Introduction

Meeting the Congressional goal of replacing 30 % of US petroleum consumption with biofuels by 2030 [1] will require advanced technologies that can efficiently and economically harvest and convert a variety of biological materials into fuel. The 30 % replacement would require approximately 1.2 billion dry megagrams of biomass each year; 334 million Mg is projected to come from forestlands as cellulosic products [1]. Previous studies conclude that major energy sources for bioenergy in the future would come from cellulosic materials [2–4]. The US Renewable Fuels Standards (RFS2), as set out in the US Energy Independence and Security Act of 2007, indicated that by 2022, about 61 billion liters of cellulosic biofuels, primarily cellulosic ethanol, could be produced and blended with gasoline. Much emphasis is being made on three energy feedstocks in the US Central and South Central regions: energy cane, perennial grasses, and sweet sorghum. However, it is also important to evaluate potential benefits of other materials as potential bioenergy feedstock. Of particular importance will be those feedstocks that are on noncroplands and that require little cultivation costs such as planting, tilling, weed and pest control, and fertilization and/or irrigation, as these costs, and in particular fertilization, will likely increase at rapid rates.

In the last 120 years, much of the noncropped native grasslands of the south central USA (defined here as Texas, Oklahoma, and New Mexico) have become invaded by woody plants, or “brush” [5]. These include several species of juniper (Juniperus spp.), Acacia spp., and, in particular, the leguminous tree/shrub, honey mesquite (Prosopis glandulosa Torr.). Some of these species may be a potential source of woody feedstock for bioenergy purposes [6]. We estimate that of the 21 million total hectares of mesquite in Texas alone [7, 8], about 20 %, or 4.2 million hectares, could be harvested for bioenergy needs. An additional ten million hectares of juniper exists in the state [7], and 10–20 % of this could also be harvested. Combining this with other species, a realistic total of about six million hectares of brush that has sufficient biomass density and occurs on acceptable topography could be harvested for biomass feedstock. Mesquite fixes nitrogen and resprouts in a multistemmed physiognomy following removal of above-ground tissue. Regrowth potential may be one of the most important features of mesquite in consideration as a bioenergy feedstock [9]. In addition, there is clear evidence that distribution and density of mesquite and other rangeland woody species are increasing at rapid rates and likely will continue with increased atmospheric CO2 and predicted climate change [5, 10]. However, the impacts of their remote locations and nonuniform distributions on harvest and feedstock transport costs need to be assessed using economic models before this potential woody feedstock source could be recommended.

A cost analysis of biomass production on rangeland grasses grown in Conservation Reserve Program areas in Oklahoma by Mapemba et al. [11] showed that the total cost of delivered feedstock varies significantly according to the production duration, harvest data, and ethanol plant capacity. Recently, Haque et al. [12] estimated amortized establishment costs for switchgrass to be $57.22 ha−1 and its production costs with a range of $39–52 Mg−1 according to various nitrogen fertilizer rates and harvest frequency per year. Harvesting constitutes a major portion of total costs of delivered biomass [11, 13–16]. McLaughlin et al. [17] showed that production cost of hybrid poplar in the Great Lakes region (at standing mass of 9.9 Mg ha–1) was $22.32 Mg–1.

There are several advantages to mesquite and similar woody plant species as a bioenergy feedstock source that may offset the lower growth rates and potentially higher transport costs. First, these species represent an abundant source of existing and drought-tolerant biomass without additional costs of planting, cultivation, irrigation, and fertilization. Ansley et al. [9] found that dry mass of 10-year-old regrowth mesquite in north Texas was 29.4 kg/tree and above-ground standing crop at a typical tree density of 750 trees/ha was 22.1 Mg ha−1. The standing mass value of existing mesquite was similar to other studies in central Texas [18] and southern California [19]. In contrast, it has been estimated that energy sorghum requires 10–20 N kg ha−1 year−1 to yield 22.4–31 Mg ha–1 [20]. A second advantage is that rangeland shrubs do not grow on land better suited for growing food or fiber and thus will not negatively impact other agricultural markets. It has been shown that the negative effects of corn production for ethanol, for example, can be devastating to other markets such as livestock feedlot feeding costs. Third, they have higher energy contents than switchgrass or short rotation woody crops (SRWCs) [16, 21]. Fourth, they can be harvested year-round because they grow in warmer latitudes and in relatively dry regions. This advantage ameliorates feedstock storage problems that plague potential bioenergy systems based on crops that are harvested only once a year and must then be stored all year. Fifth, we have found that water content of mesquite and juniper wood at harvest is lower than most cellulosic feedstocks; thus, feedstock drying costs could be reduced. Finally, unlike dedicated energy crops, harvesting mesquite and similar brush species, which have traditionally been considered as noxious plants, would yield other agricultural and ecological benefits, including increased herbaceous production for livestock forage, improved wildlife habitat that could yield increased recreational hunting income, improved environmental services such as enhanced water harvest, increased land values due to these improvements, and finally, enhanced ecological stability through increased grass cover and species diversity [22], and reduced soil erosion resulting from increased grass cover [23].

The primary weakness of this potential feedstock is the random distribution of the feedstock caused by the natural dispersion of propagules and the variable growth form from tree to tree. Any harvesting operation must contend with this variation. In addition, mesquite regrowth rates are slower, one estimate is 2.2 Mg ha−1 year−1 [9], than the 11 Mg ha−1 year−1 found for some SRWCs in the upper Midwest [4], or the 4–17 Mg ha−1 year−1 found for SRWCs in Kansas [24]. Heartwood does not begin to appear in mesquite regrowth until about the fifth year, and total regrowth mass does not begin to match preharvest biomass density levels until the tenth year [9]. Thus, any long-term harvest plan would have to incorporate more land area to account for this slower regrowth rate when compared to SRWC systems, and this would increase feedstock transport costs from harvest site to the bioenergy facility. In addition, the amount of woodland areas that are left protected and not harvested for ecological reasons such as wildlife habitat would need to be included in any long term plan.

The conversion efficiencies for woody feedstock to liquid fuels are largely unknown. However, woody material has been used as feedstock for bioelectricity generation in many countries [25]. Thus, we will base our economic projections for mesquite on bioelectricity generation as the end use. Our overall objective was to identify the delivered biomass costs of mesquite as feedstock for bioelectricity production in the south central USA. Specific objectives were to (1) estimate the total cost of delivered mesquite biomass from the field to a bioelectricity plant, (2) conduct a sensitivity analysis using key parameters, (3) simulate cost of mesquite biomass using actual biomass density distribution data from two locations in Texas, and (4) provide information as to how mesquite could be managed as a biomass feedstock for the development of the industry.

Materials and Methods

This study used two methods, i.e., standard budgeting and Monte Carlo Simulation, to estimate total cost of delivered mesquite biomass. The standard budgeting analysis was conducted in the following order: (1) baseline, (2) sensitivity, (3) best and worst scenario, and (4) change in total system area (TSA) due to changes in the percent of TSA that is left unharvested for wildlife habitat. The key parameters included in the standard budgeting analysis are annual biomass consumption by the bioelectricity plant, mesquite wood heating value, mesquite biomass density, harvesting interval, TSA needed to supply a sufficient amount of feedstock, harvesting costs, transport costs, feedstock storage costs, and land rent. The baseline and adjusted levels of these parameters for the sensitivity analyses are presented in Table 1.

Table 1 Assumed levels of selected parameters by scenarios for the budget
Full size table

The annual biomass consumption by the bioelectricity plant was determined based on existing plants in Europe and the USA that can range from 20,000 to 500,000 Mg year−1 [26]. We used a biomass consumption rate of 250,000 Mg year−1 for a mid-sized bioelectricity plant in our base model. We assumed a mesquite dry wood heating value of 19.7 MJ kg−1 based on recent laboratory analysis [21]. This energy content is about 25 % higher than other energy stocks such as switchgrass (15.8 MJ kg−1) [16].

Because the distribution of mesquite biomass density is variable, we assumed certain areas would be suitable and other areas not suitable for harvest. The suitable harvest areas (SHAs) were defined as patches of mesquite that contained a minimum of 15 Mg ha−1 dry biomass density. Any areas with lower mesquite biomass density and nonmesquite areas that occurred between the SHAs and the bioelectricity plant were designated as unsuitable harvest areas (UHA). The TSA equals SHA plus UHA.

We initially assumed that the TSA contained 50 % SHAs and 50 % UHAs (Fig. 1, left panel). Figure 1 is portrayed without units with the assumption that the number of SHAs, shown as equal-sized cells, is sufficient to sustainably supply a centrally located bioelectricity plant. The required amount of SHAs is dependent on the capacity of the bioelectricity plant, woody regrowth rates, and the amount of SHA that is protected and not harvested for wildlife habitat or other ecological purposes. We initially assumed that 10 % of the total SHAs would be protected from harvesting. The remaining 90 % of the total SHAs were divided into ten portions with one portion harvested each year under the assumption of a 10-year reharvest cycle [9]. Combining these factors, our initial assumption is that 4.5 % of the TSA will be annually harvested based on the formula:

$$ {\text{Percentage \; of \; TSA \; harvested \; each \; year}} = \left[ {0.{5} \times 0.{9} \times 0.{1}} \right] \times {1}00 $$
Fig. 1
figure 1

Phases 1 and 2 of a landscape harvest scenario for a mesquite bioenergy feedstock supply system with a central location for the bioelectricity plant. The left panel (phase 1) illustrates the total system area (TSA) needed to supply the bioelectricity plant, based on a 10-year reharvest cycle (the illustration assumes that the TSA is sufficient to supply the bioelectricity plant) and divided into equal sized cells. Cells with woody biomass densities that are suitable for harvest (suitable harvest areas, SHAs) are in green. Unsuitable harvest areas (UHA), which include non-wooded areas and mesquite areas with insufficient biomass density, are tan. In this example, 50 % of the TSA cells are SHAs. Phase 2 (right panel) shows the development of the annual harvest plan where 10 % of the SHA’s are set aside initially for wildlife habitat considerations and are never harvested (dark green cells). Of the remaining available SHAs, 10 % are harvested each year, assuming a 10-year reharvest schedule. The harvest plans for the first 2 years are shown in orange and pink. Harvesting is distributed throughout the TSA in order to even feedstock transport costs from year to year

Full size image

where 0.5 is 50 % of TSA that is SHA, 0.9 is 90 % of SHA that will be harvested, and 0.1 is 10 % of SHA that is harvested each year. Figure 1 illustrates the first 2 years of such a harvest schedule and is designed to show that in a TSA divided into 612 cells, 28 cells (colored orange), or 4.5 % of the TSA, would be targeted for harvest in the first year. The harvested areas would likely be scattered around the TSA to even transport costs over the 10-year harvest cycle.

Regarding harvesting costs, currently, specialized harvesting equipment for mesquite and similar rangeland shrubs does not exist although designs have been attempted in the past [27]. Traditional forestry equipment is designed for cutting and grappling single-stemmed trees such as pine and poplar and is not suitable for harvesting a more diffuse growth form of mesquite [9]. Thus, this portion of the economic model is based on the published literature on SRWCs [28] and industry estimates (Richard Frailey, Brush Unlimited, LLC; personal communication). Harvest costs were determined using the following rationale: We used the harvest cost $587 ha−1 from a recent paper about willow (Salix spp.) trees in a SRWC plantation [28]. Realizing that mesquite wood is much harder (more heartwood) and distribution more scattered, we assumed a 40 % greater harvest cost. Thus, our initial harvest cost baseline for a single harvest was $978 ha−1. It was also assumed that trees would be harvested and coarse-chipped in the field and directly deposited into a transport truck for delivery to the bioelectricity plant. In-the-field chipping is currently considered the most cost-effective harvesting system for recovering forest residue for biomass [29]. Our study assumed a vertical integration in that that the bioelectricity plant would supply its own crew and machinery for feedstock harvest and transport [14].

Transport costs compose a significant portion of total cost of delivered biomass [16]. Mapemba et al. [11] estimated the transport cost of switchgrass ranged from $8.20 to $21.31 Mg−1, while Graham et al. [30] calculated average delivering costs ranging from $5.62 to $8.45 Mg−1. Bennett and Anex [20] showed that transport costs of sweet sorghum varied significantly from $33 to $71 Mg−1 of fermentable carbohydrates. We used estimates of transport costs from Ashton et al. [29] and Khanna et al. [31]; total transport cost per Mg load = 1.102 + $0.082 × round trip distance in kilometers assuming an average haul rate of $0.082 Mg−1 km−1 after the drivers’ waiting time for loading and unloading was considered. Round trip distance was calculated as 2× the radius of the TSA, with the assumption that the TSA area was circular. Regarding feedstock storage costs because of potential for year-round harvest, we assumed that 1 ha would be sufficient land area for storage of woody biomass near a bioelectricity plant. Storage cost of the wood chips was assumed to be $0.49 Mg−1 [32].

Regarding land rental, we assume that the TSA would be comprised of multiple landowners that would agree through long-term leases to allow mesquite to be harvested and would also allow mesquite regrowth to occur for future harvests. The landowners would benefit from the clearing of pastures and increased grass yields for livestock for at least 7 of the 10 years but would have to tolerate relatively tall (approximately 2 m) mesquite regrowth and a leveling off of grass production in the last 2–3 years prior to harvesting the regrowth every 10 years [9]. However, this negative effect would be diluted when considering the total system because, at any single point in time there, would always be a larger portion of all the areas that had been harvested over multiple years that would have lower woody regrowth and greater grass growth. The land rental fee as assumed to be $10.83 ha−1 year−1 based on two variables such as annual grazing revenue ($130/head) and the stocking rate (12 ha/head) [32].

Sensitivity analyses using the baseline parameters consisted first of a linear variation of each parameter either 25 % up or 25 % down, with other factors remaining unchanged (Table 1). Following this, we developed a “best case” and “worst case” scenario as multiple deviations from the baseline values. The best case scenario included a greater dry biomass density of 27.50 Mg ha−1, a lower annual biomass consumption of 187,500 Mg year−1, a greater percent of the TSA that is SHA (62.50 %), a lower harvesting cost of $733.75 ha−1, a shorter harvesting interval of 7.5 year, and a lower land rent of $8.12 ha−1. The worst case scenario included a lower dry biomass density of 16.50 Mg ha−1, a greater annual biomass consumption of 312,500 Mg year−1, a lower percent of the TSA that is SHA (37.5 %), an increased harvesting cost of$1,222.91 ha−1, a longer harvesting interval of 12.5 years, and an increased land rent of $13.54 ha−1. A third linear sensitivity analysis involved varying the amount of wildlife habitat that was protected from harvest, ranging from 10 to 50 % of TSA, with all other factors at either baseline, best case, or worst case levels.

Monte Carlo Site Contrast Simulation

The Monte Carlo cost simulation model analyzed cost estimates of mesquite feedstock production for two 1,600 km2 locations (Sites 1 and 2) in Texas in which mesquite biomass density had previously been determined [33]. Site 1 had moderate and site 2 had a high biomass density (Fig. 2). Each area was divided into 64 cells of 25 km2 each and oven dry woody biomass density (Mg ha−1) was determined in each cell based on a remote sensing geographic information system (GIS)-based woody canopy cover/biomass regression method developed in our lab [33, 34]. Specific details of how biomass density was determined are beyond the scope of this paper; we use these data here only to provide two actual areas differing in woody biomass density and distribution for the model.

Fig. 2
figure 2

Map showing mesquite biomass density in two 1,600 km2 regions of Texas (left panels), with a lower density region as site 1 (top) and a higher density region as site 2 (bottom). Each region has been divided into 64, 5 × 5 km grid cells. Cells are colored coded to represent different ranges of mesquite biomass density. Mean biomass (all 64 cells) for site 1 is 13.80 [standard deviation (SD), 6.04] and site 2 is 18.62 (SD, 10.50) Mg ha−1. Calculation of biomass density was based in part on mesquite canopy cover determined for each cell. Examples of classified canopy cover (in green) in 5 × 5 km cells in each biomass density level are shown at right

Full size image

The model incorporated key variables including mesquite biomass density, annual biomass consumption by the bioelectricity plant, TSA needed to supply a sufficient amount of feedstock, harvesting costs, harvesting interval, and land rent. The result is a distribution for cost of delivered mesquite biomass, the key output variable that is important for evaluating economic viability and risk associated with future business decisions.

This simulation was conducted using SIMETAR software [35], a simulation and risk analysis software to empirically construct the probability distribution of the biomass density for the two locations. The simulation model assumptions are presented in Table 2. The GRKS distribution was developed by Gray, Richardson, Klose, and Schuman to simulate subjective probability distributions based on minimal input data. The GRKS is similar to a triangular distribution because both distributions use a minimum, middle, and maximum value to describe the population. However, the GRKS can simulate low probability events, which is beyond bounds (minimum and maximum). The GRKS distribution has been used in a recent research on the mobile fast pyrolysis system [35]. The probability distributions of mesquite biomass density were used, along with the information from Table 3, to calculate the distribution of total cost of delivered mesquite biomass. To address how biomass distribution in addition to overall density may affect risk, two scenarios (A and B) were used in the simulation analysis. Scenario A excluded cells with <5 Mg ha−1 (resulting in 64 eligible cells for site 1 and 58 cells for site 2), and scenario B excluded cells with <10 Mg ha−1 (resulting in 43 cells for site 1and 49 cells for site 2).

Table 2 Monte Carlo simulation model assumptions for two locations (central and southern Texas)
Full size table
Table 3 Estimate of costs of delivered mesquite biomass and total plant costs for the bioelectricity production in the South Central USA from the base model
Full size table

Results

Budgeting: Base Model

Estimates of base model costs of delivered mesquite biomass to a bioelectricity plant are presented in Table 3. With an assumed annual biomass consumption of 250,000 Mg year−1, the base model determined that the daily biomass supply was 714 Mg and the total system area (TSA = SHA + UHA) needed to provide this daily biomass supply was 252,525 ha. Assuming that the TSA shape was circular and had a maximum radius of 28 km, the maximum round-trip distance would be 57 km, and the total transport cost based on this distance was determined to be $5.81 Mg−1. This transport cost would obviously be lower each year because many mesquite harvest areas within the TSA would be closer to the bioelectricity plant as shown in Fig. 1. For example, if the annually harvested SHAs within the TSA are divided by three radii (28, 14, and 7 km), total transport cost can be reduced by 34 % to $3.85 Mg−1. This is the more realistic scenario as harvesting would be staggered throughout the TSA to even out transport costs from year to year.

Our transport cost value is lower than estimates in some previous studies on switchgrass including $8.80 Mg−1 in Epplin [14], $7.33 Mg−1 in Bhat et al. [36], and $8.20–21.31 Mg−1 in Mapemba et al. [11], but is similar to estimates in other studies like Graham et al. [30] ($5.62 Mg−1 for switchgrass), and Aravindhakshan et al. [16] ($5.64 Mg−1 for switchgrass and $5.65 Mg−1 for miscanthus). When compared to SRWCs, this value is slightly higher than total transport cost ($5.10 Mg−1) reported by Buchholz and Volk [28].

The total cost of delivered mesquite biomass was determined to be $51.26 Mg−1 after harvest costs, storage costs, and land rents were accounted. The budget reflects a cost of $5.81 for transport, $44.47 for harvesting and chipping, and $0.49 for rent and storage. Approximately, 87 % of the estimated total cost is for harvesting, about 11 % for transport, and 1 % for the land rent. The 86 % of the total delivered cost for harvesting was higher than the 32 % found for switchgrass [14], and the 53 % for switchgrass and 46 % for miscanthus [16]. The total delivered cost of mesquite biomass in the base model is higher than findings of previous studies on switchgrass (McLaughlin and Kszos [37], $44 Mg−1; Epplin [14], $37.08 Mg−1; Aravindhakshan et al. [16], $ 43.9 Mg−1).

Comparing energy content, if mesquite wood is 19.7 MJ kg−1 (or 19,700 MJ Mg−1) and switchgrass is 15.8 MJ kg−1 (15,800 MJ Mg−1; from [16]) and the total delivery cost for each is $49.54 Mg−1, then the cost per megajoule for mesquite is ($49.54/Mg) × (Mg/19,700 MJ) = $0.0025 MJ−1, and for switchgrass is ($49.54/Mg) × (15,800 MJ) = $0.0031 MJ−1. Using different delivery costs, if mesquite is $51.26 Mg−1 (our study) and switchgrass is $ 43.9 Mg−1 (from [16]), then the cost per megajoule is $0.0025 MJ−1 for mesquite and $0.0028 MJ−1 for switchgrass.

Budgeting: Sensitivity Analysis

Biomass Density

Summary of results for sensitivity analysis is presented in Table 4. For the base model, the biomass density was assumed to be 22 Mg ha−1. This value was changed to 27.50 and 16.50 Mg ha−1 for the 25 % increase and 25 % decrease scenarios, respectively. For the 25 % increase case, the TSA needed for each bioelectricity plant decreased by 20 % to 202,020 ha and transport cost decreased by 9 % to $5.31 Mg−1. There was also a decrease in the delivered biomass cost by 19 %. With the 25 % decrease case, the TSA increased by 33 % to 336,700 ha and transport cost increased from $5.81 to $6.54 Mg−1. Cost of delivered biomass increased by 31 % from $51.26 to $67.14 Mg−1.

Table 4 Summary of results for each scenario using the budgeting
Full size table

Annual Biomass Consumption

For the base model, it was assumed that annual biomass consumption was 250,000 Mg year−1. In the 25 % higher biomass consumption scenario (312,500 Mg year−1), the TSA increased by 25 % from 252,525 to 315,657 ha. Transport cost increased from $5.81 to $6.36 Mg−1, and the cost of delivered biomass slightly increased from $51.26 to $51.82 Mg−1. With a 25 % decrease in consumption (to 187,500 Mg year−1), the TSA decreased to 189,394 ha and transport cost decreased by 11 % from $5.81 to $5.18 Mg−1. There was a minor decrease in cost of delivered biomass from $51.26 to 50.63 Mg−1.

Suitable Harvest Area

The base model assumed that 50 % of the TSA contained SHA mesquite. With the 25 % up scenario (SHA = 62.5 %), there was a decrease in TSA, transport cost, and cost of delivered biomass. With the 25 % down scenario (SHA = 37.5 %), the TSA, transport cost, and cost of delivered biomass increased by 33, 12, and 1 %, respectively. As expected, the changes in SHA strongly affected land-related factors such as TSA and transport cost.

Harvesting Cost and Harvesting Intervals

With the 25 % up and down scenarios for harvesting costs, only the cost of delivered biomass changed by 22 %. There was no change in land associated factors including TSA and transport cost. Regarding changes in harvesting interval, with a 25 % up case (12.5 years), the TSA and transport cost increased by 25 and 9 %, respectively. There was a minor increase in cost of delivered biomass. With a 25 % down case (7.5 years), TSA and transport cost decreased by 25 and 11 %, respectively. However, there was a slight decrease in cost of delivered biomass by 1 %. A factor not considered in the current analysis related to harvest interval is that with a longer harvest interval, there would likely be a greater amount of standing mass to harvest, although this would be dependent on a variety of factors including soil type and precipitation [9, 10, 38]. Thus, a greater biomass density would reduce TSA. Ansley et al. [9] indicated that on a clay loam soil type considered typical for mesquite, regrowth biomass accumulation follows a power curve for at least the first 15 years, so the gain in biomass density by waiting 12.5 years to harvest vs. 10 years is considerable. These integrated factors will be considered as a more complete sensitivity model is constructed.

Land Rent

Change in the land rent either up or down 25 % slightly impacted cost of delivered biomass. There was no change in TSA or transport cost.

Budgeting: Best and Worst Case Scenarios

Under the best case scenario, TSA decreased by 64 % to 90,908 ha, transport cost decreased by 32 % from $5.81 to $3.93 Mg−1, and cost of delivered biomass decreased by 39 % from $51.26 to $31.20 Mg−1 when compared to the baseline (Table 5). Under the worst case scenario, the TSA increased by 178 % to 701,459 ha, transport cost increased by 54 % to $8.95 Mg−1, and cost of delivered biomass increased 64 % to $84.05 Mg−1.

Table 5 Result of best and worst scenarios
Full size table

Budgeting: Unharvested Area for Wildlife Habitat

Transport cost and costs of delivered biomass increased as more areas were left unharvested for wildlife habitat. In the baseline model, the maximum increase in unharvested area from 10 to 50 % of TSA resulted in an 80 % increase in TSA, a 28 % increase in transport costs, and a 3.14 % increase in cost of delivered biomass (Table 6). While costs of delivered biomass did not increase much, the very large increase in TSA could affect the number of landowners involved and thereby increase the complexity of long-term contract negotiations. In the best and worst case scenarios, the increase in habitat protection from 10 to 50 % resulted in a 3.08 and 3.16 % increase in cost of delivered biomass, respectively.

Table 6 Results of changes in amount of unharvested area left for wildlife habitat as a percentage of TSA
Full size table

Monte Carlo Site Contrast Simulation

Results of the SIMETAR simulation of costs of delivered biomass at two sites (moderate and high biomass density) across two harvest scenarios (A and B) are presented in Table 7. In harvest scenario A, which excluded cells with <5 Mg ha−1, the higher biomass density at the higher biomass site (site 2) had a lower mean of cost of delivered biomass than did site 1, as we expected. The variation of cost of biomass, which is expressed as a standard deviation in Table 7, was found to be similar on both sites, although site 2 had larger range of costs ($20 to $200 Mg−1) than did site 1. In harvest scenario B, which excluded cells with <10 Mg ha−1, site 2 had lower mean and variation of costs and a lower range of biomass costs compared to site 1 as both the amount of biomass density and uniformity of biomass density distribution across the harvested cells increased to a greater degree on site 2 than on site 1.

Table 7 Summary statistics for cost of delivered mesquite biomass ($ Mg−1) from a moderate (site 1) and high (site 2) biomass density site and under two harvest scenarios
Full size table

In addition to looking at the values of biomass costs in Table 7, it is also critical to look at the distributions of biomass costs to evaluate the risk associated with each site and harvest scenario. Simulation results for both sites and harvest scenarios (A and B) are presented as cumulative distribution functions (CDFs) of cost of delivered biomass ($ Mg−1) in Fig. 3. The CDF graphs on the vertical axis the probability of costs being less than (or equal to) a particular level of cost on the horizontal axis. Two panels are included in Fig. 3 to show the CDFs of the two sites and two scenarios. In both harvest scenarios, site 2 has a higher probability of lower costs and thus lower risk of mesquite biomass delivery than does site 1. Under harvest scenario B, the CDFs at site 2 are steeper, exhibiting a smaller range in costs and risks because this site had greater and more spatially consistent biomass density, compared to the site 1 after all cells with <10 Mg ha−1 were eliminated from harvest.

Fig. 3
figure 3

Results of the SIMETAR simulation model showing the cumulative distribution functions (CDFs) of cost of delivered biomass on two sites that vary in biomass density. The left panel shows when cells with <5 Mg ha−1 (blue cells in Fig. 2) were excluded from harvest. The right panel shows when the cells yielding <10 Mg ha−1 (blue and green cells in Fig. 2) were excluded from harvest

Full size image

Discussion

Our analyses indicated that mesquite biomass density and harvesting costs are major factors affecting cost of delivered biomass. Annual biomass consumption of the bioelectricity plant and the percentage of the total system area (TSA) that contains suitable harvest areas (SHA) strongly affect land-related factors including TSA and feedstock transport costs, but have minor effects on cost of delivered biomass. The energy content or heating value of mesquite wood lowers delivered cost per megajoule compared to other feedstocks such as switchgrass.

Our study is based on several key assumptions: (1) that the bioenergy industry could use rangeland biomass though a system where harvesting and transport would be vertically integrated, (2) that landowners would be willing to participate in a long-term contract with bioelectricity plants for a sustainable flow of biomass to the bioelectricity plant, and (3) that the location of the bioelectricity plant would be centered relative to the location of the feedstock source. The first assumption is based on a recommendation by Epplin [14] for harvest and transport of bioenergy crops. Another option would be to subcontract the harvesting operation, but this would be on a continual year-round basis and we believe would ultimately be more expensive. However, similar to custom cutters for wheat, if the industry grows large enough, specialized businesses that are specifically dedicated to harvesting and transporting feedstock would likely operate more efficiently than if each bioelectricity company were to maintain their own harvesting equipment. The nature of this feedstock material and the remote locations of the feedstock create a higher probability for mechanical failure and downtime that may be better absorbed by businesses that are solely dedicated to this task.

The second assumption is based on observations that many if not most ranchers in this region who are in the livestock business would prefer to reduce woody plant invasion in order to facilitate livestock production through increased grass (forage) growth [22, 23, 38]. The recent increase in hunting leases on Texas rangelands is in part related to woody plant invasion that has caused an increase in cover habitat for many game animals. However, the hunting lease option for many ranchers is more the result of necessity than choice because of declining income from cattle due to woody plant reduction of grass forage availability. A woody harvesting system where a small portion of a property is harvested and not reharvested for another 10 years (as shown in Fig. 1) we believe would be preferred by many ranchers compared to the continual management of large numbers of hunter clientele required to maintain an adequate income stream.

A condition associated with mesquite biomass that differs from dedicated energy crops is the business arrangement that would have to be made with landowners. The total feedstock required in a sustainable feedstock production plan would likely cover a large enough region as to include numerous landowners. The landowners would have to agree to have the woody plants harvested according to a long-term strategic plan that would include allowing these plants to resprout and grow to be harvested again in the future. Participation in such a plan would yield several benefits to the landowner. First, he/she would receive a land rent payment, paid by the bioelectricity plant. Second, the bioelectricity plant would harvest the wood at no cost to the landowner. Third, the landowner would receive several years of improvements in pasture visibility that would reduce livestock handling costs and enhanced grass production that would increase ranch income through increased livestock numbers [38]. Fourth, the landowner would potentially receive increased income from wildlife hunting leases if the woody plants were harvested in selected patterns that would enhance wildlife habitat and hunter access.

The locations and timing of where and when sections of woody areas would be harvested would have to be negotiated as part of a long-term lease in advance between the landowners and the bioelectricity plant. We realize that there may be some landowners within a target region that would not want to participate. The primary concern is that due to the large land area required, multiple landowners would very likely be involved and a biomass harvest and transport schedule that would include consideration of such factors as allowing time for adequate regrowth before reharvest, avoiding sensitive areas, coordinating with leased hunting schedules, etc., would all have to be negotiated with each landowner and agreed upon in a binding contract before operations began. A GIS-based long-term harvest plan would be essential toward this end. The best and worst case scenarios shown in this study clearly illustrate the extreme range in size of TSA that is possible, and this would significantly affect the number of potential landowners

The third assumption is based on the fact that feedstock transport cost can be much higher in a variably distributed feedstock source such as mesquite (or other rangeland shrubs) than in row-crop bioenergy systems, and it is therefore important for the bioelectricity plant to be located as close to the feedstock source as possible. Indeed, transport costs increased considerably when more land area was added to the TSA as a result of increased percent of unharvested areas for wildlife habitat. In addition, simulation results based on actual biomass densities determined for two sites in Texas showed, as expected, that a higher and more spatially consistent biomass density would be important factors in selecting a potential location for a bioelectricity plant and reducing risk of encountering high feedstock delivery costs.

Additional Considerations and Future Models

Any industry plan that utilizes mesquite as a biomass feedstock must simultaneously consider the factors presented in this study plus additional factors that we have not addressed in detail. These may include ecological considerations, trade-offs associated with reharvest schedules, options for comixing feedstock with other woody species, variability in harvest costs due to soil type and topography, etc. For example, the best and worst case scenarios presented in this study do not account for the possible changes in biomass density or wood energy content due to changes in harvest intervals. The longer harvest interval portrayed in the worst case scenario may not increase TSA as much as we projected if this resulted in an increased biomass density, and it may not increase cost of delivered biomass as much as projected if the energy content of the wood increased due to more heartwood in the regrowth [9]. This dynamic needs to be incorporated into more complex models in future research.

Regarding ecological impacts, in most situations harvesting for biomass will affect wildlife habitat. Thus, it will be important to consider leaving unharvested portions in order to retain habitat for species that require a woodland habitat. However, the amount needed is not well defined in the literature, and the starting value of 10 % we used in this paper is strictly an arbitrarily determined value. We varied this percentage to provide ranges of some economic factors, but a detailed treatment of potential ecological effects is beyond the scope of this paper. It is important to recognize that most of the level or nearly level rangeland areas in Texas, Oklahoma, and New Mexico that are now dominated by mesquite and similar woody plants were once grasslands or open savannas and that the process of woody invasion has destroyed much of this grassland habitat [5, 22, 38]. Thus, a woody plant harvesting industry would to some degree help restore these threatened vegetation types and their associated wildlife that have been displaced by woodlands.

Additional economic analyses are necessary to evaluate whether mesquite biomass is an economically feasible biomass feed stock in the region. If rangeland woody biomass is determined to be economically feasible in certain regions, further studies are needed to determine the optimal number and location of bioelectricity plants by simultaneously considering feedstock distribution, cost of feedstock delivery, cost of bioelectricity plants, values of end products, values of byproducts of the energy production process, and ecological considerations. An integration of biomass density under different harvesting intervals is also needed in future modeling exercises.