Ethanol production potential from conservation buffers in the inland Pacific Northwest

The Title II, Energy Security in the Energy Independence and Security Act (EISA) of 2007, required an accounting of lignocellulosic feedstocks to increase the production of biofuels.¹ The goals set by EISA called for 3.79 Gl of biofuel to be produced from lignocellulosic feedstocks by 2013, 6.62 Gl by 2014, and 60.57 Gl by 2022.¹ In practice, only 0.28 million liters (ML), or 0.007% of that called for by EISA, were produced in 2013.² The causes for this shortfall are numerous, spanning economic, political, and technical issues. However, commercial scale production of lignocellulosic ethanol was expected to begin in 2014,³ east of the 100th meridian.³ Research and development have concentrated there because of abundant precipitation (≥510 mm) per year, fertile soils, highly productive plant communities, and thus, an abundant supply of raw materials.^4–6 

In the west of the 100th meridian, particularly in the northwestern inter-mountain region between the Cascade and Rocky Mountain ranges, biomass productivity is limited by low precipitation and soil fertility. Dryland crop agriculture in this region is limited to an extent, with the exception of the interior Pacific Northwest (PNW) where native shrub-steppe transitions into prairie. With a Mediterranean climate characterized by fall, winter, and spring precipitations with cool temperatures followed by dry, hot summers, the conditions are ideal for C₃ perennial and annual small grain (SG) grasses. These conditions, combined with low mean minimum temperatures (8–10 °C) during active growth, are limiting for the C₄ grasses, e.g., switchgrass (Panicum virgatum L.) or Miscanthus (Miscanthus sp.) typically considered for lignocellulosic feedstock.^7–10 Whereas productive stands of C₄ grasses can be grown on millions of hectares east of the 100th meridian including on marginally productive land,¹¹ C₃ grass productivity is relatively low across the intermountain west, particularly in land enrolled in various conservation reserve programs. These former croplands have soils that are either drought or saline affected, sometimes shallow and rocky, and are often difficult to harvest because of steep terrain. Consequently, the crop biomass produced there is not abundant enough to warrant harvesting, except perhaps by grazing livestock.

Adequate production to warrant harvesting for lignocellulosic feedstock in the PNW might exist in waterways and streams with perennial grass buffers. Plant material typically considered forage, e.g., grasses, forbs, legumes, is generally considered a secondary, supplemental source of lignocellulosic feedstock compared to materials not seasonally limited in abundance.^5,12 As a supplemental source, annual or semi-annual harvests might be possible without damaging ecosystem functions such as filtering sediment from overland flow and nutrients from the vadose zone.

Early national estimates for biofuel production from agriculture were based on the USDA—National Resource Inventory, a county by county reporting of land use acreage and agricultural production.^11,13 Our objective was to provide a physically based estimate of biofuel lignocellulosic feedstock production, specifically from multi-species stream buffers as they are planted in the dryland cropping regions of northern Oregon, central and eastern Washington, and northwestern Idaho. Stream buffers installed in production fields occupy the part of the landscape formally populated with riparian vegetation, are the most productive areas in the arid and semiarid PNW,¹⁴ and have the greatest potential to produce sufficient biomass to make harvest feasible. Based on the values obtained for lignocellulosic production by Williams et al.,¹⁵ we compared the productivity of these sites to the productivity reported in the published literature for other areas growing grass feedstocks, and estimated the potential ethanol production in stream buffers across the inland PNW precipitation gradient.^16,17 Finally, we estimated the areal extent on which these feedstocks could be grown and potential ethanol production based on the potential buffer areas defined by the range productivity mapped in the county level surveys produced by the United States Department of Agriculture–Natural Resource Conservation Service (NRCS).

We established six research sites with simulated 50 m stream buffers in areas adjacent to croplands in 2008. Dryland agriculture conditions and practices in the interior PNW fall within in three precipitation zones, low <300 mm yr^−l, mid 300–450 mm yr^−l, and high >450 mm yr^−l.^16,18 Two sites were located along intermittent or ephemeral stream channels within each of these precipitation zones (Table I).

Sites were located within the low and mid zones in northeastern Oregon and within the high precipitation zone in east-central Washington (Figure 1).

Crop production in this region is on a plateau shaped by the sheet flows of Columbia River Basalt and subsequent fault development. River and stream networks developed on this dissected surface and in the overlying loess deposits. Cultivation for crop production eliminated the riparian habitat in the low gradient streams, with crops planted across ephemeral and intermittent channels, and up to the edge of flowing water in perennial streams. A century of farming practices that led to highly erodible soil and rapid runoff created incised stream channels; when not filled by cultivation, vary in depth from 1 m to 3 m.

We established stream buffers on abandoned terraces adjacent to incised stream channels in five of six sites. At the sixth site, located in the high precipitation zone, plots were established in a wet meadow beside a zero-order intermittent channel. Soils are derived from loess parent material (Table I) with a particle size distribution consisting of sandy loam at the driest site and silt loam at the other five sites.^19,20 This region has a semiarid Mediterranean climate with 70% of mean annual precipitation falling between November and April within each precipitation zone.

We evaluated four treatments: (1) small grains (SG) barley (Hordeum vulgare L.) or wheat (Triticum aestivum L.), (2) alfalfa (ALF) (Medicago sativa L., cultivar Ladak), (3) tall wheatgrass (TWG) (Agropyron elongatum Podp., cultivar Alkar), and (4) and a mix of alfalfa and tall wheatgrass (MIX). Because these sites represent areas that once were or still are in small grain production, small grain systems appropriate to each precipitation zone were included as treatments to serve as a comparison to traditional management productivity. In the low and mid precipitation zones, the treatment was winter wheat—summer fallow (a two year rotation), which is typical for northeastern Oregon. In the high precipitation zone where annual small grain cropping is typically practiced, the treatment was a rotation of winter wheat and spring barley. We chose alfalfa and tall wheatgrass for the perennial crop treatments because the former is commonly grown in the valley bottoms of this region for forage and the latter has been recognized as adapted to and productive in the intermountain region.^22,23 Tall wheatgrass is recommended for rehabilitation of riparian and wetland areas and commonly used in conservation buffers in the intermountain west²⁴ and has potential as a lignocellulosic feedstock.²⁵ The mix of alfalfa and tall wheatgrass was included to determine if the alfalfa would enhance tall wheatgrass productivity. Site preparation and plot management details are presented by Williams et al.¹⁵ 

Lignocellulosic feedstock biomass in the perennial plots was mechanically harvested one time each year; September of 2010, September and October of 2011, and August and September of 2012. This harvest schedule was meant to maximize the conservation benefits for wildlife, particularly avian nesting, rearing, and fledging requirements. Samples were weighed in the field and subsamples of that material returned to the laboratory, oven dried at 65 °C for 48 h, and weighed. Potential ethanol production was determined by multiplying the sum of the calculated feature areas for each precipitation zone times harvested dry biomass times the liters-per-dry-matter values determined by Kumar et al.¹⁷ for TWG and ALF. Because the MIX treatment was not specifically evaluated for liter-per-dry-matter, ethanol yield was based on the conversion and dry matter species-to-species ratios. Although not as stringent an estimate as that provided by Kumar et al.,¹⁷ ethanol yields from mixed species feedstocks have been shown to produce negligible differences from homogeneous feedstocks relative to their differences in dry matter yields.²⁶ 

At each site, the different crop treatments were planted into 3.8 m × 50 m plots in a randomized complete block (RCB) design with four replications (Figure 2). Biomass data were analyzed using a generalized linear mixed model (GLM) analysis of variance (ANOVA).^27–29 Significant differences among treatments were accepted at P ≤ 0.05. Data were tested for assumptions of normality using quantile-quantile plots and transformed as necessary. Data were evaluated to determine whether they fit a Poisson distribution and adjustment made to the GLM as recommended by Gbur et al.²⁹ Mean separation was conducted using least square means and Tukey tests.

Using the best information available for the interior PNW, NRCS soil surveys,^19,20 we then identified the extent of the areas similar to where our research plots were located to develop an estimate of regional conservation buffer biofuel feedstock production. We assumed that areas mapped with the same soil types and precipitation were capable of producing biomass quantities equal to the values we recorded in our test plots in accordance with the estimates of forage availability that have been made for western rangelands. The normalized difference vegetation index (NDVI) derived from remote sensing data is useful for estimating biomass, but would be problematic for identifying similar areas because of the mix of management types adjacent to these stream channels, where the management varies from active crop production (predominately small grains alternating with fallow) to abandoned ground given over to weeds and to marginally managed conservation buffers.

Using the ModelBuilder environment in the ESRI ArcGIS Desktop 10.3 geographic information system (GIS) software package, a methodology for defining potential buffer areas was developed (Figure 3). Features containing typical dryland crops for the region were selected from the United States Department of Agriculture–National Agricultural Statistics Service (NASS) Cropland Data Layer (CDL) for the year 2012 and included in the area of interest (AOI) for analysis. These features combine satellite imagery with the extensive ground reference data.^30–34 A generalized AOI was obtained by associating predominant areas of agriculture of dryland crops in the NASS 2012 CDL with specific units of the US Environmental Protection Agency (EPA) level IV ecoregions.³⁵ Ecoregions included in this analysis were Palouse Hills, Okanogan Drift Hills, Loess Islands, Yakima Folds, Pleistocene Lake Basins, Deep Loess Foothills, Nez Perce Prairie, Dissected Loess Uplands, Umatilla Plateau, and Grassy Potlatch Ridges. Excluded from the analysis were Native American Indian lands, military installations such as the Yakima Training Facilities, the Hanford Nuclear Reservation, National Wildlife Refuges areas, and the Channeled Scablands.³⁶ 

Stream length features within the region were obtained from the USDI-GS national hydrography dataset (NHD)³⁷ high resolution (1:24 000 or higher) flowlines. The soil data layer was obtained for the following county soil surveys:³⁸ (in Idaho) Idaho, Lewis, Nez Perce, and Latah; (in Oregon) Gilliam, Morrow, Sherman, Umatilla, and Wasco; (in Washington) Adams, Asotin, Benton, Chelan, Columbia, Douglas, Franklin, Garfield, Grant, Kittitas, Klickitat, Lincoln, Spokane, Walla Walla, Whitman, and Yakima. The county soil survey databases and associated soil survey geographic database (SSURGO) level (1:24 000) mapping were used in conjunction with the NRCS soil data viewer (SDV, v.6.1).¹³ The SDV was developed by NRCS to allow the end users to build and interpret thematic maps based on the data within more than 50 tables of each county soil database. The NRCS defines rangeland productivity in terms of favorable, normal, and unfavorable years. To provide a conservative estimate of production, a lower limit of 1.12 Mg ha⁻¹ was first used to identify the potential stream buffers in an unfavorable year. We reduced this value to 1.01 Mg ha⁻¹ to remove the production value differences found at state and county boundaries (soil survey boundaries) for the same soil types with shared values.

Average annual precipitation values for the AOI were obtained using the data from the parameter-elevation relationships on the independent slopes model (PRISM) developed by climate group website.³⁹ The average annual precipitation 800 m gridded data layer obtained from the PRISM Climate Group was derived from a 30 year record from 1981 to 2010.

Flowline features and rangeland productivity threshold features were clipped to the AOI. The 30 year normal precipitation raster was extracted to the AOI and converted to polygon features representing each precipitation zone. Flowlines were buffered 50 m independently to the left and right, and clipped to the productivity threshold features. The remaining left and right buffers were cleaned using interior buffering to eliminate features with widths less than 50 m. A minimum production area of 1 ha was used to remove the small buffer areas that might be too small to be suitable for production purposes. The remaining left and right buffers were merged. The potential stream buffers were then clipped to each precipitation zone and the area determined. Uncertainty associated with the precipitation regime boundaries derived from the 800 m PRISM raster was conservatively quantified. The buffer tool was used to define an area of 800 m adjacent to the mid precipitation boundary. The area of proposed conservation buffer strip that fell within the precipitation zone boundary area of uncertainty was calculated and used to express a range of possible production.

Treatments TWG and MIX produced significantly more lignocellulosic feedstock than ALF and SG, with the largest treatment differences in the mid and high precipitation zones (Figure 4). There was no statistical difference in production between TWG and MIX, or between ALF and SG. In the high and mid precipitation zones, TWG and MIX produced between 8.24± Mg ha⁻¹ and 10.7 ± 0.8 Mg ha⁻¹dry biomasses, with significantly lower values from plantings in the low precipitation zone and ALF and SG in all precipitation zones.

Based on the conversion rates determined by Kumar et al.,¹⁷ the best ethanol production potential was from the TWG in the high precipitation zone (3079 ± 262 l ha⁻¹) and the MIX in the mid zone (3062 ± 235 l ha⁻¹) (Table II). Within treatments, TWG and MIX were significantly more productive in the mid and high precipitation zones than in the low precipitation zone, whereas ALF and SG were not statistically different among the precipitation zones (Table II). Within each precipitation zone, TWG and MIX were significantly greater than ALF and SG treatments (Table II). Combining alfalfa and tall wheatgrass in the MIX treatment did not improve the potential ethanol productivity (Table II).

There are approximately 77 504 stream km in the rainfed cropland of the interior PNW. Along these streams, we identified a total of 55 120 ha meeting our criteria for 50 m streamside buffers and lignocellulosic feedstock production. Taking into account the uncertainty associated with the precipitation zone boundaries, the areas available for production in each of the three precipitation zones are dry (10 464 ± 704 ha), mid (20 963 ± 1916 ha), and high (23 693 ± 1221 ha). In the mid and high precipitation zones where the productivity is the most promising, there are 47 753 stream km or 44 656 ha that meet our criteria. These areas represent 1.1% and 2.4%, respectively, of the area in the mid and high precipitation zones.

The highest potential production of ethanol obtainable was from the biomass in the mid and high precipitation zones with the highest producing treatments MIX and TWG. Using the mid-range areas estimates applied to MIX and TWG, potential production of ethanol varied from 62.91 ± 5.37 ML in the MIX treatment to 72.95 ± 6.21 ML in the TWG treatment per year (Table III).

The sites in the mid and high precipitation zones where we recorded the highest production values were adjacent to ephemeral channels. Productivity of TWG and MIX in the mid and high precipitation zones met or exceeded the 9.4 Mg ha⁻¹ level needed to attract the producer's interest.⁴⁰ On a per unit area basis, dry biomass production of TWG and MIX in the mid and high precipitation zones compares favorably with the rainfed feedstocks grown in other regions of the northern hemisphere, i.e., the U.S. and Europe (Table IV). The combined potential ethanol production from the mid and high precipitation zones, based on the mid-range areas, is approximately 145 ML (mid MIX + high TWG). This volume is 0.3% of the 60 × 10⁹ l lignocellulosic ethanol production mandated by the year 2022.¹ 

Why, then, should we consider this resource? Despite the promising productivity of the TWG and MIX treatments, the use of the material grown in stream buffers for biofuel in this region faces a number of challenges. Those challenges are: (1) available facilities to process lignocellulosic feedstocks, (2) continuity of land ownership/management, and (3) current law concerning use of materials from federally contracted buffer zones. The answer to this question lies within our ability to address these challenges.

The first challenge can be addressed by development of small scale processing technology. At present, there is one industrial scale ethanol facility in this region located near Boardman, Oregon that processes lignocellulosic feedstocks (poplar trees and irrigated crop residues) obtained from local farms. The highly productive buffer sites that we identified in the mid and high precipitation zones are all >80 km (50 mile) outside the radius within which costs and energy required for feedstock transportation remain economical.⁴¹ An alternative method of alcohol production might be sub-regional, i.e., small on-farm processing facilities.^41–44 Although practical implementation of this technology has yet to be realized, material grown in stream buffers (30 × 10⁶ Mg) would be a substantial contribution to the 3.2 × 10⁶ Mg small grain straw feedstock theoretically available after leaving enough straw to maintain soil organic matter.⁴⁵ The counties with the most available residue identified by Banowetz et al.⁴⁵ are located in the mid and high precipitation zones of the region. A word of caution, however, is required in considering small grain residue availability in this region. Huggins et al.^46,47 reported that even in the high precipitation zone of this region the removal of crop residues for any purpose (biofuel feedstock, livestock forage and bedding, or mushroom production) can have negative consequences on maintenance of soil organic carbon, soil nutrient levels necessary for crop production, and soil erosion rates. The sustainable removal of crop residues depends on cropping methods and frequency. Three crops, three year rotations produced using no-tillage are the most nearly sustainable, but are only feasible in the high precipitation zone and even there are not practiced by all the producers.

There has been considerable success in establishing the stream buffers in the semi-arid croplands of the PNW, nearly 1288 km or approximately 57 870 ha, since 1999 in Washington,⁶³ and 8754 ha (stream lengths not reported) in the rainfed cropland of northeastern Oregon.⁶⁴ Although some of this contracted ground would overlap with the areas we identify, there remains approximately 46 466 km of stream length where buffers have not been but could be installed for biomass production. Annual harvest of perennial biomass from these buffers leaves substantial biomass for protection against raindrop impact and filtering of overland flow¹⁵ and has the potential of aiding in the maintenance of the species composition of the buffer.²³ 

Despite this availability, there remains the second challenge, which is the continuity of land ownership/management. Even for the small operation on farm facilities, efficient and economical operation will require contiguous stream buffers as sources of feedstock. In 2007, 29% of U.S. farmland was owned by non-operators, who also owned 77% of the rented farmland.⁶⁵ Mixed land ownership may result from the individuals trading rural for (sub)urban lifestyles, after which they may or may not relinquish their decision making to managers or family members remaining on the land. For instance, ownership of the 2.2 km long conservation buffer sampled by Williams et al.²³ was shared by five families, albeit managed by a single, conservation minded manager. But as the land is sub-divided amongst decedents of the original owners, decision making becomes more of a distributed process wherein some of the owners may have limited knowledge of farm management or conservation. Non-operators tend to be older, less likely to live on the farm, and less likely to participate in conservation programs,⁶⁵ less likely to want to be restricted by regulation,⁶⁶ and concerned about the contract lengths limiting their management options.⁶⁵ 

Land enrolled in the conservation programs would be expected to contribute lignocellulosic feedstock to meet the goals originally set out in EISA,¹¹ including land enrolled in waterway or stream buffers.⁶⁷ These areas should prove to be sufficiently productive to help meet these goals, the third challenge can be addressed by requiring legislative action to make changes in Public Law 106-78, Section 769, which currently prohibits the harvest of federally contracted conservation buffers and riparian forests for biofuel. Thus under current conditions, the establishment of stream buffers with the dual purpose of conservation and biofuel production would necessarily be the sole responsibility of the landowner or manager. Gaining feedstock for biofuel production to reduce farm operation costs could be the necessary incentive for owners concerned about restrictive contracts, but with a desire to increase their conservation efforts by reducing soil loss and stream sedimentation from fields and overbank flows, increasing infiltration of overland flow from fields, and providing avian or wildlife habitat. A thorough cost-benefit analysis would be required, but the resource, especially compared to the availability of wheat straw, is there and abundant to make a contribution to biofuel production.

Results from this study show that dry matter production rates in the mid and high precipitation zones of the inland, rainfed croplands of the PNW compare favorably to other regions where lignocellulosic biofuel feedstocks are grown. Although ethanol production from our estimated 44 656 ha of suitable land only contributes 0.3% of the 2022 goals set for lignocellulosic biofuel production, the value added to conservation buffer installation could be the incentive needed for the increased installation of conservation buffers in this region.

We thank Dr. Dave Huggins (USDA-ARS) for providing research space at the USDA Conservation and WSU Cook Farms near Pullman, Washington, and Cunningham Sheep Farms, Newtson Farms, Bracher Farms, and Bill Lorenzen for providing research sites in Oregon. We also thank students Seren Hansen and Stefka Waite, and USDA-ARS agricultural research technicians Bob Correa (Pendleton, Oregon), and Dave Uberuaga (Pullman, Washington) for field work, Joe St. Clair (chemist) for laboratory analyses, Dr. John Sulik (USDA-ARS post-doc, Pendleton, Oregon) for GIS consultation. This research funded by USDA Agricultural Research Service National Program Agricultural Competitiveness and Sustainability (NP#216). USDA is an equal opportunity provider and employer. The mention of trade names or commercial products in this publication is solely for the purpose of providing specific information and does not imply recommendation or endorsement by the U.S. Department of Agriculture.