of NMVOCs is called the ozone production efficiency (OPE). The emission of biogenic isoprene can increase the OPE (Lin et al. 1988). The presence of isoprene tends to reduce the concentration of tropospheric hydroxyl radical (OH) which slows the conversion of NO2 to HNO3 and enhances the photochemical efficiency of each mole of NOx in producing O3. The interactions between isoprene and NOx in producing O3 are complex, and highly dependent on the relative atmospheric concentrations of both. Whether NOx or hydrocarbons ultimately limit the rate of O3 production is strongly influenced by dynamics in the higher-order chemistry of NOx compounds and their dependence on the oxidative relationship between NMVOCs and OH.
Interest has recently been expressed in estimating the effect of future global change scenarios on atmospheric chemistry, particularly those involving dynamics in tropospheric O3 concentration (Brasseur et al. 1998; Johnson et al. 1999, 2001; Sanderson et al. 2003; Zeng & Pyle 2003; Fiore et al. 2005; Lathiere et al. 2005; Liao et al. 2006). These studies show that predictions of future changes in oxidative chemistry must be coupled to predictions of changes in the surface emissions of reactive biogenic hydrocarbons, including isoprene. Most global or regional models of present or future isoprene emissions are based on relationships among climate change, increases in atmospheric CO2 concentration and changes in net primary productivity (NPP; e.g. Constable et al. 1999; Tao & Jain 2005; Liao et al. 2006). The fundamental logic of such models is that changes in NPP will produce more or less biomass capable of emitting isoprene, and changes in climate (principally temperature) will stimulate or inhibit emissions per unit of biomass. These models tend to ignore the discovery that there are direct effects of changes in the atmospheric CO2 concentration on isoprene emission that tend to work in the opposite direction to that of stimulated NPP (Sanadze 1964; Monson & Fall 1989; Rosenstiel et al. 2003; Centritto et al. 2004; Rapparini et al. 2004; Scholefield et al. 2004; Possell et al. 2005; figure 1). Progress has been made in the past few years on the biochemical mechanisms underlying this direct response (Rosenstiel et al. 2003, 2004), and it is now possible to propose strategies for incorporating these effects into surface emission models. Furthermore, we now understand that most of the current generation of models use erroneous logic in assuming that the response of isoprene emission to temperature is based solely on the instantaneous effects of temperature on the enzyme isoprene synthase (e.g. the model originally proposed by Guenther et al. 1991, 1993). Several more recent studies have shown that the capacity for isoprene emission also responds to slower day-to-day changes in mean temperature, probably involving changes in the expression levels of isoprene synthase, not only its catalytic potential (Sharkey et al. 1999; Geron et al. 2000; Pétron et al. 2001; Funk et al. 2003). In the most recent global emissions model by Guenther et al. (2006), an effort has been made to modify the instantaneous algorithm to allow for some short-term acclimation of the key temperature-sensitive coefficients. While this recent effort represents a step towards providing some adjustments for recent weather conditions, it is accomplished by non-mechanistic ‘tuning’, rather than informed knowledge of underlying biochemical causes. Finally, we now understand that there are complex responses of isoprene emission to drought cycles, often involving lags in the responses between photosynthesis rate and isoprene emission rate, but typically involving an eventual decrease in isoprene emissions as the stress becomes severe (Tingey et al. 1981; Sharkey & Loreto 1993; Fang et al. 1996; Bruggemann & Schnitzler 2002; Funk et al. 2005). The response to drought is generally missing from most of the current generation of emission models. One exception is the effort by Guenther et al. (2006) to define an isoprene emission factor that is sensitive to changes in soil water content. The latter effort, while important for recognizing the effects of water stress on isoprene emission rate, forces the emission rate factor downward by an empirically determined amount in response to declining soil water content; a response that, as we show in this paper, may not be universal.
In this paper, we describe recent measurements that we have made at the sites of three prominent global change experiments in the United States, all involving entire ecosystems in close-to-natural settings. We use these measurements to make the case that the magnitude of the direct effects of increased atmospheric CO2 concentration, day-to-day changes in mean temperature and drought on the rate of forest isoprene emissions are probable to be as great as changes due to NPP and the response of isoprene synthase to instantaneous changes in temperature. Thus, the current generation of models that are founded on changes in these two factors alone are missing significant drivers of the response of isoprene emissions to future global changes. We also describe a conceptual model to explain the important biochemical controls on these direct effects in the hope that it will stimulate the development of new model schemes capable of quantifying isoprene emissions within the context of future global change.
2. Material and methods
(a) General approach
During the spring and summer of 2006, we conducted investigations at three prominent global change experimental sites in the USA: the warming and rainfall manipulation (WaRM) experiment in Texas; the Oak Ridge FACE experiment in Tennessee; and the Aspen FACE experiment in Wisconsin.
(i) The WaRM experiment
The WaRM experiment is located on a remnant post oak savannah site near Texas A&M University, College Station, Texas. This facility was constructed in 2003 to investigate the combined effects of altered precipitation distribution and warming on tree and grass dominant species of the southern oak savannah. The research infrastructure includes eight permanent 18×9×4.5 m (L×W×H) rainout shelters covered with clear polyethylene film. An overhead irrigation system in each shelter simulates a long-term ambient and redistributed (40% of summer rainfall redistributed to autumn and spring) precipitation regime. Two sets of five species combinations, post oak (Quercus stellata Wangenh.), a C4 grass little bluestem (Schizachyrium scoparium (Michx.) Nash) and an invasive tree eastern red cedar (Juniperus virginiana L.), are grown in monoculture and tree–grass mixtures in 2×2 m plots beneath each of the rainout shelters. One set of plots in each shelter is warmed with overhead infrared lamps (100 W m−2) that increase canopy and soil (depth of 3 cm) temperature by approximately 1.5 and 0.6°C, respectively. We concentrated our measurements on post oak leaves (the only isoprene emitter in the plots). No significant main effects of the various precipitation (F=1.83, p=0.19) and warming treatments (F=0.07, p=0.79) on oak isoprene emissions were found. Thus, we lumped oak trees from all treatments together and examined the potential for variation in isoprene emissions during pre- and post-watering periods, which provided an experiment on the effects of seasonal drought on isoprene emissions. We conducted the pre-watering measurements during 17–18 May 2006, which was 19–20 days since water had been added to the plots. We conducted the post-watering measurements on 19–20 May 2006, which was 1–2 days after 34 mm of water had been added to the plots (added on the evening of 18 May). In response to this watering event, soil water content increased from 11.4 to 16.9%, as measured with time domain reflectometry (TDR) integrated across the top 20 cm of the soil. The ambient temperature during the campaign in May 2006 ranged from 28 to 34°C. We repeated the measurements in August 2006, 3 days before and after a 19 mm rainfall event; in response to this latter watering event, soil water content increased from 7 to 13.2% in the upper 20 cm of the soil. The ambient temperature during the campaign in August 2006 ranged from 35 to 37°C. During both campaigns, days were generally cloud free. Both the May and August measurement campaigns occurred during the dry phase of the experimental redistribution treatment. This intensified an already severe summer drought by moving 40% of the May–September precipitation to the cooler months. Further details of the WaRM experiment can be found at http://rangeland.tamu.edu/research/nigec/index.html.
(ii) The Oak Ridge FACE experiment
The Oak Ridge free-air CO2 enhancement (FACE) site is located on the Oak Ridge National Environmental Research Park in Tennessee and consists of five 25 m diameter plots which were established in 1997 using the design of Hendrey et al. (1999). The trees in each ring are sweetgum (Liquidambar styraciflua L.) and were planted in 1988. Since the trees were originally planted, natural understory growth from the surrounding oak-hickory forest has also become established in the plots. The atmospheric CO2 concentration in each plot is maintained at ambient or elevated levels using 24 vent pipes and an automatic control system. The CO2 treatments were initiated in May 1998. The set point for the elevated CO2 plots was 565 ppmv during the day (there was no CO2 supplementation during the night). The average daytime CO2 concentration during the measurement campaign was 556 ppmv in the elevated CO2 rings and 393 ppmv in the ambient CO2 rings. We conducted measurements of isoprene emission at the Oak Ridge site during the period 10–23 June 2006. The maximum daily temperature during the campaign ranged from 28 to 32°C. Days were generally cloud free. The only measurable precipitation (0.5 cm) during the campaign fell on the evening of 19 June, halfway through the campaign. We noticed no obvious changes in the photosynthesis rate or isoprene emission rate of the leaves when comparing measurements before and after this small precipitation event. Further details of this site and the experimental treatments can be found at http://face.ornl.gov.
(iii) The Aspen FACE experiment
Full-season treatments at the Aspen FACE experiment were initiated in 1998 near Rhinelander, Wisconsin, and consist of twelve 30 m diameter treatment rings, including three ambient CO2 (control) rings, three rings with elevated CO2, three rings with elevated O3 and three rings with elevated CO2 and O3 together. The set point for the elevated CO2 plots was 560 ppmv during the day (there was no CO2 supplementation during the night). We concentrated our measurements on the ambient and elevated CO2 rings. Details about the control and elevated CO2 treatments are provided in Karnosky et al. (2003). The trees used in this experiment belong to five cloned lines of the species Populus tremuloides (Michx.) with known differences in O3 tolerance. In this paper, we will not focus on differences in isoprene emission rate among clones, but rather on general responses of this species to elevated CO2. We used all of the five clones in our sampling scheme and balanced the number of each clone used among the treatments so as not to skew the results from any single treatment due to favoured sampling of one clone over another. We conducted studies at the Aspen FACE site during 11–21 July 2006. The maximum daily temperature during the campaign ranged from 25 to 37°C. Days were generally cloud free. The only measurable precipitation (0.28 cm) during the campaign fell on the early morning of 14 July, 2 days into the campaign. We noticed no obvious changes in the photosynthesis rates or isoprene emission rates of the leaves following this small precipitation event. Further details of the Aspen FACE experiment can be found at http://www.aspenface.mtu.edu/.
(b) Isoprene emission measurements
Isoprene emission rate was measured on individual leaves using a portable gas-exchange system (LiCor, Inc., model 6400, Lincoln, Nebraska, USA) connected to a chemiluminescence continuous isoprene detector (Hills Scientific, model FIS, Boulder, Colorado, USA). In the studies in Texas and Wisconsin, we used leaves still attached to the trees. In the studies in Tennessee, we used leaves attached to branches (50–70 cm long) that had been cut from the trees using a pole pruner and immediately re-cut under water; the cut end of the branch was kept immersed in tap water during the gas-exchange measurements. We conducted initial experiments and determined that photosynthesis rates, stomatal conductance and isoprene emission rates were stable in leaves on the cut branches for at least 2 h after cutting. This was the approximate maximum time required to conduct the measurements of isoprene emission rate from the leaves. We also compared measurements of photosynthesis rate, isoprene emission rate and stomatal conductance on 10 leaves still attached to trees, to 10 leaves cut from the trees and found no significant differences. Branches used in our measurements were selected from the top 25% of the tree crowns in order to focus on leaves from sun-lit microenvironments. We restricted our measurements to the period that began 3 h after sunrise and ended 3 h before sunset in order to avoid the dynamic influences of early morning and late-day environments. Fortuitously, during all four field campaigns, we were able to make our measurements during windows of local weather that were stable, warm and without significant rain; this allowed for rather stable physiological responses in the trees and good comparability among trees from different experimental treatments.
The gas-exchange measurements were made on leaves maintained at 30°C for the May Texas campaign, and 32°C for the remainder of the campaigns, with the incident photosynthetic photon flux density (PPFD) at 1500 μmol m−2 s−1 and the chamber CO2 concentration maintained at 400 ppmv, unless otherwise noted. On some days, it was not possible to keep the leaf temperature within these set ranges due to hot weather. In those cases when the leaf temperature strayed more than 2°C from the set point, we recorded the actual leaf temperature and used a previously published model relating leaf temperature to isoprene emission rate (Guenther et al. 1993) to calculate the isoprene emission rate for the set-point temperature. In applying this model, we assumed Q10=2 for the temperature dependence of isoprene emission rate. In practice, this temperature correction was only a problem for several days during the August Texas campaign and 2 days during the July Rhinelander campaign. The ambient air source that was delivered to the leaf chamber was taken from the outlet of a clean-air generator (Aadco, Inc., model 737). The chemiluminescence isoprene detector was calibrated several times each day using a standard gas cylinder containing 6.8 ppmv isoprene with the balance being high-purity synthesized air (79% N2, 21% O2). Calibration curves were conducted at five isoprene concentrations from 0 to 400 ppbv.
(c) Sampling design and statistical analysis
To evaluate the impact of precipitation, warming and seasonal drought treatments on isoprene emission rates in the Texas WaRM experiment, we measured leaves on 17 post oak trees during the pre-watering period and 30 trees during the post-watering period in May 2006; the trees were distributed in the 2×2 m plot treatments in four polyethylene plastic shelters. In August 2006, we conducted measurements on leaves from 60 trees both pre- and post-watering events in six polyethylene plastic shelters, also distributed throughout the plot treatments. We treated the shelters as the fundamental unit of replication, n=4 in May and n=6 in August. Treatment and seasonal effects were assessed via a mixed model four-way factorial ANOVA, using Tukey–Kramer post hoc tests (α=0.05) to compare treatment means.
In order to assess the influence of growth at elevated CO2 on leaf gas-exchange rates in sweetgum trees in the Oak Ridge FACE experiment, we conducted measurements on leaves on 4–8 branches in two of the control rings (rings 3 and 4) and two of the elevated CO2 rings (rings 1 and 2) (branches were not accessible from the remaining experimental ring during our measurements). We averaged the measurements from each ring and treated the ring as the fundamental treatment unit (thus n=2 for the Oak Ridge analysis). In making observations on aspen trees in the Rhinelander FACE experiment, we used 4–6 trees in each ring and conducted measurements on three leaves from each tree. All leaves were averaged to provide a mean rate for isoprene emission, net photosynthesis or stomatal conductance (Is, A or gs, respectively) for each of the three rings for each treatment (thus n=3 for the Rhinelander analysis). In all analyses, means from each treatment were evaluated for differences using the Student's t-test with p=0.05 as the threshold of significance.
Oak leaves from the various temperature and precipitation treatments of the WaRM experiment in Texas exhibited a slight increase in isoprene emission rate (Is) when the atmospheric CO2 concentration (and thus the intercellular CO2 concentration, ci) was decreased instantaneously from ambient values (figure 2). Net CO2 assimilation rates (A) decreased as ci was decreased.
We observed significant main effects of date (May versus August) and time relative to a major watering event (pre- or post-watering) on Is from post oak leaves in the WaRM experiment (table 1). We did not observe a main effect of treatment due to seasonal redistribution of precipitation, but we did observe an interaction of this treatment effect with date. We did not observe a significant effect of the warming treatment on Is. We also observed significant interactions in the date×water (pre- versus post-watering event) and date×precipitation (control versus seasonally redistributed precipitation) treatments (table 2). Thus, overall, we observed that Is was higher in the plots with redistributed rainfall during May (plots with less summer rain had higher Is), and there was a differential influence of a major watering event depending on whether it occurred in August or May (the decrease in Is following a watering event was greater in August than in May). Rates of Is were relatively high for the oak leaves, especially during May, when compared with the sweetgum and aspen leaves we measured later in the summer (see figures 4a and 5a, respectively).
We observed a significant increase in A during May in response to the relaxation of drought by the watering treatment (p<0.05; figure 3). There was a slight upward trend in the mean ci following relaxation of the drought in May, but this did not prove significant (p>0.05). During August, we observed no significant change in A following relaxation of drought, despite seeing significant decreases in Is and increases in ci. The mean increase in ci following relaxation of the drought was 13 μmol mol−1.
We conducted measurements of Is, A and gs at both normal (400 ppmv) and elevated (600 ppmv) atmospheric CO2 concentrations in both treatments for trees at the Oak Ridge FACE experiment. When assessed at normal ambient CO2, leaves from trees in the elevated CO2 rings had lower Is when compared with leaves in the control rings (p<0.05; figure 4). Rates for A and gs were not significantly different in leaves from the two treatments. When assessed at elevated atmospheric CO2, leaves from trees in the elevated CO2 rings also exhibited lower mean Is when compared with leaves from the control rings (23.2 and 32.1 nmol m−2 s−1, respectively), but the difference in these means was not quite statistically significant (p=0.069; data not shown). As in the measurements at normal ambient CO2, rates for A and gs were also not significantly different in leaves from the two treatments when measured at elevated CO2.
In measuring Is for leaves of P. tremuloides at the Aspen FACE site, we only made measurements at the approximate atmospheric CO2 concentration of the growth rings (e.g. 400 ppmv for leaves from the control rings and 550 ppmv for leaves from the elevated CO2 rings). Leaves from the elevated CO2 rings exhibited lower Is when measured at 550 ppmv when compared with leaves from the control leaves measured at 400 ppmv (p=0.05; figure 5). Mean rates of A tended on average to be higher for leaves grown and measured at elevated CO2 concentrations, but the replicate measurements were highly variable, and the means proved not to be statistically different when formally tested (p=0.11). Mean values of gs were similar in magnitude and not statistically different in leaves from the control and elevated CO2 treatments (p>0.05).
We were not able to make observations across a long enough time span to detect changes in Is in response to changes in temperature due to weather fronts that moved through the study location, except for a short period of hot weather that developed in the middle of our observations at the Aspen FACE site. Here, the high temperature extremes only lasted for 2 days, and thus we were not able to validate our observations across several replicate hot and cold weather periods. However, we did observe that for trees in the control CO2 rings, the mean Is at a constant measurement leaf temperature of 32°C was higher (32.4±1.0 nmol m−2 s−1, mean±s.e., n=38 leaves) during the 2 days of the hot weather (mean daily maximum temperature=35.3±1.25°C, mean±s.e.) when compared with the mean Is (25.8±0.7 nmol m−2 s−1, mean±s.e., n=53 leaves) during the 3 days of cooler weather (mean daily maximum temperature=28.7±0.2°C, mean±s.e.) (means different at p<0.05). We did not have the same opportunity to observe a response to transient hot weather in trees in the elevated CO2 rings.
The isoprene emission rate (Is) from terrestrial forest ecosystems in future climatic and atmospheric regimes will be determined by complex interactions among several driving variables (Monson et al. 1995; figure 1). Most researchers who have made predictions of Is in future global change scenarios relied upon indirect coupling between NPP and Is, as altered by changes in the atmospheric CO2 concentration or changes in precipitation, to guide their models (e.g. Constable et al. 1999; Naik et al. 2004; Tao & Jain 2005; Liao et al. 2006). Additionally, when climate warming is considered as a driver of future Is, it is typically considered within the context of an Arrhenius-type model, which describes the instantaneous coupling of higher temperature to increased isoprene synthase catalysis rate (sensu Guenther et al. 1991, 1993, 1995). In this study, we focused on direct interactions between increased atmospheric CO2 concentration and drought on Is, and the influence of warmer temperatures produced by day-to-day changes in weather, rather than instantaneous second-to-second, or even hour-to-hour, increases in temperature, as are normally studied. Our intent was to fill in some of the observational gaps needed to produce a more complete modelling context for the response of Is to future global change.
Our studies of oak trees growing in the WaRM experiment in Texas produced two novel results. First, Is was higher during the spring than during the mid-summer. It is more typical for Is to increase with temperature as the growing season progresses and the weather becomes warmer (e.g. Monson et al. 1994), the opposite pattern to what we observed. In fact, there is evidence that in past studies the seasonal increase in Is is a direct response to the seasonal increase in temperature (Mayrhofer et al. 2005; Wiberley et al. 2005). It is not clear at the present time what the biochemical interactions are that control the response that we observed, but it is significant that we could not predict it based on our current understanding of seasonality in Is. We do, however, point out that the local weather was relatively hot during both the May and August campaigns (32.4±0.7 and 36.1±0.8°C, respectively) and it might be that the progressive drought over the summer caused a decline in Is, masking any increase in Is that might have been due to a small seasonal increase in the mean temperature. Second, we observed that drought during the middle of the summer caused a doubling in Is, when compared with well-watered periods. Past studies have revealed that Is is less sensitive to water stress than A, and that Is tends to remain stable, or is modestly enhanced during periods of acute water stress (e.g. Fang et al. 1996; Bruggemann & Schnitzler 2002; Pegoraro et al. 2005, 2006). In one comprehensive analysis, drought was shown to have resulted in a significant decrease in Is in a red oak forest in the northeastern United States (Funk et al. 2005). It is possible that the result we observed is similar to that of Funk et al. (2005), if indeed the decline in Is in post oak from May to August was caused by chronic drought. However, we observed that relief from the drought in August caused a further decrease in Is, not an increase; this was an unexpected result. The effect of reduced Is following relief from drought could not be explained by the small increase in ci that we observed, as has been hypothesized in past studies of the effect of water stress on Is (Pegoraro et al. 2004). The increase in ci that we observed would cause only small changes in Is through the direct CO2 response that has been described in past studies (figure 2). Rather, we hypothesize that there is an active upregulation of isoprene biosynthesis during periods of drought. Downregulation of Is during the middle of the growing season, when days are hotter, and upregulation during drought are processes not currently included in models of the response of Is to future global change. The magnitude of the changes observed in our study however—a twofold reduction through the growing season and another twofold reduction following significant precipitation—is high enough to have significant ramifications for model predictions of the coupling between Is and future climate changes. The general lack of correlation (in sign or magnitude) of the effects of drought on Is versus the effects on photosynthesis rate (also see Funk et al. 2005) further amplifies the conclusion that sensitivity of NPP to climate may not be the best basis for describing sensitivity of Is to climate.
Our observations in both a sweetgum forest in Tennessee and aspen stands in Wisconsin, revealed evidence of an active downregulation of Is during growth in an atmosphere of increased CO2 concentration. These results are among the first to show a consistent downregulation of Is in response to growth at elevated CO2, and they emphasize that while many past studies show an instantaneous inhibition of elevated CO2 on Is when measured in a leaf cuvette, a response in the similar direction is evident on whole forest stands exposed to elevated CO2 under natural field growth conditions. The response to growth CO2, however, is probably based on a mechanism that is different than that for instantaneous changes in the atmospheric CO2 concentration; this is supported by the fact that we observed downregulation when leaves were measured at the same instantaneous CO2 concentration in sweetgum trees. In a past study, Rosenstiel et al. (2003) showed that instantaneous exposure of poplar leaves to elevated CO2 causes an inhibition of Is, probably resulting from increased activities of the enzyme phosphoenolpyruvate carboxylase (PEPc), which shifts patterns of cytosolic and chloroplastic substrate use and limits the availability of pyruvate substrate for chloroplastic isoprene biosynthesis. It is possible that growth at elevated CO2 causes an upregulation in expression of the PEPc gene and concomitant reduction in Is, as has been observed when poplar trees are grown with as their only nitrogen source (Rosenstiel et al. 2004). Past studies have shown that mitochondrial densities increase when trees of several species (including sweetgum) are grown at elevated CO2 (Griffin et al. 2001), and it is possible that higher mitochondrial densities are accompanied by increased expression of the PEPc gene; the PEPc enzyme is known to provide substrate to support mitochondrial respiration.
Increases in the mean global temperature, while moving in a direction that could cause increased isoprene emission, may not be as important as increases in the frequency of extremely hot days. Past studies have shown that Is can be regulated up or down depending on recent day-to-day weather patterns (Sharkey et al. 1999; Geron et al. 2000; Pétron et al. 2001), and even within a single day (Geron et al. 2000; Mayrhofer et al. 2005). Although it is somewhat anecdotal, we observed a 26% increase in Is during the 2 days of extremely hot weather that occurred during the Aspen FACE campaign. This increase is presumably due to an upregulation of the genes underlying the isoprene biosynthetic pathway.
Figure 6 provides a cellular and biochemical context for considering the effects of elevated CO2, drought and periods of hot weather on Is. High growth [CO2] is proposed to cause an upregulation in the expression levels of PEPc and mitochondrial density, both of which would increase the channelling of PEP to the production of oxaloacetate (OAA) and decrease the channelling of pyruvate to chloroplastic DMAPP production, and thus Is. We leave open the possibility that high growth [CO2] also causes the downregulation of expression of isoprene synthase. High instantaneous [CO2] is proposed to cause an increase in the photosynthetic production of glyceraldehyde 3-phosphate (G3P), but a concomitant increase in the activity of PEPc, which would also inhibit isoprene emission by decreasing the channelling of PEP to the chloroplastic and production of chloroplastic DMAPP. Drought and periods of hot weather are proposed to increase the expression of isoprene synthase, although this has not been clearly established. Also shown are the mathematical models from Guenther et al. (1991, 1993) which are commonly used to predict the responses of isoprene emission rate to instantaneous changes in temperature and light intensity.
Researchers who model the response of biogenic VOC emissions and their associated oxidative photochemistry to changes in climate and atmospheric CO2 concentration have focused almost exclusively on influences coupled through NPP (Constable et al. 1999; Naik et al. 2004; Tao & Jain 2005; Liao et al. 2006). It is true that numerous modelling efforts have predicted changes in the amount and distribution of NPP during future global change (e.g. Cramer et al. 2001). However, models that reflect these effects as the only driving forces of future trends in tropospheric chemistry are missing a large part of the relevant dynamics. Even within the restricted context of the global change experiments we investigated, erroneous predictions are probable to emerge with sole reliance on the indirect effects of NPP. At the Oak Ridge FACE site, increases in forest NPP over the 8 years of the experiment to date have occurred through increases in fine root production, not the aboveground, isoprene-emitting shoot production (Norby et al. 2004). Thus, the only effect of increased atmospheric CO2 on isoprene emission in this forest is probable to be the direct effects, transmitted through the influences of elevated CO2 on leaf metabolism per unit leaf area and not through the indirect effects of increases in leaf area. The trees grown in elevated [CO2] at the Aspen FACE site have exhibited an approximate 15% increase in leaf area index, when compared with trees in the control rings (Karnosky et al. 2003). The inhibitory effect we observed due to the direct influence of elevated [CO2] on isoprene emission is larger in magnitude than the potential stimulation in isoprene emission predicted by the increase in leaf area index. The direct influences we observed in our experiments are large—often in the range of 2× effects—and generally greater than the changes in aboveground NPP or increase in leaf area index. Without inclusion of these effects in the current array of models being used to predict changes in atmospheric chemistry due to global change, one has to question the relevance of the predictions themselves. Other challenges lay in the potential for individualistic responses among different tree species and forest types. The potential for species with different growth and biomass allocation strategies, and different tolerances of environmental stress, to respond to global change creates immense challenges in their own rights. To a large extent, the modelling has ‘raced ahead’ of our mechanistic understanding of how isoprene emissions will respond to the fundamental drivers of global change. As a result, our understanding of the fundamental mechanisms controlling the direct responses of Is to global change needs be addressed to allow for the development of strategies and the inclusion of this knowledge in predictive models.
This work was supported by grants from the US Environmental Protection Agency (RD-83145301) and the National Science Foundation (ATM-0516610). We wish to thank Pete Casey for valuable assistance in the field measurements. The Oak Ridge FACE experiment and the Aspen FACE experiment are supported by the US Department of Energy, Office of Science, Biological and Environmental Research. The WaRM experiment is supported by the US Department of Energy, National Institute for Climatic Change Research (NICCR), Southeastern Region.
↵† Present address: Department of Biology, Portland State University, Portland, OR 97201, USA.
One contribution of 18 to a Discussion Meeting Issue ‘Trace gas biogeochemistry and global change’.
- © 2007 The Royal Society