Chapter 3

INTRODUCTION

Forest ecosystems produce copious quantities of isoprene, with tropical rain forest emitting the most isoprene per square kilometre (Rasmussen and Khalil, 1988). It is evident from the complex interactions between isoprene emission, atmosphere and climate, as summarised in Figure 1.1, p.17, that prediction of climate change effects on vegetation and the associated feedback effects on climate, will also require the prediction of the response of isoprene emission rates from forest ecosystems to atmospheric and climate change (Monson et al., 1991a). Such predictions of forest isoprene emission can now be accomplished by scaling-up the mechanistically-rich leaf emission model equations of the last chapter (Table 2.1, p.42), to the canopy level. Projections of emission rates under future conditions had previously been prevented by the lack of a process-based model for leaf emission rates (Monson et al., 1991a).

Combined macroclimate and microclimate models, based on the angle of the sun for any given latitude and time of day, take account of the differences in the radiation environment between sun and shaded leaves within a canopy, and so can be used to predict both light levels and temperature, the most important environmental factors affecting leaf isoprene emission rates (Lamb et al., 1993; Guenther et al., 1993; 1994; Baldocchi et al., 1995) (Table B, Appendix I, p.197). Simple canopy models, based on the distribution of sunlit and shaded leaf area within a canopy, are found to give improved predictions for photosynthesis over models that treat photon flux as showing an exponential decline through an homogeneously lit canopy. Also, estimates from more complex models of multi-layered components have been found to differ little from the sunlit-shaded category approach (Norman, 1980). A sunlit/shaded canopy model is, therefore, also appropriate for a model of canopy isoprene emissions, as the greater photon flux densities and higher temperatures of sunlit leaves, cause them to emit isoprene at higher rates than shade leaves, particularly on a leaf area basis (Sharkey et al., 1991b; Sharkey, 1996; Harley et al., 1996). The difference in emission rates between sunlit and shaded leaves can be logically linked to the leaf enzyme and nitrogen content, which declines with depth into the canopy (Baldocchi et al., 1995; Litvak et al., 1996).

Isoprene emission is highly temperature dependent, therefore the effects of temperature within a canopy will be critical to accurate predictions of proportionate change in Fiso. Leaf temperature, however, is not determined by ambient temperature alone, but is also influenced by physiological factors. Of particular significance, under conditions of rising [CO2] and global warming, will be the effects of stomatal closure on leaf temperature. Although stomatal conductance does not directly control isoprene emission rates, and humidity between 30 and 100% has little effect on Fiso, stomatal closure may have an indirect effect via reduced rates of transpiration and greater leaf temperatures (Tingey et al., 1981; Guenther et al., 1994; Sharkey, 1996). For example, under elevated [CO2] the high concentration of the substrate for photosynthesis will allow stomata to close against the loss of water by transpiration, whilst maintaining a constant Ca to Ci ratio. The reduced stomatal conductance will, in turn, cause higher leaf temperatures, due to reduced cooling by evapo-transpiration. Therefore, the canopy model used to predict canopy isoprene emission rates will require the incorporation of leaf energy balance equations to simulate the effects of elevated [CO2] and leaf temperatures on rates of isoprene emission. The inclusion of a leaf energy balance model has been found to give results that more closely fit peak, midday flux measurements (Baldocchi et al., 1995).

The overall aim of this chapter is to predict the proportionate response of isoprene emission rates from forest canopies of different climate zones to global warming. This is to be achieved by scaling-up the leaf isoprene emission model presented in Chapter Two (Table 2.1, p.42) to the canopy level, using the effects of direct and diffuse radiation on sunlit and shaded leaves within the canopy, together with a leaf energy balance model, and parameterisation of the model for environmental conditions relevant to global warming, using well validated macroclimate equations (Table B, Appendix I, p.197).

The objectives of this chapter are threefold. First, to present the sun/shade and energy balance canopy isoprene emission model developed from the mechanistically-rich leaf isoprene emission model presented in the previous chapter. Second, to compare the results of the simulation of the amount of isoprene emitted in one year under present day climate conditions, with published estimates of emission for different types of forest. The lack of published estimated emission rates restricts the data available for model validation to the following forest types: a temperate deciduous forest in the UK; a temperate deciduous forest in Pennsylvania; and an Amazonian tropical rain forest (Rasmussen and Khalil, 1988; Anastasi et al., 1991; Guenther et al., 1994). Third, to use the model to determine the proportionate response of forest isoprene emission rates to the effects of climate change. The putative global warming of 3oC averaged over the next century (Watson et al., 1990) was based on a projected doubling of atmospheric carbon dioxide concentrations to approximately 700 µmol mol-1. However, CO2 levels are rising at slower rates than previously anticipated, partly due to economic recession in many parts of the western world. Therefore, a more conservative 2oC increase in temperature has been used in the present research in conjunction with [CO2] elevated to 650 µmol mol-1, consistent with the predictions of various models using the IPCC business as usual scenario, IS92a, in the IPCC report of 1992 (Leggett et al., 1992; Schimel et al., 1994). To predict future trends of emission rates in a warming world, the model simulated isoprene emission rates for each of the three forest types three more times, first increasing the temperature only by 2oC, then increasing [CO2] to 650 µmol mol-1, but keeping the temperature at present day levels, and finally increasing both temperature and [CO2] concurrently. The predictions of concurrently occurring conditions are compared with those from increased temperature and [CO2] acting individually.

3.2.1 Model equations

The process-based leaf isoprene emission model of Chapter Two, based on the determination of three limiting rates of isoprene synthesis, was scaled up to the canopy level, by incorporation of the leaf model equations into the combined modules of macroclimate and microclimate equations, sun/shade canopy model equations and leaf energy balance equations, existing within the model of Humphries and Long (1995), (Tables A and B, Appendix I, pp.196 and 197). The canopy model equations are listed in Table B (Appendix I, p.197) as they are common to more than one chapter. The total amount of isoprene emitted by a forest canopy during one year (F'isotot) is calculated as the accumulated sum of isoprene emitted from both the sunlit and shaded components of a canopy, over each day of the year (Equations 3.1 and 3.2, Table 3.1, and Table B, p.197). In the model the leaf area index in direct sunlight is calculated throughout the day as a function of canopy leaf area index, the ratio of horizontally and vertically projected areas of the canopy, quantities of direct and diffuse light, and the angular distribution of radiation within the canopy, according to Campbell (1986) and Forseth and Norman (1991) (Equations B.3 to B.7). Direct and diffuse photon flux are calculated as a function of solar angle, latitude, time of year and atmospheric transmittance, after Campbell (1977), Norman (1980) and Forseth and Norman (1991) (Equations B.8 to B.14). The canopy is treated as two populations of leaves, sunlit and shaded, with mean values of irradiation and Fiso calculated for each population. The division of a canopy into two layers was shown by Norman (1980) to give adequate prediction of canopy photosynthesis, comparable to that of multiple layer canopy models, and as isoprene emission is similarly dependent on radiation, this system has also been adopted here. Equations B.15 to B.29 are used to determine evapo-transpiration and temperature using evapo-transpiration rates within a leaf energy balance model, as described by Campbell (1977). Air temperature variation through the year was calculated according to the sinusoidal pattern described by Equations B.19 to B.22.

Table 3.1: Cumulative canopy isoprene model equations. (General canopy equations are listed in Appendix I, Table B.)

The iterative method for finding the assimilation rates and stomatal conductance using the leaf temperature from the leaf energy balance equations within WIMOVAC has been described elsewhere (Humphries et al., 1996).

Isoprene emission is seasonal (Monson et al., 1994). In the UK, for example, isoprene is mainly emitted from deciduous forests during the warmer months of April to September, with emissions declining to zero during the winter months, when there are no leaves (Anastasi et al., 1991). Therefore, to prevent the overestimation of F'isoc for seasonal forest canopies, a vegetative cover coefficient, v, is included in the calculation of canopy isoprene emission (Equation 3.3). Calculation of v is based on a sine wave, with the cycle start day for vegetative cover equal to the annual temperature cycle start day, used in the macroclimate model to determine ambient air temperature (Equation B.19 to B.22, p.197). The value of v varies between 0 and 1, in phase with the temperature cycle and is set to 1 when simulating emissions from non-seasonal forests, for example, those around Belém, in Brazil.

The mean isoprene emission rates of the canopy layer components were calculated and summed to give the total canopy isoprene emission rate over the period of a day (Equation 3.2) and daily canopy emission rates were accumulated to predict long-term isoprene emission quantities over the period of one year, in terms of kg km-2 area (Equation 3.1) (Table 3.1).

All model equations and values of model parameters are presented in Appendix I and II, respectively (pp.196-208). The model can be parameterised for any global location by setting the appropriate latitude and parameterising a relevant temperature regime. This approach was used to simulate the isoprene flux from forest canopies of three locations, with the additional parameterisation of leaf area index (Table 3.2).

Table 3.2: Values used for latitude, longitude, temperature variation, leaf area index and start day of seasonal vegetation cover (DJsv) in the canopy model simulations, under present day conditions, for the three forest locations.

3.2.2 Model simulations for validation

For the purpose of model validation the simulated values of accumulated isoprene emission for one year (F'isotot) need to be compared with estimated values published elsewhere. The lack of experimental data and use of incompatible units restricted the availability of comparable published data for model validation. Simulation of the amounts of isoprene emitted from given locations require the appropriate macroclimate photon flux density and temperature regimes for that site. Using the equations of Campbell (1977), Norman (1980) and Forseth and Norman (1991) the mean course of radiation can be predicted for a given latitude (Equations B.5 to B.11, p.197). However, the temperature regimes of the locations required research. The values of mean monthly temperatures for the three locations were found from published literature. The mean monthly temperatures for the UK were based on the monthly temperature data for Oxford (52oN, 1oW) (Willett, 1992), which were within 5% of the mean monthly temperatures for the UK for the years 1951 to 1987 (Anastasi et al., 1991). The USA mean monthly temperatures were based on New York (41oN, 74oW) (Bartholomew, 1994), and the temperature data for the Amazonian rain forest was based on the mean monthly temperatures of Belém, Brazil (1oC, 48oW) (HMSO, 1963). The macroclimate module of WIMOVAC was used to determine the input values of annual mean air temperature, amplitude of annual temperature change and temperature cycle start date that gave a close fit to the published data, Figures 3.1 to 3.3. The input values determined are given in Table 3.2, along with the leaf area index values and latitude and longitude used for each of the forest canopies. The apparent seasonal variation in temperature during the year for the UK and USA, Figure 3.1 and 3.2 is noticeably lacking in the annual temperature regime of Brazil, Figure 3.3.

For the deciduous forest of the UK the leaf area index was assumed to be 4.6 (Rauner, 1976) and the latitude and longitude were set to 54oN and 2oW, respectively, to represent a deciduous woodland in the middle region of the UK. The parameterisation of the model for deciduous woodlands of the USA were a leaf area index value of 4.6 (Rauner, 1976), a latitude of 41oN and longitude of 78oW, to represent a deciduous forest in Pennsylvania.

A latitude of 1oS, a longitude of 48oW and a leaf area index of 8 (Rasmussen and Khalil, 1988) were used to simulate a typical annual quantity of isoprene emitted by the Amazonian tropical rain forest in the region, for example, as measured by Rasmussen and Khalil (1988) during the 1985 ABLE expedition. As the leaf model does not predict absolute values of species specific rates of isoprene emission, the species composition of different forest canopies are not included as variables within canopy simulations.

3.2.3 Global warming prediction model simulations

To predict the isoprene emitted under conditions of putative global warming, the total accumulated isoprene emitted was predicted for the three forest ecosystems, using the nearly all the same parameter values used for model testing simulations, apart from the values of annual mean air temperature and [CO2]. First, the canopy isoprene emission rates through the year were predicted for each forest ecosystem, with the temperature variable values as given in Table 3.2. F'isoc was simulated for a year for an area of UK deciduous forest, both with varying v and v set to 1, for comparison (Figure 3.4). Next, the annual variation in F'isoc for each forest canopy was predicted, with the annual mean temperature increased by 2oC.

Finally, the amount of isoprene emitted for a whole year was predicted (F'isotot), firstly, with "normal" annual mean air temperatures and CO2 concentrations as given in Table 3.2; secondly, with annual mean air temperatures increased by 2oC; thirdly with normal temperatures, but [CO2] increased to 650 µmol mol-1; and lastly, with concurrent increased temperatures and [CO2], consistent with the IPCC business as usual scenario, IS92a (Leggett et al., 1992; Schimel et al., 1994).

3.3.1 Model validation

3.3.1.1. Deciduous forest of the UK

The long-term canopy emission of isoprene from a square kilometre area of deciduous forest in the UK was simulated to be 3.7 x 103 kg km-2 per year (Table 3.3), which lies between the published estimates of 4.3 x 103 kg km-2 yr-1 and 2.7 x 103 kg km-2 yr-1 of Rasmussen and Khalil (1988) and Anastasi et al. (1991), respectively. Anastasi et al. (1991) used regional estimations of vegetation types based on Forestry Commission figures to estimate the isoprene emission from deciduous forests of the UK.

3.3.1.2. Deciduous forest of the USA

The model predicted the total annual emission of isoprene for a deciduous forest of the northern USA to be 7.4 x 103 kg km-2 area (Table 3.3). Although the model estimate is higher than the 4.3 x 103 kg km-2 estimated for a temperate deciduous forest by Rasmussen and Khalil (1988) and the values of isoprene emission estimated in the late 1970's, as reviewed in Altshuller (1983), which ranged from approximately 1 to 4 x 103 kg km-2 per year, it is below the lowest value in the range estimated by Guenther et al. (1994) for the area of Pennsylvania. Guenther et al. (1994) estimated the amount of isoprene emitted by deciduous USA woodlands for one year using the inventory method. Foliar emission rate measurements of volatile organic compounds (VOC) of 49 tree genera were used, in conjunction with satellite data of regional ecosystems and foliar mass and species composition data. For the area of Pennsylvania the average rate of isoprene emission was estimated as 5.5 mg m-2 h-1. Using the method of Rasmussen and Khalil (1988) to convert hours to a year total, by multiplying by 14 hours and 200 days, an estimate of 1.54 x 104 kg km-2 is reached for US deciduous woodlands, which is approximately twice the estimated value produced by our canopy model. However, the range of estimates for this area is 8.4 x 103 to 2.24 x 104 kg km-2, and the model prediction is lower than the lower limit. The estimate of Guenther et al. (1994) is also an order of magnitude above values of isoprene emission estimated in the late 1970's (reviewed in Altshuller, 1983). Therefore, overall, the model prediction falls within the range of estimates published elsewhere, but, it must be borne in mind, this range is large.

3.3.1.3. Tropical rain forest of the Amazon

The model predicted the total amount of isoprene emitted during one year as 1.16 x 104 kg km-2, which is in close agreement with the estimate of Rasmussen and Khalil (1988) of 1.2 x 104 kg km-2 per year (Table 3.3). The lack of data of the magnitude of isoprene emitted from this area make the closeness of the simulation all the more relevant. The amount predicted to be emitted by the rain forest canopy of Belém is greater than three times the amount of isoprene predicted to be emitted by a deciduous forest of the UK, and just over one and a half times that predicted to be emitted from a deciduous forest in the area of Pennsylvania, in the USA. Considering the seasonal variability of the deciduous forests, and relative lack of season variation in the climate of Belém, and also taking account of the latitudes of the three different climates, the model predictions are consistent with the relative magnitudes of emission that might be expected for the three forest types. Also, the predictions for the UK and USA fall between the extremes of estimates published elsewhere, and the prediction for Belém is close to the published estimate. Therefore, overall, the model validation can be considered successful, bearing in mind the lack of data for validation.

3.3.2 Simulation of isoprene emission over the course of one year

Using the parameters listed in Table 3.2, the canopy model was used to predict the isoprene emission rates from each of the forests from the three locations over the time course of one year, both at present day temperatures and at temperatures increased by 2oC (Figures 3.4, 3.5, and 3.6, for the UK deciduous forest, the deciduous forest of the USA and the Amazonian rain forest, respectively).

Table 3.3: Model validation results comparing predicted isoprene emitted per year (kg km-2) with published estimates.

The seasonal pattern of canopy isoprene emission rates during the course of a year follow the same pattern as the mean air temperature during the year for the deciduous forests of the UK and USA (compare Figures 3.4, p.85 and 3.5, p.89 with Figures 3.1, p.81 and 3.2, p.82). However, F'isoc from the Amazonian rain forest canopy shows very small seasonal variation, with a double cycle of shallow amplitude during the year, which, whilst not apparent in the model simulation of the annual mean air temperature variation, may be discerned from the published temperature data (HMSO, 1963) (compare Figure 3.6, p.90 with Figure 3.3, p.83). However, considering the monthly mean air temperatures vary by less than 1oC during year, the small overestimation of mean annual air temperature for the month of February is considered negligible.

3.3.3 Model Simulations of Future Emission Rates

3.3.3.1. Deciduous forests of the UK

The amount of isoprene emitted by a square kilometre of deciduous forest in the UK in one year, under present day conditions was predicted to be 3.7 x 103 kg (Table 3.4). The predicted F'isotot increased to 4.7 x 103 with a 2oC increase in temperature. But with elevated [CO2] at present day temperatures, the amount of isoprene emitted was predicted to be only 2.1 x 103, rising to 2.7 x 103 kg under concurrent elevated [CO2] and temperature increase (Figure 3.7). Thus, although higher temperatures were predicted to produce a 27.0% increase in the isoprene produced by the temperate deciduous forest, an increase in [CO2] to 650 µmol mol-1, occurring concurrently with higher temperatures, is predicted to reduce isoprene production by 27.0%.

Table 3.4: The predicted effects of increasing temperature and [CO2] on isoprene emitted from a km2 area of forest for one year (in kg).

3.3.3.2. Deciduous forests of the USA

The amount of isoprene predicted to be emitted during one year by a deciduous forest in the USA under present day conditions increases by 23.0%, from 7.4 x 103 to 9.1 x 103 kg km-2 with a 2oC increase in temperature. However, with elevated [CO2] and no concurrent temperature increase, the predicted emission declines to 4.2 x 103 (Table 3.4). Concurrent temperature rise and elevated [CO2] is predicted to cause a 29.7% decrease in the amount of isoprene emitted during one year, to 5.2 x 103 (Figure 3.8).

3.3.3.3. Tropical rain forest of the Amazon

The relatively high amount of isoprene emitted by an area of rain forest around the Belém region of Brazil, which shows a relatively non-seasonal variation in climate and leaf cover, is predicted to increase to 1.27 x 104 kg km-2 per year with a 2oC increase in temperature, a 9.5% increase (Table 3.4). The model predicts a dramatic decrease in isoprene emission under elevated [CO2], to 6.6 x 103, which is less than the amount of isoprene simulated to be emitted from the same area of a deciduous forest from Pennsylvania, under present day conditions. Meanwhile, increasing [CO2] occurring concurrently with higher temperatures is predicted to cause an overall 37.1% decrease in isoprene emitted from present day conditions, to 7.3 x 103, similar to the amount simulated to be presently emitted by a deciduous forest of the USA, Table 3.4 (Figure 3.9).

DISCUSSION

The aims of this chapter, to predict the proportionate change in isoprene emission from forest ecosystems with climate change, have been met by scaling up the leaf emission model to the canopy level using a simple sun/shade model, with leaf energy balance equations to include the additional influence of leaf temperature afforded by stomatal closure caused by higher [CO2]. The resulting model provides a method to predict the proportionate change in emission rates from forests, where no previous technique existed, due to the lack of a model that could mimic the influence of changing [CO2] on isoprene emission rates. The new canopy model, with the added leaf energy balance equations, also allows the feedback effects between ambient temperature, leaf temperature, [CO2] and stomatal conductance on canopy isoprene emission rates to be simulated. Leaf temperatures within a canopy are generally 2oC higher than ambient air temperatures and can be as much as 10 to 14oC higher under full sunlight (Lamb et al., 1993; Sharkey, 1996). However, the appropriate use of heat energy balance equations within canopy models to determine leaf temperatures, is still a matter for discussion (Baldocchi et al., 1995). For the purposes of this model, considering the high temperature dependency of isoprene emission, together with the better prediction of peak daytime values afforded by the energy equations, the inclusion of a leaf energy balance model was thought to be appropriate (Baldocchi et al., 1995). Thus the first objective, to construct the model was achieved.

The second objective, to validate the model, was achieved by testing the model simulations for three forest ecosystem types against published estimates for the three forest ecosystems. The lack of estimates of isoprene emission rates from other forest ecosystems of different climate zones is partly due to the inconsistency in measurement techniques, experimental protocol and measured units (area or weight based) between published data (Lamb et al., 1987; Martin et al., 1991; Baldocchi et al., 1995). Although the difference between estimates of ecosystem emission rates of isoprene can be as much as an order of magnitude, they are sufficiently accurate to test whether the rates predicted by the model were within an expected range of emission rates under present day climate conditions (Geron et al., 1994). At present, these estimates cannot be used to predict the variation in future emissions under changing climate conditions, despite the use of increasingly sophisticated measurement techniques, using satellite and image sensor data to determine forest canopy species composition, due to the empirical nature of the leaf emission algorithms (Monson and Fall, 1991; Geron et al., 1994; Guenther et al., 1991; 1993; 1994). Therefore, the mechanistically-rich model presented has the potential to become a powerful hierarchical model, providing a comparative method of studying the effects of global warming on the amounts of isoprene emitted from temperate forest canopies. However, at present, a lack of measurements limits the verification of model predictions of proportionate change under future climates for whole canopies. To redress this would require further research, for example, by incorporating facilities to measure isoprene flux from forest canopies, using field measurement techniques. In addition, such measurements would help to clarify the extent to which leaf level model underestimations of emission rates at low light levels might be reflected in simulated canopy emission rates (Figure 2.9, p.56).

The inclusion of a seasonal vegetation coefficient in the isoprene canopy model to simulate the seasonal variation of leaf cover and isoprene emission rates can be considered a significant contribution. It is known that young leaves do not emit isoprene until several days after the initiation of photosynthesis, and that the activity of the enzyme isoprene synthase is correlated with this time lag (Grinspoon et al., 1991; Kuzma and Fall, 1993). Also, the induction time period is dependent on cumulative springtime temperatures (Monson et al., 1994). If isoprene is synthesised from IPP transported from the cytosol, a mechanism for the spring-time induction might be the impermeability of the chloroplasts of young leaves to exogenous IPP (Heintze et al., 1990). In autumn, the decline in isoprene emission rates is suggested to be influenced by the breakdown of metabolic machinery within the leaf and loss of leaf nitrogen, as Fiso is correlated with the nitrogen content of leaves (Monson et al., 1994). Whilst the inclusion of the coefficient of seasonal vegetation cover allows the differentiation between deciduous and non-seasonal forest types, a useful development of the model would be the incorporation of species specific emission rates. This would allow the differentiation of forest canopies into woodland types, although the spatial resolution of data of forested areas would still limit the accuracy of emission rate simulations. Such differentiation has been used by other researches, for example, to estimate emission rates for US woodlands (Guenther et al, 1994). High resolution (1.1km) grid land-cover database, compiled by the EROS Data Centre, from satellite and ancillary data, were used in combination with data of species composition, foliar mass, and emission rate factors, to estimate emission rates of isoprene and other volatile organic compounds. The emission rate factors were based on the measured emission rates for leaves of 49 tree species. It would be of interest to determine whether the species specific emission rates could be modelled mechanistically, for example via differences in leaf isoprene synthase content, or Vcmax. Such developments would enhance the canopy estimates which rely on empirical algorithms. Another method for estimating canopy emission rates of volatile organic compounds, including isoprene, incorporates the data of area extent, species composition and tree diameter distributions, provided by the US Department of Agriculture, Forest Service Forest Inventory and Analysis Eastwide Database, into a simple canopy model (Geron et al., 1994). However, this method again uses empirically derived equations to simulate emission rates. The combination of such methods with process-based models of leaf isoprene emission rates would provide a significant advance in accurate estimates of canopy isoprene emission, at least in areas where such spatially high resolution forest data are available, and presumably, with the increasing use of satellite data, estimates of how climate change might affect global rates of isoprene emission would soon follow.

Further research is required to improve estimates of canopy isoprene emission, particularly to increase our knowledge of the mechanisims underlying the effects of temperature and carbon dioxide concentration on isoprene emission. An increased understanding of the temperature dependency of isoprene synthase activity should enable a more mechanistic equation to be fitted to the relationship between temperature and isoprene synthesis, and so allow the parameterisation of growth temperature effects. The optimum temperature for isoprene emission (Figure 2.12, p.60) is related to growth temperature, and so the optimum temperature for trees growing in tropical conditions would be expected to have a higher optimum temperature than those growing under temperate conditions (Monson et al., 1994). The optimum temperature will also be complicated by a presumed upper limit for the optimum temperature, related to the temperature above which isoprene synthase starts to denature. Thus, in order to increase the confidence in the model simulations for canopies from different climate zones, the canopy model would benefit from a more mechanistic equation to account for the relatively high optimum temperature of emitting species growing in warmer, tropical areas, as well as the maximum optimum temperature set by enzyme denaturation. The optimising of emission rates to growth temperature supports the suggestion that isoprene plays a thermal tolerance role in isoprene emitting plants (Sharkey and Singsaas, 1995; Sharkey, 1996).

Long term effects of temperature on isoprene synthase activity also need greater understanding. In the long term, temperatures that are higher than the optimal temperature of isoprene synthase for a species growing in a specific climate might be expected to cause an increase in F'isotot, as the plants adapt, for example, by changing mechanisms influencing isoprene synthase activity. Alternatively, if the higher temperatures exceed the temperature at which isoprene synthase starts to denature, F'isotot would be expected to be inhibited. Understanding the capacity and limitations of adaptation of isoprene emitting species is of great importance in tropical regions, where most isoprene is currently emitted (Müller, 1992).

Isoprene emissions from temperate deciduous forests are predicted by the model to cause a greater than 20% rise with a 2oC increase in air temperatures. Presumably, as P. tremuloides was used for the construction of the model, the optimum temperature for F'isotot used in the model simulations is appropriate for some temperate forests. At present, isoprene is the dominant hydrocarbon during the daytime period in deciduous forest of Eastern USA (Central Pennsylvania) and an increase in emissions would be expected to cause a significant effect on tropospheric chemistry (Martin et al., 1991). Research is needed to reduce the uncertainty of predicting the temperature dependency of isoprene emissions at high temperatures, due the denaturation of the enzyme and concurrent higher emission rates at high temperatures, particularly as high ozone episodes in the USA occur on hot summer days (Monson et al., 1994). Likewise, a greater than 20% increase in emissions would be expected to dramatically affect the atmospheric chemistry in the UK, where isoprene is currently not thought to be a significant precursor of ozone (MacKenzie et al., 1991).

Research into the effects of [CO2] on F'isotot is also needed. For example, the more surprising results of model simulations are from the interactive effects of concurrent increases in [CO2] and temperature, which the model predicts will result in a decrease in isoprene emission rates from both temperate and tropical forests, of over 25% for the deciduous forests of the United Kingdom and the USA, and over 35% from the tropical rain forest of the Amazon. This effect is due to the interactions between the direct inhibitory effect of elevated [CO2] on leaf emissions, as presented in Chapter Two (Figure 2.9, p.56), the result of the high temperatures of sunlit leaves caused by decreased stomatal closure and a reduction in evapo-transpiration rates and evaporative cooling under elevated [CO2], in addition to direct temperature effects and leaf shading, dependent upon leaf area index. This prediction, if fulfilled, could have dire consequences for forested regions should isoprene play a thermal tolerance role within plant tissues as suggested by Sharkey and Singsaas (1995), unless acclimation occurs. On the other hand, if isoprene emission plays a gas phase, anti-oxidant role within leaves, scavenging harmful oxygen radicals within the chloroplasts, for example, then elevated [CO2] would be expected to cause a decline in Oi relative to Ci and a reduction in the risk of oxidative damage within chloroplasts (Sharkey and Loreto, 1993; Harley et al., 1994). The detailed sensitivity analysis (Table 2.4 p.64) shows that at elevated [CO2] an increase in F'isotot can be caused by increasing Vc and Oi, supporting this hypothesis. However, the uncertainty of the simulation of emission rates at high Ci, as illustrated in Figure 2.11, p.59, needs to be borne in mind. Therefore, future research is required to establish the effects of elevated [CO2] on isoprene synthesis, and sensitivity analysis (Table 2.4, p.64) shows the critical factors at elevated [CO2] to be studied are Vc, Oi, and h .

Other unknowns directly concerning canopy isoprene emissions include the effects of the distribution of leaves within a canopy, for example, clumping within a deciduous forest canopy, as well as the distribution of trees, for example, edge effects together with linear and more isolated tree patterns (Monson et al., 1991a; Baldocchi et al., 1995). Also, uncertainty surrounds the effects of long term growth conditions on isoprene emission. For example, Quercus rubra and Populus tremuloides grown under elevated [CO2] are found to differ in their response to long-term elevated [CO2]. Emission rates from Q. rubra are found to increase, whilst those from P. tremuloides decrease (Monson et al., 1991a). Also, Fiso increases with increased growth PFD levels, which affect isoprene emission rates more than rates of photosynthesis (Sharkey et al., 1991b). This is consistent with hypothesis of isoprene acting as a scavenger of chloroplast oxygen radicals, as is the finding that high isoprene emission rates are associated with shade intolerant species, and low and negligible emission rates are associated with shade tolerant species (Sharkey and Loreto, 1993; Harley et al., 1994; Baldocchi et al., 1995).

The model simulations also cannot take into account the spatial and temporal shifts in ecosystems as plants and habitats shift to gradually adapt to new climate conditions. For example, it is uncertain how leaf area dynamics, phenologies and species composition with respect to emitters and non-emitters will be affected by a warmer, drier climate (Monson et al., 1991a).

Some very general assumptions have been made for the model simulations presented here concerning climate change. For example, the simulated increase in temperature of 2oC has been assumed to occur uniformly over the globe and throughout the whole year. However, general circulation models used for the IPCC Report predict non-uniform global warming, with some areas, particularly the mid to high latitudes warming more in the winter season (Mitchell et al., 1990). Nor does the model take into account other climate variation, for example, changes in spatial rainfall patterns and storm frequency. F'isotot may alter dramatically in a region where the climate becomes drier as well as warmer, if the response found in kudzu leaves (Pueraria lobata (Willd) Ohwi.) by Sharkey and Loreto (1993) should prove to be a response applicable to other isoprene emitting species. Under water stress conditions, Fiso was found to initially decrease, only to increase to several times the pre-stress rate following the relief of water stress.

[ Home ] [ Up ]

Send mail to humph@essex.ac.uk with questions or comments about this web site.
Copyright © 1998 WIMOVAC Ltd.
Last modified: August 19, 1997