Life cycle assessment of native plants and marginal lands for bioenergy agriculture in Kentucky as a

Life cycle assessment of native plants and marginal lands for bioenergy agriculture in Kentucky as a model for south-eastern USA S E T H D E B O L T *, J . E L L I O T T C A M P B E L L w , R AY S M I T H J R . z, M I C H A E L M O N T R O S S § and J O Z S E F S T O R K * *Department of Horticulture, N-318 Agricultural Science Center, University of Kentucky, Lexington, KT 40546-0091, USA, wCollege of Engineering University of California, Merced, CA 40546-0091, USA, zDepartment of Agronomy, N-200 Agriculture Science Center North, University of Kentucky, Lexington, KT 40546-0091, USA, §Department of Biosystems and Agricultural Engineering, Barnhardt Building, University of Kentucky, Lexington, KT 40546-0091, USA Abstract The Brookings Institute analysis rate both Lexington and Louisville, Kentucky (USA) as two of the nation’s largest carbon emitters. This high carbon footprint is largely due to the fact that 95% of electricity is produced from coal. Kentucky has limited options for electric power production from low carbon sources such as solar, wind, geothermal, and hydroelectric. Other states (TN, IN, OH, WV, and IL) in this region are similarly limited in renewable energy capacity. Bioenergy agriculture could account for a proportion of renewable energy needs, but to what extent is unclear. Herein, we found that abandoned agricultural land, not including land that is in fallow or crop rotation, aquatic ecosys- tems, nor plant-life that had passed through secondary ecological succession totaled 1.9 Mha and abandoned mine-land totaled 0.3 Mha, which combined accounted for 21% of Kentucky’s land mass. A life cycle assessment was performed based on local yield and agronomic data for native grass bioenergy agriculture. These data showed that utilizing Kentucky’s marginal land to grow native C4 grasses for cellulosic ethanol and bioelec- tricity may account for up to 13.3% and 17.2% of the states 2 trillion MJ energy consumption and reduce green house gas emissions by 68% relative to gasoline. Keywords: biofuel, cell walls, cellulosic biofuel, feedstock, GHG (green house gases), lignification Received 20 March 2009; revised version received 8 June 2009 and accepted 1 July 2009 Introduction Decreasing reliance on fossil energy will inevitably result in a shift in economic viability of related indus- tries, and a growth of localized energy economies with the potential to revitalize rural communities. For these reasons and in an effort to realize energy independence, public opinion, and legislation will continue to place an emphasis on reducing fossil energy consumption and switching to renewable forms of energy to stem carbon emissions. Reports from organizations such as the Brookings Institute (2005) that rate Lexington and Louisville (Kentucky) as the nation’s largest carbon emitters will pressure states like Kentucky where an extremely high carbon footprint is largely due to the fact that 95% of our electricity is produced from coal (Energy Information Administration (EIA, 2007). Other states (TN, IN, OH, WV, and IL) in this region also derive a substantial amount of electricity from coal. As with many states in this region, Kentucky has limited options for electric power production from low carbon sources such as solar thermal, photovoltaic solar, wind, geothermal, and hydroelectric. Hence, bioenergy agri- culture may be an avenue worth pursuing to reach renewable energy goals. Recent assessments of the ‘carbon cost’ from land use change (Searchinger et al., 2008) suggest that cutting forestland (3.5 Mha) to grow bioenergy crops will accelerate climate change by emit- ting carbon currently sequestered within plant matter and soil (Pin˜eiro et al., 2009). Furthermore, land use change from agricultural land used for food production converted into bioenergy agriculture poses a significant threat to global food security (Boddiger, 2007). Utilizing abandoned agricultural and mine lands for bioenergyCorrespondence: Seth DeBolt, e-mail: sdebo2@email.uky.edu GCB Bioenergy (2009) 1, 308–316, doi: 10.1111/j.1757-1707.2009.01023.x 308 r 2009 Blackwell Publishing Ltd agriculture could overcome both problems and Ken- tucky and many states in the southeast have high abandonment rates (Campbell et al., 2008). However, before this study, it was unclear how much land was marginal, defined as previously used for mining or agriculture and moreover has not yet gone through secondary ecological succession, which is then classi- fied as forestland. Life cycle assessment (LCA) and bioenergy potential analysis are currently needed to assess regional potential for renewable forms of energy using marginal land. Failing to adequately calculate the cost of land use change, resulting in increased carbon emissions, will pose significant risk to scientific validity and public perception of policy level decision making on the bioenergy issue. Analyses show that marginal land used for bioenergy agriculture could account for up to 8% of the world’s energy from biomass (Campbell et al., 2008). LCA must also take into account lower expected yield potentials on marginal land relative to agricultural land (Tilman & Downing, 1994; Haberl et al., 2007). But, utilizing mar- ginal agricultural, abandoned or reclaimed mine land, and woodland areas has unique challenges related to establishing, growing, harvesting, and transporting the crops from the land to an end user. The assessments made herein are based on cellulosic feedstock rather than corn ethanol, although we note that cellulosic feedstocks will also undermine food security if they are grown on prime agriculture lands and therefore focus on marginal land. To begin moving in the direc- tion of resolving these issues, this project addresses a life cycle roadmap for how bioenergy agriculture could fulfill a component of renewable energy needs and what the cost of production and consumption are relative to current fossil energy. Materials and methods Data sources Net energy balance of prairie biomass was generated based on data derived from various sources and are summarized in supplemental Table 1 (The Greenhouse Gases, Regulated Emissions, and Energy Use in Trans- portation (GREET Wang et al., 2007), Tilman & Down- ing, 1994; Farrell et al., 2006; Tilman et al., 2006; Haberl et al., 2007, Energy Information Administration EIA, Kentucky State Energy Use Table 2007, Kentucky Mine Mapping Information System 2008, Kentucky State Abandoned Mine Report 2008, and a 7 year collection of crop yield and input data collected for multiple entries of three different native perennial grass species at Spindetop Research farm in Lexington, Kentucky 38.0511N, 84.0611W (elevation 981 ft) (agronomic analy- sis: Stork et al., 2009). Species used were multiple entries of eastern gamagrass (EG) (Trispicum dactyloides L.), switchgrass (SW) (Panicum virgatum), and big bluestem (BB) (Andropogon gerardii Vitman). Where possible, this regionally collected and verified data was preferentially used for LCA and modeling. Briefly, these plots were randomized in a block design with four replications and 4.5� 1.5 m2 (with 1.5 m perimeter comprising each en- try) sites were established in the year 2000 into a well- drained Maury silt loam soil (fine mixed, semiactive, mesic Typic Paleudalfs). Soil analysis. Soil analysis data was used to determine what mine lands might be contaminated with that could affect yield of energy crops species and to estimate required input needs for the natural production model on a site, which reflects local constraints on abandoned land, climate, and soil types. Values for soil N, P, K, Zn, Cd, Mg, Ca, Pb, Mo, Ni, and Cr were obtained from independent soil analyses from mine sites and abandoned agricultural lands in Kentucky. Methods for analysis were derived from published sources (Soil and Plant Analysis Council, 2000; Sikora, 2005 and Sikora, 2006) Soil was oven-dried at 38 1C and ground to pass a 2 mm screen. A soil–water paste is created by adding 10 mL of water to 10 cm3 of soil and stirring for with a glass rod and letting stand for at least 15 min but not 42 h. A glass electrode is placed in the mixture to measure pH. After pH measurement, 10 mL of Sikora Buffer (a mixture of triethanoloamine, imidazole, MES, acetic acid, and KCl) was added to the soil–water paste and shaken for 10 min. A glass electrode was then placed in the mixture to measure buffer pH within 2 h after shaking. Soil pH and buffer pH were reported as unitless values. Phosphorus, K, Ca, Mg, and Zn are determined in a Mehlich III extract which contains 0.2 N acetic acid, 0.25 N NH4NO3, 0.015 N NH4F, 0.013 N HNO3, and 0.001 N EDTA. Twenty milliliters of Mehlich III extract is added to 2 cm3 soil, shaken for 5 min, and immediately filtered through Whatman #2 filter paper. Filtration was terminated at the end of 10 min. The filtrate was analyzed via inductively coupled plasma spectroscopy. Regional determination of abandoned mine and agricultural lands To calculate the abandoned lands in Kentucky, we analyzed historical data records to identify land use change that had been abandoned from use in agricul- ture but that had not transitioned to secondary forests, urban areas, aquatic ecosystems such as rivers, streams, and wetlands, or transit infrastructure by methods described by Campbell et al. (2008). Mine-lands that N A T I V E G R A S S B I O E N E R G Y A S S E S S M E N T 309 r 2009 Blackwell Publishing Ltd, GCB Bioenergy, 1, 308–316 were in production and are now in fallow (use of the term fallow is used rather than abandoned because mines have exhausted current economic viability) were estimated based on values for primarily coal and other minor mining activities such as copper, iron, and phos- phorous, that are defined in the Kentucky State Aban- doned Mine Report for coal (2008) and the Kentucky Mine Mapping Information System (2008), and Ken- tucky Geological Survey (2007) annual reports for ex- mining enterprise land use shifts other than coal. Net energy balance of native perennial grass biomass Agricultural Phase. Energy inputs were for seeding, growing, harvesting, and transporting perennial native grass biomass were calculated using the ‘cellulosic’ bioenergy agriculture spreadsheet presented by Farrell et al. (2006) with values derived from Graboski (2002). Fertilizer Application. Each plot was fertilized with 67 kg ha�1 nitrogen each spring when plant height reached 10–20 cm and P and K were maintained at 336 kg ha�1 based on annual soil analysis. Similar to the values defined by Tilman et al. (2006), we essentially replaced phosphorus, which constitutes 0.2% of the mass of dry biomass annually harvested, hence approximately 8 kg ha�1 yr�1 on abandoned lands was estimated. Productivity (yield) loss in degraded land. Life cycle analysis estimate of input and output values generated were mixed between the Stork et al. (2009) study site and those defined in the Farrell et al. (2006) model and are presented in their raw form in a supplemental online Table (S1). Loss of yield on abandoned land was previously calculated as 60–65% of the managed system (Tilman et al., 2006; Haberl et al., 2007). The Haberl paper suggests that on average (global) the existing agriculture primary production is 65% of what the natural production would be on the same land. Determining the ratio of yields on marginal lands relative to fertile agricultural land contains many Table 1 Cellulosic ethanol energy production assessment for bioenergy agriculture produced and its contribution to Kentucky’s current energy use requirements EG BB SW Annual biomass yield potential for Kentucky Yeild potential for each grass species (kg ha�1) 14455 10192 12356 Estimated abandoned land yield (65%) (kg ha�1) 9396 6625 8031 Abandoned mine lands (million ha) 0.3 0.3 0.3 Abandoned ag. (million ha) 1.9 1.9 1.9 Total abandoned land (million ha) 2.2 2.2 2.2 Total arable land (million ha) 5.6 5.6 5.6 Annual yield on abandoned mine land (kg yr�1) 2.81� 109 1.99� 109 2.41� 109 Annual yield on abandoned agricultural land (kg yr�1) 1.79� 1010 1.26� 1010 1.53� 1010 Annual yield on total abandoned land (kg yr�1) 2.07� 1010 1.46� 1010 1.77� 1010 Current annual energy usage in Kentucky (MJ) Residential usage (MJ) 3.64� 1011 3.64� 1011 3.64� 1011 Commercial usage (MJ) 2.62� 1011 2.62� 1011 2.62� 1011 Industrial usage (MJ) 9.54� 1011 9.54� 1011 9.54� 1011 Transportation usage (MJ) 5.00� 1011 5.00� 1011 5.00� 1011 Total energy usage (MJ) 2.08� 1012 2.08� 1012 2.08� 1012 Biomass energy yield produced annually compared to usage (MJ) Average LCA for native grass biomass produced on marginal land (NEV MJ L�1) 22.1 21.3 21.6 Energy produced from 1 ha biomass crop (MJ) 1.26� 105 8.81� 104 1.10� 105 Energy produced from biomass grown on total abandoned land (MJ) 2.76� 1011 1.93� 1011 2.42� 1011 Energy produced from biomass grown on total arable land (MJ) 9.84� 1011 6.90� 1011 8.63� 1011 Usage – cellulosic ethanol potential on abandoned land (MJ) 1.80� 1012 1.89� 1012 1.84� 1012 Usage – cellulosic ethanol potential on arable land (MJ) 1.10� 1012 1.39� 1012 1.22� 1012 Potential for bioenergy agriculture to fullfill energy requirements in Kentucky (%) Total MJ abandoned land (%) 13.3 9.3 11.7 Total MJ arable land (%) 59.2 41.5 51.9 Yield determinations, agronomic inputs and net energy values (NEV) were generated using data from Stork et al. (2009), GREET and EBAMM, land use estimates from Fig. 1 and energy usage from EIA (2007). 310 S . D E B O L T et al. r 2009 Blackwell Publishing Ltd, GCB Bioenergy, 1, 308–316 ancillary factors that will influence the measure and currently little data is available on this topic. Tilman et al. (2006) used an approach, which reported biomass yield energy of 68 and 111 GJ�1 ha yr�1 on degraded and fertile lands, respectively. This report may be considered a conservative approach, (tending to underestimate yields on marginal lands) because these lands were highly degraded but many of the abandoned agriculture lands might have capacity to have greater yields. Nonetheless, this value provided a 60% degradation rate. Since Kentucky had only 0.3 Mha of marginal mining land compared with 1.9 Mha of marginal agricultural land, our relative estimation of the degradation value was 65% more consistent with Haberl. Biomass Conversion to Energy. Two scenarios for utilizing native grass biomass for energy were modeled. Biorefinery energy yield potential for the production of bioethanol was performed using LCA for net energy value (NEV) resulting from cellulosic ethanol production via the Farrell et al. (2006) estimates, which proposed virtually 100% cellulose conversion. Our own experimental conversion values (without pretreatment) were up to 20% conversion potential (Stork et al., 2009), but pretreatment, albeit energetically costly are available that loosen the lignin–cellulose interaction and subsequently reduce the recalcitrance to enzymatic hydrolysis. Energy yield of biomass conversion to bioelectricity (via co-firing with coal at 0.5% decrease in overall efficiency assuming a 95% coal/5% biomass blend as compared with 100% coal (Yoshitaka, 2005) was generated herein using values obtained for SW (P. virgatum L.) pellets that had an energy yield of 18.8 MJ kg�1 (Samson et al., 2008). To calculate the LCA for bioelectricity, we used the Farrell et al. (2006) data for the agriculture phase and transportation energy inputs and Stork et al. (2009) data for the yield calculation. The co-firing processing energy inputs were calculated using Mani et al. (2004) data for grinding and Jannasch et al. (2002) data for pelleting and further processing costs were then eliminated from the Farrell et al. (2006) spreadsheet (Table S1). After energy cost data was obtained for grinding, pelleting, and agricultural phase this was subtracted from the energy produced on a per kilogram basis. In order to obtain potential for Kentucky, yield data was multiplied by per kilogram energy and total arable as well as abandoned agricultural and mining land. The resulting NEV for bioelectricity was multiplied by crop yield for abandoned land as per above and was then compared with Kentucky’s total energy consumption to reach a statewide estimation of bioelectricity potential. Calculation of net energy potential towards fulfiling regional energy needs. Biomass total was calculated based on the 7-year yield average on a per hectare basis (Stork et al., 2009) and then multiplied by the area of abandoned land within the state of Kentucky. NEV resulting from cellulosic bioenergy was calculated by multiplying the total biomass potential for Kentucky abandoned agricultural and mine land by the NEV obtained in the LCA. Energy usage for commercial, residential, industrial, and transport sectors for Kentucky were obtained from the United States Department of Energy, Energy Information Administration report of Kentucky energy use per sector (2007): energy use per sector and the total energy use was compared with the total NEV for various native perennial grass bioenergy- agriculture systems to provide a relative energy supply to energy demands output. Greenhouse gas (GHG) costs of biomass energy relative to fossil energy. We considered the total life cycle GHG savings from producing and using biomass to generate biofuels and electricity (Table S1). GHG savings results both from displacing fossil fuels and from the net GHG sequestration. To estimate net GHG savings, we subtract from this amount the total life cycle GHG release from the fossil fuels used to produce prairie biomass and transport it to its point of end use. The soil carbon levels and N2O emissions are critical to the LCA, and we have clarified our approach to quantifying these components of the life cycle. Soil carbon levels are difficult to quantify and data is scarce for marginal lands. For our analysis, we assume a soil carbon loss [emission of carbon dioxide (CO2) to the atmosphere] of 0.48 � 0.44 Mg CO2 ha�1 yr�1 and N2O emissions rate of 7.0 kg CO2e kg N �1 from previous work on monocultures planted on degraded and abandoned agriculture lands (Tilman et al., 2006). Results Soil analysis Reclaimed mine land may be contaminated leading to decreased yield in energy crops. The majority of re- claimed mine land in Kentucky was ex-coal mining enterprise. Therefore, a recently filled mountain top removal coalmine in Pike County, Kentucky (latitude: 3713401900N; longitude: 8214405600W) was selected as representative and soil samples were obtained and analyzed. These data showed that Meh3 Cd5 0.04 mg kg�1, Meh3 Cr5 0.15 mg kg�1, Meh3 Ni5 1.94 mg kg�1, Meh3 Pb5 1.63 mg kg�1, Meh3 Zn5 2.95 mg kg�1, Meh3 Cu5 1.64 mg kg�1, Meh3 Mo5o0.1 mg kg�1. The pH of the soil was 6.92, phosphorous 7.85 kg ha�1, N AT I V E G R A S S B I O E N E R G Y A S S E S S M E N T 311 r 2009 Blackwell Publishing Ltd, GCB Bioenergy, 1, 308–316 potassium 91 kg ha�1, calcium 1185 kg ha�1, magnesium 265 kg ha�1, and Zn levels were 5.7 kg ha�1. Current land use in Kentucky and land abandonment rates Kentucky’s total land is 10.3 million hectares (Mha), with 5.6 Mha classified as farmland. Of that, only 2.1 Mha are harvested with the other 3.5 Mha of farm- land in pasture, rangeland, or woodlands. In addition, 3.5 Mha of forestland is available with most of the land in private ownership. Analysis of land use showed that abandoned agricultural land totaled 1.9 Mha (Fig. 1a) and this was combined with 0.3 Mha of abandoned mine land, of which 0.266 Mha was a direct result of coal mines (Abandoned Mine Report, 2008) and the remainder a mix of mining activit