In theory, the yield of a plant is the product of the solar energy that the plant intercepts, utilizes, expends, and stores in harvestable plant biomass (Heaton et al. 2008). The amount of solar energy that a field of plants can intercept depends on the period and length of vegetative growth, the plant architecture and canopy, and planting density in a field. The solar energy utilization of a plant is largely determined by its net photosynthetic efficiency. The C3 and C4 photosynthetic pathways and theoretical solar energy conversion efficiency, along with their implications on bioenergy grass improvement, were recently reviewed (Heaton et al. 2008; Zhu et al.

2008) . Here, a few phenotypic traits, potential genes, and genetic pathways contributing to these traits are addressed.

Increasing switchgrass photosynthesis efficiency. C4 plants have a greater photosynthetic efficiency than C3 plants, primarily because of the C4 cycle. The CO2 concentrating mechanism results in the avoidance of photorespiration, which increases net CO2 assimilation and leads to a higher water use efficiency (Schmitt and Edwards 1981; Zhu et al. 2008). In most C4 plants, CO2 assimilation is processed in two distinct cell types: Kranz mesophyll cells and bundle sheath cells. The first steps occur in Kranz mesophyll cells. Initially, CO2 reacts with phospoenopyruvate (PEP) and is converted by PEP carboxylase (PEPC) into the C4 acid oxaloacetate (OAA). During C4 photosynthesis, intermediates (e. g., C4 acids) are diffused or transported between mesophyll cells and bundle sheath cells through plasmodesmata. In the bundle sheath cells, these intermediates are decarboxylated to release CO2, which is then used by RuBisCO as a substrate (Sowinski et al. 2008). According to the decarboxylation routes, C4 plants fall into one of three subtypes: 1) a NADP-Malic Enzyme (NADP — ME) subtype, 2) a NAD-Malic Enzyme (NAD-ME) subtype, or 3) a PEP — Carboxykinase (PEP-CK) subtype (Edwards et al. 2004; Weber and von Caemmerer 2010).

Most major C4 crops, such as maize, sorghum, sugar cane, and miscanthus, belong to the NADP-ME subtype, which is more efficient than the other two subtypes (Zhu et al. 2008). However, switchgrass is a NAD-ME subtype species. Notably, it is fairly unique that Panicum species have all photosynthesis types: C3, C3-C4 intermediate species, and all of the three C4 subtypes (Ohsugi and Murata 1986; Ohsugi and Huber 1987). In the same Panicum genus, the NADP-ME subtype species (e. g., P. antidotale) contain about one and a half to two times higher PEPC and RuBisCO activities than the NAD-ME subtype species (e. g., P. coloratum). The NADP-ME subtype species also exhibit about two times higher PEPC activity but one and a half times lower RuBisCO activity than the PEP-CK subtype species (e. g., P maximum) (Ohsugi and Huber 1987). The huge difference in PEPC and RuBisCO activities between species in the Panicum genus indicates a spacious room for improving photosynthesis efficiency of switchgrass by increasing the activities of these two key enzymes.

Recently, a nice review detailed the difference in photosynthetic intermediates between NADP-ME and NAD-ME subtypes (Weber and von Caemmerer 2010). In NADP-ME subtype species, OAA, after it is synthesized in the cytosol, is directly transported back to the chloroplast and converted into malate in mesophyll cells. Then, the malate is diffused or transported into the chloroplasts of bundle sheath cells and decarboxylated by the NADP-malic enzyme to release CO2. However, in NAD-ME subtype species, OAA, once synthesized in the cytosol of mesophyll cells, is not transported back to chloroplast but is instead converted into aspartate in the cytosol. Aspartate is then diffused or transported into the mitochondria of bundle sheath cells and converted into malate. The malate is then decarboxylated by NAD-malic enzymes in the mitochondria and CO2 is released in the bundle sheath cells (Weber and von Caemmerer 2010). Therefore, the photosynthetic pathway between NADP-ME and NAD — ME subtype species branches with the catalysis of OAA into aspartate or malate. The pathway further diverges with differences in their subcellular transportations, possibly because of the presence of selective membrane transporters guarding the chloroplasts and the cytosol. Identifying genetic components behind these differences, and engineering the pathway into switchgrass, will likely convert switchgrass into a "synthetic" NADP-ME subtype species with greater photosynthetic efficiency.

Although a finely constructed genomic map is not yet available for switchgrass and other NAD-ME subtype C4 grasses, high quality genomic sequences of several C4 grasses (maize, sorghum) and C3 grasses (rice, Brachypodium) are publicly available (Goff et al. 2002; Yu et al. 2002; Paterson et al. 2009; Schnable et al. 2009; Vogel et al. 2010). Comparative genomic studies have revealed that certain genetic components contribute to the difference between C3 and C4 photosynthesis. For example, comparison between the genomes of sorghum and rice showed that the "evolution of C4 photosynthesis in the sorghum lineage involved redirection of C3 progenitor genes as well as recruitment and functional divergence of both ancient and recent gene duplicates" (Paterson et al. 2009). The number of genetic components causing differences between C4 subtypes should be less than those between C3 and C4 plants. Moreover, closely related Panicum species comprise a natural pool of photosynthesis types and subtypes. Comparative studies on transcriptomes or genomes between representative Panicum species, as well as functional studies on candidate genes, will assist in elucidating the mystery of photosynthesis types and subtypes. The resultant knowledge can be readily used for genetic improvement of switchgrass and other economic plants.

Improving switchgrass plant architecture. The amount of light intercepted by a field of plants is largely determined by plant architecture (leaf angle and shapes, plant height, and tiller number), planting density, and the vegetative growth period (Heaton et al. 2008; Wang and Li 2008). The vegetative growth period can be prolonged by promoting early emergence of tillers of perennial grasses and by delaying flowering as mentioned above. The planting density of a field is dependent on plant architecture. Grass tiller number is important for field establishment. For grass cultivars with a lower tillering potential, cultivation strategies (e. g., dense planting) can compensate for their disadvantages. Here, we focus on research progress on a few aspects of plant architecture, such as leaf angle, leaf shape and plant height.

Erect leaves (small leaf angle against the stem) enhance light interception in densely planted fields (higher leaf area index), and thereby may increase biomass yield (Sakamoto et al. 2006). Decreasing brassinosteroid (BR) content or sensitivity by selecting BR-deficient mutants, e. g., brassinosteroid — dependent 1 (brd1), ebisu dwarf (d2), dwarf11, osdwarf4-1, or BR-insensitive mutants [dwarf 61 (d61) and leaf and tiller angle increased controller (oslic)] can effectively induce erect leaves in rice by altering lamina joint bending (Yamamuro et al. 2000; Hong et al. 2002; Sakamoto et al. 2006; Morinaka et al. 2006; Wang et al. 2008). Specifically, the rice osdwarf4-1 mutant has erect leaves but no alteration in reproductive development and thereby produces higher grain yields under dense planting conditions without extra fertilizer (Sakamoto et al. 2006). All of these BR-related mutants have erect and dark green (higher chlorophyll content) leaves. However, these mutants are dwarf or semi-dwarf. The dwarf to semi-dwarf stature is important for rice stand and grain yield, as the selection of one semi­dwarf mutant in a GA-biosynthesis gene, OsGA20ox2 (sd1), successfully led to the development of elite rice cultivars in the "Green Revolution" (Sakamoto et al. 2004). However, dwarf stature is not a desirable trait for bioenergy crops where the above-ground vegetative organs account for a large portion of the biomass yield. The semi-dwarf to dwarf phenotype in BR-related mutants is caused by failure of organization and polar elongation in the leaf and stem cells (Yamamuro et al. 2000). On the contrary, GA can positively regulate plant stem elongation (Kende et al. 1998). GA and BR may antagonistically regulate the expression of some downstream genes (Bouquin et al. 2001). Recently, a rice GA-stimulated transcript family gene, OsGSR1, was identified to be involved in the crosstalk between GA and BR (Wang et al. 2009). This study showed that OsGSR1 is a positive regulator of both GA signaling and BR biosynthesis (Wang et al. 2009). However, it has not yet been reported that GA can alter grass leaf angles. Therefore, it is possible to engineer grasses with erect leaves and normal, or increased, plant height by simultaneously manipulating BR and GA-related genes.

Leaf angle and leaf shape are often correlated. In BR mutants, the erect leaves are often short because of failure in elongation of leaf cells (Yamamuro et al. 2000). In several other cases, rolling (typically upward-curling) leaves create more erect leaves in rice (Shi et al. 2007; Zhang et al. 2009; Li et al.

2010) . Rolling leaves may also help prevent water loss by increasing stomatal resistance, decreasing leaf temperature, and reducing light interception per leaf, while simultaneously increasing light transmission rates to lower leaves of the plant (O’Toole and Cruz 1980). Altered expression of a few genes in rice caused rolling leaves, but anatomical reasons for the leaf­curling are different. A rice null mutant of Shallot-like 1 (SLL1, a KANADI family gene) has a broader distribution of mesophyll cells in the region where sclerenchymatous cells distribute in wild type rice (Zhang et al.

2009) . These mutants also have bulliform cells on the abaxial side of the leaf, which thereby induces upward-curling leaves (Zhang et al. 2009). Studies in Arabidopsis showed that a group of YABBY and KANADI family genes regulate abaxial organ identity (Emery et al. 2003; Eshed et al. 2004; Eckardt 2010). A group of HD-ZIP III family genes [e. g., PHABULOSA (PHB) and PHAVOLUTA (PHV)] have been found to promote adaxial organ identity. It could be the antagonism of these genes that coordinates normal leaf polarity and leaf shape (Emery et al. 2003). Similarly, a maize KANADI family gene, Milkweed Pod 1(MWP1), also functions in defining abaxial cell identity (Candela et al. 2008). Recent studies have shown that miRNAs, as well as genes in other families, are also involved in leaf shape formation. For example, overexpression of rice Argonaute 7 (OsAGO7), a gene presumably involved in miRNA metabolism, caused upward-curling leaves in rice (Shi et al. 2007). Overexpression of Abaxially Curled Leaf 1 (ACL1) induced downward rolling (abaxial-curling) leaves by increasing the number and size of bulliform cells on the adaxial side of the leaf (Li et al. 2010).

Xu et al. (2012) obtained erect leaf switchgrass by overexpressing an Arabidopsis NAC domain gene, Long Vegetative Phase 1 (AtLOV1). Interestingly, the transgenic switchgrass plants have a phenotype typical of BR-mutants (dark-green and erect leaf), but are not obviously dwarfed (except one transgenic line with an extreme phenotype). Differential gene expression analysis by microarray did not show significant expression changes of identified BR or GA-related genes in the transgenic plants (unpublished). Overexpression of AtLOV1 in rice induced dark green leaves and a dramatically dwarfed stature, but did not change the leaf angle (unpublished). Although the mechanism controlling the phenotype of transgenic switchgrass and rice is unclear, the results suggest that it is possible to alter leaf angle without causing a dramatic negative effect on other vegetative growth traits.

In summary, several important agronomic traits of switchgrass production and potential genes/genetic pathways underlying these traits are reviewed in this section. Translational and functional genomics studies will allow us to understand the functions of these gene(s) and create a more comprehensive picture of the interactions between physiological pathways. A more thorough understanding of the mechanisms underlying these traits will help improve plant yield and also help to design better strategies for plant genetic improvement (Hammer et al. 2004). We can use transgenic or "cisgenic" strategies to quickly "stack" genes of interests from foreign or native genomic origins. Then, we can use synthetic biological approaches to engineer and move entire essential genetic components of a pathway into switchgrass (Benner and Sismour 2005). For example, certain microbial metabolic pathways can be recruited and integrated into plant systems, essentially making plants bio-factories for desirable products (Somleva et al. 2008). All of these approaches are emerging at an unprecedented speed. We can imagine that many genomics tools will be successfully applied to the genetic improvement of switchgrass in the near future.