Among nutrient inputs, nitrogen is the most critical for maintaining the productivity of established switchgrass stands. Nitrogen management and the feedbacks associated with harvest management have significant consequences for biomass yield, feedstock quality, environmental impact, and system economics. Consensus regarding appropriate recommendations for nitrogen management may be harder to find, however (Parrish and Fike

2005) . The broad range of responses to applied N is a function of inherent demand and capacity for recycling, soil type and N status, precipitation and atmospheric deposition, and harvest timing. For example, in a summary of yield responses to added N (vs. a 0-N control), Brejda (2000) found biomass yield increased from 0 to 6.2 Mg ha-1.

Long-term stand sustainability will be best supported by fertility management that replaces similar amounts of N in the harvest biomass (Lemus, Parrish, and Abaye 2008). Greater N inputs are required for

biomass (or especially forage) systems that collect multiple harvests (Parrish and Fike 2005; Fike et al. 2006a, b; Guretzky et al. 2011). Given the added costs associated with such management, one end-of-season harvest is the prevailing recommendation for bioenergy cropping. However, in some cases, multiple harvests may provide value to the system as a whole (Cundiff 1996; Fike et al. 2007; Cundiff et al. 2009) by reducing logistic constraints, and we will consider this further in a subsequent section.

Compared with many other potential energy crops, switchgrass has low nutrient demand. Although N needs during the growing season may be relatively high on a mass basis, plant N concentrations decline during the growing season (Waramit et al. 2011), and N is returned to roots and rhizomes at the end of the growing season (Beaty et al. 1978; Lemus et al. 2008; Garten et al. 2011). This ability to translocate nutrients to belowground storage structures is a major component of the apparent thriftiness of many perennial, warm season grasses (e. g., see Hargrave and Seastedt 1994). End — of-season nitrogen concentrations often are in the range of 5 to 8 g kg1 for plants harvested after senescence (Madakadze et al. 1999; Fike et al. 2006a, b; Guretzky et al. 2011). Fertility practices also affect switchgrass morphology, as plants grown at a higher level of N fertility apparently conduct a greater proportion of nutrients to shoots (vs. roots) than plants grown at a lower plane of nutrition (Heggenstaller et al. 2009; Garten et al. 2011). Nitrogen- fertilized switchgrass may also have fewer tillers, particularly under one-cut management (Fike et al. 2006a; Muir et al. 2001). These changes in plant morphology may not affect biomass yields (Muir et al. 2001) but may have consequences for carbon sequestration and greenhouse gas emissions (Garten et al. 2010). The relationships of N fertility to overall system sustainability in terms of increased biomass vs. reduced soil organic carbon stocks bears further investigation (Jung and Lal 2011).

Across regions, the data regarding switchgrass N requirements—and consequent recommendations—may seem rather disparate. Some of the greatest responses to applied N have occurred in sandy soils with little nutrient retention capacity (Ma et al. 2001; Muir et al. 2001). In contrast, Stout and Jung (1995) reported little response to N for switchgrass grown on soils with high levels of N in the soil organic pool. Along with inherent soil fertility, there is increasing evidence that bacterial-based biological nitrogen fixation and plant growth stimulation occurs with switchgrass in some settings (Tjepkema 1975; Riggs et al. 2002; Ker et al. 2010; Ker et al. 2012). Such reports help to further explain the negligible responses to N often reported for switchgrass (Parrish and Fike 2005) and increase the appeal of a plant that already gets high marks for its ability to capture, sequester, and recycle N from soils.

As input costs increase, the economics of applying fertilizer nutrients may be marginal in low-value, high-volume biomass production systems. Under such circumstances, developing management strategies with alternative nutrient sources may provide an important route to the production and economic sustainability of these systems. Several researchers have reported that animal manures can support switchgrass production (Sanderson et al. 2001; Lee et al. 2007, 2009) and Lee et al. (2007) suggest they may improve stand composition, but long-term increases in soil phosphorus and other nutrients will require monitoring. Adding legumes to these systems may be another approach for reducing N input costs. However, finding compatible species that do not reduce biomass production may be a challenge in some locations (El Hadj 2000; Springer et al. 2001; Bow et al. 2008) although some researchers have reported success with this strategy (Springer et al. 2001; Bow et al. 2008).

Plant tissues produce or regulate different hormones to respond to internal and external cues during practically every aspect of plant growth and development. Bacterial endophytes have the ability to produce plant hormones and regulate their balance as well. Auxin, a hormone associated with plant growth promotion, influences many plant cellular functions and is an important regulator of growth and development. Bacterial endophytes are commonly capable of producing auxin which, at the genetic level, may either be constituently expressed or inducible (Mattos et al. 2008). Auxin producing bacterial endophytes increased the number and length of lateral roots in wheat (Barbieri and Galli 1993). Increased root length, root surface area and the number of root tips were observed in hybrid poplar inoculated with auxin producing bacteria, resulting in enhanced uptake of nitrate and phosphorus and boosting biomass by 60% compared with non-inoculated plantlets (Taghavi et al. 2009). Furthermore, Pseudomona flourescen significantly increased the growth of maize plant radicles under laboratory conditions via the production of auxin (Montanez et al. 2012). To date, multiple auxin biosynthesis pathways have been identified in bacteria, and their regulation is controlled by several different genetic and environmental factors (Bertalan et al. 2009). The production of native auxin, indole-3-acetic acid (IAA) by bacteria has been documented in species such as Rhizobium, Pseudomonas, Azospirillum, Azotobacter and Bacillus (Hayat et al. 2010).

Cytokinins are a diverse range of compounds that, like other plant hormones, are involved in many activities of plant growth and development. As a group, they have been shown to regulate cell division, seed dormancy and germination, senescence, new bud formation, and leaf expansion. They also play roles in controlling plant organ development, mediating responses to various extrinsic factors and the response to biotic and abiotic stresses (Spichal 2012). Researchers have demonstrated that certain endophytic bacteria are able to produce cytokinins and promote lateral root growth (Senthilkumar et al. 2009). Zeatin, a native plant growth promotive hormone, belonging to the cytokinin family, has been found in significantly higher levels in the beneficial bacteria B. subtilis and P. putida (Sgroy et al. 2009).

Gibberellins are native plant growth promotive hormones. Many plant growth promoting endophytes also produce gibberellins to enhance host plant growth (Joo et al. 2009; Fernando et al. 2010). For example, one Penicillium citrinum isolate, IR-3-3 from the sand dune flora, produced higher physiologically active gibberellins and stimulated Waito-c rice and Atriplex gemelinii seedling growth (Khan et al. 2008). Gibberellic acid levels were also high in the plant associated bacteria Lysinibacillus fusiformis, Achromobacter xylosoxidans, Brevibacterium halotolerans, and Bacillus licheniformis (Sgroy et al. 2009).

Ethylene, a simple organic molecule (CH2=CH2), is commonly thought to be a growth inhibitive hormone. It is typically produced when plants are exposed to environmental stress, repressing plant growth and development until the stress disappears or the levels of ethylene decrease (Gamalero and Glick 2012). Ethylene inhibits stem elongation, promotes lateral swelling of stems, and causes stems to lose their sensitivity to gravi-trophic stimulation (Glick 2005). In biomass production as in agriculture generally, it is important to keep ethylene low in order to maximize yields. An enzyme, 1-aminocyclopropane-1-carboxylate (ACC) deaminase produced by bacteria, interferes with the physiological processes of the host plant by decreasing ethylene levels (Hardoim et al. 2008) via metabolizing ACC, a precursor to ethylene so ethylene levels are reduced in plants, and plant growth is promoted. Activity of ACC deaminase is a common feature found in plant-growth promoting bacteria such as Enterobacter, Pseudomonas and Burkholderia (Shah et al. 1998; Sessitsch et al. 2005; Govindasamy et al. 2008). Burkholderia phytofirmans strain PsJN stimulates growth of many plant species, including potato, tomato, grapevine, and switchgrass (Pillay and Nowak 1997; Nowak et al. 1998; Barka et al. 2002; Kim et al. 2012) and was reported to have a high activity of ACC deaminase (Sessitsch et al. 2005). Endophytes that produce ACC deaminase have also been shown to increase host plant growth in soils with high salinity (Egamberdieva 2012; Siddikee et al. 2012) and increase drought tolerance (Arshad et al. 2008; Belimov et al.

2009) . Pseudomonas sp. strain A3R3 showed higher ACC deaminase activity and increased plant growth in nickel contaminated soil (Ma et al. 2011).

Abscisic acid (ABA) is involved in responses to environmental stresses such as heat, drought, and salt, and is also produced by endophytes.

Endophytic bacterial strains SF2, SF3, and SF4 isolated from sunflowers (Helianthus annuus) had the ability to produce ABA and jasmonic acid, which increased under drought conditions (Forchetti et al. 2007), implying these endophytes enhance stress tolerance of host plants. Two strains of Azospirillum brasilensis, successfully used to increase the yield of maize and wheat in field conditions, were both able to produce different plant growth regulators such as IAA, gibberellic acid, zeatin and ABA (Perrig et al. 2007), highlighting the ability of endophytes to confer multiple mechanisms of growth promotion.

Matrix polysaccharides (i. e., hemicelluloses), are the second most abundant polymer in nature, and, as has been discussed, consist of heterogeneous polymers of pentoses (xylose, arabinose), hexoses (mannose, glucose, galactose), and sugar acids (Girio et al. 2010). Due to the abundance and functional importance of xylan in the cell walls of switchgrass, we focus here on the GHs that degrade xylan. However, enzymes that target every class of cell wall polysaccharide have been characterized (Table 3). Many microorganisms, such as Penicillium capsulatum and Talaromyces emersonii, possess complete degradation systems for grass glucuronoarabinoxylans (Filho et al. 1991). Like cellulose biodecomposition, total degradation of xylan also requires diverse enzymes for depolymerization and side-group cleavage. Endo-xylanases attack internal bonds in the main chains of xylans; exo-xylanases hydrolyze p-1,4-xylose linkages at chain ends to release xylobiose; and then p-xylosidase further hydrolyzes xylo-oligosaccharides and xylobiose to xylose (Gilbert et al. 2008). Side chains on xylose units block the action of some xylanases, leading to the evolution of diverse accessory enzymes to remove the side-chains and render the xylan backbone accessible for complete hydrolysis (Perez et al. 2002; Gilbert et al. 2008). Hydrolases with a-arabinofuranosidase and a-glucuronidase activities are responsible for removing arabinose and 4-O-methyl glucuronic acid substituents, respectively, from the xylan backbone. Furthermore, esterases hydrolyze linkages between xylose units and acetic acid (acetylxylan esterase) or between arabinose side chain residues and the hydroxycinnamic acids, ferulic acid (feruloylesterases) and p-coumaric acid (p-coumaric acid esterase).

Cellulosomes

The individual classes of hydrolases described above function within both non-complexed and complexed cellulase systems (Fig. 6) (Fontes et al. 2010). The non-complexed systems consist of individual polypeptides that can have multiple catalytic and CBM domains, but that otherwise act without interacting physically with other classes of hydrolases. In contrast, the complexed systems, also known as cellulosomes, are superstructural, multi-polypeptide enzyme complexes that adhere to cell walls of lignocellulolytic bacteria and fungi (Fontes et al. 2010). They consist of a multi-functional integrating subunit, called a scaffoldin, that is composed of multiple cohesion modules, and diverse enzymatic subunits with dockerin modules that interact with the scaffoldin. For example, the cellulosomes of C. cellulolyticium have the potential to contain numerous cellulases, xylanases, mannases, and even protease inhibitors (Blouzard et al. 2010).

Well-studied cellulosome-producing anaerobic bacteria include Clostridium species and Ruminococcus species (Doi et al. 2004). Cellulosome composition is dynamic and heterogenous, depending on the bacteria and composition of extracellular polysaccharides, and the relative amounts of the available dockerin-containing modules consistent with this (Raman et al. 2009). Cellulosomes have higher cellulose degradation efficiency compared with non-complexed enzymes since their adhesion to the cell surface prevents their products from being lost via diffusion or uptake by neighboring bacteria (Schwarz 2001). In vitro construction of mini — cellulosomes and self-assembly of cellulosomes on the surface of yeast significantly enhances cellulose hydrolysis compared with free enzymes (Wen et al. 2010; Fan et al. 2012; You et al. 2012).

Cellulosome-generating microorganisms also exhibit diversity in cellulosomal composition and architecture. For example, the Ruminococcus flavefaciens FD-1 genome encodes over 200 dockerin-containing proteins (Berg Miller et al. 2009); whereas, the Bacteroides cellulosolvens cellulosomes may possess more than 100 enzymes (Ding et al. 2000; Xu et al. 2004). This genomic diversity is likely functional. Proteomics of isolated cellulosomes from C. cellulolyticum confirmed the expression of 50 dockerin-containing proteins out of 62 predicted by bioinformatics (Blouzard et al. 2010). The complexity of the cellulosome is related to the availability and abundance of cellulosomal components, the expression of which is influenced by substrate induction and catabolite repression. For example, C. cellulolyticum grown on cellulose, expresses 36 cellulosome component enzymes. A partially distinct set of 30 cellulosome enzymes are detected on xylan; and 48 are expressed on wheat straw (Blouzard et al. 2010). Thus, cellulosomes are heterogeneous with varied components and stoichiometries. Moreover, some microbes exhibit even more diverse cellulosomes due to the presence of multiple types of scaffoldins within a single genome (Fontes et al. 2010). C. thermocellum contains four type II cohesion-containing anchoring scaffoldins (Bayer et al. 1998). For example, the cellulosomes assembled by the type II dockerin domain of CipA are further organized into a larger complex, called a polycellulosome, via type II cohesion-containing anchoring scaffoldins (Bayer et al. 1986; Raman et al. 2009). In short, cellulosomes generally have diverse content with heterogenous composition and architecture.

Over the course of the last decade, molecular markers have revolutionized how we view and measure genetic diversity at the DNA level. Historically, the DNA marker of choice was the microsatellite or simple-sequence repeat marker because of its simple PCR-based assay and large numbers of alleles per locus. As reference genome sequencing has become routine, a radical shift in polymorphism detection was eminent. The new marker of choice, single nucleotide polymorphism (SNP), has taken polymorphism discovery and genotyping to a completely new level. In a typical grass species like maize, the level of genetic diversity is quite high (~1 substitution per 100 bp) (Tenaillon et al. 2001), and the genome complexity is largely a result of DNA rearrangements and the captured genome space in the reference contains ~70% or less of the species-wide genome space (Gore et al. 2009). As 2nd generation costs have declined, and multiplexing options increased, a new strategy to assess genetic diversity and develop SNP markers has transformed genotype-phenotype associations (trait mapping), germplasm characterization, and molecular breeding strategies (Elshire et al. 2011). The approach, termed Genotyping-by-Sequencing (GBS) or Genomic Selection essentially reduces the complexity of the genome through digestion with one or two methylation sensitive enzymes that maximizes the amount of fragmented gDNA in the 300bp range, indexed with Illumina barcodes, and sequenced in a multiplex fashion on the Illumina HiSeq. The resulting sequences are assembled bioinformatically to produce consensus sequences flanking restriction sites that can either be used from a de novo perspective or mapped to a reference genome for SNP discovery (Baird et al. 2008). With the promise as a bioenergy feedstock and urgent need for genome enablement, a GBS approach to explore genetic diversity has the potential to immediately increase the amenability of switchgrass breeding programs. A recent study conducted by Lu et al. (2013) applies GBS to 840 individuals generating a total of 350 GB of DNA sequence. Of particular importance from these authors is the development of a pipeline called Universal Network Enabled Analysis Kit (UNEAK) tailored to enable dense SNP discovery and genomic selection in genomes without reference assemblies. UNEAK removes terminal low quality bases at the ends of reads, reads are collapsed into tags, and pairwise alignment identifies tag pairs with single base mismatches as candidate SNPs (Lu et al. 2013). In large complex genomes like switchgrass, there is an additional filter that removes tags that pair as a result of repeats, paralogs, and errors (Lu et al. 2013). In switchgrass, the authors created a full-sib linkage population of 130 individuals, a half-sib linkage population with 168 individuals, and an association panel composed of 66 diverse populations and 540 individuals and after sequencing, identified ~1.2 million putative SNPs (Lu et al. 2013). An important finding through the deep genotyping efforts revealed that tetraploid switchgrass is similar to a diploid in genomic composition (Lu et al. 2013), but further genome analysis and a more comprehensive dataset through genome resequencing and reference mapping is necessary for corroboration. Through these efforts, the authors constructed a high-quality linkage map using 3,000 of the highest quality SNPs and placed into a context of the 18 chromosomes, also guided by synteny with foxtail millet. This resource will be invaluable in advancing the genome reconstruction efforts described above.

Switchgrass is considered a prime candidate as a second generation biofuel feedstock because it can produce more ethanol per unit area and triple the net energy content than ethanol derived from corn grains (Bouton 2007). However, current estimates show that it requires 45 percent more fossil fuel energy to yield one liter of ethanol from two and a half kg of switchgrass feedstock than the energy in that one liter of ethanol fuel produced (Pimentel and Patzek 2005). The average cost to produce a liter of ethanol from switchgrass feedstock is approximately 54 cents, which is nearly nine cents higher than that for corn grains (Pimentel and Patzek 2005). One of the major cost factors in converting switchgrass feedstock into bioethanol is that of microbial enzymes, which are used to hydrolyze and break down the lignocellulosic biomass into fermentable sugars that can be used for biofuel production (Ragauskas et al. 2006). Presently, microbial hydrolysis enzymes are manufactured in large industrial bioreactors (Lynd et al. 2008). This process is extremely expensive and consequently, the cost of enzymes to produce one gallon of ethanol from lignocellulosic feedstock is roughly 30 cents per gallon (Bothast and Schlicher 2005).

In order to combat the high cost of microbial hydrolysis enzymes, current investigations are working towards expressing cell-wall degrading enzymes in important crop species. The most well studied cell-wall degrading enzymes are the cellulases, a family of enzymes that are naturally found in fungi, bacteria, and some animals (Sukumaran et al. 2005). These enzymes hydrolyze cellulose to produce glucose, cellobiose, and cellooligosaccharides (Sukumaran et al. 2005). There are three major types of cellulase enzymes: cellobiohydrolases, endo-1,4-p-glucanases, and P-glucosidases (Sukumaran et al. 2005). All three types of cellulase enzymes work collectively to break down cellulose into glucose monomer subunits that can then be fermented to yield bioethanol.

The best-studied of the cellulase enzymes is endo-1,4-p-glucanase. In 2000, Ziegler et al. inserted the catalytic domain of the endo-1,4-p-D- glucanase E1 gene (subsequently referred to as E1) from Acidothermus cellulolyticus into Arabidopsis and targeted protein localization to the apoplast (Ziegler et al. 2000). The authors were able to obtain levels of recombinant endoglucanase E1 between 0.01 to 25.7 percent of the total soluble protein (TSP). Novel zymogram assays further confirmed that the catalytic endoglucanase domain was biologically active (Ziegler et al.

2000). A similar study was performed in transgenic potato in which the entire endoglucanase E1 gene from A. cellulolyticus was targeted to mature leaves. Full-length recombinant endoglucanase E1 protein accounted for 2.6 percent of TSP in these transgenic potato plants (Dai et al. 2000a), which is an improvement over the 1.3 percent of partial endoglucanase E1 in TSP extracts of tobacco plants that were transformed using the same method (Dai et al. 2000b).

The successful expression and production of cell wall degrading enzymes in model plant species, such as Arabidopsis and tobacco, opened the door for utilizing this strategy in bioenergy crops. In 2007, Oraby et al. inserted the catalytic domain of the endoglucanase E1 gene from A. cellulolyticus into the nuclear genome of embroygenic rice calli via Agrobacterium transformation (Oraby et al. 2007). After regenerating transgenic plants, the E1 enzyme accounted for 2.4 to 4.9 percent of TSP in rice leaves. The presence of E1 also greatly enhanced the conversion of cellulose to glucose in pre-treated transgenic rice straw (Oraby et al. 2007). That same year, Ransom et al. inserted the same partial endoglucanase E1 gene, containing the catalytic domain, into corn embryogenic calluses (Ransom et al. 2007). The construct was placed under control of the cauliflower mosaic virus 35S promoter and introduced via particle bombardment (Ransom et al. 2007). Using this method, the authors were able to obtain up to 1.16 percent of biologically active recombinant endoglucanase E1 in TSP extracts (Ransom et al. 2007).

A recently published study performed by researchers from Agrivida Inc. (Medford, MA) investigated expressing two xylanase genes in maize under the direction of two different promoters (Gray et al. 2011). Xylanases are another family of cell wall degrading enzymes that act in correlation with cellulases to convert hemicellulose and cellulose into fermentable pentose sugars. The xynB gene from Clostridium stercorarium and the bsx gene from Bacillus sp. were cloned and optimized for expression in maize. After removing bacterial secretion signals, each gene was fused to two signal peptides individually: the barley a-amylase signal peptide sequence (BAASS), which targets protein accumulation to the cell wall, or the rice glutelin B-4 signal peptide (GluB4SP), which would allow for kernel-specific expression. The xylanase sequences that were fused to BAASS were placed under control of the constitutive rice rubi3 promoter, whereas the sequences that were fused to GluB4SP were directed by the rice GluB-4 gene promoter. All constructs were inserted into embryogenic calluses by Agrobacterium — mediated transformation. After transformed plants were regenerated, all of the T0 transgenic maize plants that constitutively expressed both xylanase genes displayed severely stunted growth phenotypes. In GluB4SP transgenic plants, where xylanase expression was directed to the seeds, the plants exhibited normal somatic tissue development, however, the corn grains appeared shriveled. Constitutive expression of both xylanase genes resulted in relatively low accumulation of BSX and XYB proteins in corn stover (0.1 percent TSP). Given that transgenic plants were undersized, higher levels of BSX and XYB accumulation may be lethal to the plant. However, seed specific expression of BSX and XYB resulted in up to four and 16.4 percent TSP, respectively. Presently, further research is being conducted to control xylanase activity and expression in an effort to prohibit negative growth phenotypes associated with expression of these genes in maize. In another case, a gene encoding a thermostable GH10 xylanase, Xy110B, from the hyperthermophilic bacterium Thermotoga maritima, was expressed in transplastomic tobacco plants (Kim et al. 2011). The accumulation levels of the enzymatically active Xy110B were between 11 and 15 percent of the total soluble protein in tobacco leaves. The enzyme displayed "exceptional" thermostability and catalytic activities over methylglucuronoxylose (MeGXn), a major form of xylan in woody plants. The enzyme was also biologically active, hydrolyzing MeGXn into fermentable sugars between 40 and 90°C, and was stable in dry and stored leaves. The transplastomic plants, as well as the progenies, appeared morphologically normal. Due to the harsh pretreatments needed for lignocellulosic feedstocks, selection of thermostable and extreme pH tolerant cellulases and xylanases is quite important for the recombinant enzymes to remain active after the pretreatments. Alternatively, one can work with engineers to develop milder pretreatment conditions and choose appropriate enzymes that can survive the best for those conditions. Moreover, the possibility to bypass pretreatment in certain transgenic alfalfa plants has been reported (Chen and Dixon 2007).

A similar strategy could be used to improve switchgrass as a feedstock. Cellulase enzymes need to be added to the switchgrass feedstock during alcohol production in order to hydrolyze cellulose and produce sugars for fermentation. Cellulases normally include endoglucanase, exoglucanase, and cellobiase (Keshwani and Cheng 2009) and the cost of added cellulases to the process is one of the remaining major economical obstacles for commercial alcohol production from lignocellulosic feedstocks. Currently, no reports have investigated expressing cellulase genes in switchgrass, a strategy that would no doubt facilitate saccharification and reduce the production cost.

Harsh pH and high temperature conditions during pretreatment of the feedstock is a major concern for the survival of the introduced enzyme(s). To overcome this problem, the mildest pretreatment, ammonia fiber explosion (AFEX), was applied to E1-transgenic tobacco biomass and roughly one third of the heterologous enzyme activity was retained. Alternatively, to circumvent the pretreatment stage, crude extract of the E1-transgenic rice plants was added to pretreated rice straw or corn stover and approximately 30 and 22 percent of the cellulose in these plants was converted into glucose, respectively (Sticklen 2006). The expression of cellulase genes in these plants did not have an obvious detrimental effect on plant growth and development. Targeting of these genes to cellular compartments could facilitate accumulation of the heterologous enzyme(s). In switchgrass, about 26 percent of the dry weight is hemicellulose (Keshwani and Cheng 2009), which is currently underutilized for fermentable sugar production and has a great potential for biofuel production in the future.

DOE-USDA awarded Agrivida Inc. (Medford, MA) a grant for producing switchgrass with cell wall degrading enzymes that would remain inactive during plant growth but become activated after harvest. Other laboratories are working to create transgenic switchgrass plants expressing endoglucanase (data unpublished). Using the information obtained from previous research in cereal crops (Oraby et al. 2007; Gray et al. 2011), combined with an efficient transformation system (Li and Qu 2011), switchgrass is a promising candidate for producing cell-wall degrading enzymes as a value-added trait. Introducing value-added traits, such as bioplastics and cell wall degrading enzymes, into important bioenergy crops will ultimately combat the high costs associated with turning the lignocellulosic feedstock into biofuels.

Most biofuel processes currently under development require pretreatment followed by hydrolysis to produce monomeric sugars. The goal of any pretreatment is to separate the polysaccharide matrix of cellulose and hemicellulose from lignin and to loosen the structure enabling sites for chemical or enzymatic catalysis. The polysaccharide matrix may be hydrolyzed using chemical routes (such as acid hydrolysis) or by biological enzymatic routes (Huber et al. 2006). The sugars derived from hemicellulose and cellulose are fermented to produce a wide range of biofuels such as ethanol, hydrogen, biodiesel via lipid biosynthesis and butanol (Chandrakant and Bisaria 1998; Kim et al. 2008; Alvira et al. 2010; Panagiotopoulos et al. 2010; Wu et al. 2011a, b).

Many possible methods of chemical pretreatment have been reported, including steam explosion, dilute acid hydrolysis, concentrated acid hydrolysis, supercritical CO2 explosion and extraction, alkaline pretreatment (sodium hydroxide, potassium hydroxide, lime), ionic liquids, soaking in aqueous ammonia (SAA), ammonia recycles percolation (ARP) and ammonia fiber explosion (AFEX) pretreatment (Alizadeh et al. 2005; Mosier et al. 2005; Huber et al. 2006; Kim et al. 2006, 2007, 2008; Isci et al. 2009; Singh et al. 2009; Alvira et al. 2010; Panagiotopoulos et al. 2010; Wu et al. 2011a, b). In many of these pretreatments, the formation of inhibitory compounds at high temperatures (such as furfural and 5-hydroxymethylfurfural (HMF)) are one of the main constraints (Mosier et al. 2005).

Lignocellulosic biomass is highly recalcitrant to fermentation due to natural resistance mechanisms. Moreover, woody biomass such as pinewood has an even greater microbial recalcitrance than herbaceous biomass due to a tightly bound structure with high lignin content (Galbe and Zacchi

2002) . To enhance the release of polysaccharides from the lignocellulosic biomass, upstream processing (including size reduction and pretreatment) is a necessary step in biofuel production. For the production process to be economically feasible, total energy consumption in the size reduction and the pretreatment steps should be minimized as much as possible (Zhu et al. 2010).

Physical pretreatment involves size reduction to increase the available surface area and enhance enzyme hydrolysis of plant polysaccharides. Chemical and biological pretreatment methods are designed to liberate the convertible polysaccharide from the protective lignin casing, as well as to reduce the crystallinity of the cellulose so as to make the polysaccharides available to the hydrolyzing microorganisms (Hendriks and Zeeman 2009; Harmsen et al. 2010; Alvira et al. 2010). Selecting a pretreatment process is dependent on the particle size, moisture content and lignin content of the lignocellulosic biomass.

As users decide on which model is appropriate for their purposes, they will need to consider the level confidence in the inputs required and the appropriate level of model complexity to accomplish their goals. Process — based simulation models are split into those comprised of: 1) detailed leaf photosynthesis components that are integrated up to the leaf canopy level or 2) canopy level models that contain the relationship between plant functions and leaf area index (LAI), light interception, and radiation use efficiency (RUE). There are some concerns with both approaches. Leaf level photosynthetic rates are often not directly related to productivity, as described with RUE or above-ground net primary productivity (ANPP). A good example is a study by Kiniry et al. (1999) reported that sideoats

grama (Bouteloua curtipendula (Michx.) Torr.) had higher photosynthetic rates throughout the range of light levels than Alamo switchgrass, but sideoats gramma is far less productive than switchgrass. Similarly, Aspinwall et al. (2013) looked at several switchgrass ecotypes and concluded that "leaf — level physiological traits are often uncorrelated with genotype ANPP due to confounding of development with physiology, covariation among leaf traits, feedbacks with sink capacity, and increased self-shading". However, they identified "a syndrome of leaf functional traits" which aligned with genotype ANPP revealing that more productive genotypes initiated growth earlier and flowered later.

On the other hand, parameters for whole canopy models such as ALMANAC, EPIC, and SWAT are derived by measuring leaf area and dry matter destructively during the active growing period and fraction of light interception of plants assumed to be grown under nonlimiting water and nutrients conditions. Ideally such models use plant parameters derived at one site, with adequate soil moisture and soil nutrients, which are then applied for simulations under a wide range of environmental conditions. However, problems can arise when applying the model parameters, outside the regions of adaptation of a particular switchgrass ecotype. Latitudinal differences include photoperiod, number of hot days during the growing season, and number of cold days during the winter. Realistic simulation of processes controlling location differences, especially with differences in latitude, requires realistic understanding of the factors affecting such adaptation. These adaptation processes still need to be identified and quantified to more accurately simulate switchgrass ecotypes across a wide range of locations.

Another concern when simulating switchgrass with these process-based plant models is that many of these models were developed for annual crops. The perennial growth process is quite different from that of annuals, and more work is needed to accurately incorporate these differences in models, such as rooting during the establishment year as compared to subsequent years. Especially important for switchgrass modeling is N, P, and carbohydrate storage in roots in autumn and translocation out to above­ground plant parts in the spring. In addition, there has not been sufficient research regarding the possible differences in base temperature, optimum temperature, and root:shoot partitioning for different ecotypes. These will be key to simulating greenup in the spring and growth and development in the hottest part of the growing season.

Production of bioenergy requires the protection of soil and water associated with emerging bioenergy landscapes (Graham et al. 1996). The widespread degradation of the soil resource base and water quality due to past and current agricultural practices is well documented (USEPA 2009). The limited set of sustainability criteria attached to the 2007 Renewable Fuel Standard, which include stipulations about what types of land feedstocks are grown on, and the GHG intensity of biofuel production, were a promising start but may need to be expanded to include additional sustainability dimensions such as soil and water quality.

These mechanistic models will be useful for comparing switchgrass production systems to more conventional agricultural crops. Meki et al. (2011) applied a version of the EPIC model, APEX, to assess the sustainability of corn stover removal from the Upper Mississippi River Basin, based on a set of ‘acceptable planning criteria’ used in the CEAP analysis (USDA-NRCS

2010) , to judge whether or not a farm field needed additional conservation treatment. The ‘acceptable criteria’ included; (a) N in surface runoff < 16.8 kg ha1 y-1 (15.0 lb ac-1 yr-1), (b) N in sub-surface runoff < 28.0 kg ha1 y-1 (25.0 lb ac-1 yr-1), (c) total P losses < 4.5 kg ha-1 y-1 (4.0 lb ac-1 yr-1), (d) Sediment loss < 4.5 Mg ha-1 y-1 (2.0 ton ac-1 yr-1), and (e) SOC with a more ‘stringent’ restriction that the annual rate of change be positive. Given the critical functions of SOC in maintaining soil quality and productivity, biomass removal can only be justified if it does not deplete the SOC pool. These ‘acceptable’ levels represent field-level losses that are feasible to attain using traditional conservation treatment (nutrient management and soil erosion control), are agronomically feasible, and can equally be adapted to switchgrass production systems. Scientific literature on field research and edge-of-field monitoring in the U. S. Midwest, coupled with model simulations of conservation practices effects, provided guidance for identifying these thresholds (USDA-NRCS 2010).

Conclusions

Switchgrass is the "model" bioenergy crop for a potential bioenergy industry throughout the southeastern and south central USA (Wright and Turhollow

2010) . Given the infancy and urgency of the fast-evolving bioenergy industry, crop simulation models can complement and extend the applicability of information collected in field research trials, and when combined with the appropriate climate, soil, crop, and management databases, can be applied effectively to assess the sustainability and long-term impacts of converting land to bioenergy crops in a timely and cost-effective manner.

Acknowledgements

We thank Philip Fay, Daren Harmel, and Lara Reichman for comments on the manuscript. USDA is an equal opportunity provider and employer.

A recent review by Mitchell and Schmer (2012) addressed switchgrass harvest and storage considerations and provides more detailed information. Herein, an overview of harvest timing, nutrient management, and logistics will be addressed. Regionally-specific best management practices and extension guidelines have been developed from extensive research and are critical to the commercialization of switchgrass for bioenergy in a given region (Hancock 2009; Wolf and Fiske 2009; Mitchell et al. 2010b).

Individual bacterial colonies can be identified by morphology, or the observation of colony colors and physical shapes observed under a microscope. Gram positive or negative cultures can be distinguished with staining. More recently, 16S rRNA gene sequences have widely been used to identify bacterial species and to construct a phenogram. Bacterial genomic DNA needs to be isolated in order to amplify specific 16S rRNA gene sequences using a standard bacterial DNA isolation protocol (Sambrook et al. 1989). For general bacterial endophyte classification, universal PCR primers F27 (5′-AGA GTT TAT CMT GGC TCA G-3′) and R1492 (5′-GRT ACC TTG TTA CGA CTT-3′) are used to amplify partial bacterial 16S rDNA sequences (Diallo et al. 2004). The ability of bacterial endophytes to fix atmospheric nitrogen can be tested by growing bacteria in nitrogen — free medium for several cycles of cultures or PCR can be used to amplify the nifH gene, which is a conserved region in the dinitrogenase reductase gene complex. Fatty acid analysis, carbon source utilization, and antibiotic resistance (hygromycin, chloramphenicol, gentamycin, kanamycin, ampicillin, streptomycin, tetracycline, and rifampin) could be done for further identification.

Biomass yield has been widely accepted as the major trait for improvement in breeding switchgrass as a bioenergy crop since the beginning of the US Department of Energy sponsored Bioenergy Feedstock Development Program (McLaughlin and Kszos 2005). Currently more than 10 switchgrass breeding programs have breeding efforts mainly focusing on improving biomass yield as the principal trait and developing superior cultivars with improvement in biomass yield and other selected traits (Casler et al. 2011). Biomass yield is considered to be the most important factor contributing to the development of an economically viable biofuel industry if switchgrass is selected as the major feedstock crop. Genetically improving biomass yield in switchgrass through the delivery of higher-yield cultivars will increase the profit margins of producers, reduce the cost per unit biomass yield, and decrease transport delivery distance of feedstock from farmer’s fields to a target biorefinery, consequently benefitting the whole chain of biofuel production. It is recommended for one-cut system by the end of a growing season to maximize yields or after the first frost to allow a longer harvest window and to increase retranslocation of nutrients to root systems or soil (Sanderson et al. 1999; McLaughlin and Kszos 2005; Makaju et al. 2012).

Biomass yield is a highly complex trait inherited in a quantitative manner and regulated by a large number of unknown genes, and heavily affected by environmental factors and genotype by environment interaction. Selection for biomass component traits may be feasible to indirectly improve biomass yield. Biomass yield is positively correlated with plant height with a significant coefficient of 0.69 and negatively correlated with plant maturity (r= -0.45) (Talbert et al. 1983). Lemus et al. (2002) also reported a high correlation between biomass yield and plant height. Using 11 lowland populations tested in two locations, Das et al. (2004) reported biomass yield was positively correlated with tiller number per plant with a coefficient of r=0.60 to 0.68, but not with plant height. The latter study suggested selection for more tillers per plant would be the most effective method for indirectly increasing biomass yield. Leaf blade length and width, stem diameter and node number per tiller have effects on biomass yield but are not consistent across locations (Das et al. 2004). Bhandari et al. (2010) reported positive correlations between biomass yield and tillering ability (r=0.73), and plant height (0.52), and stem thickness (0.38). Boe (2007) and Boe and Beck (2008) reported significant correlations between biomass yield and tiller weight, suggesting tiller weight can be used as a trait in indirect selection for biomass yield improvement. Overall, it is promising that use of tiller number, plant height and tiller diameter size as indirect selection traits for switchgrass biomass improvement.

In addition to biomass yield, many other traits, individually or in combinational forms have been included in breeding and selection activities. Seed dormancy is a long existing issue to the successful establishment of new stands using seed in switchgrass. Efforts to select for low post harvest seed dormancy have been made and led to a substantial increase in germination rates in ‘Alamo’ germplasm and a release of ‘TEM-LoDorm’ germplasm (Burson et al. 2009). Adaptation to target environments, cell — wall recalcitrance, abiotic stress (drought, cold) tolerance, and biotic stress resistance are some additional important traits in switchgrass breeding and selection process (Casler et al. 2011). Drought tolerance is critical as switchgrass is targeted to grow on marginal lands without supplementary irrigation. Leaf rust caused by Puccinia emaculata and smut by Tilletia maclagani are important diseases for switchgrass if grown in large areas (Parrish and Fike 2005).

Several experiments were conducted to estimate heritabilities for biomass yield and related traits, and forage quality traits. Forage use of switchgrass has taken place much before the identification of the species’ potential use as a bioenergy crop (Vogel 2004). One important trait for forage use is to improve forage quality, including dry matter digestibility. Godshalk et al. (1986) and Hopkins et al. (1993) reported moderate to medium narrow- sense heritability for in vitro dry matter digestibility (IVDMD). Phenotypic recurrent selection was effective in improving populations for increased IVDMD, which led to the development and release of ‘Trailblazer’ (Hopkins et al. 1993) and ‘Performer’ (Burns et al. 2008a). Among the traits related to biomass yield, heritabilities for days to heading and days to flowering are relatively high (Bhandari et al. 2010). For plant height, Talbert et al. (1983) reported high narrow-sense heritability, while estimates were low based on variance component analysis of half-sib families, and were medium based on parent-half-sib progeny regression by Bhandari et al. (2010). Similarly, narrow-sense heritability estimates for stem thickness, tillering ability, plant spread and spring regrowth are inconsistent due to differences in genetic structures (half-sib versus full — sib families) and data sources (individual plant data versus plot mean) in the experiments.

Being the major trait as targeted for developing the species as a bioenergy feedstock crop, biomass yield has narrow-sense heritability estimates ranging from 0.12 to 0.25 (Talbert et al. 1983; Godshalk et al. 1986; Rose et al. 2008; Bhandari et al. 2010 and 2011). The low values were estimated on the basis of variance components from data of individual plants among families. But in some of the same experiments, higher heritabilities were also obtained on plot yield mean data, suggesting heritability estimates from midparent-progeny regression tend to be biased upwards (Bhandari et al.

2011) . Collectively, the tested low heritability for biomass yield indicates that direct phenotypic selection for increased biomass yield may not be effective. Breeding and selection procedures capturing additive variation and increasing frequency of additive genes responsible for the trait need rigorous progeny evaluation in the improvement of breeding populations. Improved populations can serve as germplasm pools for making synthetics and even be released as cultivars per se.