Isoprene emission variability through the twentieth century


  • N. Unger

    Corresponding author
    1. School of Forestry and Environmental Studies, Yale University, New Haven, Connecticut, USA
    • Corresponding author: N. Unger, School of Forestry and Environmental Studies, Yale University, 195 Prospect St., New Haven, CT 06511, USA. (

    Search for more papers by this author


[1] A biochemical model of isoprene emission embedded within a global chemistry-climate simulation framework is applied to investigate the transient response to environmental change over the past century. In the model, the isoprene production is directly coupled to photosynthesis and depends on intercellular carbon dioxide concentration (CO2), atmospheric CO2, and canopy temperature. Sensitivity runs are performed to isolate the relative roles of individual global change drivers: CO2, physical climate, and anthropogenic land cover change (ALCC). Between 1880 and 2000, atmospheric CO2 increased by ~30% from 291 to 370 ppmv, global average surface air temperature increased by 0.7°C, and the crop cover fraction of vegetated land area more than doubled from 15 to 37%. Over the past century, isoprene emission has decreased globally by 20% from 534 to 449 Tg C/yr, while gross primary productivity has increased by 15% from 107 to 124 Pg C/yr mostly due to CO2 fertilization. In terms of individual drivers, the global isoprene source increased by 7% due to the atmospheric CO2 concentration rise (including the opposing effects of CO2 fertilization and CO2 inhibition), decreased by 22% due to ALCC, and increased by only 3% due to physical climate change. Thus, ALCC is the dominant driver of isoprene emission change. Modeled global isoprene emissions were higher in the preindustrial than in the present day. In the industrial era, isoprene emission change represents a human-induced climate forcing, analogous to land use-driven CO2 emission, not a climate feedback because temperature-driven increase was a relatively weak driver of isoprene emission change from 1880 to 2000.

1 Introduction

[2] Terrestrial ecosystem emission of isoprene is a fundamental quantity in chemistry-climate interactions [Fiore et al., 2012]. Isoprene emission depends on vegetation cover and climate and is therefore sensitive to global change. In turn, the atmospheric photooxidation of isoprene affects the atmospheric chemical composition including ozone and aerosol particulate matter that contribute to air pollution and modify Earth's radiation budget [Unger, 2012]. Large-scale perturbations to isoprene emission may provide a powerful lever on regional climate and even trigger feedback to global climate [Pitman et al., 2012]. Therefore, understanding isoprene emission response to global change on climatically relevant spatiotemporal scales (e.g., 100–40,000 km and > months) is a critically important research area [Arneth et al., 2007a; Beerling et al., 2011; Heald et al., 2009; Lathiere et al., 2005; Liao et al., 2009; Naik et al., 2004; Sanderson et al., 2003; Valdes et al., 2005].

[3] Isoprene emission responds positively to increases in temperature and vegetation productivity but displays an inverse relationship with concomitant increases in atmospheric CO2 concentration [Monson et al., 2007; Possell et al., 2005; Possell and Hewitt, 2011; Rosenstiel et al., 2003]. This direct “CO2-inhibition effect” has been found to offset entirely the isoprene increases due to temperature and vegetation productivity in the future warming world [Arneth et al., 2007a; Heald et al., 2009]. In consequence, land use and land cover change is a major driver of global isoprene emission change in past [Lathiere et al., 2010; Tanaka et al., 2012] and future climates [Tai et al., 2013], and similarly a major driver on regional scales [Arneth et al., 2008; Constable et al., 1999; Purves et al., 2004; Steiner et al., 2002]. This sensitivity to vegetation cover arises because isoprene emission is strongly dependent on ecosystem type with broadleaf trees and shrubs exhibiting the strongest emission potentials and grassland and crops either low or nonemitting [e.g., Guenther et al., 2006].

[4] Through the twentieth century, vegetation cover has been much more influenced by anthropogenic land cover change (ALCC) than by natural vegetation dynamics [Hurtt et al., 2006]. For instance, ALCC has altered one third to one half of the Earth's land surface with the most significant change in terms of area being the conversion of forest areas to agricultural uses [Vitousek et al., 1997]. Previous studies of global change impacts on isoprene emission through the twentieth century used an isoprene emission model that had been specifically designed to reproduce the short-term response to weather variables in air quality modeling applications [Guenther et al., 1995]. In this approach that is used in the detailed canopy environment Model of Emissions of Gases and Aerosols from Nature (MEGAN, [Guenther et al., 2006]), isoprene emission rate is described using empirical functions of temperature and light in the form of serial multipliers. The effects of terrestrial productivity are manifest through a model dependence on leaf area index (LAI). Changes in LAI in different climate states are typically obtained using dynamic vegetation models, but LAI response to climate change is highly uncertain and not robust across models [Migliavacca et al., 2012; Richardson et al., 2013].

[5] A new generation of process-based global models link isoprene emission directly to photosynthesis [Arneth et al., 2007b; Pacifico et al., 2011; Unger et al., 2013]. A major difference compared to MEGAN is that the isoprene emission simulated in the photosynthesis-dependent models will respond directly to CO2 fertilization, specifically increases in leaf-level photosynthesis due to rising atmospheric CO2 concentration [De Kauwe et al., 2013; von Caemmerer, 2000]. In reality, the overall canopy-level gross primary productivity (GPP) response may be limited by water and nutrient availability depending on ecosystem and region. Net primary productivity (NPP) response to atmospheric CO2 increase is subject to greater uncertainty than the GPP response because of the simultaneous temperature driven increases in respiration [Friedlingstein et al., 2006]. For a given increase in NPP, plants may allocate the additional assimilated carbon to other biomass storage (roots, stems, and trunks) in addition to leaves.

[6] The goal of this work is to reexamine the transient evolution of isoprene emission over the past century (1880–2000) using an isoprene emission model that is directly coupled to photosynthesis and depends on intercellular carbon dioxide concentration (CO2), atmospheric CO2, and canopy temperature. In this approach, increasing atmospheric CO2 concentration influences isoprene emission through opposing mechanisms: CO2 fertilization acts to increase the isoprene emission rate through increasing photosynthesis, and the CO2-inhibition effect acts to decrease the isoprene emission rate. A large suite of sensitivity simulations is performed to isolate the relative importance of the different global change drivers (atmospheric CO2, ALCC, and physical climate change) over the historical period. In section 2, the methods and simulations are described. Results are presented in section 3, and discussion and conclusions are presented in section 4.

2 Methods

[7] The strategy employed in this study is to simulate the isoprene emission and GPP for each decade between 1880 and 2000 in a time slice approach.

2.1 Yale-E2 Global Carbon-Chemistry-Climate Model

[8] A biologically realistic leaf model of isoprene emission has been implemented into the vegetation biophysics module of the Yale-E2 global carbon-chemistry-climate model [Unger et al., 2013]. Yale-E2 is built around the new generation Intergovernmental Panel on Climate Change Fifth Assessment Report version NASA Goddard Institute for Space Studies model-E2 global climate model [Schmidt et al., 2013] and incorporates interactive terrestrial ecosystems, a dynamic carbon cycle module, and two-way coupling between the online vegetation and atmospheric chemistry. The model resolution used is 2° × 2.5° latitude by longitude with 40 vertical layers extending to 0.1 hPa. The vegetation submodel is embedded within the general circulation model that provides the key meteorological drivers for the vegetation physiology [Friend and Kiang, 2005]. The land surface hydrology submodel provides the grid cell level soil characteristics to the vegetation physiology. Comprehensive and rigorous evaluation of climatological means over the satellite era (1980–2004) indicates that the climate model skillfully reproduces the meteorological variables relevant for simulating vegetation biophysics and biogenic emissions, the main weaknesses being moist convection and overestimates in Southern Hemisphere cloud cover [Schmidt et al., 2013]. The vegetation is described using eight plant functional types (PFTs): tundra, grass, shrub, savanna, deciduous, tropical rainforest, evergreen, and crop. The canopy biophysical fluxes are computed using the well-established Michaelis-Menten leaf model of photosynthesis [Farquhar et al., 1980; von Caemmerer and Farquhar, 1981] and the stomatal conductance model of Ball and Berry [Collatz et al., 1991] described in detail elsewhere [Unger et al., 2013]. Leaf area index (LAI) for each PFT is prescribed according to regular seasonal sinusoidal variation between PFT-specific minimum and maximum seasonal LAI values. Each model PFT fraction in the vegetated part of each grid cell represents a single canopy. The model vertically stratifies each canopy into diffuse and direct light levels, and LAI profiles using an adaptive number of layers (typically 2–16) [Friend and Kiang, 2005]. Generally, the monthly average simulated LAI at the grid cell scale agrees to within a factor of 2 with the satellite-derived data from the Moderate Resolution Imaging Spectroradiometer [Myneni et al., 1997a] and the Global Modeling and Assimilation Office Modern Era-Retrospective Analysis land product data set [Rienecker et al., 2011]. Because of the unresolved uncertainties in prognostic phenology schemes and the likely small role that LAI changes have played on isoprene emission change in the past few decades, the LAI scheme in this work responds to ALCC but is insensitive to other climate drivers.

2.1.1 Isoprene Emission Model

[9] A leaf-level isoprene emission model that describes the constitutive production as a function of the electron transport-limited photosynthesis rate, Je, [Niinemets et al., 1999] has been integrated into the canopy biophysics scheme following modifications for global-scale modeling [Arneth et al., 2007b]. The leaf-level isoprene emission rate (I) in units of µmol/m2[leaf]/s is calculated:

display math(1)

where Je is the electron transport limited photosynthesis rate in units of µmol/m2[leaf]/s. The Je is a linear function of the incident photosynthetically active radiation (PAR) and the internal leaf CO2 concentration (Ci):

display math(2)

where aleaf is the leaf specific light absorptance and αqe is the intrinsic quantum efficiency for photosynthetic CO2 uptake in the chlorophyll reaction system that absorbs PAR to drive the oxidation of water and the reduction of enzymes (photosystem II). The αqe is a product of the fraction of absorbed light that reaches photosystem II and the CO2 per absorbed photon. The Γ* is the CO2 concentration compensation point in the absence of nonphotorespiratory respiration [Collatz et al., 1991].

[10] The β term in equation ((1)) translates the electron flux into isoprene equivalents given by equation ((3)):

display math(3)

[11] A detailed description of the mechanistic origin of the coefficient values is given elsewhere [Niinemets et al., 1999; Pacifico et al., 2011].

[12] The atmospheric CO2 inhibition is included via a simple parameterization (κ):

display math(4)

where Ci_s is the leaf internal CO2 concentration at standard atmospheric CO2, which is chosen to be the year 2000 global average value (370 ppmv). Equation ((4)) mimics the observed response to both short-term and long-term changes in Ci [Heald et al., 2009; Wilkinson et al., 2009]. For example, short-term reductions in Ci due to stomatal closure under drought conditions imply increases in κ, while the long-term effects of increasing (decreasing) atmospheric CO2 imply decreases (increases) in κ.

[13] The temperature relationship (τ) in the algorithm accounts for the difference in temperature optimum between photosynthesis and isoprene synthase:

display math(5)

where T = canopy temperature and Tref = standard temperature condition (30°C). Canopy temperature is computed according to the whole canopy energy balance as a function of the rate of change of canopy heat (depending on net radiation, latent heat of vaporization, heat flux of precipitation, and throughfall), water on the canopy, and the heat capacity of the dry canopy [Friend and Kiang, 2005, and references therein].

[14] The vegetation biophysics module computes the photosynthetic uptake of CO2 coupled with the transpiration of water vapor and the isoprene emission rate at the 30 min physical integration time step of the global carbon-chemistry-climate model. This version of the isoprene emission model can respond to anthropogenic climate change through: CO2 fertilization, reduced stomatal conductance/increased water-use efficiency, CO2 inhibition, and the temperature and precipitation responses of photosynthesis.

[15] The isoprene emission simulation has been extensively evaluated for conditions representative of the present-day climatic state and shown to reproduce 50% of the variance across different ecosystems and seasons in a global database of 28 measured campaign-average fluxes and to capture authentically the observed variance in the 30 min average diurnal cycle (R2 = 64–96%) at nine sites where the flux magnitude is simulated to within a factor of 2 [Unger et al., 2013].

2.2 Simulations

[16] The model is run in atmosphere-only configuration. The boundary conditions for each decadal simulation are shown in Table 1. The general circulation model component of Yale-E2 simulates its own climate state based on these boundary conditions. Decadal average monthly varying sea surface temperature and sea ice climatology from the Hadley Centre Sea Ice and Sea Surface Temperature dataset (HadSST2) data set provide the physical climatic boundary conditions for each decade between 1880 and 2000 [Rayner et al., 2006]. ALCC between 1880 and 2000 is based on a decadal harmonized gridded data set [Hurtt et al., 2011]. The observed decadal average atmospheric CO2 concentration is used in each historical time slice. A set of control simulations is performed for each decade between 1880 and 2000 that includes all drivers of global change (SimCONT). A further three sets of sensitivity simulations are performed to isolate the relative importance of each global change driver 1880–2000: SimCO2 allows only atmospheric CO2 concentration to increase; SimALCC allows only anthropogenic land cover change to vary; SimPCLIM allows only the physical climate to evolve. A total of 49 separate simulations are performed. For each simulation, integrations of 10 model years were completed and averaged for analyses. Interannual variability within each decadal time slice is due to the internal variability in the climate model. The global annual average surface air temperature increases by +0.7°C, and the global annual average precipitation rate increases by +0.06 mm/d between 1880 and 2000. In the model used here, physical climate change impacts GPP and isoprene emission through changes in temperature, precipitation, and soil moisture (which affect water availability), and radiation (cloudiness and atmospheric aerosol that affect PAR). Atmospheric CO2 concentration increases from 291 to 370 ppmv. The fraction of Earth's vegetated area covered with cropland doubles from 15 to 37%. Impacts are examined globally and in five broad emission regions: Central Africa (30°W–45°E, 15°S–15°N), Amazon Basin (80–40°W, 20S–10°N), North America (125–60°W, 15–55°N), East Asia (110–136°E, 52–70°N), and Europe (10°W–50°E, 25–65°N).

Table 1. Boundary Conditions Applied in the Decadal Time Slice Simulationsa
Decadal Time SliceObserved SSTs and Sea IceGlobal Average Surface Air Temperature (°C)CO2 (ppmv)Crop Cover (%)
  1. aThe crop cover is given as percentage fraction of vegetated land area.
1880 s1876–188513.929115
1890 s1886–189413.829416

3 Results

[17] Figures 1a and 1b show that at the global scale between 1880 and 2000, GPP has increased by 15%, while isoprene emission has decreased by 20%. Between 1880 and 2000, global GPP increases by 18% due to the CO2 increase, decreases by 3% due to ALCC, and increases by about 2% due to physical climate change (Figure 2a). Thus, the GPP increase over the past century is dominated by CO2 fertilization, mostly since The Great Acceleration of the 1950s. At the global scale, physical climate changes (temperature and precipitation) and ALCC have only relatively small impacts on GPP over the century. Between 1880 and 2000, global isoprene increases by 7% due to the CO2 increase, decreases by 22% due to ALCC, and increases by about 3% due to physical climate change (Figure 2b). Thus, the isoprene decrease over the past century is dominated by ALCC and physical climate change has only a relatively small impact on isoprene for this time period. The combined effects of CO2 fertilization and CO2 inhibition result in an overall increase in isoprene emission.

Figure 1.

(a) Historical transient annual average global and regional isoprene emissions (1880–2000) in the control and sensitivity simulations. (b) Historical transient annual average global and regional isoprene emissions (1880–2000) in the control and sensitivity simulations.

Figure 2.

(a) Year 2000 minus year 1880 change in GPP (g C/m2/d) due to all global change effects (top left), CO2 only (top right), ALCC (bottom left), and physical climate change (bottom right). (b) Year 2000 minus year 1880 change in isoprene emission (mg C/m2/d) due to all global change effects (top left), CO2 only (top right), ALCC (bottom left), and physical climate change (bottom right).

[18] Figures 1a and 1b show that regional isoprene and GPP changes over the century follow the same global trends and general driver responses and are in the range ±10–20%. ALCC is the dominant driver of the isoprene decreases in key emission regions over the past century. Figure 1b shows that in North America and Europe, isoprene emission was at a minimum around the midtwentieth century but since then followed an increasing trend, which in Europe is driven by ALCC and in North America is driven by a combination of CO2 and ALCC effects. In the Eastern United States and the United Kingdom, ALCC has caused increased isoprene emission due to the reversal of the cropland expansion (e.g., abandonment of agricultural land, reforestation, and afforestation) as shown in Figure 2b.

[19] While there is no scientific evidence to support a change in tropical LAI over the industrial era, a greening of the Northern Hemisphere (NH) high latitudes has been observed from space over the past 2–3 decades [e.g., Piao et al., 2006]. Whether the greening has been caused by CO2, temperature, precipitation, or a combination of these factors is still being debated. The LAI trend derived from satellite data is 0.0041 yr−1 from 1980 to 2000 [Myneni et al., 1997a, 1997b], which amounts an increase of about 2–3% in growing season NH average LAI at high latitudes. The NH greening is mostly in the boreal, which is not a strong isoprene emission region. It is possible that the isoprene reductions in Europe and North America may be slightly overestimated by a few percent at most because the NH greening over the past 2–3 decades has not been included here. However, such an increase would not be nearly enough to offset the large ALCC-driven decreases in these regions (~20%).

[20] The results presented here are consistent with a recent study based on a similar photosynthesis-dependent global isoprene emission scheme that found a 26% decrease in isoprene emission between 1850 and 2000 but did not isolate the roles of individual drivers [Pacifico et al., 2012]. The overall result is consistent with a study that used a modified version of the MEGAN model driven with off-line meteorology and found a 24% decrease in isoprene emission between 1901 and 2002 [Lathiere et al., 2010]. Yet the isoprene emission response to different global change drivers was somewhat different to those presented here. For instance, Lathière et al. found that global isoprene decreases by 21% due to the CO2 increase, decreases by 15% due to ALCC, and increases by about 7% due to physical climate change. The most striking difference is in the response to rising atmospheric CO2 concentration. This study and Lathière et al. both account for the direct CO2 inhibition. However, in this work, the isoprene emission response to CO2 fertilization manifests directly through photosynthesis, whereas in Lathière et al. the effects of CO2 fertilization on isoprene emission indirectly transmit through changing NPP and LAI. In a study that examined isoprene emission change over the period 1854–2000 based on the MEGAN isoprene emission algorithms but not accounting for the CO2-inhibition effect, global isoprene emission was found to decrease by only around 2% across the century [Tanaka et al., 2012]. In that study, global isoprene decreases by 7% due to ALCC, increases by about 7% due to the surface air temperature increase, and decreases by 2% due to a decrease in downward solar radiation over this time period.

4 Discussion and Conclusions

[21] The historical trend in isoprene emission over the past century has been reexamined using a global photosynthesis-dependent model that allows an informed integration of plant physiology, climate change, and ALCC. The relative importance of each of individual global change drivers (atmospheric CO2, climate change, and ALCC) on isoprene emission has been probed on climatically relevant spatiotemporal scales. Globally, GPP has increased by about 15%, while isoprene emission has decreased by about 20%. The historical GPP change is controlled by CO2 fertlization. The historical isoprene change is controlled by ALCC, and the impact is so large that is completely overrides the projected increase in GPP and forces an overall decrease in isoprene. The magnitude of the overall isoprene decrease is consistent with previous studies (~20–26%).

[22] This work demonstrates that ALCC and increased atmospheric CO2 concentration dwarf the effects of climate change and increased global average surface temperature on isoprene emission and thus atmospheric chemistry. Indeed, physical climate change, including the human-induced temperature increase, is a relatively weak driver of isoprene change over the last century. Yet in the most recent community assessment of historical anthropogenic ozone radiative forcing led by the Atmospheric Chemistry and Climate Model Intercomparison Project, only 4 of 15 participating state-of-the-science global chemistry-climate models included climate-sensitive isoprene emission [Young et al., 2013]. Those four models all projected a small increase in isoprene emission from preindustrial to present day in response to temperature change. The effects of ALCC and atmospheric CO2 on isoprene and atmospheric chemistry should be further examined with other model platforms.

[23] There are two implications of the results. First, isoprene emission was likely higher in the preindustrial than present day. This finding is consistent with previous work in which higher preindustrial isoprene emissions were evoked as an explanation for low preindustrial surface ozone observations [Mickley et al., 2001]. Second, isoprene emission change over the past century was driven by ALCC and is therefore a human-induced climate forcing mechanism [Ashworth et al., 2012; Ganzeveld et al., 2010]. Isoprene has been conventionally treated as a climate feedback mechanism, a response of the terrestrial biosphere to anthropogenically forced increases in temperature [Arneth et al., 2010].

[24] In future work, the model will be used to quantify the impact of the historical isoprene emission change on tropospheric ozone, secondary aerosol particulates, and methane. Furthermore, the model will be used to explore ALCC impacts on isoprene and atmospheric chemistry for a broad range of possible futures. In particular, the potential for replacement of native forests with bioenergy plantation forests that are major isoprene emitters to affect chemistry-climate interactions (in this case ALCC drives increases in isoprene emission) will be examined.

[25] The main limitations of this work are (1) uncertainties in the implementation of the CO2-inhibition effect that could impact the relative importance of ALCC and (2) the use of eight PFTs to describe the vegetation that do not account for plant-to-plant species level variability in isoprene emission within a particular PFT. However, inclusion of greater species diversity would not impact the main conclusions of this study that depend on the drastically different isoprene emission potential between broadleaf forests and crops.


[26] Funding for this research was provided by Yale University. Conversations with P. Friedlingstein, S. Sitch, and A. Goldstein were helpful and greatly appreciated. This project was supported in part by the facilities and staff of the Yale University Faculty of Arts and Sciences High Performance Computing Center.