The Breakeven Costs of Producing Alternative Feedstocks for Cellulosic Biofuels

Transcription

1 The Breakeven Costs of Producing Alternative Feedstocks for Cellulosic Biofuels Madhu Khanna and Haixiao Huang Energy Biosciences Institute, University of Illnois, Urbana-Champaign There is growing interest in bioenergy as a renewable alternative to fossil fuels being used for heat, power and transportation. Large scale displacement of fossil fuels requires significant production of biomass that in the long run, needs to be economically, environmentally and socially sustainable. Biomass can be obtained from several sources, crop residues, perennial grasses and short rotation trees, such as poplar and willow. The production of biomass will place demands on land with consequences for farm income, production of other crops for food and feed and for the ecosystem services provided by land. To be economically sustainable, biomass must yield an income for a landowner that covers the cost of producing them. In the case of energy crops and woody feedstocks, these costs should include not only the cost of growing these crops but also the foregone returns from alternative uses of that land. One indicator of social sustainability is the extent to which the production of these crops is likely to displace the production of food and feed and affects crop prices. The potential for growing these crops on land that is currently idle/marginal and not productive for food/feed crops makes energy crops an attractive source of feedstock for bioenergy. Additionally, higher yielding energy crops will require less diversion of land from the provision of food/feed or environmental services. The implications of biomass production for the environment will depend on the feedstocks used, the types of land on which they are grown and their implications for biodiversity, soil quality and water quality. The focus of this study is to examine the profitability of various sources of biomass feedstocks under alternative agronomic and economic scenarios. We examine this profitability in three regions of the US that differ in their climatic conditions and therefore in the yields and relative profitability of these alternative sources of feedstock. The three regions considered are: Michigan, Illinois and Oklahoma. These regions differ in the extent to which they are suitable for various crops and therefore in the trade-offs that farmers are likely to face as they decide to switch from a conventional crop to biomass production. Our analysis has implications for the extent to which government subsidies are likely to be required to induce the production of particular feedstocks because market prices based solely on their energy content of those feedstocks may not be sufficient to make them economically viable. We consider five sources of biomass feedstock: corn stover, miscanthus, switchgrass, native prairie grasses and poplar. The amount of corn stover that can be sustainably harvested depends on the tillage and rotation practices used to grow the corn and has implications for the cost of corn stover; therefore we distinguish between corn stover harvested from continuous corn and from corn-soybean rotations and from corn grown using conventional tillage or no-till practices. In the case of perennial grasses and poplar, the cost of growing them must include the opportunity cost of the land, since landowners will allocate land to them only if they provide a return that is at least equal to the existing returns from that land. Landowners have a choice of growing them on prime cropland or on marginal land that may currently be under pasture or fallow and could be brought back into crop production if profitable to do so. If marginal land is defined to include the National Agricultural Statistics Service (NASS 2009a) categories of idle 1

2 land and cropland pasture then there are 1.01, 0.81 and 3.67 million acres available in Illinois, Michigan and Oklahoma, respectively. Research is ongoing on the productivity of energy crops on marginal, low quality land. In the event that the yields of these energy crops are similar on marginal land and average cropland then it will be economically viable for farmers to first use their marginal land to grow them because it has a lower opportunity cost. However, lack of availability of marginal land that is easily accessible, un-fragmented and not subject to environmental regulations that prevent farming (e.g. land under the Conservation Reserve Program) could require the cultivation of energy crops on land currently under conventional crops. We, therefore, consider the costs and profitability of energy crops on both marginal and average cropland in these states. The yields of the feedstocks considered here differ across regions due to differences in climatic conditions, soil moisture and soil quality. With the exception of miscanthus, where only one particular variety (Miscanthus giganteus) is currently being evaluated, there are several varieties of switchgrass, mixed prairie grasses and poplar that are currently being grown. The variety best suited for a particular location will depend on the soil and climate characteristics of that location and a variety s resistance to insects, disease, frost and extreme weather. Perennial grasses senesce in the Fall and translocate their nutrients to their roots; these nutrients are then available for plant growth in the Spring. Harvesting of perennials is best delayed until after senescence so that removal of biomass does not remove phosphorus, nitrogen and potassium from the soil. Miscanthus requires frost conditions to senesce and therefore it is not well suited to the southern regions near the gulf coast where there is no frost; switchgrass is expected to perform better in this region. Miscanthus is also less cold tolerant that switchgrass and is therefore expected to have lower yields than switchgrass in the northern regions of the US such as Minnesota and Wisconsin. Higher growing degree days (warm temperatures spread over more days) and soil moisture contribute to higher biomass yields. However, the extent to which yield of an energy crop will vary across locations differs across different perennials. Yields of corn stover also differ across regions due to differences in soil quality and temperatures which lead to large differences in corn grain yields across regions in the US. This report presents crop budgets and comparative breakeven analysis for eight potential biomass production systems in each of Illinois, Michigan and Oklahoma and highlights key parameters that drive their potential profitability. We determine the breakeven price of biomass feedstocks at the farm-gate on marginal land and on land currently under the most profitable cropping system in that state: rotation corn with no-till in Michigan and Oklahoma and cornsoybean rotation with conventional tillage in Illinois. Herbaceous Feedstocks for Biomass Background Corn Stover: Corn stover yields are closely related to the yield of corn grain. A grain-to-residue ratio of 1:1 for the amount of dry matter of crop grain to dry matter of crop residues (with 15% moisture) is consistently found in the literature (Sheehan et al., 2003; Wilcke and Wyatt, 2002; Graham et al., 2007). Due to the value of biomass returned to the soil as soil organic matter and protection against wind and water erosion, a fraction of the biomass is usually left on-field. Recommended stover removal rates depend on soil characteristics, climate, management 2

3 practices (tillage), and other factors that determine the loss of soil organic matter and run-off. Assumed rates of removal range from 38% to 70%. Malcolm (2008) estimates that 50% of the residue can be removed from fields if no-till or conservation tillage is practiced and 30% can be removed if till or conventional tillage is used. Using these removal rates and 2007 crop yields for corn in the study states, under a monoculture corn system with no till, the average delivered yield at the farm-gate (after accounting for storage losses) for corn stover is the highest in Illinois at 3.41 metric tons of dry matter per hectare (Mg DM/ha), followed by Michigan at 2.69 Mg DM/ha, and Oklahoma at 1.84 Mg DM/ha. The corresponding figures under conventional tillage are 2.05 Mg DM/ha for Illinois, 1.61 Mg DM/ha for Michigan, and 1.10 Mg DM/ha for Oklahoma. The collection of corn stover has to be supplemented with additional fertilizer application to replace the loss of nutrients and soil organic matter due to removal of the crop residues from the soil. Higher rates of stover removal will therefore require higher costs of nutrient applications. They also increase the likely loss in soil organic matter; some studies indicate that the harvest of even 25% of corn stover could reduce soil organic carbon by 3-8 metric tons per hectare in the top 30 cm within the first few years (Anderson Teixeira et al., 2009). Miscanthus: Miscanthus (Miscanthus giganteus) is a perennial rhizomatous grass with the potential for high yields, low input requirements and several environmental benefits. This variety of miscanthus is a sterile, triploid clone of the species, Miscanthus giganteus, with a life span of 15 to 20 years. Miscanthus giganteus is a cross between two different miscanthus species, sinensis and sacchariflorus, and has three sets of chromosomes instead of the normal two. This prevents the normal pairing of chromosomes needed to form fertile pollen and ovules and makes it sterile. It has been grown in the European Union on a very large scale for over 20 years with no evidence of becoming invasive. Miscanthus has high efficiency of converting solar radiation to biomass and in using nutrients and water, and has good pest and disease resistance. It is planted using rhizomes and field trials indicate that miscanthus has the potential for relatively high yields in the rainfed regions of the US. It has been observed to have life-times of 14 to 18 years, during which the first two years are the establishment phase. Miscanthus biomass accumulation normally achieves the maximum potential level between August and October, following which the plant senesces and translocates nitrogen and other nutrients out of the foliage into the roots. Studies in Europe have shown that miscanthus does not respond to N fertilization using annual application rates from 0 to 60 kg/ha (Jain et al., 2010; James et al., 2010). Translocation of inputs prior to senescence of the above ground tissues of miscanthus reduces overall nutrient use, ash content and moisture content while improving the suitability of the biomass as a fuel for combustion. Similarly, field trials have not found a strong response of miscanthus to applications of potassium (K), phosphorus (P), or calcium. However, harvest is expected to remove some nutrients from the crop ecosystem and the long run nutrient requirements of the soil to maintain fertility are still unknown. Research also suggests that miscanthus could be capable of hosting N-fixing bacteria which enable it to meet its annual N requirement (Davis et al., 2010). Some studies, however, include applications of N, P, K, and lime to replenish soil reserves, especially on soils with lower fertility where application rates are assumed to replace nutrients removed that depend on yield. 3

4 Simulations using a crop productivity model, MISCANMOD, show that the farm-gate biomass yield of miscanthus is high in the Atlantic states and low in the western states due to insufficient soil moisture. Furthermore, southern states have higher yields for miscanthus as compared to northern states, in general. Yields vary across locations due to differences in climate variables. The average delivered yields and standard deviation (SD) of miscanthus at the farmgate (after accounting for storage losses) are the highest in Oklahoma at ± 5.59 Mg DM/ha followed by Illinois at ± 3.88 Mg DM/ha and Michigan at ± 4.46 Mg DM/ha. Switchgrass: Switchgrass is a warm season grass that grows primarily in the summer months June to August and like miscanthus has relatively high efficiency of converting solar radiation to biomass and in using nutrients and water, and have good pest and disease resistance. It is planted using seeds and has a stand life of ten years or more where production during the first year or two could be only a fraction of the production achieved during the remaining production years. Yields vary considerably across the varieties of switchgrass; the Cave-in-Rock switchgrass cultivar is an upland variety that originated in Southern Illinois and is cold-tolerant and wellsuited for the upper Midwest. Lowland varieties of switchgrass, like Alamo, are most suited for southern US and typically have yields that are about 50% higher than the yield of the upland variety. Analysis of data from field trials across the US shows that frequency distributions of yield for the upland and lowland varieties were unimodal, with mean (±SD) biomass yields of 8.7 ± 4.2 and 12.9 ± 5.9 MT DM/ha for the two varieties, respectively. Yields for single harvest in plot trials in Oklahoma are reported to range from 8.65 ± 1.57 MT DM/ha to ± 1.68 MT DM/ha depending on nitrogen applications (Aravindhakshan et al. 2011), while yields with commercial scale production range between 3.4 MT DM/ha in Iowa (Brummer et al., 2000) to 7 MT DM/ha in the Upper Plain states (Perrin et al., 2008). De La Torre Ugarte et al. (2003) and Brechbill et al. (2008) obtain model simulated yields of 11.2 MT DM/ha. James et al. (2010) consider yields to range between 6-17 MT DM/ha in Michigan. Field studies indicate that the uptake of nitrogen by switchgrass is half of levels achieved by miscanthus and that N-fixing bacteria do not contribute substantially to the annual N requirements of the plant (Davis et al., 2010). Unlike miscanthus, switchgrass yields have been found to be responsive to nitrogen applications with nitrogen requirements varying with sitespecific conditions. However, trials conducted across the US have not found a positive response of switchgrass yield to applications of P, K and calcium. The yields of switchgrass could be about half as large as that of miscanthus at most locations. The MISCANMOD model simulated average delivered yields and SDs at the farmgate (after accounting for storage losses) in Illinois, Michigan and Oklahoma are ± 1.23, 7.53 ± 0.58 and 9.70 ± 1.53 MT DM/ha, respectively. These simulated switchgrass yields are within the range reported in literature discussed above. Mixed Prairie Grasses: There are more than hundreds of species of grasses native to the US prairies. Mixed prairie systems may be planted to increase biodiversity and improve soil structure. The mixed prairie grasses typically consist of various species with different plant functional types such as C4 grass, forb, and legume. Commonly planted prairie grass species in 4

5 the study regions include, just to name a few, Bermudagrass, a long-lived warm season perennial that spreads by rhizome and seed; Flaccidgrass, an upright, tall, weak bunch type perennial rhizomatous subtropical, warm-season forage grass; Weeping lovegrass, a warm-season bunchgrass characterized by quick germination, an active growth period in the summer, high drought tolerance, production of thick mass of vegetative soil cover, and a deep penetrating root system; Big bluestem, a perennial warm-season grass dominant in the Midwestern tallgrass prairie; Indian grass, a native, warm-season grass with endurance of a wide range of weather extremes and ease of establishment from seed; and Showy tick trefoil, a tall native perennial warm-season legume used as a small component in a seeding mixture for prairie restoration. A field experiment by Tilman et al. (2006) in Minnesota showed that plots with 16 grassland species (low-input, high diversity or LIHD) achieved 238% more bioenergy (measured as biomass times energy release upon combustion) per hectare than monoculture switchgrass on highly degraded soil with no fertilization. Adler et al. (2009) find that the biomass yield per unit land decreased with greater plant species richness and the composition of the resulting biomass also led to a reduction in biofuel yield per unit biomass. The inclusion of tall, native C4 prairie grasses that are highly competitive and efficient users of inputs and legumes for nitrogen fixation are critical for biomass productivity with low input applications (Adler et al., 2009). Yields of prairie grasses also vary significantly across location and species and with the application rate of fertilizer and harvesting times in a year. According to the database of reported yields in the literature compiled at the University of Illinois (Dietze, 2011), harvested yields of specific prairie grasses in the US ranged from about 1.5 Mg DM/ha for cool-season grass and legume pastures in southwest Michigan (Hudson et al., 2010) to 19.2 Mg DM/ha for indiangrass with a nitrogen application rate of 220 kg N/ha in Iowa (Heggenstaller et al., 2009). However, there is a general paucity of field studies on yields of mixed prairie systems. Tilman et al. (2006) find that plant species composition and diversity have an important effect on the yield of lowinput prairie grass mixtures, with LIHD prairie mixes achieving yields as high as 3.7 and 6.0 Mg/ha/yr on degraded and fertile soils in Minnesota, respectively. On the other hand, Adler et al. (2009) find that high diversity prairie systems may lead to decreased biomass yield based on their field studies at multiple sites in the northeast region of the US, but the authors provide no specific yield estimates for the prairie mixes with different numbers of plant species. The delivered yield of mixed prairie grasses at the farm-gate (after accounting for storage losses) in Illinois is simulated using the DAYCENT model to be 7.02 ( ) Mg/ha/yr (Davis, 2011), while the delivered yield in Michigan of mixed grasses is reported to be 7.53 Mg/ha/yr and that of native prairie grasses is reported to be 4.1 Mg/ha/yr (James et al., 2010). These studies assume no nitrogen application for these grasses. So far no information on yields of mixed grasses in Oklahoma has been found in literature. For individual species, Aravindhashan et al. (2011) report that the harvested yield with single harvest and minimum nitrogen application is 4.95 ± 1.32 Mg/ha/yr for Bermudagrass, 8.4 ± 1.28 Mg/ha/yr for Flaccidgrass, and 5.98 ± 0.9 Mg/ha/yr for Lovegrass in Oklahoma Averaging these individual grass yields and accounting for storage losses, the delivered yield of mixed prairie grasses at the farm-gate in Oklahoma is derived to be 5.89 Mg/ha/yr. 5

6 Harvesting and Storage of Biomass Farm activities after establishment of an energy crop include mowing, raking, baling, and storage. A single annual harvest is found to result in lower costs than two harvests a year for perennial grasses including switchgrass and miscanthus (Aravindhakshan et al., 2010). Delaying harvest of switchgrass and miscanthus until after senescence reduces need for nutrient application in the subsequent year, reduces drying time, and improves the quality of the biomass. Yet, waiting to harvest until after senescence also decreases harvestable yield by 20-40% for miscanthus and 15-20% for switchgrass compared to peak levels in September-October (Jain et al., 2010). The gain in yield and profitability with a single harvest versus two harvests may differ across other prairie grasses; some grasses such as Bermudagrass have been found to have a high after-harvest growth and a fast recovery after a harvest in July at all levels of nitrogen application. Yields of flacidgrass and lovegrass with two harvests are found to be higher than with a single harvest but only with high levels of nitrogen application (Aravindhakshan et al., 2011). The economics of harvesting more than once in the year for mixed prairie systems is yet to be determined. Perennial grasses can be harvested using conventional hay harvesting equipment, though more specialized equipment is currently under development. A short (four month) harvest window requires considerable investment in harvest equipment. One approach to reduce harvesting costs is to extend the harvesting window and thereby spread the fixed costs of the harvest machines over more hectares and reduce the time that the harvested material must be stored. To maintain productivity, additional fertilizer applications would be needed for fields harvested prior to senescence and biomass yields would be lower for fields harvested later (Hwang et al., 2009). The harvesting window could also be extended by having a mix of different feedstocks, assuming a biorefinery can process a variety of feedstocks. Mapemba et al. (2008) consider the possibility of having an extended harvest system from June through February with wheat straw harvested in June and July, corn stover in September and October and perennial grasses from July through the following spring. They found it possible to allocate harvest equipment in a way to reduce the capital investment in harvest machines by 50%. However, results are sensitive to assumptions about harvest days available which depend on the weather. Harvest days also depend on the condition of the soil (e.g. sufficiently dry soil to hold the weight of the harvest equipment) and the moisture content of the grass. A reduction in harvest days due to weather related constraints could limit the flexibility for scheduling harvest equipment optimally and may result in substantially higher harvesting costs. Biomass can be stored after harvest in several ways including on-farm open air, on-farm covered, or storage in a centralized covered facility. Open air storage could be unprotected on the ground or on crushed rock or covered by reusable tarp. The covered storage could be a pole frame structure with open sides on crushed rock or it could be an enclosed structure on crushed rock. The loss in biomass is highest when biomass is left unprotected and lowest in the enclosed structure. These losses depend on the number of days the biomass is stored and need to be weighed against the costs of installation, land, labor, and materials as well as the biomass quality that is needed by the bio-refinery. A centralized covered storage facility could be shared by many farms but would require producers to incur biomass handling and transportation costs to move the biomass from the farm. The optimal choice of storage facility is likely to depend on the volume of biomass and the length of time that it has to be stored, the price of biomass, the 6

7 quality of biomass required, and the weather conditions within the region (Brummer et al., 2000; Duffy, 2007). Woody Biomass Hybrid Poplar: Short rotation woody crops, in particular hybrid poplar and willow, are also being considered for biomass production. Hybrid poplar and willow are planted using cuttings, or scions and have rotations of 6 to 10 years. The majority of poplars planted are either unrooted hardwood cuttings or bareroot stock and planting time is May to early June in the northern US. Weed control in the early years of poplar culture is essential for poplar growth and survival. There are a number of ways to control weeds depending upon the landowner s resources and philosophy. They include hand weeding, cultivation, mowing, cover crops, herbicides and mulching. Poplars planted in large areas can suffer from insect problems; these can be controlled by diversifying the species planted, using insecticides and practicing integrated pest management. Animal browsing is another potential problem that may require investment in fencing. Poplars have a high nutrient requirement to maintain maximum productivity. If nutrients or water are limiting, poplar growth is significantly decreased. The formulation of the fertilizer, the quantity of fertilizer, the timing of fertilization, and the number of applications are all important components of best management practices for fertilization that seek to maximize the amount of nutrients taken by poplars and minimize run off of nutrients. Fertilizers can be applied at any time during the rotation and once the poplars are established soil analyses coupled with foliar analysis are the most economical and effective way of diagnosing nutrient deficiencies. Poplars sprout readily from the stump or root collar when cut; this re-sprouting is known as coppicing. Coppicing should be done in the dormant season. Coppicing is an inexpensive way to re-establish a stand without replanting; although landowners may choose to replant improved varieties rather than coppice. But, coppicing can be productive and often provide higher yields than the original stand in the first 5 years after harvest. Poplar stands should be harvested when their annual growth increment begins to decline. There are several options for harvesting ranging from low tech labor intensive methods to the use of sophisticated high tech harvesting machines and chippers. In the northern states, poplars should be harvested in the winter months to minimize soil compaction and to maximize resprouting. Poplar trees resprout better if cut during the dormant season from November through April. Winter is also a better time for harvest because foliage is left on site to recycle essential nutrients and organic matter. The choice of a harvest system depends upon the planting area, tree size, and landowner objectives. For relatively small areas a labor intensive system or a small tractor may be used. But, for large areas a highly mechanized approach may be necessary. For bioenergy there are mobile whole tree chippers under development. These approaches vary greatly in cost and energy use. Post-harvest the farmer could kill the stumps of the former planting with herbicide and replant with new improved poplar clonal stock. Stumps could be removed by a bulldozer or left as is the site replanted within the old rows. Another post-harvest option is to maintain the subsequent stand as a coppice stand. Each stump will have multiple stems and the coppice stand will be more productive than the old stand. This approach may be suitable when poplar is grown for bioenergy. Biomass yields in Minnesota ranged from 5.8 to 10.1 Mg/ha/yr, and currently operational plantings in Minnesota yield over 9.0 Mg/ha/yr. A goal 7

8 of 15 Mg/ha/yr has been set by geneticists for new poplar clones in the future. Other studies report delivered yields of poplar of 7.3 and 13.4 Mg/ha/yr in the US (De La Torre Ugarte et al., 2003). Poplar variety NM6 yielded 8.3 Mg/ha/yr in the Upper Peninsula of Michigan over 10 years (Miller and Bender, 2008). Based on the maximal annual yields and the rotation to achieve the maximal yields projected by Wang et al. (2011), the delivered yields and SDs of hybrid poplar at the farm-gate (after accounting for storage losses) are 9.74 ± 1.12 Mg/ha/yr with a 10-year rotation in Illinois, 9.4 ± 0.69 Mg/ha/yr with an 8-year rotation in Michigan, and 3.15 ± 1.58 Mg/ha/yr with a 6-year rotation in Oklahoma. Method for Determining Break-even Prices This study estimates the break-even prices of these different biomass production systems using the following steps: i) Construct costs of production for each crop system for each year over the life of the crop; ii) Discount these to obtain the present value of these costs; iii) Determine the annual value of the residual return to cropland if used for its most profitable alternative use (continuous corn or rotation corn) at given prices for corn and soybeans. iv) Add this value of land to the discounted net present value of production costs to obtain discounted total costs of production. v) Determine the time path of biomass yields and its discounted level using the same discount rate as in (ii). vi) Divide the discounted total costs by the discounted level of yield to obtain the breakeven cost of producing biomass with a particular feedstock in terms of $ per dry metric ton. More specifically, as described in Jain et al., (2010), we calculate the present discounted T Ct value of the sequence of annual costs, C t t, over the life of each crop using a discount t= 0 (1 + d) rate of 4%. We similarly calculate the present value of yields, given the sequence of annual T Yt yields over the life of the crop, using the same discount rate as above. Note that Y t t is t= 0 (1 + d) yield after losses during harvesting and storage in year t and we refer to the annualized yields after losses during harvesting and storage as yield at farm-gate. The breakeven farm-gate price P B ($ per ton of dry matter) for each crop is the minimum price per dry metric ton of the bioenergy crop that a cropland owner would need to receive each year to cover all the costs of production over the life of the crop. This price would result in the present value of revenues from the crop being just equal to the present value of costs of producing it over its life as follows: P Y C T T t t B[ ] = t t t= 0 (1 + d) t= 0 (1 + d) 8

9 T Ct t t= 0 (1 + d) Thus, PB = T Yt t t= 0 (1 + d) where T is the life of the crop, C t is the cost of the bioenergy crop per hectare in period t, and d is the discount rate. C t includes the cost of producing the crop at time t (C pt ) and opportunity cost of land (C Lt ), both measured in $ per hectare. We estimate C Lt as follows: C Lt = (P ct *Q ct C ct + P st *Q st C st )/2 where P ct, Q ct and C ct are the price ($ per metric ton), yield (metric tons of dry matter per hectrare) and production cost of corn ($ per hectare), respectively while P st, Q st and C st are the corresponding values for soybeans at time t. In the case of marginal land that is currently idle, we use the average soil rental rate for land enrolled in the Conservation Reserve Program (CRP) in that state as a proxy for the opportunity cost of that land. Data and Assumptions We develop state-specific enterprise budgets of the costs of the eight biomass feedstock production systems examined (see Table 1) over their lifetime for Illinois, Michigan, and Oklahoma. Specifically, we estimate rotation and tillage specific costs of production in 2007 prices for corn, soybeans, corn stover, switchgrass, miscanthus, native perennial grass mix, and hybrid poplar for each of the states included in our analysis. Agronomic Assumptions: Corn stover yields are estimated based on a grain-to-residue ratio of 1:1 and a moisture content of 15% in grains reported in Sheehan et al. (2003), Graham et al. (2007) and Wilcke and Wyatt (2002). We also assume that corn yield in a monoculture corn system is 88% of the yield level achieved in a rotation corn system. The application rates of N, P, and K to replace the loss of nutrients and soil organic matter due to the removal of each dry metric ton of stover are assumed to be 3.5, 0.8, and 7.6 kg, respectively (Sheehan et al., 2003). Similar to Malcolm (2008), we assume that 50% of stover can be removed from fields if no tillage is practiced and 30% can be removed if conventional till is used. The delivered corn stover yield is also subject to a 7% storage loss as is the case for biomass feedstocks. In estimating the costs of miscanthus and switchgrass we rely on agronomic assumptions about fertilizer, seed, and pesticide application rates described in Jain et al. (2010) for all the three study states. As show in Table 2, to allow flexibility in input requirements for these variables, we consider two alternative scenarios, a high cost and a low cost scenario. The low cost scenario considers a low fertilizer application rate, low replanting probability, high secondyear yield, and low harvest loss while the high cost scenario captures the opposite case of production. In both scenarios, we assume a life span of 10 years for switchgrass and 15 years for miscanthus for our economic analysis as suggested in numerous studies (Duffy, 2001; Qin et al., 2006; Khanna et al., 2008; Brechbill and Tyner, 2008; Perrin et al., 2008; Mooney et al., 2009; 9

10 Lewandowski et al., 2003; Clifton-Brown et al., 2007; Hansen et al., 2004; Christian et al., 2008; Jain et al., 2010). Similarly, based on James et al. (2010) and Haque et al. (2009), we develop state-specific low and high cost scenarios for prairie grass mixes with agronomic assumptions differing mainly in the reseeding rate and harvestable yield in the first two year (Table 3). For hybrid poplar, the state-specific agronomic assumptions in the low cost scenario are based on James et al. (2010) featuring low-input, no replanting, and low maintenance between planting and harvest while assumptions in high cost scenario are based on Lazarus (2010) featuring low input for plant establishment but a 20% replanting rate and higher requirements for nitrogen fertilizer application and weed mowing for maintenance purposes (see Table 4 for details). A stand of poplar is assumed to last for two rotations and harvestable biomass yields for each of the two rotations are assumed to be the same. Assumptions about the lifetime of the poplar are described in Table 4 and differ across states. Biomass Yields: Corn stover yields under different rotation and tillage are shown in Table 5. The yield of corn stover varies from less than 1 Mg per ha in Oklahoma under rotation corn and no-till to 3.4 Mg per ha in Illinois under monoculture and no-till. In the absence of long term observed yields for miscanthus and limited data for switchgrass, we use a crop productivity model MISCANMOD to simulate their potential yields. The MISCANMOD estimates yields of miscanthus and Cave-in-Rock variety of switchgrass using GIS data on climate, soil moisture, solar radiation and growing degree days as described in Jain et al. (2010). The Cave-in-Rock switchgrass cultivar is an upland variety that originated in Southern Illinois and is cold-tolerant and well-suited for the upper Midwest but has relatively low yields. Lowland varieties of switchgrass, like Alamo, are most suited for southern US and have higher yields (Lemus and Parrish, 2009, Lewandowski et al., 2003). The estimate of harvested yields for mixed prairie grass systems in Illinois is based on the DAYCENT model (Davis, 2011) while the estimates for Michigan and Oklahoma are obtained from James et al. (2010) and Aravindhakshan et al. (2011), respectively. Due to lack of sufficient field experimental data on hybrid poplar yields in the study states, the state-specific poplar yield estimates are obtained from a Predictive Ecosystem Analyzer model (Wang et al., 2011), which shows that annual yield maximizing rotations for poplar are 10, 8, and 6 years for Illinois, Michigan, and Oklahoma, respectively. Table 5 shows the assumptions we make about the yields of biomass feedstocks included in the study. For energy crops, these yields differ across the low cost and high cost scenarios because of differences in harvest losses and the time taken to establish the grasses. Lobell et al. (2009) note that actual yields achieved by farmers in a region are typically smaller than the yield potential, because achieving the yield potential requires almost perfect management and soil conditions that might be possible only under experimental conditions. They find evidence that average yields of row crops like wheat and rice in rainfed conditions were commonly 50% or lower than yield potential. In the absence of data on actual yields obtained by farmers growing energy crops, we examine the sensitivity of cost estimates to having 25% lower yields of switchgrass, miscanthus, mixed prairie grasses, and hybrid poplar as compared to the 10

11 maximum potential yields projected by crop simulation models calibrated using data from experimental plots. Costs of production: The farm-gate production cost of biomass includes (i) the cost of inputs, such as chemicals, fertilizers and seeds, (ii) the cost of field operations, such as planting and harvesting, and (iii) the costs of storage. Costs of production for each county are obtained using statespecific input prices and machinery costs for The per hectare costs of land, overhead (such as farm insurance and utilities), building repair and depreciation, and labor are not included in the costs of perennials or row crops since they are assumed to be the same for all crops and do not affect the relative profitability of alternative crops. Instead, these are included as the opportunity costs of using existing farmland, labor and capital to produce bioenergy crops. Fertilizer application rates in Table 6, and input prices in Table 7 are obtained from the crop budgets complied for that state by state extension services with tillage or inflation adjustments (FBFM, 2006; Dartt and Schwab, 2001; Oklahoma State University Extension, 2007). Seed prices for switchgrass and mixed prairie grasses in Michigan are obtained from James et al. (2010) and we assume that seed prices for the two perennial systems are the same in Illinois and Oklahoma. There is considerable uncertainty about the cost of miscanthus rhizomes since they are not yet commercially available for large scale plantations. Personal communication with developers and producers of miscanthus varieties, rhizomes and plugs indicates that they expect miscanthus plugs to cost between $0.30 and $0.80 per plug as they begin commercial sales in 2010 and close to $0.25 per plug by Plugs are vegetatively produced from rhizomes and have a dominant meristem. Costs are expected to be lower for rhizomes compared to plugs and to decrease as the scale of production increases. Cost of propagating rhizomes at the University of Illinois is estimated to be $0.10 per rhizome. Therefore, we assume a rhizome cost of $0.25 and a planting rate of 10,000 rhizomes per ha for all the three states following Jain et al. (2010). For lack of site-specific information on price of cuttings for planting poplar, we assume the cost is $0.22 per cutting and 2717 cuttings per ha for all three states based on James et al. (2010). The costs of producing corn stover includes the cost of fertilizer that needs to be applied to replace the loss of nutrients and soil organic matter due to removal of residue from the soil. Machinery costs of corn stover and dedicated energy crop production and harvest as shown in Table 8 are obtained from various sources. Cost estimates for Illinois are based on the state s crop budgets on hay alfalfa (FBFM 2007, 2008) while for Michigan and Oklahoma, these estimates are based on James et al. (2010) and Haque et al. (2009), respectively. For poplar harvesting, feller-buncher and chipping -grounding costs reported in James et al. (2010) are used for all the three states. Poplar stump removal costs are based on Lazarus (2010) and assumed to be identical for the three states. The costs of staging and loading are assumed to be $6.38 per Mg for all three states based on Duffy (2007). We also assume that biomass is stored outside on crushed rock on reusable tarp; a method found to be most cost-effective at 3.22 $ per Mg by Brummer et al. (2000). Storage is assumed to result in 7% loss in yield and to reduce the moisture content of miscanthus bales to 15% through natural ventilation and without any 11

12 additional drying costs. In addition, in the low cost scenario of biomass feedstock production, baling costs are split into fixed costs of baling (tractor and implement overhead) and variable costs of baling (fuel, lube, and labor). The fixed costs of baling is estimated to be $34.7 per ha and invariant with yield while the variable costs of baling are estimated to be $6.16 per Mg in Illinois (FBFM, 2008). Therefore, harvest costs would decrease with the increase of yields per hectare. Similar breakup is also applied to the baling costs in Michigan and Oklahoma based on the same ratio of the fixed and variable cost of baling in Illinois due to lack of data to estimate the breakeven cost of biomass feedstock production in the low cost scenario. In the high cost scenario, a flat baling cost per Mg of biomass as shown in Table 8 is used. The estimate the opportunity cost of cropland is based on determination of the most profitable use of that land. We find that, among the crop choices examined, a corn-soybean rotation with conventional tillage is the most profitable land use in Illinois with a total revenue above operating costs of $771 per ha, while a corn-soybean rotation with no till is the most profitable land use in Michigan and Oklahoma with a total revenue above operating costs of $458 per ha and $409 per ha, respectively. These are estimated using state-specific five year ( ) historical average corn and soybean yields per acre as well as state-specific three year average ( ) corn and soybean prices obtained from USDA/NASS (2009b). Illinois has the highest opportunity costs of land since its corn and soybean yields and prices are high while the costs of production of corn and soybeans are relatively low. On the other hand, Oklahoma has the lowest opportunity cost of land under a corn-soybean rotation due to its low corn and soybean yields and relatively high costs of production of these crops. Since a rotation corn system is systematically more profitable than a monoculture corn system in the study states, an opportunity cost of land is also considered for the collection of corn stover for a monoculture corn system. The opportunity cost of land in this case is the foregone profits with rotation corn if the production of corn stover leads the farmer to switch to monoculture corn and measured as the difference in net profits between these two production systems. There is also an opportunity cost of land for switching across tillage practices with rotation corn, from conventional till to no-till in Illinois and from no-till to conventional till in Michigan and Oklahoma. For marginal land costs for energy crops we use the average CRP payments for CRP general sign-ups in 2007 in the state as a measure of the alternative income from that land (USDA Farm Service Agency, 2008). Results Among the three perennial grass production systems examined, miscanthus has the lowest costs of production, ranging $35 87/Mg DM when planted on marginal land and $43 103/Mg DM when planted on cropland. In contrast, the costs of biomass production under the two types of land range respectively between $46 100//Mg DM and $73 135/Mg DM for switchgrass and between $69 109/Mg DM and $99 177/Mg DM for mixed prairie grasses. Thus, production of energy crops is more likely to be observed first on marginal land in these states and it will be viable on cropland only if the price of biomass is sufficiently high and low cost marginal land is unavailable. Across the states, Oklahoma has the lowest costs of switchgrass and miscanthus production regardless of the type of land used because of its higher switchgrass and miscanthus 12

13 yields and lower opportunity costs of land relative to Illinois and Michigan. Michigan has the highest costs of switchgrass and miscanthus production in general due to its low yields for these two crops, though it has lower opportunity costs of land than Illinois. Costs of mixed prairie grass production in most cases are lowest in Michigan since it has the highest grass yields. In addition, the costs of production for perennial grasses between the low and high cost scenarios differ by about $17/Mg to $38/Mg, suggesting that agronomic decisions about input application rates and crop attributes such as length and ease of establishment and timing of harvest have a substantial impact on the costs of biomass grass production. In addition to its advantage in mixed prairie grass production compared with Oklahoma or Illinois, Michigan also has the lowest costs of production for hybrid poplar, with a cost of $88-95/Mg DM when planted on marginal land and $ /Mg DM when planted on cropland. Costs of poplar production in Illinois are slightly higher than in Michigan because of its higher marginal and crop land costs. Oklahoma has the highest costs of poplar because poplar yields are much lower, only about one third of those in Illinois and Michigan. Among the source of biomass production on cropland we find that the breakeven costs of producing corn stover are lower than those of dedicated energy crops (except miscanthus in Oklahoma in the low cost scenario); the breakeven cost of corn stover ranges between $51/Mg DM and $60/Mg DM under a rotation corn system with no till in the three states. Costs of corn stover are higher with rotation corn with conventional tillage in all three states because of the lower residue harvest rate. However, breakeven costs of corn stover are much higher when a monoculture corn system is considered even with no-till and even more so with conventional tillage; this is due to the opportunity cost of land in the former case and compounded with lower corn stover yields in the latter case. This suggests that farmers in these states are unlikely to have an economic incentive to convert their production system from rotation corn to monoculture corn simply for greater stover collection. The components of the costs of biomass feedstock production vary across species but are not too different across states. For corn stover, in all three states, harvesting costs account for about 80% or higher in the total operating costs of production excluding the opportunity cost of land, with the rest being the costs of fertilizer needed to replace the nutrient loss due to stover removal. When collected from a monoculture corn system and after taking into account the opportunity cost of land, corn stover becomes much more costly, with land costs accounting for 18-34% of the total cost in Illinois, 44-56% in Michigan, and 47-60% in Oklahoma. In contrast, when collected from a rotation corn system, total costs of stover collection are much lower and land costs due to alternative tillage practices account for 12% of the total cost in Illinois for no till, 6% in Michigan for conventional till, and 44% in Oklahoma for conventional till. For switchgrass, of the total operating costs excluding the opportunity cost of land, the largest cost component is harvesting expenses, accounting for more than half of the total operating cost in general, except in the high cost scenario in Michigan where harvesting costs are only 42% but still higher than a percentage of 38% for fertilizer costs. Fertilizer costs are the second largest cost item in switchgrass production accounting for 24-38% of the total operating costs across states. The cost of land for switchgrass is only about 11-31% of the total cost of production when it is planted on marginal land. However, the share of land cost can rise significantly to 35-59% when switchgrass is planted on cropland. 13

14 Harvesting expenses are also the largest component in the costs of miscanthus production, with a share of around 60% in total operating costs excluding the cost of land in Illinois and Oklahoma and 41-48% in Michigan. Miscanthus establishment costs in the three states range 26-38% in the total operating costs, ranking the second largest cost component. Fertilizer costs account for about 8-14% and are consistently the third largest component in the total operating costs in all three states. The share of land cost in the total cost of miscanthus production varies significantly across states, ranging from 6-8% for marginal land and 21-27% for cropland in Oklahoma to 14-18% for marginal land and 27-40% for cropland in Illinois and Michigan. For low input mixed prairie grasses, the dominant cost item in total operating costs excluding the cost of land is harvesting expenses, accounting for 57-63% of the total across the study states. Seed costs are consistently the second largest operating costs in the study states, with a share of 25-30% in the total. Chemical costs and preharvest machinery costs have about the same share of around 4-6% in the total in Illinois and Michigan; In Oklahoma, however, preharvest machinery costs have a share of 7-9% in the total, much higher than the share of chemical costs, 2%. The share of land costs in the total cost of production of low input prairies depends significantly on the type of land used. When mixed prairie grasses are planted on marginal land, land costs account for 16-20% in Oklahoma, 25-27% in Michigan, and 29-34% in Illinois. However, when cropland is used, the share of land costs can be 42-51% high in Michigan and Oklahoma and 57-62% in Illinois. For hybrid poplar, in Illinois and Michigan, the dominant component in total operating costs is harvesting expenses with a share of 77-85% and followed by preharvest machinery costs and seed costs with a share of 6-12% each in the total. In Oklahoma, the largest cost component is still harvest expenses but the share of them in the total is only 34-57%. Though preharvest machinery costs and chemical costs are still the second and third largest components in total operating costs in Oklahoma, respectively, the shares of preharvest costs are substantially higher than those in Illinois and Michigan, ranging 22-34% in the total for preharvest costs and 19-27% for chemical costs. In Oklahoma, the share of land costs in the total costs of poplar production is a little above 20% when planted on marginal land and 51-56% when planted on cropland. In Illinois and Michigan, the share of land costs in total is relatively low when poplar is planted on marginal land, being 17% in Michigan and 23% in Illinois. Even when cropland is used for polar production, the costs of land still account for about 32% of the total costs in Michigan and around 49% in Illinois. Sensitivity Analysis We examine the sensitivity of the farm-gate breakeven prices of biofuel feedstocks including the opportunity cost of land to changes in the prices of corn, soybean, seed, and fertilizer, crop and biomass yield, harvest costs, preharvest machinery expenses, and discount rate. The breakeven prices of biofuel feedstocks planted on cropland will also be affected by the value placed corn stover when estimating the opportunity cost of land. The sensitivity of breakeven prices to stover values is ignored in this study since as shown in Jain et al. (2010), the effect of inclusion of corn stover in the opportunity cost of land on the breakeven prices of 14

15 biomass feedstocks is very modest on average, ranging between 3-7% across grass species and the Midwestern states. Table 10 summarizes the sensitivity of the breakeven prices of corn stover while Table 11 and Table 12 report the sensitivity of the breakeven prices of perennial biomass feedstocks with respect to the inclusion of marginal land rent and the opportunity cost of cropland, respectively. In all three tables, sensitivity is reported in terms of percentage changes relative to the benchmark estimates in Table 9. Table 10 shows that the breakeven price of corn stover collected from a rotation corn system is most sensitive to changes in harvest costs. A 25% increase in harvest costs will lead the breakeven price of stover to increase by 12-21% since harvest costs are always the dominant component in the cost of stover collection in this rotation system. For stover collected from a monoculture corn system, because of the influence of the opportunity cost of land, the increase in the breakeven price of stover with respect to a 25% increase in harvest costs is 9-17%, only slightly higher than changes to a 25% increase in crop yields and fertilizer prices. With a 25% increase in crop yields, the breakeven price of stover will decrease by 5-13% when collected from a rotation corn system and 5-13% when collected from a monoculture corn system; while with a 25% increase in fertilizer prices, breakeven prices will increase by 2-11% and 7-15% for stover collected from the two systems, respectively. A 25% increase in corn and soybean prices will decrease the breakeven price of stover by 3-7% across the states, suggesting that the opportunity cost of land for stover collected from the monoculture corn system is decreasing as crop prices rise. We do not include a sensitivity analysis for the impact of a change in the prices of corn and soybeans in Table 11 since the opportunity cost of marginal land is fixed at the level of the CRP payments. Over time, however, CRP payments could change with sustained increases in crop prices and raise the opportunity costs of using land for energy crops. Table 11 shows that breakeven prices of biomass feedstocks on marginal land are also insensitive to seed or fertilizer prices changes in general: when seed or fertilizer prices increase by 25%, the increases in breakeven prices of biomass feedstocks are less than 10%, with miscanthus in the high cost scenario being the most sensitive feedstock to seed price changes (leading to a 6-9% increase in breakeven price across states) and switchgrass in the high cost scenario being the most sensitive feedstock to fertilizer prices changes (6-8% across states). Breakeven prices of biomass feedstocks are generally most sensitive to crop/biomass yield and harvest cost changes. A 25% increase in biomass yields can reduce the breakeven prices by 9-14% for perennial grasses and 6-25% for poplar across states, while a 25% increase in harvest costs will lead the breakeven prices to increase by 7-15% for perennial grasses and 7-17% for poplar. In conducting the sensitivity analysis to changes in yields of energy crops, it should be noted that input requirements are essentially being assumed to be unchanged with changes in yield. As shown in Tables 2-4, seed/rhizome requirements and fertilizer and herbicide applications are assumed to be determined on a per hectare basis (except for some nutrient application rates for switchgrass) instead of on a per Mg basis. This is due to lack of empirical evidence to support nutrient application rates that are related to biomass yield. This may result in some over-estimate of the extent to which an increase in yield reduces breakeven costs. 15

16 Table 11 also shows feedstock prices are not sensitive to pre-harvest cost changes since, as mentioned earlier, the share of pre-harvest costs in total operating costs is relatively low for the feedstocks examined. In addition, changing discount rate from 4% to 8% leads to a relatively smaller increase in the breakeven prices of switchgrass and mixed prairie grasses (1-6%) but this impact on miscanthus and poplar breakeven prices can be significantly larger, ranging between 8-17% because of their longer lifetime. Table 12 shows that when biomass feedstocks are produced on cropland, their breakeven prices are most sensitive to corn and soybean price changes. A 25% increase in corn and soybean prices can lead the breakeven prices of feedstock to increase by 11-35% for perennial grasses and poplar, with switchgrass and mixed prairie grasses in the low cost scenario being the most impacted species with a breakeven price change around 30% across states on average. Another significant difference in the sensitivity results in comparison to marginal land case is the sensitivity of breakeven prices of feedstocks to changes in biomass yields. An increase in biomass yields leads to a substantial reduction in breakeven prices for all biomass feedstocks. The percentage reduction is larger when the energy crops are planted on cropland as compared to when they are planted on marginal land. The implications of a reduction in yields of these feedstocks relative to the benchmark case are simply an opposite impact of the same magnitude as indicated in Table 12. If biomass yields of switchgrass turn out to be 25% lower than in the benchmark case, its costs would increase by 12-15% than in the benchmark case. On the other hand, the costs of mixed prairie grasses would increase by 14-17% and of miscanthus by 11-15%. The increase in costs would be much higher in the low cost scenario than in the high cost scenario where the yields are relatively lower and thus the effects of a 25% increase in yields is relatively smaller. It can also be seen in Table 12 that the sensitivity of the breakeven prices of biomass feedstocks to other parameter changes is very similar to that discussed in the case when biomass feedstocks are produced on marginal land, with the breakeven prices being most sensitive to changes in harvest costs relative to other operating cost components and the breakeven price of feedstocks with longer lifetimes being more sensitive to the increase in discount rate. Conclusions We find that the breakeven costs of production of various biomass feedstocks differ widely across feedstocks, regions and scenarios regarding the ease of establishing and harvesting them. The farm-gate costs of biomass feedstocks vary dramatically from $35 per Mg for miscanthus in the low cost scenario on marginal land in Oklahoma to $389 per Mg in hybrid poplar in the high cost scenario on cropland also in Oklahoma. In general, corn stover is less costly than most dedicated energy crops (on cropland) when a rotation corn system is considered, with a cost of production ranging from $51 per Mg under a rotation corn system with no till in Michigan to $114 per Mg under a rotation corn system with conventional till in Oklahoma. We also find no significant difference in costs of stover production in the rotation corn system with a no-till practice across the three study states. However, under a monoculture corn system, the costs of stover collection increase substantially and in Oklahoma in particular the cost of stover collection under a monoculture corn system is dominated by the opportunity cost of land, with a 16

17 share of land cost of 47 60% in the total, suggesting that farmers in these states do not have an economic incentive to convert their production system from rotation corn to monoculture corn simply for greater stover collection unless the price of corn stover or subsidies for biomass feedstock provision is high enough to compensate the opportunity cost of the conversion. Energy crops have the potential to provide significant environmental benefits in the form of soil carbon sequestration, reduction in sediment run-off, improvement in soil quality and habitat for wild life. To induce farmers to switch from a corn-soybean rotation to production of an energy crop, the minimum price they would need to be paid differs across regions. To the extent that this price is higher than the market price of biomass, our analysis shows the magnitude of the subsidy that will be needed per metric ton to induce the production of a particular feedstock. The production of corn stover was restricted to sustainable levels to preserve soil quality and water quality by preventing run-off. If there is interest in preventing farmers from collecting excessive residues, they will need to be compensated for the foregone income from corn stover. The analysis here shows their income from corn stover collection and the compensation they will need to be provided for reducing the level of stover harvested. Sensitivity analysis can be use to determine the payments that would be needed to prevent harvest levels from being even higher than levels considered here. We also analyze the sensitivity of the breakeven prices of biomass feedstocks to various parameter changes and find that when biomass feedstocks are produced on cropland, their breakeven prices are most sensitive to corn and soybean prices changes. When planted on marginal land, their breakeven prices then are most sensitive to changes in biomass yields and harvest costs. Moreover, biomass feedstocks with longer lifetimes are more price sensitive to changes in discount rate. The breakeven costs estimated above together with information about the market price of biomass can be used to determine the extent to which farmers would need to be compensated through subsidies to produce biomass from various sources and on the two types of land (marginal and average cropland). The subsidy needed will depend on the particular scenario about costs of production, yields and crop prices that are relevant. As shown in Table 9 and Figures 1 and 2, the lowest cost biomass production would occur in Oklahoma where the breakeven cost is $35 per Mg DM with miscanthus. At a biomass price of $50 per Mg DM, it would be profitable to produce miscanthus on marginal land in Illinois and Oklahoma, switchgrass on marginal land in Oklahoma and even miscanthus on cropland in Oklahoma if the low cost scenario prevails. If the production of other energy crops is desired for environmental reasons or if the costs of production of miscanthus and switchgrass turn out to be high, the subsidy required for their production can be calculated as the difference between the breakeven price and the market price of biomass. If the market price of biomass happens to be high, say, $60 per Mg DM, it would also be profitable to produce miscanthus in Michigan and switchgrass in Illinois on marginal land in the low cost scenario and harvest corn stover in all three states from rotation corn with no-till practice without any subsidies. 17

18 If the production of energy crops is desired on cropland as a mechanism to reduce soil erosion and nitrogen leaching and increase biodiversity, then considerable subsidies would be required. In particular, at a biomass price of $50 per Mg, the subsidies required to induce the production of a high yielding perennial like miscanthus, would range from $12-$19 Mg if the costs of production are low and $23-$52 per Mg if they are high. Mixed prairie grasses have attracted attention due to their LIHD attributes. Considerably high subsidies ($19-34 per Mg) would be required to motivate landowners to grow mixed grasses even on marginal land if their costs of production are low. These would be even higher if costs of production turn out to be high and/or policy makers seek to induce their production on cropland. For example, in the low cost scenario, the subsidy needed to trigger the production of mixed prairie grasses as bioenergy feedstocks on cropland (if the market price of biomass is $50 per Mg) is $95 per Mg in Illinois, $49 per Mg in Michigan, and $63 per Mg in Oklahoma. Also in the low cost scenario, the subsidy needed for poplar production on cropland is $84 per Mg in Illinois, $56 per Mg in Michigan, and $163 per Mg in Oklahoma. The production of perennials involves lags between planting and harvest, upfront investment in establishment of these crops and risks and uncertainty about returns over the life of the perennial. The subsidy estimated above is based on a comparison of the breakeven cost of producing a bioenergy crop with the market price of biomass, and it ignores farmers cash flow constraints and concerns about the riskiness of the investment. To the extent that cash flow constraints and concerns about risk and uncertainty are significant barriers to investment in perennial crops, per unit output based subsidies may need to include a risk premium and be larger than those indicated above. They may also need to be supplemented by subsidies that share the establishment costs and reduce the upfront investment needed in perennials by a landowner. 18

19 References Adler, P.R., M.A. Sanderson, P.J. Weimer, and K.P. Vogel Plant species composition and biofuel yields of conservation grasslands. Ecological Applications 19(8): Aravindhakshan, S.C., F.M. Epplin, and C.M. Taliaferro "Economics of Switchgrass and Miscanthus Relative to Coal as a Feedstock for Generating Electricity." Biomass and Bioenergy 34(2010): Aravindhakshan, S.C., F.M. Epplin, and C.M. Taliaferro "Switchgrass, Bermudagrass, Flaccidgrasss, and Lovergrass biomass yield response to nitrogen for single and double harvest. Biomass and Bioenergy 35(2011): Brechbill, S.C., W.E. Tyner, and K.E. Ileleji The Economics of Biomass Collection and Transportation and Its Supply to Indiana Cellulosic and Electric Utility Facilities, In B.C. English, R.J. Menard and K. Jensen (eds.), Risk, Infrastructure and Industry Evolution, Proceedings of a Conference June 24-25, 2008, in Berkeley, California, Brummer, E.C., C.L. Burras, M.D. Duffy, and K.J. Moore Switchgrass production in Iowa: economic analysis, soil suitability, and varietal performance. Iowa State University, Ames, Iowa. Prepared for Bioenergy Feedstock Development Program, Oak Ridge National Laboratory, Oak ridge, TN. Christian, D.G., and E. Haase Agronomy of miscanthus. In: Jones MB, Walsh M, editors. Miscanthus for energy and fibre. London: James and James (science Publishers) Ltd.: p Clifton-Brown. J.C., J. Breuer, and M.B. Jones Carbon mitigation by the energy crop, miscanthus. Global Change Biology 13: Dartt, B., and G.D. Schwab Crops and livestock budgets estimates for Michigan. Agricultural Economics Report No. 609, Department of Agricultural Economics, Michigan State University, East Lansing, MI, September, Davis, S Personal communication. Energy Biosciences Institute, University of Illinois, Urbana, IL. Davis, S., Parton, W., Dohleman, F., Smith, C., Grosso, S., Kent, A. & Delucia, E Comparative Biogeochemical Cycles of Bioenergy Crops Reveal Nitrogen-Fixation and Low Greenhouse Gas Emissions in a Miscanthus Giganteus Agro-Ecosystem. Ecosystems, 13, DEFRA A review of the potential of giant grasses for UK agriculture. Research and development final project report NF0419; URL: SARS. De La Torre Ugarte, D.G., M.E. Walsh, H. Shapouri, and S.P. Slinsky "The Economic Impacts of Bioenergy Crop Production on U.S. Agriculture." U.S. Department of Agriculture. URL: Dietze, M Grass Yield Data. Energy Bioscience Institute, University of Illinois, Urbana, IL. Duffy, M.D., and V.Y. Nanhou Costs of producing switchgrass for biomass in southern Iowa. In: Janick J, Whipkey A, editors. Trends in new crop and new uses. Alexandria, VA: ASHS Press: p Duffy, M Estimated Costs for Production, Storage and Transportation of Switchgrass, University Extension Report PM 2042, Iowa State University, Ames, Iowa. 19

20 Ellinger, P Ethanol Plant Simulator. Dept. Agr Econ, University of Illinois at Urbana- Champaign, Urbana, IL. URL: FBFM Farm Economics Facts and Opinions. Crop Budgets, Illinois, Department of Agricultural Economics, University of Illinois at Urbana Champaign, October, URL: FBFM Farm Economics Facts and Opinions. Farm Business and Farm Management Newsletters. Department of Agricultural Economics, University of Illinois at Urbana Champaign, Graham, R.L., R. Nelson, J. Sheehan, R.D. Perlack and L.L Wright Current and Potential U.S. Corn Stover Supplies. Agronomy Journal 99:1-11. Haque, M., F.M. Epplin, and C. Taliaferro Nitrogen and harvest frequency effect on yield and cost for four perennial grasses. Agronomy Journal 101(6): Heggenstaller, A.H., K.J. Moore, M. Liebman, and R.P. Anex Nitrogen influences biomass and nutrient partitioning by perennial, warm-season grasses. Agronomy Journal 101(6): Hwang, S., F. E. Epplin, B. Lee, and R. Huhnke "A probabilistic estimate of the frequency of mowing and baling days available in Oklahoma USA for the harvest of switchgrass for use in biorefineries." Biomass and Bioenergy 33: Huisman W, Venturi P, Molenaar J Costs of supply chains of Miscanthus giganteus. Industrial Crops and Products 6(3): Jain, A.K., M. Khanna, M. Erickson, and H. Huang An Integrated Bio-geochemical and Economic Analysis of Bioenergy Crops in the Midwestern United States. Global Change Biology BioEnergy 2(5): James, L.K., S.M. Swinton, and K.D. Thelen Profitability analysis of cellulosic energy crops compared with corn. Agronomy Journal 102(2): Khanna, M., B. Dhungana, and J. Clifton-Brown Costs of Producing Miscanthus and Switchgrass for Bioenergy in Illinois. Biomass and Bioenergy 32: Lazarus, W.F Minnesota crop cost & return guide for Department of Applied Economics and University of Minnesota Extension, University of Minnesota, Twin Cities, MN. URL: Lemus, R., and D. L. Parrish "Herbaceous Crops with Potential for Biofuel Production in the USA." CABI Reviews: Perspectives in Agriculture, Veterinary Science, Nutrition and Natural Resources 4: Lewandowski,.I, J.M.O. Scurlock, E. Lindvall, and M. Christou The development and current status of potential rhizomatous grasses as energy crops in the US and Europe. Biomass and Bioenergy 25(4): Lobell, D. B., K. G. Cassman, and C. B. Field "Crop yield gaps: their importance, mangnitudes, and causes." Annual Review of Environment and Resources 34: Malcolm, S. (2008) "Weaning Off Corn: Crop Residues and the Transition to Cellulosic Ethanol." Paper Presented at the Transition to A BioEconomy: Environmental and Rural Development Impact, Farm Foundation, St. Louis, MO. Mapemba, L. D., F. M. Epplin, R. Huhne, and c. M. Talaferro "Herbaceous plant biomass harvest and delivery cost with harvest segmented by month and number of harvest machines endogenously determined." Biomass and Bioenergy 32: