Reducing the negative human-health impacts of bioenergy crop emissions through region-specific crop selection

An expected global increase in bioenergy-crop cultivation as an alternative to fossil fuels will have consequences on both global climate and local air quality through changes in biogenic emissions of volatile organic compounds (VOCs). While greenhouse gas emissions may be reduced through the substitution of next-generation bioenergy crops such as eucalyptus, giant reed, and switchgrass for fossil fuels, the choice of species has important ramifications for human health, potentially reducing the benefits of conversion due to increases in ozone (O3) and fine particulate matter (PM2.5) levels as a result of large changes in biogenic emissions. Using the Community Earth System Model we simulate the conversion of marginal and underutilized croplands worldwide to bioenergy crops under varying future anthropogenic emissions scenarios. A conservative global replacement using high VOC-emitting crop profiles leads to modeled population-weighted O3 increases of 5–27 ppb in India, 1–9 ppb in China, and 1–6 ppb in the United States, with peak PM2.5 increases of up to 2 μg m−3. We present a metric for the regional evaluation of candidate bioenergy crops, as well as results for the application of this metric to four representative emissions profiles using four replacement scales (10–100% maximum estimated available land). Finally, we assess the total health and climate impacts of biogenic emissions, finding that the negative consequences of using high-emitting crops could exceed 50% of the positive benefits of reduced fossil fuel emissions in value.


Introduction
As bioenergy crops continue to replace both existing agricultural crops and natural landscapes, the choice of crop species will become increasingly important given their likely impacts on air quality and climate. The use of crops such as poplar, eucalyptus, and switchgrass as bioenergy feedstocks has increased globally over the past decade, and while ongoing adoption trends are highly dependent on both economic outcomes and policy decisions, significant increases are expected to continue (Energy Information Administration 2013). The large scale land-use changes associated with bioenergy production will have consequences on many aspects of environmental and human health, including food supply, watershed cleanliness, soil quality, and ecological diversity (e.g. Service 2007, Dominguez-Faus et al 2009, Ditomaso et al 2010, Pedroli et al 2013. In some cases, the environmental costs associated with bioenergy production and consumption have been estimated to outweigh the benefits of reduced fossil fuel combustion completely (Hill et al 2009, Tessum et al 2014. In addition to these concerns, many of the most popular candidates for bioenergy crop feedstocks are highemitters of isoprene, monoterpenes, and other biogenic volatile organic compounds (BVOCs), precursors of surface-level ozone (O 3 ) and fine particulate matter (PM 2.5 ). Recent observational and modeling studies of existing bioenergy crop plantations have highlighted the importance of examining ambient chemical and climatological conditions, along with crop emission profiles, when assessing the potential for large-scale cultivation to effect negative air quality consequences in any given area (Hewitt et al 2009, Porter et al 2012. Thus far, however, strategies to inform regional energy crop selection and limit the negative impacts of bioenergy cultivation on air quality and human health have not been adequately addressed.
With a global single-year death toll of seven million, air pollution has now been identified as the most significant environmental risk to human health (WHO 2014). Over half of the total estimated air pollution related deaths are linked to outdoor exposure, particularly to elevated levels of O 3 and PM 2.5 (Brauer et al 2012). These pollutants largely are secondary, forming in situ as a result of chemical reactions involving both naturally and anthropogenically emitted precursors (Sillman 1999, de Gouw andJimenez 2009), making effective control of their ambient concentrations particularly challenging. For both O 3 and PM 2.5 , fluxes of BVOCs can play a crucial role in determining peak daily pollutant concentrations. While all plants emit BVOCs to some extent, there is enormous variability in the magnitude of emission rates between different plant species (Guenther et al 1995). Thus, large-scale landscape-level changes in the type and quantity of plants can have profound consequences on regional air quality by affecting the rates of BVOC emissions, and therefore rates of O 3 and PM 2.5 formation (Wiedinmyer et al 2006, Ashworth et al 2012.
While recent work has examined the air quality consequences of both theoretical and observed landuse changes driven by bioenergy crop cultivation (Porter et al 2012, Ashworth et al 2013), a global study comparing targeted likely bioenergy crop emission profiles has not yet been performed. This comparison is important, since the differences between the highest and lowest emitting candidate crops are large: eucalyptus and other woody energy crops rank among the highest of known BVOC emitters (Guenther et al 1994, Street et al 1997, Padhy and Varshney 2005, Hewitt et al 1990, Winters et al 2009, Owen 2001, while rapidly growing cellulosic alternatives such as switchgrass (Panicum sp.) and Miscanthus (Miscanthus x Giganteus) are among the lowest (Graus et al 2011, Eller et al 2011. Since the conversion of existing fossil-fuel based energy sources to these modern bioenergy cropping systems will presumably be driven by climate and air quality concerns, choosing feedstocks that do not undermine these efforts through drastic changes in biogenic emissions will be critical. As modern bioenergy methods continue to be developed and applied, the regional selection of feedstock crop will naturally be based on a number of agronomic criteria, including growth rates, hardiness, ease of cultivation and harvesting, and suitability for efficient conversion to energy. It is the purpose of this paper to assess whether biogenic emissions must also be added to emerging feedstock crop-selection criteria in areas demonstrating high air quality sensitivities.

Methodology
In this work, the air quality and climatic impacts of increased biogenic emissions due to a global transition to bioenergy crop cultivation are examined, and considerations for reducing these impacts are proposed and evaluated. To compare these impacts, we model the climate and air quality consequences of large-scale conversion of underutilized land to bioenergy crop cultivation using the NCAR chemistry climate model Community Earth System Model (CESM) 1.1, including version 4 of the Community Atmosphere Model (Lamarque et al 2012, Neale et al 2013 additional model and crop replacement details available in supplementary data available at stacks.iop. org/ERL/10/054004/mmedia). We use three characteristic replacement crop emission profiles representing a range of emission types: trace isoprene emissions (LOW case), high isoprene emissions (ISO case), and high isoprene and monoterpene emissions (ISO + MT case). We also generate a fourth case using an average of all three emission types to represent an intermediate emissions scenario (MIXED). We then use each of these four profiles to simulate varying degrees of landuse change in areas identified as underutilized (Cai et al 2010) under both current and projected future anthropogenic emissions scenarios , and compare them to base cases using nonemitting land rather than a replacement crop.
We run a total of three base cases and 48 replacement cases (table 1), representing a wide range of total replaced area (143-1430 Mha) and emissions profiles (low to high BVOC emitters) using both current and projected anthropogenic emissions and ambient temperatures. Projections of future total biomass energy production potentials are highly variable, mostly due to uncertainties in quantifying land availability and likely crop yields (Berndes et al 2003). For much of this study, the 'Moderate' replacement scale, representing 25% of the full potential area, or around 277 Mha (supplementary table 1  In addition to comparing pollutant changes between crop profiles and anthropogenic emissions scenarios, we perform an analysis of human-health impacts of alternative cropping scenarios. Using population distribution data and concentration response estimates we estimate the number of lives affected by degraded air quality and assess a health cost on a regional basis (Jerrett et al 2009, Anenberg et al 2010. We then weight changes in O 3 and PM 2.5 by population density and normalize results by replacement scale to generate an aggregate 'energy crop air quality score' (ECAQS) for each crop and region. Following the examination of urban tree selection and air quality effects presented in Donovan et al 2005, the ECAQS score can be understood as the population-weighted average change in air quality (measured relative to WHO standards for O 3 and PM 2.5 ) per 10 Mha of energy crop planted in that region, allowing for a normalized measure of AQ sensitivity to variation in crop type and regional abundance. An ECAQS score of negative ten, for example, indicates that the population will see an average increase in pollutant concentrations equivalent to 10% of their respective standards.

Air quality impacts
Our model results show highly non-linear, regionspecific air quality impacts of bioenergy crop selection (figure 1), demonstrating the need for regional crop selections taking local sensitivities into account. Simulated O 3 generally increase for all three high-emitting crop types, with especially large summertime effects predicted in the following regions: India (average increases of 5-27 ppb population-weighted summertime O 3 for the Moderate replacement cases), China (1-9 ppb O 3 ), and the United States (1-6 ppb O 3 ). However, changes in O 3 levels are low or even negative in areas exhibiting already low NO x /volatile organic compound (VOC) ratios, such as South America, where changes in O 3 range from −1 ppb in the summer to +1 ppb in the winter for the Moderate replacement scale ISO and ISO + MT cases, respectively. Secondary organic aerosol formation in all regions is primarily driven by monoterpene emissions, characteristic of the ISO + MT and MIXED cases. ISO replacement cases show some PM 2.5 response, but at much reduced levels, while the LOW cases show negligible impacts on PM 2.5 concentrations. Unlike O 3 , PM 2.5 levels consistently increase with greater crop replacement area. PM 2.5 levels also show much less sensitivity to future changes in climate and emissions. In China and India maximum population-weighted changes in PM 2.5 approach 2 μg m −3 for the ISO + MT case.
Since the health risks associated with PM 2.5 exceed those of tropospheric O 3 relative to the magnitudes of changes evident in this study, the high monoterpene emissions of the ISO + MT cases result in the greatest estimated health costs among the four crop types, with over 410 000 premature deaths worldwide associated with the Moderate replacement level (see supplementary data for more details on mortality estimate calculations (available at stacks.iop.org/ERL/10/054004/ mmedia)). Using the reduced emissions of the ISOonly profile leads to 328 000 premature deaths, while conversion of land using LOW emissions shows only very small changes in O 3 and PM 2.5 , and therefore negligible increases in premature deaths. Notably, mixing the three crop types (MIXED) proves to be more detrimental than might be expected, with mortality increases almost identical to those of the ISO case worldwide, and greater than the average of the three emissions cases for each individual region. This may be related to nonlinearities in pollutant formation, and therefore in the resulting changes in mortality estimates. The scale of estimated land availability and dense population distributions in India, Africa, and Table 1. Settings used for all 46 simulated cases (above), and changes to population-weighted NO x resulting from modified anthropogenic emissions (below).
The extent to which air quality concerns should be taken into account in the selection of bioenergy crops on a regional basis can be evaluated quantitatively by comparing differences in the ECAQS for each crop and region (table 2). Total scores for ISO + MT emissions at the Moderate replacement scale range from The potential increase in productivity of higheremitting crops when compared to low-emitting crops such as miscanthus or switchgrass may be justifiable when the population-weighted impacts of the higher emissions are relatively low, as in Africa and South America. On the other hand, the larger impacts shown in the other four regions suggest that finding lowemitting solutions may be crucial for any region-wide implementation of a bioenergy cultivation strategy. Furthermore, comparing the ISO + MT and ISO cases highlights the relative importance of monoterpene and isoprene emissions in each region. While PM 2.5  impacts are larger for the ISO + MT case in all regions, in some areas the health effects of increased O 3 may outweigh those of PM 2.5 , reducing the difference between isoprene-only emitters such as giant reed and monoterpene emitters such as eucalyptus. The combined health benefits of using lower-emitting crops for bioenergy production is evident in figure 2, where summed impacts of O 3 and PM 2.5 changes for the ISO + MT case are shown in red, with regional benefits of lower-emitting options (in deaths prevented per energy produced) shown as inset bar plots.

Climatic impacts
In addition to local air quality effects, a worldwide shift towards high VOC-emitting bioenergy crop cultivation would have global climatological impacts, including possible changes to methane (CH 4 ) lifetimes and the aerosol radiative effect. The increase in total organic aerosol burden would have complex effects on the net energy budget of the atmosphere, affecting not only incoming radiation directly, but also a variety of cloud properties such as albedo and lifetime. Greater levels of PM 2.5 are predicted for most replacement scenarios, contributing an overall cooling effect. Taken together, the increased CH 4 lifetime and aerosol burden resulting from enhanced BVOC emissions would be expected to have competing effects on radiative forcing, with the CH 4 effect greater by a factor of approximately five in these simulations. Since the reduction of radiative forcing from greenhouse emissions is one of the main goals driving the present move towards bioenergy development and expansion, taking such BVOC related feedbacks into account will be important when assessing bioenergy crop selection and expansion from a global perspective.

Total costs
Finally, we perform economic assessments of climatic and human-health impacts using estimates for the social cost of carbon, and value of a statistical life. Literature values for both of these costs vary greatly, depending heavily upon future economic and atmospheric assumptions. For this work we use conservative, low-end estimates of $21 per ton CO 2 (IAWG 2010) and $1 million per statistical life (Viscusi and Aldy 2003). We then convert the climatic and human health costs associated with each modeled crop type into equivalent CO 2 (CO 2 e) and compare them to estimated benefits of coal combustion reductions (Burnham et al 2012), to produce a comparison of costs and benefits for each crop emission type (figure 3). Costs associated with increased CH 4 (minus the cooling effects of increased organic aerosol) are approximately 8-10% of the expected value of reduced fossil fuel emissions for ISO + MT, ISO, and MIXED emission cases, and negligible for the LOW case. Globally summed health costs are larger, totaling 24-45% of the dollar value of reduced fossil fuel emissions for the high BVOC-emitting cases. These costs make the reduction in negative climate and human health impacts associated with a low-emitting bioenergy crop like switchgrass especially attractive. Under these assumptions, high-emitting crops would dramatically mitigate any net benefits of greenhouse gas emission reductions, primarily through impacts on human health.

Land-use change assumptions and uncertainties
In simplifying and evaluating the impacts of largescale global trends as we have done here, important assumptions must be made. For one, we use a single land-availability map to determine replacement areas worldwide, scaling each grid cell down uniformly for reduced area cases. In practice, bioenergy crops will probably not be distributed in this manner, and are more likely to be clustered in dense areas around processing facilities. This difference in distribution and density would likely affect the resulting air quality impacts. Likewise, land-use change consequences will be highly dependent on the original land being replaced, a sensitivity that remains unexplored in our study.
While we have chosen values representative of typical candidate crops, the specific phenology, yields, and cultivation needs of selected feedstocks will also vary greatly between species and regions. For this reason, the comparisons presented here should be considered emblematic of the wide-ranging bioenergy crop options available, rather than evaluations of specific crops.

Summary
As alternatives to fossil fuels are increasingly adopted, it will be crucial to do so in ways that seek to maximize carbon reduction while also minimizing associated negative consequences. The results of this study suggest that, while cultivation of high productivity bioenergy crops may offer an attractive option for biomass-based energy production in many areas of the world, air quality and climate impacts of crop emissions must also be considered alongside other concerns such as water availability, biodiversity, and food security. In terms of biogenic emissions, using a lowemitting crop such as switchgrass or Miscanthus rather than a higher-emitting alternative could maximize any expected environmental benefits of reduced fossil fuel use. However, the possibility of lower potential yields of these crops compared to higher VOC-emitting options may affect the final costbenefit analysis of any proposed bioenergy development. By evaluating the local and global consequences of crop selection on a case-by-case basis, including calculation of ECAQS values for candidate crops as performed here, these considerations can help to further inform energy policy and maximize the benefits of alternative energy production while minimizing the negative impacts of bioenergy crop cultivation on air quality, human health, and climate.