Sexual and asexual reproduction systems and associated mechanisms are essential for the breeders to determine both general features of a breeding program and specific procedures of hybridization, selection and cultivar development. The inflorescence of switchgrass is a typical open and diffuse panicle of 20-60 cm in length and 20-40 cm in width (Fig. 1). Each panicle consists of many to hundreds of spikelets, with two dissimilar florets in each spikelet (Tyrl et al. 2002). The lower floret is staminate while the upper one is perfect (Fig. 2). Stigma of each perfect floret exerts out of lemmas about 1-3 days earlier than anthers.

But the stigma is still receptive when anthers shed pollen grains on the same floret. The morphological structure and blooming behavior of florets provide specific conditions which may favor outcrossing over selfing. Switchgrass has long been recognized as a naturally cross-pollinating species (Jones and Brown 1951). The peak period of pollen shedding in a day occurs from 10 am to 4 pm (Jones and Brown 1951). Dispersal of pollen grains is facilitated by wind. Occasionally, pollen grains can form a visible tunnel (like a swirl) in the air over a large switchgrass planting. Anthesis duration of a panicle is about one to two weeks long, which may be affected by genotypes and environmental factors.

Talbert et al. (1983) reported that bagged inflorescences on average produced less than 1% seed yield as compared to open-pollinated inflorescences on the same plants in an experiment of 33 plants. Their results indicated that switchgrass produces seed primarily through cross pollination while self-pollination is minimal. Using verified tetraploid and octoploid plants in a greenhouse, Taliaferro and Hopkins (1996) concluded that there is a strong genetic barrier between tetraploid and octoploid plants. Martinez-Reyna and Vogel (2002) reported the cross­fertility between octoploid and tetraploid plants is inhibited by a post­fertilization incompatibility system. They observed tetraploid by octoploid or reciprocally fertilized zygotes have a much slower growth than tetraploid by tetraploid or octoploid by octoploid zygotes. However, the cross-fertility between tetraploid upland and lowland plants is substantial or quite normal, and switchgrass outcrossing behavior is enforced by self-incompatibility (Taliaferro and Hopkins 1996; Martinez-Reyna and Vogel 2002). However,

seed set of selfing octoploid plants was as high as 6% (a range of 0-35%) while less than 1% (0-1.5%) selfed seed is produced in bagged inflorescences of tetraploid plants (Taliaferro and Hopkins 1996). Martinez-Reyna and Vogel (2002) noted the self-incompatibility in switchgrass is controlled by a gametophytic prefertilization incompatibility system, which is similar to the S-Z incompatibility system commonly seen in many members of the Poaceae plants. It appears tetraploid plants tend to be less self-fertile than octoploid plants. In later experiments, Taliaferro (2002) reported some lowland switchgrass plants of ‘Alamo’ and ‘Kanlow’ produced more than 20 selfed (S1) seeds from bagged plants while selfing plants of ‘Caddo’, ‘Blackwell’ and ‘Cave-in-Rock’ produced 100 or more seeds.

Recently, molecular markers have been available and used to accurately identify breeding origin of progeny of two controlled crosses (Okada et al. 2010b; Liu and Wu 2012). Okada et al. (2010b) found about 4% progeny of a full-sib cross between a ‘Kanlow’ (female) and an ‘Alamo’ (male) parents are selfed. In an attempt to make a full-sib cross-fertilized mapping population, two lowland plants: ‘NL94 LYE16x13’ (NL94) and ‘SL93 7×15’ (SL93) were grown in a large growth chamber before they were blooming. From the seed harvested from the female parent NL94 plant, 456 progeny were grown in a greenhouse (Liu and Wu 2012). Using 12 simple sequence repeat markers, they identified 279 of the progeny population were selfed progeny of NL94 whereas only 39% (177) were crossed between the two parents. The selfed progeny percentage is much higher than those reported before. However, whether the NL94 plant would produce a similar amount of selfed and crossed progeny when grown under field conditions and when other crossing compatible switchgrass plants are available, is not known. The authors speculated that limited wind flow in the growth chamber enhanced NL94 self-fertilization and reduced cross fertilization. Similarly, it has been observed that some other switchgrass genotypes have a self­pollination rate as high as 50% as cited by Casler et al. (2011).

Switchgrass can be reproduced in asexual ways. In breeding programs, switchgrass clones are normally produced by digging selected field-grown stands and separating into individual ramets, which are used to establish new crossing blocks. But it is difficult to use the labor intensive method to produce a large amount of clones. A micropropagation method to produce switchgrass by nodal culture has been reported (Alexandrova et al. 1996). With nodal segments cultured on an optimized MS medium, 500 plantlet clones can be produced from one parent plant within a period of three months. In our own experience, mature nodes of switchgrass genotypes can be cut into 1- or 2-nodal segments, when grown into a good soil medium and maintained moisture for two or three weeks, new clonal plants will grow out from the buds on the nodes (Fig. 3a and 3b). The greenhouse-type nodal propagating methods may be useful in producing clonal plants of the same size, which are important in quantitative trait loci (QTL) mapping experiments.

It appears there is large variation for breeding behavior in switchgrass. This is not surprising because switchgrass is genetically very diverse. Switchgrass is a predominantly wind-facilitated outcrossing species. Conventional breeding and selection methods have been developed and used based on the major mating system. It is also possible to inbreed switchgrass, at least in some genotypes. Breeding methods exploring the mating systems will be discussed in detail in latter sections.