Isoprene and monoterpene emissions since the preindustrial era

Introduction Conclusions References


Introduction
Monoterpene secondary organic aerosols (SOAs) are the most important organic aerosol components on a global scale (Chung and Seinfeld, 2002;Engelhart et al., 2008). SOAs act as cloud condensation nuclei (Novakov and Penner, 1993) and absorb solar radiation (Andreae and Crutzen, 1997). The photooxidation of isoprene gen-5 erates SOAs (Jang et al., 2002). However, previous estimates of isoprene-related SOA levels are being reconsidered (Claeys et al., 2004;Henze and Seinfeld, 2006;Paulot et al., 2009), as an organic aerosol that has been found in several forested areas is strongly indicative of an isoprene precursor (Matsunaga et al., 2003;Ion et al., 2005;Kourtchev et al., 2005). Vegetation is thought to emit about 90 % of volatile organic car- 10 bon compounds (VOCs) (Kuhn et al., 2004), and Guenther et al. (1995) estimated that the annual global VOC flux is 1150 Tg C, composed of 44 % isoprene, 11 % monoterpenes, 22.5 % other reactive VOCs, and 22.5 % other VOCs. That study also showed that the contribution to VOCs from vegetation should not be ignored when considering carbon cycles. The estimation of VOC emissions from vegetation, especially isoprene 15 and monoterpene emissions, is essential for understanding global tropospheric chemistry and regional photochemical oxidant formation, for balancing the global carbon cycle, and for understanding the production of organic acids (Fehsenfeld, 1992).
Vegetation is influenced by climate changes, and much vegetation has been rapidly replaced with croplands since the preindustrial era (Ramankutty and Foley, 1999; Thus, the observed changes in global VOC emissions may have influenced spatial and temporal SOA composition since the preindustrial era (1850s) (Tsigaridis et al., 2006). Lathière et al. (2005) estimated global annual isoprene and monoterpene emissions from the terrestrial biosphere between the preindustrial era and present day, using static mode simulation; annual values were found to be 409 Tg C and 127 Tg C for the 10 preindustrial era, and 402 Tg C and 131 Tg C for the present day, respectively. Lathière et al. (2010) also estimated global annual isoprene emissions from the terrestrial biosphere during the 20th century with a dynamic vegetation model that considered the negative effect of an increase in leaf area under rising atmospheric CO 2 , and found that anthropogenic cropland expansion contributed the most (15 %) to the reduction in iso- 15 prene emissions that occurred by 2002, while climate changes and rising CO 2 caused a 7 % increase and a 21 % reduction, respectively. Thus, the authors estimated that the present day annual global isoprene emissions would decrease and that monoterpene emissions would increase in the current era compared to the preindustrial era and to the early 20th century, and they demonstrated the influences of climate change 20 and cropland expansion on both types of emissions (Lathière et al., 2005). The authors also demonstrated the influence of ambient CO 2 on isoprene emissions (Lathière et al., 2010).
In the present study, we estimated the annual global isoprene and monoterpene emissions from the preindustrial era to the present. We used the Model of Emissions of Introduction how the expansion of cropland, and climate factors such as air temperature and solar radiation, influenced the annual global isoprene and monoterpene emissions from the preindustrial era to the present. Simulations also considered historical emissions from areas including and excluding large expansions of cropland and how each vegetation type in each area contributed to both annual emissions from 1854 to 2000.

2 Materials and methods
To estimate emissions for isoprene and monoterpenes (classified by eight components: myrcene, sabinene, limonene, 3-carene, ocimene, β-pinene, α-pinene, and other monoterpenes), we used the MEGAN model (Guenther et al., 2006) and monthly climatic data including ambient solar radiation and air temperature at 2 m above the land surface (Watanabe et al., 2011), reproduced by a historical run from 1850 to 2005 with MIROC5 (Watanabe et al., 2010), which is an atmosphericocean circulation model with the standard resolution of the T85 atmosphere and one-degree ocean models. The model considered historical solar irradiance data (Lean et al., 2005) and surface aerosols emission data, and it reproduced the ob- 15 served global mean surface air temperature during the 20th century well (Watanabe et al., 2011). The expansion of cropland is described as the ratio of cropland to each grid (Hurtt et al., 2006). The global distribution of potential vegetation types shown by Ramankutty and Foley (1999)

A model for emissions of isoprene and monoterpenes
The emission of VOCs (in this case, isoprene and monoterpenes) is described in 10 MEGAN as follows: where ε is the emission factor of isoprene or monoterpenes that represents the emission of a compound into the canopy under standard conditions, and λ LAI , λ age , λ L , and λ T are emission activity factors for LAI, age, light (or photosynthetic photon flux den-15 sity, PPFD), and temperature, respectively.
An emission activity factor for age is estimated as follows: where F is a fraction of foliage, A is relative emission activity, and the subscripts new, gro, mat, and old are new, growing, mature, and old foliages, respectively. The A new , A gro , A mat , and A old values were set at 0.05, 0.6, 1.125, and 1 for isoprene emission, and 2, 1.8, 0.95, and 1 for monoterpene emission, respectively. The F new , F gro , F mat , and 10 F old are estimated based on the current LAI (LAI c ), LAI of the previous month (LAI p ), and monthly temperature (T m ), in the following three cases: when LAI c = LAI p , F new = 0, F gro = 0.1, F mat = 0.8, and F old = 0.1; when LAI c < LAI p , F new = 0, F gro = 0, F mat = 1−F old , and F old = (LAI p −LAI c )/LAI p ; and when LAI c > LAI p , F gro = 1−F new −F mat , F old = 0, and F new and F mat are estimated as follows: and 20 where t is time step (in this case, 30 days), t i is the number of days between bud break and the induction of isoprene emission, and t m (or 2.3 · t i ) is the number of days between bud break and the initiation of peak isoprene emission. The value for t i is estimated as follows: where T i is the average ambient air temperature (K) of the preceding time step interval, and T m was used in place of T i in the study. An emission activity factor of light is estimated as follows: λ L = sin(a) 2.46 · {1 + 0.0005 · (P m − 400)}φ − 0.9φ 2 LDF + 1 − LDF at sin(a) > 0, or λ L = 0 at sin(a) < 0, 5 where a is solar angle in degrees (in this case, the monthly average value of solar angle only in the daytime), P m is the monthly average (original daily average) above canopy PPFD (µmol m −2 s −1 ), LDF is a light-dependence fraction that varies depending on the compound under consideration, and ϕ is the above canopy PPFD transmission, which 10 is estimated as follows: where P ac is the above canopy PPFD (here, P m ), and DOY is the day of year. An emission activity factor of temperature is estimated as follows: where .6 · (T 240 − 297)}, C T1 (= 80) and C T2 (= 200) are empirical coefficients, T d is the daily average temperature (K), T h is the hourly average temperature (K), T 240 is the 20 average air temperature over the past 240 h (K), and β is a temperature dependence (K −1 ), the value of which was set at 0.1. Equation (9) considers that leaves exposed to a past higher temperature release more isoprene than those exposed to a lower past temperature. The influence was disregarded, however, because monthly average ambient temperature at 2 m above the land surface (T m ) was used in place of T d , T h , 25 and T 240 .  Figure 1 shows the distribution of potential vegetation. Here, ten vegetation types including continental ice (Ice), broadleaf evergreen forest (BEF), broadleaf deciduous forest and woodland (BDFW), mixed coniferous and broadleaf deciduous forest and woodland (MCBDF), coniferous forest and woodland (CFW), high-latitude decidu-5 ous forest and woodland (HLDFW), wooded C4 grassland (WC4G), shrubs and bare ground (SBG), tundra (Tundra), and C3 grassland (C3G) were classified. These were determined according to Takata et al. (2003), with reference also to Ramankutty and Foley (1999). The values of ε and LDF for isoprene or monoterpenes are shown in Table 1. They 10 were set based on Guenther et al. (2006) and Sakulyanontvittaya et al. (2008), respectively. The values were the highest for isoprene in BEF and BDFW, which correspond to tree species extending to low latitudes. The values were the highest for most monoterpenes in CFW and HLDFW, which correspond to tree species extending to high latitudes, or in SBG for limonene and ocimene, respectively. The value for cropland was 15 the lowest for isoprene and most monoterpenes. The values of LDF indicate the higher dependency on land type for isoprene emission and the smaller dependency on land type for monoterpenes, excluding ocimene. VOC levels (e.g., isoprene, myrcene, sabinene, limonene) were calculated for both vegetation type and cultivation in a grid, using LAI c , LAI p , monthly DSR (S m ; W m −2 ), a, 20 and T m . Monthly above canopy PPFD (P m ; µmol m −2 s −1 ) was calculated as kS m , where k (µmol J −1 ) is an empirical coefficient and the value was set at 2. We estimated both isoprene and monoterpene emissions in the 11 regions shown in Fig. 2. The A1-A8 regions included parts of Europe, Africa, East Asia, India, Southeast Asia, Oceania, North America, and South America, respectively. These regions had the largest expansion of cultivation since 1850, as shown in the results. On the other hand, minimal expansion was estimated for regions A9, A10, and A11, which are regions 10 that include latitudes from 90 • N to ∼21 • N, from ∼21 • N to ∼21 • S, and from ∼21 • S to 90 • S, excluding the areas in the A1-A8 regions, respectively (Fig. 2). Figure 2 shows the area of each region; region A2 is the largest and A5 is the smallest in A1-A8.

Estimation of the influence of both cultivation and climate on emissions
Both isoprene and monoterpene emissions were calculated with the expansion of crop- 15 land in 1850 and climate from 1850 to 2005, and described as 10-yr running means (VOC veg1850 (y)). The value of VOC veg1850 (y) was compared to the value for VOC(y) calculated with the changes in both expansion of cropland and climate, to estimate the influence of expansion of cropland on both types of emissions. Here, y is the year from 1854 to 2000 (see Sect. 2.2). The influence of the expansion of cultivated land on the 20 emissions was estimated as VOC(y)/VOC veg1850 (y) − 1. The influence of climate was estimated as VOC veg1850 (y)/VOC veg1850 (1854) − 1. The influence of changes in temperature (SAT) and light conditions (DSR) were estimated as λ T (y)/λ T (1854) − 1 and λ L (y)/λ L (1854) − 1 for isoprene and monoterpene emissions, respectively. Here, the λ L (y) values for monoterpenes were averaged among the eight species, whose val- 25 ues were different as the emissions for most monoterpene species react very little to 16524 ACPD 12,2012 Isoprene and monoterpene emissions since the preindustrial era changes in light conditions while the ocimene emissions react strongly, as mentioned above.  2012) found that the anomalies were similar to those in observed SAT data, whose magnitude of interannual fluctuation was near a 10-yr running mean of reconstructed interannual SAT rather than a temporal one during the period from 1948 to 2006, as well as a reanalysis by the National Centers for Environmental Prediction/National Center for Atmospheric Research (NCEP/NCAR) (Kalnay et al., 1996). The DSR gradually decreased until the 1950s, fell again sharply from the 1960s to the 15 1980s, and then increased slight from the mid-1980s. Figure 4a, b show the extent of cropland in 1850 and in 2000, respectively. In 1850, extensive cultivation was already found in regions A1, A3, A4, and A7 (shown in Fig. 2). Figure 4c shows the difference in the extent of cropland between 2000 and 1850. By 1850, cultivation was already extensive in Europe but more extensive in the Ukraine 20 (region A1) and extended eastward into Russia by 2000 (Fig. 4b). Cultivation in 1850 was minimal in western regions of Africa (region A2), but expanded due to a change in distribution of WC4G-type vegetation (yellow area of Fig. 1). Cultivation was initially concentrated around the Yellow River basin in region A3, but over time extended eastnorthward and southward. Region A4 (Indian subcontinent) was already cultivated in ACPD 12,2012 Isoprene and monoterpene emissions since the preindustrial era cropland into other forms of vegetation was found in some areas of both A1 and A7 (Fig. 4c). Cropland expansion of more than 10 % (or 0.1) was found only minimally in regions A9-A11. Figure 5 shows the increase in cultivation and changes in vegetation types in A1 through A11 during the period 1850-2005. Cropland increased from 11.5 % to 24.4 %, 15 coinciding with a decrease in both C3G and MCBDF vegetation during the period 1850-1960 in region A1, subsequently decreased gradually until 1980, and then decreased more rapidly in the 1980s until it reached a value of 20.6 % (Fig. 5a). As shown in Fig. 5b, cropland area in region A2 increased from 2.4 % to 10.1 %, mainly due to the replacement of WC4G. Region A3 exhibited a steep increase in cropland after 20 1980 with cultivation in both MCBDF and SGB lands (Fig. 5c). In region A4, cropland increased from 19.5 % to 33.1 %, with the steeper increase from 1930 to 1960, but the increase rate slowed after about 1960 (Fig. 5d). Region A5 initially exhibited BEF vegetation of 76.3 % and BDFW vegetation of 10.9 %, respectively, but large portions of both were replaced with cropland by 2005 (Fig. 5e). As shown in Fig. 5f, the cropland 25 in region A6 increased clearly from about 1910 with a coinciding decline in SBG and WC4G. In region A7, cultivation increased constantly along with a decline in both C3G and MCBDF during the period 1850-1940, after which the rate of cropland expansion slowed (Fig. 5g). As shown in Fig. 5h, cropland in region A8 constantly increased along ACPD 12,2012 Isoprene and monoterpene emissions since the preindustrial era with a decrease in WC4G, SBG, and BDFW after ∼1900, while the decrease in BEF in this region was relatively lower than these three vegetation types. Regions A9 through A11 exhibited only small changes (<∼1 %) in cultivation (Fig. 5i, k); region A11, which has the third largest area, was 90.1 % ice cover without VOC emissions (Table 1). Overall, region A5 featured decreases in both BEF and BDFW with the highest emis-5 sion factors (ε) of isoprene (Table 1). The decrease in BDFW was also obvious in region A4. However, the decrease in both CFW and HLDFW, with the highest emission factors (ε) of monoterpenes (Table 1), were relatively small in higher latitudes such as A1, A3, and A7. In these areas, both C3G, which had the lowest ε, and MCBDF, which had a relatively high ε for monoterpenes, were replaced the most with cropland. Thus, compared to isoprene, the expansion of cropland had a lesser impact on the ε values of monoterpenes (Table 1).

Annual isoprene and monoterpene emissions and the influence of both climate and cultivation during the period 1854-2000
3.2.1 Global scale 15 Global annual isoprene and monoterpene emissions during the entire study period are shown in Fig. 6a Both annual emissions were estimated with cropland expansion from 1850 fluctuating with changes in SAT and DSR (Fig. 3), mainly SAT, during the period from 1854 to 2000 (Fig. 6a, b). The values were estimated to be 604 and 73.2 Tg C yr −1 for isoprene and monoterpenes, respectively, in 2000. Although the changes were similar to the annual emissions estimated by changes in cropland expansion, the differences be- respectively. The influence of a gradual decrease in DSR on isoprene emissions was evident as a 2.2 % reduction in emissions by 2000. In contrast, monoterpene emissions were minimally influenced by changes in DSR. Figure 7a, b shows the distribution of estimated annual global emissions for isoprene and monoterpenes, respectively, in 2000. Estimated isoprene emissions were concentrated in low latitudes with BEF and BDFW, in particular in BEF (or tropical rain forests), and also in MCBDF (Fig. 1). On the other hand, estimated monoterpene emissions occurred in high latitudes with CFW, HLDFW and WC4G, as well as in low latitudes.

Contribution of isoprene and monoterpene emissions from each area to
the global emissions Figure 8 shows how regions A1 through A11 contributed to global emissions during the study period. Isoprene emissions were largest in the following order: low latitudes of 20 region A10 with minimal expansion of cultivation, an area of South America in region A8, an area of Africa in region A2, and an area of Southeast Asia in region A5, all of which are distributed in lower latitudes. Monoterpene emissions were the largest in the following order: region A2, region A8, region A10, and an area of North America in region A7. Annual isoprene emissions were much larger in these four areas than 25 in any other areas, because both BEF and BDF with the highest ε are distributed in the areas with high temperature and solar radiation all year round (Fig. 7a) with high temperature during a few months in the summer season, the annual emissions were larger in lower latitudes, as in regions A2, A8, and A10, due to their high year-round temperatures. However, the differences (e.g., between A10 and A7) were relatively small, compared to those between the annual isoprene emissions from different regions.

5
The fluctuation in isoprene emissions in region A8 over the study period was similar to that in region A10, with relatively less change in land use, and the two areas contributed to the global annual emissions the most, which indicates that the emissions were dependent on climate changes rather than on the expansion of cropland. These areas had lower emissions with lower SAT from around 1950 to the 1960s (Fig. 3) than 10 in 1854. The annual emissions in region A5 were similar to those in region A2 from 1850 to the 1890s, but they decreased obviously starting around 1930, and the difference in annual emissions between regions A5 and A2 were the largest at the end of the time series. The influence of expansion of cropland on the annual emissions was the highest in region A5; region A5 likely contributed significantly to the decreases in the 15 annual emissions since the 1950s (Fig. 6a). The annual emissions also decreased with time in regions A4, A3, and A1, although the expansion of cropland had a lesser influence. Region A1 had slight increases in isoprene emissions by reforestation (Fig. 5) in the 1990s. The time series for annual monoterpene emissions were similar among regions A2 and A8, both of which had relatively large expansions of cropland, and region 20 A10, which had minimal expansion of cropland. Figure 9 shows how each vegetation type contributed to global annual emissions. Cropland minimally emitted isoprene according to the lowest emission factors (ε) (Table 1). Simulations showed that loss of both BEF and BDFW with the highest ε effectively decreased the annual isoprene emissions in region A5. A slight decrease due to 25 BEF loss appeared in region A2. In region A8, a decrease caused by BDFW loss of was more obvious than that caused by BEF loss. Decreases in regions A3 and A4 appeared because of the loss of MCBDF and BDFW, respectively. As previously mentioned, the ACPD 12, 2012 Isoprene and monoterpene emissions since the preindustrial era decrease caused by the loss of MCBDF appeared in the 1980s in region A1, but then reforestation increased the annual emissions around the 1990s. On the other hand, annual monoterpene emissions from cropland increased with its expansion, and offset the decreases due to other vegetation types to some degree. Thus, the replacement of natural vegetation with cropland decreased the annual 5 emissions of monoterpenes to a lesser extent than it did for isoprene. In region A1, reforestation also increased the annual monoterpene emissions around the 1990s, as it did for isoprene, but the contributions to both annual global emissions was quite small, as shown by Lathière et al. (2006). In regions A2 and A8, WC4G played an important role increasing the annual monoterpene emissions, but not isoprene emissions.

Discussion
We demonstrated the effects of the expansion of cropland and climate change on annual global isoprene and monoterpene emissions during the period 1854-2000. The expansion of cropland had a greater effect on isoprene emissions than on monoterpene emissions (Fig. 6). The expansion of cropland contributed to the annual global 15 emissions of monoterpenes to some degree but contributed only minimally to isoprene emissions (Fig. 9). The annual global emissions increased for monoterpenes mainly due to global warming, and they decreased for isoprene in the 1990s due to a decline in cropland, compared to those in the preindustrial era. The changes for both compounds between the preindustrial era and the present were consistent with those 20 demonstrated using a static mode simulation by Lathière et al. (2005).
DSR gradually decreased over the entire period, except for a rapid decline from the 1950s to the 1980s and a small increase in the 1990s, while SAT increased overall but with larger fluctuations (Fig. 3). The decline reconstructed by MIROC5 could describe the significant reductions in solar radiation during the past 50 yr, reported by ACPD 12,2012 Isoprene and monoterpene emissions since the preindustrial era addition, the reduction can be attributed to changes in optical properties caused by an increase in atmospheric water vapor due to global warming. The decrease in DSR decreased the annual global isoprene emissions by 2 %, while it had little impact on monoterpene emissions. The influence of DSR was the smallest of the three factors considered (Fig. 6). Considering the increase in the ratio of the diffused to direct radia-5 tion and the consequent increase in infiltration of solar radiation into regions with deep canopy (e.g., Mercado et al., 2009), the influences may become smaller. We disregarded the influences of ambient CO 2 , soil moisture, and other factors on global emissions, and the contributions to both annual global emissions from BEF in Southeast Asia with low latitudes such as in the Amazon were the largest. These find- 10 ings are further discussed in Sects. 4.1 and 4.2, respectively. We demonstrated that the influence of cropland on annual global isoprene emissions was the largest in Southeast Asia. The expansion of oil palm plantations with high emissions has occurred on the largest scale in Southeast Asia. However, the influences of this were not considered in the present study. In Sect. 4.3, we discuss how the expansion of this crop in Southeast 15 Asia may influence the estimate of annual global isoprene emissions.

Effects of ambient CO 2 , soil moisture, and other factors neglected in the study
Our estimate did not consider the influence of ambient CO 2 and soil moisture on either isoprene or monoterpene emissions. According to a review by Laothawornkitkul  (Sharkey et al., 1991;Staudt et al., 2001), decrease (Sharkey et al., 1991;Loreto et al., 2001;Rosenstiel et al., 2003;Possell et al., 2004;Vuorinen et al., 2004;Wilkinson et al., 2008), or have no significant effects (Penuelas and Llusia, 1997;Constable et al., 1999;Buckley, 2001;Centritto et al., 2004)  design and implementation. On balance, increasing CO 2 likely causes a decrease in isoprene emissions from the leaf surface. On the other hand, the decrease might be offset by increases in emissions as a result of increasing vegetation productivity and leaf area growth caused by elevated CO 2 levels (Possell et al., 2005;Arneth et al., 2007). Lathière et al. (2010) (Guenther et al., 2006), mainly because including soil moisture decreased emissions by more than 20 %. Moderate drought may decrease, enhance, or have no effect on isoprene and monoterpene emissions, although severe and long-lasting water stress significantly reduces BVOC emissions (Laothawornkitkul 15 et al., 2009). Vegetation classified here as BEF corresponds to tropical or seasonal tropical forests with a dry season and a wet season. The evergreen vegetation is likely to have deep roots (e.g., Canadell et al., 1996;Nepstad et al., 1994), and the consequent large water capacity may maintain leaves all year round (Tanaka et al., 2004). Thus, the emissions from BEF could be minimally reduced by soil moisture even in 20 a dry period. On the other hand, the emissions from SBG and BDFW, with high ε for both isoprene and monoterpenes around BEF, can be significantly reduced by soil moisture stress during dry periods (Table 1, Fig. 1). Therefore, both of our estimated global emissions may be overestimated because we disregarded the effects of CO 2 and soil moisture, even though our findings were within the ranges of the published 25 annual global emissions.
We used monthly instead of hourly data for SAT and DSR for our estimates of emissions. Because the monthly SAT value includes lower air temperatures at nighttime, when isoprene emissions do not occur, the use of monthly data might reduce estimated  (1997), to estimate isoprene emissions. But they did not consider the influence of diurnal patterns on the estimates, and neither did we. Müller et al. (2008) also examined how the dif-5 ferences between air temperature and leaf temperature influence estimated isoprene emissions, and showed that leaves are about 1 or 2 K warmer than their environment in most forested areas, resulting in emission enhancements of about 10 %. 10 Our estimates demonstrated that region A5 may have made the greatest contribution to annual global isoprene emissions, in particular from BEF (Fig. 9). The data also suggest that this region may have contributed to the annual global monoterpene emissions with constant emissions all year round. These results are consistent with many previous reports (e.g., Guenther et al., 1995Guenther et al., , 2006Müller et al., 2008). However, mea-15 surements of BVOC emissions from BEF at the canopy scale in Southeast Asia have only been done by Langford et al. (2010), while a relatively larger number of measurements have been done in Amazon forests (e.g., Helmig et al., 1998;Rinne et al., 2002;Greenberg et al., 2004;Karl et al., 2007;Kuhn et al., 2007;Müller et al., 2008) and in Africa (Greenberg et al., 1999;Serca et al., 2001). Langford et al. (2010) mea-20 sured BVOC emissions over a tropical rainforest in Malaysian Borneo and found that the emission rates for isoprene and monoterpenes were 4 and 1.8 times lower, respectively, than the default value for tropical forests in the MEGAN model used here, so our estimated emissions for region A5 may be underestimates. On the other hand, the estimated emissions in the abovementioned studies on Amazon forests varied widely. the need for more direct canopy-scale flux measurements of VOCs from the world's tropical forests.

Decrease in annual isoprene emissions in Southeast Asia and the expansion of oil palm
The simulated isoprene emissions also demonstrated that the influence of land use 5 changes on annual isoprene emissions during the study period were remarkable, in particular in Southeast Asia  and that the effective reduction by the expansion of cropland from BEF, BDFW, and other vegetation was −15.7 % between 1854 and 2000. In the area (4.47 × 10 6 km 2 ; see Fig. 2), an expansion of palm oil cultivation has occurred since the early 1980s; by 2006, the planted area had reached 10 around 6.2 × 10 4 km 2 (BPS, 2008) and 4.2 × 10 4 km 2 (MOPB, 2008) in Indonesia and Malaysia, respectively. Oil palm is one of highest isoprene emitters (Owen and Penuelas, 2005;Wilkinson et al., 2006;Geron et al., 2006;Misztal et al., 2011), and the total area occupied 2.3 % of the area of region A5. Moreover, the expansion likely continues at the expense of natural forest. The influence of expanded palm oil cultivation on iso-15 prene emission, however, was not considered in the present study. Misztal et al. (2011) measured BVOC emissions, including isoprene from a 15-year-old palm oil plantation in Malaysian Borneo, with an eddy correlation system, compared the measurements to those from a nearby rainforest measured by Langford et al. (2010) with the same system, and suggested that the isoprene concentrations from the oil palm site were 20 4 to 8 times greater than the values from the rainforest. Thus, the 9.6 % loss of BEF areas in region A5 during the period from 1854 to 2000 was likely offset or exceeded by subsequent positive effect of substitution with oil palm. Nonetheless, the estimated influences are appropriate before the 1990s, when oil plantations expanded exponentially. The influence will become increasingly important with further expansion after the 25 2000s. The LAI of an oil palm plantation changes with age, being 2.14 for 5-to 9-yearold palms and 2.37 for 15-year-old palms in the Malay Peninsula (Awal et al., 2010). Trees aged 9 to 15 yr are the most productive (Sheil et al., 2009) isoprene may be emitted the most during those years. Trees become too tall to harvest the fruits after 25 to 30 yr, and some long-established plantations in Malaysia have already been replaced for the third time (Basiron, 2007), indicating that the isoprene emissions should weaken during the replacement. Palms mature so rapidly that the fruit can be harvested as soon as 2 to 3 yr after planting (Basiron, 2007). Thus, such 5 changes in the characteristics of oil palm with age and plantation management will be essential for estimating isoprene emissions with the expansion of cropland area in Southeast Asia.