The advancement of DNA sequencing technologies and the associated rapid decrease in sample prep and run costs has facilitated routine transcriptional analysis in non-model species, especially those with large genomes, such as switchgrass. Analyzing the transcribed fraction of a genome using Expressed Sequence Tags (ESTs) through RNAseq or Second-generation sequencing technologies allow researchers unprecedented resolution for gene mapping, annotation, functional analysis, discovery, and quantitation (Andersen and Lubberstedt 2003; Lister et al. 2009; Wang et al. 2009; Lu et al. 2010; Kaur et al. 2011). ESTs have been invaluable in the development of molecular markers (Gupta et al. 2003; Varshney et al. 2005; Barbazuk et al. 2007), comparative genomics (Fei et al. 2004; Ribichich et al. 2006; Tobias 2008; Mayer et al. 2011), and the elucidation of important traits (Moralejo et al. 2004; Shen et al. 2006). Without a complete reference genomic sequence, studies focused on unraveling the transcriptional complexity of switchgrass on a genomic scale have just begun to be presented and are still in their infancy. The earliest larger-scale efforts date back to 2005 (Tobias et al. 2005) where authors looked at a brief glimpse of the transcriptional landscape of 4 tissues (stem, callus, crown, and leaf). The authors identified 7,810 tentatively unique genes (TUGs) (Tobias et al. 2005) and focused analysis on gene targets with bioenergy conversion traits. In 2008, Tobias et al. (Tobias

2008) collected 61,585 EST sequences from three different cDNA libraries (crown, seedling, and callus tissue) of an elite high-yielding northern lowland ecotype that shows a high degree of phenotypic variation (Casler et al. 2007) to produce minimally redundant consensus sequences that were used for genome wide comparisons to other grasses, such as sorghum, to identify critical genes and gene families involved in photosynthetic efficiency, stress tolerance, and cell wall structure; as well as an EST-SSR marker set for genetic analysis (Tobias 2008). The sequences were produced using Sanger-style dye terminator sequencing techniques and a clone-based cDNA system. The result of a Sanger-style sequencing endeavor is long, high-quality reads (gold standard) that can be easily mapped to closely related reference genomes. These data were also used to assess the degree of subgenome similarity in this all suspected autopolyploid species, and search for genome duplication signatures (Tobias 2008). A more recent transcriptome study that employs the power of 2nd generation sequencing technologies (i. e., massively parallel sequencing), Roche 454, focuses on four different tissues that are underrepresented in the current EST databases (dormant seeds, germinating seedlings, emerging tillers, and flowers) (Wang et al. 2012). This study produced nearly 1 million pyrosequencing reads (approximately 360 Mbp of sequence) with an average cleaned read length of 367bp that represented the respective switchgrass transcriptome (Wang et al. 2012). A comprehensive dataset of 545,894 switchgrass ESTs obtained from Genbank was combined with this study and assembled using the foxtail millet genome sequence (Zhang et al. 2012) as a guide that resulted in 98,086 switchgrass EST contigs (Wang et al. 2012). The unigene assembly approach used is conservative because of the tetraploid genomic nature and the result is more contigs than traditional unigene assemblies, but intended to reduce erroneous assemblies of homologs, paralogs, and splice variants (Wang et al. 2012). In convergence with other functional studies, the authors focused the analysis on genes associated with C4 photosynthesis, a process that is critical to biomass accumulation, in addition to tillering and dormancy.

Damaging population thresholds of plant-parasitic nematodes are currently unknown (Mekete et al. 2011). However, based on damage threshold value ranges for other monocotyledon hosts, several US states (e. g., Illinois, Iowa, Kentucky, Tennessee, and Georgia) have potential for to switchgrass yield losses due to plant-parasitic nematodes. Specifically, Helicotylenchus, Xiphinema, Pratylenchus, Hoplolaimus, Tylenchorhynchus, Criconemella, and Longidorus spp. were all found to have population densities within or above the threshold value ranges reported for other monocotyledon hosts (Mekete et al. 2011). Nematodes are most likely to damage either young or stressed plants (Griffin et al. 1996) by feeding on switchgrass roots. In many cases, nematode damage may be mistaken for environmental stress symptoms (Leath et al. 1996). Evers and Butler (2000) found that soil fumigation in Texas improved switchgrass establishment and seedling vigor compared to a weed-free check indicating soil borne pathogens were responsible for poor establishment. Cassida et al. (2005a) observed that in fully established (5 yr) switchgrass stands, nematode numbers were greater in higher rainfall regions such as Louisiana, Arkansas than in drier sites in Texas. They also reported that dry matter yield and persistence of switchgrass were reduced as nematode populations increased.

AM fungi are obligate and must be grown with their host plant. Soil collected or spores isolated from soil can be used as inoculum, which is called a soil trap culture. To create a plant trap culture, plants containing mycorrhizal fungi are collected from the field and are transplanted to a potting medium (sterile soil or soil-sand mixture, which should have low available P and not rich in organic matter) (http: //www. scribd. com/doc/58675784/4-3- Mycorrhiza0403). More details can be found in the International Culture Collection of (Vesicular) Arbuscular Mycorrhizal Fungi (INVAM) (http: // invam. caf. wvu. edu).

Identification

Several methods can be used to identify endophytic bacteria and fungi as well as AM fungi, such as morphological characterization, genomic sequencing, and staining (Zinniel et al. 2002).

Switchgrass is a native widely distributed species in North American continent, from southern provinces (i. e., Saskatchewan, Manitoba, Ontario, Quebec and New Brunswick) of Canada to a majority of the United States (except Washington, Oregon and California), to 15°N latitude in Mexico (Hitchcock and Chase 1950; USDA NRCS 2012). The broad geographic distribution of the species in extremely diverse climatic environments and edaphic conditions dictates its enormous genetic diversity through natural selection and evolution over a long period of time. Currently existing switchgrass germplasm has been classified into two groups: lowland ecotype and upland ecotype on the basis of morphology and habitat preference (Porter 1966). Lowland plants are generally larger than upland ones in most morphological characters (see an example in Fig. 5). Native

lowland plants occur on alluvial soils while native upland plants are found on typically drier and less fertile non-alluvial soils. The lowland switchgrass is largely distributed from the lower south to about 40° N latitudes while the upland ecotype is widely distributed throughout the range of switchgrass adaptation. Huang et al. (2003) reported switchgrass progenitor’s genomes may have been divergent from other species less than 2 million years ago (MYA) and the polyploidization events occurred within 1 MYA.

In recent years, especially after the selection of switchgrass as a model crop for feedstock development, molecular markers have been extensively used to quantify genetic differentiation and variation in switchgrass germplasm (Gunter et al. 1996; Hultquist et al. 1996; Huang et al. 2003; Narasimhamoorthy et al. 2008; Cortese et al. 2010; Todd et al. 2011; Zalapa et al. 2011). Combined use of nuclear DNA markers and chloroplast markers allow grouping switchgrass germplasm into two clusters, which correspond well with lowland and upland ecotypes (Todd et al. 2011; Zalapa et al. 2011). Within ecotypes, a large portion of the genetic diversity has been found to reside within populations and the remaining smaller part to be among populations (Gunter et al. 1996; Narasimhamoorthy et al. 2008; Cortese et al. 2010; Zalapa et al. 2011). Okada et al. (2010a) reported genetic diversity of octoploids is larger than that in tetraploid germplasm due to polysomic inheritance of octoploids. These results are well consistent with and reflect the outcrossing reproduction mechanisms in the species. Using SSR markers, Narasimhamoorthy et al. (2008), Cortese et al. (2010) and Zalapa et al. (2011) demonstrated genetic diversity is related to the geographic locations where the germplasms have been collected. Based on the results of Zalapa et al.

(2011), Casler et al. (2011) proposed the concept of regional gene pools, which can be used in developing complementary heterotic groups for breeding superior cultivars harnessing heterosis in biomass yield.

Although a large number of switchgrass accessions have been collected in public (USDA NRCS, ARS, and universities) and private research programs, it appears more collections are needed to fully sample naturally available genetic diversity. Large amounts of native germplasm are preserved in remnant prairie sites and fragmented habitats scattering across the species range. Individual programs, however, are not generally charged with or funded for collecting and maintaining large collections for a long term. Casler et al. (2011) correctly indicated concerted efforts are essential to coordinate effective collection and to ensure public availability of the germplasm in switchgrass. Full collection, characterization, long term preservation can be best accomplished by agencies like USDA NPGS, which has the dedicated responsibilities.

Hormones are important regulators in plant growth and development, playing pivotal roles not only in regulating plant cell division, elongation, and differentiation, but also in plant organ formation and responses to environmental stresses (Srivastava 2002; Woodward et al. 2005). In Arabidopsis, it was discovered that some transcription factor genes, Auxin Response Factors (ARFs) including ARF10, ARF16, ARF17, ARF6 and ARF8 have complementary sites for miRNAs (Bartel 2003; Wu et al. 2006). ARF10, ARF16 and ARF17 are the targets of miR160, and ARF6 and ARF8 are the targets of miR167 (Bartel 2003; Wu et al. 2006). GAMYB is a target of miR159 (Achard et al. 2004). GAMYB is regulated by GA and coordinates flower formation and floral organ development. miR159, also regulated by GA, is a negative regulator of GAMYB mRNA, adjusting flowering time in short — day photoperiod and anther development (Achard et al. 2004). NAC1 is a target of miR164 (Xie et al. 2002; Guo et al. 2005). The expression of NAC1 is positively regulated by auxin, controlling lateral root development. It was reported that miR164 is also induced by auxin. As it accumulates in response to auxin treatment, the NAC1 mRNAs decrease, exhibiting an auxin-induced, miR164-guided mRNA cleavage process in plants. This process indicates an intricate auto-regulatory loop in auxin signaling transduction pathways controlling lateral root development (Xie et al. 2002; Guo et al. 2005).

MiRNAs are Involved in Plant Biotic and Abiotic Stress Responses

Recent studies have demonstrated that miRNAs play an important role in mediating plant responses to many kinds of biotic and abiotic stresses (Kasschau et al. 2003; Navarro et al. 2006; Lu et al. 2007; Jung and Kang 2007; Pandey and Baldwin 2007; Liu et al. 2008; Li et al. 2010; Zhou et al.

2010) . Increasing evidence indicates that miRNAs are involved in plant response to viruses suggesting their possible role in virus-induced post­transcriptional gene silencing (PTGS) (Kasschau et al. 2003; Li et al. 2010). MiRNAs have also been implicated in plant responses to bacteria, fungi and insects (Navarro et al. 2006; Lu et al. 2007; Pandey and Baldwin 2007). Recently, it has been found that several miRNA families, such as miR1507, miR2109, miR482/2118, miR6019, and miR6020, target genes that encode nucleotide binding site-leucine rich repeat (NBS-LRR) plant innate immune receptors (Zhai et al. 2011; Li et al. 2012; Shivaprasad et al. 2012).

Plants have developed multiple mechanisms and strategies to cope with various abiotic stresses, in which miRNAs play an important role. MiRNAs can be induced by abiotic stresses, either up-regulated or down-regulated (Sunkar and Zhu 2004; Jung and Kang 2007; Sunkar et al. 2007; Liu et al. 2008; Zhou et al. 2010). For example, miR319 was up-regulated by cold stress (Sunkar and Zhu 2004); miR398 was down-regulated by oxidative stress (Sunkar et al. 2006); and miR171 was up-regulated by cold, drought and salinity stresses respectively (Liu et al. 2008). Sunkar et al. (2007) found that miR393 was up-regulated in response to cold stress, leading to the down-regulation of its targets encoding E3 ubiquitin ligase, which could degrade its target protein including some positive regulators for adapting cold stress. Thus, the up-regulation of miR393 positively contributed to plant response to cold stress.

Genetics and Origin of Upland and Lowland Ecotypes

Before one can begin to make appropriate decisions about cultivar selection, it is import to know something of switchgrass ecotypes and their origins. Switchgrass is characterized by a wide degree of genetic diversity, which conveys broad adaptation. The plant’s native range extends from Canada to Mexico and from the Atlantic coast to the Sierra Nevada mountain ranges. In addition to this wide geographic adaptation, switchgrass also divides out into two groups—uplands and lowlands—that describe the species’ typical adaptation relative to position on the landscape.

Essentially all switchgrass is categorized as either upland or lowland "ecotypes", although some indistinct plant types may represent hybrids of these two forms (Zhang et al. 2011a). These two ecotypes may be able to interbreed and produce fertile offspring, and both can be found in switchgrass populations within a habitat (Hultquist et al. 1997). However, the differences between ecotypes promote fitness for survival in unique environments. Upland and lowland ecotypes have also been described as "cytotypes", which reflects small but distinct differences in their chloroplastic DNA sequences (Hultquist et al. 1996).

In terms of ploidy, lowland ecotypes have been thought to be tetraploid (2n = 4x = 36), although recent discoveries suggest there may be octoploid lowlands (Zhang et al. 2011a, b). Upland ecotypes can contain either tetraploid and octoploid genotypes (2n = 8x = 72) (Zalapa et al. 2011). Although there is significant overlap in their regional distribution, lowlands generally predominate in the southern USA while uplands generally originate from the drier, colder northern Great Plains.

McMillan (1959) postulated that all current switchgrasses likely hailed from three regions of North America—safe havens, really—that provided refuge from the glacial conditions of the Ice Ages. More recent research (e. g., Cortese et al. 2010; Zhang et al. 2011a, b) provides evidence of the geographic origins of switchgrass ecotypes. Lowland ecotypes appear to have their origins in the Eastern Gulf Coast and Southern Great Plains while uplands hail geographically from the Central and Northern Great Plains and the Eastern Savannah regions (Zalapa et al. 2011; Zhang et al. 2011a). While such consideration may seem to be an academic exercise, they have very real, practical implications for cultivar selection in the context of bioenergy production systems, which we discuss in the section of the same name below.

The first interaction of soil bacteria with the plant occurs at the rhizoplane, and a sufficient titer of robust bacteria are required in the soil region in close proximity to the root surface (rhizosphere). An experiment demonstrated that sustained, high rhizosphere soil populations of the endophyte Bacillus subtilis GY-IVI were required for efficient endophytic colonization of the root (Zhao et al. 2011). It has also been suggested that these robust and high titer levels of the colonizing bacteria in the soils help bacterial endophyte competition, indicating that these bacteria are highly competent at rhizosphere/rhizoplane colonization (Whipps 2001; Compant et al. 2005a).

A variety of bacterial traits are known to be required for rhizosphere and/ or rhizoplane colonization competence (Compant et al. 2010). Numerous studies have shown that bacterial colonization of the rhizoplane occurs initially with localization across various regions of the root, including root tips, sites of lateral root emergence, and root hair zones (Compant et al. 2008; Prieto and Mercado-Blanco 2008; Zhang et al. 2010). During rhizoplane colonization, single cells have been observed, leading to the development of colonies along the root surface, and to the establishment of biofilms (Hansen et al. 1997; Benizri et al. 2001). However, rhizoplane colonization does not occur uniformly (Compant et al. 2010). For example, Pseudomonas fluorescens PICF7 predominantly colonized the root differentiation zone (Prieto and Mercado-Blanco 2008), and the more mature parts of the root exhibited little colonization by Pantoea agglomerans YS19 (Zhang et al.

2010) . This variation on rhizoplane colonization distribution may be due to differences in root exudate production (Lugtenberg and Dekkers 1999), the protective microenvironment of different regions of the root (Prieto and Mercado-Blanco 2008), and/or the presence of specific or preferential cell surface binding sites for the bacteria (Miao et al. 2008). Regardless of the sites of rhizoplane colonization, it has been reported that the population densities of bacteria in the soil are approximately 2 orders of magnitude higher (107-109 CFU per g of rhizosphere soil) than are found on the root surface (Benizri et al. 2001; Bais et al. 2006).

One promising strategy for improving switchgrass biomass quality, and possibly yield, is to utilize knowledge about the synthesis of cell walls. The last decade has seen an enormous increase in our understanding of the enzymes and regulators of cell wall synthesis. Most of this work has been in the reference dicotyledenous plant and forward and reverse genetics workhorse, Arabidopsis thaliana. However, as described above, the differences between grass and dicot cell walls limit some of the transferability of work between these clades. Indeed, a recent analysis of expression of cell wall synthesis genes in ovary tissues with primary walls revealed a wide divergence between Arabidopsis expression patterns and those of rice and maize (Penning et al. 2009). Fortunately, there has also been a steady increase in the study of cell wall synthesis and regulation in grass species, with work in rice, maize, and the diminutive wheat

relative, Brachypodium distachyon (BGI 2010), paving the way for more facile improvement of bioenergy grasses, such as switchgrass. The genomic colinearity of grasses (Devos 2005), as well as conventional phylogenetic analysis, facilitate transfer of information among grasses with sequenced genomes. Importantly, the effort to reconstruct the switchgrass genome sequence is underway, with the first draft assembly (v0.0) released in early 2012 (http://www. phytozome. org/panicumvirgatum. php) (Casler et al.

2011) . As will be described below, researchers have already characterized in switchgrass a number of lignin biosynthesis genes and a secondary cell wall transcriptional regulator.

How can functional information about cell wall synthesis and regulation be utilized? On a basic level, many mutants in cell wall related genes are in themselves easier to digest. Though many single gene cell wall mutants have reduced stature or disease resistance, for others, at least when grown in greenhouses, vegetative development is apparently unaffected or increased, as is resistance to specific pathogens. Geneticists have already used information about cell wall synthesis to design molecular markers that correlate with altered cell wall content (Truntzler et al. 2010; Wegrzyn et al. 2010). Furthermore, examination of cell wall synthesis variants with increased or reduced amounts of specific polymers permits dissection of the function of various cell wall constituents. From this information in the longer term, researchers may be able to engineer grass cell walls, with widely divergent, but still functional, cell walls that are optimized for conversion into biofuels via specific conversion technologies. In this section, we will discuss the current understanding of the synthesis of the major components of the switchgrass cell wall, namely, cellulose, xylan, and lignin. Table 1 and Table 2 list genes that function in the synthesis of cellulose and xylan, respectively, in Arabidopsis and grass species.

**All rice loci are abbreviated from their full identifier by removing the LOC_ that is part of the normal MSU annotation.

The construction of highly saturated maps is often a time-consuming process, especially if investigators are employing different parental stocks and markers are not easily transferable. Merging maps are attractive since their integration allows for an increase in marker density without the need of additional genotyping, increased marker portability (i. e., polymorphic markers can be used in more than one population), and improved marker alignment precision (i. e., congruent anchor maker position). Using JoinMap, a number of integrated linkage maps have been developed in numerous economically important crop plants including wheat (Triticum aestivum

aF= female; M=male; b RFLP=restriction fragment length polymorphism; SSR=simple sequence repeat; STS= sequence tagged site; c Segregation distortion, percentage of markers showed segregation distortion.

L. Somers et al. 2004), maize (Zea mays L., Falque et al. 2005), red clover (Trifolium pratense L., Isobe et al. 2009), and ryegrass (Lolium ssp. Studer et al.

2010) . So far no similar studies have been reported in switchgrass probably due to limited linkage maps available.

MiRNA-mediated gene repression is a conserved mechanism in plants and animals. Although most of the genes in the genome are not the targets of miRNAs, artificial miRNA (amiRNA) can be designed to repress expression of target genes by replacing new miRNA duplex from a natural miRNA precursor (Schwab et al. 2006; Ossowski et al. 2008; Khraiwesh et al. 2008; Molnar et al. 2008). The amiRNA technology is currently becoming a powerful tool for gene silencing and has been used successfully in many plant species, including Arabidopsis thaliana (Schwab et al. 2006; Ossowski et al. 2008), rice (Warthmann et al. 2008), moss (Physcomitrella patens) (Khraiwesh et al. 2008), and alga (Chlamydomonas reinhardtii) (Molnar et al.

2008) . There are several examples where down-regulating gene expression level or loss of gene function results in improved traits in switchgrass (Fu et al. 2011; Xu et al. 2011). AmiRNAs as a highly specific approach for effective post-transcriptional gene silencing (PTGS) provides a new molecular tool in switchgrass genetic improvement and can significantly contribute to cost-effective and environmentally friendly production and utilization of renewable bioenergy.